Block

4
FIELD THEORY
Block Introduction 111
Notations and Symbols 112
UNIT 11 113
An Introduction to Field Theory
UNIT 12 141
Splitting Fields – Finite and Infinite
UNIT 13
A Gateway to Galois Theory 163
UNIT 14
Fundamental Theorem of Galois Theory 187
 Course Design Committee*
Prof. Gurmeet Kaur Bakshi Prof. Parvati Shastri
Deptt. of Mathematics, Centre for Excellence in Basic Sciences
Panjab University Mumbai University

Dr. Anuj Bishnoi Faculty members

Deptt. of Mathematics, School of Sciences, IGNOU
University of Delhi Dr. Deepika
Prof. K. N. Raghavan Prof. Poornima Mital
Deptt. of Mathematics Prof. Parvin Sinclair
Institute of Mathematical Sciences, Chennai Prof. Sujatha Varma
Prof. Ravi Rao Dr. S. Venkataraman
School of Mathematics
Tata Institute of Fundamental Research
Mumbai

* The course design is based on the recommendations of the Expert Committee for the programme
M.Sc (Mathematics with Applications in Computer Science).

Block Preparation Team

Prof. B. Sury (Editor) Prof. Parvati Shastri
Deptt. of Mathematics Centre for Excellence in Basic Sciences
Indian Statistical Institute, Bengaluru Mumbai University
Dr. Anuj Bishnoi Prof. Parvin Sinclair
Deptt. of Mathematics, School of Sciences
University of Delhi IGNOU

Course Coordinator: Prof. Parvin Sinclair

Acknowledgement: To Dr. S. Venkataraman for his comments on some units, and to Sh. S. S. Chauhan
for the preparation of the CRC of this block.

110
BLOCK INTRODUCTION
The little cares that fretted me,
I lost them yesterday
Among the fields above the sea,
Among the winds at play.

So far you have studied several aspects of groups, rings and domains in this course. You
will now build on that understanding, while studying about fields in this block
comprising five units.

To start with, in Unit 11, you will study about extension fields. You will find several
interesting properties of different kinds of extensions. In particular, we will discuss
finite extensions and algebraic extensions.

In Unit 12, the focus is on irreducible polynomials and their roots. Given such a
polynomial f ( x ) over a field K, you will find that there is an extension field containing
all the roots of f ( x ). Such an extension field is called the splitting field of f ( x ) over
K. We will also discuss finite fields in this context, and their properties, in this unit.

In Unit 13, we will introduce you to Galois theory. This was developed by the famous
French mathematician Evariste Galois, who died very young. The focus on this unit is to
develop a background for the main force of Galois theory, which is discussed in the next
unit. In this context we shall also introduce you to normal extensions, separable
extensions and Galois extensions.

In Unit 14, the focus is on the main theorem of Galois theory. We state this theorem
here, and discuss some of its applications. It is here that you will consider answers to
questions like squaring a circle, or doubling a cube.

With this block we come to the end of our discussion on algebra. Whatever you study
here will be needed throughout your further studies. So, as for the other blocks, please
try and achieve all the objectives of the units in this block.

Talking about fields brings to mind John Charles Fields (1863-1932). He was a
Canadian mathematician who founded the Fields Medal for outstanding contributions in
mathematical research. This medal is given once in 4 years, and is considered by some
to be more prestigious than the Nobel prize. The late Prof. Maryam Mirzakhani and Fig. 1: The Fields
Prof. Akshay Venkatesh, Stanford University, are some of the recent awardees. medal.

111
 NOTATIONS AND SYMBOLS
(Please refer to the notations given in Blocks 1, 2, 3 also.)

Fpr the field with p r elements, p a prime, r ≥ 1

F( x ) the field of rational functions in x , with coefficients in the field F

char F the characteristic of the field F
LF the extension field L over F
[ L : F] the degree of the extension field L over F
k[α1 ,K , α n ] the subring of F generated by k and the elements α1 , K, α n ,
where F k is an extension field
k (α1 , K , α n ) the smallest subfield of F containing k and α1 ,K, α n
m α, k ( x) the minimal polynomial of α over k
deg α the degree of α
f ′( x ) the formal derivative of the polynomial f ( x )
Aut (K ) the group of automorphisms of a field K
Aut (L K ) the group of those automorphisms of a field L which fix K
σ K
the restriction of σ : L → F to K ⊆ L
G (L F) the Galois group of the Galois extension L F
LH the fixed field of H

112
 An Introduction to
UNIT 11 AN INTRODUCTION TO FIELD Field Theory

THEORY
Structure Page No.
11.1 Introduction 113
Objectives
11.2 Fields 113
11.3 Extension Fields 121
11.4 Algebraic Extensions 126
11.5 Summary 134
11.6 Solutions / Answers 134

11.1 INTRODUCTION
In the previous block, the focus was on rings. You also worked with examples
of rings like C, R and Q. As you would recall, these are not just examples of
rings, but also of fields. In this unit, you will study the notion of fields in
detail, as well as related concepts. Cardan’s formula, dating from the 16th
century, used elements of Q, R, C. Lagrange, and others, worked with other
fields. However, the first truly abstract notion of field is due to Dedekind’s
student, the mathematician Weber, who gave the first fully abstract definition
of a field, in 1893.

In Sec.11.2, we will define, and discuss, the concepts of fields, subfields,

prime subfields and the characteristic of a field. You will see that the
characteristic of a field is the same as that of its prime subfield.

In the next section, Sec.11.3, we introduce you to extension fields. In Sec.11.4

the extension fields we look at are algebraic extensions. You will study several
examples and properties of algebraic extensions here.

Objectives
After studying this unit, you should be able to
• define, and give examples of, a field, a subfield and a quotient field of an
integral domain;
• define, and obtain, the prime subfield and characteristic of a given field;
• define extension fields, and find their degrees;
• define, and give examples of, algebraic and transcendental elements over
a field;
• prove some fundamental properties of finite and algebraic extensions.

11.2 FIELDS
In this section, we will first help you recall the definition, and some examples,
of fields. You will also study about subfields, prime subfields and the
characteristic of a field. So, let us begin by recapitulating what a field is.
113
Field Theory Definition: A non-empty set F, with two binary operations (denoted by ‘+’
and ‘· ’), is called a field if the following axioms are satisfied:
i) (F , +) is an abelian group,

ii) (F∗ , .) = (F \ {0}, .) is an abelian group, and

iii) ‘·’ distributes over ‘+’, i.e., a.(b + c) = a.b + a.c ∀ a , b , c ∈ F.

Thus, you can see that a field is a non-zero integral domain in which every
non-zero element is a unit.

Let us consider some examples, and non-examples, of fields which you would
be familiar with.
• Q , R and C are fields with respect to addition and multiplication.

• Z p is a field w.r.t addition and multiplication (mod p), where p is a

prime.
• Z is not a field, since a is not a unit ∀ a ∈ Z \ {0 ,1, − 1}.

• Z 4 is not a field, since, for example, 2 is not a unit.

As you have seen in the examples above, a field can have infinitely many
elements, like C, or finitely many elements, like Z 3 . If it has finitely many
elements, it is called a finite field. In the next unit, you will study such fields
in detail.

Let us now look at an intriguing property of fields.

Proposition 1: Let k be any field, and k ∗ = k \ {0}. Then every finite

subgroup of (k ∗ , ⋅) is cyclic.

Proof: Let G be a finite subgroup of k ∗ . So G is a finite abelian group. Let

m = max{o(g ) g ∈ G}, and let h ∈ G s.t. o(h ) = m. Then m o(G ).
Thus, m ≤ o(G ).

Since o(g) m ∀ g ∈ G (see E2), g m = e for every g ∈ G. Hence, g is a root of

x m − 1 over k ∀ g ∈ G. As x m − 1 can have at most m roots, o(G ) ≤ m.
Hence o(G ) = m = o(h ). That is, G = h .

From Proposition 1, you can conclude, for example, that any finite subgroup of
Q∗ is cyclic, and that Z ∗p is cyclic for any prime p.

Try some exercises now.

E1) Show that if d is a square-free integer, then Q( d ) = {a + b d a , b ∈ Q}

is a field with respect to the usual addition and multiplication. In
114 particular, Q( 2 ) , Q( 3 ) and Q(i) are fields, i = − 1.
 An Introduction to
E2) Let G be a finite abelian group, and let m = max {o(g) g ∈ G}. Then Field Theory
prove that o(g) m ∀ g in G.

Let us now recall what the ideals in a field look like. You would recall, from
Block 3, that if I is an ideal of R and 1 ∈ I, then I = R. Now if F is a field
and I is a non-zero ideal of R , let x ∈ I. Then x −1 ∈ F. So xx −1 = 1 ∈ I, and
hence I = F. Thus, a field F has only two ideals, {0} and F.

Try the following exercises now.

E3) Prove Theorem 1 of Unit 9, namely, an ideal I in a commutative ring R

with identity is a maximal ideal iff R I is a field.

E4) Show that if φ : F → F′ is a homomorphism between two fields, then φ

is 1-1 or φ is the zero map.

For some more examples of fields, let us look at the notion of ‘quotient field’,
also called ‘field of fractions’. Here we generalise the procedure of obtaining
the quotient field of Z, which you learnt in your undergraduate studies.

Theorem 1: Every integral domain can be embedded in a field, i.e., for every
integral domain R , there exists a field F and a monomorphism from R into
F. Further, every element of F is of the form rs −1 , where r, s ∈ R , s ≠ 0.

Proof: Let R be an integral domain R. Let K = {(a , b) a , b ∈ R and b ≠ 0}.

Let us define a relation ~ in K by ‘ (a , b) ~ (c , d ) if ad = bc. ’
You can check that ~ is an equivalence relation (see E5).
Let us denote the equivalence class that contains (a , b) by [a , b]. Thus,
[a , b] = {(c , d) c , d ∈ R , d ≠ 0 and ad = bc}.

Let F be the set of all equivalence classes of K with respect to ~.

Let us define + and · in F as follows. (This is analogous to the rules for adding
and multiplying rational numbers.)
[a , b] + [c , d ] = [ad + bc , bd ], and
[a , b] ⋅ [c , d] = [ac , bd ].
Note that b ≠ 0 and d ≠ 0 in the integral domain R , so that bd ≠ 0. So, the
right-hand sides of the equations given above are well-defined equivalence
classes.

Let us now check that + and · are binary operations on F.

So, let [a , b] = [a ′ , b′] and [c , d] = [c′ , d′]. We have to show that

[a , b] + [c , d ] = [a ′ , b′] + [c′ , d′], and [a , b] ⋅ [c , d] = [a ′ , b′] ⋅ [c′ , d′].

Now, (ad + bc )b′d′ − (a ′d′ + b′c′)bd

= ab′dd′ + cd′bb′ − a ′bdd′ − c′dbb′
115
Field Theory = (ab′ − a ′b)dd′ + (cd′ − c′d)bb′
= (0)dd′ + (0)bb′, since (a , b) ~ (a ′ , b′) and (c , d ) ~ (c′ , d′)
= 0.
Hence, [ad + bc , bd ] = [a ′d′ + b′c′ , b′d′], i.e., + is well-defined.

Next, (ac)(b′d′) − (bd )(a ′c′) = ab′cd′ − ba ′dc′

= 0, since ab′ = ba ′ and cd′ = dc′.
Therefore, [ac , bd ] = [a ′c′ , b′d′]. Hence, · is well-defined.

We will together prove that (F , + , ⋅) is a field.

i) + is associative: For [a , b] , [c , d ] , [u , v] ∈ F, you should check that

([a , b] + [c , d ]) + [u , v] = [a , b] + ([c , d ) + [u , v]).

ii) + is commutative: For [a , b] ,[c , d] ∈ F, you should check that

[a , b] + [c , d ] = [c , d] + [a , b].

iii) [ 0 ,1] is the additive identity for F : For [a , b] ∈ F,

[0 ,1] + [a , b] = [0.b + 1.a ,1.b] = [a , b].

iv) The additive inverse of [a , b] ∈ F is [−a , b], since [0 ,1] = [0 , b 2 ].

You should check that the rest of the requirements for F to be a field also hold
(see E6).

Now, let us define f : R → F : f (a ) = [a ,1]. f is well-defined since for

a , b ∈ R , a = b ⇒ [a ,1] = [b ,1] ⇒ f (a ) = f (b).

Next, for a , b ∈ R , f (a + b) = [a + b ,1] = [a ,1] + [b ,1]

= f (a ) + f (b), and
f (ab) = [ab ,1] = [a ,1] ⋅ [b , 1]
= f (a ) ⋅ f (b).
Thus, f is a ring homomorphism.

Finally, let a , b ∈ R such that f (a ) = f (b). Then [a ,1] = [b ,1], i.e., a = b.

Therefore, f is 1-1.
Thus, f is a monomorphism.

So, Im f = f (R ) is a subring of F which is isomorphic to R. As you know,

isomorphic structures are algebraically identical. So, we can identify R with
f (R ), and think of R as a subring of F.

Now, any element of F is of the form

[a , b] = [a ,1][1, b] = [a ,1][b ,1]−1 = f (a )f (b) −1, where b ≠ 0. Thus, identifying
x ∈ R with f ( x ) ∈ f (R ), we can say that any element of F is of the form
ab −1 , where a , b ∈ R , b ≠ 0.

In fact, the field in Theorem 1 is the smallest field containing R in the sense
116
that if K is any field which contains R , then K must contain ab −1 ∀ a , b ∈ R .
So K contains a field isomorphic to F. Ignoring the isomorphism, we can An Introduction to
Field Theory
α 
assume that the field F =  α , β ∈ R , β ≠ 0 in Theorem 1 is the smallest
β 
field containing R. This leads us to the following definition.

Definition: Given an integral domain R , the smallest field containing it is

called its quotient field, or its field of fractions.
For example, Q is the quotient field of the ring of integers Z. Let us consider
another example.

Example 1: Find the quotient field of F[ x ], where F is a field.

Solution: You know that F[ x ] is an integral domain with respect to the usual
addition and multiplication of polynomials. Its quotient field, denoted by
F( x ), is the field of rational functions in x , i.e.,
F( x ) = {f (x ) g( x ) f ( x )∈ F[ x ], g ( x ) ≠ 0}.
***

Try some exercises now, including those required to complete the proof of
Theorem 1.

E5) Show that ~, in the proof of Theorem 1, is an equivalence relation.

E6) Show that + and · in F (in the proof of Theorem 1) are associative,
commutative, · is distributive over +, [1,1] is the multiplicative identity
for F, and [a , b]−1 = [b , a ] ∀ [a , b] ≠ [0 ,1].

E7) Find the quotient field of a field F.

E8) Check whether or not the quotient field of Z[ 2 ] is R.

E9) Find the quotient field of R[ x ] .

x2 + 1

Let us now look at a concept analogous to that of a subring.

Definition: A non-empty subset E of a field F is called a subfield of F if E

is a field under the induced operations of addition and multiplication on F. If
E ⊆/ F, E is called a proper subfield of F.

The following examples would have immediately come to your mind.

• Q is a subfield of both R and C, and R is a subfield of C.

• Q( 2 ) is a subfield of R as well as C. However, Q(i), where i 2 = −1,

is a subfield of C but not a subfield of R.

• Z p is a subfield of Z p ( x ), where p is a prime.

117
Field Theory Let us now consider an important subfield.

Proposition 2: Every field F has a unique least subfield, that is, a subfield
contained in every subfield of F.

Proof: Let R = {n .1 n ∈ Z}. This is a non-trivial subring of F containing 1,

and hence it must be an integral domain. Let K denote the quotient field of R.
Then K = {(n .1) (m.1) n , m ∈ Z , m.1 ≠ 0} and K ⊆ F.

Now, let L be any subfield of F. Then 1 ∈ L, and hence n .1 ∈ L for every

n ∈ Z. So the ring R is contained in the field L. Hence its quotient field K is
contained in L. Thus, K is contained in every subfield of F. This shows that
K is a least subfield of F.

Moreover, if E is another least subfield of F, then from what we have shown

above, K ⊆ E. But, by the minimality of E, we must have E ⊆ K. So K = E,
and hence K is the unique least subfield of F.

The subfield found in Proposition 2 has a very special place in field theory. Let
us formally define it.

Definition: The unique least subfield of a field F is called the prime subfield
of F.

Let us consider some examples.

Example 2: Find the prime subfield of C.

Solution: The prime subfield of C is

 n ⋅1  =  n n , m ∈ Z, m ≠ 0 = Q.
 m ⋅1 n , m ∈ Z, m ≠ 0   
  m 
***

Example 3: Let K be the prime subfield of a field F1 , and let F2 be a field

such that F1 ⊆ F2 . Then show that K is the prime subfield of F2 .

Solution: Let L be any subfield of F2 . Then L ∩ F1 is a subfield of F1. Hence

K ⊆ L ∩ F1 ⊆ L.
Thus, K is the least subfield of F2 also, that is, its prime subfield.
***

From Examples 2 and 3, you can see that Q is the prime subfield of Q , R and
Q( d ) (where d is a square-free integer). But what about fields that do not
contain Q ? You will find an answer to this later. For now try the following
exercises.

E10) Prove or disprove the statement, ‘Every subring of a field F is a subfield

118 of F ’.
E11) Show that given a field F, a subfield K is the prime subfield of F if any An Introduction to
Field Theory
one of the following holds:

i) K is the least subfield of F.

ii) K is generated by 1F , the identity element of F.

iii) Every subfield of F contains K.

E12) Check whether or not the intersection of a non-empty family of subfields

of a field F is a subfield of F.

E13) If F1 and F2 are fields, will F1 ∪ F2 be a field? Why, or why not?

E14) Show that Z p is the prime subfield of Z p ( x ).

We will now move towards relating prime subfields to another concept, which
we now define.

Definition: The characteristic of a field F is defined to be

i) the least positive integer n such that n .1 = 0 in F, if such an n exists;
ii) 0, otherwise.
It is denoted by char F .

Let us consider some examples.

• Q is of characteristic zero since n ⋅ 1 ≠ 0 for any n ∈ N. Similarly,

char R = 0, char C = 0.

• For any prime p, Z pZ is of characteristic p, and char (Z pZ ) ( x ) = p

since p ⋅1 = 0, and n ⋅ 1 ≠ 0 for any n < p.

The examples above may have given you a hint about what we are now going
to prove.

Proposition 3: The characteristic of a field is either zero or a prime number.

Proof: Let F be a field such that char F = n ≠ 0. We will prove that n is a

prime, by the method of contradiction.

So, assume that n is not a prime, say n = rs, with 1 < r < n , 1 < s < n. Then
n ⋅ 1 = 0 ⇒ rs ⋅ 1 = 0 ⇒ (r ⋅1)(s ⋅ 1) = 0 ⇒ r ⋅ 1 = 0 or s ⋅1 = 0, since F is without
zero divisors.
However, n is the least positive integer such that n ⋅ 1 = 0. So, we reach a
contradiction.
Hence n must be a prime.

Thus, char F = 0 or char F = p, for some prime p.

The examples, and proposition, above may give you a hint about the
relationship between the prime subfield of a field and its characteristic, which
we now state and prove. 119
Field Theory Theorem 2: Let F be a field. If char F = 0, then the prime subfield of F is
The only prime subfields isomorphic to Q, and if char F = p, then the prime subfield of F is isomorphic
are Q and Z p , p a to Z pZ.
prime.
Proof: Let R = {n .1 n ∈ Z}. Then by Proposition 1, R is a subring of F and
its quotient field K is the prime subfield of F.
Define φ : Z → R by φ(n ) = n .1. Then you should check that φ is a well-
defined ring epimorphism. Also, Ker φ = {n ∈ Z n.1 = 0}.
So, by the Fundamental Theorem of Homomorphism, Z ~
− R.
Ker φ
Now two cases arise:

Case 1 (Char F = 0) : In this case, n .1 = 0 ⇒ n = 0.

So Ker φ = {0}, and hence, Z ~ − R.
This isomorphism can be extended to Q by defining φ(m n ) = (m.1) (n .1) , to
obtain an isomorphism between Q and the quotient field K of R. Thus, in
this case we have Q ~
− K, the prime subfield of F.

Case 2 (Char F = p) : In this case, p .1 = 0, for some prime p. p is the

smallest such positive integer. ∴ Ker φ = pZ. Hence, Z pZ − ~ R.
Since Z pZ is a field, R must also be a field. Hence, R coincides with its
quotient field K, i.e., R = K. Thus, Z pZ ~−K.

From Theorem 2, it follows that the characteristic of a field is the same as

that of its prime subfield.

Let us look at another example involving these concepts.

Example 4: Find the characteristic, and prime subfield, of Q( − 5 ).

Solution: Since Q ⊆ Q( − 5 ), char Q( − 5 ) = 0, and its prime subfield is Q.

***

Try some exercises now.

E15) Find the smallest subfield of R that contains 2 . Also, find its
characteristic.

E16) Let k be a subfield of F. Prove that F and k have the same

characteristic.

E17) Let k be a field and σ be an automorphism of k. Prove that σ keeps

the prime subfield of k elementwise fixed.

We will now move on to consider fields that have a given field as a subfield.
120
 An Introduction to
11.3 EXTENSION FIELDS Field Theory

In the previous section, you have studied the concept of a subfield. For
instance, Q is a subfield of R. In such a situation, we also say that R is an
extension field of Q, according to the following definition.

Definition: A field F is called an extension field of a field k (or an extension

of k ) if k is a subfield of F. This is denoted by F k . Here k is called the
base field, or the ground field, of the extension.

For example, C is an extension of both R and Q. Also, as you have seen, A subfield of C is called
every field is an extension of its prime subfield. a number field.

Now, if F k is an extension, then k is a subfield of F. So F can be

considered as a vector space over k. Therefore, the following definition makes
sense.

Definition: The degree of an extension field F k is the dimension of the

k-vector space F, and is denoted by [F : k]. It is also called the degree of F
over k.
F is called a finite extension of k if [F : k ] is finite, and an infinite extension
of k otherwise.

Let us look at some examples.

Example 5: Find [C : R], [R : Q],[R ( 2 ) : R ] and [Q( d ) : Q] (where d is a

square-free integer).

Solution: [C : R ] = 2, since {1, i} is a basis of the R -vector space C.

For finding [R : Q], note that Q is countable. So any finite-dimensional
vector space over Q will be countable. But R is not countable. ∴[R : Q] is
not finite.
Since 2 ∈ R, R( 2 ) = R. ∴[R( 2 ) : R] = [R : R ] = 1.
Q( d ) has degree two over Q, with a basis being {1, d }.
***

Try some exercises now.

E18) Let F k be an extension field and σ be a homomorphism of F into a

field E. Show that [σ(F) : σ(k )] = [F : k ], i.e., σ preserves the degree of
If F k and F′ k are
the extension. two extensions, a
k-homomorphism
E19) Let F k be an extension field of finite degree. Show that every k- σ : F → F′ is a
endomorphism of F is an automorphism of F. Give an example, with homomorphism such
justification, to show that this may fail to hold for arbitrary extensions. that σ is the identity
k
map.

Now, let us look at a particular type of extension. Let F k be an extension

field, and let S be a subset of F. We will denote by k[S] the smallest subring
of F containing k and S, i.e., the subring of F generated by the elements of 121
Field Theory k and S. We will also denote by k(S) the smallest subfield of F containing k
and S.
In particular, if S = {α1 , α 2 ,K , α n }, we write k[α1 , α 2 , K , α n ] for k[S] and
k(α1 , α 2 ,K , α n ) for k (S).
An example of this that we will be looking at in detail is the case when
S = {α}. Then k[S] = k[α] and k (S) = k (α).
This leads us to the following definitions.

Definitions: Let F k be an extension field.

i) F is said to be finitely generated over k if there exist
α1 , α 2 ,K , α n ∈ F such that F = k (α1 ,K , α n ).
ii) If there exists α ∈ F such that F = k (α), then F is called a simple
extension of k, and α is called a primitive element of F over k.

Note that a primitive element is not unique.

For example, 2 and − 2 are both primitive elements of Q( 2 ) over Q.

