This action might not be possible to undo. Are you sure you want to continue?
Andrew Baker
[14/01/2008]
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 (2004)
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 5
1.3. Identifying irreducible polynomials 10
1.4. Finding roots of complex polynomials of small degree 14
1.5. Automorphisms of rings and ﬁelds 16
Exercises on Chapter 1 20
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 31
3.1. Algebraic extensions 31
3.2. Splitting ﬁelds and Kronecker’s Theorem 34
3.3. Monomorphisms between extensions 37
3.4. Algebraic closures 39
3.5. Multiplicity of roots and separability 42
3.6. The Primitive Element Theorem 45
3.7. Normal extensions and splitting ﬁelds 47
Exercises on Chapter 3 47
Chapter 4. Galois extensions and the Galois Correspondence 49
4.1. Galois extensions 49
4.2. Working with Galois groups 50
4.3. Subgroups of Galois groups and their ﬁxed ﬁelds 52
4.4. Subﬁelds of Galois extensions and relative Galois groups 53
4.5. The Galois Correspondence and the Main Theorem of Galois Theory 54
4.6. Galois extensions inside the complex numbers and complex conjugation 55
4.7. Galois groups of even and odd permutations 56
4.8. Kaplansky’s Theorem 58
Exercises on Chapter 4 61
Chapter 5. Galois extensions for ﬁelds of positive characteristic 63
5.1. Finite ﬁelds 63
5.2. Galois groups of ﬁnite ﬁelds and Frobenius mappings 66
5.3. The trace and norm mappings 67
Exercises on Chapter 5 69
Chapter 6. A Galois Miscellany 71
6.1. A proof of the Fundamental Theorem of Algebra 71
6.2. Cyclotomic extensions 71
6.3. Artin’s Theorem on linear independence of characters 75
iii
iv CONTENTS
6.4. Simple radical extensions 77
6.5. Solvability and radical extensions 79
6.6. Symmetric functions 82
Exercises on Chapter 6 82
Bibliography 85
Solutions 87
Chapter 1 87
Chapter 2 93
Chapter 3 95
Chapter 4 97
Chapter 5 99
Chapter 6 101
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.
• 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
2 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
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)!
.
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).
1.1. RECOLLECTIONS ON INTEGRAL DOMAINS AND FIELDS 3
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.
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
¸
4 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
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
.
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.
1.2. POLYNOMIAL RINGS 5
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.
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.
6 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
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
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.
1.2. POLYNOMIAL RINGS 7
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.
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.
8 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
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.
1.29. 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)) = 1, hence
a(X)p(X) +b(X)u(X) = 1.
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].
1.2. POLYNOMIAL RINGS 9
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 is 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.
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 = .
10 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
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
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),
1.3. IDENTIFYING IRREDUCIBLE POLYNOMIALS 11
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].
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.
12 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
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).
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.
1.3. IDENTIFYING IRREDUCIBLE POLYNOMIALS 13
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.
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].
14 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
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
♠
♥ ♦
♣
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`a 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
1.4. FINDING ROOTS OF COMPLEX POLYNOMIALS OF SMALL DEGREE 15
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
3
+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
_
.
16 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
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
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) = (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
(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).
1.5. AUTOMORPHISMS OF RINGS AND FIELDS 17
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).
1.59. 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].
18 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
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).
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
.
1.5. AUTOMORPHISMS OF RINGS AND FIELDS 19
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.
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.
20 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
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?
(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.
21
(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 ϕ: k −→ R 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
+ ,
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.
22 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
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.
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
).
23
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/Kto indicate this and 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/L.
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].
(ii) If the dimensions [L : K] and [M : L] are both ﬁnite then so is [M : K] and
[M : K] = [M : L] [L : K].
25
26 2. FIELDS AND THEIR EXTENSIONS
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.
2.12. 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
.
2.2. SIMPLE AND FINITELY GENERATED EXTENSIONS 29
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
.
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
).
30 2. FIELDS AND THEIR EXTENSIONS
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.
(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.
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.
31
32 3. ALGEBRAIC EXTENSIONS OF FIELDS
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.
3.10. 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].
3.1. ALGEBRAIC EXTENSIONS 33
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.
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.
34 3. ALGEBRAIC EXTENSIONS OF FIELDS
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.
3.22. 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.
3.2. SPLITTING FIELDS AND KRONECKER’S THEOREM 35
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.
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.
36 3. ALGEBRAIC EXTENSIONS OF FIELDS
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;
• 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.
3.3. MONOMORPHISMS BETWEEN EXTENSIONS 37
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.
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 (this
happens precisely when p(X) has no root in K).
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.
38 3. ALGEBRAIC EXTENSIONS OF FIELDS
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
.
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).
3.4. ALGEBRAIC CLOSURES 39
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.
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.
40 3. ALGEBRAIC EXTENSIONS OF FIELDS
(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.
(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.
3.4. ALGEBRAIC CLOSURES 41
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)
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 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
42 3. ALGEBRAIC EXTENSIONS OF FIELDS
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.
(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.
3.5. MULTIPLICITY OF ROOTS AND SEPARABILITY 43
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.
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).
44 3. ALGEBRAIC EXTENSIONS OF FIELDS
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)[.
3.69. 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].
3.6. THE PRIMITIVE ELEMENT THEOREM 45
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
)].
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
46 3. ALGEBRAIC EXTENSIONS OF FIELDS
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 separable extension of a ﬁeld of characteristic 0.
Then L has a primitive element.
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).
47
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]?
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.]
48 3. ALGEBRAIC EXTENSIONS OF FIELDS
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
θ: F −→ F restricts to an automorphism θ
F
E
: E −→ E. The proof of the next result is left as
an exercise.
49
50 4. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE
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.
4.7. 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 a
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.
4.2. WORKING WITH GALOIS GROUPS 51
(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
.
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).
52 4. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE
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[.
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 53
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).
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.
54 4. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE
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).
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.
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).
4.6. GALOIS EXTENSIONS INSIDE THE COMPLEX NUMBERS AND COMPLEX CONJUGATION 55
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
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.
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.
56 4. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE
(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
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.
4.7. GALOIS GROUPS OF EVEN AND ODD PERMUTATIONS 57
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) = a
0
+a
1
X + +a
m
X
m
,
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:
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
.
58 4. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE
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)
) = sgnσ
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.
4.26. 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
,
4.8. KAPLANSKY’S THEOREM 59
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.
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,
60 4. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE
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).
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
,
61
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
.
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
.
62 4. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE
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.
(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 characteristic char K = p (where
p > 0 is a prime) 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.
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
.
63
64 5. GALOIS EXTENSIONS FOR FIELDS OF POSITIVE CHARACTERISTIC
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 an 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
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).
5.1. FINITE FIELDS 65
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.
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
.
66 5. GALOIS EXTENSIONS FOR FIELDS OF POSITIVE CHARACTERISTIC
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).
5.22. 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,
5.3. THE TRACE AND NORM MAPPINGS 67
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.
Figure 5.2 shows the subgroup diagram corresponding to the lattice of subﬁelds of F
p
24
shown in Figure 5.1.
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).
68 5. GALOIS EXTENSIONS FOR FIELDS OF POSITIVE CHARACTERISTIC
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
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.
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.
69
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].
(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.
70 5. GALOIS EXTENSIONS FOR FIELDS OF POSITIVE CHARACTERISTIC
(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
nth cyclotomic polynomial
Φ
n
(X) =
t=1,...,n−1
gcd(t,n)=1
(X −ζ
t
n
).
71
72 6. A GALOIS MISCELLANY
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
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
6.2. CYCLOTOMIC EXTENSIONS 73
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.
Proof. By Theorem 1.43 we have [Q(ζ
24
) : Q] = 8. Also,
ζ
6
24
= i, ζ
3
24
=
1
2
+
√
3
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.
74 6. A GALOIS MISCELLANY
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,
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).
6.3. ARTIN’S THEOREM ON LINEAR INDEPENDENCE OF CHARACTERS 75
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
.
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.
76 6. A GALOIS MISCELLANY
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
= 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.
6.17. 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 6.17.
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,
6.4. SIMPLE RADICAL EXTENSIONS 77
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.
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.
78 6. A GALOIS MISCELLANY
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
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 79
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.
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 (or has a quotient group
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) over K.
Then E/L is a radical Galois extension. In particular, if L/K is a radical Galois extension
then so is E/K.
80 6. A GALOIS MISCELLANY
Proof. Suppose that u
d
∈ L. By Proposition 6.22, there is a primitive dth root of unity
ζ ∈ E and the subﬁeld L(ζ, u) E is normal over L, hence 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.
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 radical Galois extension. Then Gal(E/K)
is a quotient group of Gal(E
t
/K), so it is 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.
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
.
6.5. SOLVABILITY AND RADICAL EXTENSIONS 81
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.
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.
82 6. A GALOIS MISCELLANY
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
].
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.
83
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).
(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)
_
84 6. A GALOIS MISCELLANY
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.]
85
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,
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,
88 SOLUTIONS
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
.
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
89
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=min¡k
1
,k
2
¦
(a
k
+b
k
)X
k
,
(
∞
k=k
0
a
k
X
k
)(
∞
=
0
b
X
) =
∞
k=min¡k
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
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]
90 SOLUTIONS
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 X
4
− 2 = (X
2
− 2)(X
2
+ 2) in R[X] and we have to consider
which factor α is a root of. If α
2
−2 = 0 then
ε
α
R[X] = ε
−α
R[X] = R[α] = ¦a +bα : a, b ∈ R¦ ⊆ R, ker ε
α
= ker ε
−α
= (X
2
−2) R[X].
If α
2
+ 2 = 0 then
ε
α
R[X] = ε
−α
R[X] = R[α] = ¦a +bα : a, b ∈ R¦ ,⊆ R, ker ε
α
= 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
.
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.
91
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),
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,
92 SOLUTIONS
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.
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
93
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.
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
94 SOLUTIONS
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.
(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.
95
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.
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.
96 SOLUTIONS
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
.
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] = 4 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).
97
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)[;
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),
98 SOLUTIONS
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
_
,
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)).
99
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.
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.
100 SOLUTIONS
(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 1)1,
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
.
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
101
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
.
(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).
102 SOLUTIONS
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 the 4 primitive 8th roots of unity ±
1
2
±
√
2
2
i as well as the 4th
roots.
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),
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
).
103
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¦.
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,
104 SOLUTIONS
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.
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
)
∼
= Z/4 and has 3 subgroups ¦1¦ ¦1, 4¦ Gal(Q(ζ
5
), 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),
105
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).
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
.
106 SOLUTIONS
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
.
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 ggggnnnnn qq gg qq ggggg nnnn 2 qq ggg 2 nn 2 gg qq n ggggg √ 2 √ gg √ qq3 qq Q( 3 2 ζ3 ) Q( 3 2) Q( 3 2 ζ3 ) qq yyy N N ff I qq yyy ff qq yyy ff qq yyy ff q yyy3 ff3 yyy ff 3 Q(ζ3 ) yyy f Q jj yyy fff jjj 2 jj yyy ff jjj yyy ff jjj yy jjj jjj QS
√ 3
vv vv vv vv 3 vvv {id, (1 2)} {id, (1 3)} {id, (2 3)} vv vv vv vv 2 vv 2 2 vv
ooo  ooo  oo ooo  3 oooo 3  o 3  ooo  o  ooo o  ooo  ooo 
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 (2004)
5. ﬁelds and polynomial rings Basic notions.2. Automorphisms of rings and ﬁelds Exercises on Chapter 1 Chapter 2.1. Normal extensions and splitting ﬁelds Exercises on Chapter 3 Chapter 4.3. Algebraic extensions of ﬁelds 3. Algebraic closures 3. Simple and ﬁnitely generated extensions Exercises on Chapter 2 Chapter 3. Cyclotomic extensions 6. The Primitive Element Theorem 3.2. Galois extensions inside the complex numbers and complex conjugation 4. Finding roots of complex polynomials of small degree 1. The Galois Correspondence and the Main Theorem of Galois Theory 4. Monomorphisms between extensions 3. Polynomial rings 1.1.2.Contents Introduction: What is Galois Theory? Chapter 1. Artin’s Theorem on linear independence of characters iii ii 1 1 1 5 10 14 16 20 25 25 27 30 31 31 34 37 39 42 45 47 47 49 49 50 52 53 54 55 56 58 61 63 63 66 67 69 71 71 71 75 . Identifying irreducible polynomials 1.2.1.2. Galois extensions and the Galois Correspondence 4.3.2. Splitting ﬁelds and Kronecker’s Theorem 3.7. Recollections on integral domains and ﬁelds 1. Fields and their extensions 2. Galois groups of ﬁnite ﬁelds and Frobenius mappings 5. Multiplicity of roots and separability 3. Algebraic extensions 3. convention.5.6.3.3. Working with Galois groups 4. Fields and subﬁelds 2.1.1. A Galois Miscellany 6.5. Kaplansky’s Theorem Exercises on Chapter 4 Chapter 5. Subgroups of Galois groups and their ﬁxed ﬁelds 4. A proof of the Fundamental Theorem of Algebra 6.3.1. etc 1. Galois extensions 4. Galois groups of even and odd permutations 4. The trace and norm mappings Exercises on Chapter 5 Chapter 6. Integral domains.8.4.7.6.4.4. Finite ﬁelds 5. Subﬁelds of Galois extensions and relative Galois groups 4. Galois extensions for ﬁelds of positive characteristic 5.
4.iv CONTENTS 6. Symmetric functions Exercises on Chapter 6 Bibliography Solutions Chapter Chapter Chapter Chapter Chapter Chapter 1 2 3 4 5 6 77 79 82 82 85 87 87 93 95 97 99 101 .5. Simple radical extensions 6.6. Solvability and radical extensions 6.
• ϕ is an epimorphism if it is surjective. 5] or other books containing introductory ring theory. This means that for u. then (q) = (p) if and only if q = up for some unit u ∈ R. Definition. if it is invertible (in which case its inverse is also an isomorphism).. • R is a principal ideal domain if it is an integral domain and every ideal I R is principal. An ideal I R will always mean a twosided ideal. q ∈ R. 1. Under a ring homomorphism ϕ : R −→ S.. i. Hence Z is a principal ideal domain. v ∈ R. if for every s ∈ S there is an r ∈ R with ϕ(r) = s. a ring will always be a ring with unity 1 = 0. C. Example.1. A commutative ring R in which there are no zerodivisors is called an integral domain or an entire ring. Definition. Notice that if p. the ring of integers modulo p.e. • I is a prime ideal if for u. • p ∈ R is prime if (p) R is a prime ideal. 1 . if for r1 .2. 1. ϕ(1) = 1. Let ϕ : R −→ S be a ring homomorphism. Z. r2 ∈ R. i. Details may be found in [3. this is equivalent to the requirement that whenever p  xy with x.CHAPTER 1 Integral domains. R[X] and C[X] are all integral domains. Example.. real numbers.. (iv) The polynomial ring R[X].3.e. R. y ∈ R then p  x or p  y. Let I R be a proper ideal in a commutative ring R.e. in particular. 1 ∈ R is sent to 1 ∈ S. ﬁelds and polynomial rings Basic notions. Fp = Z/p = Z/(p). 1. Most of the rings encountered will also be commutative. and complex numbers. or equivalently if I R. i.. where R is an integral domain. i. Q. etc In these notes. We also write p  x if x ∈ (p). n 0. (ii) If p is a prime. The following rings are integral domains.e. the polynomial rings Z[X]. • ϕ is an isomorphism if it is both a monomorphism and an epimorphism. Q[X]. (i) The ring of integers. • I is a maximal ideal R if whenever J R is a proper ideal and I ⊆ J then J = I. convention.e. Every ideal I Z is principal. so I = (n) for some n ∈ Z which we can always take to be nonnegative. • ϕ is a monomorphism if it is injective. (iii) The rings of rational numbers. uv = 0 =⇒ u = 0 or v = 0. 1. v ∈ R. or equivalently if ker ϕ = {0}.5. i. An ideal I R in a ring R is proper if I = R. 1.1. • I R is principal if I = (p) = {rp : r ∈ R} for some p ∈ R. Definition. ϕ(r1 ) = ϕ(r2 ) =⇒ r1 = r2 . Recollections on integral domains and ﬁelds The material in this section is standard and most of it should be familiar.4. uv ∈ I =⇒ u ∈ I or v ∈ I. 1.
j . This subring is sometimes called the characteristic subring of R. 1. Lemma. The rings Z and Z/n = Z/(n) for n > 0 are often called core rings.1) (u + v)p = up + u v+ u v + ··· + uv p−1 + v p . n η(n) = n1 = −(1 + · · · + 1) if n < 0.6. Proposition.2 1. there is a unique nonnegative integer p 0 such that ker η = (p). Since 1 ∈ R is nonzero. INTEGRAL DOMAINS. hence pt = 0 for any t ∈ R.8. j! or (p − j)!.7. hence Z/ ker η is an integral domain. ker η = (p) is prime ideal and so by Example 1. so since this number is an integer it must be divisible by p. this p is called the characteristic of R and denoted char R. FIELDS AND POLYNOMIAL RINGS 1. j j! (p − j)! j! (p − j)! There are no factors of p appearing in (p − 1)!. −n 0 if n = 0. 1. 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. When discussing a ring with unit R. 1 2 p−1 Now suppose that 1 j p − 1. the rings Z and Fp = Z/p = Z/(p) for p > 0 a prime are called prime rings.e. Now by Proposition 1.9. 1. Proof. p is a prime. So suppose that p > 0. By Example 1. 0. Remark. Example. The Binomial Expansion yields p p−1 p p−2 2 p (1. Then we have p p (p − 1)! (p − 1)! = =p× .7. v ∈ R.5. Proof. 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 n n is a prime. If u. (i) The quotient ring R/I is an integral domain if and only if I is a prime ideal. When considering integral domains.10. Let R be a commutative ring and I R an ideal. p (1. Here is a useful and important fact about rings which contain a ﬁnite prime ring Fp . then (u + v)p = up + v p .2a) p . The quotient monomorphism η : Z/ ker η −→ R identiﬁes Z/ ker η with the subring im η = im η of the integral domain R. Let R be a commutative ring with unit containing Fp for some prime p > 0. If R is an integral domain. allowing us to identify these rings. Consider p = char 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) −→ im η. its characteristic char R is a prime. If p = 0 we are done. i. Theorem (Idiot’s Binomial Theorem).2b) ≡ 0 (mod p). 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. j or equivalently p (1. 1. In particular. (ii) The quotient ring R/I is a ﬁeld if and only if I is a maximal ideal.6(i). But every subring of an integral domain is itself an integral domain. We have p1 = 0 in R.
This is certainly true for the integers Z which are contained in the ﬁeld of rational numbers Q. A commutative ring k is a ﬁeld if every nonzero element u ∈ k is a unit. An element u ∈ R is a unit if it is invertible. Proposition. Proof. Example. j Combining the divisibility conditions of (1. We will denote the set of all invertible elements of R by R× and note that it always forms a group under multiplication.1. 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. Suppose that p = uv for some u.2) with the expansion of (1. 1. Lemma. the prime ring Fp is itself a ﬁeld. Then p  u or p  v. Suppose that u. If n 1. 1. (i) There is a ﬁeld of fractions of D.16. If u = 0. (u + v)p = up + v p . v ∈ k and uv = 0. Fr(D).e. Let R be an integral domain. v must be 0. Proof.17. Definition. 1. and for a prime p > 0. Let D be an integral domain. This is equivalent to requiring that k× = k − {0}. A natural question to ask is whether D is isomorphic to a subring of a ﬁeld. Definition. So at least one of u. the quotient ring Z/n is a ﬁeld if and only if n is a prime. A nonzero element p ∈ R is irreducible if for u. (ii) If ϕ : D −→ F is a ring monomorphism into a ﬁeld F . we can multiply by u−1 to obtain v = u−1 uv = 0. v ∈ R.1). hence v = 0. The ﬁelds Q and Fp where p > 0 is a prime are the prime ﬁelds. 1.13.. 1. which is necessarily unique and is called the (multiplicative) inverse of u in R.12. showing that v is a unit with inverse t. R and C are all ﬁelds. Every ﬁeld is an integral domain.15. there is always a ‘smallest’ such ﬁeld which is unique up to an isomorphism. Theorem. and we might as well assume that u = tp for some t ∈ R. Now let D be an integral domain. 1. RECOLLECTIONS ON INTEGRAL DOMAINS AND FIELDS 3 Hence in R we have p 1 = 0. Let R be a ring. Of course. v ∈ R. Let k be a ﬁeld.11. which contains D as a subring. p = uv =⇒ u or v is a unit.14. Then (1 − tv)p = 0 and so tv = 1. i. We usually write u−1 for this element v. If p ∈ R is a nonzero prime then it is an irreducible. Definition. there is and element v ∈ R for which uv = 1 = vu. 1. D inc ϕ / <F Fr(D) ∃! ϕ . there is a unique homomorphism ϕ : Fr(D) −→ F such that ϕ(t) = ϕ(t) for all t ∈ D ⊆ Fr(D). The familiar rings Q. we obtain the required equation in R.1.
. a] = [ab. a] as an inverse. b = 0}. Corollary.18. b ) ∼ (a. 1] which is a ring homomorphism. so [a. (i) Consider the set P(D) = {(a. 1] ∈ Fr(D) and D with the subring im j ⊆ Fr(D). and so (a d + b c . b] = [0. b) by [a. so Φ is constant on each equivalence class of ∼. it is easy to check that it is a monomorphism. 1] and unit 1 = [1. b] ∈ Fr(D) be nonzero. Then b = 0 and [a. bd]. Corollary. i. In fact. We also have [a. bd). It is now easy to verify that ϕ is a ring homomorphism which agrees with ϕ on D ⊆ Fr(D). a] ∈ Fr(D) satisfy [a. b] and the set of equivalence classes by Fr(D). then Fr(D) ⊆ Fr(F ) = F and Fr(D) is the smallest subﬁeld of F containing D. d ] = [c. so multiplication is also well deﬁned. namely (a . The next three corollaries are left as an exercise. bd]. b] = [1. b]. as we will soon see. If F is a ﬁeld then F = Fr(F ). this is left as an exercise. b d ) ∼ (ad + bc. Therefore we may identify t ∈ D with j(t) = [t. b) ⇐⇒ ab = a b. Now introduce an equivalence relation ∼ on P(D). We deﬁne addition and multiplication on Fr(D) by [a. 1]. We can view D as a subring of Fr(D) using the map j : D −→ Fr(D). b). d]. It is now straightforward to show that Fr(D) is a commutative ring with zero 0 = [0. Now let [a. Φ(a. notice that for any b = 0. b]) = Φ(a. A similar calculation shows that (a c . Fr(D) is a ﬁeld. b) which is equivalent to requiring that a = 0. b][b. We denote the equivalence class of (a. b) : a. FIELDS AND POLYNOMIAL RINGS Proof. if [a . Hence we may deﬁne the function ϕ : Fr(D) −→ F . b) = ϕ(a)ϕ(b)−1 . 1] if and only if a = b. 1) ∼ (a.4 1. 1. [a.e. b][c. ba] = [1. If D is a subring of a ﬁeld F . (ii) Consider the function Φ : P(D) −→ F . d] = [ad + bc. 1. b] = [0. b] ∈ Fr(D). b). . b).19. [b. Then [a. ϕ([a. d] = [ac. We need to verify that these operations are well deﬁned. 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. b] has [b. then Φ(a . [0. bd). b ∈ D. hence addition is well deﬁned. it is necessary to check that this relation is an equivalence relation. b ) ∼ (a. b] + [c. If (a . Of course. b] and [c . For example. 1] if and only if (0. 1] = 1. j(t) = [t. b d ) ∼ (ac. b ] = [a. then (a d + b c )bd = a d bd + b c bd = ab d d + b bcd = (ad + bc)b d . a = 0. This shows that Fr(D) is a ﬁeld. Let [a. 1]. INTEGRAL DOMAINS.
. .. 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 . The rules for algebraic manipulation of such symbols are the usual ones for working with fractions. Polynomial rings Let R be a commutative ring. 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 .e. the identity homomorphism id : D −→ D induces the identity homomorphism (id)∗ = id : Fr(D) −→ Fr(D). pm ∈ R.20 is sometimes said to imply that the construction of Fr(D) is functorial in the integral domain D. D1 ϕ / D2 inc inc Fr(D1 ) ϕ / Fr(D2 ) ∗ Moreover. Then there is a unique induced ring homomorphism ϕ∗ : Fr(D1 ) −→ Fr(D2 ) which satisﬁes ϕ∗ (t) = ϕ(t) whenever t ∈ D1 ⊆ Fr(D1 ). D1 ϕ / D2 θ / D3 D id /D inc inc inc inc inc / Fr(D3 ) Fr(D1 ) ϕ / Fr(D2 ) ∗ θ∗ / Fr(D) Fr(D) id∗ = id 1. Then R[X] is a commutative ring with the constant polynomials 0 and 1 as its zero and unit. • If ϕ : D1 −→ D2 and θ : D2 −→ D3 are monomorphisms between integral domains then θ∗ ◦ ϕ∗ = (θ ◦ ϕ)∗ as homomorphisms Fr(D1 ) −→ Fr(D3 ). We identify r ∈ R with the obvious constant polynomial. p1 . (a) When working with a ﬁeld of fractions we usually adopt the familiar notation a = a/b = [a. b] b for the equivalence class of (a. 1. however as the word quotient is also associated with quotient rings we prefer to avoid using that terminology. . this construction has the following properties.2. . i. . a1 a2 a1 b2 + a2 b1 + = . this allows us to view R as a subring of R[X] and the inclusion function inc : R −→ R[X] is a monomorphism. such p(X) are called polynomials.1. Remarks.20. POLYNOMIAL RINGS 5 1. Let D1 and D2 be integral domains and let ϕ : D1 −→ D2 be a ring monomorphism. Corollary. b1 b2 b1 b2 a1 a2 a1 a2 a1 a2 × = = . b1 b2 b1 b2 b1 b2 The ﬁeld of fractions of an integral domain is sometimes called its ﬁeld of quotients. This consists of elements of form p(X) = p0 + p1 X + · · · + pm X m where m 0 and p0 . (b) Corollary 1. We will make frequent use of the ring R[X] of polynomials over R in an indeterminate X. b). • For any integral domain D.2.21.
.. . . ... . there is a unique ring homomorphism ϕs1 .rn : R[X1 . . . . we deﬁne (1. .. .sn R[X1 . . . . . rn ∈ R. . sn ) = ϕs1 . . INTEGRAL DOMAINS. ... .3) ϕs (p(X)) = p0 + p1 s + · · · + pm sm ∈ S. .. .... sn ∈ S. .. .. Xn ] −→ R. . Xn−1 ][Xn ] for n 1. Again there is an inclusion monomorphism inc : R −→ R[X1 ... sn . These polynomial rings have an important universal property. (ii) is proved by induction on n using (i). . . . . .. .. . (Sketch) (i) For a polynomial p(X) = p0 + p1 X + · · · + pm X m ∈ R[X]. .. . .. sn ] = εs1 . .sn : R[X1 . . . .. R inc ϕ /S = R[X] (ii) For n ∃! ϕs 1 and s1 . then we have the homomorphism εr1 ... . . . . .sn : R[X1 . Xn ] ⊆ S. . Theorem (Homomorphism Extension Property). .. sn and we denote its image by R[s1 . Slightly more generally we may take the inclusion of a subring inc : R −→ S and s1 . .. . sn over R. . .sn (p(X1 . Xn ] −→ S is called evaluation at s1 . . . FIELDS AND POLYNOMIAL RINGS More generally..6 1.. An extremely important special case occurs when we start with the identity homomorphism id : R −→ R and r1 . . Xn ))... It is standard to write p(s1 . Then R[s1 . .. . . (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.sn = incs1 .. . .sn (r) = ϕ(r) for all r ∈ R.. R inc ϕ 8/ S R[X1 . .. Here is an example illustrating how we will use such evaluation homomorphisms.. . Let ϕ : R −→ S be a ring homomorphism. called the subring generated by s1 .sn as the extension of ϕ by evaluation at s1 . . . Xn ] −→ S for which • ϕs1 . sn ] is a subring of S. . Xn ] = R[X1 . • ϕs (X) = s. . . It is then straightforward to check that ϕs is a ring homomorphism with the stated properties and moreover is the unique such homomorphism. Xn ] ∃! ϕs1 . . 1. sn ∈ S.. . ..22.. • ϕs1 . .sn (Xi ) = si for i = 1.. Xn ] which sends each element of R to itself considered as a constant polynomial. .. Xn over R.. we inductively can deﬁne the ring of polynomials in n indeterminates X1 . .rn = idr1 . . ..sn Proof. . . . then εs1 ... . R[X1 . ... . We will refer to ϕs1 . . . .. n. .
. (i) The ring R[X] of polynomials in an indeterminate X over R is an integral domain. hence we also have Q[i] ∼ Q[X]/(X 2 + 1). Let k be a ﬁeld. 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. Example. These evaluation homomorphisms are related by complex conjugation since ε−i (p(X)) = εi (p(X)). 3X 2 d(X) = the divisor . Xn is an integral domain. full details can be found in a basic course on commutative rings or any introductory book on this subject. 1. We refer to the process of ﬁnding q(X) and r(X) as long division of f (X) by d(X). 1. . Let k be a ﬁeld and n 1. . f (X) = the dividend . (ii) The ring R[X1 . Let R be an integral domain. . ﬁnd the quotient and remainder when f (X) = 6X 4 − 6X 3 + − 3X + 1 is divided by d(X) = 2X 2 + 1. r(X) = the remainder . For k = Q. Xn ] of polynomials in the indeterminates X1 .23.26. . . Example. . . later we will write In fact (X Q(i) for this subﬁeld. Notice also that in these examples we have ker ε−i = ker εi = (X 2 + 1) Q[X]. Notice that if we had used −i instead of i. Corollary. evaluation at −i. for which εi (X) = i. 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. In the situation discussed in this result. which is equivalent to the functional equation ε−i = ( ) ◦ εi . Also. Then the polynomial ring k[X1 . .25. We have the evaluation at i homomorphism εi . . Theorem (Long Division). Consider the inclusion homomorphism inc : Q −→ C. .1.2. Proposition. Then there are unique polynomials q(X). As we will make considerable use of such rings we describe in detail some of their important properties. = 2 + 1) is actually a maximal ideal and so Q[i] ⊆ C is a subﬁeld. d(X) ∈ k[X] and assume that d(X) = 0 so that deg d(X) > 0. . First we recall long division in a polynomial ring k[X] over a ﬁeld k. q(X) = the quotient. 1. . 1. giving q(X) = 3X 2 − 3X and r(X) = 1. Let f (X). Solution. ε−i . 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) = 0.27. . In the usual notation we have the following calculation. .24. POLYNOMIAL RINGS 7 1. Xn over R is an integral domain. b ∈ Q. . . the following names are often used. we would also have ε−i Q[X] = Q[i]. Xn ] in the indeterminates X1 .
(mod 5). Proof.. INTEGRAL DOMAINS.29. We have the following calculation.15 we already know that p(X) is irreducible if it is prime. f (X) = 10X 5 + 6X 4 − 6X 3 + 3X 2 − 3X + 1 ≡ X 4 + 4X 3 + 3X 2 + 2X + 1 Solution. 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).31. Proposition. Then by Corollary 1. I = (h(X)) for some h(X) ∈ k[X]. g(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 giving q(X) = 3X 2 + 2X and r(X) = 1. 1. An important consequence of Theorem 1. g(X). This shows that p(X)  u(X) or p(X)  v(X) and so p(X) is prime. But since p(X) is irreducible. Here the greatest common divisor gcd(f (X).8 1. g(X) is the monic polynomial of greatest degree which divides both of f (X). u(X)). b(X) ∈ k[X] such that a(X)p(X) + b(X)u(X) = gcd(p(X). g(X)) of f (X). 4−1 ≡ 4 (mod 5). By Lemma 1. FIELDS AND POLYNOMIAL RINGS 6X 3 1. gcd(p(X). . i. Then a nonconstant polynomial p(X) ∈ k[X] is an irreducible if and only if it is a prime. Example.26 is the following which makes use of the Euclidean Algorithm. ﬁnd the quotient and remainder when f (X) = 10X 5 + 6X 4 − + 3X 2 − 3X + 1 is divided by d(X) = 2X 2 + 1. For k = F5 . First notice that working modulo 5 we have (mod 5).e. (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]. Theorem. Let f (X). Let k be a ﬁeld and X an indeterminate. there are a(X). (i) Every ideal I k[X] is principal. Notice also following multiplicative inverses in F5 : 2−1 ≡ 3 (mod 5). Let k be a ﬁeld and X an indeterminate.28. v(X) ∈ k[X].30. (iv) The quotient ring k[X]/(p(X)) is a ﬁeld if and only if p(X) is an irreducible in k[X]. Then there are a(X). hence a(X)p(X) + b(X)u(X) = 1. Corollary. So suppose that p(X) is irreducible and that p(X)  u(X)v(X) for u(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]. 1. u(X)) = 1. 1. 3−1 ≡ 2 (mod 5). g(X) ∈ k[X] be nonzero.29. b(X) ∈ k[X] such that a(X)f (X) + b(X)g(X) = gcd(f (X). Let k be a ﬁeld and X an indeterminate.
Proposition (Unique Factorization Property). there are q(X).30. Now let p(X) ∈ I. 1. . then either f (X) is already irreducible. q(X)). hence must equal it. Then for any q(X) ∈ k[X] with q(X) ∈ (p(X)). where q(X) = p−1 p0 + p−1 p1 X + · · · + p−1 pd−1 X d−1 + X d . and therefore deg fj (X) < deg f (X). so a(X)p(X) + b(X)q(X) = 1. then provided d = deg p(X) > 0. or f (X) = f1 (X)f2 (X) with both factors of positive degree. each pi (X) is prime. p(X) is an irreducible by (iii). (Sketch) Existence is proved by induction on the degree of f (X) and begins with the obvious case deg f (X) = 1. (i) Let I k[X] and assume that I = (0). we have for some pd = 0. c is unique and the sequence of polynomials p1 (X). Suppose that p(X) is irreducible (and hence is nonzero). So we can always ﬁnd a monic polynomial as the generator of a given ideal. Thus I ⊆ (p(X)) ⊆ I and therefore I = (p(X)). Then there must be at least one element of I with positive degree and so we can choose h(X) ∈ I of minimal degree. Then by Proposition 1. pk (X) ∈ k[X] are irreducible monic polynomials. (ii) This follows from Proposition 1.2. But gcd(p(X). POLYNOMIAL RINGS 9 Proof. b(X) ∈ k[X] for which / a(X)p(X) + b(X)q(X) = gcd(p(X). by Corollary 1.29 we can ﬁnd suitable a(X).1. This gives the inductive step.30. it is follows that if k[X]/(p(X)) is a ﬁeld then because it is an integral domain. hence divides one of the qj (X). . say d = deg h(X). and p1 (X). notice that if p(X) ∈ k[X]. . and so we see that k = .31(i). In connection with Theorem 1. q(X)) = 1 since p(X) is irreducible. Moreover. qj (X) ∈ k[X] are irreducible monic polynomials. After cancelling common factors we obtain qk+1 (X) · · · q (X) = 1. 1. so we must have r(X) = 0. suppose that p1 (X) · · · pk (X) = q1 (X) · · · q (X) where pi (X). this would contradict the minimality of d. . (iii) This follows from Proposition 1. If deg f (X) > 1. . By Long Division. Proof. (p(X)) = (q(X)). where c ∈ k.32. we also have r(X) = p(X) − q(X)h(X) ∈ I. Every nonconstant polynomial f (x) ∈ k[X] has a factorization f (x) = cp1 (X) · · · pk (X). To prove uniqueness. . p(X) = p0 + p1 X + · · · + pd X d = pd q(X). . . showing that p(X) = q(X)h(X). and this monic polynomial is unique. 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. pk (X) is unique apart from the order of the terms. By reordering we can assume that pi (X) = qi (X) and k . r(X) ∈ k[X] such that p(X) = q(X)h(X) + r(X) and deg r(X) < d or r(X) = 0. d d d This easily implies that as ideals of k[X]. .6(i). Since p(X) and h(X) are in the ideal I.33. If r(X) = 0. Remark. (iv) Since k[X] is an integral domain and not a ﬁeld.
Corollary. ud ∈ k. . Definition. The version here is slightly more general than the more usual one which corresponds to taking s = 0. Working in Z[X].10 1. Then the sequence of roots u1 . Then f (X) is irreducible in Q[X] and hence also in Z[X]. ≡ (X − 1)p . Let p 2 be a prime. FIELDS AND POLYNOMIAL RINGS 1. Our next result is a special case of the Eisenstein Irreducibility Test. h(X) ∈ R[X] with deg g(X) > 0 and deg h(X) > 0. we say that f (X) has a proper factorisation over R if f (X) = g(X)h(X) for some g(X). bj ∈ k and b0 + b1 X + · · · + bn X n = 0. Corollary. 1. Suppose that f (X) ∈ k[X] factors into linear factors f (X) = c(X − u1 ) · · · (X − ud ). If k is a ﬁeld and X an indeterminate. 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). INTEGRAL DOMAINS. So to ﬁnd factors of f (X) it is suﬃcient to look for factors in Z[X]. . The elements of k(X) are fractions of the form a0 + a1 X + · · · + am X m b0 + b1 X + · · · + bn X n with ai . If f (X) ∈ Z[X] then we can also consider f (X) as an element of Q[X]. . d − 1.36. . The number of distinct roots of a nonconstant polynomial f (X) ∈ k[X] is at most deg f (X). Then the polynomial Φp (X) = 1 + X + · · · + X p−1 ∈ Z[X] is irreducible in Q[X] and hence also in Z[X]. • a0 ≡ 0 (mod p2 ). vr are the distinct roots. then f (X) = c(X − v1 )m1 · · · (X − vr )mr . where u1 . If R = Z or Q. Identifying irreducible polynomials When k is a ﬁeld. Let us consider factorisation of polynomials over Q. Choose ai ∈ Z so that f (X) = a0 + a1 (X − s) + · · · + ad−1 (X − s)d−1 + ad (X − s)d . 1. where mi > 0 and this factorization is unique apart from the order of the pairs (vi . 1. . Proposition (Eisenstein Test). ud is unique apart from the order. .34. Proof. .3. . mi ). k(X). Proposition (Gauss’s Lemma). . then the ﬁeld of fractions of k[X] is the ﬁeld of rational functions. . 1. we will need some eﬀective methods for deciding when a polynomial in k[X] is irreducible. if v1 . Let f (X) ∈ Z[X]. Let f (X) ∈ Z[X] and s ∈ Z. . . Suppose that p > 0 is a prime for which the following three conditions hold: • ak ≡ 0 (mod p) for k = 0.38.35. In particular. . . • ad ≡ 0 (mod p).37. where d = deg f (X). 1. . Then f (X) has a proper factorisation over Z if and only it has a proper factorisation over Q.39. Example. 1. .
Φ7 (X) = 1 + X + X 2 + X 3 + X 4 + X 5 + X 6 . . For example. So the Eisenstein Test can be applied here with s = 1 to show that Φp (X) is irreducible in Z[X]. Example. As examples we have the irreducible polynomials Φ2 (X) = 1 + X.2a). Φ5 (X) = 1 + X + X 2 + X 3 + X 4 . . p − 1. Hence Φp (X) ≡ (X − 1)p−1 Also. Φ3 (X) = 1 + X + X 2 .5b) Φn (X) = . X 3 − 1 = (X − 1)(X 2 + X + 1) = Φ1 (X)Φ3 (X). n for which gcd(n.1.5a) Φd (X). p divides p k when k = 1. Of course. 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). 1. IDENTIFYING IRREDUCIBLE POLYNOMIALS 11 since by (1. p 1 giving (1. In particular. k) = 1 = (Z/n)×  = number of units in Z/n = number of generators of the cyclic group Z/n. (1. 1.40. . Φd (X) dn d<n The degree of Φn (X) involves a function of n probably familiar from elementary Number Theory. X 2 − 1 = (X − 1)(X + 1) = Φ1 (X)Φ2 (X). . If we know Φk (X) for k < n. . . . if p 2 is a prime then ϕ(p) = p − 1. ϕ(1) = 1. . then Xn − 1 (1. X 5 − 1 = (X − 1)(X 4 + X 3 + X + 1) = Φ1 (X)Φ5 (X). Φ11 (X) = 1 + X + X 2 + X 3 + X 4 + X 5 + X 6 + X 7 + X 8 + X 9 + X 10 . Definition. X 6 − 1 = (X − 1)(X + 1)(X 2 + X + 1)(X 2 − X + 1) = Φ1 (X)Φ2 (X)Φ3 (X)Φ6 (X).4) Φp (X) = (X − 1)p−1 + cp−2 (X − 1)p−2 + · · · + c1 (X − 1) + c0 with cr ≡ 0 (mod p) and c0 = p.5a). Cyclotomic polynomials can be computed recursively using Equation (1.41. n These are examples of the cyclotomic polynomials Φn (X) ∈ Z[X] which are deﬁned for all 1 by Xn − 1 = dn = p! k! (p − k)! (mod p) = p ≡ 0 (mod p2 ). . where the product is taken over all the positive divisors of n. The Euler function ϕ : N −→ N is deﬁned by ϕ(n) = number of k = 1. X 4 − 1 = (X − 1)(X + 1)(X 2 + 1) = Φ1 (X)Φ2 (X)Φ4 (X).3.
the latter can be used to solve Equation (1. if p and q are distinct primes.5) lead to (1. 1. . then ϕ(pq) = pq − (ϕ(p) + ϕ(q) + ϕ(1)) = pq − (p − 1) − (q − 1) − 1 = (p − 1)(q − 1).2.n)=1 Proof. if p is a prime and r > 0. the cyclotomic polynomial Φn (X) is irreducible in Q[X] and hence in Z[X]. i. then s 1 Furthermore. Φn (X) = t=1. The complex roots of Φn (X) are the primitive nth roots of unity.43. (1. k ζn = e2πik/n (0 k n − 1. Theorem. . 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) . 2πi/6 =e πi/3 √ 1 3 i.8) (1.11) Φn (X) = dn (X n/d − 1)µ(d) . 1. Notice that we can inductively determine ϕ(n) using this equation. . It is also true that whenever m. n) = 1. if p. 1.. (1.. n) = 1). For example. So for example. For those who know about the M¨bius function µ (which takes values 0. ϕ(n) = ϕ(pr1 ) · · · ϕ(prs ).. when gcd(m. Thus if n = pr1 · · · prs where p1 < p2 < · · · < ps are the prime factors of n and rj > 0.. For each n 1. We will give a reformulation and proof of this in Theorem 6.6) dn ϕ(d) = n. t (X − ζn ). n − 1. then Notice that as a result. ±1) o and M¨bius inversion. For n = 6 we have ζ6 = e Then ϕ(6) = 2 and 5 Φ6 (X) = X 2 − X + 1 = (X − ζ6 )(X − ζ6 ).44. . Hence.n−1 gcd(t. q are distinct primes. Remark.9) ϕ(mn) = ϕ(m)ϕ(n). Then every k complex nth root of unity has the form ζn = e2πik/n for k = 0. ϕ(n) is even when n > 2.10) ϕ(n) = d dn Similarly. Example.42. the formulae of (1. and the number of these is deg Φn (X) = ϕ(n).12 1. 1. n are coprime. We think of ζn = e2πi/n as the standard complex primitive nth root of unity. FIELDS AND POLYNOMIAL RINGS It can be shown that for each natural number n.e. . giving o n µ(d) .. s 1 ϕ(pr ) = (p − 1)pr−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.6) for ϕ. INTEGRAL DOMAINS. then using standard properties of µ. = + 2 2 It is also worth recording a related general result on cyclic groups. gcd(k. (1.7) (1.
45. Now if θG (d) < ϕ(d) for some d. . Let f (X) ∈ k[T ][X] and s ∈ k[T ]. Recall the Euler ϕfunction satisﬁes Equation (1. 1. Then f (X) is irreducible in k(T )[X] and hence also in k[T ][X].46. Proposition (Gauss’s Lemma). so G is cyclic. Suppose that p(T ) ∈ k[T ] is an irreducible for which the following three conditions hold: • ak ≡ 0 (mod p(T )) for k = 0. Proposition. . the number of such elements of C is ϕ(n). hence there must be an element of order G. where k is a ﬁeld. If R = k[T ] or k(T ). Then f (X) has a proper factorisation over k[T ] if and only it has a proper factorisation over k(T ).1. 1.12) dG θG (d) = dG ϕ(d). there are ϕ(G) elements of order G.12).48. . IDENTIFYING IRREDUCIBLE POLYNOMIALS 13 1. Then G is cyclic and in particular is abelian. again we state a version which is slightly more general than the usual one which corresponds to the case where s = 0. this gives θG (d) ϕ(d). . A polynomial f (X) ∈ k[T ][X] is an element of k(T )[X]. n) = 1. d − 1. By Lagrange’s Theorem. Let d be a divisor of G. Lemma. Let f (X) ∈ k[T ][X]. 1. Here is another version of the Eisenstein Test. Proof. the cyclic subgroup g G has ϕ(d) generators.45. Let θG (d) denote the number of elements in G of order d. θG (d) = 0 unless d divides G. Let G be a ﬁnite group satisfying the following condition: • For each n 1. each of order d. So θG (d) dG dG ϕ(d). Choose ai ∈ k[T ] so that f (X) = a0 + a1 (X − s) + · · · + ad−1 (X − s)d−1 + ad (X − s)d . h(X) ∈ R[X] with deg g(X) > 0 and deg h(X) > 0.3. Then an element g r ∈ C is a generator if and only if gcd(r. we would have a strict inequality in place of Equation (1. Hence θG (d) = ϕ(d) for all d. In particular. dG θG (d). • ad ≡ 0 (mod p(T )). hence G = dG ϕ(d).47. for each element g ∈ G of order d. Let n 1 and C = g be a cyclic group of order n and a generator g. • a0 ≡ 0 (mod p(T )2 ).6). Combining these we obtain (1. Since G= we have G = dG {g ∈ G : g = d}. there are at most n solutions of xn = ι in G. where d = deg f (X). By Proposition 1. This leads to a useful group theoretic result. we say that f (X) has a proper factorisation over R if f (X) = g(X)h(X) for some g(X). The above results for polynomials over Q and Z have analogues over the ﬁeld of fractions k(T ) and polynomial ring k[T ]. Proposition (Eisenstein Test). As there are at most d such elements g in G. .
i. Cubic polynomials: Cardan’s method.. Suppose that x ∈ C is a root of f (X). Clearly ﬁnding the roots of f (X) is equivalent to ﬁnding those of g(X). In practice we will usually have k = R or k = C. 1. 3]. a 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 a1 a2 2a3 1 g(X) = f (X − a2 /3) = X 3 − a1 − a2 X + a0 − + 2 2 3 3 27 ∈ C[X].49.13) 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 . Example.14 1. Let k be a ﬁeld. FIELDS AND POLYNOMIAL RINGS 1.e. Finding roots of complex polynomials of small degree ♠ ♥ ♦ ♣ In this section we work within the complex numbers and take k ⊆ C. 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 . 27 p3 + q = 0. 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 = 0.4. so we may as well assume that we want to ﬁnd the complex roots of f (X) = X 3 + pX + q ∈ C[X]. 27u3 p 3u 3 x3 + px + q = 0. Let us consider the cases of cubic (degree 3) and quartic (degree 4) polynomials. For monic linear (degree 1) or quadratic (degree 2) polynomials. (1. p . 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. INTEGRAL DOMAINS. Then the polynomial X n − T is irreducible in k(T )[X]. 27 p3 = 0.
diﬀering by factors of the form ω r for r = 0. The following method of ﬁnding roots of quartics was publicised by Cardan who attributed it to his student Lodovicio Ferrari. 5 + 26 ω. since 4p3 4p3 −q − q 2 + −q − + 27 1 27 = −27 = 2 . we also have √ −1 √ .1. Here we have x = u − 1/u. Notice that 5 + 26 > 0 and 5 − 26 < 0. FINDING ROOTS OF COMPLEX POLYNOMIALS OF SMALL DEGREE 15 we obtain the desired root of f (X) as x = u − p/3u. 2). 3 16 256 3 4 . Applying the method above. namely 3 √ 5 + 26 − 1 3 3 5+ √ = 26 5+ 3 2 √ 26 − 1 . 1.50. 2 where ω = e2πi/3 is a primitive cube root of 1. Example. Find the complex roots of the polynomial f (X) = X 3 + 3X − 10 ∈ R[X].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 . A general monic quartic polynomial f (X) = X 3 + 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 . Notice that one of these is real. we reduce to the quadratic equation U 2 − 10U − 1 = 0 √ √ √ whose roots are 5 ± 26 ∈ R. 5 − 26 = 5 + 26 √ Now 5 + 26 has the complex cube roots √ √ √ 3 3 3 5 + 26. so the 3 complex roots of f (X) are 3 5+ √ 26 − 3 √ 5 + 26 1 ωr (r = 0. Notice that we have a choice of 2 values for u3 and for each of these a choice of 3 values for u. 4p3 q2 + 27 q 1 − + 2 2 1.14) or more precisely as (1. Solution. 2 + 4p3 /27) 3 q − (q 4p 4p3 −q + q 2 + 27 it is easy to verify that there are in fact only 3 choices of the root x which we can write symbolically as q2 (1. However. √ 5 + 26 Quartic polynomials: Ferrari’s method. 5 + 26 ω 2 .4. 1.
17) is a cubic in y. This can be done by requiring the vanishing of the discriminant 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. Automorphisms of rings and ﬁelds 1. Later we will see that this is not true in general for polynomials of degree at least 5. we can use the method of solution of cubics to ﬁnd a root y = t say. 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]. Then z 2 = x4 + 2x2 y + y 2 = −px2 − qx − r + 2xy + y 2 = (2y − p)x2 − qx + y 2 − r. β : R −→ R is also an automorphism of R as is the inverse of α. Now choose y to make the last quadratic expression in x a square.5. FIELDS AND POLYNOMIAL RINGS Clearly ﬁnding the roots of f (X) is equivalent to ﬁnding those of g(X). Remark. Definition.52.53. 1. 1. q 2 − 4(2y − p)(y 2 − r) = 0. The composition α◦β of two automorphisms α. The set of all automorphisms of R over R0 is denoted AutR0 (R).17) (2y − p)x2 − qx + (y 2 − r) = (Ax + B)2 . Proposition..16 1. Proof. The identity function id = idR : R −→ R is an automorphism. i. For a ring R with a subring R0 ⊆ R. z such that z = x2 + y (we will ﬁx the values of these later).54. Since Equation (1. INTEGRAL DOMAINS. Thus taking the two square roots of the right hand side we obtain 4 values for x. then α restricts to the identity on R. Then for Equation (1. • An automorphism of R is a ring isomorphism α : R −→ R.e. Aut(R) = {id}. (1. Proposition. The argument for AutR0 (R) is similar.51. Let R be one of the core rings Z or Z/n with n > 1. α(r) = r for all r ∈ R. Then (i) The only automorphism of R is the identity. 1. The set of all such automorphisms is denoted Aut(R). Suppose that x is a root and introduce numbers y. which we write symbolically as x = ± −t ± (Ax + B).16) (1. • An automorphism of R over R0 is a ring isomorphism α : R −→ R for which α(r) = r whenever r ∈ R0 . 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). Hence Aut(R) forms a group under composition. (ii) If S is a ring containing a core ring R and α ∈ Aut(S).. whence x2 = −t ± (Ax + B). this is one of the great early successes of this theory. Aut(R) and AutR0 (R) form groups under composition of functions. i.16) we have (x2 + t) = (Ax + B)2 . Consequently such polynomials are said to be solvable by radicals. Let R be a ring and R0 ⊆ R a subring.e. Hence. 1. Aut(S) = AutR (S).
56. 1.54(ii) we obtain another useful result. Proposition.58. −k α(0) if k = 0 1 + · · · + 1 if k > 0. k α(k1) = −(α(1) + · · · + α(1)) if k < 0. Corollary.1. Recalling Deﬁnition 1.e. Then the induced homomorphism gives an automorphism α∗ : Fr(D) −→ Fr(D). α −→ α∗ is an isomorphism and hence Aut(Q) = {id}. . Aut(R) = Autk (R).57. (ii) For α ∈ Aut(S). Here we take D = Q[X] and Fr(Q[X]) = Q(X).5. Thus α = id. 1. There is a monomorphism of groups ( )∗ : Aut(D) −→ Aut(Fr(D)).20 to show that (α−1 )∗ ◦ (α)∗ = id = (α)∗ ◦ (α−1 )∗ . we can apply Corollary 1. γ : Q(X) −→ Q(X). then every automorphism of R restricts to the identity on k. AUTOMORPHISMS OF RINGS AND FIELDS 17 Proof. Given α. The inverse automorphism α−1 : D −→ D also gives rise to an induced homomorphism (α−1 )∗ : Fr(D) −→ Fr(D). Let D be an integral domain and α : D −→ D be an automorphism. Example. Proof. The homomorphism ( )∗ : Aut(Q[X]) −→ Aut(Q(X)).55. every element has the form k1 for some k ∈ Z. Combining this example with Proposition 1. i. we have an example which shows that the monomorphism of Corollary 1. 1.36. Proposition. 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]. (i) For such a core ring R. α(1) = 1 and a similar argument to that for (i) shows that α(r) = r for all r ∈ R. For an automorphism α of R. α(1) + · · · + α(1) if k > 0. If R is a ring containing k as a subring. α −→ α∗ . 1.59.. The ﬁeld of fractions of the ring of integers Z is the ﬁeld of rationals Q.56 need not be an epimorphism. Let k be one of the prime ﬁelds Q or Fp with p > 0 prime. 1. −k 0 if k = 0 =k1. Hence (α)∗ is invertible with inverse (α−1 )∗ . Since α−1 ◦ α = id = α ◦ α−1 . The homomorphism ( )∗ : Aut(Z) −→ Aut(Q). k = −(1 + · · · + 1) if k < 0. the induced homomorphism α∗ : Fr(D) −→ Fr(D) exists and we need to show it has an inverse. Example.
αA2 A1 = αA1 ◦ αA2 . Notice that GL2 (k) has another interesting subgroup called the aﬃne subgroup. 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]. this is also an epimorphism and we leave the proof of this as an exercise. PGL2 (k) = GL2 (k)/ Scal2 (k). 1. Let k be a ﬁeld and X an indeterminate. It is straightforward to check that for A1 . By applying ( )∗ to this we obtain a homomorphism (known as a fractional linear transformation) αA : k(X) −→ k(X) for which a11 X + a12 . t = 0} GL2 (k). In fact. a11 a12 GL2 (k) = : aij ∈ k. Then Autk (k[X]) and hence Autk (k(X)). = Proof. Example. (ii) In fact. Then (i) Autk (k(X)) contains a subgroup isomorphic to PGL2 (k). a = 0 0 1 GL2 (k). We begin by choosing the element (a11 X + a12 )/(a21 X + a22 ) ∈ k(X) and then using Theorem 1.55 (or doing a direct calculation) we see that αA−1 is the inverse of αA an so αA ∈ Autk (k[X]). Example.60. In fact. contains a subgroup isomorphic to Aﬀ 1 (k).18 1. INTEGRAL DOMAINS.22(i) to obtain a homomorphism k[X] −→ k(X) that sends X to (a11 X+a12 )/(a21 X+a22 ). A2 ∈ Aﬀ 1 (k). 1. which is easily seen to be a monomorphism.22(i) to obtain a homomorphism αA : k[X] −→ k[X] with αA (X) = aX + b. 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. Aﬀ 1 (k) = a b : a. Autk (k[X]) ∼ Aﬀ 1 (k). (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). t) : t ∈ k. The group of invertible 2 × 2 matrices over k is the 2 × 2 general linear group over k. αA (X) = a21 X + a22 . (note the order!) hence there is a homomorphism of groups Aﬀ 1 (k) −→ Autk (k[X]). Composing with ( )∗ we see that there is a monomorphism Aﬀ 1 (k) −→ Autk (k(X)). For this we use the element aX + b ∈ k[X] together with the extension result of Theorem 1. Autk (k(X)) ∼ PGL2 (k). The quotient group is called the 2 × 2 projective general linear group over k. b ∈ k. = Proof. A −→ αA−1 . a11 a22 − a12 a21 = 0 a21 a22 The scalar matrices form a normal subgroup Scal2 (k) = {diag(t. FIELDS AND POLYNOMIAL RINGS Let k be a ﬁeld. Using the same line of argument as in the proof of Proposition 1. Let k be a ﬁeld and X an indeterminate.61.
but an easy argument then shows the general case. this is not an injection in general since for each scalar matrix diag(t. p1 (X) ∈ C[X] and deg p0 (X) < deg q(X). Let α ∈ AutC (C(X)). Now let c ∈ C.1.57. Thus aX + b f (X) = = constant cX + d a b must be invertible. where α(X) = f (X). if deg p(X) deg q(X) then the number of poles of f counted with algebraic multiplicity is also ord f .62. and the matrix c d 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. First we note that Q ⊆ R is a subring and if α ∈ Aut(R) then α(q) = q for q ∈ Q by Example 1. Also. t). αdiag(t. where qn ∈ Q. Then the number of solutions counted with algebraic multiplicity of the equation f (z) = c turns out to be ord f . Therefore passing to the quotient PGL2 (k) = GL2 (k)/ Scal2 (k) we obtain a monomorphism PGL2 (k) −→ Autk (k(X)). There is one case where Scal2 (k) is the trivial group.t) is the identity function. deg q(X)}. Then for a continuous function f : R −→ R. We give a sketch proof for the case of k = C. if deg p(X) > deg q(X) then we can write p0 (X) f (X) = p1 (X) + . Proposition. In fact it is easy to see that Scal2 (k) GL2 (k) is the kernel of this homomorphism. However.5. 1.t) (X) = tX = X. There is an associated homomorphism of groups GL2 (k) −→ Autk (k(X)) sending A to αA . 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 . q Now it is easy to see that since α is invertible so is the function f . Now we turn to a more familiar ﬁeld R. namely k = F2 . the real numbers. Then the number of poles of f counted with algebraic multiplicity is p0 deg p1 (X) + ord . its value at r depends on its values on Q since f (r) = f ( lim qn ) = lim f (qn ). hence Aut(R) = {id}. AUTOMORPHISMS OF RINGS AND FIELDS 19 Again we ﬁnd that αA2 A1 = αA1 ◦ αA2 . hence ord f = 1. then the order of f (X) is ord f = max{deg p(X). If we write f (X) = p(X) q(X) where p(X). q(X) where p0 (X). deﬁned on C with the poles of f deleted. But this can only happen if the function f is injective which means that all of these numbers must be 1. (ii) To show that every automorphism of k(X) is a fractional linear transformation is less elementary. actually this argument can be modiﬁed to work for any algebraically closed ﬁeld. q(X) ∈ C[X] have no common factors of positive degree. There is an associated rational (hence meromorphic) function f given by z −→ f (z). . Finally. The only automorphism of the ﬁeld R is the identity function. t −1 showing that αdiag(t. n→∞ n→∞ We will show that an automorphism α ∈ Aut(R) is continuous. Proof.
So α preserves order and ﬁxes rational numbers. Let R be a ring. INTEGRAL DOMAINS. It is easy to see that complex conjugation ( ) : C −→ C is an automorphism of C and ﬁxes every real number. However.. we have α(s2 ) = α(s)2 . 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. If R is a commutative ring with unit containing the prime ﬁeld Fp for some prime p > 0.e.3. 1. so there are many elements of order 2 in the group Aut(C). A is an arbitrary commutative ring. . (a) Any subring R ⊆ C. a1n . an1 . . Thus every automorphism of R is continuous function which ﬁxes all the rational numbers. in fact. . 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}. show that the function ϕ : R −→ R given by ϕ(t) = tp . . . −δ < y − x < δ) we have −δ = α(−δ) < α(y) − α(x) < α(δ) = δ.4. (c) The ring of n × n matrices over A. Hence. (a) Show that every subring S ⊆ R is also an integral domain. AutR (C) = {id. 1. . ij R = Matn (A) = . y ∈ R. ﬁnd the characteristic char R and the characteristic subring of R.63. . (b) The polynomial ring R = A[X].e.1.5. Now for α ∈ Aut(R) and s ∈ R. In (b) and (c).2. a11 . :a ∈A . This shows that α is continuous at x. i. Now let x ∈ R and ε > 0. give an example to show that a subring of R need not be a ﬁeld. ann 1. If we try to determine Aut(C) the answer turns out to be much more complicated. . . Note that given an automorphism α ∈ Aut(C). Let R be an integral domain. . 1. Determine which of these rings is an integral domain. x<y ⇐⇒ 0<y−x ⇐⇒ y − x = t2 for some nonzero t ∈ R. Show that {n ∈ Z : n > 0 and n1 = 0} = {n ∈ Z : n > 0 and nr = 0 for all r ∈ R}.20 1.. hence α(y) − α(x) < δ ε. ( )}. ( ) ∈ AutR (C). For each of the following rings R. the composition α ◦ ( ) ◦ α−1 is also self inverse. 1. hence it must be the identity function. Give examples to show that ϕ need not be surjective or injective. What is the relationship between char S and char R ? (b) If R is a ﬁeld. x < y =⇒ α(y) − α(x) = α(t)2 for some nonzero t ∈ R =⇒ α(x) < α(y). deﬁnes a ring homomorphism. Let R and S be rings with unity and Q S a prime ideal.. ε. Remark. FIELDS AND POLYNOMIAL RINGS First recall that for x. Exercises on Chapter 1 1. . 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 ( ).
Let k be a ﬁeld. k=0 with ak ∈ k.b : R[X] −→ R[X] for which ψa. 1. (i) If f (X) ∈ k[X]. Which of the following polynomials in Z[X] is irreducible? X 3 − X + 1. Show that Q ∩ R = P .b ◦ ψc. 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. (b) If R ⊆ S is a subring. (ii) If p(X) ∈ k[X] is a prime then so is ψa. are the results in parts (a) and (b) still true? [Hint: look for a counterexample. determine ψa. suppose that Q S is a prime ideal for which P ⊆ Q. 1. the deg ψa. Let p > 0 be a prime. Q(i) is a ﬁeld and Z(i) is a subring of Q(i).b (r) = r if r ∈ R and ψa. (iii) If p(X) ∈ k[X] is an irreducible then so is ψa. b ∈ R. X 5 − X + 1. If a is a unit. 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. Q(i) = {u + vi : u. 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. show that ψa.21 (a) If ϕ : R −→ S is a ring homomorphism.b (X) = aX + b. 1. (c) Show that inc∗ is an isomorphism. show that ϕ−1 Q = {r ∈ R : ϕ(r) ∈ Q} ⊆ R is a prime ideal of R. Prove the following. d ∈ R. 1. (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. show that Q ∩ R is a prime ideal of R. 1. v ∈ Q} ⊆ C. show that there is a unique ring homomorphism ψa. 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. b ∈ k with a = 0. (b) Show that the inclusion homomorphism inc : Z(i) −→ Q(i) extends to a monomorphism inc∗ : Fr(Z(i)) −→ Q(i). (a) Show that Z(i) and Q(i) are subrings of C.7. 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 f (X) is irreducible.9.b (f (X)) = deg f (X). (b) Show that ∞ ak X k ∈ k[[X]] is a unit if and only if a0 = 0.6.b (p(X)).8. Consider the sets Z(i) = {u + vi : u. R be a ring with unit and ϕ : k −→ R a ring homomorphism. (c) If the word ‘prime’ is replaced by ‘maximal’ throughout. Show that ϕ is a monomorphism. . (a) If a.d . Taking k = Q. v ∈ Z} ⊆ C. Taking k = F3 . X 3 + 2X + 1. (b) Now suppose that R = k is a ﬁeld and a.10.] (d) If R ⊆ S is a subring and P R is a maximal ideal. If c. If f (X) is irreducible. 1.12. X 5 + X − 1.b is an isomorphism and ﬁnd its inverse.b (p(X)). 5X 3 − 10X + X 2 − 2. so Fr(Z(i)) = Q(i). X 3 + X − 1.11. Also show that Z(i) is an integral domain. Let R be a commutative ring.
Determine the image εω Q[X] ⊆ C.13. = 1.21. Determine the image εα Q[X] ⊆ C and the ideal ker εα Q[X]. Let ω = (−1 + 3i)/2 ∈ C.22 1. 1. Use Cardan’s method to ﬁnd the complex roots of the polynomial f (X) = X 3 − 9X 2 + 21X − 5. Find generators for each of the following ideals: I1 = {f (X) ∈ Q[X] : f (i) = 0} Q[X]. Determine ker εω Q[X] and decide whether it is maximal. What is ε−√2 Q[X] ⊆ C? Find ker ε√2 Q[X] and ker ε−√2 Q[X]. 3 Find rational cubic polynomials f (X) and g(X) for which f (α) = 0 = g(β). .16. 1. 1. When k = F2 . √ I4 = {f (X) ∈ R[X] : f ( 2) = 0} R[X]. β= 3 1+ 2 3 7 + 3 3 1− 2 3 7 .18. Consider the 6 automorphisms αj : k(X) −→ k(X) (j = 1. α4 (f (X)) = f ((X − 1)/X). √ I3 = {f (X) ∈ Q[X] : f ( 2) = 0} Q[X].20.17. Let p > 0 be a prime.60 by showing that there is an isomorphism of groups Aﬀ 1 (k) ∼ Autk (k[X]). Hence determine these real numbers. Prove the ﬁnal part of Example 1. show that Γk ∼ GL2 (k). Consider the inclusion inc : Q −→ C and its extension to εω : Q[X] −→ C.19. . FIELDS AND POLYNOMIAL RINGS 1. α2 (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 . α5 (f (X)) = f (1/(1 − X)). 1.15. . α6 (f (X)) = f (X/(X − 1)). are these maximal ideals? √ 1. 6) deﬁned by α1 (f (X)) = f (X). 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]. (a) Show that for k 1. then Φn (X) = Φp1 ···pk (X p1 r −1 r1 −1 ···pkk ). INTEGRAL DOMAINS. . Let k be any ﬁeld. √ I5 = {f (X) ∈ R[X] : f ( 2 i) = 0} R[X]. I6 = {f (X) ∈ R[X] : f (ζ3 ) = 0} R[X]. = 1. Find another evaluation homomorphism with the same kernel and image. 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. 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. α3 (f (X)) = f (1/X). Consider the inclusion inc : Q −→ C and its extension to ε√2 : Q[X] −→ C.14. 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. Determine the cyclotomic polynomial Φ20 (X). (b) Show that Φpk (X) ∈ Q[X] is irreducible.22. 1. √ I2 = {f (X) ∈ Q[X] : f ( 2 i) = 0} Q[X]. Consider the real numbers α= 3 10 + √ 108 + 3 10 − √ 108. Determine the image ε√2 Q[X] ⊆ C.
(b) If a ∈ K. ζn + ζn = 2 cos(2π/n). Deduce that if m. For n 2. −1 1. n > 0 and p n. show that X ϕ(n) Φn (X −1 ) = Φn (X).24. −1 −2 2 Find expressions for ζ5 +ζ5 and ζ5 +ζ5 in terms of cos(2π/5). 1.23 1. then every npm th root of 1 in K is an nth root of 1. Let p > 0 be a prime and K be a ﬁeld with char K = p. . Show that for n 1. show that the polynomial X p − a ∈ K[X] has either no roots or exactly one root in K.25. Hence ﬁnd a rational polynomial which has cos(2π/5) as a root. (a) Show that if ζ ∈ K is a pth root of 1 then ζ = 1.23.
.
however R and C are uncountable. i. v √ √ which we know √ be false. these elements are also linearly independent / over R and therefore they form a basis.5. √ Show that [Q( 2) : Q] = 2. Since i ∈ R. 25 . Consider the extension Q( 2)/Q where √ √ Q( 2) = {x + y 2 : x. (i) If one or both of the dimensions [L : K] or [M : L] is inﬁnite then so is [M : K].3. in fact it is easy to see that we would then also have u. y ∈ Q}.1. The elements 1.1. v both nonzero. i span C as a vector space over R.4. Theorem. The inﬁniteness of R/Q and C/Q are consequences of the fact that any ﬁnite dimensional vector space over Q is countable. M 2. 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. 2. L is also said to be an extension (ﬁeld ) of K. Then we say that K is a subﬁeld of L. We will call dimK L the degree or index of the extension L/K and use the notation [L : K] = dimK L. otherwise it is inﬁnite (dimensional ). We write K L or L/Kto indicate this and K < L if K is a proper subﬁeld of L. Given two extensions L/K and M/L. if K = L.6. Hence 1. √ 2. Definition.. Solution. whence [C : R] = 2. Let K and L be ﬁelds and suppose that K ⊆ L is a subring. so 1. we say that L/K is a subextension of M/K and sometimes write L/K M/L.e.2. Definition. A basis for the Qvector space R is known as a Hamel basis. while R/Q and C/Q are both inﬁnite. We have C = {x + yi : x. 2. √ √ Now recall that √ Solution. √ √ 2 ∈ Q.CHAPTER 2 Fields and their extensions 2. 2 were linearly dependent we would have u + v 2 = 0 for some / u. 2 clearly span the Qvector space Q( 2). 2. Definition. If the elements 1. Thus we would have √ u 2 = − ∈ Q. Fields and subﬁelds 2. An extension of ﬁelds L/K is ﬁnite (dimensional ) if [L : K] < ∞. Show that the extension C/R is ﬁnite. 2 are linearly independent and so form a basis for Q( 2) to over Q and [Q( 2) : Q] = 2. v ∈ Q not both zero. (ii) If the dimensions [L : K] and [M : L] are both ﬁnite then so is [M : K] and [M : K] = [M : L] [L : K]. Let L/K be a subextension of M/K. 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. Example. x ∈ L). y ∈ R}. Example.