To understand finitely generated extensions, consider the following

propositions.

Proposition 4: Let F k be an extension field and S a subset of F. Then k (S)

is the field of fractions of k[S].

Proof: Let L denote the field of fractions of k[S] in F. Then L is the smallest
subfield of F containing the subring k[S], and hence contains both k and S.
Also k[S] ⊆ k (S). Therefore, L ⊆ k (S). But, by definition, k (S) is the smallest
subfield of F which contains both k and S. Therefore, k (S) ⊆ L.
Thus, L = k (S).

Proposition 5: Let F k be an extension field and α∈ F. Then

k[α ] = {p(α ) p( x ) ∈ k[ x ]}, and

k (α ) = p(α ) p( x ), q( x ) ∈ k[ x ], q (α ) ≠ 0.
 q ( α ) 

Proof: Let R = {p(α) p(x ) ∈ k[ x ]} = {a 0 + a 1α + L + a n α n a i ∈ k, n ≥ 0}. Then

R is a subring of F containing k and α.

Further, if T is any subring of F containing k and α, then T contains all

elements of the form a 0 + a1α + L + a n α n , where a i ∈ k and n ≥ 0. Therefore,
R ⊆ T. Hence, R is the smallest subring of F containing k and α, that is,
R = k[α], i.e., k[α] consists of all the elements of F that can be expressed as
a polynomial in α with coefficients from k.

Now, k (α) is the quotient field of k[α], and hence consists of all elements of
F of the form p(α) q (α) , q (α) ≠ 0, p( x ) , q (x ) ∈ k[ x ].

Consider the following comment about a generalisation of Proposition 5.

122
 An Introduction to
Remark 1: If S = {α1 , α 2 ,K, α n }, then it can be shown, as in Proposition 5,
Field Theory
that k[α1 , α 2 ,K , α n ] consists of all elements of F that can be expressed as
polynomials in α1 , α 2 ,K , α n with coefficients from k, and k (α1 , α 2 ,K, α n )
is its quotient field.

Try some exercises now.

E20) Show that, if F is a field, then F( x ) F is an infinite extension, where x

is a variable.

E21) Let F k be an extension field and α, β ∈ F. Then show that

k (α , β) = k (α)(β) = k (β)(α). [Similarly,
k (α1 , K , α n ) = k (α1 , K , α n −1 )(α n ) for α1 ,K , α n ∈ F.]

Let us now consider properties of finite extensions. The first is the so-called
‘tower theorem’.
E
Theorem 3: Let E be an extension field of F and F be an extension field of
k (see Fig. 1). Then [E : k ] is finite if and only if both [E : F] and [F : k ] are
finite. Further, if E k is finite, [E : k ] = [E : F][F : k ]. F

Proof: Suppose first that both E F and F k are finite, with bases
{v1 , v 2 ,K, v n } and {u1 , u 2 ,K , u m }, respectively. We shall prove that the set k
Fig. 1: A tower of extensions!
{u i v j 1 ≤ i ≤ m;1 ≤ j ≤ n} is a basis of the k-vector space E, which will then
prove that
[E : k ] = [E : F][F : k ]. …(1)
Let x ∈ E. Since {v1 , v 2 ,K, v n } spans E over F, we have
n
x = ∑ α jv j , where α j ∈ F ∀ j = 1, K, n. …(2)
j =1

Since α j ∈ F for each j ∈ {1, 2 ,K , n}, we write α j in terms of {u1 , u 2 ,K , u m },

m
i.e., α j = ∑ a iju i , where a ij ∈ k ∀ i = 1, K , m; j = 1, K, n.
i =1

Substituting for α j , in Eqn. (2), we get

n
 m 
x = ∑  ∑ a iju i  v j
j=1  i =1 
n m
= ∑∑ a iju i v j .
j=1 i =1

This shows that the mn elements u i v j span E over k.

n m
For checking linear independence, assume that ∑∑ b u v
j=1 i =1
ij i j = 0, for some
n
m 
bij ∈ k, i.e., ∑  ∑ b ij u i  v j = 0.
j=1  i =1 
123
Field Theory m
Since ∑b u
i =1
ij i ∈F ∀ j, the linear independence of {v1 , v 2 ,K, v n } over F
m
implies that ∑b u
i =1
ij i = 0 for each j. Now as b ij ∈ k , the linear independence

of the u js over k tells us that b ij = 0 for each i and j.

Thus, {u i v j 1 ≤ i ≤ m;1 ≤ j ≤ n} is linearly independent, and hence a k-basis of
E, which proves Eqn. (1).

Conversely, let E k be a finite extension. Since k ⊆ F, the elements of E

which are linearly independent over F are also linearly independent over k.
The finiteness of [E : k ] implies that [E : F] is also finite.
Now, as F is a subfield of E containing the field k, the field F can be
considered as a subspace of the k-vector space E. Therefore, the finiteness of
the extension E k implies that [F : k ] is finite.

Another important property of the degree of a finite extension is proved below.

Theorem 4: Let F be a field, and f ( x ) ∈ F[x ] be irreducible over F. Then

E = F[ x ] is an extension field of F, with [E : F] = deg f ( x ).
< f (x ) >

Proof: Since f ( x ) is irreducible over F, F[ x ] is a field.

< f (x) >
Next, consider the canonical map π : F[x ] → F[x ] : π(g( x )) = g( x ).
< f (x) >
This is a well-defined ring homomorphism.
Consider its restriction, φ : F → F[ x ] : φ(a ) = a , which is a
< f (x) >
homomorphism between fields, and hence φ is 1-1 or φ ≡ 0 (from E4).
Since φ is not the zero map, φ is 1-1, so that F ~ − φ(F) O F[x ] .
< f (x) >
Identifying F with φ(F), we see that F[ x ] is an extension field of F.
< f (x ) >
Finally, you should check that {1 , x ,K, x n−1} forms an F-basis of
E = F[ x ] , where n = deg f (x ).
< f (x) >
Hence, [E : F] = n = deg f ( x ).

As you can see, Theorem 4 gives us infinitely many examples of finite

extensions. For example, since x 2 + a is irreducible over R ∀ a ∈ R + ,
R[ x ] is an extension of R of degree 2 ∀ a > 0.
< x2 + a >
Let us look at one example to help you understand how the elements of such
extensions behave.

Example 6: Consider ( x 3 − 7) ∈ Q[ x ]. Let α ∈ C be a root of ( x 3 − 7) (in C).

Find (1 + α) −1.

Solution: Q[ x ] ~ {a + bx + cx 2 a , b , c ∈ Q}.
−
124 < x3 − 7 >
 An Introduction to
Now α3 = 7. Also, since (1 + x , x 3 − 7) = 1, by the Euclidean algorithm Field Theory
∃ f ( x ), g ( x ) ∈ Q[x ] such that
(1 + x )f (x ) + (x 3 − 7)g (x ) = 1, …(3)
1 1
In fact, here you can explicitly find f ( x ) = ( x 2 − x + 1) and g( x ) = − .
8 8
Now, putting x = α in (3), we get
(1 + α)f (α) = 1
1
i.e., (1 + α) −1 = f (α) = (α 2 − α + 1).
8
***

Try an exercise now.

E22) i) Let α ∈ C \ R be a root of the irreducible polynomial

x 3 − 3x + 4 ∈ Q[ x ]. Find the inverse of α 2 + α + 1 in Q[α].
ii) Let α be a root of an irreducible polynomial
a 0 + a1 x + L + a n −1 x n −1 + x n ∈ k[α], k a field, α ∉ k.
Find α −1∈ k[α] explicitly in terms of a i , i = 1,K, n.

E23) Show that Q( 2 + 3 ) = Q( 2 , 3 ), and this is an extension of Q of

degree 4.

Continuing our discussion on finite extensions, consider an interesting thing

that happens if F k is a finite field extension. In this case we see that k[α]
and k[α1 , K , α n ] are actually fields, and hence they coincide with their
respective quotient fields. We will now prove this. But first, we need to prove
a result that will be used in the proof.

Lemma 1: Let R be an integral domain containing a field k (as a subring). If

R is finite-dimensional, when regarded as a k-vector space, then R is a field.

Proof: To show that R is a field, we only need to show that every non-zero
element of R is invertible. So, let α ≠ 0 ∈ R and consider
L α : R → R : L α ( x ) = αx , x ∈ R .
Then, you can check that Lα is an injective linear transformation from R to
itself. Since R is a finite-dimensional k-vector space, Lα must be surjective
also. This means that every element of R is of the form αx for some x ∈ R .
In particular, there exists β ∈ R such that αβ = 1, i.e., α is invertible in R ,
and hence the lemma follows.

Now let us state and prove the theorem.

Theorem 5: Let F k be a finite extension of fields, and S be a subset of F.

Then k[S] is a field.

Proof: The subring k[S] is a finite-dimensional k-vector space, being a

subspace of the finite-dimensional k-vector space F. So k[S] satisfies the
hypothesis of Lemma 1. Therefore, k[S] is a field. 125
Field Theory In the next section you will see this result used, for simple extensions
particularly. However, a word of warning follows.

Remark 2: Theorem 5 need not hold for infinite extensions. For instance,
Q(x ) Q is infinite, x ∈ Q( x ) and Q[ x ] is not a field.

Why don’t you try some exercises now?

E24) i) If E is a finite extension of k and F is a subfield of E containing

k, then show that [F : k ] [E : k ].

ii) Use (i) above to show that if E k is an extension of degree p , p a

prime number, then there are no fields properly lying between k
and F.

E25) Q has a finite extension of degree n ∀ n ∈ N. True, or false? Why?

E26) Prove that if F k is an extension of prime degree, then F k is simple. Is

the converse true? Give reasons for your answer.

Let us now focus on a particular kind of extension, examples of which are

galore, as you will see.

11.4 ALGEBRAIC EXTENSIONS

So far you have seen different types of extensions – finite, infinite, simple. In
this section we will look at certain simple extensions to start with. But first,
some definitions.

Definitions: Let F k be an extension of fields.

i) α ∈ F is called algebraic over k if there is a non-zero polynomial f ( x )

in k[ x ] such that f (α) = 0.

ii) If α ∈ F is not algebraic over k, then α is called transcendental over

k.
iii) F k is called an algebraic extension if every element of F is algebraic
over k.

For example, i ∈ C is algebraic over R, since it is a root of x 2 + 1. Also, note

that the real numbers e and π are examples of transcendental elements over
Q. (We shall not prove this here though.) Hence R Q is not an algebraic
extension.

Let us consider an example of an algebraic extension.

Example 7: Show that C R and Q( p ) Q are algebraic extensions, where

126
p is a prime.
 An Introduction to
Solution: First, let us consider C R. Let a + ib ∈ C, with a , b ∈ R. Then
Field Theory
(a + ib ) satisfies x 2 − 2ax + (a 2 + b 2 ) = 0 ∈ R[ x ]. Hence (a + ib ) is algebraic
over R. Since (a + ib ) was an arbitrary element of C, C R is an algebraic
extension.

Next, for any prime p, any element of Q( p ) is of the form

a + b p , a , b ∈ Q.
Since (a + b p )2 = pb 2 − a 2 + 2a (a + b p ),
a + b p satisfies x 2 − 2ax + a 2 − pb 2 ∈ Q[ x ].
Hence a + b p is algebraic over Q. Since a + b p was an arbitrary element
of Q( p ), this shows that Q( p ) Q is algebraic.

***

Try some exercises now.

E27) Show that k k is an algebraic extension, for any field k.

E28) Show that if k ⊆ E ⊆ F are fields, and α ∈ F is algebraic over k, then

α is algebraic over E.

Now, consider 2 . It satisfies x 2 − 2 ∈ Q[ x ]. But is also satisfies

x 3 − x 2 − 2x + 2 ∈ Q[x ]. Of these two polynomials, the first is irreducible over
Q. So, the question is: Does every algebraic element over a field k satisfy an
irreducible polynomial over k ? If so, is there any point in looking for such a
polynomial? The next two theorems will answer such questions.

Theorem 6: Let F k be an extension field, and let α ∈ F be algebraic over k.

Then there exists a unique least degree non-zero irreducible monic polynomial
p( x ) ∈ k[ x ] s.t. p(α) = 0. Further, if f (α) = 0 for some polynomial
f ( x ) ∈ k[x ], then p( x ) f ( x ).

Proof: Since α is algebraic over k, there exists a non-zero polynomial g( x )

in k[ x ] such that g(α ) = 0. This shows that the set S = {f ( x ) ∈ k[ x ] f (α) = 0}
is non-empty. So we can find p( x ) ∈ S of smallest degree, say n. Multiplying
p( x ) by the inverse of its leading coefficient (if necessary), we can assume
that p( x ) is a monic polynomial. Thus, there exists a monic polynomial p( x )
over k, of least degree, satisfied by α.

Irreducibility: Suppose, to the contrary, that p( x ) is reducible over k. Then

there exist non-constant polynomials a ( x ) , b( x ) ∈ k[ x ] with
deg a ( x ) , deg b( x ) < n such that p( x ) = a ( x )b( x ). Then
p(α) = 0 ⇒ a (α)b(α) = 0 ⇒ a (α) = 0 or b(α) = 0.
⇒ a (x ) ∈ S or b( x ) ∈ S.
127
Field Theory In either case, this contradicts the minimality of n. Hence, p( x ) must be
irreducible.

Uniqueness: Suppose q( x ) ∈ S is another such irreducible monic polynomial

of least degree n. Then h ( x ) = p( x ) − q ( x ) is a non-zero polynomial
(Q p( x ) ≠ q( x )) of degree less than n such that h (α) = p(α) − q (α) = 0. This
contradicts the minimality of n. Hence our assumption must be wrong. That
is, p( x ) is the unique monic polynomial of least degree satisfied by α.

Finally, let f ( x ) ∈ S. Then, by the division algorithm in k[ x ], there exist

polynomials q( x ) and r( x ) in k[ x ] such that f ( x ) = p( x )q( x ) + r (x ), where
either r( x ) = 0 or deg r( x ) < deg p( x ).
Now if r( x ) ≠ 0, then f (α) = 0 implies that r(α) = 0. But
deg r( x ) < deg p(x ), which again contradicts the choice of p( x ). So r( x ) = 0,
and hence f ( x ) = p( x )q( x ), i.e., p( x ) f ( x ).
Thus, p( x ) f ( x ) ∀ f ( x ) ∈ S.

The situation in Theorem 6 leads us to the following definition.

Definition: Let F k be a field extension, and let α in F be algebraic over k.

The unique monic polynomial of least degree of which α is a root is called
the minimal polynomial of α over k , and is denoted by m α , k (x), or simply
m α (x), if k is understood.
Further, the degree of m α (x ) is called the degree of α over k, denoted by
deg α.

For example, mi , R ( x ) = x 2 + 1, so that deg i = 2 over R. Similarly, the degree

of d over Q is 2, where d is a square-free integer.

Let us prove an important result, which relates [k (α) : k ] with deg m α ( x ).

This result actually led to the term ‘degree of an extension’ being used more
generally.

Theorem 7: Let F k be an extension field and α ∈ F be algebraic over k.

Then the following holds:

i) k[ x ] m α ( x ) ~
− k[α];

ii) k[α] = k (α);

iii) [k[α] : k ] = deg α.

Proof: Let φ α : k[ x ] → F : φ α (f ( x )) = f (α), be the evaluation map at α. You

should check that φα is a well-defined ring homomorphism. Let deg α = n.
i) Since α is algebraic over k, Ker φα is a non-trivial ideal of k[ x ]. We
claim that Ker φ α = m α (x ) .
Since m α (α) = 0, and Ker φ α consists of all the non-zero polynomials
128
 An Introduction to
in k[ x ] satisfied by α, we see that m α ( x ) ∈ Ker φα . Hence
Field Theory
m α (x ) ⊆ Ker φ α .
Now, let g( x ) ∈ Ker φα , then g(α) = 0 and so p( x ) g (x ) in k[ x ].
Consequently, g( x ) ∈ m α (x ) , and hence Ker φα ⊆ m α (x ) , proving
the claim. Now, by the Fundamental Theorem of Ring Homomorphisms,
k[ x ] ~
− Im φα = k[α].
mα (x )

ii) By (i) above, k[ x ] ~

− k[α]. Since m α ( x ) is irreducible,
mα (x )
k[ x ] is a field. Consequently, k[α] is a field. Therefore, it must
mα (x )
coincide with its quotient field, i.e., k[α] = k (α).

iii) To prove that [k[α] : k ] = n, we show that the set {1, α ,K , α n−1} forms a
k-basis of k[α]. Since Im φα = k[α], an arbitrary element of k[α] has
the form g(α) for some polynomial g( x ) ∈ k[ x ]. By the division
algorithm, there exist polynomials q( x ) , r ( x ) ∈ k[x ] such that
g( x ) = m α ( x )q( x ) + r ( x ), where either r( x ) = 0 or
deg r (x ) < deg m α (x ) = n. As m α (α) = 0, g(α) = r(α), and hence, g(α)
is a k-linear combination of 1, α ,K , α n−1 .
If {1, α ,K , α n−1} were linearly dependent, then there would exist scalars
a 0 , a1 ,K, a n −1 , not all zero, in k such that a 0 + a 1α + L + a n−1α n −1 = 0.
This would imply that α satisfies a non-zero polynomial
f ( x ) = a 0 + a 1 x + L + a n −1 x n −1 over k of degree less than that of m α (x ),
which is not possible.
Thus, {1, α ,K , α n−1} is linearly independent, and forms a k-basis of
k[α].
∴ [k[α] : k ] = n = deg α.

Using the theorem above, you can see that R[ x ] ~ R[ − 1] = C.

−
x2 +1

Similarly, Q[ x ] ~
− Q( 2 ). Here, − 1 is of degree 2 over R, and 2
x2 − 2
is of degree 2 over Q.

Let us do some examples involving the minimal polynomial.

Example 8: Show that x 4 − 10x 2 + 1 is the minimal polynomial of 2+ 3

over Q.

Solution: Firstly, 2 + 3 satisfies the given polynomial. Next, from E23,

you know that Q( 2 + 3 ) Q is of degree 4, which is deg ( x 4 − 10x 2 + 1).
Hence, x 4 − 10x 2 + 1 is irreducible, and is the minimal polynomial of 2+ 3
over Q.
*** 129
Field Theory Example 9: Let F k be a field extension and α , β ∈ F be algebraic over k,
with minimal polynomials f ( x ) and g( x ), respectively. Prove that f ( x ) is
irreducible over k (β) if and only if g( x ) is irreducible over k (α).

Solution: Let m and n denote, respectively, the degrees of f ( x ) and g (x ).

Then [k (α) : k ] = m and [k (β) : k ] = n.
Suppose, first, that f ( x ) is irreducible over k (β). As k ⊆ k (β), f ( x ) remains
the minimal polynomial of α over k (β), and hence [k (β)(α) : k (β)] = m. Now,
[k (α, β) : k ] = [k (β)(α) : k ] = [k (β)(α) : k (β)][k (β) : k ] = mn.
So, mn = [k (α, β) : k ] = [k (α)(β) : k (α)][k (α) : k ], and [k (α) : k ] = m.
Thus, [k (α)(β) : k (α)] = n , i.e., the minimal polynomial of β over k (α) is of
degree n. Also, g( x ) ∈ k (α)[ x ] s.t. g(β) = 0, so that mβ , k ( α ) (x ) g ( x ). But both
have the same degree. Hence, g( x ) is irreducible over k (α).
You can use a similar argument to prove the converse.
***

Try the following exercises now.

E29) i) Find the minimal polynomials of ω over Q , R and C,

respectively, where ω is a cube root of unity. Further, show that
m ω, C (x ) m ω, R ( x ) in C[ x ] and m ω, R ( x ) m ω, Q ( x ) in R[ x ].

ii) Show that if k ⊆ E ⊆ F are fields, and α ∈ F is algebraic over k,

then m α , E ( x ) m α , k (x ) in E[x ].

E30) Let F k be an extension field, and let α ∈ F. Show that k[α] is a field if
and only if α is algebraic over k.

E31) Give an example each of an extension of Q in which π is algebraic and

in which π is transcendental, respectively.

Let us now use Theorem 7 to prove a very useful property of algebraic

elements, that we now state.

Theorem 8: Let F k be an extension field, and α ∈ F. Then k (α) k is a

finite extension if and only if α is algebraic over k.

Proof: Suppose first that k (α) k is a finite extension of degree n, i.e.,

[k (α) : k ] = n. Consider the set {1, α ,K , α n } consisting of n + 1 elements.
This must be linearly dependent over k. So there exist a 0 , a1 ,K , a n ∈ k , not all
zero, such that a 0 + a1α + L + a n α n = 0. Thus, α satisfies the non-zero
polynomial f ( x ) = a 0 + a 1x + L + a n x n over k, i.e., α is algebraic over k.
The converse follows from Theorem 7.

An immediate corollary to Theorem 8 is the following.

130 Corollary 1: Every finite extension is an algebraic extension.

 An Introduction to
Proof: Let F k be a finite extension, and α ∈ F. Then k (α) is a subspace of
Field Theory
the finite-dimensional vector space F over k. ∴[k (α) : k ] ≤ [F : k ] < ∞. Hence,
by Theorem 9, α is algebraic over k. Since α was an arbitrary element of
F, F k is algebraic.

We can apply Corollary 1 to note that the extension R Q is not finite, since it
is not an algebraic extension.
The converse of the result in the corollary above is not true. For a counter-
example we require the following property about the set of all algebraic
elements in an extension field.

Proposition 6: Let F k be an extension field. Let

E = {α ∈ F α is algebraic over k}. Then E is a subfield of F.

Proof: Firstly, k ⊆ E, since k k is algebraic.

Next, let α, β ∈ E and let n , m denote, respectively, their degrees over k.
Then, by Theorem 7, [k (α) : k ] = n and [k (β) : k ] = m. Since β is algebraic
over k and k ⊆ k (α), β is also algebraic over k (α). Hence,
[k (α)(β) : k (α)] ≤ m. Now, by Theorem 3, it follows that
[k (α)(β) : k ] = [k (α)(β) : k (α)][k (α) : k ] ≤ mn .
So k (α)(β) k is a finite extension, and hence an algebraic extension. Since
k (α)(β) is a field containing α and β, it contains α ± β , αβ and α β (for
β ≠ 0 ). Hence all these elements are algebraic over k, and thus, lie in E.
Thus, E is a subfield of F.

Now, let us consider a counterexample to the converse of Corollary 1.

Example 10: Consider the extension C Q , and let The subfield F of C in

F = {α∈ C α is algebraic over Q}. Example 10, is called the
field of algebraic numbers,
Show that F Q is an infinite algebraic extension. and its elements are called
algebraic numbers.
Solution: By Proposition 6, F is a field. By its very definition, F Q is an
algebraic extension.
Suppose, if possible, that F Q is a finite extension, say [F : Q] = n for some
n ∈ N. Now, p( x ) = x n +1 − 2 is irreducible over Q by Eisenstein’s
1 1
irreducibility criterion, and 2 is a real root of p(x ). Hence 2 n+1 ∈ F, with
n +1

p( x ) being its minimal polynomial over Q. So, by Theorem 7,

  1  
Q  2 n+1  : Q = n + 1.
   
 1 
Also Q  2 n+1  ⊆ F. Therefore, by the Tower Theorem,
 
  1
 
Q  2 n +1  : Q ≤ [F : Q], i.e., n + 1 ≤ n, which is not possible.
   
Hence our assumption must be wrong. Thus, F Q is not finite.
*** 131
Field Theory The example above shows that not every algebraic extension is finite. But in
the next result we will prove that an extension generated by a finite number of
algebraic elements is finite, that is, a generalisation of Theorem 8.

Theorem 9: Let F k be an extension field generated by a finite number of

algebraic elements over k. Then F k is a finite extension.

Proof: We prove the result by applying the principle of mathematical

induction on the number of generators of F over k.
Suppose first that F is generated over k by a single algebraic element say α1 ,
i.e., F = k (α1 ). Since α1 is algebraic over k, by Theorem 9, F k is finite.

Next, assume that if a field is generated over k by n algebraic elements, then

it is a finite extension.

We will now prove the result for extensions generated over k by n + 1

algebraic elements. For this, let F = k (α1 , α 2 ,K, α n , α n+1 ), where
α i (1 ≤ i ≤ n + 1) are algebraic over k. Since α n+1 is algebraic over k, it is also
algebraic over k (α1 , α 2 ,K, α n ). Therefore,
[k (α1 , α 2 ,K , α n ) (α n+1 ) : k (α1 , α 2 ,K, α n )] is finite.
By the induction hypothesis, [k (α1 , α 2 ,K, α n ) : k ] is finite. Thus
[F : k ] = [k (α1 , α 2 ,K, α n , α n +1 ) : k ]
= [k (α1 , α 2 ,K, α n ) (α n +1 ) : k (α1 , α 2 ,K , α n )][k (α1 , α 2 ,K, α n ) : k ]
is finite.
Hence, F k is a finite extension for any F generated by a finite number of
algebraic elements over k.

Let us consider an example now.

Example 11: Let F k be an extension and α , β ∈ F be algebraic over F, with

degrees m and n respectively. Show that if g.c.d. (m , n ) = 1, then
[k (α , β) : k ] = mn.
Solution: Since k (α , β) k is a finite extension (generated by algebraic
elements α and β over k ) containing k (α) and k (β), by the Tower Theorem
you can see that m and n both divide [k (α , β) : k ]. Therefore, mn also
divides [k (α, β) : k ] (as m , n are coprime). Hence [k (α , β) : k ] ≥ mn.
For the other inequality, note that β is also algebraic over k (α). Therefore,
k (α)(β) = k (α , β) is a finite extension of k (α) of degree at most
deg mβ , k ( x ) = n, i.e., [k (α , β) : k (α)] ≤ n.
∴ [k (α , β) : k ] = [k (α , β) : k (α)][k (α) : k ] ≤ nm.
Hence, [k (α, β) : k ] = mn .
***

Try an exercise now.

E32) Let α be algebraic over a field k, with deg α being odd. Show that
k (α) = k (α 2 ). Does this hold true if deg α is even? Give reasons for
your answer.
132
You have proved, and applied, a ‘tower theorem’ (Theorem 3) regarding finite An Introduction to
Field Theory
extensions, several times. Let us look at another ‘tower theorem’ now, which
we prove by applying Theorem 9.

Theorem 10: Let k ⊆ F ⊆ E be fields. If E F and F k are algebraic

extensions, then so is E k.

Proof: Let α ∈ E. Since E F is algebraic, α is algebraic over F. Let

m α , F ( x ) = a 0 + a1x + L + a n x n . Since F k is algebraic, each a i (0 ≤ i ≤ n ) is
algebraic over k. Hence, by Theorem 10, [k (a 0 , a 1 ,K, a n ) : k ] is finite. Also,
α is algebraic over k (a 0 , a 1 ,K, a n ). Therefore,
[k (a 0 , a 1 ,K, a n ) (α) : k (a 0 , a 1 ,K, a n )] is finite.
Now, using Theorem 3, you can see that [k (a 0 , a1 ,K, a n ) (α) : k ] is finite.
Hence α is algebraic over k.
Thus, E k is algebraic.

Is the converse of Theorem 10 true? You can check this by doing E33 below.

E33) Prove, or disprove, the converse of Theorem 10.

E34) Prove that Q( p1 , p 2 ) Q is an algebraic extension of degree 4, where

p1 and p 2 are distinct primes.

So far our discussion has focused on algebraic extensions. You may wonder
what happens with extensions having transcendental elements. In fact, as a
contrast to Theorem 7, we have the following theorem about transcendental
elements over a field.

Theorem 11: Let F k be a field extension and α ∈ F. If α is transcendental

~ k ( x ).
over k, then k (α) −

Proof: Consider the evaluation map φα : k[ x ] → F : φα (f (x )) = f (α).

As you have shown in the proof of Theorem 7, φα is a well-defined ring
homomorphism. Also, since φ α (a ) = a ∀ a ∈ k and φα (x ) = α, k[α] ⊆ Im φ α .
Further, given the definition of k[α], Im φα ⊆ k[α]. Hence, Im φα = k[α].
Since α is transcendental, it is not the root of any polynomial over k.
∴ Ker φ α = {0}. Hence, by the Fundamental Theorem of Ring
Homomorphisms, k[ x ] − ~ k[α]. Since isomorphic domains have isomorphic
~ k (α).
quotient fields, we find that k ( x ) −

Applying this theorem, we know that Q(e) ~

− Q(π), since both are isomorphic
to Q(x ).

Try an exercise now.

133
Field Theory
E35) Let F = k (α), where α ∈ F is transcendental over k. Show that F is
finitely generated but F k is not a finite extension.

With this we come to the end of this unit. However, in the next unit, we shall
continue our discussion on extension fields.

11.5 SUMMARY
In this unit, you have studied the following points.

1. The definition, and examples, of a field, a quotient field of an integral

domain, a subfield, and the prime subfield of a field.

2. If k is a field, every finite subgroup of k ∗ is cyclic.

3. Every field contains its prime subfield, which is either Q or Z p , for

some prime p.

4. The characteristic of a field is 0 or p, for some prime p.

5. What an extension field F k is, and what its degree is.

6. Finite and infinite extensions.

7. If E F and F k are extension fields, then E k is finite iff E F and F k

are finite, and then [E : k ] = [E : F][F : k ].

8. − k[ x ]
If α is algebraic over a field k, then k[α] = k (α) ~ , where
mα (x)
m α (x ) is the minimal polynomial of α over k. Further, [k (α) : k ] < ∞.
Conversely, if k[α] is a field, then α is algebraic over k.

9. If α is transcendental over k, then k (α) ~

− k ( x ), so that k (α) k is an
infinite extension.

10. k (α1 ,K, α n ) k is finite if αi is algebraic over k ∀ i = 1,K, n.

11. Let k ⊆ F ⊆ E be fields. E F and F k are algebraic extensions iff E k

is an algebraic extension.

11.6 SOLUTIONS / ANSWERS

E1) (a + b d ) + (r + s d ) = (a + r) + d (b + s ), and
(a + b d ) ⋅ (r + s d ) = (ar + bsd ) + d (as + br ) ∀ a , b , r , s ∈ Q.
Hence + and · are binary operations on Q( d ).
134 Since + and · are associative and commutative in R, they have these
 An Introduction to
properties in Q( d ) also. The additive and multiplicative identities in Field Theory
R are 0 and 1, which lie in Q( d ).
The additive inverse of (a + b d ) is [(−a ) + (−b) d ] ∈ Q( d ).
The multiplicative inverse of (a + b d ) is
(a 2 − b 2d ) −1 (a − b d ) ∈ Q( d ) since a 2 − b 2d ≠ 0, d being square-free.
Since · distributes over + in R, it does so in Q( d ) also.
Hence Q( d ) is a field.

E2) Let h ∈ G s.t. o(h ) = m. Suppose ∃ x ∈ G s.t. o(x )/ m. Let p be a

prime such that p o( x ) and pα m, p α +1 / m, where α ≥ 0. Let
o( x ) = p t r, (p , r) = 1, t ≥ 1, and y = x r . Then o( y) = p t and yh ∈ G s.t.
o( yh ) = o( y)o(h ) > m, which is a contradiction. Hence o(g) m ∀ g ∈ G.

E3) I is maximal in R iff R I has no proper non-trivial ideals, iff R I is a

field.

E4) Ker φ is an ideal of F, a field. Hence, Ker φ = {0} or Ker φ = F.

Accordingly, φ is 1-1 or φ ≡ 0.

E5) i) (a , b) ~ (a , b) ∀ (a , b) ∈ R , since R is commutative. Thus, ~ is

reflexive.

ii) Let (a , b) , (c , d) ∈ R such that (a , b) ~ (c , d). Then ad = bc, i.e.,

cb = da. Therefore, (c , d ) ~ (a , b). Thus, ~ is symmetric.

iii) Finally, let (a , b) , (c , d ) , (u , v) ∈ R such that (a , b) ~ (c , d) and

(c , d ) ~ (u , v). Then ad = bc and cv = du.
Therefore, (ad) v = (bc )v = bdu , i.e., avd = bud. Thus, av = bu ,
since d ≠ 0.
∴ (a , b) ~ (u , v). Thus, ~ is transitive.
Hence, ~ is an equivalence relation.

E6) For [a , b],[c , d ],[u , v] ∈ F,

([a , b] + [c , d ]) + [u , v] = [ad + bc , bd ] + [u , v]
= [(ad + bc ) v + bdu , bdv ] = [adv + b(cv + du ) , bdv ]
= [a b] + [cv + du , dv] = [a , b] + ([c , d] + [u , v]).
Hence + is associative.
You can similarly show that · is associative, and + and · are
commutative, as well as · distributes over +.
Finally, [1,1] ⋅ [a , b] = [a , b] and [a , b][b , a ] = [ab , ab] = [1,1], since
(α , α) ~ (1,1) ∀ α ∈ R .

E7) Since F is the smallest field containing itself, it is its own quotient field.

E8) Since Z[ 2 ] ⊆/ Q( 2 ) ⊆/ R , R is not the smallest field containing

Z[ 2 ], and hence not its quotient field. 135
 E9) Since x 2 + 1 is irreducible over R, R[ x ]
Field Theory
is a field, and hence
x2 +1
is its own quotient field.

E10) Since Z is a subring of Q, and Z is not a field, we have a

counterexample to the given statement. Hence, the statement is false.

E11) i) By definition, K is the prime subfield of F.

ii) {
Here K = n ⋅ 1
m ⋅1
}
n , m ∈ Z, m ⋅ 1 ≠ 0 , and hence is the smallest
field contained in F. Hence it is the prime subfield.

iii) This means that K is the least subfield of F, and hence is its
prime subfield.

E12) Let {Fλ λ ∈ I} be a family of subfields of F, indexed by the set I.

Let α , β ∈ I Fλ . Then α , β ∈ Fλ for every λ ∈ I. Now, for each λ ∈ I, Fλ
λ∈I

is a field. Hence, α ± β, αβ, α β(β ≠ 0) ∈ Fλ ∀ λ ∈ I, and hence belong

to I F . Thus , I F
λ
λ∈I
λ is a subfield of F.
λ∈I

E13) No. For instance, consider Q( 2 ) ∪ Q( 3 ) = L, say. So

2 ∈ L, 3 ∈ L, but 2 + 3 ∉ L since any element of L is of the form
a + b 2 or a + b 3 , for a , b ∈ Q.

E14) Since Z p is a field, it is its own prime subfield. It is also the prime
subfield of any field containing Z p . Hence, it is the prime subfield of
Z p ( x ).

E15) The prime subfield of R is Q, which doesn’t contain 2 . So, the

required field is Q( 2 ).
Since this contains Q, its characteristic is zero.

E16) Since k is a subfield, it will contain the least subfield of F, and hence
will have the same prime subfield as that of F. Thus, both F and k will
have the same characteristic.

E17) The prime subfield of k is

P = {(m ⋅ 1) (n ⋅ 1) m , n ∈ Z, n ⋅ 1 ≠ 0}.
For any α ∈ P, α = (m ⋅1) (n ⋅1).
 m ⋅1  σ(m ⋅1)
Now, σ  =
 n ⋅1  σ(n ⋅1)
m ⋅ σ(1)
= , and σ(1) = 1.
n ⋅ σ(1)
 m ⋅1  m ⋅1
∴σ  = .
136  n ⋅1  n ⋅1
E18) Since a field homomorphism is injective (see E4), k ~
− σ(k ) and An Introduction to
Field Theory
~
F − σ(F). Let B be any k-basis of F and B ′ = σ (B ) its image in σ(F).
We claim that B ′ is a σ(k ) -basis of σ(F).

Linearly Independent: Let {α1′ , α′2 ,K, α′n } be any finite subset of B ′
such that
a1′α1′ + a′2α′2 + L + a′n α′n = 0, for some a1′ , a ′2 ,K, a ′n ∈ σ(k ). …(4)
Then there exist unique α1, α 2 ,K, α n ∈ B and a1 , a 2 ,K, a n ∈ k such that
σ(α i ) = α′i and σ(a i ) = a′i for 1 ≤ i ≤ n. Then (4) becomes
σ(a 1 )σ(α1 ) + σ(a 2 )σ(α 2 ) + L + σ(a n )σ(α n ) = 0
⇒ σ(a1α1 + a 2α 2 + L + a n αn ) = σ(0),
⇒ a1α1 + a 2α 2 + L + a n α n = 0.
Since α1, α 2 ,K, α n ∈ B are linearly independent over k,
a1 = a 2 = L = a n = 0, which implies that a′i = σ(a i ) = 0 ∀ i ∈{1,2, K , n},
i.e., α1′ , α′2 ,K, α′n are linearly independent over σ(k ).

Spanning: For an arbitrary element α′ ∈ σ(F) there exists α ∈ F such

that α′ = σ(α). Since B forms a k-basis of F, there exist unique
elements α1, α 2 ,K, α n ∈ B and a1 , a 2 ,K, a n ∈ k such that
α = a1α1 + a 2α 2 + L + a n αn .
⇒ σ(α) = σ(a1α1 + a 2α 2 + L + a n α n )
⇒ α′ = σ(a1 )σ(α1 ) + σ(a 2 )σ(α 2 ) + L + σ(a n )σ(α n ),
i.e., σ(α1 ), σ(α 2 ),K , σ(α n ) ∈ B′ spans α′ over σ(k ).

E19) Let F k be a finite extension and σ be a k-endomorphism of F. Since

every field homomorphism is injective, it only remains to show that σ is
surjective. Now, by the Fundamental Theorem of Homomorphism,
F−~ σ(F). Also σ is identity on k. Thus, by E18,
[F : k ] = [σ(F) : σ(k )] = [σ(F) : k ], i.e., both F and σ(F) are finite-
dimensional vector spaces over k having the same dimension.
Also, σ(F) is a subfield, and hence a subspace, of the k-vector space F,
having the same dimension as that of F. Hence, σ(F) = F, and σ is an
automorphism of F.

For the second part, note that Q(π) Q is not finite. Consider
σ : Q(π) → Q(π) given by σ(π) = π2 . You can verify that σ is a Q -
endomorphism of Q(π). We will now show that σ is not surjective.
Suppose, to the contrary, that σ is surjective. Then π must have a pre-
image, i.e., there exist p(π), q (π)(≠ 0) ∈ Q[π] such that
σ(p (π) q(π)) = π ⇒ p(π2 ) − πq(π2 ) = 0
⇒ π is a root of f ( x ) = p( x 2 ) − xq ( x 2 ) over Q,
contradicting the fact that π is transcendental over Q.
Hence, σ is not an automorphism of Q(π).

E20) Note that F( x ) ⊇ F[ x ], and F[ x ] has {1, x , x 2,K} as an F-basis. Hence

F( x ) is an infinite extension of F.
137
Field Theory E21) Since k (α, β) is the smallest field containing k , α and β ,
k (α , β) ⊆ k (α)(β). Further, since k (α)(β) is the smallest field
containing k (α) and β , k (α)(β) ⊆ k (α , β).
Hence k (α , β) = k (α)(β). Similarly, k (α , β) = k (β)(α).

E22) i) Since α is a root of x 3 − 3x + 4, α3 = 3α − 4. …(5)

Therefore, α 4 = 3α 2 − 4α …(6)
2 2
Now, for a , b , c ∈ Q, (α + α + 1) (cα + bα + a ) = 1
⇒ α 2 (a + b + 4c) + α(a + 4b − c) + (a − 4b − 4c − 1) = 0, using (5)
and (6).
Since α cannot satisfy an equation over Q of degree less than 3,
a + b + 4c = 0, a + 4b − c = 0, a − 4b − 4c − 1 = 0.
Solving these equations for a , b , c we obtain the inverse as
1
(17 − 5α − 3α 2 ).
49

ii) Since α is a root of f ( x ) = x n + a n −1 x n −1 + L + a 0 , and f ( x ) is

irreducible, a 0 ≠ 0. So,
α n + a n −1α n −1 + L + a 0 = 0.
⇒ α(−a 0−1α n −1 − a 0−1a n −1α n −1 − L − a 0−1a1 ) = 1
⇒ α −1 = −a −01α n −1 − a 0−1a n −1α n −2 − L − a 0−1a 1.

E23) Since 2 , 3 ∈ Q( 2 , 3 ), 2 + 3 ∈ Q( 2 , 3 ).
Hence Q( 2 + 3 ) ⊆ Q( 2 , 3 ). …(7)
So [Q( 2 + 3 ) : Q] divides [Q( 2 , 3 ) : Q].
Now, Q ⊆/ Q( 2 ) ⊆/ Q( 2 ) ( 3 ) (since 3 ∉ Q( 2 )), so that
[Q( 2 , 3 ) : Q] = [Q( 2 , 3 ) : Q( 2 )][Q( 2 ) : Q] = 2 × 2 = 4 …(8)
So [Q( 2 + 3 ) : Q] = 1, 2 or 4.
Since 2 + 3 ∉ Q, [Q( 2 + 3 ) : Q] ≠ 1.
Also ( 2 + 3 ) satisfies the irreducible polynomial
( x − 2 )2 − 3 ∈ Q( 2 ). Thus, [Q( 2 + 3 ) : Q( 2 )] = 2.
Hence [Q( 2 + 3 ) : Q] = [Q( 2 + 3 ) : Q( 2 )][Q( 2 ) : Q] = 4.
Thus, [Q( 2 + 3 ) : Q] = 4. …(9)
Hence, (7), (8), (9) tell us that Q( 2 + 3 ) = Q( 2 , 3 ).

E24) i) By the ‘Tower theorem’, [E : k ] = [E : F][F : k ]. Hence the result.

ii) If [E : k ] = p, then [E : F] = 1 or [E : F] = p in (i) above. Thus,

F = E or F = k.

E25) True. For instance, x n − 2 is irreducible over Q ∀ n ∈ N, using

Eisenstein’s criterion. Hence Q[x ] n is a finite extension of
x −2

138
degree n of Q.
E26) Let [F : k ] = p and α ∈ F \ k. Then k ⊆/ k (α) ⊆ F . An Introduction to
Field Theory
By E24 (ii), k (α) = F, i.e., F is simple.

E27) ∀ α ∈ k , x − α ∈ k[ x ] is satisfied by α. Hence k k is algebraic.

E28) Since α is algebraic over k, ∃ f ( x ) ∈ k[ x ] s.t. f (α) = 0. Since

k ⊆ E, f ( x ) ∈ E[x ] also. Hence, α is also algebraic over E.

E29) i) Since ω ∉ Q and 1 + ω + ω2 = 0,

m ω , Q ( x ) = x 2 + x + 1.
Similarly, mω , R ( x ) = x 2 + x + 1.
Since ω ∈ C, mω , C ( x ) = x − ω.
As ( x − ω) (x 2 + x + 1) in C[ x ], the rest follows.

ii) Let p( x ) = m α, k (x ). Since p(α) = 0 in E[x ], m α , E ( x ) p( x ). Hence

the result.

E30) Suppose first that k[α] is a field. Then every non-zero element of k[α]
is invertible. In particular, α is invertible. So ∃ β ∈ k[α] such that
αβ = 1. Now, β = a 0 + a1α + L + a n α n , where a i ∈ k for 0 ≤ i ≤ n and n
is a non-negative integer.
So, αβ = 1 ⇒ −1 + a 0α + a1α 2 + L + a n α n +1 = 0, i.e., α is a root of a
polynomial over k, and hence, α is algebraic over k.
Conversely, assume that α is algebraic over k. Then, by Theorem 7,
k[α] is a field.

E31) R Q and Q Q are the required extensions. There can be several other
examples.

E32) Suppose, if possible, that k (α 2 ) ⊆/ F = k (α), then α ∉ k (α 2 ). As

α 2 ∈ k (α 2 ), α satisfies the polynomial f ( x ) = x 2 − α 2 over k (α 2 ).
Since α ∉ k (α 2 ), it follows that f ( x ) is irreducible over k (α 2 ), and
hence [k (α) : k (α 2 )] = [k (α 2 )(α) : k (α 2 )] = deg f ( x ) = 2.
In view of the Tower Theorem, this implies that 2 divides the degree of
k (α) k, contradicting the fact that k (α) k is an odd degree extension.
Hence, k (α) = k (α 2 ).
For the second part, consider 2 ∈ Q. Its degree is even. Also
Q( 2 ) ≠ Q( 22 ) = Q. Hence, it need not be true if deg α is even.

E33) The converse is: If k ⊆ F ⊆ E are fields such that E k is algebraic, then
E F and F k are algebraic. You have already proved that E F is
algebraic.
Now, consider α ∈ F. Then α ∈ E, and hence, is algebraic over k. Thus,
F k is algebraic also.
139
Field Theory E34) First, let us prove that p1 ∉ Q( p 2 ) by contradiction. So, suppose
p1 ∈ Q( p 2 ).
Then p1 = α + β p 2 , α, β ∈ Q∗. So p1 = α 2 + β 2 p 2 + 2αβ 2 , and
hence p 2 ∈ Q, a contradiction.
Hence p1 ∉ Q( p 2 ).
Similarly, p 2 ∉ Q( p1 ).

Now, we have a tower Q ⊆/ Q( p1 ) ⊆/ Q( p1 , p 2 ). So

[Q( p1 , p 2 ) : Q] = [Q( p1 , p 2 ) : Q( p 2 )][Q( p 2 ) : Q] = 2 × 2 = 4
(since p 2 has the minimal polynomial x 2 − p 2 ∈ Q[ x ] and p1 has the
2
minimal polynomial x − p1 ∈ Q( p 2 )[x ]).
Further, since Q( p1 , p 2 ) Q is finite, it is algebraic.

E35) As F = k (α), it is finitely generated. Suppose, if possible, that F k is a

finite extension. Then it must be algebraic, which implies that every
element of F is algebraic over k. As α ∈ F, α must also be algebraic
over k, which contradicts the hypothesis that α is transcendental. Thus,
F k is not finite.

140
 Splitting Fields
UNIT 12 SPLITTING FIELDS – FINITE AND – Finite and Infinite

INFINITE
Structure Page No.
12.1 Introduction 141
Objectives
12.2 Splitting Fields 142
12.3 Finite Splitting Fields 148
12.4 Finite Extensions and Subfields of Finite Fields 152
12.5 Summary 157
12.6 Solutions / Answers 158

12.1 INTRODUCTION
In the previous unit, you studied algebraic extensions. You now know that
given an irreducible polynomial p( x ) over a field K, K[ x ] < p( x ) > is an
extension field of K containing one root of p(x ). However, this extension
field may not contain all the other roots of p(x ).

In Sec.12.2, we look at extensions F k containing all the roots of a given

polynomial f ( x ) over the field k. You will see that such extensions exist. You
will also see that there is a “smallest” extension field with this property, which
is called a splitting field of f ( x ) over k.

In the next section, Sec.12.3, you will see that every finite field is a splitting
field of some polynomial. You will also see that given any prime p, and any
n ∈ N, there is a finite field with p n elements, and these are all the finite
fields. Hence, any splitting field over Z p has cardinality p n for some n ∈ N.

Finally, in Sec.12.4, you will study finite extensions of finite fields.

Surprisingly, every such extension is a simple extension, as you will see. In this
section you will also see under what condition a field is a subfield of a given
field. Finally, in this section, we will also consider irreducible polynomials
over finite fields.

A word here about finite fields! They arose in the study of congruences modulo
a prime p. But the more general notion of finite fields was first introduced by
the mathematician Galois, in 1830, for solving congruences of the type
f ( x ) ≡ 0(mod p), where p is a prime number and f ( x ) ∈ Z[x ] is irreducible
modulo p. In his honour, finite fields are also know as Galois fields. Finite
fields are of fundamental importance in many areas of mathematics such as
number theory, coding theory, cryptography and algebraic geometry.

Let us now consider the expected learning outcomes of this unit.