. . Also L is a Kvector subspace of the ﬁnite dimensional Kvector space M . . By the linear independence of the i over K. j=1 i=1 On collecting terms we obtain s j=1 r tij i=1 i mj = 0. j = 1. . . where t1 . . . (i = 1. We can build up sequences of extensions and form towers of arbitrary . r of L over K and a basis m1 . (ii) Setting r = [L : K] and s = [M : L]. . . .26 2. Hence the i mj form a basis of M over K and so [M : K] = rs = [M : L] [L : K]. . Now any element u ∈ M can be expressed as u = t1 m 1 + · · · + tr m r . Then there are elements y1 . . . Now let v ∈ M . Now suppose that for some tij ∈ K we have s r tij ( i mj ) = 0. . . where each coeﬃcient means that for each j. r i=1 tij i is in L. . s) span the where each coeﬃcient xij is in K. ys ∈ L for which v = y1 m1 + · · · + ys ms . Hence. (i) If [M : K] is ﬁnite. this r tij i=1 i = 0. . . choose a basis m1 . . By the linear independence of the mj over L. . . this means that m1 . . Such towers of extensions are our main objects of study. but since K ⊆ L. . r. tr ∈ K. each tij = 0. ms of M over L. Thus the elements Kvector space M . mr of M over K. . choose a basis 1 . s r s r 1 + · · · + xrj r v= j=1 i=1 xij i mj = j=1 i=1 i mj xij ( i mj ). . . mr spans M over L and so [M : L] < ∞. . 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. . . . But each yj can be expressed in the form yj = x1j for suitable xij ∈ K. . hence [L : K] < ∞. . FIELDS AND THEIR EXTENSIONS Proof.
un . . . The extension K(u1 . . v) = K(u)(v) = K(v)(u). K(u)/K is inﬁnite and u is said to be transcendental over K. .. . Given elements u1 . if L1 /K. . . ur . Xr ). SIMPLE AND FINITELY GENERATED EXTENSIONS 27 length. . 2. . . For a simple extension K(u)/K. . L2 /L1 . . .2. . . ur . .. . ur ) = which is the smallest subﬁeld in F that contains K and the elements u1 . un ) = K(u1 . . An extension of the form K(u)/K is called a simple extension of K with generator u.8. . Lk /Lk−1 is a such a sequence of extensions. . ur .9. ur ∈ F we set L K L F u1 . g(X1 . un−1 )(un ) and this is independent of the order of the sequence u1 . we also say that K(u1 . 2. . . Then K(u. . . . . . ur ) = 0 . . . . .2. . . . . . . . . . g(u1 . ur ) ∈ F : f (X1 .2. ur ) = f (u1 .) F the smallest extension ﬁeld of K containing all the elements ur . . It can be shown that (2. Proposition. Theorem. . More generally. . v)/K(u) be simple extensions. . . . g(u1 . . Xr ) ∈ K[X1 . . exactly one of the following conditions holds. . . . . . . ur . Thus. . . . . . . .. We can extend this to the case of an inﬁnite sequence u1 . there is a diagram Lk Lk−1 L1 K 2.. . . ur ) Reordering the ui does not change 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. Simple and ﬁnitely generated extensions 2. Xr ]. . Definition. Let F be a ﬁeld and K F . . . . in F and denote by K(u1 . (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. . ur )/K is said to be generated by the elements u1 . . un ). .7. Let K(u)/K and K(u. In this case. . . . . . (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). . ur )/K is a ﬁnitely generated extension of K. . . . .1) K(u1 . . . . . . . .ur ∈L K(u1 . K(u1 .
So 1. pn ) : Q] = 2n . 3):Q]=2[Q( 2. 3):Q( 2)] √ Q( 2) 2 √ √ √ √ √ [Q( 2. Example. . . 2. √ √ 2. Theorem 1. There are some other subﬁelds of Q( 2. √ √ √ √ 2. . .31(iv) implies that the image of εu is a subﬁeld of K(u) and since it contains u it must equal K(u). . where deg f (X) < deg p(X). √ √ √ √ √ so 1. . b ∈ Q(√2). Using Long Division. 3):Q( 2)] Q √ √ √ We will show that [Q( 2. . . √3) √Q( 2)] = 2. . w ∈ Q. . √ : √ √ √ Notice that if u ∈ Q( 2. 3) over Q( 2). pn−1 )] = 2 and [Q( p1 . 3) √ √ √ [Q( 2. This shows that [Q( 2. √ and hence 2vw 2 ∈√ The possibilities v = 0 or w = 0 are easily ruled out. Via the isomorphism εu under which εu (X k + (p(X))) = uk .10. 3) which are conveniently displayed in the following diagram. u. all that needs checking is that (εu )∗ is an epimorphism. . . . . . 3 are linearly√ independent over Q( 2) √ √ √ √ and√ √ therefore form a basis of Q( 2. .4 we know that [Q( 2) : Q] = 2. √ √ Q( 2. Hence εu is an isomorphism. 3). 3) : Q] = 4. w √ 0 Q. For the extension Q( 2. (i) If ker εu = (0).11. . X + (p(X)). . . Let p1 . / √ √ √ √ √ √ Hence [Q( p1 . . 3) : Q] = 4. but as u is in the image of (εu )∗ this is obvious. . pn be a sequence of distinct primes pi > 0. Remark. Writing √ √ 3=v+w 2 with v. 3) = Q( 2)( 3) then u = a + b 3 for some a. We have the following tower of extensions. By Example 2. where d = deg p(X). √ √ Q( 2. 3 span Q( 2. we ﬁnd that √ v 2 + 2w2 + 2vw 2 = 3 ∈ Q. . √ Proof. . . = √ would implies that 2 ∈ Q which is false. 3)/Q we have [Q( 2.10 can be extended to provide a useful general result whose proof is left as an exercise. But if these are linearly dependent then 3 ∈ Q( 2). we see that the elements 1. . . (ii) When ker εu = (0).12. . X d−1 + (p(X)). Proposition. Then √ √ √ pn ∈ Q( p1 . pn−1 ). 3) √ Q( 2) rr rrr 2 rrrr r 2 ww ww ww ww 2 ww w √ Q( 3) 2 vvv vvv 2 vvv v Q rrr rrr rrr 2 rrr √ Q( 6) One idea in the veriﬁcation of Example 2. 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)). .28 2. 3) : Q( 2)] = 2 and so [Q( 2. FIELDS AND THEIR EXTENSIONS Proof. while v. ud−1 form a basis for K(u) over K. pn ) : Q( p1 . we ﬁnd that every element of K[X]/(p(X)) can be uniquely expressed as a coset of the form f (X) + (p(X)). X 2 + (p(X)). .
i) : Q] = [Q( 2. (ii) If m 1 with m  n.14. we have Q( 2. hence Q(21/m . (i) Consider the evaluation homomorphism ε21/n : Q[X] −→ En . and the induced homomorphism ε21/n : Q[X]/(X n − 2) −→ En is an isomorphism. (i) Show that [En : Q] = n. C 2 R ∞ Q( 2) tt tt tt tt √ t 2 ∞ √ Q( 2. 21/m = (21/n )n/m ∈ En . where 21/n ∈ R denotes the positive real nth root of 2. mn mn n m This shows that 21/mn = (21/n )r (21/m )s ∈ Q(21/m . (iii) If m. s for which rm + sn = 1 and so 1 rm + sn r s = = + . whence [En : Em ] = n/m. i ∈ Q( 2) since i is not real and Q( 2) R. √ √ √ Proof. 21/n ). i) = Q( 2)(i) and [Q( 2. show that Em En and determine [En : Em ]. (iii) By (ii) we have Em Emn and En Emn . Example. so Em = Q(21/m ) ⊆ En .38 using the prime 2 to the polynomial X n − 2 ∈ Z[X]. i) ∩ R being a subﬁeld of R. 21/n ). whence Emn Emn . / √ √ √ √ Since i2 + 1 = 0. let En = Q(21/n ) R. 21/n ). Q(21/m . Solution. For the extension Q( 2. n are coprime. . i) : Q( 2)] = 2. We know that [Q( 2) : Q] = 2. 21/n ) there are integers r. we ﬁnd that ker ε21/n = (X n − 2) Q[X].2. i) : Q( 2)] [Q( 2) : Q]. i) : Q] = 4. n) = 1.2. For n 1. Also. √ we obtain [Q( 2. Hence [En : Q] = n. By Theorem 2. (ii) Since n/m is an integer. 21/n ). show that Emn = Q(21/m . with only Q( 2) = Q( 2. Combining these inclusions we obtain Emn = Q(21/m . √ √ This example also has several other subﬁelds. Example. i)/Q we have [Q( 2. SIMPLE AND FINITELY GENERATED EXTENSIONS 29 √ √ 2. As gcd(m. i) uuu uuu 2 uuu 2 u r rrr rrr2 r rrr uuu uuu u 2 uuu u Q(i) 2 √ Q( 2 i) Q 2.6 we have n = [En : Q] = [En : Em ] [Em : Q] = m[En : Em ]. Applying the Eisenstein Test 1.13. i) : Q] = 4. Using the formula √ √ √ √ [Q( 2.
v ∈ K(u+v).7. Show that the extension Q( p)/Q has [Q( p) : Q] = 2. Show that K(u.2. (b) Let E = En R. (d) Which of the subﬁelds in part (c) contain the element 21/2 + 21/3 ? . In this question we continue to consider the situation described in Example 2. [Hint: ﬁrst show that u±v = 0 and deduce that u−v ∈ K(u+v). Let ζ5 = e2πi/5 ∈ C. {id.3. Show that the numbers cos(2kπ/7) with k = 0. (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.30 2. Prove Proposition 2. sin(2π/7) i ∈ Q(ζ7 ). n 1 (c) Display the 6 subﬁelds of E12 in a diagram. sin(2π/5) i ∈ Q(ζ5 ). FIELDS AND THEIR EXTENSIONS Exercises on Chapter 2 √ √ 2. AutQ (En ) = ∼ Z/2 if n is even. q > 0 be distinct primes. . suppose that u. Let p. . Show that AutQ (E) = {id}. Let K a ﬁeld with char K = 2 and suppose that L/K is an extension. Let p ∈ N be an prime. (a) Explain why [Q(ζ7 ) : Q] = 6. 3. b ∈ K are distinct. (b) Show that cos(2π/5). 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. i) : Q] = 4.6. Show that [Q(i) : Q] = 2. √ √ √ 2. (a) Explain why [Q(ζ5 ) : Q] = 4. 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. (f) Is i ∈ Q(ζ7 )? 2. This question is for those who like lots of calculation or using Maple. (c) Show that for t ∈ R. If a. then show that u. .12 by induction on n. v) = K(u + v).1. q) : Q( p)] = 2. (b) Show that cos(2π/7). 2. cos 5t = 16 cos5 t − 20 cos3 t + 5 cos t.8. τn } = where τn has composition order 2. 2. Find the three subﬁelds L Q( 3. Q(cos(2π/5)).9. (e) Display the relationship between the ﬁelds Q. What is [Q(sin(2π/7) i) : Q]? (e) Display the relationship between the ﬁelds Q. indicating which ones are subﬁelds of R. 2. and Q(ζ5 ) in a suitable diagram. Q(sin(2π/7) i) and Q(ζ7 ) in a diagram. (d) Show that the numbers cos(2kπ/5) with k = 0.4. 1. . Q(cos(2π/7)). Show that [Q( p.] 2. 2. 1. (a) Show that {id} if n is odd. Let ζ7 = e2πi/7 ∈ C. √ √ 2. Show that [Q( 3.14. (c) Show cos 7t = 64 cos7 t − 112 cos5 t + 56 cos3 t − 7 cos t. i) with [L : Q] = 2 and display their relationship in a diagram. v ∈ L satisfy u2 = a and v 2 = b.5. 2.
(i) t is algebraic over K. p(X) = minpolyK. Definition.3. so t is algebraic over K. By Example 2.t (X) = degK t. Remark.2. Then minpolyK. we have minpolyQ.t (X) for some u ∈ K.CHAPTER 3 Algebraic extensions of ﬁelds 3.i (X) = X 2 + 1. (iii) The extension K(t)/K is ﬁnite dimensional. This follows from Theorem 2. (ii) The evaluation homomorphism εt : K[X] −→ L has nontrivial kernel.9 allows us to characterize algebraic elements in other ways.6. 31 . Algebraic extensions Let L/K be an extension of ﬁelds.9(ii).2. hence minpolyQ. In particular. recall the following notion. Example. The minimal polynomial of i ∈ C over Q is X 2 + 1. Let t ∈ L. Notice in particular that for an element t ∈ K. From Theorem 2.√2 (X) = X 2 − 2. Clearly X 2 + 1 ∈ ker εi since i2 + 1 = 0. If t ∈ L is algebraic over K then [K(t) : K] = deg minpolyK.√2 (X) = [Q( 2) : Q] = 2. 3. 3.4. Example. Consider C/Q. As [Q(i) : Q] = 2. 3. 3.1.t (X) is called the degree of t over K and is denoted degK t.t (X)) = (0).4.7. Then the following conditions are equivalent.5.√2 (X) = X 2 − 2.t (X). √ deg minpolyQ. Proposition. √ Proof. ker εt = (minpolyK. 3. Proof. We will often use this without further comment. the polynomial p(X) = X − t ∈ K[X] satisﬁes p(t) = 0. Proposition. Clearly X 2 − 2 ∈ ker ε√2 since ( 2)2 − 2 = 0. 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). 3.t (X)  p(X) and so p(X) = u minpolyK. If t ∈ L is algebraic over K then by Proposition 3.t (X) ∈ K[X] is an irreducible monic polynomial called the minimal polynomial of t over K.9(ii). Theorem 2. when p(X) is monic. 3. Definition. Proof. The degree of minpolyK. where minpolyK. The minimal polynomial of √ 2 ∈ C over Q is minpolyQ. Suppose that t ∈ L is algebraic over K and that p(X) ∈ ker εt with deg p(X) = deg minpolyK.1. Consider C/Q.
Find the minimal polynomial of the primitive 6th root of unity. Let L/K be a ﬁnite extension.10. 2 √ √ √ √ √ √ 1 √ ( 2 + 3) + ( 3 − 2) ∈ Q( 2 + 3). √ minpolyQ(√2). Then Φ6 (X) ∈ ker εζ6 so minpolyQ. ALGEBRAIC EXTENSIONS OF FIELDS 3. Since this is monic and of degree 2.10 or Proposition 2. Consider p(X) = minpolyQ(√2). ζ6 ∈ C over Q. 3. Find the minimal polynomial of 2 + 3 over Q.√2+√3 (X) = X 2 − 2 2X − 1. Example. Solution. 3− 2= =√ ( 3 + 2) 2+ 3 So we have √ √ √ √ √ √ 1 √ 2= ( 2 + 3) − ( 3 − 2) ∈ Q( 2 + 3). An element u ∈ L for which L = K(u) is called a primitive element for the extension L/K. Consider C/Q. √ One polynomial in ker ε√2+√3 Q( 2)[X] is √ √ √ √ √ (X − ( 2 + 3))(X − ( 2 − 3)) = X 2 − 2 2X − 1.12 we see that 2 + 3 ∈ Q( 2). √ √ √ Referring to Example 2.9.ζ6 (X) = Φ6 (X) and so degQ ζ6 = 2. we have minpolyQ. Consider C/Q. √ √ 3. Since Φ6 (X) is irreducible and monic.√2+√3 (X) minpolyQ(√2). Since Q( 2 + 3) Q( 2. 3) Q( 2 + 3). Notice that √ √ √ √ √ √ √ √ ( 3 − 2)( 3 + 2) 1 √ √ √ ∈ Q( 2 + 3). Referring to Example 2.8. Example. hence / √ √ degQ(√2) ( 2 + 3) = 2.10 we see that √ √ degQ ( 2 + 3) = 4. we must have minpolyQ. 3.44 that ζ6 is a root of the irreducible cyclotomic polynomial Φ6 (X) = X 2 − X + 1. Let us ﬁnd a nonzero polynomial in ker ε√2+√3 Q[X]. √ minpolyQ(√2). = X 4 − 10X 2 + 1. 3= 2 √ √ √ √ √ √ √ √ hence Q( 2. Later we will see that when char K = 0 every ﬁnite extension L/K has a primitive element. Recall from Example 1.32 3. Similarly. Solution. Definition. 3).√2+√3 (X) = X 4 − 10X 2 + 1. .11. Since deg p(X) = 4 and p(X) is monic. Lemma.−√2+√3 (X) = X 2 + 2 2X − 1.−√2+√3 (X) √ √ = (X 2 − 2 2X − 1)(X 2 + 2 2X − 1) √ √ Then p( 2 + 3) = 0 so p(X) ∈ ker εt . Let L/K be a ﬁnite extension and u ∈ L. 3) we must have √ √ √ √ Q( 2 + 3) = Q( 2.ζ6 (X)  Φ6 (X). Then u is a primitive element for L/K if and only if degK u = [L : K].
In fact.3. suppose that q(X) = q1 (X)q2 (X) with deg qi (X) < deg q(X). The irreducible monic polynomial minpolyK. u)/K(p0 . . Then L/K is algebraic. . Let M/L and L/K be algebraic extensions. Sometimes the minimal polynomial of an element in an extension is introduced in a diﬀerent but equivalent way. . Proof. pm )/K is ﬁnite and so is K(p0 . The next Lemma will often be useful. q(X) = minpolyK. Proof. Let t ∈ L. . so by Proposition 3. 3. K(p0 . u + v and uv are in K(u. . . tn . Proof.17. For an extension L/K. v) and if u = 0. Use induction on n together with Proposition 2. pm .6(ii). 3. K(u) ⊆ L is a ﬁnite dimensional Kvector subspace.13. . .12. Let u ∈ M . Lemma.t (X). Since degK u = dimK K(u) and [L : K] = dimK L the result follows. .15. Let u. we must have q1 (t) = 0 or q2 (t) = 0. It is easy to see that I(t) K[X] and therefore I(t) = (q(X)) for some monic generator q(X). By Lemma 3.t (X) is in I(t) so q(X)  minpolyK. . pm . Since the Kvector space L is ﬁnite dimensional. We must show that Lalg L. u)/K is ﬁnite. . . . let Lalg = {t ∈ L : t is algebraic over K} ⊆ L. 3. Let L/K be an extension and suppose that u1 . the extension K(p0 . . Proposition.6(ii). . Let L/K be a ﬁnite extension. Proposition. Definition. so there is a polynomial p(X) = p0 + p1 X + · · · + pm X m ∈ L[X] of positive degree with p(u) = 0.15. v)/K is a ﬁnite dimensional extension.13. . t. .8 and Theorem 2. Then K(u1 . But this means that t is algebraic over K. Let t ∈ L be algebraic over K. u is algebraic over K. . . Proof. un )/K is a ﬁnite extension. . For an extension L/K. Proposition.14. Then the extension M/K is algebraic. 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].16. 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). . In particular. . u−1 is also in K(u. the powers 1. 3. We now come to an important notion for extensions. Clearly K ⊆ Lalg . K(u. hence q1 (X) ∈ I(t) or q2 (X) ∈ I(t). un ∈ L are algebraic. . ALGEBRAIC EXTENSIONS 33 Proof. The extension L/K is algebraic or L is algebraic over K if every element t ∈ L is algebraic over K. . Definition. must be linearly dependent over K. To see that q(X) is irreducible. when viewed as elements of this vector space.t (X) and therefore q(X) = minpolyK. 3. v ∈ Lalg .1. Proof. Then u is algebraic over L. Hence for suitable coeﬃcients cj ∈ K not all zero and some m 1 we have c0 + c1 t + · · · + cm tm = 0. Now as q1 (t)q2 (t) = 0. pm ).13. . Proposition. . Lalg is a subﬁeld containing K and Lalg /K is algebraic. .18. Therefore u + v. By Theorem 2. . hence every element of K(u. . . . . Then K(u) = L if and only dimK K(u) = dimK L. Then by Lemma 3. v). contradicting the above assumption that deg qi (X) < deg q(X). uv and u−1 are all algebraic over K.t (X). 3. v) is algebraic over K. 3.
. 3.u (X) has only ﬁnitely many monic divisors in K(u)[X]. whereas C and R are uncountable. which divides minpolyK. . each of which has the form K(a0 . −1 X −1 + X ∈ K(u)[X] . . . so there are only a ﬁnite number of possibilities for minpolyF. p(X) ∈ K[X] splits in E/K or over E if it factorizes into linear factors in E[X]. the extension ﬁeld of K generated by the coeﬃcients of minpolyF. Theorem (Kronecker’s Theorem: ﬁrst version).u (X) = deg minpolyF0 . . Definition. Similarly.u (X) in K(u)[X]. . We end this section with a technical result. ALGEBRAIC EXTENSIONS OF FIELDS 3. . . in the extension R/Q the subﬁeld Ralg = Calg ∩ R C consists of all the real algebraic numbers. .u (X). When dealing with algebraic extensions of Q we will usually work with subﬁelds of Q = Calg . if we have such a ﬁeld E then the distinct roots u1 . Now for any subextension F/K K(u)/K we can also consider minpolyF. Proof. Question. Let K(u)/K be a ﬁnite simple extension. Consider the minimal polynomial minpolyK. Splitting ﬁelds and Kronecker’s Theorem We can now answer a basic question.25.u (X) = [K(u) : F0 ]. 3. Let K be a ﬁeld and p(X) ∈ K[X] be a polynomial of positive degree.21.21.u (X). Now consider F0 = K(c0 . uk ) E which is the smallest subﬁeld of E that answers Question 3. 3.u (X) = c0 + c1 X + · · · + ck−1 X k−1 + X k ∈ F [X]. c1 . Let K be a ﬁeld and p(X) ∈ K[X] be a polynomial of positive degree. The sets Calg and Ralg are both countable.24. In the extension C/Q we can consider Calg C which is called the subﬁeld of algebraic numbers.20. Definition.19. .22. Then there is a ﬁnite extension L/K for which p(X) has a root in L.u (X) = minpolyF0 .u (X) ∈ K[X]. a −1 ). . This shows that there are only ﬁnitely many subextensions F/K K(u)/K. We have [K(u) : F ] = deg minpolyF.u (X) ∈ F0 [X] is irreducible since it is irreducible in F [X]. Then there are only ﬁnitely many subextensions F/K K(u)/K. . We already know how to answer Question 3. uk of p(X) in E generate a subﬁeld K(u1 . examples are e and π. The Unique Factorization Property 1.22. hence F = F0 . where a0 + a1 X + · · · + a is a factor of minpolyK. 3. Of course. 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. .23. Example. Is there an extension ﬁeld E/K for which p(X) factorizes into linear factors in E[X]? 3. . A more usual notation for Calg is Q since this is the algebraic closure of Q which will be discussed later. Then F0 F and so minpolyF. 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 .33 implies that minpolyK. Question. so there are in fact many more transcendental numbers but it can be hard to determine whether a given number is transcendental or not. 3. ck−1 ). Proposition. . . a1 .u (X).u (X) in F [X].34 3. hence minpolyF.2. Elements of C−Calg are called transcendental complex numbers. 3.
3. When working over Q (or any other subﬁeld of C) we can always ﬁnd roots in C by the Fundamental Theorem of Algebra. √ √ Next consider the polynomial X 2 + 2 ∈ Q( 2)[X]. 3. Theorem (Kronecker’s Theorem: second version). 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). Hence p(X) has a root in K[x]/(qj (X)). i) and the extension Q( 2. 3. i)/Q( 2) which has degree 2. Let K be a ﬁeld and p(X) ∈ K[X] be a polynomial of positive degree. 2i) = Q( 2. this construction is only interesting if qj (X) to has degree bigger than 1 since a linear polynomial already has a root in K. We then refer to a subﬁeld of C which is a splitting ﬁeld as the splitting subﬁeld. 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. Of course. Namely. In practise we often have extension ﬁelds ‘lying around in nature’ containing roots and we can work inside of these. The √ complex roots of X 2 + 2 are ± 2i and these are not √ real. 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]. Find a splitting ﬁeld E/Q for p(X) = X 4 − 4 over Q and determine [E : Q]. Hence we need to consider in √ √ √ Q( 2. 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)).2. .26. Solution. i) and [Q( 2. At each stage we either already have another root in L1 or we need to enlarge the ﬁeld to obtain one. − 2) = Q( 2) which gives an √ extension Q( 2)/Q of degree 2. C ∞ √ Q( 2. so this polynomial is irreducible √ Q( 2)[X]. √ √ √ √ so ﬁrst we adjoin the roots ± 2 of (X 2 − 2) to form Q( 2. Example. SPLITTING FIELDS AND KRONECKER’S THEOREM 35 Proof.27. 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. To answer Question 3. i) : Q] = 4. Then there is a ﬁnite extension E/K which is a splitting ﬁeld of p(X) over K. Notice that p(X) = (X 2 − 2)(X 2 + 2).22 we iterate this construction.
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 . . . . . Proposition.e. 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. .38. it is unique as a subﬁeld of F . 3 2 ζ3 lying in the extension √ √ √ Q( 3 2. 3.1 shows all the subﬁelds of the extension Q( 3 2. uk ) must be contained in any splitting subﬁeld. If E1 .28. 2a 2a if ∆ = δ 2 for some nonzero δ ∈ K. • two distinct roots −b − δ −b + δ = (2a)−1 (−b + δ). F are Proof. ζ3 ) with [Q( 3 2. .4. . By deﬁnition. Then the quadratic polynomial p(X) = aX 2 + bX + c ∈ K[X] has • no roots in K if ∆ is not a square in K. K(u1 . So the splitting subﬁeld of X 3 − 2 in C over Q is Q( 3 2. . 3. ζ3 )/Q(ζ3 ) of degree 3 is obtained by adjoining one and hence all of the roots. Let u1 . Example. uk ) = E2 . . . 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. But K(u1 . √ Figure 3. so E1 = K(u1 . Let K be a ﬁeld of characteristic diﬀerent from 2. . Proposition. = (2a)−1 (−b − δ). ζ3 )/Q. Since we will frequently encounter quadratic polynomials we record a useful result on roots of such polynomials. . ζ3√ )/Q( 3 2) of degree 2. 3. the splitting ﬁeld of p(X) over K is K if ∆ is a square in K and K(δ) otherwise. .29. uk ∈ F be the distinct roots of p(X) in F . • one root −b/(2a) = −(2a)−1 b if ∆ = 0. 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. thus giving the tower C ∞ √ Q( 2. ζ3 ) : Q] = 6. The proof of the next result is the standard one which works provided 2 has an inverse in K. i. ALGEBRAIC EXTENSIONS OF FIELDS Of course we could have started by ﬁrst adjoining roots of X 2 + 2 and then adjoining roots of X 2 − 2.36 3. √ Solution. . We could also have begun by adjoining ζ3 to form the extension Q(ζ3 )/Q. In particular.. . Recall that p(X) = aX 2 + bX + c ∈ K[X] is quadratic if a = 0 and its discriminant is ∆ = b2 − 4ac ∈ K. Let F/K be an extension ﬁeld and p(X) ∈ K[X].30. Find a splitting ﬁeld E/Q for p(X) = X 3 − 2 over Q and determine [E : Q]. when char K = 2. uk ) is the smallest subﬁeld containing K and all the uj . but none of √ the roots of p(X) lie in this ﬁeld so the extension Q( 3 2. E2 splitting subﬁelds for p(X) over K then E1 = E2 . p(X) is irreducible over Q. . By the Eisenstein Test 1. √ √ 2 An alternative strategy would have been to adjoin one of the other roots 3 2 ζ3 or 3 2 ζ3 ﬁrst. .
hence every such homomorphism arises this way. If F/K is ﬁnite. F ) is closed under composition but is not always a group since elements are not necessarily invertible.34. Set Roots(p.3. then we do have MonoK (F. Then if t ∈ F is a root of p(X). F ) ←→ MonoK (K(t0 ).33. 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 each root of p(X) in F gives rise to a monomorphism ϕt = εt ◦ ε−1 : K(t0 ) −→ F for which ϕt (t0 ) = t. Definition. Let F/K be a ﬁeld extension. F ) denote the set of all monomorphisms L −→ F which ﬁx the elements of K. and let F/K be an extension. This is always a ﬁnite set which may of course be empty (this happens precisely when p(X) has no root in K). the set of roots of p(X) in F . Monomorphisms between extensions 3. 3. F ) = {u ∈ F : p(u) = 0}. F ) and MonoK (F. let MonoK (L. F ) given by t ←→ ϕt . where ϕt : K(t0 ) −→ F has the eﬀect ϕt (t0 ) = t. The subﬁelds of Q( 3 2. Then there is a bijection Roots(p. Of course. . the evaluation homomorphism εt : K[X] −→ F factors through the quotient monomorphism εt : K[X]/(p(X)) −→ F whose image is K(t) F . This discussion is summarized in the following result. We always have AutK (F ) ⊆ MonoK (F. If we ﬁx one such root t0 and identify K[X]/(p(X)) with K(t0 ) via εt0 .1. Let F/K be an extension and p(X) ∈ K[X]. Proposition. ζ3 )/Q 3. Let p(X) ∈ K[X] be an irreducible polynomial with t0 ∈ F be a root of p(X). For extensions F/K and L/K. 3. 3.31.3. MONOMORPHISMS BETWEEN EXTENSIONS 37 C 2 R ∞ ∞ √ √ iiii Q( 3 2 ζ3 ) Q( 3 2) iii iiii 2 iiiii 2 qqqq i iiii q qq qqq √ Q( 3 2. Suppose that p(X) ∈ K[X] is an irreducible polynomial which we might as well assume is monic.3. F ) = AutK (F ) since every injective Klinear transformation is surjective and so invertible. Definition. We will also use the following notation. there is one such monomorphism for each root of p(X) in F .32. ζ3 ) www `` `` www `` www `` www 3 `3 www 3 Q(ζ3 ) www ``` r www `` 2 rrr www `` rr ` www ` rrr w rr VV V 2 VV VV √ 2 VVV3 3 Q( 2 ζ3 ) VV VV VV VV Q √ Figure 3. then the coset X + (p(X)) must be sent by ϕ to a root of p(X) in F . Remark.
√ We can replace C by Q( 2) to obtain √ √ √ √ MonoQ (Q( 2). Q( 3 2 ζ3 ) or Q( 3 2 ζ3 ) √ of C and there are 3 monomorphisms α0 . L) −→ Roots(p. L) into Roots(p. In the fact both possibilities occur. hence it is also surjective by the Pigeon Hole Principle. F ) restricts to a function αp : Roots(p. extensions it is the key to understanding the automorphism group and this is a fundamental insight of Galois Theory. As two of these roots are not real. √ √ Solution. We will see that this is not always true. 3 2 ζ3 . L) ⊆ L is also an injection. each monomorphism α ∈ MonoK (L. We build up the list of monomorphisms √ stages. (ii) From (i). ζ3 ). Determine MonoQ (Q( 3 2. C). Example. This gives us a strong hold on the possible automorphisms. These form the subset √ √ 3 3 MonoQ( √2) (Q( 2.38. We have already met the extension Q( 3 2. L) is a bijection. Q( 3 2)) contains only the identity function. in √ 3 First consider monomorphisms that ﬁx 2 and hence ﬁx the subﬁeld Q( 3 2).30 and we will make use of information from there. √ Solution. then αp : Roots(p. ζ3 ). Q( 3 2)) contains only the identity √ since Q( 3 2) R.35. (i) For p(X) ∈ K[X].37. √ Solution. 3. (ii) If α ∈ MonoK (L. C) has 3 elements but MonoQ (Q( 3 2). so α maps Roots(p. Q( 2)) = AutQ (Q( 2)). L) is a bijection. C). 3 . or more generally algebraic. F ). L) −→ Roots(p. Example. giving monomorphisms id. Thus αp : Roots(p. C) ⊆ MonoQ (Q( 2. α2 : Q( 3 2) −→ C given by √ √ √ √ 3 3 3 3 α0 (a + b 2 + c( 2)2 ) = a + b 2 + c( 2)2 . We have Q( 2) ∼ Q[X]/(X 2 − 2) where X 2 − 2 is irreducible over Q. ζ3 ). √ √ √ √ 3 3 3 3 2 α1 (a + b 2 + c( 2)2 ) = a + b 2 ζ3 + c( 2)2 ζ3 . L) −→ Roots(p. L) we have p(α(u)) = α(p(u)) = α(0) = 0. Show that MonoQ (Q( 2). Since α is an injection its restriction to Roots(p. Show that MonoQ (Q( 3 2). Part (ii) says that any automorphism of L/K permutes the set of roots in L of a polynomial p(X) ∈ K[X]. ALGEBRAIC EXTENSIONS OF FIELDS √ 3. C) = MonoQ (Q( 2). α : Q( 2) −→ C. αp : Roots(p. In the case of ﬁnite. L). Proof. (i) For u ∈ Roots(p. √ √ √ 3. 3 √ 2 √ √ 3 2 ζ3 .38 3. L) is an injective function from a ﬁnite set to itself. √ √ 2 √ 2 √ 2 3 3 3 3 α2 (a + b 2 + c( 2) ) = a + b 2 ζ3 + c( 2) ζ3 . where √ √ α(a + b 2) = a − b 2. Let F/K and L/K be extensions. These mappings have images √ √ 3 3 α0 Q( 2) = Q( 2). Example. L) −→ Roots(p. F ) which is an injection. Proposition. √ √ √ 2 Each of the above roots corresponds to one of the subﬁelds Q( 3 2). ζ3 )/Q in Example 3. √ √ 2 3 3 α2 Q( 2) = Q( 2 ζ3 ). MonoQ (Q( 3 2). Here minpolyQ. Hence = √ √ the Qmonomorphisms we want send 2 to ± 2 which are √ complex roots of X 2 − 2. √ √ 3 3 α1 Q( 2) = Q( 2 ζ3 ). C) has two elements. α1 . √2 (X) = X 3 − 2 and there are 3 complex roots 3 2. √ 3.36.