Objectives
After studying this unit, you should be able to
• define, and find, a splitting field of a polynomial over a given field; 141
Field Theory • prove that the splitting field of a polynomial over a given field is unique
upto isomorphism;
• prove that every finite field is a splitting field of a polynomial;
• prove the existence and uniqueness of splitting fields having p n
elements, for every prime p and n ∈ N;
• prove, and apply, the fact that every finite extension of a finite field is
simple;
• characterise the subfields of a given finite field;
• characterise the irreducible polynomials of degree d over Z p .

12.2 SPLITTING FIELDS

In the previous unit, we ended our discussion with looking at extension fields
generated by elements that satisfied some polynomial over the base field. Now,
the question is if one root of a polynomial is in the extension field, will all its
roots be in that field? For instance, (x + 1)(x 2 + 1) ∈ Q[ x ] has one root in Q,
but not the other. Again, there are polynomials like x 2 + 1 ∈ R[ x ] that have no
root in R. In fact, you can find several examples of fields k, and polynomials
over them, with no root in k. However, as you will see, there always exists an
extension of k containing all the roots of a given polynomial over k. Let us
begin our search for such an extension field.

Theorem 1: Let p( x ) be an irreducible polynomial of degree n over a field k.

There exists an extension field F, of k, in which p( x ) has a root. Moreover,
[ F : k ] = n.

Proof: Since p( x ) is irreducible over k, k[x ] I is a field, where I = p( x ) .

Define φ : k → k[ x ] I : φ(a ) = a + I. Then, you should check that φ is an
isomorphism of k onto Im φ = k′, say. (In fact, φ is the composition of the
inclusion map k → k[ x ] and the canonical surjection k[x ] → k[ x ] I. )
Identifying k with k′, we can consider k[ x ] I = F as an extension of k. Also,
x + I ∈ F is a root of the polynomial p( x ) (considered over F ) since
p( x + I) = p( x ) + I = 0, as p( x ) ∈ I.

Next, you should check that {1 + I, x + I,K, x n −1 + I} is a basis of F over k.

Thus, [F : k ] = n.

An immediate corollary to this theorem is the following.

Corollary 1 is due to the Corollary 1: Let f ( x ) be a polynomial of degree n ≥ 1 over a field k. Then
mathematician Kronecker,
who proved it in 1887.
there exists a field F, containing k, in which f ( x ) has a root, and [F : k ] ≤ n.

Proof: If f ( x ) is irreducible over k, then the result follows from the theorem
above. Otherwise, f ( x ) has an irreducible factor, p( x ), in k[ x ]. So again by
Theorem 1, there exists an extension F k such that F contains a root of p( x ),
and hence of f ( x ).
142 Also [F : k ] = deg p( x ) ≤ deg f (x ) = n.
 Splitting Fields
Now consider x 3 − 2 ∈ Q[x ]. You know that it has a root in an extension field – Finite and Infinite
F = Q(21/ 3 ) of Q. But does it have all its roots in F ? Not so, since the other
roots are in C \ R a further extension of Q. This example may give you an
idea about what we now prove.

Theorem 2: Let f ( x ) ∈ k[ x ] be a polynomial of degree n ≥ 1. There exists an Recall that n! denotes ‘n

extension field F of k in which f ( x ) has n roots, and such that [F : k ] ≤ n!. factorial’.

Proof: We will prove the result by induction on deg f ( x ). If deg f ( x ) = 1,

then f ( x ) is linear, and F = k has all the roots of f ( x ), and [k : k ] = 1. So the
theorem holds in this case.
Suppose, now, that deg f ( x ) = n, and the theorem holds for all polynomials of
degree less than n. By Corollary 1, there is an extension field F0 of k
containing a root (say α ) of f ( x ), and [F0 : k ] ≤ n. So in F0 [x ], f ( x ) factors as
f ( x ) = ( x − α)g( x ), where g( x ) ∈ F0 [ x ] is a polynomial of degree n − 1.
By our induction hypothesis (applied to g( x ) over F0 ), there is an extension
field F of F0 which contains all the n − 1 roots of g( x ), and [F : F0 ] ≤ (n − 1)!.
Since any root of f ( x ) is either α or a root of g( x ), you can see that the field
F contains all the roots of f ( x ). Now,
[F : k ] = [F : F0 ][F0 : k ] ≤ (n − 1)!n = n!.
Hence the theorem follows in this case too.
Thus, by the principle of induction, it is true in general.

Theorem 2 tells us that given f ( x ) ∈ k[ x ] of degree n, there is a finite

extension F k in which f ( x ) has all its roots. If
f ( x ) = a 0 + a 1 x + L + a n x n , a n ≠ 0, n ≥ 1, a i ∈ k, and if the n roots of f ( x ) in F
are α1, α 2 ,K, α n , then f ( x ) can be factored in F[ x ] as
f ( x ) = a n ( x − α1 )(x − α 2 ) L (x − α n ).
Thus, f(x) splits completely over F, as a product of linear factors in F.
Since a finite extension F of k exists with this property, there is a finite
extension of k of minimal degree in which f ( x ) can be factored as a product
of linear factors. For such a minimal extension field, no proper subfield would
have this property. This leads us to the following definition.

Definition: Let f ( x ) be a polynomial over a field k. An extension field F of

k is called a splitting field of f ( x ) over k, if

i) f ( x ) can be written as a product of linear factors in F[ x ], and

ii) there is no proper subfield E of F, containing k, such that f ( x ) splits in

E[x ].

For example, a splitting field of x 2 + 1 ∈ R[x ] is C. This is because

x 2 + 1 = ( x − i) ( x + i) in C, and C is the smallest such field since [C : R ] = 2,
and i ∉ R.

Now consider the following comment about splitting fields.

143
Field Theory Remark 1: You know that x 2 − 1 ∈ Q[ x ] factors completely as ( x − 1)(x + 1) in
Q[ x ] itself. Thus, Q is its splitting field over Q. Similarly, its splitting field
over R will be R, not Q. Thus, it is important to keep in mind the base field
you are working with.

Now, by Theorem 2, you know that you can always find a splitting field. In
some simple cases, you can actually find it also. But what does it look like for
polynomials of large degree, say 3x 5 + x 3 + 2 x + 7 ∈ Z11[ x ].

Theorem 3: Let f ( x ) be a polynomial over a field k, of degree n, having

roots α1 , α 2 , K , α n in some extension field F of k. Then k (α1 , α 2 ,K, α n ) is a
splitting field of f ( x ).

Proof: k (α1, α 2 ,K, α n ) certainly contains k and all the roots of f ( x ). Suppose
E is another subfield of F with k ⊆ E and α1 , α 2 , K , α n ∈ E, then E must
contain the subfield generated by k and α1 , α 2 , K , α n , i.e.,
k (α1 , α 2 ,K , α n ) ⊆ E. Thus, k (α1 , α 2 , K , α n ) is the smallest subfield of F
containing all the roots of f ( x ). Hence, it is a splitting field of f ( x ) in F.

Let us consider some examples.

Example 1: Find a splitting field of each of the following:

i) x 3 − 2 ∈ Q[x ], ii) x 3 − 2 ∈ Z 3[ x ].

Solution: We shall apply Theorem 3 here.

i) Let f ( x ) = x 3 − 2. Then f ( x ) is irreducible over Q by Eisenstein’s
criterion. The roots of f ( x ) are 21/ 3 , 21/ 3 ω and 21/ 3 ω2 , where ω is a
complex cube root of unity. So a splitting field of f ( x ) = x 3 − 2 over Q
is the field Q(21/ 3 , 21/ 3 ω, 21/ 3 ω2 ), which is the same as the field
Q(21/ 3 , ω).

ii) x 3 − 2 = x 3 + 1 = ( x + 1 )3 in Z 3 [x ]. Hence, a splitting field over Z 3 is

Z 3 ( − 1 ,− 1 ,− 1 ) = Z 3 .
***

So far we have been talking about a splitting field. Can there be more than one
over the same base field? It turns out that up to isomorphism, there is only one
splitting field of a given polynomial over a given field. To prove this, we will
need two lemmas.

Lemma 1: Let σ : k → k ′ be a field isomorphism. Then σ induces an

~ ( a x i ) = σ(a )x i .
isomorphism σ~ : k[ x ] → k ′[ x ] : σ
∑ i ∑ i
~
Further, σ maps irreducible polynomials to irreducible polynomials.
γ:A →~ B denotes
that γ is an isomorphism We leave the proof of the lemma to you to do (see E1). Let’s consider the
from A to B , also second lemma we need.
γ
denoted by A → ~ B. Lemma 2: Let k , k ′ be fields and σ be an isomorphism from k to k ′,
~ : k[ x ] → k ′[ x ]. Let p( x ) ∈ k[ x ] be an irreducible polynomial.
extended to σ
144
Then σ~ (p( x )) ∈ k ′[ x ] is also irreducible (by Lemma 1). If α , α′ denote the Splitting Fields
~ (p( x )), respectively, in some extension fields of k and k ′, – Finite and Infinite
roots of p( x ) and σ
respectively, then σ can be extended to an isomorphism σ : k[α] → k ′[α′]
such that σ (α) = α′. σ
k[ α ] k′[α′]

Proof: In view of Lemma 1, ∃ σ ~ : k[x ] → ~ (p( x )) is

~ k ′[x ], such that σ
θ φ
irreducible over k ′. Under this isomorphism, the maximal ideal p( x ) is
mapped onto the maximal ideal σ ~ (p( x )) , and hence ψ
k[ x ] k′[ x ]
~ (p( x )) : ψ (f ( x ) + p(x ) ) = σ~ (f ( x )) + σ
~ (p(x )) is an p( x ) ~ (p( x ))
σ
ψ : k[ x ] p(x ) → k′[ x ] σ
isomorphism of fields.
Since α and α′ are roots of p( x ) and σ ~ (p( x )), respectively, by Theorem 5 of
Unit 11, k[x ] p( x ) −~ k[α ] and k′[ x ] σ ~ (p( x )) −
~ k ′[α′]. Recall that under ~′
σ
k[ x ] k′[ x ]
these isomorphism x + p( x ) a α and x + σ ~ (p( x )) a α ′, respectively.

If θ and φ denote, respectively, the isomorphisms from k[ x ] p( x ) to k[α]

~ (p( x )) to k′[α′], then the composition σ = φ o ψ o θ −1 is an
and k′[ x ] σ σ
k k′
isomorphism from k[α] to k′[α′] (see Fig. 1). Fig. 1
Now, for any a ∈ k, we have
σ (a ) = φ o ψ o θ −1 (a ) = φ o ψ (a + p(x ) ) = φ(σ(a ) + σ ~ (p( x )) )
= σ(a ).
Therefore, σ k = σ.
Also
σ (α) = φ o ψ o θ −1 (α) = φ o ψ ( x + p( x ) ) = φ( x + σ ~ (p( x )) )
= α′.
Thus, σ is the required isomorphism between k[α] and k′[α′], which extends
σ, and maps α to α′.

Now, let us use this lemma to prove the theorem showing the uniqueness of the
splitting field of a polynomial over a given field.

Theorem 4: Let σ : k → k ′ be a field isomorphism, extending to the ring

isomorphism σ ~ : k[ x ] → k ′[ x ]. Let f ( x ) ∈ k[x ], and let F and F′ be splitting
fields of f ( x ) and σ ~ (f ( x )), respectively. Then there exists an isomorphism
from F to F′, which is an extension of σ.

Proof: We will prove this by induction on deg f ( x ).

If deg f ( x ) = 1, then F = k , F′ = k ′, and the theorem holds trivially.
Assume that the theorem holds for polynomials of degree not exceeding n − 1.
Now, let f ( x ) ∈ k[x ], with deg f ( x ) = n. Let p( x ) be an irreducible factor of
f ( x ) over k, α be a root of p( x ) and α′ be a root of σ ~ (p( x )). By Lemma 2,
σ can be extended to an isomorphism σ from k[α] to k ′[α′] mapping α to
α ′.
Let us write f ( x ) = ( x − α)g( x ) in k (α)[x ], so that
~ (f (x )) = σ (f ( x )) = ( x − α′) σ (g (x )).
σ
Since F is a splitting field of f ( x ) over k, it is a splitting field of g( x ) over
k (α). Similarly, F′ is a splitting field of σ (g( x )) over k ′(α′). As
145
 Field Theory deg g (x ) = n − 1, by induction σ can be extended to an isomorphism σ ′ from
F onto F′. Since σ′ is an extension of σ, σ ′ k = σ.
Thus, the theorem is true for a polynomial of degree n, and hence for all
polynomials.

We are now in a position to prove what we had aimed for, as an immediate

corollary of Theorem 4.

Corollary 2: The splitting field of a polynomial over a field k is unique up to

isomorphism.

Proof: The result follows immediately from Theorem 4, on taking k = k ′ and

σ as the identity map.

Because of Corollary 2, we can say ‘the’ splitting field, instead of a splitting

field, of a given polynomial over a given field.

Let us consider some examples now.

Example 2: Find the splitting field of x 4 + 4 over Q. Hence show that the
degree of the splitting field extension can be strictly less than the degree of the
polynomial concerned.

Solution: Now, to find the roots of x 4 + 4 (in its splitting field), note that
x 4 + 4 = ( x 2 + 2) 2 − 4 x 2 = (x 2 − 2 x + 2)(x 2 + 2 x + 2) in Q[x ].
By Eisenstein’s criterion, ( x 2 − 2 x + 2) and ( x 2 + 2 x + 2) are irreducible over
Q. Their roots, in C, are 1 ± i and − 1 ± i. Thus, the splitting field of
x 4 + 4 over Q will be Q(1 + i,1 − i, − 1 + i, − 1 − i), that is, Q(i).
Here, you can see that [Q(i) : Q] = 2 < deg ( x 4 + 4).

***

Example 3: Find the splitting field of x n − 1 over Q, where n ∈ N.

Solution: As you know, the roots of x n − 1 = 0 are the nth roots of unity. Using
De Moivre’s theorem, you know that the roots are
2πi k
2πk 2πk
e n = cos + i sin ∀ k = 0,1, K , n − 1.
n n
 2nπ i 4nπ i 2 ( n −1) π i

Thus, the required splitting field is F = Q1, e , e ,K, e n .

 
2π i 2 kπ i
Q(ζ n ) is called the Now let ζ n = e . Then e
n n
= ζ kn ∀ k = 0,1, K , n − 1.
cyclotomic field of the nth Thus, F = Q(ζ n , ζ 2n ,K , ζ nn−1 ) = Q(ζ n ). (Note that ζ nn = 1 ∈ Q.)
roots of unity.

***

In the next two units, you will come across the cyclotomic fields several times.
See Fig. 2 for an example of a geometric representation of the points ζ i , where
i = 1,K ,6.
146
 Splitting Fields
Y – Finite and Infinite

ζ2 ζ1

ζ6 = 1
ζ3 O X

ζ4 ζ5

Fig. 2: The geometric view of the 6th roots of unity.

Try some exercises now.

E1) Prove Lemma 1.

[Hint: First prove that σ~ is an isomorphism; then, by the method of
contradiction, you should prove that if f ( x ) ∈ k[ x ] is irreducible, so is
~ (f (x )) ∈ k ′[ x ]. ]
σ

E2) Show that if f ( x ), g( x ) ∈ k[x ], have a common factor in some extension

field F of k, then they have a common factor over k.

E3) Identify the splitting fields of the following polynomials:

i) f ( x ) = ( x 3 − 2)( x 2 + x + 1) over Q,

ii) f ( x ) = ( x 3 − 5) 2 over Q,

iii) f ( x ) = x 2 − x + 1 over Q(ω),

iv) f ( x ) = x 6 + 1 over Z 2 .

E4) Let f ( x ) be a polynomial over Q of degree n and let F denote its

splitting field over Q. Give examples of polynomials f ( x ) for which:
i) [ F : Q] = n ,
ii) [ F : Q] < n ,
iii) n < [F : Q] < n!, and
iv) [F : Q] = n!.

Before ending this section, we shall just briefly discuss a property that C has,
namely, any polynomial over C has all its roots in C. This is not true for Q or
R, as you have seen earlier. However, there are other fields that have this
property too. We give them a name.
147
Field Theory Definition: A field k is called algebraically closed if every polynomial over
k has a root in k.

In fact, as you can prove, a field k is algebraically closed if and only if

i) every polynomial over k splits completely in k or, equivalently,
ii) it has no proper algebraic extension.

A property of such fields is that they cannot be finite, which you can prove too.

Let us now move our discussion to splitting fields over Z p , for some prime p.

12.3 FINITE SPLITTING FIELDS

Finite fields are also In the previous unit, you have studied fields of different kinds, including Q
called Galois fields, in
and Z , where p is a prime. You have noted that Z has p elements, and
honour of the pZ pZ
mathematician, Galois.
(Galois is pronounced
hence it is a finite field. As you know, a field has at least two elements – 0 , 1.
‘Gal-waa’.) Thus, the smallest field is the field having two elements Z .
2Z

Further, recall that any finite field contains its prime subfield, Z p , for some
prime p. Hence the characteristic of a finite field is always a prime number.
Henceforth, we will denote Z pZ by Fp , for a prime p, as many authors
do.

And now, for something that may surprise you.

Example 4: Show that Fp is the splitting field of some polynomial over Fp .

Solution: Consider f ( x ) = x p − x. Note that every element of Fp satisfies it, by

Fermat’s little theorem. Thus all p roots of f ( x ) are in Fp , and hence, F p is
the splitting field of f ( x ) over Fp .
***

What you have seen in Example 4 is not just true for Fp . As you will see, any
finite field is the splitting field of some polynomial over Fp . But first, you will
see that a finite field is not necessarily of the form Fp , for some prime p.
There are actually infinitely many other finite fields. In fact, the focus of this
section is to prove, and apply, Theorem 5. It says that for each n ∈ N and
prime p, there is a ‘unique’ splitting field extension of Fp with p n elements.
The first stage towards this is the following lemma.

Lemma 3: Let F be a finite field. Then F has p n elements, for some prime p
and positive integer n.

Proof: Since F is a finite field, its characteristic must be a prime (say) p.

Therefore, the prime subfield P of F is isomorphic to Fp . So F can be
considered as a vector space over Fp . Since F is finite, [F : P] is finite, say n.
148 Thus, F has p n elements.
The strategy in the proof above is built on the idea of considering an extension Splitting Fields
– Finite and Infinite
field as a vector space over the base field. This idea is due to the mathematician
Richard Dedekind.

Try an exercise now.

E5) On the lines of Example 4, show that every finite field is the splitting
field of some polynomial over Fp .

To prove Theorem 5, we need to take a bit of a diversion into the relation

between the type of roots of a polynomial and those of its derivative. So let us
consider some definitions first.

Definition: Let k be a field and let f ( x ) ∈ k[ x ] split over its splitting field F,
as f ( x ) = a ( x − α1 ) n1 ( x − α 2 ) n 2 L ( x − α r ) n r ,
where α1 , α 2 , K , α r are the distinct roots of f ( x ) in F and n i ≥ 1 for all
i ∈ {1, 2 , K , r}. The root α i is called a multiple root of f ( x ) if n i > 1, and is
called a simple root if n i = 1. The integer n i is called the multiplicity of the
root α i .

For instance, x 4 − 4x 2 + 3 ∈ Q[ x ] has only simple roots in R, namely,

± 1, ± 3. But ( x + 2i)2 ( x − 5) ∈ C[x ] has one multiple root, − 2i, and one
simple root, 5.

Next is a definition with no surprises!

Definition: Let k be a field and f ( x ) = a 0 + a 1x + L + a n x n be a polynomial

over k, then the formal derivative of f(x) is the polynomial
f ′(x) = a 1 + 2a 2 x + L + na n x n −1 ∈ k[ x ].

For instance, if f ( x ) = 3x 5 + 0.7 x 2 + πx ∈ R[ x ], then f ′( x ) = 15x 4 + 1.4 x + π.

Similarly, if f ( x ) = 3x 5 + 2x ∈ F 5[ x ], then f ′( x ) = 2 ∈ F5[ x ], since 5 = 0 in
F5 .

Some properties of the derivative, that you would be using, are given in the
following proposition.

Proposition 1: If f ( x ) and g( x ) are polynomials over k, then

i) (f ( x ) + g ( x ))′ = f ′( x ) + g ′( x ),

ii) (af ( x ))′ = af ′( x ) ∀ a ∈ k ,

iii) (f ( x )g( x ))′ = f ′( x )g ( x ) + f (x )g′(x ).

n m
Proof: Let f ( x ) = ∑ a i x i and g( x ) = ∑ b j x j . Assume n > m.
i =1 j −1
149
Field Theory m n
i) f ( x ) + g ( x ) = ∑ (a i + b i ) x i + ∑a x . i
i

i =0 i = m +1
m n
Hence (f ( x ) + g ( x ))′ = ∑ i(a i + bi ) x i −1 + ∑ ia x i
i −1

i =1 i = m +1
n m
= ∑ ia i x i−1 + ∑ jb j x j−1
i =1 j=1

= f ′( x ) + g′( x ).

We leave the proof of (ii) and (iii) to you.

Let us now move to the next stage of the proof of Theorem 5.

Lemma 4: Let k be a field and f ( x ) be a non-constant polynomial over k.

Then all the roots of f ( x ) are simple (in the splitting field of f ( x )) if and only
if f ( x ) and its derivative f ′(x ) are relatively prime.

Proof: Suppose, first, that the roots of f ( x ) are all simple. Let us assume, to
the contrary, that f ( x ) and f ′(x ) have a common factor in k[ x ]. This would
imply that they have a common factor, say ( x − α), in the splitting field of
f ( x ). Since all the roots of f ( x ) are simple, we can write f ( x ) as
f ( x ) = ( x − α)g( x ) for some polynomial g( x ), with g(α) ≠ 0.
Taking the derivative, we see from Proposition 1 that
f ′( x ) = g( x ) + (x − α)g′(x ), and hence f ′(α) = g(α) ≠ 0.
This contradicts the fact that α is a common root of f ( x ) and f ′( x ).
Thus, f ( x ) and f ′(x ) must be relatively prime.

Conversely, suppose that f ( x ) and f ′(x ) are relatively prime. Let us assume,
to the contrary, that there exists a root α of f ( x ) with multiplicity m > 1.
Then f ( x ) = (x − α)m g( x ), with g(α) ≠ 0.
On taking the derivative, we have,
f ′( x ) = m(x − α)m −1 g (x ) + ( x − α)m g′( x )
= ( x − α) [m(x − α)m − 2 g ( x ) + ( x − α) m −1 g′( x )].
This shows that f ( x ) and f ′(x ) have a common factor ( x − α), which
contradicts the fact that f ( x ) and f ′(x ) are relatively prime.
Hence, f ( x ) only has simple roots.

Now we will prove the theorem we were aiming for.

Theorem 5 (Existence, and Uniqueness, of Finite Fields): For each prime

number p and each integer n ≥ 1, there exists exactly one field (up to
isomorphism) with p n elements, namely, the splitting field of the polynomial
n
f ( x ) = x p − x over Fp , and these are the only finite fields.

Proof: We shall first prove that there exists a field of order p n for each prime
p, and n ∈ N.
n
Consider f ( x ) = x p − x over the field Fp , and let F be the splitting field of

150 f ( x ) over Fp . Since the characteristic of Fp is p, f ′( x ) = −1. Therefore, f ( x )

and f ′(x ) are relatively prime, and hence, by Lemma 4, all the roots of f ( x ) in Splitting Fields
– Finite and Infinite
F are distinct.
n
Now, let K be the set of all the roots of f ( x ) in F. Let α , β ∈ K, then α p = α
n
and β p = β. Since char F = p, (α ± β)p = α p ± βp .
n n −1
Now, (α ± β) p = ((α ± β) p ) p , and by induction, we get
n n n
(α ± β) p = α p ± β p = α ± β.
n n n
Also (αβ) p = α p βp = αβ.
Therefore, if α and β are two roots of f ( x ), then so are α ± β and αβ.
Clearly, 0 , 1 ∈ K. Thus, K is a subfield of F. As the prime subfield is
contained in every subfield of F, Fp ⊆ K.

Thus, K is the splitting field of f ( x ), and contains exactly p n elements. Since

the splitting field is unique up to isomorphism, it follows that F = K. Hence,
for any positive integer n and prime number p the splitting field of the
n
polynomial f ( x ) = x p − x over Fp is a field having p n elements.

Now we will prove the uniqueness part of the theorem. So, let F and K be two
finite fields having the same number of elements, say p n , for some positive
integer n and prime p. By E5, both F and K are the splitting fields of
n
f ( x ) = x p − x over Fp , as both the fields F and K have the same
characteristic p. Since the splitting fields of a polynomial are unique upto
isomorphism, F ~ − K. Thus, up to isomorphism, there is exactly one field with
n
p elements.

Theorem 5 tells us that if F is a finite field then F = p n , for some prime p and
positive integer n. We denote this finite field by Fp n or Fq , where q = p n .

Now, an important remark about fields having the same cardinality.

Remark 2: Note that there can be several fields of order p n . However, they
will all be isomorphic, not equal. For example, you can check that x 2 + 1 and
x 2 − x − 1 are both irreducible over F3 . Hence K = F3[ x ] 2 and
x +1

L = F3[ x ] are both fields having 9 elements. Hence both are

x2 − x −1
isomorphic to F9 . But K ≠ L, since, for example, x 2 + 1 = 0 in K, but
x 2 + 1 = x + 2 ≠ 0 in L.

Let us now consider an example of the use of Theorem 5 to construct a finite

field of a given cardinality.

Example 5: Give an explicit construction of a field with 4 elements.

Solution: By Theorem 5, you know that there exists a finite field with 22 = 4
151
Field Theory elements, viz. F2 2 = F4 , with characteristic 2. You also know that F4 is the
splitting field of x 4 − x over F2 . Let F4 = {0 ,1, α , β}.
Now, x 4 − x = x (x − 1)(x 2 + x + 1).

By a direct verification, you can see that x 2 + x + 1 is irreducible over F2 .

Also, α , β have to be the roots of x 2 + x + 1. So α 2 + α + 1 = 0. Also, relating
the coefficients and the roots of x 2 + x + 1, α + β = −1 = 1, since char F4 = 2.
So β = α + 1.
Thus, F4 = {0 ,1, α , 1 + α}, and a basis of F4 over F2 is {1 , α}.
***

Try some exercises now.

E6) i) Construct a field with 9 elements.

ii) Determine the irreducible polynomial satisfied by each element of
the field in (i).

E7) Show that the product of the non-zero elements of Fpn is − 1.

So, you have seen that every finite field is a splitting field. Further, given a
prime p and n ∈ N, there is a splitting field over Fp of order p n . Let us now
consider extensions of these fields.

12.4 FINITE EXTENSIONS AND SUBFIELDS OF

FINITE FIELDS
In Unit 11, you studied various types of extension fields. Here we shall focus
on F k , where k = Fp n , and [F : k ] < ∞. We will also see whether or not
Fpm ⊆ Fpn ∀ m < n , m, n ∈ N.

So, let us see what K looks like if K Fp n is finite.

Theorem 6: Every finite extension of a finite field is simple.

Proof: Let F be a finite field. Then F = Fq , where q = p n for some prime p

and positive integer n. If E is any finite extension of Fq , then E is also a
finite field. So, by Proposition 1 of Unit 11, E ∗ is a cyclic group, generated by
α, say.
So Fq (α) ⊆ E. …(1)
Now Fq (α ), being a field, contains 0 ,1 and all the powers of α. Since every
non-zero element of E is a power of α, we conclude that
E = E ∗ ∪ {0} ⊆ Fq (α ). …(2)
Thus, from (1) and (2), E = Fq (α ), i.e., E is a simple extension of Fq .
152
An immediate corollary to the theorem above is the following. Splitting Fields
– Finite and Infinite

Corollary 3: Fp n Fp is a simple extension, for every prime p and n ∈ N.

Proof: Fpn is a finite extension of its prime field Fp . Hence, by the theorem
above, Fp n Fp is simple.

Let us consider an example of the application of the results above.

Example 6: Obtain a primitive element of F4 over F2 .

Solution: From Example 5, you know that F4 = {0 ,1, α ,1 + α}, where α is a

root of the polynomial x 2 + x + 1. Since the characteristic of F4 is 2, and
α 2 = α + 1, we have F4 = {0 ,1, α , α 2 }.
Also, α 3 = α ⋅ α 2 = α(α + 1) = α 2 + α = −1 = 1.
Hence F4∗ = α , and F4 = F2 (α). Thus, α is a primitive element of F4 over
F2 .
***

Try some exercises now.

E8) Find a 13th root of 3 in F13 .

E9) Find two distinct primitive elements of F9 over F3 .

We now give a characterisation of subfields of a finite field.

Theorem 7: Let p be a prime number and m , n be positive integers. Then

Fpm is a subfield of Fpn if and only if m n.

Proof: Suppose, first, that Fpm is a subfield of Fpn . Then there exists an
injective homomorphism θ : Fpm → Fpn . So Fpm ~
− Im θ = E, say.
Now, E is a finite subfield of Fpn .
Since the characteristic of Fpn is p, Fp ⊆ E ⊆ Fpn .
Now [Fpn : Fp ] = n and [E : Fp ] = [Fpm : Fp ] = m.
So [Fpn : Fp ] = [Fpn : E][E : Fp ]
i.e., n = [Fp n : Fp m ]m,
i.e., m n.

Conversely, assume that m n. As you know, Fpn and Fpm are, respectively,
n m
the splitting fields of the polynomials x p − x and x p − x over Fp .
153
 m n
Field Theory Since m n , x p − x divides x p − x over Fp . (Why? See E10). So all the
m n
roots of x p − x are also roots of x p − x, and hence Fpm ⊆ Fpn , i.e., Fpm is a
subfield of Fpn .

Here is an important comment about the theorem above.

Remark 3: Let p be a prime and n a positive integer. Since any two finite
fields with the same number of elements are unique upto isomorphism, the
theorem above tells us that Fpn contains a unique subfield with p m elements
(Fpm ) if and only if m n. Thus, to each divisor of n there corresponds a
unique subfield of Fp n , and conversely.

Try the following exercises now.

m n
E10) Prove that x p − x divides x p − x over Fp if and only if m divides n.

E11) Find all the proper subfields of Fp8 and Fp12 , where p is a prime.

As an application of the theorems above, let us now consider irreducible

polynomials over Fpn . We first see whether or not for any positive integer d
there exists an irreducible polynomial of degree d over any given finite field.

Theorem 8: Let p be a prime and n a positive integer. For any positive

integer d, there is an irreducible polynomial of degree d over Fp n .

Proof: Let q = p n, then for d ≥ 1, q d = p nd . So Fqd is the splitting field of the

d
polynomial x q − x over Fp . Since n nd , Fq = Fpn is a subfield of Fq d , and
Fqd
contains Fp . Now, since is finite, it is a simple extension. So, there
Fq
exists ζ ∈ Fqd such that Fq d = Fq (ζ ). Thus,
[Fq d : Fp ] = [Fq d : Fq ][Fq : Fp ],
i.e., [Fp nd : Fp ] = [Fq (ζ ) : Fq ][Fp n : Fp ]
i.e., nd = [Fq (ζ) : Fq ] n.
Hence [Fq (ζ) : Fq ] = d.
If f ( x ) is the minimal polynomial of ζ over Fq , then f ( x ) is the required
irreducible polynomial of degree d over Fq .

As an immediate corollary, we have the following.

Corollary 4: For any prime p and any integer d ≥ 1, there exists an

154 irreducible polynomial of degree d over Fp .
The proof of the corollary is left to you. We move on to make an important Splitting Fields
– Finite and Infinite
comment about f ( x ) in the proof of Theorem 8.
d
Remark 4: You know that Fqd is the splitting field of x q − x over Fq . Hence
d
all the elements of Fqd are roots of the polynomial x q − x. In particular, ζ (in
d
Theorem 8) is also a root of x q − x. Since f ( x ) is the minimal polynomial of
d
ζ, f ( x ) (x q − x ) over Fq .

Now, a surprising result about irreducible polynomials over Fp , which Remark

4 may have given you an inkling of.

Theorem 9: Let p be a prime and n ≥ 1 an integer.

n
i) Every irreducible polynomial of degree n, over Fp , is a factor of x p − x.
n
ii) x p − x is the product of all the distinct irreducible polynomials over Fp
of degree d, where d varies over all the divisors of n.

r
Proof: Let F be the splitting field of x p − x over Fp . Then, by Theorem 5,
F = Fp r . Further, by Theorem 6, Fp r = Fp (α) for some α ∈ Fp r . As α is a root
r r
of x p − x, its minimal polynomial over Fp , f ( x ), must divide x p − x.
r
Moreover, as x p − x splits completely in Fp r , so does f ( x ), i.e., all the roots
of f ( x ) lie in Fp r .

Using the above, let us now prove (i) and (ii).

i) Let p( x ) be an irreducible polynomial over Fp of degree n. If θ is a
root of p( x ), in its splitting field, then the extension Fp (θ) is a subfield
of Fpn of degree n over Fp . Therefore, Fp (θ) = p n . Hence θ is a root
n n
of x p − x. Since p( x ) is irreducible, p( x ) ( x p − x ).

ii) Let d n. Then, as in (i), any irreducible polynomial of degree d is a

d n
factor of x p − x, which is a factor of x p − x, by E10.
So, every irreducible polynomial whose degree divides n is a factor of
n
x p − x.

Now, suppose f ( x ) is irreducible over Fp with degree r n. Then

r n n
( x p − x ) ( x p − x ). Hence f ( x ) ( x p − x ). Hence (ii) is proved.

So, what Theorem 9 says is that an irreducible polynomial of degree d over Fp

n
must be a factor of x p − x , where n is a multiple of d. This theorem gives us
an alternative method of constructing finite fields. It is also the basis of
methods by which polynomials are factored modulo p. 155
Field Theory Now, an important comment about counting the number of irreducible
polynomials.
n
Remark 5: As Theorem 9 shows us, x p − x is the product of all the
irreducible polynomials over Fp of degree d, taken over all the divisors d of
n. If N (d ) denotes the number of irreducible polynomials of degree d over
Fp , then on comparing degrees we get
p n = ∑ d N (d ).
d n

Using this equation, we can recursively determine the numbers N (d).

Let us consider some examples of the application of Theorem 9 and Remark 5.

Example 7: Find all the irreducible polynomials over F2 with roots in F4 .

Solution: You know that F4 is the splitting field of x 4 − x over F2 , and has
the factorisation x 4 − x = x ( x − 1) (x 2 + x + 1).
Here the factors x and x − 1 are distinct irreducible polynomials over F2 of
degree 1, corresponding to the divisor 1 of 2.
The factor x 2 + x + 1 is the irreducible polynomial of degree 2 over F2 ,
corresponding to the divisor 2 of 2.
***

Example 8: Obtain all the irreducible polynomials over F2 , which divide

x 8 − x. Hence obtain all the elements of F8.

Solution: As in Example 7, x and ( x − 1) are the two irreducible polynomials

over F2 of degree 1. Since F8 = 2 3 , any other irreducible polynomials have to
be of degree 3.
By Remark 5, 23 = N (1) + 3N (3), and you know that N (1) = 2. Hence,
N (3) = 2.
Therefore, x 8 − x has two irreducible cubic factors.
Now x 8 − x = x ( x − 1)( x 6 + x 5 + L + x + 1).
You can check that x 6 + x 5 + L + x + 1 is factored over F2 as
( x 3 + x 2 + 1) ( x 3 + x + 1). You should also verify that both these cubic factors
are irreducible over F2 . So their roots are the elements of F8 \ {0, 1}.
Suppose α is a root of x 3 + x 2 + 1. Then {1, α, α 2 } is an F2 -basis of F8. So
F8 = {0, 1, α , α 2 , 1 + α, 1 + α 2 , α + α 2 , 1 + α + α 2 }. Note that α 3 + α 2 + 1 = 0.
***

Example 9: Write x 16 − x as a product of distinct irreducible polynomials over

F2 .

Solution: F24 is the splitting field of x 16 − x over F2 . As [F24 : F2 ] = 4, an

irreducible factor of x 16 − x can have degree 1, 2 or 4. Now, the only linear
156
factors possible over F2 are x and x − 1. Also
 Splitting Fields
x 16 − x = x ( x15 − 1) = x ( x 5 − 1)( x10 + x 5 + 1) – Finite and Infinite
= x (x − 1)(x 4 + x 3 + x 2 + x + 1)( x10 + x 5 + 1).
Since the polynomial f ( x ) = x 4 + x 3 + x 2 + x + 1 has no roots in F2 , it is either
irreducible over F2 or has an irreducible factor of degree 2. But the only
irreducible polynomial of degree 2 over F2 is x 2 + x + 1, which does not divide
f ( x ). So f ( x ) must be irreducible over F2 .
On dividing x 10 + x 5 + 1 by x 2 + x + 1, we have the factorisation
x 10 + x 5 + 1 = ( x 2 + x + 1)( x 8 + x 7 + x 5 + x 4 + x 3 + x + 1). …(3)
On adding 2x 4 (= 0 in F2 ) to the second factor on the right hand side of (3),
we get
x 8 + x 7 + x 5 + x 4 + x 3 + x + 1 + 2x 4 = x 4 (x 4 + x 3 + 1) + x ( x 4 + x 3 + 1) + ( x 4 + x 3 + 1)
= (x 4 + x 3 + 1)(x 4 + x + 1). …(4)
Arguing as before, you can check that both the factors on the right hand side of
(4) are irreducible over F2 . Hence, the irreducible factorisation of x 16 − x over
F2 is x (x − 1)(x 2 + x + 1)(x 4 + x 3 + x 2 + x + 1)(x 4 + x 3 + 1)(x 4 + x + 1).

***

Try some exercises now.

E12) Write x 27 − x as a product of irreducible factors over F3 . Is this also the

irreducible factorisation over Q of x 27 − x ? Give reasons for your
answer.

E13) With reference to Example 8, let K = F2 (α) and L = F2 (β), where β is a

root of x 3 + x + 1. Give an explicit isomorphism from K to L.

E14) i) Show that x p − x − a is irreducible over Fp , where a ≠ 0.

ii) Show that x p − x − a is irreducible over Fpn , n ≥ 1, iff it has no

linear factor.

With this we come to the end of this discussion on finite (splitting) fields and
irreducible polynomials over Fp . Let us take a brief look at what you have
studied in this unit.

12.5 SUMMARY
In this unit, we have covered the following points.

1. The existence, and uniqueness, of a splitting field of a given polynomial

over a given field.

2. Every finite field is a splitting field over Fp , for some prime p.

157
Field Theory 3. For every prime p, and n ∈ N, there is a unique field (up to
n
isomorphism) of order p n, which is the splitting field of x p − x over Fp .
These are the only finite fields.

4. How to explicitly construct a finite field of order p n.

5. Every finite extension of a finite field is simple.

6. Fpm is a subfield of Fpn iff m n.

7. Given a prime p and n ∈ N, there is an irreducible polynomial of degree

d over Fpn ∀ d ∈ N.

n
8. Given a prime p and n ∈ N, x p − x is the product of all the distinct
irreducible polynomials over Fp of degree d, where d varies over all the
divisors of n.

9. p n = ∑ d N (d ), where N (d ) is the number of irreducible polynomials of

d n

degree d over Fp .

12.6 SOLUTIONS / ANSWERS

E1) Let f ( x ) = a 0 + a 1x + L + a n x n and g( x ) = b 0 + b1x + L + b m x m be two

arbitrary polynomials over k. We can assume, without loss of generality,
that n ≤ m. Then
f ( x ) + g( x ) = (a 0 + b 0 ) + (a 1 + b1 ) x + L + (a n + b n ) x n + b n +1x n +1 + L + b m x m
and
 
f ( x )g ( x ) = (a 0 b 0 ) + (a 0 b1 + a 1b 0 ) x + L +  ∑ a i b j  x k + L + (a n b m ) x n + m .
 i + j=k 
Now, on applying σ, we get ~
σ~ (f (x ) + g( x ))
= σ(a 0 + b 0 ) + σ(a 1 + b1 )x + L + σ(a n + b n ) x n + σ(b n +1 ) x n+1 + L + σ(b m ) x m
= (σ(a 0 ) + σ(b 0 )) + (σ(a 1 ) + σ(b1 ))x + L + (σ(a n ) + σ(b n ))x n + L + σ(b m ) x m
= [σ(a 0 ) + σ(a 1 ) x + L + σ(a n ) x n ] + [σ(b 0 ) + σ(b1 ) x + L + σ(b m ) x m ]
=σ ~ (f ( x )) + σ
~ (g( x )),
and
~ (f (x )g( x ))
σ
 
= σ(a 0 b 0 ) + σ(a 0 b1 + a 1b 0 )x + L + σ ∑ a i b j  x k + L + σ(a n b m ) x n +m
 i+ j=k 
= σ(a 0 )σ(b 0 ) + (σ(a 0 )σ(b1 ) + σ(a 1 )σ(b 0 ))x + L
 
+  ∑ σ(a i )σ(b j )  x k + L + σ(a n )σ(b m ) x n +m
 i+ j=k 
158
 Splitting Fields
= [σ(a 0 ) + σ(a 1 ) x + L + σ(a n ) x n ][(σ(b 0 ) + σ(b1 ) x + L + σ(b m )x m ] – Finite and Infinite
=σ~ (f ( x ))σ
~ (g( x )).
Thus, σ ~ is a ring homomorphism.

~ (f ( x )) = σ
Next, if σ ~ (g ( x )), then
~ (a + a x + L + a x n ) = σ
σ ~ (b + b x + L + b x m )
0 1 n 0 1 m

⇒ σ(a 0 ) + σ(a 1 ) x + L + σ(a n ) x n = σ(b 0 ) + σ(b1 ) x + L + σ(b m ) x m

⇒ n = m, and σ(a i ) = σ(b i ) ∀ i ∈ {0, 1, K , n}.
Since σ is one-one on k and a i , b i ∈ k, a i = b i ∀ i.
∴ f (x) = g(x)
∴σ~ is 1-1.

To show surjectivity, let f ′( x ) = a ′0 + a ′1x + L + a ′n x n be an arbitrary

polynomial in k′[ x ]. Since a ′0 , a 1′ , K , a ′n ∈ k ′ and σ : k → k ′ is
surjective, there exist a 0 , a 1 , K , a n ∈ k such that σ(a i ) = a ′i , 0 ≤ i ≤ n.
Thus, ∃ f ( x ) = a 0 + a 1x + L + a n x n ∈ k[x ] such that
~ (f (x )) = σ(a ) + σ(a ) x + L + σ(a ) x n = f ′(x ).
σ 0 1 n

Finally, let f ( x ) be an irreducible polynomial over k and assume that its

image f ′( x ) = σ ~ (f ( x )) is reducible over k ′. Then there exist non-constant
polynomials g′(x ), h ′( x ) ∈ k ′[ x ] such that f ′( x ) = g′( x )h ′( x ).
By the surjectivity of σ ~ on k[ x ], there exist g( x ), h ( x ) ∈ k[x ] such that
σ~ (g ( x )) = g′( x ) and σ ~ (h (x )) = h′( x ).
Since σ preserves degree, g( x ) and h ( x ) must be non-constant. Also
σ~ (f (x )) = f ′( x ) = g′( x )h ′(x ) = σ~ (g ( x ))σ
~ (h ( x ))
⇒ f ( x ) = g (x )h ( x ), as σ ~ is 1-1.
This contradicts the irreducibility of f ( x ).
Hence f ′( x ) is irreducible in k′[ x ].

E2) Let f ( x ) , g( x ) ∈ k[ x ] have a common factor over some extension field F

of k. Suppose they are relatively prime in k[ x ], then there exist
polynomials a ( x ) , b( x ) in k[ x ] such that a ( x )f ( x ) + b(x )g( x ) = 1. Since
k ⊆ F, this relation will also hold over F, i.e., f ( x ) and g( x ) must be
relatively prime in F[ x ], which is a contradiction.
Hence f ( x ) and g( x ) have a common factor over k also.

E3) i) The roots of f ( x ) are 21/ 3 , 21/ 3 ω, 21/ 3 ω2 , ω, ω2 , where ω is a non-

real cube root of unity.
Thus, the splitting field of f ( x ) over Q is
Q(21/ 3 , 21/ 3 ω, 21/ 3 ω2 , ω, ω2 ) = Q(21/ 3 , ω).

ii) Q(51/ 3 , 51/ 3 ω, 51/ 3 ω2 ) = Q(51/ 3 , ω).

iii) The roots of f ( x ) are 1 + ω, 1 + ω2 ∈ Q(ω).

Hence Q(ω) is the splitting field over itself.
159
Field Theory iv) Over Z 2 , x 6 + 1 = x 6 − 1 = ( x 3 − 1) 2 = ( x + 1) 2 ( x 2 + x + 1) 2 .
The roots are 1,1, α, α, 1 + α, 1 + α, where α is a root of the
irreducible polynomial x 2 + x + 1 over Z 2 .
Hence, the splitting field of f ( x ) over Z 2 is Z 2 (α).

E4) i) The splitting field of x 2 + 1 over Q is Q(i), which is a degree 2

extension of Q.

ii) The polynomial f ( x ) = x 4 + x 2 + 1 factors over Q as

x 4 + x 2 + 1 = ( x 2 + 1) 2 − x 2 = ( x 2 − x + 1)( x 2 + x + 1), so that the
roots of f ( x ) are ± ω, ± ω2 . Hence the splitting field of x 4 + x 2 + 1
over Q is Q(ω, ω2 ) = Q(ω), which is a degree 2 extension of Q,
since ω satisfies the irreducible polynomial x 2 + x + 1 over Q.

iii) The splitting field of x 4 + 2 over Q is Q(21/ 4 , i). Now,

[Q(21/ 4 , i) : Q] = [Q(21/ 4 , i) : Q(21/ 4 )][Q(21/ 4 ) : Q] = 2 × 4 = 8,
since i satisfies the irreducible polynomial x 2 + 1 ∈ Q(21/ 4 ) and
21/ 4 satisfies the irreducible polynomial x 4 + 2 ∈ Q (irreducible by
Eisenstein’s criterion).
Hence 4 < [Q(21/ 4 , i) : Q] < 4!.

iv) Consider the polynomial f ( x ) = x 3 − 2. You can check it is

irreducible over Q. Its roots are 21/ 3 , 21/ 3 ω, 21/ 3 ω2 . So the splitting
field of f ( x ) over Q is Q(21/ 3 , ω). Hence,
[Q(21/ 3 , ω) : Q] = [Q(21/ 3 ) (ω) : Q(21/ 3 )][Q(21/ 3 ) : Q] = 2 ⋅ 3 = 6 = 3!.

E5) Let F be a finite field. The multiplicative group F∗ = F \ {0} is a finite

n −1
group of order p n − 1. So a p = 1 for every non-zero element a in F.
n
Therefore, the polynomial x p − x has p n distinct roots in F. Hence the
field F is the splitting field of f ( x ) over FP .

E6) i) The field is F9 , the splitting field of x 9 − x, over F3 .

Now x 9 − x = x ( x − 1)(x 7 + x 6 + L + x + 1)
= x (x − 1)(x − 2)(x 2 + 1)( x 2 − x − 1)(x 2 + x − 1) as a product of
irreducible polynomials over Z 3 .
If α is a root of x 2 + 1 over Z 3 , then {1, α} is an F3 -basis of F9 .
So F9 = {0, 1, 2, α, 2α, 1 + α, 1 + 2α, 2 + α, 2 + 2α}.

ii) 0, 1, 2 satisfy x , x − 1, x − 2, respectively.

α, 2α satisfy x 2 + 1,
1 + α, 1 + 2α satisfy x 2 + x − 1
2 + α, 2 + 2α satisfy x 2 − x − 1.

n
−1
E7) Let K = p n . Then x p −1 = ∏ ( x − α ).
i
160 αi ∈K∗
 Splitting Fields
n  
Putting x = 0 in this gives (−1) p −1  ∏ αi  = −1. – Finite and Infinite
 i 
n
For p ≠ 2, p − 1 is even, and hence the result.
If p = 2, − 1 = 1, and hence the result.