2 ζ3 −→ ζ3 √ √ 3 Next we consider monomorphisms that send √ 2 to 3 2 ζ3 . say p(u) = 0 for some p(X) ∈ K[X] of positive degree. . Question.42. It is natural to pose the following question. ζ3 )) ∼ S3 . 3. This shows that v ∈ im α and so αalg is surjective.4. Then αq : Roots(q. and it is also ﬁnite. Let L/K be an extension and α ∈ MonoK (L. ζ3 )) = AutQ (Q( 2. ζ3 )) = [Q( 3 2. 3. ζ3 ). Q( 2. ζ3 ) = Q( 3 2)(ζ3√ and that ζ3 is a root of the irreducible cyclotomic ) polynomial Φ3 (X) = X 2 + X + 1 ∈ Q( 3 2)[X]. where α0 has the eﬀect √ √ 3 2 −→ 3 2 α0 : .37(ii). . Algebraic closures An important property of the complex numbers is that C is algebraically closed. Let K be a ﬁeld. It is also worth remarking that  AutQ (Q( 2. ζ3 ). 3. ud ∈ C and this is unique apart from the order of the roots uj . This time we have 2 distinct ways to √ extend to elements of MonoQ (Q( 3 2. It is also the case here that √ √ √ √ 3 3 3 3 MonoQ (Q( 2. Definition. 2 ζ3 −→ ζ3 ζ3 −→ ζ3 These are all 6 of the required monomorphisms. L) is a bijection by Proposition 3. Let K be a ﬁeld. We must show that αalg is a bijection by showing it is surjective. So there are two monomorphisms id. 3. We end this section with another useful result. ζ3 ).4. α0 ﬁxing √ Q( 3 2). u1 . the symmetric = √ √ 3 group on 3 objects. It is a nice exercise to show that AutQ (Q( 3 2. ζ3 ) : Q]. Then α restricts to an automorphism αalg : Lalg −→ Lalg . . 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. . C) = MonoQ (Q( 2. Corollary.39. The possibilities are √ √ √ √ 3 3 2 −→ 3 2 ζ3 2 −→ 3 2 ζ3 . Every nonconstant polynomial p(X) ∈ C[X] has a factorization p(X) = c(X − u1 ) · · · (X − ud ). Now Roots(q. Proof. α1 : . (i) There is an algebraic closure of K.44. 3. L) = ∅ since it contains v.43. Q( 3 2. α1 : 2 ζ3 −→ ζ3 ζ3 −→ ζ3 √ √ 2 Finally we consider monomorphisms that send 3 2 to 3 2 ζ3 . ζ3 )) since again we can send ζ3 to either ζ3 or 2 ζ3 . ζ3 )). L) −→ Roots(q. . Every nonconstant polynomial p(X) ∈ C[X] has a root in C. L). 3. 3. hence v = αq (w) = α(w) for some w ∈ Roots(q. Then p(α(u)) = α(p(u)) = α(0) = 0. Theorem (Fundamental Theorem of Algebra for C). There are again two possibilities √ √ 2 √ √ 2 3 3 2 −→ 3 2 ζ3 2 −→ 3 2 ζ3 α2 : . alg ⊆ L into itself and therefore gives rise to a restriction αalg : Lalg −→ Lalg which so α maps L is also a monomorphism. L) ⊆ Lalg .3. ALGEBRAIC CLOSURES 39 √ √ We know that Q( 3 2. Theorem.41. α2 : . Let v ∈ Lalg and suppose that q(v) = 0 for some q(X) ∈ K[X] of positive degree. An extension F/K is called an algebraic closure of K if F is algebraic over K and algebraically closed. where c. Proposition. √ so these form a group.40. Suppose that u ∈ Lalg . Let K be a ﬁeld.
then L = K. Suppose that ϕ0 : L −→ K is a monomorphism ﬁxing the elements of K. then so is L/K and (L)alg = K. See [3] for a proof using Zorn’s Lemma (see Axiom 3.44. y or y x is true. A partially ordered set (X. There is a stronger result than Theorem 3.47. y. . Axiom (Zorn’s Lemma). Let (X. ) be a partially ordered set in which every totally ordered subset has an upper bound. E1 = E2 to indicate that for extensions E1 /K and E2 /K there is an isomorphism E1 −→ E2 ﬁxing the elements of K. the Monomorphism Extension Theorem. .40 3. ALGEBRAIC EXTENSIONS OF FIELDS (ii) Let F1 and F2 be algebraic closures of K. referring to it as the algebraic closure of K. Theorem (Monomorphism Extension Theorem). Proposition. which we will ﬁnd useful. • x x. • y ∈ X is an upper bound for Y if for every y ∈ Y . Again the proof uses Zorn’s Lemma which we state below. There are some immediate consequences of Theorem 3. Hence these roots lie in (L)alg . K } eee ee }} ee }} e }} ~} ϕ / F2 F1 Hence algebraic closures are essentially unique. Let K be a ﬁeld.46. . all of its roots are in fact algebraic over K since (L)alg is algebraic over K. x then x = y. (i) By Proposition 3. Let M/K be an algebraic extension and L/K M/K. z ∈ X. Definition. 3. at least one of x 3. which shows that it is algebraically closed. y • An element x ∈ X is a maximal element of X if x y =⇒ y = x. Definition. Since L is algebraically closed it is an algebraic closure of K.48) which is logically equivalent to the Axiom of Choice. For example. First we need some deﬁnitions.49. (X. indeed. • if x y and y • if x y and y z then x z. ) be a partially ordered set and Y ⊆ X. Then there is an isomorphism ϕ : F1 −→ F2 which ﬁxes the elements of K. Because of the uniqueness we usually ﬁx some choice of algebraic closure of K and write K or K alg cl . Let (X. y. ) is totally ordered if for every pair x.48. ) consists of a set X and a binary relation such that whenever x. every element of L is algebraic over K. y ∈ X.44(ii). by Proposition 3. Then X has a maximal element. we have Q = Calg and R = C. Proof.16. We are already familiar with the example C = C. (i) If L/K is an algebraic extension.45. (ii) If L/K is an extension. We will temporarily write .16. 3. 3. Proof. (ii) Every nonconstant polynomial in (L)alg [X] has a root in L. 3.
50. θ) ∈ Y . (M0 . We consider the set X consisting of all pairs (F. θ) is an upper bound for Y . θ0 ). We order X using the relation for which (F1 . Suppose that Y ⊆ X is a totally ordered subset. It is straightforward to check that if u ∈ F for (F . Then for any other root of p(X). Suppose that v = ϕ(u) for some ϕ ∈ MonoK (K. But then (M0 . v ∈ K say.u (X) ∈ K[X]. Let u ∈ K and suppose that p(X) = minpolyK. θ1 ) (F2 . Since M is algebraic / over K it is also algebraic over M0 .u (X) = a0 + · · · + an−1 X n−1 + X n . so θ is welldeﬁned. By Zorn’s Lemma there must be a maximal element of X.51.u (v) = 0. where F/L M/L and θ : F −→ K extends ϕ0 . v ∈ K. Also there is a function θ : F −→ K deﬁned by θ(u) = θ(u) whenever u ∈ F for (F. ) is a partially ordered set. so there is an element u ∈ M for which u ∈ M0 . 3. then v is conjugate to u over K if and only if minpolyK. If u. θ0 ) (M0 . θ ) ∈ Y then θ (u) = θ(u). θ0 ).u (X)) to (M0 (u). 8@ K ÑÑ Ñ ÑÑ ÑÑ ÑÑ M ÑÑ ϕ0 ÑÑ ÑÑ ÑÑ ϕ L K = /K Proof. This extends to a monomorphism ϕ : K −→ K. contradicting the maximality of (M0 . so (F . Lemma. θ2 ) whenever F1 F2 and θ2 extends θ1 . Example.4. θ) ∈ Y we have (F. The Homomorphism Extension Property 1. This factors through the quotient ring M0 [X]/(minpolyM0 . θ) (F . 3. ALGEBRAIC CLOSURES 41 Then there is an extension of ϕ0 to a monomorphism ϕ : M −→ 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 for which v = ϕ(u). Proof. Then for every (F. 3. 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). hence u is algebraic over M0 . θ0 ).θ)∈Y F.52. Let F = (F. Definition. θ0 ) and give a monomorphism θ0 : M0 (u) −→ K extending θ0 .3. Hence M0 = M and so we can take ϕ = θ0 . Then F /L M/L. θ0 ) = (M0 (u).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. Suppose that M0 = M . If minpolyK. there is a monomorphism ϕv : K(u) −→ K with ϕv (u) = v. v ∈ K.u (X) = a0 + a1 X + · · · + ad−1 X d−1 + X d . θ). then a0 + a1 u + · · · + ad−1 ud−1 + ud = 0 . K). Let u. Then (X. If minpolyM0 . θ).
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 ). ALGEBRAIC EXTENSIONS OF FIELDS and so a0 + a1 v + · · · + ad−1 v d−1 + v d = ϕ(a0 + a1 u + · · · + ad−1 ud−1 + ud ) = 0. i. (iv) If char K = p > 0. im ∂ is spanned by the monomials X k for which ∂(X k+1 ) = 0. for f (X). Proof. So ∂(f (X)) = 0 if and only if f (X) = a0 ∈ K. If f1 (u) = 0 then u is a simple root of f (X). 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). Proposition.e. The formal derivative ∂ : K[X] −→ K[X] has the following properties.12) and which vanishes at u.54. where f (X) = a0 + a1 X + a2 X 2 + · · · + ad X d with aj ∈ K.42 3. If f1 (u) = 0 then u is a multiple or repeated root of f (X). g(X) ∈ K[X]. f (u) = 0. the function ∂ : K[X] −→ K[X] given by ∂(f (X)) = f (X) = a1 + 2a2 X + · · · + dad X d−1 ..53. then ker ∂ = {h(X p ) : h(X) ∈ K[X]} and im ∂ is spanned by the monomials X k with p (k + 1).55. so f (u) = f1 (u).5. Proof. ∂(X m ) = mX m−1 and this is zero if and only if p  m. then ker ∂ = K and ∂ is surjective. Multiplicity of roots and separability Let K be a ﬁeld. Working in L[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. (iv) For a monomial X m .. (iii) If char K = 0. The converse follows from Example 3. i. (iii) If f (X) = a0 + a1 X + a2 X 2 + · · · + ad X d then ∂(f (X)) = 0 ⇐⇒ a1 = 2a2 = · · · = dad = 0. . 3. 3. s 0. Using this we see that ∂(a0 + a1 X + a2 X 2 + · · · + ad X d ) = 0 ⇐⇒ am = 0 whenever p m. let f (X) = (X − u)f1 (X). Definition. Also.e. (i) ∂ is Klinear. Then f (X) = f1 (X) + (X − u)f1 (X). Let f (X) ∈ K[X] have a root u ∈ L for some extension L/K. 3. We now apply the formal derivative to detect multiple roots. which are the ones with p (k + 1).e. Proposition. 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. ∂(f (X)g(X)) = ∂(f (X))g(X) + f (X)∂(g(X)). Then we can factor f (X) as f (X) = (X − u)f1 (X) for some f1 (X) ∈ K[X]. Suppose that f (X) ∈ K[X] and u ∈ K is a root of f (X).50. i. We need to understand more clearly when an irreducible polynomial has a multiple root since this turns out to be important in what follows.. (ii) ∂ is a derivation. (i) This is routine. 3. Consider the formal derivative on K[X]. it suﬃces to verify this for the case where f (X) = X r and g(X) = X s with r. (ii) By Klinearity.
and therefore the multiplicity of v must be at least m. Using Long Division and the Euclidean Algorithm we ﬁnd that gcd(f (X). Hence 2i is a multiple root of f (X).56. Let K be a ﬁeld and let K be an algebraic closure. We have f (X) = ∂(X p − 1) = pX p−1 = 0. . this is equivalent to requiring that p (X) = 0. Let u ∈ K be a root of p(X) and suppose that it has multiplicity m. We have f (X) = 4X 3 +16X. 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]. . Example. Show that 2i is a multiple root of f (X) = X 4 + 8X 2 + 16. Corollary.63. uk ∈ K. Show that each of the roots of f (X) = X p − 1 in L is multiple. m where m is the multiplicity of u in minpolyK. Let K be a ﬁeld and let K be an algebraic closure. Proposition. 3. Proof.5. . 3. (X − v)m must divide p1 (X) in K[X]. We have f (X) = ∂(X n − 1) = nX n−1 .3. . so if ζ is any root of f (X) then f (ζ) = 0. Now by Corollary 1.64. f (ζ) = nζ n−1 = 0. Then the number of distinct conjugates of u is deg minpolyK.60.57. then the multiplicities of the uj are equal. Corollary. If f (X) is irreducible in K[X] then a root u is a multiple root if and only if f (X) = 0. 3. where c ∈ K and k = deg p(X). If the irreducible polynomial p(X) ∈ K[X] has distinct roots u1 . X 4 + 8X 2 + 16 = (X 2 + 4)2 . An irreducible polynomial p(X) ∈ K[X] is separable over K if every root of p(X) in an extension L/K is simple.61. uk ∈ K which are all simple then in K[X]. Later we will see that 1 is the only root of X p − 1.62. p(X) = c(X − u1 ) · · · (X − uk ).56. the coeﬃcients of p(X) are ﬁxed by ϕv . For n 1.34. Corollary. In particular. show that each of the roots of f (X) = X n − 1 in C is simple. If char K = 0 and f (X) is irreducible in K[X].59. . Example. 3. Solution. When p(X) is viewed as an element of K(u)[X]. Definition. Now let v ∈ K be any other root of p(X). .u (X) . If u ∈ L is a multiple root of p(X). p(X) = c(X − u1 )m · · · (X − uk )m . 3. and so (X − v)m p1 (X) = (X − u)m p1 (X). If the irreducible polynomial p(X) ∈ K[X] has distinct roots u1 . then every root of f (X) is simple. Hence in K[X]. where 2i is also a root of X 2 + 4. By Proposition 3. Corollary. 3. Let K be a ﬁeld and let u ∈ K. Interchanging the rˆles of u and v we ﬁnd that the multiplicities of u and v are in fact equal. Solution. 3. Example.u (X). so for any root ζ of f (X). so this is obvious. By Corollary 3. where c ∈ K and m 1. there is a monomorphism ϕv : K(u) −→ K for which ϕv (u) = v. f (X)) = X 2 + 4. MULTIPLICITY OF ROOTS AND SEPARABILITY 43 3. . Let p > 0 be a prime and suppose that L/Fp is an extension. o 3. Solution. this can only happen if char K > 0.58. so we can write p(X) = (X − u)m p1 (X) where p1 (X) ∈ K(u)[X] and p1 (u) = 0. where p1 (X) ∈ K[X] is obtained applying ϕv to the coeﬃcients of p1 (X). .34. In fact. . Then ϕv ((X − u)m p1 (X)) = (X − u)m p1 (X).
. · · · . Proposition. Proof. so the desired result follows. . then β −1 ◦ β (u) = β −1 (β(u)) = u. So we can use the following to compute (L : K) = (K(u1 . so (M : L) =  MonoL (M.6(ii) which gives [L : K] = [K(u1 ) : K] [K(u1 . If K(u)/K is separable. K) agrees with the number of extensions of β to a monomorphism M −→ K. each γ ∈ MonoL (M. . K) let αL ∈ MonoK (L. K) be its restriction to L. Lemma. Proposition.70. Therefore we have the desired formula (M : K) = (M : L)(L : K). Let L/K and M/L be ﬁnite extensions. 3. this shows that there is a bijection extensions of β to monomorphism a M −→ K ←→ MonoL (M. . . K). by Proposition 3. K) has the form β = αL for some α ∈ MonoK (M. K). . .65.1) K(u1 )/K. K(u1 . K). . uk−1 ).66. K) extends to a monomorphism M −→ K. K). Of course. . K). 3. . So given β ∈ MonoK (L.49. restricting to M K we obtain a monomorphism M −→ K. . ALGEBRAIC EXTENSIONS OF FIELDS 3. uk−1 )) divides [K(u1 . Since (L : K) =  MonoK (L. . each element of MonoK (L. . Let L/K be a ﬁnite extension. 3. 3. . If L/K is a simple extension then by Propositions 3. hence β −1 ◦ β ∈ MonoL (M. β is an automorphism. An algebraic extension L/K of a ﬁeld of characteristic 0 is separable by Corollary 3.39. . Definition. uk ) : K(u1 . For α ∈ MonoK (M. notice that if u ∈ L. u2 )/K(u1 ). . . This follows from Proposition 3. . Any ﬁnite extension L/K can be built up from a succession of simple extensions (3.67. u2 ) : K(u1 )] · · · [K(u1 . The general result follows by building up L/K as a sequence of simple extensions as in (3. Example. . . 3. .34 we know that this is true. then [K(u) : K] = (K(u) : K).44 3.69. . An algebraic element u ∈ L in an extension L/K is separable if its minimal polynomial minpolyK. K). . .34 applied to the case L = K. By the Monomorphism Extension Theorem 3. In eﬀect. Corollary. Let L/K be a ﬁnite extension. choose any extension to a monomorphism β : K −→ K. . .71. Then L/K is separable if and only if (L : K) = [L : K]. . . Now for any extension β : M −→ K of β we can form the composition β −1 ◦ β : M −→ K. . . . uk−1 )]. (K(u1 . Proof. uk ) : K). Then (L : K)  [L : K]. . The separable degree of L over K is (L : K) =  MonoK (L. uk )/K(u1 . Definition. Conversely. For each k.u (X) ∈ K[X] is separable. Proof. L = K(u1 .68.72. (K(u) : K) =  Roots(minpolyK. so every element β ∈ MonoK (L. K).1) and then using Theorem 2. uk−1 )]. we need to show that the number of such α is always (M : L) =  MonoL (M.62 and 3. uk ) : K(u1 . . Then (M : K) = (M : L)(L : K).u . K) gives rise to a monomorphism β ◦ γ : M −→ K which extends β. . For a ﬁnite simple extension K(u)/K. 3. uk ) : K(u1 . . . An algebraic extension L/K is called separable if every element of L is separable over K.57. . Let L/K be a ﬁnite extension. K). Definition. 3.
u (X)  minpolyK. ur .u (X).6. v).3. and in the polynomial ring E[X] we have minpolyE. Then M/K is separable if and only if L/K and M/L are separable.u (X) is separable. .1). then for any u ∈ L.75. Now since the polynomials h(X). . 3. Conversely. Since L is built up from a sequence of simple extensions it suﬃces to consider the case L = K(u. Definition. Then h(v) = p(u) = 0. . by separability their greatest common divisor in K(w)[X] is a linear polynomial which . Thus (K(u) : K) = [K(u) : K] and so u is separable. 3. As minpolyK. . Therefore M/K is separable. Hence (L : K(u))(K(u) : K) = rs(L : K(u))(K(u) : K). Then L has a primitive element.6. . r = deg p(X) and s = deg q(X). namely v. For a ﬁnite simple extension L/K. By Proposition 3. Let L/K and M/L be ﬁnite extensions. If K E L.70. The case where K is a ﬁnite ﬁeld will be dealt with in Proposition 5. Let p(X). Deﬁne the polynomial (of degree r) h(X) = p(w − tX) ∈ K(w)[X] ⊆ L[X]. while the distinct roots of q(X) are v = v1 . q(X) ∈ K(w)[X] have exactly one common root in K. Expressing L/K in terms of a sequence of simple extensions as in (3. Since K is inﬁnite. Proof. Suppose that the distinct roots of p(X) in K are u = u1 . we can choose an element t ∈ K for which u − ui t= vj − v whenever j = 1. Let L/K be a ﬁnite separable extension. Proof. So we will assume that K is inﬁnite. By Corollary 3. . but h(vj ) = p(ui ) = 0 for any j = 1 by construction of t. hence [M : K] = [M : L][L : K] = (M : L)(L : K) = (M : K). 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]. Then taking w = u + tv ∈ L. if L/K and M/L are separable then [M : L] = (M : L) and [L : K] = (L : K).72 this implies that L/K and M/L are separable.74. . uk−1 )]. . [M : L][L : K] = (M : L)(L : K). so is minpolyE. . [L : K] = [K(u1 ) : K] · · · [L : K(u1 . . . we ﬁnd that w = ui + tvj whenever j = 1. u is separable.71. Suppose that L/K is separable. This can only happen if [M : L] = (M : L) and [L : K] = (L : K). u is algebraic over E. an element u ∈ L is called a primitive element for the extension if L = K(u). .16. Now we can apply Lemma 3. If M/K is separable then [M : K] = (M : K) and so 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]. which can only happen if r = s = 1. The Primitive Element Theorem 3. We must show that for each u ∈ L.u (X). so none of the other vj is a zero of h(X). vs . By the separability assumption.73.69 to each of these intermediate separable simple extensions to obtain (L : K) = [L : K]. . we have (L : K) = (K(u1 ) : K) · · · (L : K(u1 . q(X) ∈ K[X] be the minimal polynomials of u and v over K. s for which [L : K(u)] = r(L : K(u)) and [K(u) : K] = s(K(u) : K). and therefore L/E is separable. 3. uk−1 )). Theorem (Primitive Element Theorem). THE PRIMITIVE ELEMENT THEOREM 45 Proof. . We have (L : K) = (L : E) (E : K) and [L : K] = [L : E] [E : K]. . Proposition. suppose that (L : K) = [L : K]. there are some positive integers r. Clearly E/K is also separable. . since (M : L) [M : L] and (L : K) [L : K]. For the converse.
i). i) given by √ √ √ √ √ √ 3 −→ 3 3 −→ − 3 3 −→ − 3 α1 = id. ( 3 + i) = √ = 3+1 4 ( 3 + i)( 3 − i) √ √ since ( 3 + i)−1 ∈ Q( 3 + i). Consider 3 + i. Thus we must have deg minpolyQ.75. K(w) 3.√3+i (X) = 4. Corollary.u (X) = [K(u) : K] = (K(u) : K). 2 2 √ √ √ √ √ are both in Q( 3 + i). showing that Q( 3. . we see that minpolyQ. v). i) −→ Q( 3.u (X) = (X − α1 (u)) · · · (X − αd (u)).46 3. K is inﬁnite and by Example 3. where α1 . Then minpolyK. i) : Q] = [Q( 3. This shows that K(u. i −→ −i i −→ i i −→ −i Then so √ √ α2 ( 3 + i) = ( 3 − i). In particular. Let L/K be a ﬁnite separable extension of a ﬁeld of characteristic 0. To ﬁnd a primitive element we can always use the method suggested by the proof of Theorem 3. 3. Hence √ √ √ 1 √ 1 √ 3 = (( 3 + i) + ( 3 − i)). Since Q K. hence v ∈ K(w) and so u = w − tv ∈ K(w). Then working over the subﬁeld Q( 3) Q( 3. In Example 3. √ √ √ √ 3 − i)(X − 3 + i)(X + 3 − i)(X + 3 + i) = X 4 − 4X 2 + 16 ∈ Q[X]. Proof. i) : Q( 3)][Q( 3) : Q] = 2 · 2 = 4.76. αd are the elements of MonoK (K(u).78. i = (( 3 + i) − ( 3 − i)). i) Q( 3 + i) and so Q( 3. α2 = . Then L has a primitive element. So we have [Q( 3 + i) : Q] = 4 and Q( 3 + i) = Q( 3. √ √ There are four monomorphisms αk : Q( 3. Let u ∈ K be separable over K.√3+i (X)  (X 4 − 4X 2 + 16) in Q[X]. √ √ (X 2 − 2 3X + 4)(X 2 + 2 3X + 4) = X 4 − 4X 2 + 16 ∈ Q[X]. the polynomial (X − α1 (u)) · · · (X − αd (u)) ∈ K[X] is in K[X] and is irreducible therein. √ √ α4 ( 3 + i) = (− 3 − i). i) we ﬁnd that √ i ∈ Q( 3) R and / √ √ √ √ (X − ( 3 + i))(X − ( 3 − i)) = X 2 − 2 3X + 4 ∈ Q( 3)[X].77 we have √ √ √ √ [Q( 3. d = deg minpolyK. √ √ √ Solution. Example.77. v) and therefore K(w) = K(u. . . Notice that √ √ √ √ 1 √ ( 3 − i) ( 3 − i) −1 √ = ( 3 − i) ∈ Q( 3 + i). There is a general phenomenon illustrated by Example 3. i)/Q.77. ALGEBRAIC EXTENSIONS OF FIELDS must be X − v. .√3+i (X) = X 4 − 4X 2 + 16. α4 = . K). Since K(u) is separable then by Lemma 3. √ 3.52.√3+i (X). however a ‘try it and see’ approach will often be suﬃcient. hence Now taking √ X 2 − 2 3X + 4 = minpolyQ(√3). √ √ √ Hence this polynomial is irreducible. (X − √ √ α3 ( 3 + i) = (− 3 + i). Proof.67 L/K is separable. and so minpolyQ. Find a primitive element for the extension Q( 3. i) = Q( 3 + i). α3 = . Proposition. .
. o Exercises on Chapter 3 3. C over Q and determine [E : Q] for each of the following p3 (X) = X 4 +2.2. . [Hint: for p4 (X). as claimed in the = solution of Example 3. Now any monomorphism θ ∈ MonoK (E. Now suppose that E is a splitting ﬁeld for g(X) ∈ K[X]. . K). . Prove that AutQ (Q( 3 2. If E/K is a normal extension then whenever an irreducible polynomial p(X) ∈ K[X] has a root in E. Finding splitting subﬁelds E polynomials. Also. . vk are the distinct roots of g(X) in E. consider p4 (Y − 1) ∈ Q[Y ]. it splits in E since by Lemma 3. . Definition. p2 (X) = X 6 −2. u2 ) ··· K(u1 . . [Hint: work out the eﬀect of each automorphism on the three roots of the polynomial X 3 − 2. .38. If E is the splitting ﬁeld of a polynomial f (X) ∈ K[X] over K. .3. Proof. √ Q( 4 3. or use the method of proof of Theorem 3. . . un ) = E Construct a polynomial by taking f (X) = minpolyK. 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 is generated by some of the roots of f (X). 3. Find primitive elements for the extensions Q( 5. . i)/Q. 3. Show that the polynomial g(X) = X p − T ∈ k(T )[X] is irreducible and has a multiple root in k(T ).34). . Let E/L and L/K be ﬁnite extensions. . .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. then E is the splitting ﬁeld of f (X) over L. Since θ permutes the roots vj . 3. where v1 . vk ) = E. ζ3 )) ∼ S3 .82. Remark. . 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.79. K) must map these roots to θ(v1 ). 3. Then there is a sequence of extensions K K(u1 ) K(u1 . Proof. vk ). . we have θE = θK(v1 . K). f (X) splits in E. E/K is normal if ϕE = E for every ϕ ∈ MonoK (E.7. Corollary. 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. . Prove Proposition 3. p4 (X) = X 4 +5X 3 +10X 2 +10X+5. . indeed these are known as Galois extensions.75. In Chapter 4 we will see that separable normal extensions play a central rˆle in Galois Theory.5. p1 (X) = X 4 −X 2 +1. 3. Theorem.] √ 3. Suppose that E/K is normal. .] . then by Remark 3.u2 (X) · · · minpolyK. If ϕ ∈ MonoK (E.un (X).] 3. ϕE = E if and only if ϕE E.4. 10)/Q.2. . . .32. . Q( 3.u1 (X) minpolyK.1. in each case ﬁnding it minimal polynomial over Q. How does g(X) factor in k(T )[X]? √ √ √ √ 3. Hence E is a splitting ﬁeld for f (X) over K. the symmetric group on 3 elements. If E/K is normal then E/L is normal. θ(vk ) which are also roots of g(X) and therefore lie in E (see Proposition 3. Q( 2. θ(vk )) = K(v1 .81. vk ) = K(θ(v1 ). . Then by Remark 3. .80. [Hint: look for elements of high degree over Q. i)/Q.47 3. Let k be a ﬁeld of characteristic char k = p > 0 and k(T ) be the ﬁeld of rational functions in T over k. .80. so that E = K(v1 . i)/Q. .
of degree 2) extension E/K is normal. Let f (X) ∈ Q[X] be an irreducible polynomial of odd degree greater than 1 and having only one real root u ∈ R.. . v). . i. . Show that Q(u)/Q is not a normal extension..e. ALGEBRAIC EXTENSIONS OF FIELDS 3. Prove the following converse of Proposition 3.e.8.6. If there are only ﬁnitely many subextensions F/K L/K. [Hint: First deal with the case where L = K(u.7. .] 3. Is such an extension always separable? 3. L = K(w) for some w ∈ L. Show that every quadratic (i. Let K be a ﬁeld. then use induction on n to prove the general case L = K(u1 .48 3.20: Let L/K be a ﬁnite extension. then L/K is simple. . un ).
Definition. K) maps E into itself.1)  AutK (E) = (E : K) = [E : K]. the group . [E : K] = (E : K) and also every monomorphism ϕ ∈ MonoK (E.1.1. The elements of Gal(E/K) are called (Galois) automorphisms of E/K. K) ←→ AutK (E) and we have (4.2)  Gal(E/K) = (E : K) = [E : K]. Gal(E/K) = AutK (E) is called the Galois group of the extension or the Galois group of E over K.49.3. Similarly. Definition. in particular we will explain how these are related through the Galois Correspondence. hence restricts to an automorphism of E which will be denoted ϕE . every automorphism F θ : F −→ F restricts to an automorphism θE : E −→ E. Notice that Equation (4.51. For a ﬁnite Galois extension E/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. every automorphism α ∈ AutK (E) extends to a monomorphism E −→ K ﬁxing elements of K. A ﬁnite extension E/K is a (ﬁnite) Galois extension if it is both normal and separable. Galois extensions 4. 49 4. hence gives rise to an automorphism ϕE : E −→ E. From Section 3. It is easy to see that for u. 4. if F/K is any ﬁnite normal extension with E F . Throughout the chapter. So there is a bijection = MonoK (E. Here is a slightly diﬀerent way to understand this. v ∈ E. >K ~~ ~ ~~ ~~ ∼ = E ϕ / E  ϕ E /K K Also. K) restricts to a monomorphism E −→ K whose image is contained in E. The proof of the next result is left as an exercise.CHAPTER 4 Galois extensions and the Galois Correspondence In this Chapter we will study the structure of Galois extensions and their associated Galois groups. by the Monomorphism Extension Theorem 3. 4.2. First notice that every element ϕ ∈ AutK (K. We can also reformulate the notion of conjugacy introduced in Deﬁnition 3. Let E/K a ﬁnite Galois extension and u. Definition. we also say that v is a conjugate of u. let K be a ﬁeld.5 we know that for such a Galois extension E/K. Then v is conjugate to u if there is a ϕ ∈ Gal(E/K) for which v = ϕ(u).1) implies (4. v ∈ K.
2 − 3. the action is faithful or eﬀective if for every nonidentity element h ∈ G. Using a primitive element u for the extension. Describe the Galois group Gal(Q( 2. α3 ←→ (1 2)(3 4). 4. Recall that for an extension F/K and a polynomial f (X) ∈ K[X].2. Then we know that E is a splitting ﬁeld for some polynomial over K since E/K is normal. We have √ √ √ √ √ √ [Q( 2. y ∈ X. Listing these as √ √ √ √ √ √ √ √ 2 + 3.7. ϕ −→ ϕE F is a surjective group homomorphism. If E/K is a ﬁnite Galois extension. F ) denotes the set of roots of f (X) in F .5. Suppose that E is the splitting ﬁeld of a separable irreducible polynomial f (X) ∈ K[X] of degree n. describe the Galois group Gal(Q( 2. E).8. 3) : Q] = [Q( 2. α2 = √ √3 −→ √3 √3 −→ −√3 √3 −→ −√3 − 3 −→ − 3 − 3 −→ 3 − 3 −→ 3 √ √ √ √ Writing the roots in the list 2. Proposition. K) we have (ϕF )F = ϕE . Example. − 3 and numbering them from 1 to 4. α4 ←→ (1 4)(2 3).50 4. 3) = Q( 2 + 3) and the conjugates of u = 2 + 3 are √ √ ± 2 ± 3. 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. F ) −→ AutK (E. F θ −→ θE . then the function AutK (K. there is a element z ∈ X such that hz = z. we ﬁnd the correspondences α2 ←→ (1 3)(2 4). K) −→ AutK (E. 4. 3. α3 = −√2 −→ −√2 . − 2 + 3. We have Q( 2.6. Recall that an action of a group G on a set X is transitive if for every pair of elements x. − 2.√ √ Example. there is an element g ∈ G such that y = gx (so there is only one orbit). Solution. and the following nontrivial elements of the Galois group together with the element identity α1 = id: √ √ √ √ √ √ √2 −→ −√2 √2 −→ √2 −→ −√2 √2 − 2 −→ √2 . Recollection. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE 4. for ϕ ∈ AutK (K. Then the following are true. . α4 = −√2 −→ √2 . this minimal polynomial has degree [E : K]. Next we summarize the properties of Galois groups that can be deduced from what we have established so far. Let E/K be a ﬁnite Galois extension. 3) : Q( 2)] [Q( 2) : Q] = 4. Hence E is a splitting ﬁeld for the minimal polynomial of any primitive element for E/K. If F/K K/K is any ﬁnite normal extension with E then there is a surjective group homomorphism AutK (F. α4 ←→ (1 2)(3 4). these automorphisms correspond to the following permutations in S4 expressed in cycle notation: α2 ←→ (1 2).4. √ √ √ √ √ √ Solution. Theorem. E). 3)/Q) as a subgroup of the group of permutations of the roots of (X 2 − 2)(X 2 − 3) ∈ Q[X]. E 4. Working with Galois groups Let E/K be a ﬁnite Galois extension. Furthermore. − 2 − 3. 4.u (X) ∈ Q[X]. Roots(f. We also know that E is a simple extension of K since E/K is separable. α3 ←→ (3 4). and after numbering them accordingly. 3)/Q) as a subgroup of the group of permutations of the roots of minpolyQ.
Now list the roots of f (X) in the order 2. 5 β : ζ8 −→ ζ8 .2. (iii)  Gal(E/K) divides n! and is divisible by n. i. (ii) Gal(E/K) can be identiﬁed with a subgroup of the group of permutations of Roots(f. Then Gal(Q(f (X))/Q) ∼ S3 . We also have Q(f (X)) ∩ R = Q(u1 ). u3 . i ∈ Q(ζ8 ). (1 2)(3 4)} S4 . Example. 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 .6. 4.√ hence Q( 2. = Proof. 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 . So the Galois group Gal(Q(ζ8 )/Q) corresponds to {id. Let u1 ∈ R be the real root of f (X) and let u2 . If we order the roots u1 .5 and 4. . special arguments can sometimes be exploited. i) is the splitting ﬁeld of f (X) = (X 2 − 2)(X 2 + 1) over √ √ Q. − 2. Notice that Q( 2. ζ8 . Since [Q( 2. Hence Gal(Q(f (X))/Q) is isomorphic to a subgroup of S3 and so ∼ S3 since the orders agree. un then Gal(E/K) can be identiﬁed with a subgroup of Sn . ←→ (1 2). It also contains an element of order 2 obtained by restricting complex conjugation to Q(f (X)). i) : Q] = 4. u3 . i) Q(ζ8 ). . . ζ8 . β ←→ (1 3)(2 4). so it corresponds to the transposition (2 3). this ﬁxes u1 and interchanges u2 .4. −i −→ i √ In this description. we ﬁnd that these automorphisms correspond to the following permutations in S4 : α ←→ (1 2)(3 4). and observe that √ √ √ √ 2 −→ −√2 2 −→ −√2 √ √ − 2 −→ − 2 −→ 2 2 ←→ (1 2)(3 4). u2 . 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. u3 be the remaining complex roots. (3 4). E).ζ (X) = Φ8 (X) = X 4 + 1 3 5 7 in the order ζ8 . i) = Q(ζ8 ).10. we have Q( 2. 4. If we list the roots of the minimal polynomial minpolyQ.9. the Galois group Gal(Q(ζ8 )/Q) = Gal(Q( 2. u3 ) and in fact [Q(f (X)) : Q] = 6 since [Q(f (X)) : Q]  6 and u2 ∈ Q(u1 ) / R. Then Q(f (X)) = Q(u1 . u2 . in practise it is often easier to use a not necessarily irreducible polynomial to determine and work with a Galois group. −i. (1 3)(2 4). E). WORKING WITH GALOIS GROUPS 51 (i) Gal(E/K) acts transitively and faithfully on Roots(f. ζ8 . S4 . (1 2)(3 4). (1 4)(2 3)} Noticing that 1 1 ζ8 = √ + √ i. γ ←→ (1 4)(2 3). (1 2). α: β: i −→ −i i −→ i −i −→ i −i −→ −i √ √ √2 −→ √2 − 2 −→ − 2 γ: i −→ −i ←→ (3 4). Example. While it can be hard to determine Galois groups in general. we can assume that this is (1 2 3). 2 2 √ √ √ we√ easily ﬁnd that 2. i)/Q) corresponds to the subgroup {id. . . As we have seen in Examples 4. 7 γ : ζ8 −→ ζ8 .