E8) 3 ∈ F13∗ . Hence, 312 = 1, so that 313 = 3.

Thus, 3 is a 13th root of 3 in F13 .

E9) From E6, you see that F9 = F3 (α), where α 2 + 1 = 0.

Thus, α is one primitive element.
Similarly, you can check that F9 = F3 (β), where β = 1 + α.
Clearly, β ≠ α.

E10) Let n = km + r, 0 ≤ r < m. Then

x n − 1 = x r (x m ) k − 1 = x r ( x mk − 1) + (x r − 1)
 k −1 
= x r ( x m − 1) ∑ x im  + ( x r − 1).
 i =0 
Therefore, x − 1 divides x n − 1 iff x r − 1 = 0, i.e., iff r = 0, i.e., iff m
m

divides n. On taking x = p, a prime number, we have p m − 1 divides

p n − 1 if and only if m divides n.
m n m n
−1 −1
Thus, x p − x divides x p − x iff x p − 1 divides x p − 1 iff p m − 1
divides p n − 1 iff m divides n.

E11) Since both the fields Fp8 , Fp12 have characteristic p, they have the same
prime field, Fp . Now, by Theorem 7, Fpm is a subfield of Fpn if and only
if m divides n.
Therefore Fp , Fp2 , Fp4 are the only proper subfields of FP8 , corresponding
to the divisors 1, 2 and 4 of 8.
Similarly, as 1, 2, 4 and 6 are the only divisors of 12, Fp , Fp2 , Fp4 , Fp6
are the proper subfields of Fp12 .

E12) F33 is the splitting field of x 27 − x over F3 .

x , x − 1, x − 2 are the only linear factors, corresponding to the divisor 1
of 3. So N (1) = 3.
Now, by Theorem 9, any other irreducible factor has to be of degree 3.
Also, by Remark 5, 33 = N (1) + 3N (3), so that N (3) = 8.
Hence, we need to find 8 cubic irreducible polynomials over F3 , with
their roots in F27 . You should check that these are
x 3 + x + 1, x 3 − x + 1, x 3 + x − 1, x 3 − x − 1, x 3 + x 2 + 1, x 3 − x 2 + 1,
x 3 + x 2 − 1, x 3 − x 2 − 1.
Thus, x 27 − x is the product of these 8 cubic polynomials and the 3 linear
polynomials.
161
Field Theory Over Q, x 27 − x = x ( x 13 − 1)(x 13 + 1)
= x (x − 1)(x + 1)(x 12 + x11+ L + x + 1)(x 12 − x11 + x10 − L − x + 1) …(5)
From Example 11, Unit 9, you know that x 12+ L + x + 1 is irreducible
over Q. By a similar argument to that in the example, you can check that
x 12− x11+ x10 − L + x 2 − x + 1 is irreducible over Q. Hence, the
factorisation in (5) above is the required one, which is very different from
the factorisation over F 3 .

E13) Since β = 1 + α, K = L. Thus, the identity map is an explicit isomorphism

from K to L.

E14) i) Let f ( x ) = x p − x − a , a ≠ 0. For any b ∈ Fp , b p = b. Therefore,

f (b) = b p − b − a = −a ≠ 0, i.e., f ( x ) has no roots in Fp . So f ( x )
has no linear factors over Fp . Moreover, if α is a root of f ( x ) in
some extension of Fp , then so is α + i for every i ∈ {0, 1, K , p − 1}.
p−1
So f ( x ) = ∏ [ x − (α + i)] in the splitting field K of f ( x ).
i=0

Now, if p( x ) is an irreducible factor of f ( x ) over Fp of degree m,

then p( x ) f ( x ).
∴ p( x ) = ( x − α − i1 )(x − α − i 2 ) L (x − α − i m ) in K.
But p( x ) ∈ Fp [ x ], so that the coefficient of x m −1 belongs to Fp ,
m m
i.e., ∑ (α − i k ) ∈ Fp , i.e., mα − ∑ ik ∈ Fp .
k =0 k =0
m
As ∑i
k =0
k (mod p) ∈ Fp . So mα ∈ Fp . Since 0 < m < p, m and p are

coprime. So m −1 ∈ Fp , and hence α ∈ Fp , which contradicts the

fact that f ( x ) has no linear factor over Fp .

ii) The second part follows on similar lines as (i) above.

162
 A Gateway to Galois
UNIT 13 A GATEWAY TO GALOIS THEORY Theory

Structure Page No.

13.1 Introduction 163
Objectives
13.2 Some Special Finite Extensions 164
13.3 Homomorphisms of Extensions of Fields 169
13.4 Normal Extensions and Separable Extensions 174
13.5 Galois Extensions 177
13.6 Summary 181
13.7 Solutions / Answers 182

13.1 INTRODUCTION

In maths, his rationality was sound;

In life his levelheadedness unwound.
From rational and real one may abscond
To meddle in complexities unbound.

Unsolved are quintics and beyond,

Unsolvable, the academic monde;
Revolt can soothe a young man’s fitful zeal;
A uniform, a sword, an aim au fond.
From ‘Evariste Galois’, by Diane Hine

In the previous units of this block, you have studied finite extensions of fields
and finite fields. In this unit, we shall focus on finite extensions of the field Q,
and lay the foundations to help you understand Galois theory. For simplicity,
we will restrict our attention to subfields of C. So, unless otherwise stated, all
fields considered in this unit are subfields of C.

Throughout this unit, you will be walking the path that takes you into the
precincts of Galois theory. To start with, in Sec.13.2, you will be building on
what you studied in Unit 11 about finite extensions. Here we will discuss
extensions of degree 2, 3 and 4, to start with, in the context of finding the roots
Fig. 1: Evariste Galois
of polynomials of degree 2, 3 and 4 in Q[x ]. Then you will study some (1811-1832), a
properties of cyclotomic extensions and simple extensions whose primitive remarkable French
elements are the sum or product of radicals in C. mathematician who
died young, after a
duel.
In the next section, Sec.13.3, we focus on isomorphisms between extensions of
a field. You will see here that if L K is a finite extension, then there are
[L : K ] K-homomorphisms from L into C. Here you will also prove that every
finite extension is simple.

In Sec.13.4, the focus is on two particular types of algebraic extensions that

Galois particularly worked with. Here you will build on your understanding of
splitting fields from Unit 12.

Finally, we will use the background developed so far to introduce you to a

certain kind of finite extension, called a Galois extension. While studying these
extensions, you would realise that you have already worked with many such
extensions in Unit 12. It is these extensions, and their properties, around which
the next unit is developed.
163
Field Theory Let us now clearly state the expected outcomes of your study of this unit.

Objectives

After studying this unit, you should be able to

• state, prove and apply, properties of finite extensions of degree ≤ 4 ;
• find the degree of the cyclotomic extension Q(ζ ) over Q, where ζ is a
primitive pth root of unity, p being a prime;
• obtain the number of isomorphisms and automorphisms of finite
extensions of Q;
• prove, and apply, the theorem that every finite extension of Q is simple;
• define, and give examples of, normal extensions;
• define, and give examples of, separable extensions;
• define, and give examples of, Galois extensions;
• obtain the Galois group of some Galois extensions.

13.2 SOME SPECIAL FINITE EXTENSIONS

In Unit 11, you studied several examples of finite extensions of Q. You have
also seen, in Unit 9, that the irreducible polynomials over C are linear, those
over R are either linear or quadratic; and over Q you can obtain irreducible
polynomials of any degree. However, finding all the roots of an irreducible
polynomial over Q (in its splitting field), may not be easy, or even possible.
And what if the polynomial is over K, a subfield of C ? Let us consider some
cases.

We first look at degree 2, 3 and 4 extensions of K. Let us define some related

terms.

Definitions: Let K be a field. A finite extension L K is called:

i) a quadratic extension if [L : K ] = 2;
ii) a cubic extension if [L : K ] = 3;
iii) a quartic extension if [L : K ] = 4.

These terms come from the fact that if f ( x ) ∈ K[x ] is an irreducible quadratic
(respectively, cubic, quartic) polynomial, then L = K[x ] is a quadratic
< f (x) >
(respectively, cubic, quartic) extension of K.

As you have studied in Unit 11, k (α) k will be a quadratic, cubic or quartic
extension, depending on whether m α , k ( x ) is a quadratic, cubic or quartic
polynomial. For example, since x 2 + 1 ∈ Q[ x ] is irreducible, with roots ± i,
[Q(i) : Q] = 2. Hence Q(i) Q is a quadratic extension.
Similarly, Q (21 / 3 ) Q is a cubic extension and Q(71 / 4 ) Q is a quartic
164
extension.
 A Gateway to Galois
You know that if ax 2 + bx + c ∈ K[ x ], K a subfield of C, then its roots are Theory
− b ± b 2 − 4ac
. Recall that D = b 2 − 4ac is called the discriminant of
2a
2
ax + bx + c. The discriminant tells us many things about the nature of the
roots of the quadratic, namely, if D > 0, they are real; if D = 0, they are
repeated; if D < 0, the roots are non-real complex numbers.

To help you recall some properties of quadratic polynomials and quadratic

extensions, do the following exercises.

E1) Prove that if α1, α 2 ∈ C are the roots of f ( x ) = ax 2 + bx + c ∈ K[ x ], then

the discriminant of f ( x ), D = [a (α1 − α 2 )]2 , so that D = a (α1 − α 2 ) .

E2) Prove that a quadratic polynomial over K is reducible over K if and

only if its roots are in K, where K is a subfield of C.

E3) i) Prove that a subfield L of C is a quadratic extension of Q if and

only if L = Q( d ), for some rational number d which is not the
square of a rational number. In fact, this is true if Q is replaced by
any subfield K of C.
[Hint: d can be taken to be the discriminant of the quadratic
polynomial.]

ii) Prove that we can take d to be a square-free integer in (i) above.

iii) Let ω be a root of x 2 + x + 1 ∈ Q[ x ]. Prove that Q(ω) = Q( − 3 ).

Now consider cubic polynomials and cubic extensions. You know that
x 3 ± 2, x 3 ± 3, x 3 ± 10, x 3 ± 25 are all irreducible over Q. In fact, x 3 − a is
irreducible over Q ∀ a ∈ Z such that a ≠ b 3 for any b ∈ Z.
As you know, it is not easy to find the roots of a cubic polynomial over K.
However, there is a method, due to the Italian mathematician Scipione del
Ferro (1465-1526). Strangely, it is named after the Italian mathematician
Fig. 2: Scipione del Ferro
Cardano, probably because it became widely known after he published it in
1545 in his book ‘Ars Magna’.

According to Cardano’s method, the roots of x 3 + px 2 + qx + r = 0 are

If you are interested in
p p p −1+ i 3 learning more about
α + β − , αω + βω 2 − , αω 2 + βω − , where ω = , α is a cube
3 3 3 2 Cardano’s method,
please refer to ‘Higher
 B B2 A 3   B B2 A 3  Algebra’ by Hall and
root of − + + , β is a cube root of − − + , Knight, for example.
 2 4 27   2 4 27 
p2 2p 3 pq
A=q− , B= − + r.
3 27 3
Here, the discriminant is
D = −(27B 2 + 4A 3 ).
165
Field Theory For instance, the discriminant of x 3 − 3x + 1 is − 4(−3)3 − 27 = 81, and of
x 3 − 3x + 2 is − 4(−3)3 − 27(4) = 0.

If you go into the details, you will find that for

x 3 + px 2 + qx + r = 0, p, q, r ∈ K,
1) all the roots are real and distinct iff D > 0,
2) exactly one root is real and two are complex conjugates iff D < 0,
3) exactly two roots are equal, and all the roots are real, iff D = 0,
4) all three roots are equal iff D = 0 and B = 0.
Also, on using the relation between the roots and coefficients, you will find that
D = (α1 − α 2 ) 2 (α1 − α3 ) 2 (α 2 − α3 )2 , where α1, α 2 , α3 are the roots of the
cubic x 3 + ax 2 + bx + c ∈ K[ x ]. See how much information the discriminant of
a cubic gives us about the nature of its roots!

We can generalise the concept of the discriminant to that of a polynomial of

any degree over K[x ].

Definition: Let f ( x ) be a monic polynomial of degree n over K, with roots

α1 , α 2 , K, α n in C. Then D = ∏ (α i − α j ) 2 is called the discriminant of
i< j

f ( x ).

From this definition, you know that D is zero iff the polynomial has
repeated roots. Thus, for instance, x 3 − 3x + 2 has repeated roots.

Let us now look at quartic extensions. As you may know, quartic (or
biquadratic) equations have been studied for several centuries. The ancient
Arabs were known to study them from a geometric point of view. The 16th
century Italian mathematician Ferrari, who worked with Cardano, gave an
algebraic method for solving such equations. He reduced the problem to that of
finding a solution of a related cubic equation, called the resolvent cubic.

Another method for obtaining an algebraic solution for a quartic was given by
the mathematician René Descartes in 1637. In his method, he first removed the
cubic term. Then he wrote the biquadratic as a product of two quadratic
polynomials. On solving the resultant quadratic equations, he obtained all the
Fig. 3: Descartes
roots of the original quartic polynomial.
(pronounced they-cart)
Thus, in the case of a quartic too, as you can see, one can obtain its roots
algebraically, though in a cumbersome manner. Here, the discriminant of
For more details on x 4 + ax 3 + bx 2 + cx + d ∈ K[ x ] would be
Ferrari’s method or
Descartes’ method, you D = (α1 − α 2 ) 2 (α1 − α3 ) 2 (α1 − α4 )2 (α 2 − α3 ) 2 (α 2 − α 4 ) 2 (α3 − α 4 )2 ,
may refer to ‘Higher where α1, α 2 , α3 , α 4 are the roots of the given quartic.
Algebra’ by Hall and
Knight, for example. Again, D = 0 iff α i = α j for some i ≠ j.

Now, assume f ( x ) is a quartic polynomial which is irreducible over K (⊆ C),

and α ∈ C is a root of f ( x ). Then, from Unit 11, you know that
[K (α) : K ] = 4. Conversely, if L is a subfield of C containing K, and if
166
[L : K ] = 4, then you will see in Sec.13.3 that there exists a complex number α A Gateway to Galois
Theory
such that K (α) = L. You can also prove this directly (see E5 below).

E4) i) Prove the statement, ‘A cubic polynomial over Q is reducible over

Q if and only if it has a root in Q ’. Is the statement true if ‘cubic’
is replaced by ‘quartic’? Why, or why not?
ii) Can you replace Q by any subfield of C in the statement in (i)
above? Justify your answer.

E5) Let [L : K ] = 4. By proving the following parts, you will be able to prove
that L = K ( γ ) for some γ in L.
i) Let α ∈ L \ K and K (α) ≠ L. Prove that
[L : K (α)] = 2 = [K (α) : K ].

ii) In the situation of (i) above, let K (α) = K′. Show that K′ = K ( d )
and L = K ′( β ) = K ( d , β ), where d ∈ K , β ∈ K′ are not
squares in K and K′, respectively.

iii) Show that L = K ( γ), where γ = α + β , where α, β ∈ K but

α , β ∉ K.

So far you have seen that there are formulae for solutions of quadratic, cubic
A radical is an nth
and quartic equations. You have also seen that these formulae involve taking root of a number,
radicals and various rational combinations of these radicals. So, you may n ∈ N \ {1}.
wonder whether one can write down such formulae for the solutions of
polynomials of any degree. In fact, historically that is what generations of
mathematicians tried to prove, but in vain. Today, due to the work of many
mathematicians, we know that we cannot write such a formula for equations
of degree higher than 4. But it was the crowning achievement of the young
mathematician, Evariste Galois, who solved this mystery completely. You shall
study more about this theory in the next unit. For now, let us continue our
exploration of some finite extensions.

We shall now discuss simple extensions generated by an nth root of unity. As

you would recall from the previous unit, a complex number ζ is called an nth
root of unity, if ζ n = 1. Also, ζ ∈ C is called a primitive nth root of unity if
ζ n = 1 but ζ m ≠ 1 for any m < n.
For instance, i = − 1 is a primitive 4th root of unity, while (−1) is a 4th root of
unity which is not primitive.
You have also studied that if ζ is a primitive nth root of unity and K = Q(ζ ),
then K is called a cyclotomic field, sometimes called the nth cyclotomic
field.

In Block 1, you have seen that the nth roots of unity in C are given by
m
2πi
ζ = e n ,0 ≤ m ≤ n − 1, and that they form a cyclic group of order n with
respect to multiplication. Now here is a related exercise.
167
Field Theory
m
2 πi
E6) Prove that ζ = e n
is a primitive nth root of unity if and only if
g.c.d (m , n ) = 1.

Let us now consider a special case of cyclotomic fields, namely, when n = p, a

prime.

Theorem 1: Let K = Q(ζ ) , where ζ is a primitive pth root of unity, p a

prime. Then, [K : Q] = p − 1.

Proof: x p − 1 = (x − 1)g ( x ), …(1)

p −1 p −2
where g( x ) = x + x + L + x + 1.
Since ζ is a primitive pth root of unity, ζ − 1 ≠ 0. Therefore, ζ is a root of
g( x ). You have seen in Unit 9 that g( x ) is irreducible over Q. Therefore, the
minimal polynomial of ζ is g( x ), which is of degree p − 1. Therefore,
[K : Q] = p − 1.

The theorem above does not generalise to the nth cyclotomic field, where n is
composite. (For instance, you know that [Q(i) : Q] = 2, where Q(i) is the 4th
cyclotomic field.) What is true is that for a general n,
[Q(ζ ) : Q] = φ(n), where ζ is a primitive nth root of unity and φ is the Euler
function defined in Definition 3, Unit 10. We shall not prove this fact in this
course. However, let us consider an example of its application.

Example 1: Show that the 8th cyclotomic field is Q(i, 2 ).

Solution: Now [Q(ζ ) : Q] = φ(8) = 4, where ζ is a primitive 8th root of unity.

Also, Q(ζ ) contains all the 4th roots of unity, and hence the 4th cyclotomic
field. So Q(i) ⊆ Q(ζ).
1+ i 1− i
Further, ζ = e2 π i / 8 , so that ζ + ζ 7 = + = 2.
2 2
So, 2 ∈ Q(ζ).
∴ Q(i, 2 ) ⊆ Q(ζ ).
Also [Q(i, 2 ) : Q] = [Q(i, 2 ) : Q( 2 )][Q( 2 ) : Q] = 4.
So Q(i, 2 ) = Q(ζ ).
***

Try an exercise now.

E7) Prove that if K contains the nth roots of unity for an odd n, then it also
contains the (2n)th roots of unity. Also show that this need not be true if
n is even.

The final type of finite extension that we will mention in this section is
168 important for Galois theory.
Definition: Let K be a subfield of C and a ∈ K, where a is not an nth power A Gateway to Galois
Theory
of any element in K, where n > 1. Then the extension K (n a ) is called a
(simple) radical extension of K.
More generally, K (α1 ,K, α n ) is called a radical extension if
K ⊆ K (α1 ) ⊆ K (α1 , α 2 ) ⊆ K ⊆ K (α1 ,K, α n ) forms a tower of simple radical
extensions.

For example, Q(5 2 ) and Q( 2 , 3 ) are radical extensions of Q. You have

already worked with several such extensions in the earlier units. It is such
extensions that come into play while looking for roots of polynomials over Q,
as you will see in this unit, and the next one.

Now we shall focus on isomorphisms and automorphisms between extensions

of a field.

13.3 HOMOMORPHISMS OF EXTENSIONS OF

FIELDS
In this section, we will discuss homomorphisms between fields. As you know
from Unit 11, any such homomorphism is a monomorphism. Let us define
some terms now.

Definitions: Let K be a field.

i) An isomorphism σ of K with itself is called an automorphism of K.
The set of automorphisms of K is denoted by Aut(K).
ii) An automorphism σ ∈ Aut (K ) is said to fix an element α ∈ K if
σ(α) = α. An automorphism σ is said to fix a subset S of K if it fixes
every element of S, i.e., if σ(a ) = a for all a ∈ S.
iii) Let K F be an extension of fields. Then Aut(K F) denotes the set of
automorphisms of K which fix F.

Here is a related remark.

Remark 1: Note that

i) any field has at least one automorphism, the identity map, denoted by I,
and sometimes called the trivial automorphism.
ii) Aut (K F) ⊆ Aut (K ), where K F is an extension of fields.

Let us consider some properties of Aut (K ) now.

Proposition 1: If K is a field and F its prime subfield, then

Aut (K ) = Aut (K F). In particular, Q and Fp have only the trivial
automorphism: Aut (Q) = {I} and Aut (Fp ) = {I}.

Proof: The prime subfield of K is generated by 1 ∈ K. Since any

automorphism σ of K fixes 0 and 1, it follows that σ(a ) = a ∀ a ∈ F.
Hence Aut (K ) ⊆ Aut (K F).
Also, by definition, Aut (K F) ⊆ Aut K. Hence the result.
169
Field Theory We shall find Aut (K ) and Aut (K F) in some cases, for which we need to
prove a theorem. But first, do the following exercise.

E8) Prove that (Aut (K ), o) is a group, and Aut (K F) ≤ Aut (K ), where K F

is an extension.

Now, let us recall some terminology that we have used earlier.

i) Let L K be a finite extension and σ be a homomorphism from K to a
Recall that τ K field F. Then we say that a homorphism τ from L to F extends σ if
denotes ‘ τ restricted τ K = σ.
to K ’.
ii) In (i) above, when σ is the identity map of K, we say that τ is a
K-homorphism of L, where L K is finite.

So, for example, every element of Aut (L K ) is a K-isomorphism.

Now let us consider a basic theorem that will help us in finding isomorphisms
of a finite extension L K with another field F.

Theorem 2: Let α ∈ C be algebraic over K and β ∈ C. Then, there is a K-

γ isomorphism γ : K (α) → K (β), which maps α to β, if and only if β is
K (α ) K (β)
algebraic over K and the minimal polynomial of β is the same as the minimal
K K
polynomial of α over K.
Id
K K Proof: The statement of the theorem is depicted in Fig.4.
Fig. 4 Let f ( x ) be the minimal polynomial of α over K.
First, let us assume that β is algebraic over K and its minimal polynomial
over K is also f ( x ).
Let θ : K[x ] → C : θ(p( x )) = p(α) be the evaluation map at α, and
µ : K[x ] → C : µ(p( x )) = p(β) be the evaluation map at β.
Then θ(K[x ]) = K[α] and µ(K[ x ]) = K[β].
Also, Ker θ = < f ( x ) > = Ker µ.
So, by the Fundamental Theorem of Homomorphism, we have induced
isomorphisms θ : K[ x ] ~ K[ α] = K (α), since α is algebraic over
−
< f (x ) >
K, and µ : K[ x ] ~ K[β] = K (β), since β is algebraic over K.
−
< f (x) >
Let γ = µ θ −1. Then γ is a K-isomorphism from K[α] to K[β], and γ (α) = β.

Conversely, suppose γ : K (α) → K (β) is a K-isomorphism, and γ (α) = β.

Since f (α) = 0, γ (f (α)) = 0, i.e., f ( γ(α)) = f (β) = 0.
Therefore, β is algebraic over K. Since f ( x ) is irreducible over K, f ( x ) must
be the minimal polynomial of β over K.
The K-isomorphisms
are symmetries of the
set of roots of the What Theorem 2 says is that a K-isomorphism from K(α ) to C will take a
minimal polynomial of root of the minimal polynomial to another of its roots. This describes all the
α over K. K-isomorphisms from K (α) to C. Let us consider some examples of the
application of this result.
170
 A Gateway to Galois
Example 2: Find Aut (Q( 2 )) and Aut (Q( 2 ) Q). Theory

Solution: Firstly, by Proposition 1, Aut (Q( 2 )) = Aut (Q( 2 ) Q).