. By Proposition 3. γ(uv) = γ(u)γ(v) = uv. . Combining these two inequalities we obtain [E : E Γ ] =  Gal(E/E Γ ) = Γ = h and therefore Γ = Gal(E/E Γ ). Γ [E : E Γ ] =  Gal(E/E Γ ) = Γ.3. focusing on the relationship between objects of these types. we will identify its Galois group. Lemma. we also have h = Γ  Gal(E/E Γ ) = [E : E Γ ]. γh . This happens for example with f (X) = X 3 − 3X + 3 which has local extrema at ±1 and f (1) = 1. 4. For Γ Gal(E/K). E Γ EΓ. if u = 0. Finally. Proof. Notice that [E : E Γ ] = (E : E Γ ) =  Gal(E/E Γ ). Now let the distinct elements of Γ be γ1 = id. so by the Primitive Element Theorem 3. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE 4. f (c− ) < 0.13. E is the ﬁxed subﬁeld of Γ. Consider the subset of E Γ = {u ∈ E : ∀γ ∈ Γ. Such examples occur when the cubic polynomial f (X) has local maximum and minimum at real values c+ and c− with f (c+ ). E/E Γ is also normal. Hence.52 4. We know that E/E Γ is separable. .73. For u. Given a Galois extension E/K. say E = E Γ (u). v ∈ E Γ and γ ∈ Γ. E Γ Gal(E/K). we will next study subextensions L/K E/K and subgroups Γ Gal(E/K).11. so K 4.14. f (c− ) > 0 or f (c+ ). Subgroups of Galois groups and their ﬁxed ﬁelds Let E/K a Galois extension and suppose that Γ elements of E ﬁxed by Γ. γ(u−1 ) = γ(u)−1 = u−1 . Now each element of Gal(E/E Γ ) is also an element of Gal(E/K) and Gal(E/E Γ ) Gal(E/K).75 it is simple. This shows that [E : E Γ ] = [E Γ (u) : E Γ ] Since Γ Gal(E/E Γ ). Proposition. we have Gal(E/E Γ ) = Γ and the equations [E Γ : K] =  Gal(E/K) . the extensions E/E Γ and E Γ /K are separable. so Lagrange’s Theorem implies that Γ divides Γ ). In fact we have  Gal(E/E 4. Also. . if t ∈ K then γ(t) = t. Proof. f (X) ∈ E Γ [X]. Remark. . Consider the polynomial of degree h f (X) = (X − u)(X − γ2 (u)) · · · (X − γh (u)) ∈ E[X]. γ(u + v) = γ(u) + γ(v) = u + v. Notice that f (X) is unchanged by applying any γk to its coeﬃcients since the roots γj (u) are permuted by γk . Notice that by deﬁnition Γ Gal(E/E Γ ). E is a subﬁeld of E containing K. where h = Γ. γ(u) = u}. so this is a Galois extension. h = Γ. γ2 . 4. f (−1) = 5.12. Definition.
This shows that ϕ−1 θϕ ∈ Gal(E/L).. Proof. αL : L −→ L. Gal(E/L) Gal(E/K). 4. there is an automorphism θ ∈ Gal(E/L) such that θ(u) = u. Lemma. [E : L] Hence this homomorphism is an isomorphism. Proof. then αL = L since L/K is normal.e. ϕ Gal(E/L)ϕ−1 = Gal(E/L). Subﬁelds of Galois extensions and relative Galois groups Let E/K a Galois extension and suppose that L/K E/K (i. Then αL is the identity function on L if and only if α ∈ Gal(E/L). then for every ϕ ∈ Gal(E/K) and v ∈ L. so for every θ ∈ Gal(E/L). 4.  Gal(E/K)/ Gal(E/L) = . Thus we obtain an injective homomorphism Gal(E/K)/ Gal(E/L) −→ Gal(L/K) for which [E : K] = [L : K] =  Gal(L/K). The next result explains the connection between the two uses of the word normal which both ultimately derive from their use in Galois theory. every monomorphism L −→ K ﬁxing K extends to a monomorphism E −→ K which must have image E. (ii) If L/K is normal and hence a Galois extension. The following is immediate. we have βαβ −1 ∈ Gal(E/L). It is easy to see that the function Gal(E/K) −→ Gal(L/K). SUBFIELDS OF GALOIS EXTENSIONS AND RELATIVE GALOIS GROUPS 53 4. and its order is  Gal(E/L) = [E : L].. (ii) If α ∈ Gal(E/K). Proposition. Suppose that u ∈ E − L. since γ −1 αγ ∈ Gal(E/L). Now if u ∈ L. hence u ∈ E Gal(E/L) . (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.e.17. Hence we can restrict α to an automorphism of L. 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. By Theorem 4. so the above argument shows that L/K is normal. We need to understand when Gal(E/L) Gal(E/K) is actually a normal subgroup. hence γ(u) ∈ E Gal(E/L) = L.e. Hence for every ϕ ∈ Gal(E/K).16. i. Proposition. K L E). if L/K is normal.. The relative Galois group of the pair of extensions L/K E/K is a subgroup of Gal(E/K). i. Let L/K E/K. α −→ αL is a group homomorphism whose kernel is Gal(E/L). γ(u) ∈ E satisﬁes αγ(u) = γ(γ −1 αγ(u)) = γ(u). 4. Conversely.8(i). This shows that / E Gal(E/L) L and therefore E Gal(E/L) = L.4.49.15. which shows that Gal(E/L) Gal(E/K). θ(ϕ(v)) = ϕ(v) and therefore ϕ−1 θϕ(v) = v. → ∼ = α Gal(E/L) −→ αL .4. αL (u) = α(u). By the Monomorphism Extension Theorem 3. Then L = E Gal(E/L) . (i) Suppose that Gal(E/L) Gal(E/K). ϕ(v) ∈ L. Clearly L E Gal(E/L) . then for any γ ∈ Gal(E/K) and α ∈ Gal(E/L). Let E/K be a ﬁnite Galois extension and L/K E/K.4. for all α ∈ Gal(E/L) and β ∈ Gal(E/K). then there is a group isomorphism Gal(E/K)/ Gal(E/L) − Gal(L/K).
19. Each of these sets is ordered by inclusion. α0 . S(E/K) is also a ﬁnite set. It = is useful to make this isomorphism explicit. First take the √ roots of the polynomial X 3 − 2 for 3 √ √ 2 which E is the splitting ﬁeld over Q. Example. Let E/K be a ﬁnite Galois extension. α1 . α1 . 3 2 ζ3 which we number in the order they are listed. Similarly. α2 = (1 3 2). Hence ΦE/K (L2 ) ΦE/K (L1 ) and so ΦE/K reverses order. . Also.54 4. each of which permutes these 3 roots in the following ways given by cycle notation: α0 = (2 3). We will use the following notation. Then the functions ΦE/K and ΘE/K are mutually inverse bijections which are orderreversing. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE 4. Corollary. Let E/K be a ﬁnite Galois extension. Theorem (Main Theorem of Galois Theory). 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. This shows that ΦE/K and ΘE/K are mutually inverse and so are inverse bijections. Proof.18 which is closely related to Proposition 3. When dealing with a ﬁnite Galois extension E/K.1 shows the Galois Correspondence for the extension of Example 3.5. 4. by Proposition 4. the Galois Correspondence inverts these diagrams. the Galois group here is Gal(Q( 3 2. We can also do this with the subgroups of the Galois group Gal(E/K) with labels indicating the index of the subgroups. 3 2 ζ3 . ΘE/K (ΦE/K (L)) = ΘE/K (Gal(E/L)) = E Gal(E/L) = L. F(E/K) = the set of all subextensions L/K of E/K. 4. α2 extend to automorphisms of E. ζ3 )/Q) ∼ S3 . In eﬀect.16 that for an extension L/K in F(E/K). Γ2 ∈ S(E/K) and Γ1 Γ2 . if Γ1 . α2 = (1 3). α1 = (1 2 3). ΦE/K (L) = Gal(E/L). Proof. Since every subgroup of a ﬁnite group is a ﬁnite subset of a ﬁnite set. Then there are only ﬁnitely many subextensions L/K E/K. α2 .18.14 for H ∈ S(E/K) we have ΦE/K (ΘE/K (Γ)) = ΦE/K (E Γ ) = Gal(E/E Γ ) = Γ. 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 ). ΘE/K (Γ) = E Γ . ΘE/K : S(E/K) −→ F(E/K). 4.20.30. let S(E/K) = the set of all subgroups of Gal(E/K). Hence ΘE/K reverses order. We know from Proposition 4.20. so is F(E/K). normal subextensions of E/K correspond to normal subgroups of Gal(E/K) and vice versa.38. Then the monomorphisms id. α1 = (1 2). we indicate the subextensions in a diagram with a line going upwards indicating an inclusion. There is an immediate consequence of the Main Theorem 4. ΦE/K F(E/K) o / S(E/K) ΘE/K Under this correspondence. Figure 4. √ As noted at the end of Example 3. Let L1 /K. L2 /K ∈ F(E/K) satisfy L1 /K L2 /K. these are 3 2. Deﬁne two functions by ΦE/K : F(E/K) −→ S(E/K). Since the set S(E/K) is ﬁnite. For a ﬁnite Galois extension E/K. Then Gal(E/L2 ) Gal(E/L1 ) since L1 ⊆ L2 and so if α ∈ Gal(E/L2 ) then α ﬁxes every element of L1 .
Proof. α1 } ∼{id. . Complex conjugation ( ) : C −→ C restricts to an automorphism of E over Q. Furthermore. (1 2)}. = = Notice that {id. Setting ER = R ∩ E. ( )E/Q : E −→ E. (2 3)}. The Galois Correspondence for E = Q( 3 2. (1 2 3). α2 } ∼{id. minpolyQ.u (X) ∈ Q[X] splits over E. (1 3 2)}. But ( ) permutes the roots of this minimal polynomial. α0 } ∼ ={id. √ 2 √ 3 3 Gal(E/Q( 2 ζ3 )) = {id. (1 3)}. 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.1. α2 } ∼ Gal(E/Q( 2)) = {id. 4. Of course Q(ζ3 ) is the splitting ﬁeld of X 3 − 1 over Q. Therefore ( ) maps E into itself. (1 3 2)} S3 and so Q(ζ3 )/Q is a normal extension. (1 2 3). so all of its complex roots lie in E. ζ3 ) H ii eee ii eeeeeekkkkkk eee ii kk 2 eeeeee ii eeee 2 kkk 2 eee 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 rr 3 rr 3 Q(ζ3 ) rr Q kk kkk rrr 2 k rr kkk kk rr kkk r kkkk QS ΦE/K √ 3 kk v 2 kkk vv kkk vvv kkk kk vv v kk 3 kkkk 3 vv v k Gal(E/Q(ζ3 )) 3 v kk k vv y kkk vv kkk yy vv kkk yy v kk v k yy √ v √ kkk √v yy 3 3 3 2 )) 3 yy Gal(E/Q( 2 ζ3 yy Gal(E/Q( 2)) Gal(E/Q( 2 ζ3 )) y yy yy 2 yy 2 2 yy Gal(E/Q) {id} √ Figure 4. Let u ∈ E. Proposition. Gal(E/Q( 2 ζ3 )) = {id. ( )E/Q } ∼ Z/2. ={id.6. ζ3 )/Q We ﬁnd that √ 3 Gal(E/Q(ζ3 )) = {id.4. = hence. α1 . (i) ( )E/Q agrees with the identity function if and only if ER = E.21. 4. Let E/Q be a ﬁnite Galois extension with E/Q C/Q. (ii) If ER = E. As E/Q is normal. then ( )E/Q = {id. GALOIS EXTENSIONS INSIDE THE COMPLEX NUMBERS AND COMPLEX CONJUGATION 55 E = Q( 2. ER = E ( )E/Q and [E : ER ] = 2.6. we have Q ER E.
Here u1 . . and E ( )E/Q = {u ∈ E : u = u} = ER . exactly one of the inequalities σ(i) < σ(j) or σ(j) < σ(i) must hold and the ratio (σ(j) − σ(i))/(j − i) is either positive or negative. z = z if and only if z ∈ R. j−i Note that this is sometimes used as the deﬁnition of sgn σ. the Galois group of its splitting ﬁeld E over K can naturally be thought of as a subgroup of the symmetric group Sn .3) sgn σ = 1 i<j n σ(j) − σ(i) . 4.23. 2 C R ∞ E nnn nnn n nnn nnn 2 n ER e ee ee ee e ∞ Q We will usually write ( ) rather than ( )E/Q when no confusion seems likely to result. Definition. . Example.22. where we view the latter as permuting the roots of f (X). Galois groups of even and odd permutations We have seen that for a monic separable polynomial f (X) ∈ K[X] of degree n. 4. as we have assumed that f (X) is separable. . sgn σ = ±1. i). the ui are distinct. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE (i) For z ∈ C. The discriminant of f (X) is Discr(f (X)) = 1 i<j n (uj − ui )2 ∈ E. Consider the cyclotomic extension Q(ζ8 )/Q where 1 1 ζ8 = eπi/4 = √ + √ i. (ii) Here  ( )E/Q  = 2. It is reasonable to ask when Gal(E/K) An rather than just Gal(E/K) Sn . We ﬁrst recall an interpretation of the sign of a permutation σ ∈ Sn .7. Notice that Discr(f (X)) = 0 since ui = uj if i = j. 4. It is easily veriﬁed that the righthand side of the following equation must have value ±1 and so (4. .9 we know that √ Q(ζ8 ) = Q( 2. [Q(ζ8 ) : Q] = 4. . j with 1 i<j n. un ∈ E are the roots of f (X). and we easily see that √ Q(ζ8 )R = Q( 2). Suppose that f (X) factorizes over E as n f (X) = (X − u1 ) · · · (X − un ) = i=1 (X − ui ). For each pair i. 2 2 From Example 4.56 4.
. . . . . . . .. . . . . . . . . bn Then if f (X) is monic with d = deg f (X). . . Discr(X 5 + a4 X 4 + a0 ) = a3 (3125a0 + 256a5 ). . . . . .. . . . . . . . . . q(X)) = . . . . . . . . . . .. . . . . . Discr(X 3 + pX + q) = (−1)3 Res(X 3 + pX + q. .. . . 0 1 2 2 1 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 .. . . .5) So for example. . . . . . . . . . . . . . Here are some low degree examples of discriminants obtained with the aid of Maple. . . . . . . . .. . . 3X 2 + p) q 0 = (−1) p 0 0 n = 2: n = 3: n = 4: p q 0 p 0 0 p 3 0 p 1 0 0 3 0 0 1 0 = −4p3 − 27q 2 .. . . a0 0 . . . . . . . a1 a0 .4) Res(p(X). . . . . . .. . . b0 b1 . . Remark. (4. . . . . . . . 1 Discr(a0 + a1 X + a2 X 2 + X 3 ) = −27a2 + 18a0 a1 a2 + a2 a2 − 4a3 a0 − 4a3 . . . . . . . .4.. .. There is an explicit formula for computing Discr(f (X)) is terms of its coeﬃcients. . . . . . . their resultant is the (m + n) × (m + n) determinant (with n rows of ai ’s and m rows of bi ’s) . 1 0 + 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 . ..7. . . 0 . am 0 . . . . . . . . . Discr(a0 + a1 X + X 2 ) = −4a0 + a2 . .24. . . 0 4 Discr(X 5 + a1 X + a0 ) = 256a5 + 3125a4 . . .. .. . 0 a0 a1 .. . am 0 . . . . . . . . . . . . . . 0 3 Discr(f (X)) = (−1)d(d−1)/2 Res(f (X). am (4. . . . . . . . . . . 0 . . .. . . . . 0 . . . . . . . . . . . . . .. . . . .. . . . . . . . . . 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 . . . 0 . 0 0 b0 b1 . .. . . . .. bn 0 . .. .. . .. . . . . . .. . . For polynomials p(X) = a0 + a1 X + · · · + am X m . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . q(X) = a0 + a1 X + · · · + am X m .. . 0 b0 b1 . GALOIS GROUPS OF EVEN AND ODD PERMUTATIONS 57 4. . f (X)). . . 0 a1 ... . bn 0 . . So for example. . . . . . . . . . .
26. we have Discr(f (X)) ∈ K.28. Let f (X) = X 4 + aX 2 + b ∈ Q[X] be irreducible. so the square roots of Discr(f (X)) are ±δ(f (X)). Of course  Gal(E/K)/ Gal(E/K) ∩ An  = 2. 1 i<j n Then δ(f (X))2 = Discr(f (X)). Let g(X) = X 2 + aX + b ∈ Q[X].8. (uσ(j) − uσ(i) )2 = Now for each pair i. = 2 − 4b) is a square in Q then Gal(Q(f (X))/Q) ∼ Z/4. this means that sgn σ = 1. For every σ ∈ Gal(E/K). Hence Discr(f (X)) ∈ E Gal(E/K) = K. hence (a2 − 4b) is not a square in Q.42 we obtain Discr(X 5 − 35X 4 + 7) = −4611833296875 = −33 · 56 · 74 · 29 · 157. For σ ∈ Gal(E/K) σ(Discr(f (X))) = 1 i<j n Sn . ±v. Proposition. if δ(f (X)) ∈ K then / K(δ(f (X))) = E Gal(E/K)∩An .58 4. we ﬁnd that the roots of g(X) are (−a±δ)/2 ∈ Q. σ(uj − ui ) = uσ(j) − uσ(i) .3) (4. where u2 = (−a + δ) . (i) If b is a square in Q then Gal(Q(f (X))/Q) ∼ Z/2 × Z/2. The following result describes the Galois groups that occur and the proof introduces some useful computational techniques. Since Now let δ(f (X)) = E Gal(E/K) = K. √ √ δ(X 5 − 35X 4 + 7) = ±53 · 3 · 72 · 3 · 29 · 157 i = ±18375 13659 i ∈ Q. σ(δ(f (X))) = sgn σ δ(f (X)) = ±δ(f (X)). The Galois group Gal(E/K) Discr(f (X)) is a square in K. / / Then the roots of f (X) are ±u. 4. / Discr(X 5 + 20X + 16) = 1024000000 = 216 · 56 . By Equation (4.27. Example. Setting d = (a2 − 4b) ∈ Q and taking δ to be a square root of d (so δ ∈ Q). Notice that g(X) must be irreducible since otherwise f (X) would factorize.40 and 6. If δ(f (X)) ∈ K. 4. Now consider the eﬀect of σ ∈ Gal(E/K) on δ(f (X)) ∈ E. 2 v2 = (−a − δ) . (uj − ui ) ∈ E. 4. σ(Discr(f (X))) = Discr(f (X)). Hence σ(Discr(f (X))) = Discr(f (X)). GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE 4. j with i < j. = Proof. Proof.6). On the other hand.6) 1 i<j n (uσ(j) − uσ(i) ) = sgn σ 1 i<j n (uj − ui ) = (±1) 1 i<j n (uj − ui ). 2 . we have 1 i<j n 2 (uσ(j) − uσ(i) ) . 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].25. Proposition. δ(X 5 + 20X + 16) = ±28 53 ∈ Q. For the polynomials of Examples 6. Sn is contained in An if and only if 4. (ii) If b(a = (iii) If neither b nor b(a2 − 4b) is a square in Q then Gal(Q(f (X))/Q) ∼ D8 . and by Equation (4. Theorem (Kaplansky’s Theorem).
(i) We have a2 − d 4b (uv)2 = u2 v 2 = = = b. στ = τ σ. Gal(Q(δ)/Q) = Gal(E/Q)/N. τ σ(δ) = στ (δ) = −δ. E = Q(u) 2 Q(δ) 2 Q In particular [E : Q] = 4 =  Gal(E/Q). so we can write uvδ = c ∈ Q or equivalently v = c/(uδ) ∈ Q(u) since Q(δ) Q(u). hence we might as assume that σ(v) = u since if necessary we can replace our original choice by τ σ. Setting c = uv ∈ 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 . given u we might as well choose v so that σ(u) = v. Since deg f (X) = 4.8. we ﬁnd that v = c/u ∈ Q(u). σ(v) = σ(uδ) vδ . Given u we might as well choose v so that σ(u) = v. then (uvδ)2 = u2 v 2 d = bd. Hence there is an element σ ∈ Gal(E/Q) for which σ(δ) = −δ. This shows that E = Q(u) and we have the following Galois tower. we also have 4  [E : Q]. 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). τ (u) = −u. This element must also have the eﬀects σ(u) = ±v and σ(v) = ±u. σ. This shows that Gal(Q(f (X))/Q) = Gal(E/Q) = {id. since E is obtained by at most 3 successive quadratic extensions we also have [E : Q]  8. In fact. We now have c c σ(u) = . = (ii) If bd is a square in 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. Notice that c c = − = −u. τ σ(u) = στ (u) = − . v) which contains the quadratic extension Q(δ)/Q. στ } ∼ Z/2 × Z/2 = the Klein 4group. There is also an element τ ∈ N for which τ (u) = −u and we also have τ (v) = −v. τ.4. Notice that if σ(v) = −u then easy calculation shows that τ σ(v) = στ (v) = u. u u These satisfy σ 2 = τ 2 = (στ )2 = id = the identity. KAPLANSKY’S THEOREM 59 so the splitting ﬁeld of f (X) over Q is E = Q(u. which is a square in Q. This shows that E = Q(u. 4 4 hence uv is a square root of b which is in Q.
ασ −1 form a group isomorphic to the dihedral group of order 8. This shows that Gal(Q(f (X))/Q) = Gal(E/Q) = {id. = . So we have a Galois tower which includes the following subﬁelds. ασα = βσβ = σ −1 . E = Q(u. σ −1 . q ∈ Q.60 4. By renaming −v to v if necessary. σ 2 σ 3 } ∼ Z/4 = a cyclic group of order 4. α. implying that bd was a square in Q. we may assume that v = α(u) and so u = α(v). δ) xxx xxx xxx xx vvv vvv vvv 2 vvv Q(uv) Q p ppp 2 ppp pp 2 ppp Q(uvδ) Choose α ∈ Gal(E/Q(uv)) Gal(E/Q) so that α(δ) = −δ. then uv = p+qδ for some p. Choose γ ∈ Gal(E/Q(δ. γ. By squaring we obtain b = (p2 + q 2 d) + 2pqδ. The eight elements id. σ. so uv ∈ E is a square root of b in E. if p = 0 then b = q 2 d and so bd = (qd)2 . By an easy calculation we ﬁnd that (uv)2 = b. Also. v) Q(δ) rrr rrr r rrr Q(uv. so by interchanging ±δ if necessary we can assume that β(u) = −u and β(v) = v. ασ. Then we must have γ(v) = −v since γ(uv) = uv. = (iii) Suppose that d. Thus we have Q(uv) ∩ Q(δ) = Q. b and bd are not squares in Q. Then σ 2 = γ and σ has order 4. Therefore we have Gal(Q(f (X))/Q) = Gal(E/Q) ∼ D8 . Setting σ = αβ we ﬁnd σ(u) = −v and σ(v) = u. A similar discussion shows that Q(uvδ) ∩ Q(δ) = Q = Q(uvδ) ∩ Q(uv). Notice that α2 = id. D8 . and so pq = 0. σ. We cannot have q = 0 since this would imply that b was a square in Q. uv)) Gal(E/Q) so that γ(u) = −u. Notice that γ 2 = id. Choose β ∈ Gal(E/Q(δ)) Gal(E/Q) with β(uv) = −uv. Notice that β 2 = id. αγ. Suppose that uv ∈ Q(δ). GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE so σ 2 (u) = −u. We must have either β(u) = −u or β(v) = −v.
b ∈ K. suppose that w1 . Gal(Q(X + 2X + 2)/Q) = Exercises on Chapter 4 4. v) 2 Q(δ) r 2 rrr 2 rrr rrr vvv vvv vvv 2 vvv Q(uv. deduce that a. βα2 . α3 . (a) Show that every nonabelian ﬁnite group has order at least 6. Show that f (X) = X 3 −3X +1 ∈ Q[X] is irreducible over Q.29. The corresponding Galois tower is E = Q(u. Show that ∆ = −4b3 − 27a2 . δ) xxx xxx2 xxx xx Q(uv) Q p ppp 2 ppp pp 2 ppp Q(uvδ) 4. and hence ∆ ∈ K. This is a revision exercise on ﬁnite groups of small order. If f (X) is monic. α2 . show that Gal(E/K) ∼ = [Hint: Consider K(δ) in Gal(E/K) S3 . β.2. and show that its discriminant is a square in Q. = 4 2 ∼ D8 .4. (c) Continuing with the notation and assumptions of (b). 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. (a) Show that there u. Prove that the Galois group of f (X) over Q is cyclic. α. βα3 satisfying α4 = ι.1. prove that the splitting ﬁeld of f (X) over K is a ﬁnite Galois extension of K. Example. if δ ∈ K. explain why Gal(E/K) is isomorphic to one of the groups S3 or A3 . / E and the eﬀect on the element δ of even and odd permutations 4. 4. b ∈ K. 4. w2 . = Gal(Q(X 4 + 4X 2 + 2)/Q) ∼ Z/4. βα.3. 3 and suppose that f (X) ∈ K[x] is a cubic polynomial.] A3 S3 if δ ∈ K. w3 are the distinct roots of g(X) in E and let ∆ = (w1 − w2 )2 (w2 − w3 )2 (w1 − w3 )2 ∈ E. If δ = (w1 − w2 )(w3 − w3 )(w1 − w3 ). βαβ = α−1 = α3 . . If f (X) ∈ K[X] is a separable polynomial.61 and [E : Q] = 8. We have the following Galois groups: Gal(Q(X 4 + 1)/Q) ∼ Z/2 × Z/2. β 2 = ι. v ∈ K with u = 0 such that f (uX + v) = X 3 + aX + b for some a. (b) Let D8 be the dihedral group with the eight elements ι. Let K be a ﬁeld for which char K = 2.
8. (b) Suppose that K contains p distinct pth roots of 1. √ √ √ Q( 23 i)(X 3 − X − 1)/Q( 23 i).62 4. Q( 5). Suppose that 0 = a ∈ K and f (X) = X p − a ∈ K[X]. Determine all the subextensions F E for which F/Q is Galois. Show that either f (X) is irreducible over K or it factors into p distinct linear factors over K.7. 4.28 to ﬁnd the Galois group of the splitting ﬁeld E of the polynomial X 4 + 3 ∈ Q[X] over Q. 4. Use Kaplansky’s Theorem 4. Q( 3 i)(X 3 − 10)/Q( 3√ i). If u ∈ L is one such root. Find the Galois groups for each of the following extensions: √ √ √ √ Q(X 3 − 10)/Q. Let K be a ﬁeld with char K = p. 4. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE Find all the normal subgroups of D8 . 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. Show that either f (X) is irreducible over K or it has a root in K.6.5. Let p > 0 be a prime. describe the remaining roots and show that L contains p distinct pth roots of 1. . (c) Suppose that the only pth root of 1 in K is 1. (a) Show that f (X) has p distinct roots in L. Q( 2)(X 3 − 10)/Q( 2). Q(i). Suppose that 0 = a ∈ K and f (X) = X p − a ∈ K[X]. Q( 5 i). K(X 3 − X − 1)/K for K = Q. 4. Let L/K where L is a splitting ﬁeld for f (X) over K.
. For each d 1. We start with the algebraic closure Fp of Fp and consider the polynomial Θpd (X) = X p − X ∈ Fp [X]. p Proof. Furthermore. Let F be a ﬁnite ﬁeld with q elements and let V be an F vector space. hence by Proposition 3. Then in Fp [X] we have d X p − X = X(X − u1 ) · · · (X − upd −1 ). . . Let F be a ﬁnite ﬁeld and E/F an extension. Corollary. (u + v)p − (u + v) = (up + v p ) − (u + v) = (up − u) + (v p − v) = 0.4. and each root is separable over Fp . then K is an Fp vector space. (uv)p − uv = up v p − uv = uv − uv = 0. hence F0d = F× . Hence Fpd Fp . . Notice that Fp Fpd .10. upd −1 . v ∈ Fpd then by the Idiot’s Binomial Theorem 1. If d = dimF V < ∞. Our next task is to show that for each power pd there is a ﬁnite ﬁeld with pd elements. vd we can express each element v ∈ V uniquely in the form v = t1 v1 + · · · + td vd . Let Fpd = {u ∈ Fp : Θpd (u) = 0} ⊆ Fp . Let K be a ﬁnite ﬁeld. Then dimF V < ∞ if and only if V is ﬁnite in which case V  = q dimF V . td ∈ F . Therefore by Corollary 1. 5. Then K/Fp is ﬁnite and K = p[K:Fp ] . 5. Conversely. say 0. u1 . Lemma. 5.3. . where t1 . . Our ﬁrst goal is to count the elements of K. Finite ﬁelds If K is a ﬁnite ﬁeld.CHAPTER 5 Galois extensions for ﬁelds of positive characteristic In this chapter we will investigate extensions of ﬁelds of positive characteristic. . p pd 63 d d d d d d d d d d .2. if V is ﬁnite then any basis has ﬁnitely many elements and so dimF V < ∞. the extension Fpd /Fp is a separable splitting ﬁeld. . 5. Proposition. so Fpd /Fp is a ﬁnite extension. Notice that Θpd (X) = −1. Fpd is a ﬁnite subﬁeld of Fp with pd elements and × F0d = Fpd . so V  = q d . especially ﬁnite ﬁelds. Notice that u ∈ F0d if and only if up p d −1 d F0d = {u ∈ Fpd : u = 0}. A thorough account of ﬁnite ﬁelds and their applications can be found in [6].55 every root of Θpd (X) in Fp is simple. Then E is ﬁnite if and only if E/F is ﬁnite and then E = F [E:F ] . Here is a more general result. . then for a basis v1 . . p = 1.1. Throughout this chapter we will assume that K is a ﬁeld of characteristic char K = p (where p > 0 is a prime) containing the prime subﬁeld Fp . 5. Clearly there are exactly q d such expressions. In any ﬁeld the nonzero elements are always invertible. Proof. If u. Furthermore. .35 Θpd (X) must have exactly pd distinct roots in Fp . Corollary.1. . if u = 0 then up −1 = 1 and so u has multiplicative inverse up −2 .
Proposition. Consider the polynomial X 4 − X ∈ F2 [X]. d d (i) Fpd Fp is the splitting subﬁeld for each of the polynomials X p − X and X p −1 − 1 over Fp . b ∈ F2 . b. so if u ∈ F3 is a root of X 2 + 1 then F3 (u)/F3 has degree 2 and F3 (u) = F9 . This tells us that every element of F4 = F2 (w) can be uniquely expressed in the form a + bw with a. The ﬁnite subﬁeld Fpd Fp is called the Galois ﬁeld of order pd . 5. (ii) Fpd Fp is the unique subﬁeld with pd elements.7. By (ii). Corollary. Let d 1. Similarly. We can ﬁnd their roots in F9 using the quadratic formula since in F3 we have 2−1 = (−1)−1 = −1. (ii) Let F Fp have pd elements. Fp1 = GF(p1 ) = GF(p) = Fp and [Fpd : Fp ] = d. it is the splitting subﬁeld. d The nonzero elements of Fpd are the roots of X p −1 − 1. To calculate products we use the fact that w2 = w + 1. the image of the resulting monomorphism must be Fpd . The notation GF(pd ) is often used in place of Fpd . d d each element u ∈ F × has order dividing pd − 1. Consider the polynomial X 9 − X ∈ F3 [X]. hence F ∼ Fpd . Then K/Fp is a ﬁnite Galois extension.6. This group is abelian and has pd − 1 elements. Of course. (i) As Fpd consists of exactly the roots of Θpd (X) in Fp . d ∈ F2 we have (a + bw)(c + dw) = ac + (ad + bc)w + bdw2 = (ac + bd) + (ad + bc + bd)w. therefore F ∼ Fpd . Example. 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). So X 2 + X − 1 and X 2 − X − 1 are also quadratic irreducibles in F3 [X]. hence X 2 + 1 ∈ F3 [X] is irreducible. GALOIS EXTENSIONS FOR FIELDS OF POSITIVE CHARACTERISTIC 5. = It is worth noting the following consequence of this result and the construction of Fpd . 5. c. equality follows since these d elements. Definition. Then we have F9 = F3 (u) = F3 (1 ± u) = F3 (−1 ± u). therefore up −1 = 1 and so up = u. By inspection. The discriminant of X 2 + X − 1 is 1 − 4(−1) = 5 = 2 = u2 .49 for K = L = Fp and M = F . Let K be a ﬁnite ﬁeld of characteristic p.9. Multiplication is carried out using the relation u2 = −1 = 2.8. so by Lagrange’s Theorem. (iii) If F is any ﬁeld with pd elements then there is an monomorphism F −→ Fp with image Fpd . Notice that X 2 + 1 has no root in F3 . Its splitting ﬁeld is a quadratic extension F2 (w)/F2 where w is one of the roots of X 2 + X + 1. 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). subﬁelds both have p (iii) Apply the Monomorphism Extension Theorem 3. Example. b ∈ F3 . the other being w + 1 since the sum of the roots is the coeﬃcient of X. Now X 2 + X + 1 has no root in F2 so it must be irreducible in F2 [X]. But this means every element of F is a root of Θpd (X) and so F Fpd .64 5. 5. = Proof.5. 5. so Fpd is also the splitting subﬁeld for this polynomial. so its roots are (−1)(−1 ± u) = 1 ± u. By inspection. Every element of F9 can be uniquely expressed in the form a + bu with a. Let us ﬁnd an irreducible polynomial of degree 2 in F3 [X]. . the discriminant of X 2 − X − 1 is 1 − 4(−1) = 5 = 2 = u2 and its roots are (−1)(1 ± u) = −1 ± u. Notice that the nonzero elements of F form a group F × under multiplication. so for a.
{ {{ {{ { {{ { {{ {{ { {{ Fp24 Fp8 Fp12 Fp4 Fp2 { {{ {{ { {{ Fp6 gg gg gg gg { {{ {{ { {{ gg gg gg gg Fp3 Fp Figure 5. The separability statement follows from Corollary 5. = up m(k−1) = · · · = up = u. .2. 5. Theorem.35 and Lemma 1. If Fpm Fpn . Proposition. so m  n. w ∈ F× is called a primitive root if it is a primitive (pd − 1)th root of pd K × is cyclic. 5.1. pd This is a special case of a more general result about arbitrary ﬁelds. Proposition.1. each element u ∈ Fp is separable over Fp . Then for u ∈ Fpm we have up = u. write n = km with k up = up n mk 1.10. The subﬁelds of Fp24 5. Proposition 5. pd pd 5. n 1 Furthermore. Then u is algebraic over Fp and the extension Fp (u)/Fp is ﬁnite. 5.5.46. so m m(k−1) m = (up )p Fpn . i. Proposition.13. Then every ﬁnite subgroup U Proof. Let Fpm and Fpn be two Galois ﬁelds of characteristic p.. its order in the group F× is (pd − 1). The group of units F× in Fpd is cyclic. Proof. The algebraic closure of Fp is the union of all the Galois ﬁelds of characteristic p. then by Corollary 5. Let u ∈ Fp . Use Corollary 1.11. pn = (pm )[Fpn :Fpm ] = pm[Fpn :Fpm ] .1 which shows extensions with no intermediate subextensions. unity. The diagram of subﬁelds for Fp24 can be seen in Figure 5.e. Hence by Corollary 5. We will require a useful fact about Galois ﬁelds.10 now implies that Fp (u) = Fpn for some n. FINITE FIELDS 65 There are two issues we can now clarify.12. 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 K be a ﬁeld.2. Then Fpm Fpn if and only if m  n. Fp (u) Fp is a ﬁnite subﬁeld. Proof.7. Fp = Fpn .14. hence w = F× . If m  n. Definition.
20. 5.15.75 which we had previously only established for inﬁnite ﬁelds. We record a fact that is very important in Number Theory. This completes the proof of the Primitive Element Theorem 3. so Fp2 ∼ Fp [X]/(X 2 +1). By Proposition 5. for u ∈ Fpd we have Fd (u) = up = u.16. Here is a generalization of Proposition 5.8 we ﬁnd that F4 = F2 (w) has the two primitive roots w and w + 1. Example. Definition.73. Example.17. i.9 we have F9 = F3 (u) and F× is cyclic of order 8. In Example 5. . this extension is Galois and  Gal(Fpnd /Fpd ) = [Fpnd : Fpd ] = n. (ii) If p ≡ 3 (mod 4) the polynomial X 2 +1 ∈ Fp [X] is irreducible. The extension of Galois ﬁelds Fpnd /Fpd is simple. 5. Then every element of Fpnd can be expressed as a polynomial in w. d d d d d d d × × Fd (uv) = (uv)p = up v p . The order of Fd is n.22. we have the identities Fd (u + v) = (u + v)p = up + v p .12. Fpnd = Fpd (u) for some u ∈ Fpnd . but that conﬂicts with our usage.. 5. Fpnd has a primitive root w say. Since 9 ϕ(8) = 4.e. Remark. 5. the polynomial X 2 + 1 ∈ Fp [X] has two roots in Fp .66 5.20. so Gal(Fpnd /Fpd ) = Fd . Proposition. Proof.6(i). 5.19.18. 5. This implies that Fpnd = Fpd (w). We next introduce an important element of the Galois group Gal(Fpnd /Fpd ). Remark. For u. which is p p impossible since p − 1 ≡ 2 (mod 4). (i) We have 4  (p − 1) = F× . Let p > 0 be an odd prime.7 and Proposition 3. (i) If p ≡ 1 (mod 4). so Fpnd Fpd (w) Fpnd . 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. Proof. (ii) If v ∈ Fp is a root of X 2 + 1 then v has order 4 in F× . But then 4  (p − 1) = F× . so Fd is a ring homomorphism. although as we will in the next result. the cyclic group generated by Fd . = Proof. so Fd ∈ Gal(Fpnd /Fpd ). Proposition. some authors use the term primitive element for what we have called a primitive root. By Proposition 5.23.2. so if u ∈ F× is a generator of this cyclic group. every primitive root is indeed a primitive element in our sense! 5. Proposition. hence this is a root of X 2 + 1 (the other root is −uFp /4 ). 5. Proposition. the p p order of uFp /4 is 4. The relative Frobenius mapping Fd : Fpnd −→ Fpnd is an automorphism of Fpnd that ﬁxes the elements of Fpd . Unfortunately the word primitive has two confusingly similar uses in the context of ﬁnite ﬁelds. The (relative) Frobenius mapping for the extension Fpnd /Fpd is the funcd tion Fd : Fpnd −→ Fpnd given by Fd (t) = tp . Galois groups of ﬁnite ﬁelds and Frobenius mappings Consider an extension of Galois ﬁelds Fpnd /Fpd . Indeed.21. Corollary 5. Also. In Example 5. Fpd contains a primitive nth root of unity if and only if pd ≡ 1 (mod n) and p n. 5. v ∈ Fpnd . GALOIS EXTENSIONS FOR FIELDS OF POSITIVE CHARACTERISTIC 5.
Suppose the order is k with k n. 5. Fd (t) = tp . for each k with k  n there is the cyclic subgroup Fk of order d  Fk  = n/k. This is left as an exercise. then every element u ∈ Fpnd satisﬁes the equation Fk (u) = u which expands to up d u ∈ Fpkd . d Fk Fpnd n/k F Fk d pnd = Fpdk k Fpd Proof. Fd (u) = u if and only if u ∈ Fpd . In fact for u ∈ Fp . + u = TFpnd /Fpd (u). To see that Fd is an automorphism. THE TRACE AND NORM MAPPINGS 67 so Fd ﬁxes the elements of Fpd .24.1. d d Figure 5. (ii) The restriction of Fd to the Galois subﬁeld Fpdn agrees with the relative Frobenius mapping Fd : Fpnd −→ Fpnd . Proposition.2 shows the subgroup diagram corresponding to the lattice of subﬁelds of Fp24 shown in Figure 5. Then Fd is invertible with inverse F−1 = Fn−1 . hence 1. (i) Fd : Fp −→ Fp is an automorphism of Fp which ﬁxes the elements of Fpd . This also shows d d that the order of Fd in the group AutFpd (Fpnd ) is at most n. . hence = id. so AutFpd (Fp ) is inﬁnite. hence Fk (u) = u if and only if u ∈ 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. then Fk = Fkd . For k  n. consider the function 5. Let d 1. Fd has inﬁnite d order. the ﬁxed subﬁeld of Fk in Fpnd is Fpnd = Fpdk . Proof. d kd = u. Proposition.5. In Gal(Fpnd /Fpd ) = Fd . For d Fd : Fp −→ Fp . notice that the composition power Fn = Fd ◦ · · · ◦ Fd (with n factors) satisﬁes d Fn (t) = tp d Fn d nd =t for all t ∈ Fpnd . Hence in the automorphism group AutFpd (Fp ). The trace and norm mappings For an extension of Galois ﬁelds Fpnd /Fpd . But this can only be true if k = n. (ii) If k 1. 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 (n−1)d d 2d (n−1)d dk = u + Fd (u) + F2d (u) + · · · + F(n−1)d (u).25. Frobenius mappings exist on the algebraic closure Fp .3.3. For u ∈ Fpnd we have Fk (u) = up . d d 5.
5. N(u) ∈ Fpd . For t ∈ Fpd we have tp = t. pd p2d u u p3d ···u p(n−1)d 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 . TFpnd /Fpd (u) ∈ Fpd . There is a multiplicative version of this construction. 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). GALOIS EXTENSIONS FOR FIELDS OF POSITIVE CHARACTERISTIC 3 F2 l F lll ll lll lll ll lll 4 6 F F ll lll lll l lll lll Gal(Fp24 /Fp ) = F ∼ Z/24 = F8 F12 F24 = {id} Figure 5. Clearly TrFpnd /Fpd is additive. The relative trace TrFpnd /Fpd is a surjective Fpd linear mapping and whose kernel is an Fpd vector subspace of dimension n − 1. pnd pd NormFpnd /Fpd (u) = uup up · · · up d 2d (n−1)d . If we modify TFpnd /Fpd to have codomain Fpd . TrFpnd /Fpd (u) = u + up + up + · · · + up d 2d (n−1)d . TrFpnd /Fpd is surjective since its codomain has dimension 1.68 5. The relative norm NormFpnd /Fpd is a surjective group homomorphism. 5. The subgroups of the Galois groups of Fp24 /Fp So by Proposition 5. Proposition.24(i). we obtain the relative trace TrFpnd /Fpd : Fpnd −→ Fpd . . = up up up · · · up = uu u = N(u).26. Proposition. To see that TrFpnd /Fpd is surjective.24(i). Proof.2. notice that TrFpnd /Fpd (u) = 0 if and only if u is a root of the polynomial d 2d (n−1)d X + Xp + Xp + · · · + Xp ∈ Fpd [X] which has degree p(n−1)d and so has at most p(n−1)d < pnd roots in Fpnd . By redeﬁning the codomain we obtain the relative norm NormFpnd /Fpd : F× −→ F× .27. This means that ker TrFpnd /Fpd cannot be the whole of Fpnd .
(b) X 8 − 1 ∈ F5 [X]. 5. Let p > 0 be a prime. Let w ∈ F× be a primitive root.] (b) If d = 1. 5. (c) X 8 − 1 ∈ F11 [X]. Deduce that if t ∈ Fq then the set t − Σq = {t − u2 ∈ Fq : u ∈ Fq } has t − Σq  = (q + 1)/2 elements. then so is u + t for every t ∈ Fp .3. (a) If the polynomial ga (X) is irreducible over K. If pd < d. Show that the number of elements of Σq is Σq  = (q + 1)/2.2. and d  ϕ(pd − 1). let d ga (X) = X p − X − a ∈ K[X]. Is λq always q q surjective? (b) Consider the set of all squares in Fq .6. show that the splitting ﬁeld E of ga (X) over K is separable and Gal(E/K) ∼ Fpd . 5. 5. Let f (X) ∈ Fpd [X] be an irreducible polynomial with deg f (X) = n. Show that there is a unique group homomorphism λq : F× −→ {±1} which is characterized by the requirement q that for every u ∈ F× . Deduce that degFp w = d / p 5. (c) If K is a ﬁnite ﬁeld and d > 1. [Hint: show that if u ∈ E is a root of ga (X) = in an extension E/K. and K/Fpd is an extension. so pnd − 1  ker NormFpnd /Fpd  1 + pd + · · · + p(n−1)d = d . d 1 and write q = pd . the splitting ﬁelds of f (X) and g(X) over Fpd agree. show that w ∈ F× .7. Let p be an odd prime. For a ∈ K. −1} as a group under multiplication. (a) Consider {±1} = {1.1.4.5. Find the smallest Galois ﬁelds containing all the roots of the following polynomials.20 if we try to take p = 2. p −1 Hence pnd − 1  im NormFpnd /Fpd  = pd − 1. Suppose that d 1. Find the splitting ﬁeld of f (X). Σq = {u2 ∈ Fq : u ∈ Fq } ⊆ Fq . we also have pd  im NormFpnd /Fpd  therefore im NormFpnd /Fpd = F× . What happens to Theorem 5. Show that Proposition 5. λq (u) = 1 if and only if u = v 2 for some v ∈ F× . (d) X 8 − 1 ∈ F2 [X].13 also applies to an integral domain in place of a ﬁeld.69 Proof. Multiplicativity is obvious. . explain why ga (X) can never be irreducible over K.  ker NormFpnd /Fpd  Since im NormFpnd /Fpd F× . pd pd − 1. Deduce that for any other irreducible polynomial g(X) ∈ Fpd [X] with deg g(X) = n. 5. show that ga (X) is irreducible over K if and only if it has no root in K. in each case ﬁnd a primitive root of this Galois ﬁeld: (a) X 8 − 1 ∈ F41 [X]. 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]. Exercises on Chapter 5 5.
(d) Deduce that the equation x2 + y 2 + z 2 = 0 has at least one nontrivial solution in Fq . (e) What can you say about the case p = 2? . GALOIS EXTENSIONS FOR FIELDS OF POSITIVE CHARACTERISTIC (c) If t ∈ Fq . Deduce that every element of Fq is either a square or can be written as the sum of two squares.70 5. show that Σq ∩ (t − Σq ) 1.
Then any root u of p(X) in the algebraic closure C is algebraic over R. Cyclotomic extensions We begin by discussing the situation for cyclotomic extensions over Q using material discussed in Section 1. so irreducibility implies that v has degree 1 over R and therefore E P = R.n−1 gcd(t. The splitting ﬁeld of p(X) over C is contained in the splitting ﬁeld E of minpolyR. We must show that  Gal(E/C) = 1.1. But from known properties of C (see Proposition 3. 6. we have  Gal(E/R) . t=1. This shows that Gal(E/R) = P . 6. so suppose not. So we cannot have an irreducible quadratic polynomial in C[X]. Let p(X) ∈ C[X] be irreducible. Proof. Let ζn = e2πi/n . From the theory of 2groups.. hence Gal(E/R) is a 2group.u (X). Complex Analysis and Algebraic Geometry.75 allows us to write E P = R(v) for some v whose minimal polynomial over R must also have odd degree. 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). But by the Intermediate Value Theorem. Historically.1. Galois Theory has always been an important tool in Number Theory and Algebra. it was claimed that the irreducible polynomial over Q which has ζn as a root was the nth cyclotomic polynomial t Φn (X) = (X − ζn ). our proof only uses the connectivity of the real line (via the Intermediate Value Theorem) together with explicit calculations in C involving square roots.43. It is interesting to compare the proof below with others which use the topology of the plane and circle or Complex Analysis. so C/R is algebraic. we have 2  [E : R] and so 2   Gal(E/R). A proof of the Fundamental Theorem of Algebra We will prove the Fundamental Theorem of Algebra for the complex numbers C.. Ring Theory and such diverse areas as Differential Equations.3. As C/R is a Galois extension. For the ﬁxed subﬁeld of P .2.CHAPTER 6 A Galois Miscellany In this chapter we will explore some miscellaneous topics in Galois Theory. Therefore  Gal(E/C) = 1 and E = C. This proof is essentially due to Gauss but he did not use the historically more recent Sylow theory.29).u (X)(X 2 + 1) over R. [E P : R] = P  which shows that E P /R has odd degree. Many of the ideas introduced in this chapter are of great importance in these and other mathematical areas. stimulating the development of subjects such as Group Theory. 6. there is a normal subgroup N Gal(E/C) of index 2. Now consider a 2Sylow subgroup P Gal(E/R) and recall that  Gal(E/R)/P  is odd. In Theorem 1. the standard primitive nth root of 1 in C. Theorem (The Fundamental Theorem of Algebra).. The ﬁeld of complex numbers C is algebraically closed and R = C. The Primitive Element Theorem 3. so in C[X] we have p(X)  minpolyR. Since C E. so we can consider the Galois extension E N /C of degree 2.. every quadratic aX 2 + bX + c ∈ C[X] has complex roots (because we can ﬁnd square roots of every complex number). every real polynomial of odd degree has a real root.n)=1 71 .
. p 1 f (ζn1 ) r 1 = 0 in Z[ζn ].ζn (X) = minpolyQ. Theorem. n) = 1. Let p > 0 be a prime which does not divide n. n − 1. Consider r Z(ζn ) = {a0 + a1 ζn + · · · + ar ζn : r 0. this is a cyclic subgroup of order n. n) = 1. where the p are primes with 2 p1 < · · · < pm and rj 1 with. Let P Z(ζn ) be a maximal ideal which contains p.ζn (ζn1 p r1 r2 p2 )=0 . therefore Φn (X) = minpolyQ. To do this we will show that every power ζn as above is actually a Galois conjugate of ζn over Q. . say Fpd for some d. we ﬁnd that ζ n is conjugate to ζ n1 . Since the complex roots of Φn (X) are the powers ζn with t = 1. By elementary Group Theory we can assume that r  n and so p r. . π gives an injective r group homomorphism. This shows that minpolyQ. To see this. n − 1 and gcd(t. ζn Z(ζn )× . Now consider a maximal ideal P1 Z[ζn ] containing p1 . Since t = p1 · · · pm j gcd(t. • [Q(ζn ) : Q] = ϕ(n). t tp F(ζ n ) = (ζ n )p = ζn . so ζn − 1 ∈ P or r ∈ P since maximal ideals are prime.ζn (X) t for some monic polynomial f (X) ∈ Z[X] and f (ζn ) = 0. But Z ∩ P = (p) and so r ∈ P . π : ζn −→ F× . Then • Q(ζn ) = Q[X]/(Φn (X)). The main step in the proof has the above properties and so ζn t is to show that Φn (X) ∈ Z[X] is irreducible. Writing u = π (u). it is a ﬁnite ﬁeld. Let n 2. Then Z(ζn ) is a subring of Q(ζn ) and so is an integral domain. 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. we ﬁnd that p 1 f (ζn1 ) r 1 p r1 = 0 and so conjugate to ζn . Now let t = 1. n) = 1 we also have pj n s. we see that n ∈ P and hence p  n. . 2. then ζn − 1 ∈ P . Consider the prime power factorization r1 rm . On factoring we have r−1 (ζn − 1)(ζn + · · · + ζn + 1) ≡ (ζn − 1)r (mod P ). aj ∈ Z} ⊆ Q(ζn ). we can consider the eﬀect of the absolute Frobenius map F : Fpd −→ Fpd t on ζ n = ζn . hence / ζn − 1 ∈ P . suppose that π (ζn ) = 1 for some r = pd r 1. . . A GALOIS MISCELLANY 6. Suppose there is a factorization t k t t t Φn (X) = f (X) minpolyQ. then the quotient ring Z(ζn )/P is a ﬁeld of characteristic p. ζ n is conjugate to ζn .72 6. by iterating this we ﬁnd tp that ζ n is conjugate to every power of the form ζn . thus contradicting our original assumption on 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. So π is injective. Recalling that n−1 ζn + · · · + ζn + 1 = 0. Let π : Z(ζn ) −→ Fpd be the quotient homomorphism. By separability and the fact that the reduction map π1 : Z[ζn ] −→ Fpd1 is injective on the powers of ζn . Reducing modulo P1 and working in the resulting extension Fpd1 /Fp1 .2. . tp This shows that in the Galois extension Fpd /Fp . . = t Proof. . n − 1 and t gcd(t. . Its group of units contains the cyclic subgroup ζn of order n.ζn (X) t and hence Φn (X) is irreducible. . In fact. t • Gal(Q(ζn )/Q) ∼ (Z/n)× . . Inside the group of units of Z(ζn ) is the subgroup of powers of ζn . where the element tn ∈ (Z/n)× acts on Q(ζn ) by tn · ζn = ζn . We claim that when restricted to ζn .
hence Q( 2. Then Q(ζn )R = Q(ζn ). .2. ( )} = Gal(Q(cos(2π/n))/Q). Notice that ζn is a root of the polynomial X 2 − 2 cos(2π/n)X + 1 ∈ Q(cos(2π/n))[X]. Q(ζn )R = Q(ζn ) and [Q(cos(2π/n)) : Q] = Proof.43 we have [Q(ζ24 ) : Q] = 8. It is easy to check that √ √ [Q( 2. Recall that ϕ(n) . which implies that Using this we ﬁnd that √ √ Q(ζ24 ) = Q( 2.6. Example. 2 1 1 −1 cos(2π/n) = (ζn + ζ n ) = (ζn + ζn ). . Making use of the identities eθi = cos θ + sin θ i. we obtain 1 cos θ = (eθi + e−θi ). i). 3. n − 1. = .ζn (ζn ) = 0. . for each j = 1.3. The residue class of r acts by sending cos(2π/n) to cos(2πr/n). m we have minpolyQ. We have Proof. When j = m. Gal(Q(ζn )/Q) ∼ Z/n× . so we also have (6. ζ24 = + 2 2 2 2 √ √ and all of these numbers are in Q(ζ24 ). 2 ( ) = Q(ζn + ζ n ) = Q(cos(2π/n)). this shows that minpolyQ. consider the cyclotomic extension Q(ζn )/Q where ζn = e2πi/n . Hence ( ) = {id. 2. [Q(ζ24 ) : Q] = ϕ(24) = 8 and Gal(Q(ζ24 )/Q) ∼ Z/2 × Z/2 × Z/2. and so [Q(cos(2π/n)) : Q] = ϕ(n)/2. . Also. Complex conjugation corresponds to the residue class of −1 ≡ n − 1 (mod n).4. Hence t is conjugate to ζ in the extension Q(ζ )/Q. it is elementary to show that cos(2πr/n) = cos(2π/n) unless r ≡ 1 (mod n). CYCLOTOMIC EXTENSIONS 73 and so ζn1 p r1 r2 p2 is conjugate to ζn . ζn n n r r p11 ···pj j 6. By Theorem 1. Thus we have Q(ζn ) ( ) = Q(cos(2π/n)). 3. i) Q(ζ24 ). ζ24 = i.ζn (X) = X 2 − 2 cos(2π/n)X + 1. ζ24 = − + i.1) minpolyQ(cos(2π/n)). i) : Q] = 8. . Furthermore. 3. For n > 2. Continuing in this fashion. Theorem. 2 2 Complex conjugation ﬁxes each of the real numbers cos(2πk/n) for k = 1. Gal(Q(ζ24 )/Q) ∼ Z/2 × Z/2 × Z/2. .ζn (ζn1 p r r1 r2 p2 ···pj j )=0 t and so ζn is conjugate to ζn . . = 6. √ √ 1 1 3 3 8 6 3 i. . = r where the residue class of r acts by sending ζn to ζn .
. T7 (X) = 64X 7 − 112X 5 + 56X 3 − 7X. Since √ 3 .74 6. T4 (X) = 8X 4 − 8X 2 + 1. hence since Φ24 (X) = X 8 − X 4 + 1. A GALOIS MISCELLANY We also have cos(2π/24) = cos(π/12) ∈ Q(ζ24 ). i. Note that case (i) of Kaplansky’s Theorem 4.5. 16X 4 − 16X 2 + 1 = 16 minpolyQ. when n > 2. and so 16 cos4 (π/12) − 16 cos2 (π/12) + 1 = 0. Gal(Q(ζ24 )/Q) has 23 − 1 = 7 subgroups of each of the orders 2 and 4. Similarly. 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]. −4 Then after multiplying by ζ24 we have −4 4 ζ24 − 1 + ζ24 = 0. The minimal polynomial for cos(π/12) can also be found as follows. Now −1 −4 −2 4 2 (ζ24 + ζ24 )4 = (ζ24 + ζ24 ) + 4(ζ24 + ζ24 ) + 6. 6.cos(π/12) (X). These form a system of orthogonal polynomials which can be computed in Maple using the command orthopoly[T](n. T5 (X) = 16X 5 − 20X 3 + 5X. T6 (X) = 32X 6 − 48X 4 + 18X 2 − 1. For this example.X). Remark. This method will work for any n where ϕ(n) is even. it is an interesting exercise to ﬁnd them all together with their ﬁxed subﬁelds. giving −4 4 (ζ24 + ζ24 ) − 1 = 0.28 applies to the polynomial minpolyQ. we obtain 8 4 ζ24 − ζ24 + 1 = 0. T3 (X) = 4X 3 − 3X. Here are the ﬁrst few of these polynomials: T2 (X) = 2X 2 − 1.e. 6. cos(2π/12) = cos(π/6) = 2 √ 3 2 cos (π/12) − 1 = 2 2 we have and so giving Then 3 4 cos4 (π/12) − 4 cos2 (π/12) + 1 = .cos(π/12) (X). Remark. Combining these we have −1 −1 (ζ24 + ζ24 )4 − 4(ζ24 + ζ24 )2 + 1 = 0. We have Φ24 (ζ24 ) = 0.6. hence −4 −1 −2 4 2 ζ24 + ζ24 = (ζ24 + ζ24 )4 − 4(ζ24 + ζ24 ) − 6. so −2 −1 2 ζ24 + ζ24 = (ζ24 + ζ24 )2 − 2.. 4 16 cos4 (π/12) − 16 cos2 (π/12) + 1 = 0. −1 −2 2 (ζ24 + ζ24 )2 = ζ24 + ζ24 + 2.
. Theorem (Artin’s Theorem). . X 4 − 1 splits over F5 . K× 6. this Galois group only depends on the largest subﬁeld of K which is algebraic over Fp . Example. When p = char K > 0.9. 6. X 4 − 1 is separable over F3 (T ) and has irreducible factors (X − 1).21. (X + 1) and (X 2 + 1). χα : K × −→ K × is a character of in K. Then χ1 . Given an automorphism α : K −→ K. Given any ring homomorphism ϕ : R −→ K we obtain a character of R× in K by restricting ϕ to a map χϕ : R× −→ K × . = 6. χn be characters of a group G in a ﬁeld K. Example. 6.14. Let E/K be a Galois extension and σ ∈ Gal(E/K). we know that Φn (ζ) = 0. . Artin’s Theorem on linear independence of characters Let G be a group and K a ﬁeld.ζ (X) ∈ K[X] are primitive roots of 1. However. (ii) By Proposition 5.6. . = (ii) By Proposition 5. . In either case it can happen that Φn (X) factors in K[X]. . This generator clearly extends to an automorphism of F9 (T ) which ﬁxes T . . So X n − 1 splits over K(ζ) and each element α ∈ Gal(K(ζ)/K) has the action α(ζ) = ζ rα .15. In this deﬁnition. Theorem. 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. hence it is abelian with order dividing ϕ(n).13. where gcd(rα . For example. . . . hence the roots of minpolyK. where ζ 2 + 1 = 0. Definition. ARTIN’S THEOREM ON LINEAR INDEPENDENCE OF CHARACTERS 75 Now let K be a ﬁeld with characteristic char K n.7. 6. Proof. We have the following splitting ﬁelds and Galois groups. tn ∈ K. . .12. Let χ1 . so (X 2 + 1) splits over F9 (T ). = = with generator σ satisfying σ(ζ) = ζ −1 = −ζ. Example. the functional equation means that for all g ∈ G. Example.3. The polynomial Φn (X) has integer coeﬃcients. Also. .20. 6. χn are not linearly independent then they are linearly dependent. then the splitting ﬁeld of X n − 1 over K is K(ζ). .3. The precise outcome can be determined with the aid of Proposition 5. The Galois group Gal(K(ζ)/K) is isomorphic to a subgroup of (Z/n)× . χn be distinct characters of a group G in a ﬁeld K. where ζ ∈ K is a primitive nth root of unity. 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).11. . Remark. t1 χ1 (g) + · · · + tn χn (g) = 0. .10. . 6. 6. .8. A group homomorphism χ : G −→ K × is called a character of G with values in K. .20. Let χ1 . . .20. 6. . The splitting ﬁeld of (X 2 + 1) over F3 is F9 = F3 (ζ). t1 χ1 + · · · + tn χn = 0 =⇒ t1 = · · · = tn = 0. χn are linearly independent if for t1 . . 6. we can still describe the splitting ﬁeld of X n − 1 over K and its Galois group. X 4 − 1 splits over F5 (T ) and the Galois group Gal(F5 (T )/F5 (T )) is trivial. If char K n. Then χσ : E × −→ E × is a character. (i) By Proposition 5. Definition. Working in K. n) = 1. . . Gal(F9 /F3 ) ∼ (Z/4)× ∼ Z/2. Then χ1 . If χ1 . χn are linearly independent. Proof. (i) The splitting ﬁeld of X 4 − 1 over F3 (T ) is F9 (T ) and ∼ Gal(F9 (T )/F3 (T )) = (Z/4)× ∼ Z/2. if K = Fpd (T ) then the value of d is the crucial factor.
there must be an element g0 ∈ G for which χ1 (g0 ) = χ2 (g0 ).2) by χ1 (g0 ) and subtracting gives t2 (χ2 (g0 ) − χ1 (g0 ))χ2 + t3 (χ3 (g0 ) − χ1 (g0 ))χ3 + · · · + tk+1 χk+1 = 0. . NE/K : E × −→ K × . . in which the coeﬃcient t2 (χ2 (g0 )−χ1 (g0 )) is not zero. χr+1 . . . . χr−1 . Suppose that E/K is a ﬁnite Galois extension with cyclic Galois group Gal(E/K) = σ of order n.16. . . Hence. NE/K is called the norm mapping for E/K and generalizes the norm mapping for ﬁnite ﬁelds of Section 5.17. . We proceed by induction on n. Corollary. . Then there is a z ∈ K for which t1 α1 (z) + · · · + tn αn (z) = 0. αn apart from its order. Corollary. For each u ∈ E × . . . Multiplying Equation (6. For n = 1 the result is easily veriﬁed. . . Hence the associated Klinear transformation TrE/K : E −→ K. say tr = 0. again contradicting the inductive assumption. . tn ∈ K be a sequence of elements. Suppose that α1 . . χk+1 is linearly independent. . Hence all of the ti must be nonzero. For the inductive assumption. . not all of which are 0. . δE/K (u) = uσ(u)−1 . αn are distinct automorphisms of the ﬁeld K. . . . α1 (x) + · · · + αn (x) ∈ E Gal(E/K) = K. 6. Hence the Klinear transformation t1 α1 + · · · + tn αn : K −→ K is nontrivial.2) t1 χ1 + · · · + tk+1 χk+1 = 0. χk+1 is linearly dependent. . Let χ1 . Then the function α1 + · · · + αn : E −→ E is a nontrivial Klinear transformation whose image is contained in K. 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. As χ1 = χ2 . Notice that for u ∈ E × . . 6. χk+1 be a set of k + 1 distinct characters for which there are t1 . . Let t1 . hence in uσ(u) · · · σ n−1 (u) ∈ E σ = K. .76 6.17. . TrE/K (x) = α1 (x) + · · · + αn (x) is surjective. . contradicting the inductive assumption. Proof. χk+1 is linearly dependent. αn be the distinct elements of Gal(E/K). So χ1 .2) applied to g0 g yields t1 χ1 (g0 g) + · · · + tk+1 χk+1 (g0 g) = 0. A GALOIS MISCELLANY Proof. since the list γα1 . . First note that for x ∈ E and γ ∈ Gal(E/K). γ(α1 (x) + · · · + αn (x)) = γα1 (x) + · · · + γαn (x) = α1 (x) + · · · + αn (x). So for all g ∈ G. Let E/K be a ﬁnite Galois extension of degree n and let α1 . .3. Now using this we deﬁne a group homomorphism NE/K (u) = uσ(u) · · · σ n−1 (u). . . . There is another homomorphism δE/K : E × −→ E × . . and therefore since χj (g0 g) = χj (g0 )χj (g). . . suppose that it holds for any n k. . . . . NE/K (δE/K (u)) = (uσ(u)−1 )(σ(u)σ 2 (u)−1 · · · σ n−1 (u)σ n (u)−1 ) = 1. . then χ1 . tk+1 ∈ K not all zero and such that (6. which demonstrates the inductive step. . The function TrE/K : E −→ K is called the trace mapping of E/K. . Equation (6. . The rest of the statement follows directly from Corollary 6. . Hence χ2 . . If one of the ti is zero. γαn is the same as α1 . we see that t1 χ1 (g0 )χ1 + · · · + tk+1 χk+1 (g0 )χk+1 = 0.
uk ) where un = 1 a1 . . where u is a root of f (X). u)/K(ζ)) is cyclic and Gal(K(ζ)/K)) ∼ Gal(K(ζ. SIMPLE RADICAL EXTENSIONS 77 since σ n (u) = u. Theorem (Hilbert’s Theorem 90). . This cannot be identically zero. The Galois group Gal(K(u)/K) is cyclic of order n with a generator σ for which σ(u) = ζu. . Proposition. K(ζ.19. where Gal(K(ζ. Then the splitting ﬁeld of f (X) over K has the form K(u. . .4. where u is a root of f (X) and ζ is a primitive nth root of 1. ζ). u) a Gal(K(ζ. Let u ∈ ker NE/K .20. The Galois Correspondence identiﬁes the following towers of subﬁelds and subgroups. Then L/K is a simple nKummer extension if L = K(u) where un = a for some a ∈ K.27. the element v = w + uσ(w) + uσ(u)σ 2 (w) + · · · + uσ(u) · · · σ n−2 (u)σ n−1 (w) is nonzero. . 6. un = ak for some elements a1 . Simple radical extensions In this section we will investigate splitting ﬁelds of polynomials of the form X n − a. Let E/K be a ﬁnite Galois extension with cyclic Galois group Gal(E/K) = σ of order n. Let f (X) = X n − a ∈ K[X] be irreducible and separable over K.19. ker NE/K .6. u)/K(ζ)) = is abelian. . ak ∈ K. since Thus we have u = vσ(v)−1 uσ(u)σ 2 (u) · · · σ n−1 (u)σ n (w) = w. . . . . Then im δE/K = ker NE/K . n−1 are distinct and linearly independent by Artin’s Theorem 6. Consider the function id +uσ + uσ(u)σ 2 + · · · + uσ(u) · · · σ n−2 (u)σ n−1 : E × −→ E.21.15. Let K be a ﬁeld with char K n and which contains a primitive nth root of 1. 6. In the more general situation of Proposition 6.33 we introduce a more general notion of radical extension.4. Proof. {id} Gal(K(ζ. then the splitting ﬁeld of f (X) = X n − a over K has the form K(u). u)/K(ζ)) 8 ' K {id} 6. . ζ say. . 6. u)/K(ζ)) Gal(K(ζ. then there is a v ∈ E × such that u = vσ(v)−1 . if u ∈ E × and u σ(u) · · · σ n−1 (u) = 1. . where char K n. Our next result is an important generalization of 6. Explicitly. The characters σ k : E × −→ E × with k = 0. as required. k Note that in this deﬁnition we do not require the polynomials X n − aj ∈ K[X] to be irreducible. u)/K) 7 K(ζ) f } Gal(K(ζ. . . 1.18. . ζ. so for some w ∈ E. L/K is an (iterated ) nKummer extension if L = K(u1 . 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. If K contains a primitive nth root of 1. We call these simple radical extensions and later in Deﬁnition 6. Definition. Corollary. So im δE/K Proposition 5. u)/K))/ Gal(K(ζ. u)/K).
In fact. Let K(u)/K be a simple nKummer extension. Gal(K(u)/K) is cyclic since every subgroup of a cyclic group is cyclic. −1} Here is a converse to Proposition 6. Let n 1 and q ∈ Q. hence  Gal(K(u)/K) divides n. hence an ∈ K since it is ﬁxed by σ. A GALOIS MISCELLANY 6. this shows that [E : K] = n i . Hence the function ρ : Gal(K(u)/K) −→ ζ . and so rβα = rα + rβ . This means that K(u)/K is Galois.18. n − 1. Proposition. so they must be the n conjugates of a. Example. Since every element of Gal(K(u)/K) is determined by its eﬀect on u. whence K(a) = E. If E/K is a ﬁnite Galois extension with cyclic Galois group of order n. is a group homomorphism. Proof. 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. ρ(α) = ζ rα . Clearly the roots of X n − a are distinct and so K(u)/K is separable over K. Suppose that char K n and there is an element ζ ∈ K which is a primitive nth root of unity. Then in K[X] we have X n − a = (X − u)(X − ζu) · · · (X − ζ n−1 u). .24. √ The corresponding group homomorphism ρ : Gal(Q(i)( 2)/Q(i)) −→ i has image im ρ = {1. . 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)). Proposition. E. Then K(u)/K is a Galois extension and Gal(K(u)/K) is cyclic with order dividing n. but X 4 − 4 = (X 2 − 2)(X 2 + 2).22.22. √ √ 6. Lagrange’s Theorem implies that the image of ρ has order dividing n. √ Proof. . 1. 6. there is an element a ∈ E for which ζ −1 = aσ(a)−1 . . n q)/Q(ζn ) is a simple nKummer extension. . We have ( 2)4 − 4 = 0.23. n − 1 are distinct. Hence E/K is a simple nKummer extension.√2 (X). so by Hilbert’s Theorem 6. βα(u) = β(ζ rα u) = ζ rα β(u) = ζ rα ζ rβ u = ζ rα +rβ u. Then Q(ζn . .78 6. and X 2 − 2 = minpolyQ(i). Q(i.25. Proof. . √ 6. K(u) is a splitting ﬁeld of X n − a over K. ρ is injective. Then σ(a) = ζa and the elements σ k (a) = ζ k a for k = 0. 1 . . in fact. Example. 2)/Q(i) is a simple 4Kummer extension with Gal(Q(i. As ζ is cyclic of order n. 2)/Q(i)) cyclic of order 2. Notice that for β ∈ Gal(K(u)/K). Suppose that un = a ∈ K. We have NE/K (ζ −1 ) = ζ −n = 1. For each α ∈ Gal(K(u)/K) we have α(u) = ζ rα u for some rα = 0. Since K(a) n = [K(a) : K] and therefore [K(a) : K] = [E : K] = n.