Next, let K = Q( 2 ). If τ ∈ Aut (Q( 2 )), then τ( 2 ) = 2 or τ( 2 ) = − 2 ,
since these are the two roots of the minimal polynomial for 2 .
Since τ fixes Q and {1, 2} is a basis of Q( 2 ) over Q, this determines τ
completely.
The map 2 a 2 is just the identity automorphism I of Q( 2 ).
The map σ : 2 a − 2 is the other isomorphism.
Hence τ = I or σ. Thus, Aut (Q( 2 )) = Aut (Q ( 2 ) Q) = {I, σ} is the cyclic
group of order 2 generated by σ.
***

Example 3: Find Aut (Q(3 2 Q) .

Solution: Let K = Q(3 2 ). If τ ∈ Aut (K Q), then τ is completely determined

by its action on 3
2 since {1, 3 2 , (3 2 ) 2} generates K over Q, and
τ(a + b3 2 + c(3 2 )2 ) = a + bτ(3 2 ) + c(τ(3 2 ))2 .
Now τ(3 2 ) must be a root of x 3 − 2, and the other two roots of this equation
are ω(3 2 ) and ω2 (3 2 ), which are not elements of K. Thus, the only
possibility for τ is τ(3 2 ) = 3 2 , i.e., τ = I.
Hence Aut (Q (3 2 ) Q) = {I}, the trivial group.
***

Try some exercises now.

E9) Let L be a quadratic extension of K ⊆ C. Prove that every

K-isomorphism θ from L to C is necessarily an automorphism of L.

E10) Find all the Q -homomorphisms from K = Q(3 2 ) into C.

Let us now apply Theorem 2 to obtain the number of distinct K-

homomorphisms of L K. But first a lemma.

Lemma 1: Let f ( x ) be an irreducible polynomial over K (⊆ C). Then the

roots of f ( x ) are all distinct.

Proof: Let α ∈ C be a root of f ( x ). Since f ( x ) is irreducible over K, it is the

minimal polynomial of α over K. Now, by Lemma 4 of Unit 12, you know
that α is a repeated root of f ( x ) if and only if α is a root of f ′( x ), where
f ′(x ) is the derivative of f ( x ). Also, deg f ′(x ) = deg f ( x ) − 1 and
f ′( x ) ∈ K[ x ]. Hence, α cannot be a root of f ′( x ). Therefore, α is not a
repeated root. Since α is an arbitrary root of f ( x ), it follows that f ( x ) has no
repeated root.

We now use Lemma 1, to prove the result we have aimed for.

171
Field Theory Theorem 3: Let K be a subfield of C. Let σ be a homomorphism from K
into C. Let L K be a finite extension of degree n. Then, there exist n distinct
homomorphisms from L into C which extend σ.
L
Proof: In Fig. 5, we depict the situation in the theorem.
τ
Let us first look at the special case: σ = I and L = K (α), where α is algebraic
K
σ
over K. Let f ( x ) be the minimal polynomial of α over K, with deg f ( x ) = n.
K C
K By Lemma 1, f ( x ) has n distinct roots in C, α = α1 , α 2 ,K , α n .
Fig. 5 By Theorem 2, for every i, we have a K-homomorphism τi from L to C,
where τi (α) = α i . They are all distinct, since the αi are distinct.
From Theorem 2, it also follows that there are no other homomorphisms of
L = K (α) into C since any such homomorphism must map α to a root of
f ( x ). In this case, note that [L : K ] = n since L = K (α). Thus, we have proved
the theorem in the special case when L = K (α) and σ K
= I.

The basic idea of the proof in the general case is the same as above. So, let
σ:K → ~ K ′ ⊆ C.
Now, by Lemma 1, Unit 12, you know that
 
σ′ : K[ x ] → K ′[ x ] : σ′ ∑ a i x i  = ∑ σ(a i )x i is an isomorphism of rings that
 i  i
extends σ. Also, f ( x ) is irreducible over K if and only if σ′(f ( x )) is
irreducible over K′.
In this situation, Theorem 2 can be modified to the following:
Let α ∈ C be algebraic over K, and β ∈ C. Then there is an isomorphism
σ : K (α ) →~ K ′(β) which extends σ, and maps α to β, if and only β is
algebraic over K ′ and the minimal polynomial of β over K ′ is σ (f (x )),
γ
K (α ) K ′(β) where f ( x ) is the minimal polynomial of α over K.
K K
Now, look at Fig. 6. As in the case of σ = I , it follows that the number of
σ homomorphisms of K (α) into C is deg f ( x ) = [K (α) : K ].
K K′
So we have succeeded in proving the theorem in the case L = K (α) and for any
Fig. 6
homomorphism σ from K into C.

To complete the proof of the theorem, we use induction on the degree of the
extension. So, let L be an extension of K of degree n, where L K is not
simple. Let us assume that the theorem is true for any finite extension F′ F of
fields, with [F′ : F] < n and any homomorphism τ from F into C. Let
α ∈ L \ K. Take F = K (α). We have already proved that there are
[F : K ] = m(≤ n ) homomorphisms from F into C extending σ. Let
τi ,1 ≤ i ≤ m, be these homomorphisms.
Now, α ∈ L \ K implies that [L : F] < n. Therefore, by the induction hypothesis,
there exist exactly [L : F] homomorphisms whose restriction to F is τ i . This is
true for each i = 1,K, m. Moreover, if γ is any homomorphism from L into C
whose restriction to K is σ, then its restriction to F must be τi , for some i.
Thus, the number of homomorphisms from L into C extending σ must be
equal to [L : F] ⋅ [F : K ], i.e., [L : K ].
Hence, by induction, this completes the proof of the theorem.
172
The following corollary is immediate from the theorem, by taking K′ = K and A Gateway to Galois
Theory
σ to be the identity map on K.

Corollary 1: Let L K be a finite extension. Then there are [L : K ]

K-homomorphisms from L into C.

Let us now consider an application of Theorem 3. In most of the examples of

finite extensions L of Q, that you have studied so far, L = Q(α) for some
element α ∈ L, that is, L Q is simple. In fact, the following amazing theorem,
due to the mathematician Emil Artin, tells us that these are the only possible
finite extensions of Q.

Theorem 4 (Primitive Element Theorem): Let L K be a finite extension,

where K ⊆ C. Then ∃ α ∈ L such that L = K (α).
You will see, in the next
Proof: Since L K is a finite extension, it is generated by finitely many section, that L K (in
elements over K, say, L = K (α1 , α 2 ,K, α m ). We shall prove that every finite Theorem 4) is separable.
extension L K which is generated by two elements is simple. Then the
theorem will follow by induction on the number of generators.

So, let L = K (α , β) and [L : K ] = n. If α ∈ K (β), then L = K (β), and we are

through. Similarly, if β ∈ K (α), then L is simple.
So, assume that α ∉ K (β) and β ∉ K (α). We shall construct a γ ∈ L \ K such
that L = K ( γ).

By Theorem 3, there are n K-homomorphisms σ1 , σ 2 ,K, σ n from L to C.

These homomorphisms are completely determined by their values on α and β,
that is, for two K-homomorphisms of L, τ and τ′, τ = τ′ if and only if
τ(α) = τ′(α) and τ(β) = τ′(β). Thus, if i ≠ j, then either σi (α ) ≠ σ j (α ) or
σ i (β) ≠ σ j (β).
Now, we claim that there exists an element c ∈ K such that
σ i (α + cβ) ≠ σ j (α + cβ) ∀ i ≠ j. To prove this, first observe that
σ i (α + cβ) = σ i (α) + cσ i (β), since the σ i are homomorphisms.
Next, note that there are only finitely many solutions of the linear system
σi (α ) + xσi (β) = σ j (α ) + xσ j (β), for 1 ≤ i ≠ j ≤ n.
But the field K is infinite. So choose c ∈ K such that
σi (α ) + cσi (β) ≠ σ j (α ) + cσ j (β), for i ≠ j; that is, such that the σ i (α) + cσ i (β)
are all distinct.
Now, let γ = α + cβ for such a c, and let F = K ( γ ). Then the σi ,1 ≤ i ≤ n , are
distinct homomorphisms from F to C. Therefore, by Theorem 3, [F : K ] ≥ n.
Also F is a subfield of L, so that [F : K ] ≤ [L : K ] = n.
Thus, [F : K ] = [L : K ], and hence, F = L.

Here is a related remark.

Remark 2: Note that by the theorem above, and by Theorem 6 of Unit 12, a
A simple extension can
finite extension of any field is simple.
also be generated by
In Unit 11, you already saw one application of Theorem 4, when you proved more than one element.

that Q( 2 , 3 ) = Q( 2 + 3 ). Let us consider another example of this.

173
Field Theory Example 4: Show that Q( 2 , 3 , 17 ) = Q( 2 + 3 + 17 ).
Solution: From Unit 11, you know that
L = Q( 2 , 3, 17 ) = Q( 2 + 3 , 17 ).
Now, 2 + 3 and 17 are in L, so 2 + 3 + 17 ∈ L.
Hence Q( 2 + 3 + 17 ) ⊆ L.
Also 17 ∉ Q( 2 + 3 ), which you can verify just as you checked that
3 ∉ Q( 2 ).
So [L : Q( 2 + 3 )] = 2.
Now, there are two Q( 2 + 3 ) -homomorphisms of L into C, I and
σ : L → C : σ( 17 ) = − 17 .
Note that 2 + 3 + 17 ≠ 2 + 3 − 17 .
So, by Theorem 4, taking c = 1 gives us γ = 2 + 3 + 17 .
Hence L = Q( 2 + 3 + 17 ).
***

Here is a comment about Theorem 4.

Remark 3: Note that finiteness is an essential condition in Theorem 4. For
instance, Q( x1 , x 2 ) Q is an infinite extension, where Q( x1 , x 2 ) is the field of
rational function in two variables x1 and x 2 . This extension cannot be simple.

Try an exercise now.

E11) Show that Q(α , ω) = Q(α + ω), where α = 3 2 and ω is a primitive cube
root of unity.

Let us now consider two kinds of extensions of fields that Galois worked with.

13.4 NORMAL EXTENSIONS AND SEPARABLE

EXTENSIONS
In the previous unit, you have studied splitting fields, and several examples of
them over finite and infinite fields. So, given a polynomial f ( x ) of degree ≥ 1
over K, we can find an extension field L in which f ( x ) factors (splits) as a
product of linear factors. Do all polynomials with one root in this extension L,
split in L ? There actually are extensions L K such that whenever a
polynomial has one root in L, it has all the roots in L. Such fields do exist, and
we have a name for them.

Definition: An algebraic extension of fields L K is called normal if whenever

an irreducible polynomial f ( x ) ∈ K[x ] has one root in L, then all its roots are
in L.

For example, C R is normal, by the Fundamental Theorem of Algebra.

However, Q(21 / 3 ) Q is not normal since x 3− 2 ∈ Q[ x ] has one root in it, but
174
its other roots are non-real and hence not in Q(21 / 3 ).
The normal extensions that you will study here will be mostly finite. However, A Gateway to Galois
Theory
not all normal extensions are finite. Consider the following example.

Example 5: Give an example of an infinite normal extension of Q.

Solution: Consider L = {α ∈ C α is algebraic over Q}. In Example 10, Unit

11, you have seen that L Q is an infinite extension.
Also, by definition, if p( x ) ∈ Q[ x ] is irreducible, all its roots are algebraic over
Q, and hence lie in L. Thus, L Q is a normal extension.
***

Now, interestingly, if L K is a finite normal extension, then it must be the

splitting field of a polynomial over K, as you will now see.

Theorem 5: Let L K be a finite extension of fields. Then L K is normal, if

and only if L is the splitting field of a non-constant polynomial over K.

Proof: Let L K be a finite extension. By Theorem 4, L = K (α) for some

α ∈ C. Let f ( x ) be the minimal polynomial of α.
Now, first suppose L K is normal. Since f ( x ) has one root α in L, all the
roots of f ( x ) belong to L. Let L′ be the splitting field of f ( x ). Then L′,
being the smallest field containing all the roots of f ( x ), is contained in L.
Also, L = K (α) ⊂ L′, since α is a root of f ( x ). Thus, L = L′.

Conversely, suppose L K is the splitting field of a polynomial h ( x ). If

α1 , α 2 , K, α n are the roots of h ( x ), then L = K (α1 , α 2 ,K , α n ). For any
K-homomorphism σ of L into C , σ must map each αi to some α j .
Hence, σ(L) = L. …(2)
Now suppose g( x ) is an irreducible polynomial over K[x ], with one root β in
L. Let β′ be another root of g( x ). Then there is a K-homomorphism from L
to C which maps β to β′. Since the image of every such homomorphism is L,
by (2) above, it follows that β′ ∈ L. Since β′ was arbitrary, it follows that all
the roots of g( x ) lie in L. This is true for every irreducible polynomial g( x )
having a root in L. Therefore, L K is normal.

By applying this theorem, you know that Fpn Fp is normal, for any prime p,
and n ∈ N.

Try an exercise now.

E12) Which of the following extensions are normal? Give reasons for your
answers.
i) Q( x ) Q,
ii) Q( − 7 ) Q,
iii) Q( 2 , 3 ) Q( 2 ),
iv) Q(51/ 5 ) Q.
175
Field Theory Let us now consider another kind of extension field.

Definitions: Let K be a field, of characteristic 0 or of characteristic p.

i) An irreducible polynomial f ( x ) is called separable over K if it has no

repeated roots (in its splitting field). If f ( x ) is not separable over K, it is
called inseparable over K.

ii) An arbitrary polynomial f ( x ) ∈ K[ x ] is called separable over K if all its

irreducible factors are separable over K.
iii) An algebraic element α over K is called separable over K if its
minimal polynomial over K is separable over K.

iv) An algebraic extension L K is called a separable extension if every α

in L is separable over K. If L K is not separable, it is called
inseparable.

For example, x 2 + 1 and ( x 2 + 1) 2 are separable over Q. Similarly, x 2 + x + 1

is separable over F2 since its roots are α and 1 + α, where α 2 + α + 1 = 0.

Now, from Lemma 4, Unit 12, you know that f ( x ) ∈ K[ x ] has repeated roots
iff f ( x ) and f ′( x ) have a common factor. This fact will be used to prove the
following theorem.

Theorem 6: If K is a field of characteristic 0, then every algebraic extension

L K is separable.
If K is a field of characteristic p > 0, then an irreducible polynomial
f ( x ) ∈ K[ x ] is separable if and only if f ( x ) ≠ b 0 + b1x p + L + b t x tp ,
that is, f ( x ) ≠ g( x p ) for any g( x ) ∈ K[ x ].

Proof: Let f ( x ) ∈ K[ x ] be irreducible. As you have seen in Unit 12, f ( x ) is

inseparable iff f ( x ) and f ′( x ) have a common factor. Since f ( x ) is
irreducible over K, this is possible only if f ′( x ) = 0.
Now, let f ( x ) = a 0 + a 1 x + L + a n x n , a i ∈ K.
So f ′( x ) = 0 ⇒ ma m = 0 ∀ m, 1 ≤ m ≤ n.
Now, consider both possibilities for K.
1) char K = 0 : In this case, ma m = 0 iff a m = 0 ∀ m, 1 ≤ m ≤ n,
i.e., iff f ( x ) = a 0 , a contradiction.
Hence, if char K = 0, every irreducible polynomial over K is separable.

2) char K = p : In this case, ma m = 0 iff a m = 0 or p m, for m = 1, K, n ,

i.e., iff f ( x ) = a 0 + a p x p + L + a pt x pt .
Thus, f ( x ) is inseparable iff f ( x ) = g( x p ), for some g( x ) ∈ K[ x ].
Hence, if char K = p, f (x ) is separable iff f ( x ) ≠ g( x p ) for any
g( x ) ∈ K[ x ].

176 Let us look at some examples.

Example 6: Show that every finite field is a separable extension of its prime A Gateway to Galois
Theory
subfield.
n
Solution: Let the finite field be Fpn. Since x p − x ∈ Fp [ x ] has no repeated
roots, all its irreducible factors have no repeated roots. Hence, Fpn Fp is
separable.
***

Example 7: Give an example of a polynomial in Z[ x ], which is separable over

Q, but not separable when considered as a polynomial over Fp , for some
prime p.

Solution: Consider x 3 + 1 ∈ Z[x ]. Its roots over Q are − 1, − ω, − ω2 . Hence it

is separable over Q.
However, by Theorem 6, x 3 + 1 ∈ F3[ x ] is inseparable.
***

Try some exercises now.

E13) Let L K be separable, and E be a field s.t. L ⊇ E ⊇ K. Then prove that

L E and E K are separable.

E14) Show that a finite extension L K is separable iff there are [L : K ]

distinct K-homomorphisms of L into N, where N is any normal
extension of K containing L.

E15) Show that if α is separable over F, then F(α) F is separable.

E16) i) Every separable extension is a normal extension. True or false?

Why?

ii) Is the converse of the statement in (i) true? Why, or why not?

Now let us look at the kind of extensions which Galois built his theory around.

13.5 GALOIS EXTENSIONS

In the previous unit we proved the existence and uniqueness of the splitting
field of a given polynomial f ( x ) whose coefficients are in F. The main idea of
Galois theory (named after Evariste Galois) is to consider the relation of the
group of permutations of the roots of f ( x ) with the algebraic structure of its
splitting field. The connection, given by the Fundamental theorem of the next
unit, is a beautiful application of the unifying idea in mathematics that an
algebraic object acting on another provides structural information about both.

We begin with a definition.

177
Field Theory Definition: An extension L K , of subfields of C, is called a Galois extension
if it is finite and normal, that is, if it is the splitting field of a non-constant
polynomial over K.

Consider the following theorem, that gives an alternative definition of a Galois

extension.

Theorem 7: A finite extension L K , of subfields of C, is a Galois extension

if and only if Aut (L K ) = [L : K ].

Proof: Let L K be a Galois extension. Then L = K (α), for some α ∈ L, and

[L : K ] = deg α. Further, by Theorem 3 and Corollary 1, you know that the
number of K-homomorphisms of L into C is [L : K ], and each such map
permutes the roots of m α , K ( x ).
Since L K is a Galois extension, σ(α) ∈ L for every K-homomorphism σ
from L into C, so that σ ∈ Aut (L K ).
Thus, Aut (L K ) = [L : K ].

Conversely, let Aut (L K ) = [L : K ], and let f ( x ) be irreducible over K with

one root α in L. Then, for each root αi of f ( x ), ∃ σi ∈ Aut (L K ) s.t.
σi (α) = αi . Thus, all the roots of f ( x ) lie in L.
Hence L K is a Galois extension.

This theorem leads us to some definitions.

Definitions: Let L K be a Galois extension.

i) Aut (L K ) is called the Galois group of L K , and is denoted by
G(L K).
ii) For α ∈ L and σ ∈ G (L K ), σ(α) is called a conjugate (or Galois
conjugate) of α over K.

Every cyclic extension is

iii) L K is called an abelian extension if G (L K ) is an abelian group.
abelian.
iv) L K is called a cyclic extension if G (L K ) is a cyclic group.

Let us consider some examples.

Example 8: Show that a quadratic extension is a Galois extension. Also find its
Galois group.

Solution: Let L = K ( α ) , α ∉ K, be a quadratic extension, where

char K ≠ 2. Then the minimal polynomial of α is x 2 − α, and ± α are its
roots. Thus, there are two K-isomorphisms of L into C, namely, σ1 = I, which
maps α to itself, and the other being σ 2 , which maps α to − α . Both
σ1 and σ 2 map L to L, i.e., they are automorphisms of L. Thus,
Aut (L Q) = [L : Q]. Hence L Q is a Galois extension, and its Galois group is
a cyclic group of order 2, generated by σ 2 .

178 ***
 A Gateway to Galois
Example 9: Let L = Q( 2 , 3 ). Show that L Q is a Galois extension, and its Theory
Galois group is isomorphic to Klein’s four group; i.e., C2 × C2 , where C 2
denotes a cyclic group of order 2.

Solution: Let K = Q( 2 ). Then L = K ( 3 ). You have verified, in earlier

units, that 3 is not a square in K. So L K is a quadratic extension. Hence,
[L : Q] = [L : K ][K : Q] = 2 × 2 = 4.
Now σ1 ,σ 2 are automorphisms of Q( 2 ). There are two K-isomorphisms of
L K over each of the automorphisms σ1 and σ2 . These isomorphisms are
completely determined by their values on { 2 , 3}, since L = Q( 2 , 3 ). Let
us write them explicitly.
τ1 ( 2 ) = 2 , τ1 ( 3 ) = 3 
 extensions of σ1
τ 2 ( 2 ) = 2 , τ 2 ( 3 ) = − 3 
τ3 ( 2 ) = − 2 , τ3 ( 3 ) = 3 
 extensions of σ 2
τ 4 ( 2 ) = − 2 , τ 4 ( 3 ) = − 3.
Note that τ2 , τ3 , τ 4 are elements of order 2 and τ1 is the identity of this group.
Since Aut (L Q) = 4 = [L : Q], L Q is a Galois extension, and its Galois group
is < τ 2 > × < τ3 >, isomorphic to C 2 × C 2 .
***

Example 10: Let L = Q(3 2 , ω). Show that L Q is a Galois extension, and its
Galois group is isomorphic to the symmetric group S3 .

Solution: Since L is the splitting field of x 3 − 2 ∈ Q[x ], L K is a Galois

extension.
Let K = Q(3 2 ). Look at the tower of extensions Q ⊂ K ⊂ L. We know the
three Q -isomorphisms from K into C. They are given by
τ1 (3 2 ) = 3 2 , τ 2 (3 2 ) = ω(3 2 ), τ 3 (3 2 ) = ω2 (3 2 ), respectively.
Now L = K (ω), and [L : K ] = 2. Therefore, there are two isomorphisms of L
over τi , for each i = 1, 2, 3. And all these isomorphisms are all automorphisms
of L K. (Why?)
3
Let us list them, by specifying the images of 2 and ω under each.
i) γ1 (3 2 ) = 3 2 , γ1 (ω) = ω
ii) γ 2 (3 2 ) = 3 2 , γ 2 (ω) = ω2
iii) γ 3 (3 2 ) = ω(3 2 ), γ 3 (ω) = ω
iv) γ 4 (3 2 ) = ω(3 2 ), γ 4 (ω) = ω2
v) γ 5 (3 2 ) = ω2 (3 2 ), γ 5 (ω) = ω
vi) γ 6 (3 2 ) = ω2 (3 2 ), γ 6 (ω) = ω2 .

You can verify that γ 3 γ 4 (3 2 ) = ω2 (3 2 ), whereas γ 4 γ 3 (3 2 ) = 3 2 .

Thus, γ 3 γ 4 ≠ γ 4 γ 3 . Thus, G (L Q) is a non-abelian group of order 6, and hence,
isomorphic to S3 .
***
179
Field Theory Example 11: Let K = Q(ω) and L = K (3 2 ). Check whether or not L K is a
Galois extension. If it is, is it a cyclic extension? If L K is not a Galois
extension, find an extension F L such that F K is a Galois extension.

Solution: In this case [L : K ] = 3 and L is the splitting field of x 3 − 2 ∈ K[ x ].

So L K is a Galois extension. Its Galois group is of order 3, and hence is
cyclic.
***

Example 12: Let ζ be a primitive pth root of unity, p an odd prime. Show
that Q(ζ ) Q is an abelian extension of order p − 1.

Solution: From Theorem 1 you know that [Q(ζ ) : Q] = p − 1, since

g( x ) = x p−1 + x p−2 + L + x + 1 is the minimal polynomial of ζ.
Further, if ζ is a primitive pth root of unity, then ζ i ,1 ≤ i ≤ (p − 1) are primitive
pth roots of unity. Hence Q(ζ ) is the splitting field of g( x ) over Q.
Therefore, Q(ζ ) Q is a Galois extension. Hence G (Q(ζ ) Q) = p − 1.

Now we find the Galois group. Since ζ i is also a root of the irreducible
polynomial g( x ), σ i : ζ a ζ i is an automorphism of Q(ζ ) Q for all
i = 1,K , p − 1. Thus, G (Q(ζ ) Q) = {σ i i = 0, K , p − 1}.
Since σ i σ j (ζ ) = ζ ij = σ jσ i (ζ ) ∀ i, j = 1, K , p − 1, G (Q(ζ ) Q) is abelian.
***

Here are some important remarks about cyclotomic fields.