In the opposite direction we can sometimes see that a group is insolvable. A group G is solvable. (ii) If N G and G/N are solvable then so is G. Example.34. L/K is a radical extension of K if it has the form L = K(a1 . Let G be a group. . Proposition. . Let E/L be an extension where E is a splitting ﬁeld for the polynomial minpolyK. An is a simple group and An Sn with Proof.5. Thus every element of L is expressible in terms of iterated roots of elements of K. Proposition. a2 . SOLVABILITY AND RADICAL EXTENSIONS 79 6. . Solvability and radical extensions We begin by recalling some ideas about groups. Then G is insolvable if any of the following conditions holds: (i) G contains a subgroup which is a nonabelian simple group (or has a quotient group which is a nonabelian simple group). an ) with adk ∈ K(a1 . Example. the alternating and symmetric groups An and Sn are insolvable. Let G be a ﬁnite pgroup. We will need the following Lemma and its Corollary. If each composition factor is a cyclic group of prime order the subnormal series is called a composition series. we assume also that char K = 0. Definition. only their order varying for diﬀerent composition series. (ii) G has a quotient group which is a nonabelian simple group.28. so beware when reading other sources! 6. Then E/L is a radical Galois extension. ak−1 ) k for some dk 1.u (X) over K.29. (i) If G is solvable then every subgroup H G and every quotient group G/N is solvable. 6. In fact. if L/K is a radical Galois extension then so is E/K. . Lemma.31. so in this case we can assume each quotient Gk /Gk+1 is cyclic of order p. This follows from the fact that if n quotient group Sn /An ∼ Z/2. For n 5. (iii) G has a composition series in which one of the terms is a nonabelian simple group. Example. 5. add extra terms) a subnormal series of a solvable group to obtain a composition series.. For simplicity. It is a standard result that we can always reﬁne (i. 5] for further details. . Let G be a ﬁnite abelian group.5.30. .32. 6. Let K be a ﬁeld and L/K a ﬁnite extension. 6. there is always a normal subgroup of a pgroup with index p. . where p is a prime. a2 . 6. = Now we explain how this relates to ﬁelds and their extensions. 6. soluble or soluable if there is a chain of subgroups (called a subnormal series) {1} = G {1} = G G G −1 ··· G1 G0 = G in which Gk+1 Gk and each composition factor Gk /Gk+1 is abelian. 6.27. Recall that a group is simple if it has no nontrivial proper normal subgroups. Let G be a ﬁnite group. Then G is solvable. Then G is solvable. see [3. A group which is not solvable is called insolvable. Definition. In particular.6. . for a ﬁnite pgroup G. several text books make subtle errors or omissions related to this result.26.e. The primes appearing as well as the number of times each occurs are all determined by G. 6. According to [4]. 6. Remark. we usually write −1 · · · G1 G0 = G.33. . Let L/K be a ﬁnite Galois extensions and let L(u)/L be a radical extension.
. . an ) as in Deﬁnition 6. Also. . as is E /K.35.80 6. . there is a primitive dth root of unity ζ ∈ E and the subﬁeld L(ζ. which is a radical Galois extension of K. an ). Clearly Q(ζ3 .37. . . so it is solvable by Proposition 6. 6. .u (X) in E. Writing L = K(a1 . . Now Gal(E /E) Gal(E /K) and by Theorem 6. ut ). . . Gal(E /E) is abelian. 2 ζ3 . then E = L(ζ. √ We know that Gal(Q(ζ3 . . Hence E /K(ζ) is radical. . L/K is solvable if L of K. But L(ζ. Proof. an . . However. u. Corollary.30.25. But this is clearly a radical extension of L. this is proved by induction on n using Lemma 6. Clearly K(ζ)/K is radical. 3 2) is a radical extension of Q and √ √ 3 3 Q(ζ3 . u1 . L where L /K is a ﬁnite radical Galois extension 6. a2 . u. Suppose that E E where E /K is a radical Galois extension. 6. u1 . . .30 and 4. u)/L is a radical Galois extension. each extension (E )Gk+1 /(E )Gk is radical. The Galois group of the extension Q(ζ3 . u)/K need not be Galois. Proof. 3 2)/Q is solvable. . In the next deﬁnition. Let E be the splitting ﬁeld of X n − 1 over E. then f (X) is solvable by radicals over K if L is contained in a radical (Galois) extension of K. ut ∈ E are the distinct roots of minpolyK. .36. . hence L(ζ. the word Galois is superﬂuous because of the preceding results. then E = L(a1 . where we identify each element of the Galois group with = √ a permutation of the three roots of X 3 − 2 in Q(ζ3 . Definition.33. If L is the splitting ﬁeld of a polynomial f (X) ∈ K[X]. Let E/K be a ﬁnite Galois extension. 2) = Q(ζ3 )( 2). ζ. Let {1} = G G −1 · · · G1 G0 = Gal(E /K(ζ)) be a composition series. .22. . Gal(E /K)/ Gal(E /E) ∼ Gal(E/K) which is = solvable.30. 6. A GALOIS MISCELLANY Proof. By Proposition 6. Now suppose that Gal(E/K) is solvable and let n =  Gal(E/K). 3 2)/Q) ∼ S3 . If L/K is a radical Galois extension. If L/K is a radical extension then it is contained in a radical Galois extension L /K. Suppose that ud ∈ L. 3 2) which we list in the order √ √ √ 2 3 3 3 2. u) E is normal over L. √ 6.20.34. Example. . Then Gal(E /K(ζ)) Gal(E /K) is solvable. . Similarly. We have already studied this extension in Example 3. Then E/K is solvable if and only if the group Gal(E/K) is solvable. √ Proof. ut ). Definition. We will now show that E /K is a radical extension.7.39. so Gal(E /K) is solvable by Proposition 6. 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 Gal(E/K) is a quotient group of Gal(E /K). . if u = u1 .38. The extension (E )G1 /K(ζ) is radical by Proposition 6. . say L = K(a1 . . 2 ζ3 . Theorem.
Suppose that H (1 2 · · · n).26. 5 divides the order of Gal(E/Q). Then H = Sn . The next lemma shows that Gal(E/Q) ∼ S5 . It is also interesting to reverse the question and ask whether there are extensions which are not solvable. we ﬁnd that f (X) is irreducible over Q.40.40. Furthermore. u4 . so by Cauchy’s Lemma this group contains an element of order 5. = = so we have the following composition series for S3 : {id} A3 S3 . = hence it is not solvable. 6. u2 being the nonreal roots. SOLVABILITY AND RADICAL EXTENSIONS 81 We have the following towers of subﬁelds and subgroups related under the Galois Correspondence. u3 . Then f (0) = 7 > 0 > f (28) = −4302585. Notice that A3 ∼ Z/3 and S3 /A3 ∼ Z/2. There has been a great deal of research activity on this question in the past few decades. 3 2)A3 i 2 A3 = Gal(Q(ζ3 . The splitting ﬁeld of the polynomial f (X) = X 5 − 35X 4 + 7 ∈ Q[X] is not solvable. √ 5 S3 Q(ζ3 . we ﬁrst recall that the smallest nonabelian simple group is A5 which has order 60.41. 6. Sn and H contains the elements (1 2) and The proof is left as an exercise. Let n 1. Example. = 6. The Galois group of f (X) = X 5 +20X+16 over Q is Gal(Q(f (X))/Q) ∼ A5 . u2 . By Theorem 4. 3 2) e 3 2 √ Q(ζ3 ) = Q(ζ3 . then the transposition (1 2) ∈ S5 corresponds to this element. Here is an example whose Galois group is A5 . Then complex conjugation restricts to an element of order 2 in Gal(E/Q) which interchanges the nonreal roots and ﬁxes the others. but apparently not all simple groups are known to occur as Galois groups of extensions of Q or other ﬁnite subextensions of C/Q. Now observe that f (X) = 5X 4 − 140X 3 = 5X 3 (X − 28). u5 with u1 . Example. . Using the Eisenstein Test 1. Here is an explicit example over Q.6.38 with p = 7. It is worth remarking that the most extreme version of this occurs when we ask for a Galois group which is simple.8(iii). 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 . namely a maximum at x = 0 and a minimum at x = 28. If we list the roots of f (X) as u1 . hence there are three real roots of f (X) and two nonreal complex ones. f (X) = 20X 4 − 420X 2 = 20X 2 (X − 21). 5 3 √ 3 2)/Q(ζ3 )) Q y ) {id} Here Q(ζ3 )/Q is itself a Galois extension and A3 S3 . this is veriﬁed using Proposition 4.5. Let E C be the splitting ﬁeld of f (X) over Q. the only elements of S5 of order 5 are 5cycles. To ﬁnd examples. 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. This completes the veriﬁcation of Example 6. There are two turning points. Lemma. This was a famous problem pursued for several hundred years. Proof.42.
The ﬁeld of symmetric functions on n indeterminates is Symn (k) = k(X1 . . Remark. Now Sn Gal(k(X1 . Combining these inequalities we obtain  Gal(k(X1 . . Hence every element of Symn (k) is a rational function in the ei . . . . . eσ = ek . . Let p > 0 be a prime and G a group of order G = pn for some n 1. Xn )/ Symn (k) is not solvable. . Xn ] is a ring isomorphism. Xn ) ∈ Symn (k) ⇐⇒ ∀σ ∈ Sn f (X1 . .2. . . . . . . Xn ). Xn )/ Symn (k)) ∼ Sn . . . Varying σ we obtain actions of the group Sn on k[X1 . . 6. .1. . . . so ek ∈ Symn (k). 6. Xn )Sn So if f (X1 . . . . The extension k(X1 . . Xn ) −→ k(X1 . . Xn ) by ring isomorphisms ﬁxing k and in the latter case it is by ﬁeld automorphisms ﬁxing k. . . . . Xn ) : Symn (k)] =  Gal(k(X1 . . . . . Since k(e1 . Xn ] and its ﬁeld of fractions K = k(X1 . If n 5. . . Xn )/ Symn (k)) = Sn . . 6. hence the roots of this polynomial are the Xi . Xn ] ⊆ k(X1 . . . . .] 6. . . Xn )/ Symn (k)) permutes the roots of fn (Y ) and is determined by this permutation. . . . . . . Xn ) = f σ (X1 . Each permutation σ ∈ Sn acts on k[X1 . . . . .e.82 6. . . i. . . Viewed as a function σ· : k[X1 . . . Show by induction on n that there is a normal subgroup N G with N  = pn−1 .6. . Xn ). . . Xi1 Xi2 · · · Xik . But as every element of Gal(k(X1 . . . Xn ) is the splitting ﬁeld of fn (Y ) over Symn (k). Xn ] and k(X1 . Xn )/k(e1 . the extension k(X1 . . A GALOIS MISCELLANY 6. Let K be a ﬁeld for which char K = 2 and n 1 be odd. . . If K contains a primitive nth root of unity. . . Definition. this extends to a ring isomorphism σ· : k(X1 . . . . [Hint: what do you know about the centre of G? Use this information to produce a quotient group of smaller order than G. . .46. . . . . . . So k(X1 . . . k[X1 . . . . .45. . .44. . Analogous results are true for polynomials. . . we also have n!  Gal(k(X1 . . . . Xn )/ Symn (k)). . . . Xn )/ Symn (k) is a ﬁnite Galois extension for which Gal(k(X1 . Theorem. . . . Working in k the ring k(X1 . Corollary. . Xn ) ∈ k(X1 . ek = i1 <i2 <···<ik k(X1 . . . Xn ] −→ k[X1 . . . en ]. . Xn ) = f (Xσ(1) . . . . . . . . In fact. . Exercises on Chapter 6 6. Xn ) called elementary symmetric functions. . . . . en ) is Galois of degree n!. . . en ) Symn (k) we can also deduce that k(e1 . where 1 k n. . . . . Symmetric functions Let k be a ﬁeld. . Xn ] by σ · f (X1 . . Xn ). . It is easy to see that for every σ ∈ Sn . . . . There are elements of k[X1 . . . . . Xn )/ Symn (k)) = n! and therefore Gal(k(X1 . Xn )/ Symn (k)). . . . Xn ). . then f (X1 . . . . . . Consider the polynomial ring on n indeterminates k[X1 . . . . show that then K contains a primitive 2nth root of unity. = Proof. . Xσ(n) ).43. . . hence [k(X1 .. en ) = Symn (k). Xn )[Y ] we have fn (Y ) = Y n − e1 Y n−1 + · · · + (−1)n−1 en−1 Y + (−1)n en = 0. . . 6. . Xn ) = f (Xσ(1) . . this proof shows that the extension k(X1 . Xσ(n) ). . . Xn ]Sn = k[e1 . Xn )/ Symn (k)) Sn  = n!. .
83
6.3. Find all values of n in the following ﬁelds:
1 for which ϕ(n)  4. Using this, determine which roots of unity lie √ √ √ 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(ζ24 ) 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(ζ20 ) 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(ζ5 ) where ζ5 = e2πi/5 . (a) Describe the Galois group Gal(Q(ζ5 )/Q) and its action on Q(ζ5 ). (b) Determine the minimal polynomial of cos(2π/5) over Q. Hence show that √ −1 + 5 cos(2π/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(ζ5 ) and express them as ﬁxed ﬁelds of subgroups of Gal(Q(ζ5 )/Q). 6.7. In this question, let p be an odd prime and let ζp = e2πi/p ∈ Q(ζp ) (a) Consider the product
(p−1)/2
C.
ξ=
r=1
r −r (ζp − ζp ) ∈ Q(ζp ).
Show that ξ = (−1) (b) Deduce that ξ2 = (c) Conclude that ξ= p −p √ ± p √ ± pi
2 (p−1)/2
p−1 r (1 − ζp ). r=1
if p ≡ 1 (mod 4), if p ≡ 3 (mod 4).
if p ≡ 1 (mod 4), if p ≡ 3 (mod 4). √ √ and also p ∈ Q(ζp ) if p ≡ 1 (mod 4) and p i ∈ Q(ζp ) 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 TrE/K : E −→ K. (b) If v ∈ E has TrE/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 TrE/K t = 0, 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)
84
6. A GALOIS MISCELLANY
and adapt the proof of Hilbert’s Theorem 90 in Theorem 6.18, using TrE/K in place of NE/K .] 6.10. (a) For n 1 and 1 k n, the kth power sum sk ∈ k[X1 , . . . , Xn ]Sn is deﬁned by sk =
1 i n
Xik .
Prove the formula sk = e1 sk−1 − e2 sk−2 + · · · + (−1)k−1 ek−1 s1 + (−1)k kek . (b) For n 1 and 1 k n, the total symmetric function is deﬁned by hk =
j1 j2 ··· jk
Xj1 Xj2 · · · Xjk ,
i.e., the sum of all the monomials in the Xi of degree k. (i) For large values of n, express h1 , h2 , h3 in terms of the elementary symmetric functions e1 , e2 , e3 . (ii) Show that the power sum functions sk of the previous question satisfy sk = −(h1 sk−1 + h2 sk−2 + · · · + hk−1 s1 ) + khk .
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.]
85
.
In fact Q only needs to be a proper ideal of S for this argument to work. (c) If we identify A with the subring of scalar matrices in Matn (A). . (a) Recall from the Isomorphism Theorems of basic Ring Theory that ϕ−1 Q R. If n > 1 then Matn (A) is not commutative.Solutions Chapter 1 1. while for R = Fp [X]/(X 2 ). v ∈ R and R is an integral domain.4. To see that Q[i] is a ﬁeld. (a) Let u. R is an integral domain with characteristic subring Z and char R = 0. Then for n ∈ Z. 1. so this result follows from (a). (c) Consider Z ⊆ Q. (u − vi)(u + vi) = (u + vi)(u − vi) = u2 + v 2 = 0. in any case it always has zerodivisors since any singular matrix is a zerodivisor. then ϕ(u)ϕ(v) = ϕ(uv) ∈ Q and so ϕ(u) ∈ Q or ϕ(v) ∈ Q. 1. The main thing to check is that ϕ(u + v) = ϕ(u) + ϕ(v) which is a consequence of the Idiot’s Binomial Theorem. ϕ is not surjective. v ∈ S and suppose that uv = 0. Now by deﬁnition of characteristic. 1. v ∈ Q. 1. (u1 +v1 i)(u2 +v2 i) = (u1 u2 −v1 v2 )+(u1 v2 +u2 v1 )i.1. 1. (d) We have P ⊆ Q ∩ R R with P R maximal. v ∈ R with uv ∈ ϕ−1 Q.3. Consider the unit homomorphisms η : Z −→ R and η : Z −→ S. Clearly {n ∈ Z : n > 0 and nr = 0 for all r ∈ R} ⊆ {n ∈ Z : n > 0 and n1 = 0}.6.5. (a) For any subring R ⊆ C. they are both integral domains. (b) Q is a ﬁeld and Z ⊆ Q is a subring which is not a ﬁeld. char R = min{n ∈ Z : n > 0 and n1 = 0} = min{n ∈ Z : n > 0 and nr = 0 for all r ∈ R}. Suppose that u. nr = r + · · · + r = (1 + · · · + 1)r = (n1)r = 0r = 0. 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. Hence these sets are in fact equal. so ker ϕ = (0). For R = Fp [X]. notice that if u + vi = 0 with u. with Z[i] and Q[i] both closed under these operations and containing 1 = 1 + 0i as a unity. then inc−1 Q = Q ∩ R. The only proper ideal of k is the zero ideal (0). If 0 < n ∈ Z and n1 = 0. (p)Z Z is a (maximal) ideal that properly contains (0)Z . so they are subrings of the ﬁeld C. 1. η (n) = η(n). (b) Consider the inclusion function inc : R −→ S. When char R = p > 0 they must both be nonempty. n n so {n ∈ Z : n > 0 and n1 = 0} ⊆ {n ∈ Z : n > 0 and nr = 0 for all r ∈ R}. (a) Addition and multiplication follow from the obvious formulae (u1 +v1 i)+(u2 +v2 i) = (u1 +u2 )+(v1 +v2 )i. so ker η = ker η and therefore char S = char R. ϕ is not injective. by Qu. then the characteristic subring of Matn (A) is the same as that of A and char Matn (A) = char A. A[X] is an integral domain if and only if A is an integral domain. 1.2.7. then for every r ∈ R. then u = 0 or v = 0 since u. we need to show it is a prime ideal.1. hence u ∈ ϕ−1 Q or v ∈ ϕ−1 Q. so P ⊆ Q ∩ R. (b) The characteristic subring of A[X] is the same as that of A and char A[X] = char A.
ψa. Choose k(X) ∈ k[X] so that g(X)h(X) = k(X)ψa. (iii) This follows from (ii) and Proposition 1.9.b follow from the Homomorphism Extension Property 1.−ba−1 = id = ψa−1 . −1 Therefore these are inverse isomorphisms. therefore Q[i] is a ﬁeld. u. Since cn an = 0.b (f (X)) = f (aX + b) = i=0 ri (aX + b)i .30.cb+d (X). v ∈ Z and n = 0. ∞ k (b) Let a = k=0 ak X ∈ k[[X]]. b ∈ k[[X]] we may write ∞ ∞ a= k=k0 ak X k . in particular this forces a0 = 0 since otherwise the lowest term . Then a has an inverse in k[[X]] only if there is a b = ∞ k=0 b X ∈ k[[X]] with ab = 1.b ◦ ψc.b (k(X))p(X) −1 −1 and as p(X) is prime.−ba−1 : R[X] −→ R[X] has the property that ψa−1 .−ba−1 = ψa. Since ψa.b (f (X)) = deg f (X). then deg f (X) = n. By the uniqueness part of the Homomorphism Extension Property. Now i=0 n ψa.b (p(X))  g(X) or ψa. So k[[X]] is an integral domain.b (cX + d) = c(aX + b) + d = caX + (cb + d) = ψca. (a) Existence and uniqueness of such an ψa. Then the lowest degree term in ab is ak0 b 0 X k0 + 0 with ak0 b 0 = 0.cb+d . Hence ab = 0. p(X)  ψa.b (h(X)).−ba−1 ◦ ψa. we have −1 −1 −1 ψa.b (a−1 X − ba−1 ) = X.b ◦ψc.d = ψca.b (p(X))  h(X) and so ψa.b (p(X)) is prime.b (g(X))ψa. n so so the latter element is in the image of inc∗ which must therefore be a surjection.22 and its eﬀect on f (X) = n ri X i ∈ R[X] where ri ∈ R is i=0 n ψa. Hence every nonzero element of Q[i] has an inverse. b= 0 bX with ak0 = 0 = b 0 . ψa−1 . this shows that deg ψa. k ( k=0 ak X )( k=0 k bk X ) = k=0 k ( =0 a bk− )X k .−ba−1 (aX +b) = X and ψa. We have ψa.b (f (X)) = i=0 ci (aX + b)i = cn an X n + terms of lower degrees in X. so by the uniqueness part of the Homomorphism Extension Property. (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 . If a is a unit then ψa−1 .b (p(X))  g(X)h(X) for g(X). we have ψa.b ◦ ψa−1 . 1.b . 1 (b) & (c) The crucial point is that every element of Q[i] can be written as (u + vi) with n n.b (p(X)).b is an isomorphism.b (g(X)) or p(X)  ψa. Given two nonzero elements a.b (h(X)) = ψa. Then u2 inc∗ (u + vi) n = inc∗ (u + vi) n + 0i = 1 (u + vi). Hence ψa.d (X) = ψa.8.b . h(X) ∈ k[X].88 SOLUTIONS u v + 2 i ∈ Q(i) 2 +v u + v2 is the inverse of u + vi. (b) (i) If f (X) = n ci X i ∈ k[X] with ci ∈ k and cn = 0. Clearly k[X] ⊆ k[[X]] is a subring. (ii) Suppose that ψa. 1.
b ∈ Q}. X −k0 r=0 so inc∗ is surjective. 1. I6 = X 2 +X +1. So f (X) must be irreducible. We have ker ε√2 = ker ε−√2 = (X 2 − 2) Q[X] . Conversely. =0 a bn− = a0 bn + a1 bn−1 + · · · + an bn = 0 (n 1). is a ring homomorphism. 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).11. h(X) ∈ Z[X]. ρ(f (X)) = f (X). √ is ε−√2 Q[X] = Q[ 2] = ε√2 Q[X]. Notice that every element k0 < 0 can be written as ( ar+k0 X r )X k0 . where deg g(X) < deg f (X) = deg f (X) and deg h(X) < deg f (X) = deg f (X). deg g(X) < deg f (X) and deg h(X) < deg f (X). k=min{k1 .13. to ensure that ab = 1. I5 = (X 2 +2). 0 } j=0 a bk−j )X k .14. If f (X) = g(X)h(X) with g(X). I2 = (X 2 +2). The image is The image of ε−√2 √ √ ε√2 Q[X] = Q[ 2] = {a + b 2 : a. The reduction modulo p function ρ : Z[X] −→ Fp [X].10. X 3 + X − 1 reduces modulo 2 to an irreducible since it has no roots modulo 2. (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=k2 ∞ ∞ k bk X ) = ∞ k (ak + bk )X k . So X 3 + X − 1 is irreducible. So X 3 − X + 1 is irreducible. X 5 − X + 1 is irreducible modulo 3 and 5 so is itself irreducible. if a0 = 0.89 in X in ab would be of degree greater than 0. √ 1.12. X 3 − X + 1 reduces modulo 3 to an irreducible since it has no roots modulo 3. ∞ k=k0 Clearly k[[X]] ⊆ k((X)) is a subring. then we can inductively solve the system of equations n a0 b0 = 1. 1. 1. then f (X) = ρ(g(X)h(X)) = ρ(g(X))ρ(h(X)) = g(X)h(X). 1. I1 = (X 2 +1). I4 = (X − 2). 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. X 3 + 2X + 1 ≡ X 3 − X + 1 (mod 3) so this polynomial reduces modulo 3 to an irreducible and so is irreducible. Here f (X) = (3X − 3)d(X) + (−9X + 7). I3 = (X 2 −2). But this is impossible since f (X) is irreducible.k2 } k ( k=k0 ak X k )( = 0 b X )= ( k=min{k0 . 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 .
. b ∈ R} ⊆ R. Therefore β = 1. then X 4 − 2 = (X 2 − 2)(X 2 + 2) in R[X] and we have to consider which factor α is a root of. ker εα = ker ε−α = (X 2 + 2) R[X]. u = 2e 2 √ √ √ √ For the roots of g(X) we obtain 2. For α. −α i (notice that two of these are real while the other two are not). 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]. The other complex root of X 2 + X + 1 is ω 2 . 2 108 so q = −20 and p = 6. b ∈ Q}. while for f (X) we have 5. 3 + 2. 1. ker εα = (X 4 − 2) Q[X]. and the latter ideal is maximal. α i. ker εω = (X 2 + X + 1) Q[X]. d ∈ Q} = Q[α]. 1. 1. √ √ πi/4 r 2 ω = √ (1 + i)ω r = (1 + i)ω r (r = 0. b ∈ R} ⊆ R. d ∈ Q}. 3). = 108. take 27q 2 + 4p3 28 q = . If α2 + 2 = 0 then εα R[X] = ε−α R[X] = R[α] = {a + bα : a. First change variable to obtain g(X) = f (X + 3) = X 3 − 6X + 4. 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).17. − = 1. Work backwards with Cardan’s method. 1. − 3 − 1. c. For β. b.16. which has roots Thus we can take √ −2 ± 2i = ( 2)3 e3πi/4 . so the evaluation homomorphism εω2 has εω2 Q[X] = εω Q[X] and ker εω2 = ker εω .15. Then εω Q[X] = Q[ω] = {a + bω : a. Then ker ε−α = ker εα i = ker ε−α i = (X 4 − 2) Q[X] but although ε−α Q[X] = Q[α]. − 3 + 2.90 SOLUTIONS √ 1. If we replace Q by R. We have εα Q[X] = Q[α] = {a + bα + cα2 + dα3 : a. If α2 − 2 = 0 then εα R[X] = ε−α R[X] = R[α] = {a + bα : a. b. where X 2 + 2X + 10 has no real roots. take q 27q 2 + 4p3 − = 10. Thus α is a real root of f (X) = X 3 + 6X − 20. which is a maximal ideal. 3 − 1. we have εα i Q[X] = ε−α i Q[X] = Q[α i] = {a + bα i + cα2 + dα3 i : a. The other complex roots of X 4 − 2 are −α.18. Using Cardan’s method we have to solve the quadratic equation U 2 + 4U + 8 = 0. ker εα = ker ε−α = (X 2 − 2) R[X]. where (X 2 + X + 1) Q[X] is a maximal ideal. c. 2 108 27 so q = −2 and p = 1. so εα i Q[X] = εα Q[X] since one of these is a subset of R but the other is not. Notice that 2 is a real root of this polynomial and f (X) = (X − 2)(X 2 + 2X + 10). where X 2 + X + 2 has no real roots. Therefore α = 2.
(c) First notice that r r deg Φn (X) = ϕ(n) = (p1 − 1) · · · (pk − 1)p11 −1 · · · pkk −1 . described in the Proof of Example 1. so the Eisenstein Test can be applied to show that Φpk (X) is irreducible over Q. The complex roots of Φp (X) are the primitive pth roots of 1. 1. . 1 k r1 −1 so deg Φn (X) = deg Φp1 ···pk (X p1 ···pkk r −1 ). hence we must have n 1. Φpj (X) = Φp (X p 0 j k k−1 k−1 k k−1 )p − 1 = (X p ) k−1 − 1)Φp (X p k−1 ). ck ∈ k} = k[X]. In fact. k−1 k−1 Φpk (X) = (X − 1)(p−1)p k−1 + cp−2 (X − 1)(p−2)p k−1 + · · · + c1 (X − 1)p k−1 + c0 . so X ∈ ϕk[X]. . )p1 ···pk = ξ n = 1. c0 = Φpk (1) = Φp (1) = c0 = p. 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 . Let ϕ(X) = a0 + a1 X + · · · + an X n with ai ∈ k and an = 0. Then ϕk[X] = {c + 0 + c1 ϕ(X) + · · · + ck ϕ(X)k : c0 . Suppose that show that n > 1. If n = 0 then ϕk[X] = k ⊆ k[X] so ϕ would not be surjective. 0 j k−1 and therefore Φpk (X) = Φp (X p ). (ξ r −1 r −1 p11 ···pkk . Therefore ϕ(X) = a0 +a1 X and so ϕ = αA for some A ∈ Aﬀ 1 (k). we must have Φ20 (X) = X 8 − X 6 + X 4 − X 2 + 1.5). suppose that ϕ ∈ Autk (k[X]) is any automorphism. But if k > 0 and ck = 0 then deg(c + 0 + c1 ϕ(X) + · · · + ck ϕ(X)k ) = kn > 1. Also. / which gives a contradiction.20. since cyclotomic polynomials are irreducible. The Idiot’s Binomial Theorem gives Xp so where cr ≡ 0 (mod p). so the roots of Φpk (X) are their pk−1 st roots which are the primitive pk th roots of 1. To see that the homomorphism Aﬀ 1 (k) −→ Autk (k[X]).19. (a) We have X p − 1 = (X p so by (1. A −→ αA−1 .22.91 1. where cr ≡ 0 (mod p) and c0 = p. Since the roots of X 10 − 1 are the 10th roots of unity. Φpj (X). we ﬁnd that Φ20 (X)  (X 2 + 1)(X 8 − X 6 + X 4 − X 2 + 1). (b) Using the formula of Equation 1. So we must have n = 1.4.21. each root ξ of Φn (X).60 is surjective. Calculation. . 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). . c1 . 1. 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 (mod p) k−1 − 1) + c0 .
.. Thus a suitable polynomial is 4X 2 + 2X − 1 ∈ Q[X]. hence 4 cos2 (2π/5) + 2 cos(2π/5) − 1 = 0.43. 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). the only root of this polynomial in K must be 1.. We also have Φ5 (X) = X 4 + X 3 + X 2 + X + 1..n)=1 t (ζn − X) t=1. As these are monic polynomials of t (X − ζn ).n−1 gcd(t.n−1 gcd(t.92 SOLUTIONS r1 −1 and no smaller power of (ξ p1 ···pkk r −1 ) has this property.n−1 gcd(t. X p − 1 = X p + (−1)p = (X + (−1))p = (X − 1)p . This shows that Φn (X)  Φp1 ···pk (X the same degree they are equal.n−1 gcd(t.n)=1 −t (1 − Xζn ) t=1..24..n)=1 ···pkk r −1 ). m m X np − 1 = (X n − 1)p . 1. Now we have −1 ζ5 + ζ5 = 2 cos(2π/5)... Similarly. Since 2  ϕ(n) when n > 2 and the result is immediate when n = 2..23.. hence (ξ p1 p11 r −1 r1 −1 ···pkk r −1 ) is a root of Φp1 ···pk (X). Φn (X) = t=1. 1.n)=1 t (1 − Xζn ) t=1.n)=1 = (−1)ϕ(n) X −ϕ(n) t=1.33. −2 −1 2 ζ5 + ζ5 = (ζ5 + ζ5 )2 − 2 = 4 cos2 (2π/5) − 2... by the Idiot’s Binomial Theorem 1. we have −2 −1 2 (ζ5 + ζ5 ) + (ζ5 + ζ5 ) + 1 = 0... 1.... By Theorem 1.n)=1 t (X − ζn ) = (−1)ϕ(n) X −ϕ(n) Φn (X)..n−1 gcd(t. By the Unique Factorization Property 1.25.n)=1 n−t (1 − Xζn ) t=1.. (a) In K[X]. so Φn (X −1 ) = = X −ϕ(n) = X −ϕ(n) = X −ϕ(n) = X −ϕ(n) t (X −1 − ζn ) t=1.n−1 gcd(t... ζ5 = ζ5 .. so 4 3 2 ζ5 + ζ5 + ζ5 + ζ5 + 1 = 0...... −1 −2 4 3 Rearranging and using the formulae ζ5 = ζ5 . we see that desired equation always holds..n−1 gcd(t.10.
i) : Q] = [Q( 3. b2 ) = 1. b2 ∈ Z.4. √ √ q = (a + b p)2 = (a2 + b2 p) + 2ab p. As in (a) we have X p − a = X p − up = (X − u)p .6. 2. Also. So q ∈ Q( p). if [Q( p. √ The following three subﬁelds of Q(√ 3. Writing b = b1 /b2 with b1 . giving the simultaneous pair of equations a2 + b2 p = q. hence (b2 )2 p = (b1 )2 q. b2 ∈ Z and gcd(b1 . so u is the only root in K. Q( 3. i is a basis for Q(i) over Q. v) 1 ((u + v) + (u − v)) ∈ K(u + v). so . 2ab = 0. q) : Q( p)] √ √ say q = a + b p for some a. L2 = Q(i). i span the Qvector space Q(i). Chapter 2 2. then x + yi = 0 ⇐⇒ x = y = 0. b ∈ Q.5.1. we obtain b2 q = b2 p 2 1 and so p  b2 and q  b1 . √ √ √ If b = 0 then q ∈ Q which contradicts the result of Qu. i) are distinct and are extensions of Q having degree 2: √ L1 = Q( 3). 3) √ Q( 3) R.2. q) : Q( p)] = 1 then q ∈ Q( p). we obtain (b2 )2 p2 q = (b1 )2 q 2 p. (b) If u ∈ K is a root of this polynomial then up = a. L3 = Q( 3 i). By Theorem 2. / 2.93 and the only roots of this must be nth roots of 1.3. From this we obtain p  b1 and q  b2 . Then [Lr ∩ Ls : Q] > 1 ⇐⇒ Lr ∩ Ls = Lr = Ls . First notice that [Q( 3) : Q] = 2 (with Qbasis 1. √ √ √ √ [Q( 3. Then √ √ √ √ √ 2. 2. Notice that if v = ±u then b = v 2 = u2 = a which is impossible. √ √ √ √ 2. so 1. If a = 0 then q = b p. √ √ √ 2. 2 (u − v)(u + v) u2 − v 2 a−b = = ∈ K(u + v). But also if x. i ∈ Q( 3) and / √ √ √ and since i2 + 1 = 0.1.6(ii). i) : Q( 3)] [Q( 3) : Q] = 2 × 2 = 4. Hence [Q(i) : Q] = 2. Arrange the induction carefully. v) and therefore K(u + v) = K(u. Since 1. Then u−v = Hence u= So K(u. 2 K(u + v) v= 1 ((u + v) − (u − v)) ∈ K(u + v). 2. so v = ±u. but then p  b1 as well as p  b2 . i) : Q( 3)] = 2. v). y ∈ R. contradicting the fact that √ √ gcd(b1 . b2 ) = 1. u+v u+v u+v K(u. we have [Q(i) : Q] 2. i) = Q( 3)(i) has [Q( 3. Writing b1 = b1 q and b2 = b2 q for suitable b1 . It is obvious that [Q( p.4. This is similar to Example 2.
[Q(ζ5 ) : Q] = [Q[X]/(Φ5 (X)) : Q] = ϕ(5) = 5 − 1 = 4.94 SOLUTIONS Lr ∩ Ls = Q whenever r = s. an automorphism α ∈ AutQ (E) has the eﬀect α(21/n ) = 21/n ±21/n if n is odd. i) uuu uuu 2 uuu 2 u rr rrr rrr 2 rrr uuu uuu u 2 uuu u Q(i) 2 √ Q( 3 i) Q 2. if n is even. = (e) Q(ζ5 ) 2 Q(cos(2π/5)) 2 Q 2. 4. (d) For k = 0. −1 −1 (b) We have ζ5 = cos(2π/5) + sin(2π/5) i ∈ Q(ζ5 ). see Remark 6. If n is even. C 2 R ∞ Q( 3) tt tt tt tt √ t 2 ∞ √ Q( 3. (a) Since 5 is a prime. So each of the numbers cos(2kπ/5) is a root of the polynomial T5 (X) − 1 = (X − 1)(4X 2 + 2X − 1)2 .8. therefore Q(cos(2π/5)) ∼ Q[X]/(4X 2 + 2X − 1). But also ζ5 ∈ Q(ζ5 ) and ζ5 = cos(2π/5) − sin(2π/5) i ∈ Q(ζ5 ). the possibilities are α(21/n ) = ±21/n . Hence we have 1 1 −1 −1 ζ5 + ζ5 ∈ Q(ζ5 ). sin(2π/5) i = ζ5 − ζ5 ∈ Q(ζ5 ). 3.7. 2. The only real subﬁeld amongst these is L1 . 2. For k = 1. [Q(cos(2kπ/5)) : Q] = 2. 4. so α(21/n ) ∈ En is also a real nth root of 1. (a) If α ∈ AutQ (En ) then α(21/n )n = α(2) = 2. cos(2kπ/5) is a root of 4X 2 + 2X − 1. so T5 (cos 2kπ/5) − 1 = 0. If n is odd. 1. the only possibility is α(21/n ) = 21/n . cos(5(2kπ/5)) = cos(2kπ) = 1. 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 2 (c) This can be found by repeated use of the double angle formula cos(2π/5) = cos(A + B) = cos A cos B − sin A sin B. 2. 3. The polynomial Tn (X) expressing cos nθ in terms of cos θ is called the nth Chebyshev polynomial. On passing to the quotient homomorphism of ε21/n we obtain an automorphism τn of En under which τn (21/n ) = −21/n . This is similar to the previous question. so α = id.6. . (b) Since E R.9.
Explicitly these are √ √ √ √ 3 1 3 1 3 1 3 1 + i. √ p3 (X) = X 4 + 2: The roots are the four 4th roots of −2. The diagrams at the bottom indicate useful subﬁelds of the splitting ﬁelds occurring in each of these examples.e.95 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. ±e±πi/6 . (c) Assuming there are only 6 such subﬁelds. − i. (ii) ⇐⇒ (iii). 4 2e(2k+1)πi/4 for k = 0. − 2. − + i. . − √ + √ i.e. √ − √ i. 3. 3 {{ { { E12 2 E4 2 {{ {{ E2 g { {{ {{ { {{ 3 E6 g gg 2 gg gg g gg gg g 2 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]. 3. 4. + i. 2. 1. i. Writing ω = e2πi/3 . Explicitly. − √ − √ i.e. 1.e. 4 4 4 4 4 4 4 4 2 2 2 2 2 2 2 2 √ 4 The splitting ﬁeld is E = Q( 2. so p1 (X) = X 2 the four roots of p1 (X) are the complex square roots of these numbers. 2 √ 2 2 2 2 2 2 2 The splitting ﬁeld is E = Q( 3. / 1/2 + 21/3 ∈ E . 6 2e2kπi/6 = 6 2ekπi/3 for k = 0. 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.9. they form the following tower.. 3i) = Q( 6 2)( 3i) which has degree [E : Q] = 12. Chapter 3 3.1. − i. i) and [E : Q] = 8.2. i) and [E : Q] = 4. By Theorem 2. (i) ⇐⇒ (ii). − − i.. 3. so it is certainly an element of E6 which is the only degree 2 extension of E3 . − − i. so it cannot lie in E4 since 21/2 + 21/3 ∈ E2 and 3 [E4 : E2 ] = 2. i. i. t is algebraic over K if and only if ker εt = (0). If 21/2 + 21/3 ∈ E3 then 21/2 ∈ E3 . Hence these three conditions are indeed equivalent. so α = id. Clearly. Explicitly these are 1 1 1 1 1 1 1 1 √ + √ i. 5. So 21/2 + 21/3 is in E6 and / E12 and none of the others. √ 4 − X 2 + 1: The polynomial X 2 − X + 1 has the complex roots e±πi/3 = 1 ± 3 i .. i. 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]. which would imply 2 = [E2 : Q]  [E3 : Q] = 3 which is false. so 21/2 + 21/3 ∈ E3 . √ √ p2 (X) = X 6 − 2: The roots are the six complex 6th roots of 2.. This contradiction shows that α(21/n ) = 21/n for every n. 2. 2 2 2 2 2 2 2 2 √ √ √ √ The splitting ﬁeld is E = Q( 6 2.
u2 = 3 2ζ3 . √ √ √ Q( 3. = σid = id. in fact we have Q(ζ5 ) = Q(cos(2π/5). Irreducibility is a consequence of the polynomial version of the Eisenstein Test 1.48. 10) : Q] = 4 and the element 5 + 10 has degree 4 with √ √ minimal polynomial X 4 − 30X 2 + 25 which has roots ± 5√ 10. Suppose that t ∈ k(T ) is a root of g(X). i) : Q] = 4 and the element 3 + i has degree 4 with minimal √ polynomial X 4 − 4X 2 + 16 which has roots ± 3 ± i. therefore AutQ (Q( 3 2. σα1 = (1 2 3). ± √ √ Q( 2. 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. 3) 6 √ Q( 4 2. ζ3 )) ∼ S3 . so can be identiﬁed with the unique permutation σα ∈ S3 for which α(ui ) = uσα (i) (i = 1. . hence v ∈ K(u + tv). √ √ √ √ √ √ 3. so t is in fact a root of multiplicity p. ζ3 )) permutes these roots. i) 2 √ √ Q( 6 2.4. 3. σα2 = (1 3). u3 = 3 2ζ3 . 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 3.96 SOLUTIONS 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 ).6. polynomial X 3. σα0 = (2 3). 3. √ These are the six elements of S3 . This also gives the factorization of g(X) into linear factors over k(T ). σα2 = (1 3 2). √ √ Q( 6 2. This implies that u = (u + tv) − tv ∈ K(u + tv). hence K(u. Then (s − t)v = (u + sv) − (u + tv) ∈ K(u + tv). Then each automorphism √ α ∈ AutQ (Q( 3 2. 2.5. Then there are only ﬁnitely many of these. i)/Q: Here [Q( 3. sin(2π/5)i). List the three roots of X 3 −2 as u1 = 3 2. Consider the subﬁelds K(u + tv) K(u. i) 2 Q(cos(2π/5). 3). i)/Q: Here [Q( 2. t ∈ K such that s = t and K(u + sv) = K(u + tv).3. hence it is the only root of g(X) in k(T ). Q( 5. v) with t ∈ K. Here is the argument that K(u. We assume that K is inﬁnite since otherwise the result will be proved in Proposition 5. i)/Q: Here [Q( 4 3. i) 2 √ Q( 3.16. i) : Q] = 4 and √ element 2 + i has degree 4 with minimal the polynomial X 4 − 2X 2 + 9 which has roots ± 2 ± i. The induction is straightforward. v) K(u + tv) K(u. i) : Q] = 4 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. then using the Idiot’s Binomial Theorem we have (X − t)p = X p − tp = X p − T. v)/K is simple. v) = K(u + tv). so there must be s. We ﬁnd that (using cycle notation for permutations) σα1 = (1 2). 10)/Q: Here [Q( 5. v) and so K(u. √ √ √ Q( 4 3.
so we can take δ = 2 3i. So  Gal(Q(u. w)/Q) A3 ∼ Z/3. w)/Q) is a subgroup of S3 (viewed as the permutation group of {u. (a) Suppose that f (X) = c3 X 3 + c2 X 2 + c1 X + c4 with c3 = 0. Now a3 − 3ab2 + b3 = 0. (b) Viewing Gal(E/K) as a subgroup of S3 .26 implies that Gal(Q(u. α3 }. so f (X) does not split over Q(u) even though it has a root in this ﬁeld. v = 3 ζ8 = (1 − i). v. α. βα3 = {ι. βα3 }. so if we take u to be any cube root of c3 and u = −c2 /3c3 then f (uX + v) has the desired form. βα. The Galois group Gal(Q(u. w))/Q) = 3 = and Gal(Q(u. v. Chapter 4 4. the discriminant of f (X) is ∆ = −27 − 4(−3)3 = 81 = 92 . βα = {ι. w. v. βα = {ι. v. βα}. so both its roots in K lie in E.7 for the rest of this question. so [K(u) : K] = 2 = [E : K] and therefore K(u) = E.8. 3. Q) for which ϕ(u) = v.81 we know that splitting ﬁelds are always normal. v. (c) This is a tedious calculation! See Section 4. βα2 }. (a) This should be a familiar result. v. This is an example of case (iii) of Kaplansky’s Theorem and we use the notation of the √ proof. 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 ). Let E C be a splitting subﬁeld for f (X) over Q. 2 2 . βα2 }. 3.5.97 3. w}). β = {ι. w)) is cyclic of order 3 whose generator is a 3cycle which cyclically permutes u. βα2 = {ι. α2 . If a/b is a rational root of f (X). by Theorem 4. v.1. If E/K is a quadratic extension then for any u ∈ E − K we have 1 < [K(u) : K] 2. / This means that there is a monomorphism ϕ ∈ MonoQ (Q(u). This shows that E is normal over K. β}.3. w) = Q(u. β. and there are three normal subgroups of order 4. we have f (uX + v) ∈ K. βα3 }. but 1 is certainly not a root.4. then L(v)/K is separable.8 we know that 3 divides  Gal(E/K).4). Then if v ∈ C is a nonreal root of f (X) we have v ∈ Q(u). Since E is obtained by repeatedly adjoining roots of p(X). The roots of X 2 + 3 are √ ± 3i. If the distinct roots of f (X) in C are u. α2 . α2 .25. α2 . the splitting subﬁeld K(v. v. Then minpolyK. Proposition 4.7. so it is only necessary to show that the splitting ﬁeld E of p(X) over K is separable over K. (b) The centre of D8 is α2 which has order 2. b) = 1. namely α = {ι. v. which easily implies that a. b = ±1. 4. hence ϕQ(u) = Q(u) and so Q(u)/Q is not normal. By Theorem 3. Since the discriminant is a square in Q. Hence there are no rational roots and so no proper rational factors. w) : Q] and [Q(u. β = {ι. C) = MonoQ (Q(u). so we may assume √ √ √ √ √ √ −1 243 243 4 4 u = 3 ζ8 = (1 + i). Notice that there are also four nonnormal subgroups of order 2. w) C satisﬁes 3  [Q(u. α2 .u (X) must factor into linear factors over E. so provided that we can ﬁnd a cube root of 1/c3 in K. Notice that v ∈ K(u) and then f (uX + v) ∈ K(u). 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. but the only subgroups of S3 with this property are S3 and A3 . 4. w. v.2. 4. By the formula following Proposition 4. 4. the result follows from Proposition 3. If char K = 2 then all quadratic polynomials over K are separable. The discriminant here is δ 2 = −12.73 together with the fact that if L/K E/K is separable and v ∈ E is a root of p(X). w) : Q]  3! = 6. we may assume that gcd(a.
√ notice that [E ∩ R : Q] = 3. [Q( 2. we have √ √ √ −1 √ √ 4 4 4 4 4 β( 3 ζ8 ) = − 3 ζ8 . Gal(E/Q) ∼ S3 and = √ ∼ A3 . so 5 ∈ E. Q(i): Continuing the preceding discussion. 4 3 ζ8 ) = Q( 4 3. ζ3 ) and √ Gal(Q( 3 10. √ The discriminant of the polynomial X 3 − X − 1 is ∆ = −23 and so δ = 23 i. Hence we have uv = 3 and uvδ = 6i. .α σ 2 .6. √ √ √ Since ζ3 is not real. √ 4. hence / / = √ √ Gal(Q( 5 i)(X 3 − X − 1)/Q( 5 i)) ∼ S3 ∼ Gal(Q(i)(X 3 − X − 1)/Q(i)). 3 10) : Q( 2)] = 3 and √ √ √ √ 3 3 Q( 2. = √ for√ which σ( 3 10) = 3 10 ζ√ 3. with splitting ﬁeld Q( 3 10. σ2. 1. ζ3 ) and [Q( 3√ ζ3 ) : Q(ζ3 )] = 3. Q( 5 i). ζ3 ). hence Gal(Q( 3 10. 4. with [Q(ζ3 ) : Q] = 2. X 3 + X + 1 ∈ F2 [X].ασ E = Q(i). σ 2 . σ 2 . β( 3 ζ8 ) = β(− 3 ζ8 i) = − 3 ζ8 i. and these have ﬁxed ﬁelds √ 2 E σ = Q( 3.2 above (see also Section 4. This gives the diagram of subﬁelds of E √ −1 √ √ E = Q( 4 3 ζ8 .98 SOLUTIONS √ √ where as usual ζ8 = e2πi/8 = (1 + i)/ 2. ασ .10). while β( 3i) = 3i and β( 3) = − 3. Using the previous question (but beware that the notation there is inconsistent with that of the present situation!) we have the normal subgroups σ2 . Then if √ E = Q( 23 i)(X 3 − X − 1) is the splitting ﬁeld of X 3 − X − 1 over Q. Q(X 3 − 10)/Q: This is similar to Example 4. i) j jjjj jjjj j jjjj jjj jj √ √ Q(i) Q( 3i) Q( 3) jj jjjj jj 2 jjjj 2 jjjj 2 jj jjjj Q √ √ √ √ Then α is the restriction of complex conjugation to E. and these generate a dihedral subgroup of S4 . 3 10. 5 i ∈ E and i ∈ E.20. ζ8 ) √ Q( 3. 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. The Galois group is isomorphic to S3 . √ √ Q( √3 i)(X 3 − 10)/Q( 3 i): Here Q( 3 i) = Q(ζ3 ). 3 10. ζ3 ) :√ Q( 2)] = 6. ζ3 )/Q(ζ3 )) ∼ Z/3 with generator σ 10. = = each of which is a normal extension of Q. ζ3 )/Q) ∼ S3 . − 4 3 ζ8 of f (X) are given in permutation notation by σ = (1 4 3 2) and α = (1 2)(3 4). The splitting ﬁeld is √ √ Q( 3 10. Using the choices of the proof.7). − 4 3 ζ8 . ζ3 ). Similarly. √ √ √ hence Gal(Q( 5)(X 3 − X − 1)/Q( 5)) ∼ S3 . √ = Q( 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. 4 3 ζ8 . Gal(E/Q( 23 i)) = √ √ K(X 3 − X − 1)/K for K = Q. √ √ −1 √ √ −1 The eﬀects of σ and γ on the four roots 4 3 ζ8 . hence also β(i) = −i. i). α . [Q( 2. has no root in F2 and hence has no linear factor (see Qu. To proceed further we can use the ideas of Qu. Q( 5). E E √ = Q( 3 i). 10) Q( 2. σ σ . 10. = √ √ √ √ √ √ √ Q( 2)(X 3 − 10)/Q( 2): The splitting ﬁeld is Q( 2.
F can be taken to be the ﬁeld of fractions of D). 25 Consider the polynomial X 2 + X + 1 ∈ F5 [X]. To ﬁnd an element of order 24 in F× .99 4. hence they are (−1 ± u) = −3 ± 3u. Since the extension Fpdn /Fpd is normal. . Since 8  (41 − 1). Since deg f (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). As gcd(p. where w2 = (1 − 4) = −3 = 2.10 X p − a = X p + (−u)p = (X − u)p . where the left hand side is in K. Since γ(ξγ ) = ξγ . we ﬁrst ﬁnd one of order 3. which can only equal u if p  r. we know that each root of g(X) has the form wξ where ξ is some pth root of 1. (b) Here 5 is prime 4  (5 − 1). and if U is ﬁnite so is jU . Therefore jU and U becomes isomorphic to a subgroup jU F are cyclic. an integral domain D always admits a monomorphism into a ﬁeld j : D −→ F (e. 5. s such that rd + sp = 1.4. so any subgroup U D× × . f (X) = pX p−1 = 0.6 implies that it is isomorphic to Fpdn . there are integers r.3. Therefore F25 is the splitting ﬁeld for X 8 − 1 over F5 and we have F25 ∼ F5 (u) = F5 (v). (b) If there is a root u ∈ K. By Theorem 1. there are no multiple roots. This shows that u ∈ K. Notice that in F5 [X].2. Arguing as in (a). where the polynomials X 2 − 2 and X 2 − 3 are irreducible. Hence if f (X) is not irreducible in K[X] it has a root in K. p) = 1. so we have a = (ar )p (ad )s = a pth power in K. where u2 = 2 and v 2 = 3. every root of f (X) is conjugate to u. The only root of X 2 + 1 in F2 is the multiple root 1.g. 4. so v = uξ for ξ ∈ K a pth root of 1 with ξ = 1. Fpdn is a splitting ﬁeld for f (X) over Fpd . We can write γ(u) = uξγ r where ξγ = 1 is a pth root of 1.1. v ∈ L are distinct roots.. 2 Now the elements ±(2 ± 2u)u = ±(±4 + 2u) = ±4 ± 2u all have order 8 × 3 = 24. 2 and 3 have order 4. hence it is a ﬁnite ﬁeld with pdn elements. By Proposition 3. then (vu−1 )p = 1. the Galois group Gal(L/K) acts in the following way. In fact. from which it follows that ad is a pth power in K. hence p distinct roots. so f (X) must be irreducible over K. there must be an element γ ∈ Gal(L/K) with γ(u) = u.55. These roots are given by (−1 ± w)/2. d) = 1. 6 is a primitive root for F41 and 65 ≡ 27 (mod 4)1 has order 8. If u ∈ L is a root of f (X) in an extension L/K then by the Idiot’s Binomial Theorem 1. so there is a primitive 4th root of unity in F× . L must contain p distinct pth roots of 1. Hence either f (X) has a root in K or it must be irreducible over K. moreover. So u must have at least p conjugates which are all roots of f (X). 5. X 8 − 1 = (X 4 − 1)(X 4 + 1) = (X 4 − 1)(X 2 − 2)(X 2 − 3). this has roots which have order 3. so ±u and ±v are = primitive 8th roots of unity. so these are primitive roots for F5 . Hence (ud )s (up )r = u.8. in F5 . 5. (c) Suppose that f (X) = g(X)h(X) with g(X) ∈ K[X] monic irreducible and 0 < d = deg g(X) < p. (a) Since char K = 0. Chapter 5 5. If (X − u)d ∈ K[X] for some d with 1 < d < p then ud ∈ K. Now the constant coeﬃcient of g(X) is g(0) = (−1)d ξ0 wd ∈ K where ξ0 is a pth root of 1. s such that rp + sd = 1. Since gcd(d. (a) Here 41 is prime. so if u ∈ L is any root of f (X) then f (u) = pup−1 = 0. The ﬁeld Fpd [X]/(f (X)) is an extension of Fpd which has degree n.17. By The/ orem 4. If u. so u is the only such root in L and f (X) splits over L. there is a primitive 8th root of unity in F41 . but no primitive 5 8th root of unity.7. So p g(0)p = (−1)dp ξ0 (wp )d = (−1)dp ad . there are integers r. for r 1 we have γ r (u) = uξγ .8. hence Proposition 5.
But Fpd is not cyclic. Then τt0 must be isomorphic to a nontrivial subgroup of Fp . if w ∈ F× then Fp (w) Fp . (c) If K is a ﬁnite ﬁeld and d > 1 then if ga (X) were irreducible over K. So now 2d suppose that p is odd. there must be an element τt0 ∈ Gal(E/K) for which τt0 (u) = u + t0 . Hence there is an isomorphism Gal(E/K) ∼ = Fpd with σt corresponding to t ∈ Fpd . F× is a cyclic group. whose only root in F2 is 1. If this element has order 8. By the same approach as in (b).23 that = ∼ Z/d is cyclic. b = ±7. [E : K] = pd =  Gal(E/K) and so the following pd automorphisms are the elements of Gal(E/K): σt : E −→ E. σs ◦σt = σs+t . The set of squares in F× is the pd pd normal subgroup (F× )2 = {u2 : u ∈ F× } F× pd pd pd d d d d . This means that E = K(u) since all the other roots of ga (X) lie in K(u). 5. As ga (X) is irreducible over K. If u + t0 = u is a conjugate of u with 0 = t0 ∈ Fp . each orbit has exactly d elements. Now 6 is not a square in F11 but 72 ≡ −6 ≡ 42 (mod 1)1. pd = d ϕ(pd − 1). So let us solve (a2 − b2 ) + 2abz = z. Since Fp (w) = Fpd we have d p The number of conjugates of w is d. The polynomial X 2 + 1 is irreducible over F11 so F121 = F11 (z) where z 2 = −1. but this must be Fp since this group is simple. so . 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 . If p = 2 then F×  = 2d − 1. 10 ± 4z.. Hence. Suppose that a + bz ∈ F121 with a.12. it is suﬃcient to ﬁnd elements of order 8. which is even.e. hence b = ±a. 5. so we have a = 4. u must have p conjugates and so ga (X) is irreducible over K. Therefore the elements of order 8 in F× are 121 4 ± 4z and 7 ± 7z. we may therefore take λ2d (u) = 1 for all u ∈ F× . So the splitting ﬁeld is F2 . suppose that ga (X) has no root in K. Notice that Fp (w) is a splitting ﬁeld of the separable polynomial X p −1 − 1 over Fp . (a) First note that ga (X) = −1. we also have degFp w = d. then for t ∈ Fpd . hence u + t is also a root of ga (X). Now we have 2a2 = ±1 and so a2 = ±1/2 = ±6. the other roots are the p elements u + t ∈ E (t ∈ Fp ). so ga (X) is separable. (b) If ga (X) is irreducible over K then it cannot have a root in K since its degree is greater than 1. 3 and 5 whose product will have order 120. q 2d × so every element of F2d is a square. hence E/K is separable. b ∈ F11 . the elements of order 3 in F121 are (−1 ± 5z)/2 = 5 ± 8z. Hence d  ϕ(pd − 1).5. 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). If u ∈ E is a root of ga (X). Then F×  = pd − 1. It is easy to check that for s. Gal(Fpd /Fp ) = 5. i.100 SOLUTIONS (c) Here 11 is prime and 8  (121 − 1) = 120. E would be ﬁnite and Gal(E/K) ∼ Fpd . yet we know from Proposition 5. Then 2ab = 1 and b2 = a2 .7. (a) By Proposition 5. (d) In F2 [X] we have X 8 − 1 = (X − 1)8 . ga (u + t) = (u + t)p − (u + t) − a = (up − u − a) + (tp − t) = (up − u − a) = 0. Conversely.6. Since 120 = 8 × 3 × 5. which is odd. then by (a). t ∈ Fpd . b = ±4 and a = 7. 2 is a primitive root for F11 so 4 = 22 has order 5. σt (u) = u + t (t ∈ Fpd ). = so the number of orbits is ϕ(pd − 1)/d which is an integer. Combining these we obtain the following primitive roots for F121 : 7 ± z. then (a + bz)2 = ±z. Then if u ∈ E is any root of ga (X) in a splitting ﬁeld over K. so F121 is the splitting ﬁeld of X 8 − 1 over F11 .
v ∈ Fq (possibly 0) for which u2 = t − v 2 . In this situation. we may write −1 = a2 + b2 for some a. 6. where each pj is an odd prime. 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. 6. = pd pd We may use this group isomorphism to deﬁne λq . whence t = u2 + v 2 . which is abelian. then (−ζ)n = (−1)n ζ n = −1. Hence G/Z has order G/Z = pk with k < n. This establishes the inductive step and hence the desired result. 2 2 Chapter 6 6. G ∼ Z/p. 3 (hence n = 1. 8). (q + 1) − Σq ∩ (t − Σq ) Σq ∪ (t − Σq ) = Σq  + t − Σq  − Σq ∩ (t − Σq ). 1. Now if G = pn . there are u.. 12). Thus we have q Σq  = 1 + Then t − Σq  = Σq  = (c) Since Σq ∪ (t − Σq ) ⊆ Fq . pd λq is surjective if and only if p is odd. Write n = 2k pr1 · · · prs . there is a normal subgroup M G/Z with M  = pk−1 . Hence −ζ ∈ K is a primitive 2nth root of unity. Now when n = 1.e. By the inductive hypothesis. If ζ ∈ K is a primitive nth root of unity. • s = 0 and k = 0. (q + 1) . 4. the centre Z of G is nontrivial. Thus for every t ∈ Fq . hence F× /(F× )2 ∼ {±1}.3. while (−ζ)2n = (−1)2n ζ 2n = 1. for any nonzero t ∈ K. b ∈ Fq . then either u = 0 or u = 0 and u = (±v)2 for some v ∈ F× . λp (u) = . . 1. −t = t (since otherwise 2t = 0 and so t = 0). Suppose that the result holds whenever = G = pk with k < n. Clearly N  = Z M  = pn−1 . p1 < p2 < · · · < ps . we have q This implies that q and so Σq ∩ (t − Σq ) 1. rj s 1 k 0.101 and it is easily seen that its quotient group has order 2. By one of the Isomorphism Theorems.1. 2 (hence n = 3. i. Then r ϕ(n) = ϕ(2k )ϕ(pr1 ) · · · ϕ(prs ) = ϕ(2k )(p1 − 1)p11 −1 · · · (ps − 1)prs −1 . s = 1 and k = 0 (hence n = 5). 2. 2 (q − 1) (q + 1) = . u Remark: when d = 1. the Legendre symbol of u from Number Theory. • p1 = 3. 6. 12 + a2 + b2 = 0. (d) By (c). recall that by Cauchy’s Lemma. p (b) If u ∈ Σq . Clearly we have ker λq = (F× )2 .2. there is a normal subgroup N G containing Z and satisfying N/Z = M ⊆ G/Z. s = 1 and k = 0.
(a) If 4 n then writing n = 2k with k odd. 7. 19.5. We also have −1 ζn − ζn = 2 sin(2π/n) i ∈ Q(ζn ). For each of these numbers r. / sin(2π/n) = and by Theorem 6.3. Notice that 23 acts like complex conjugation. satisﬁes r2 = 1. hence these all have order 2 except 1 which has order 1. The eﬀect of r these elements on Q(ζ24 ) is given by r · ζ24 . = 6. −2 2 ζ2n − ζ2n ∈ Q(ζ2n ). 3. 17. the residue class modulo 24. If it contained a 3rd root of unity then it has √ √ would contain 3 and so Q( 3. (b) This is similar to (a). i) Q(i) which is impossible since [Q( 3. 2 2 √ Q( 5 i): This ﬁeld contains only the square roots of unity ±1. so we might as well assume that n is even from now on. For any n 1. with minpolyQ(cos(2π/n)). 2i sin(2π/n) ∈ Q(ζ2n ) ∩ R = Q(cos(π/n)). 13.4. then Q(ζn ) = Q(−ζn ) where −ζn is a primitive 2nth root of unity. Then we have (Z/20)× ∼ Z/2 × Z/4. 5. for instance 7 and 13. Q(ζn ) cannot contain ζ2n and by another simple argument it cannot contain i = ζ2n . we can write n = 4 . 7. 9.102 SOLUTIONS Q(i): Here degree [Q(i) : Q] = 2 and clearly the four 4th roots of unity ±1. Then i = ζn . 19. √ √ 1 2 Q( 2 i): This ﬁeld contains the 4 primitive 8th roots of unity ± ± i as well as the 4th 2 2 roots. we have [Q(ζn ) : Q] = ϕ(2k) = ϕ(2)ϕ(k) = ϕ(k). The eﬀect on Q(cos(π/12)) is given by r · cos(π/12) = cos(πr/12). so in particular. whence sin(π/2 ) = sin(2π/n) = −1 ζn − ζn ∈ Q(ζn ). it√ no 5th roots of unity except 1. Notice that since i = ζ2n . 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. r. Also. 11. 17. Since (Z/24)× is abelian. so i ∈ Q(ζn ). Notice that if n is odd. 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. ±i. ±i lie in this ﬁeld. 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.sin(2π/n) (X) = X 2 + cos2 (2π/n) − 1. let ζn = e2πi/n = cos(2π/n) + sin(2π/n) i. This time there are elements of order 4. we have [Q(cos(π/n)) : Q] = 2[Q(cos(2π/n)) : Q]. k Hence. So k we see that sin(2π/n) ∈ Q(ζn ) in this situation. i . 13. As ϕ(5) = 4. The elements of Z/24 which are invertible are the residue classes modulo 24 of the numbers 1. ± ± i. 23. √ √ 1 3 Q( 3 i): This contains the six 6th roots of unity ±1. (a) We have ϕ(24) = ϕ(8)ϕ(3) = 4 × 2 = 8. i) : Q] = 4. 11. If 4  n. −r · cos(π/12) = cos(−πr/12) = cos(πr/12) = r · cos(π/12). while [Q(ζ2n ) : Q] = ϕ(4k) = ϕ(4)ϕ(k) = 2ϕ(k).
if is even. −1}. 2 Consider the automorphism σ ∈ Gal(Q(ζn )/Q)) for which σ(ζn ) = ζn see that σ has order 2. 4 √ √ − 6− 2 7 · sin(π/12) = −7 · sin(π/12) = − sin(7π/12) = . 2 and so Then and Gal(Q(sin(π/12))/Q)) ∼ (Z/4 )× /{1. i Explicitly we have √ √ 6− 2 1 · sin(π/12) = −1 · sin(π/12) = sin(π/12) = . 11 · 3 = − 3. 1 · 3 = 3. −1} √ 1 − cos(π/6) 2− 3 sin (π/12) = = . 11 · sin(π/12) = −11 · sin(π/12) = − sin(11π/12) = 4 √ √ In terms of the generators 2 and 3 these act by √ √ √ √ √ √ √ √ 1 · 2 = 2. since cos(π/ ) = 1 − 2 sin2 (π/2 ) ∈ Q(sin(π/2 )). 3). −1 −ζn + ζn = 2(−ζn ) +1 = −ζn . σ(sin(2π/n)) = σ(sin(π/2 )) = From this we ﬁnd that when is odd. 4 √ √ − 6+ 2 . Q(cos(2π/n)) = Q(cos(π/2 )) = Q(cos(π/ ))(sin(π/2 )) = Q(sin(π/2 )). 5 · 2 = − 2. 6. √ √ √ √ √ √ √ √ 7 · 2 = 2. . is even. −1} ∼ Z/2 × Z/2. (a) We have  Gal(Q(ζ5 )/Q) = [Q(ζ5 ) : Q] = deg Φ5 (X) = ϕ(5) = 4.103 Clearly sin(π/2 ) ∈ Q(ζn ) ∩ R = Q(cos(2π/n)). = = Here the eﬀect of the coset of the residue class of r ∈ (Z/4 )× is given by ζr − ζr r · sin(π/12) = 24 r 24 = sin(rπ/12) i1−r .6. 5 · 3 = − 3. 11 · 2 = − 2. Similarly. √4 √ 6+ 2 5 · sin(π/12) = −5 · sin(π/12) = sin(5π/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 (b) We have Gal(Q(sin(π/2 ))/Q) = Gal(Q(cos(π/2 ))/Q) = (Z/4 )× /{1. σ(cos(π/ )) = cos(π/ ). sin(π/12) = 2 4 √ √ √ √ Q(sin(π/12)) = Q( 6 − 2) = Q( 2. 2 4 √ √ √ 2− 3 6− 2 = . it is easy to sin(π/2 ) if − sin(π/2 ) if is odd. 5 · 3 = 3. Then σ(cos(2π/n)) = σ(cos(π/2 )) = − cos(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.
so this is this polynomial is irreducible over Q. (b) Since (−1)(p−1)/2 = −1 if p ≡ 1 (mod 4). 2 4 √ √ −1 ± 5 −1 + 5 The roots of this are .104 SOLUTIONS and (Z/5)× is cyclic generated by the residue class 2. As sin(2π/5) > 0. 4 2 · ζ5 = ζ5 . Q(ζ5 ) 2 Q(ζ5 ) 4 √ = Q(cos(2π/5)) = Q( 5) 2 Q 6. 1 if p ≡ 3 (mod 4). sin (2π/5) = 1 − cos (2π/5) = 1 − 16 8 √ 5+ 5 hence sin(2π/5) = .cos(2π/5) (X) = X 2 + X − . . 3 2 · ζ5 = ζ5 .7. therefore 1 1 minpolyQ. 2 3 2 · ζ5 = ζ5 . We 4 √ 4 −1 − 5 also have cos(4π/5) = . The quadratic polynomial 4X 2 + 2X − 1 ∈ Z[X] has discriminant 20 which is not a square in Q. (a) We have (p−1)/2 (p−1)/2 r −r −r r (ζp − ζp )(ζp − ζp ) r=1 (p−1)/2 ξ2 = r=1 r −r (ζp − ζp )2 = (−1)(p−1)/2 = (−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. 4} Gal(Q(ζ5 ). 4 −1 −2 −1 3 4 (b) We have ζ5 + ζ5 = 2 cos(2π/5) and Φ5 (ζ5 ) = 0. As cos(2π/5) > 0 we must have cos(2π/5) = . The action is given by 2 2 · ζ5 = ζ5 . 4 √ √ 5+ 5 1+5−2 5 2 2 = . so since ζ5 = ζ5 and ζ5 = ζ5 . 8 (c) Gal(Q(ζ5 ) ∼ Z/4 and has 3 subgroups {1} {1. Hence 4 cos2 (2π/5) + 2 cos(2π/5) − 1 = 0. giving the following tower = of subﬁelds. −1 −2 2 (ζ5 + ζ5 ) + (ζ5 + ζ5 ) + 1 = 0 and therefore −1 −1 (ζ5 + ζ5 )2 + (ζ5 + ζ5 ) − 1 = 0.
(a) This can be proved by induction on n. 6.8.105 and p−1 s (1 − ζp ) = Φp (1) = p. (c) Taking square roots we ﬁnd that if p ≡ 1 (mod 4). r n − 2 we have r √ ± p √ ± pi (1 2 · · · n) (1 2)(1 2 · · · n)n−r = (1 2 · · · n)r (1 2)((1 2 · · · n)r )−1 = (r + 1 r + 2). so T (u) is ﬁxed by σ and all its powers. 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. (b) Let v ∈ E and suppose that TrE/K (v) = 0. Then for 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).15. . Write e[m] = r i1 <i2 <···<ir m Xi1 · · · Xir . By Artin’s Theorem 6.9. Recall the wellknown formula σ(i1 · · · ir )σ −1 = (σ(i1 ) · · · σ(ir )). 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). so there is an element t ∈ E for which TrE/K t = t + σ(t) + · · · + σ n−1 (t) = 0. if p ≡ 3 (mod 4). Hence H = Sn . Also recall that every permutation ρ ∈ Sn is a product of 2cycles. s=1 the result follows. It is straightforward to verify that the resulting function TrE/K : E −→ K is Klinear. Therefore T (u) is in E Gal(E/K) = K. hence by Gal(E/K). so it suﬃces to show that every 2cycle (a b) ∈ Sn is a product of 2cycles of the form (r + 1 r + 2). s[m] = r 1 i m Xir . the linear combination of characters id +σ + · · · + σ n−1 must be linearly independent. Assuming that a < b.10. and this is in H. This means that every such 2cycle (r + 1 r + 2) is in H. So we obtain v= 1 TrE/K t u−σ 1 TrE/K t u . (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). ξ= 6. 6. √ √ As ξ ∈ Q(ζp ). we see that p ∈ Q(ζp ) if p ≡ 1 (mod 4) and p i ∈ Q(ζp ) if p ≡ 3 (mod 4).
106 SOLUTIONS Then we easily ﬁnd that [m] er = e[m−1] + er−1 Xm . h2 = e2 − e2 and h3 = e3 − 2e1 e2 + e3 . r When n = 1 we have sr = X1 and e1 = X1 from which the result follows. while [n] 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−1 k−2 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 ) [n] [n] [n] [n] [n] [n] + (sk−1 − e1 sk−2 + · · · + (−1)k−1 ek−2 s1 + (−1)k kek−1 )Xn+1 [n] k−1 [n] k 2 + (Xn+1 − e1 Xn+1 + · · · + (−1)k−1 ek−2 Xn+1 ) [n] [n+1] k . 1 1 (ii) This can be done by induction on n in a similar way to part (a). Now suppose that [n+1] [n] k the result is true for some n 1. r [m−1] [m] r sr = s[m−1] + Xm . . The desired result is that for all n [n] [n] [n] [n] [1] [n] [n] [n] [1] 1 and k 1. = sk + Xn+1 = sk which demonstrates the inductive step. r [m] Notice also that er = 0 whenever r > m. sk = e1 sk−1 − e2 sk−2 + · · · + (−1)k−1 ek−1 s1 + (−1)k kek . Then sk = sk + Xn+1 . (b)(i) We have h1 = e1 .