Remark 4: Note that Q(3 2 ) Q is not a Galois extension. But K (3 2 ) K is a

Galois extension if K = Q(ω), where ω is a primitive cube root of unity. In
general, a radical extension L = K ( n α ) is a Galois extension if K contains
the nth roots of unity, where K ⊆ C.

The extension Q(ζ ) Q in Example 12 is, in fact, cyclic. But the proof is more
involved and we shall skip it. However, we shall show this for some special
values of p, in the following examples and exercises.

Example 13: Find G (Q(ζ) Q) of Example 12 in the cases p = 3, 5, 7.

Solution: Take p = 3 , 5. In both these cases, consider the automorphism

τ : Q ( ζ ) → Q ( ζ ) : τ( ζ ) = ζ 2 .
Then, τ is an element of order p − 1 in the Galois group G = G (Q(ζ) Q) , and
hence G is the cyclic group generated by τ.

When p = 7, the automorphism τ, given above, is not of order 6. But if we

take the automorphism σ(ζ ) = ζ 3 , then the order of σ is 6, so that G is
generated by σ.
180 ***
Try some related exercises now. A Gateway to Galois
Theory

E17) Use Theorem 7 to decide if C R is a Galois extension.

E18) i) Find a generator for the Galois group of Q(ζ ) Q , where ζ is a

primitive 11th root of unity.

ii) Find all the generators for the Galois group G (Q(ζ ) Q), where ζ
is a primitive pth root of unity, for p = 3 , 5 , 7 and 11, respectively.

E19) How many generators are there for a cyclic group of order p − 1, where
p is a prime?

E20) Let L = Q(3 a ), where a is an integer which is not the third power of an
integer. Find Aut (L Q). Is L Q a Galois extension?

E21) Let L = Q( p , q ), where p and q are distinct primes. Show that L Q

is a Galois extension. What is its Galois group?

E22) Let L = Q(ζ), where ζ is a primitive 8th root of unity. Find [L : Q].
Prove that L Q is a Galois extension and the Galois group is isomorphic
to Klein’s four group C 2 × C 2 , where C 2 denotes a cyclic group of order
2.

E23) Let L = Q( 2 , 3 , 7 ). Show that the Galois group of L Q is

isomorphic to C 2 × C 2 × C 2 where C 2 is a cyclic group of order 2.
[Remark: After you study the next unit, you can solve this problem
easily. However, it is possible and instructive for you to do it with the
learning you have from this unit.]

With this we come to the end of our discussion on the preliminaries required
for studying the next unit. Let us take a brief look at what you have studied in
it.

13.6 SUMMARY
In this unit, we have discussed the following points.

1) Algebraic formulae for solutions of quadratic, cubic and quartic

polynomials exist in terms of radicals and their rational combinations.

2) If K = Q(ζ ), where ζ is a primitive pth root of unity, for an odd prime

p, then [K : Q] = p − 1.

3) The definition, and examples, of isomorphisms and automorphisms of

finite extensions.

4) A K-isomorphism of L to C maps a root of an irreducible polynomial to

another of its roots, where L K is a finite extension.
181
Field Theory 5) If L K is a finite extension, there are [L : K ] K-homomorphisms from
L into C.

6) Every finite extension of a field is simple.

7) The definition, and examples, of a normal extension.

8) A finite extension L K is normal iff it is the splitting field of a

polynomial over K.

9) The definition, and examples, of a separable extension.

10) i) If char K = 0, every algebraic extension L K is separable.

ii) If char K = p, an irreducible f ( x ) ∈ K[x ] is separable iff

f ( x ) ≠ g (x p ) for any g( x ) ∈ K[ x ].

11) The definition, and examples, of a Galois extension.

12) Computation of the Galois group of the pth cyclotomic field, where p is
an odd prime, and for some other Galois extensions.

13.7 SOLUTIONS / ANSWERS

E1) Let α1 , α 2 be the roots of ax 2 + bx + c. Then

ax 2 + bx + c = a ( x − α1 ) ( x − α 2 ). Expanding the right hand side and
comparing the coefficients, we get b = −a (α1 + α 2 ) and c = aα1α 2 .
Therefore, the discriminant D = b 2 − 4ac = (α1 + α 2 ) 2 − 4α1α 2 .
On simplifying, we get D = a 2 (α1 − α 2 ) 2 , i.e., D = a (α1 − α 2 ) .

E2) Suppose f ( x ) = ax 2 + bx + c is reducible over K. Then f ( x ) must be a

product of two linear polynomials:
f ( x ) = (dx + e) (d′x + e′), d, e, d′, e′ ∈ K , d ≠ 0 and d′ ≠ 0.
Then α = −ed −1 , α′ = −e′d′−1 are in K, and are the roots of f ( x ).

Conversely, suppose that the roots of f ( x ) are α1 , α 2 ∈ K. Then

f ( x ) = a (x − α1 ) ( x − α 2 ), and hence it is reducible over K.

E3) i) Let [L : Q] = 2. Then L = Q(α), for some α with minimal

polynomial x 2 + bx + c ∈ Q[ x ]. Let D be the discriminant of the
−b± D
minimal polynomial of α. Then α = . Note that D is not
2a
a square in Q since if it were, then α ∈ Q, which is not the case.
Then, verify that α ∈ Q( D ) and D ∈ Q(α ). Hence, L = Q( D ).
Conversely, if L = Q( d ), for some d as given, then
deg ( d ) = 2, and hence L Q is a quadratic extension.
182
 A Gateway to Galois
ii) First observe that Q(n m ) = Q( m ), n ∈ Z. Theory
γ
Let L = Q( D ). Suppose D = , where γ , δ are integers. Then
δ
1
D = 2 ⋅ ( γ ⋅ δ). Therefore, L = Q( γ ⋅ δ ), and γ ⋅ δ is an integer.
δ
Now write γ ⋅ δ = n 2 ⋅ d, where d is a square-free integer.
Then L = Q( d ).

− 1+ i 3
iii) Note that ω = . Now prove the statement.
2

E4) A cubic polynomial is reducible over K if and only if it has a linear

factor, i.e., if and only if it has a root in K, where K ⊆ C. This answers
both (i) and (ii). We have not used any special property of Q here.

For the second question in (i), a quartic polynomial can be reducible even
when it has no root in K. For example, f ( x ) = x 4 − 4 = (x 2 − 2)(x 2 + 2)
over Q, but f ( x ) has no root in Q.

E5) i) We have, K ⊂ K (α) ⊂ L. Therefore [L : K ] = [L : K (α)][K (α) : K ].

Therefore, [K (α) : K ] = 2 or 4, since [K (α) : K ] ≠ 1.
If [K (α) : K ] = 4, K (α) = L, which is not the case.
Hence [K (α) : K ] = 2. Hence [L : K (α)] = 2.

ii) By E3 (i), K′ = K ( d ) for some d ∈ K , where d is not a square in

K. Similarly, there exists β = a + b d ∈ K ′ = K ( d ), which is not
a square in K′, such that L = K ′( a + b d ) = K ′( β ).

iii) Note that if α + β is a root of a polynomial over K, then so are

α − β , − α + β , − α − β . Thus, the minimal polynomial is
f ( x ) = x 4 − 2(β + α) x 2 + (β − α)2 ∈ K[x ].

E6) You can do this by direct computation, and by applying De Moivre’s

theorem as well as the fact that e i θ = 1 if and only if θ is a multiple of
2π.

E7) Note that Q(ζ ) ⊆ Q(α), where ζ and α are primitive nth and (2n ) th
roots of unity. Also, [Q(α) : Q] = φ(2n ) = φ(2)φ(n ), since (2, n ) = 1.
So [Q(α) : Q] = φ(n ) = [Q(ζ) : Q].
Hence Q(α) = Q(ζ).
Hence Q(α) ⊆ K.

For the second part, consider n = 2 for a counterexample. Then Q is the

2nd cyclotomic field, and Q ⊆ Q. Q(i) is the 4th cyclotomic field, and
Q⊇ / Q(i).
E8) You can check that if σ and τ are automorphisms of K, then σ o τ is
defined and is an automorphism of K.
Next, o satisfies associativity.
183
Field Theory Further, I is the identity w.r.t. o, and if σ ∈ Aut (K ), so does σ −1.
Thus, (Aut (K ), o) is a group.
Now, Aut (K F) ≠ « since I is in it.
Also, for σ and τ ∈ Aut (K F), στ and σ −1 are the identity on F, so that
στ and σ −1 are in Aut (K F). Thus, Aut (K F) is a subgroup of Aut (K ).

E9) Let L = K ( d ), d ∉ K. Let θ : L → L′ ⊆ C be an isomorphism. Since

the minimal polynomial of d is x 2 − d, and θ( d ) is also a root of
x 2 − d , θ( d ) = ± d .
So θ(L) ⊆ K ( d ) = L.
Thus, θ ∈ Aut (L).

E10) There are three Q -homomorphisms from K = Q(3 2 ) into C. They are
given by τ1 , τ 2 , τ3 , where
τ1 (3 2 ) = 3 2 , τ 2 (3 2 ) = ω(3 2 ), τ3 (3 2 ) = ω2 (3 2 ).

E11) Let L = Q(3 2 , ω) and K = Q(3 2 + ω). Write down 6 distinct

homomorphisms from L to C, and show that the images of 3 2 + ω are
all distinct under these maps. This proves that the number of
homomorphisms of K into C is at least 6. Therefore, [K : Q] ≥ 6.
However, K ⊂ L and [L : Q ] = 6. Therefore, K = L.

E12) i) This is not normal, since it is not algebraic.

ii) This is the splitting field of x 2 + 7 over Q, and hence is normal.

iii) This is the splitting field of x 2 + 3 over Q( 2 ), and hence is

normal.
iv) One root of x 5 − 5 ∈ Q(51/ 5 ) is in Q(51/ 5 ). The other roots are
51 / 5 ζ t , t = 1, 2, 3, 4, which lie in C \ R, and hence not in Q(51/ 5 ).
Thus, Q(51/ 5 ) Q is not normal.

E13) First, assume L K is separable.

Consider any α ∈ L. Its minimal polynomial over E divides its minimal
polynomial over K. Since L K is separable, m α , K ( x ) has no repeated
roots. Hence m α , E (x ) has no repeated roots. Thus, L E is separable.
You can easily show that E K is separable.

E14) Let L K be separable. Then L = K (α) for some α ∈ L. Let f ( x ) be the

irreducible polynomial of α over K, where deg f ( x ) = n.
Let N L be a normal extension. Since α ∈ N, all the roots of f ( x ) are in
N. So, for each root α1 (= α), α 2 , K, α n of f ( x ), we can define
σ i : L → N : σ i (α) = α i . Each σi defines a distinct K-homomorphism of
L into N. Thus, there are [L : K ] = n distinct K-homomorphisms from
L into N.
184
 Conversely, suppose the condition holds and α ∈ L. Let the irreducible A Gateway to Galois
Theory
polynomial of α be f ( x ) over K and its splitting field over L be N.
Let [L : K ] = n and σ1 , σ 2 , K, σ n be the distinct K-homomorphisms of
L into N. Then σ1 (α), K, σ n (α) are roots of f ( x ). The maximum
number of roots can only be [L : K ] since Aut (L K ) ≤ [L : K ]. Thus, all
the roots of f ( x ) are distinct, and hence f ( x ) is separable.

E15) Use E14 to show this.

E16) i) This is false. For example, Q(21/ 3 ) Q is separable, but not normal.

ii) Since every algebraic extension in characteristic 0 is separable,

every such normal extension is separable. Since every extension
Fpn Fp is separable, any such normal extension will be separable.
Hence, the statement is true in all cases.

E17) Aut (C R) is a cyclic group of order 2, with a generator being the

complex conjugation z a z. Hence, Aut (C R ) = 2 = [C : R]. Thus,
C R is a Galois extension.

E18) i) The Galois group is of order 10 and is isomorphic to F11∗ . Since

211 ≡ 1(mod 11) and no smaller power of 2 is congruent to
1(mod 11), 2 is a generator for this group. Therefore, the
automorphism τ, defined by τ(ζ) = ζ 2 , is a generator of the Galois
group of Q (ζ ) Q .

ii) Observe that there are φ(n ) generators for a cyclic group of order
n. Find this number for the special values given in the problem.
Now, you have to find elements of the respective groups whose
order is equal to the order of the group.
For example, for p = 7, the multiplicative group is of order 6. So
we expect φ(6) = 2 generators for this group.
For p = 11, verify that the elements 2 , 3, 5 , 7 are of order 10, and
hence they are generators. It will be good practice for you to find
the orders of all the elements of this group. (e.g., I is of order 1, 10
is of order 2.) Find the others.

E19) There are φ(p − 1) generators for a cyclic group of order p − 1. For
example, for p = 5, p − 1 = 4. So there are φ(4) = 2 generators. For
p = 31, p − 1 = 30. So there are φ(30) = φ(2)φ(3)φ(5) = 1 × 2 × 4 = 8
generators for a cyclic group of order p − 1 = 30.

E20) Aut (L Q) is trivial, i.e., it consists of the identity only. It is not a Galois
extension since [L : Q] = 3 and the automorphism group is of order 1.

E21) Let L = Q( p , q ), where p ≠ q are primes. You know that it is a

Galois extension. Its Galois group is isomorphic to the Klein four group
C 2 × C 2 , just like the special case p = 2 , q = 3, in Example 9.
185
Field Theory E22) Note that x 8 − 1 = (x 4 − 1)(x 4 + 1).
And x 4 + 1 is irreducible over Q. If ζ is a complex number such that
ζ 8 = 1 and no smaller power of ζ is 1, then x 4 + 1 is the minimal
polynomial of ζ; for example, you can take ζ = e 2 π i / 8 . Therefore,
[L : Q] = 4. It is a Galois extension, since L is a splitting field over Q.
Let τ1 (ζ ) = ζ 3 , τ 2 (ζ ) = ζ 5 , τ3 (ζ ) = ζ 7 .
Then τ12 (ζ ) = ζ 9 = ζ. Similarly, τ22 (ζ) = ζ 25 = ζ and τ23 (ζ) = ζ 49 = ζ. This
shows that there are three elements of order 2 and the fourth element is
the identity element. Hence there is no element of order 4. Therefore, the
Galois group is not cyclic. Since a group of order 4 which is not cyclic
must be Klein’s four group, we get the result.
(Remark: The group of invertible elements of Z 8Z is {1 , 3 , 5 , 7},
which is Klein’s four group. The Galois group is isomorphic to this
group.)

E23) The extension L Q is a Galois extension, since L is a splitting field over

Q. Now, prove it step-by-step, as below

i) Let K be a field and d1 , d 2 be non-squares in K, such that d1d 2 is

not a square in K. Let K′ = K ( d1 , d 2 ). Show that [K′ : K ] = 4.

ii) Let d1 = 2 , d 2 = 3 and K = Q( 7 ). Apply (i) to get [L : Q] = 8.

iii) Let K′′ = Q( 2 , 3 ) Q . Since [L : Q] = 8 and [K′′ : Q] = 4 , 7 is not

a square in K′′. Thus, L is a quadratic extension of K′′. Write
down the isomorphisms (automorphisms) of K′′ = Q( 2 , 3 ) Q .
Extend each of these isomorphisms to L = K′′( 7 ). You should be
able to work out the details and show that the Galois group is
abelian, and except for the identity, all the other elements have
order 2.
iv) From Block 1, you know that an abelian group of order 8, in which
all the elements other than the identity are of order 2, is isomorphic
to C 2 × C 2 × C 2 .

186
 Fundamental Theorem
UNIT 14 FUNDAMENTAL THEOREM OF of Galois Theory
GALOIS THEORY
Structure Page No.
14.1 Introduction 187
Objectives
14.2 The Galois Correspondence 188
14.3 The Galois Group as a Subgroup of the Symmetric Group 190
14.4 Applications of Galois Theory 193
14.5 Summary 200
14.6 Solutions / Answers 200

14.1 INTRODUCTION
In the previous unit, you studied that polynomial equations upto degree 4 have
algebraic formulae that give their roots in terms of the radicals of the
coefficients of the polynomial. However, the work of many mathematicians,
notably Abel and Ruffini (early 19th century), has shown that in general, there
cannot exist such formulae for polynomials of degree n ≥ 5. But it was the
pioneering work of Evariste Galois, who gave a criterion for deciding whether
the roots of a given polynomial can be expressed in terms of radicals of the
coefficients of the polynomial. This, in turn, would also prove the Abel-Ruffini
Fig. 1: Paolo Ruffini
theorem, which had been left incomplete by Paolo Ruffini. (1765-1822) was an
Italian
Galois died very young, but became immortal with his profound theory, known mathematician and a
as Galois Theory today. His Fundamental Theorem finally established a medical doctor.
necessary and sufficient condition for a polynomial to be solvable by radicals.
In this unit, we state the Fundamental Theorem of Galois theory (Galois’
theorem) without proof, and explain how this theorem can be used to prove that
there cannot be a formula for obtaining the roots of certain fifth degree
equations in terms of the radicals of the coefficients of the polynomial. You can see a
documentary on Galois
As in the last unit, in this unit too, unless mentioned otherwise, we continue to at the site
http://youtu.be/J6dsan
assume that all fields considered are subfields of C, for the sake of pnpt0
simplicity.
Also refer to ‘The
Objectives Theory of Equations
and the Birth of Modern
After studying this unit, you should be able to Group Theory’, by B.
Sury, Resonance, 4(10),
• state, and apply, the Fundamental Theorem of Galois Theory; pp47-60, available at the
website
• define, and give examples of, the Galois group of an irreducible ias.ac.in/describe/article
polynomial; /reso/004/10/0047-0060.
• decide when the Galois group of an irreducible polynomial is contained
in A n, and when it is not;
• relate the solvability of a polynomial by radicals to the solvability of its
Galois group;
• explain how Galois theory can be applied to determine those values of
primes p for which a regular p-gon can be constructed by ruler and
compass. 187
Field Theory
14.2 THE GALOIS CORRESPONDENCE
Recall, from Unit 13, that if a finite extension L K is a Galois extension
G (L K ) = [L : K ]. The question we shall look at in this section is whether or
not there is a 1-1 correspondence between the subfields of L containing K and
the subgroups of G (L K ). In fact, this is what the fundamental theorem
(Theorem 1) is about. To see how this correspondence is defined, we first
prove two propositions.

Proposition 1: Let L K be a finite Galois extension. Let F be a field such that

K ⊂ F ⊂ L. Then L F is a Galois extension, and G (L F) is a subgroup of
G (L K ).

Proof: In Unit 13 you have seen that a finite extension is a Galois extension iff
it is normal, i.e., iff it is the splitting field of a polynomial with coefficients in
the base field. Thus, L is the splitting field of a polynomial f ( x ) ∈ K[ x ]. Since
K ⊂ F, f ( x ) ∈ F[ x ] and L is still the splitting field of f ( x ) over F. Hence,
L F is a Galois extension.

Next, every element of the group G (L F) is an F-automorphism of L, and

hence, is a K-automorphism of L. Therefore, G (L F) is a subgroup of
G (L K ).

So, given any subfield F of L containing K, we can associate a subgroup

G (L F) of G (L K ). On the other hand, given any subgroup H of G (L K ),
can we associate with it a subfield of L containing K ? To answer this, we
define a field associated with any subgroup of G (L K ) in the following
proposition.

Proposition 2: Let H be a subgroup of G (L K ). Then

H
L is called the fixed LH = {α ∈ L σ(α ) = α ∀ σ ∈ H} is a subfield of L containing K.
field of H.

Proof: For any σ ∈ H, σ K = I. Hence, K ⊆ LH . Thus, K ⊆ LH ⊆ L.

Now, for α, β ∈ LH and ∀ σ ∈ H, you should check that
σ(α ± β) = α ± β, σ(αβ) = αβ, σ(α −1 ) = α −1. Thus, LH is a field.

Propositions 1 and 2 may have already given you an inkling about the
correspondence between the subgroups of G (L K ) and the subfields of L
containing K. In fact, this is what the Fundamental Theorem of Galois Theory
tells us about.

Theorem 1 (Fundamental Theorem of Galois Theory): Let L K be a Galois

extension. Let Γ = {F F is a subfield of L containing K}, and
Σ = {H H ≤ G (L K )}. Then the following hold:

i) The maps µ : Γ → Σ : µ(F) = G (L F) , and ν : Σ → Γ : ν(H) = LH ,

188 are inverses of each other.
 Fundamental Theorem
ii) µ and ν are inclusion reversing, that is, for F1 , F2 ∈ Γ, F1 ⊆ F2 if and of Galois Theory
only if G (L F2 ) ⊆ G (L F1 ).
iii) For F ∈ Γ, F K is a Galois extension if and only if G (L F) G (L K ) , If you are interested in
− G (L K )
studying the proof of
and in that case G (F K ) ~ . Theorem 1, please refer
G (L F)
to ‘Abstract Algebra’ by
We shall not prove Theorem 1 in this course. However, consider an important Dummit and Foote, or
point about the theorem. any other such book.

Remark 1: In Theorem 1, note that G (L K ) : G (L F) = [F : K ].

Let us now see what Theorem 1 tells us in the case of some extensions you
have already studied in Unit 13.

Example 1: Let L K be a cyclic extension. Show that if m divides

G (L K ) = n , then there exists a unique field F lying between L and K
such that [L : F] = m.
Solution: Since L K is cyclic, G (L K ) is a finite cyclic group. Hence, for
every m dividing n, G (L K ) has a unique normal subgroup H of that order.
Hence, by Theorem 1, there exists a unique field F, such that K ⊆ F ⊆ L,
with G (L F) = H and [L : F] = m.
***
The inclusion reversing bijection between the subgroups of G (L K ) and the
subfields of L containing K can be nicely explained using diagrams for a
lattice of subgroups of G (L K ) and a corresponding lattice of subfields of
L K , where L K is a Galois extension. Let us do it for a few examples.

Example 2: Let L = Q( 2 , 3 ). Show the correspondence given in Theorem 1

for L Q diagrammatically.
Solution: From Example 9, Unit 13, you know that L Q is a Galois extension,
and the Galois group is isomorphic to the Klein-four group C 2 × C 2 . The lattice
of subgroups of G (L Q) and of the subfields of L containing Q are illustrated
in the following diagram, where the τi are as given in Example 9, Unit 13.
Q( 2 , 3) {I}

Q( 3 ) Q( 6 ) Q( 2 ) {I, τ3 } {I, τ τ3 } {I, τ }

2 2

Q {I, τ , τ , τ τ3 }
2 3 2

(a) (b)
Fig. 2: (a) The lattice of subfields of Q( 2 , 3 ), (b) the lattice of the corresponding
subgroups of G ( Q( 2 , 3 ) Q) .
*** 189
Field Theory Example 3: Let L = Q(ω , 21/ 3 ). Show the correspondence given in Theorem 1
for L Q diagrammatically.

Solution: In Example 10, Unit 13, you have seen that L Q is a Galois
extension, and that G (L Q) is S3 . Here we give a diagrammatic representation
of the correspondence given in Theorem 1. You can see the strong similarity
between the lattice for the subfields of L and the lattice for the subgroups of
G (L K ).
1/ 3
Q(ω, 2 ) {I}

1/ 3 1/3 2 1/ 3 2
Q( 2 ) Q( ω 2 ) Q( ω 2 ) {I, g} {I, gf } {I, gf }

2
Q(ω) {I, f , f }

2 2
Q {I, f , f , g, gf , gf }
(a) (b)
1/3
Fig. 3: The lattices for Q( ω, 2 ), where f = γ 3 , g = γ 2 in Example 10, Unit 13.
***

Try an exercise now.

E1) Let L = Q( 2 , 3 , 5 ). Find all the subgroups of the Galois group,

G (L Q) , and the corresponding subfields of L.

In Examples 2 and 3, notice that G (L Q) is isomorphic to a subgroup of S4

and S3 , respectively. This is true in general. Let us discuss that now.

14.3 THE GALOIS GROUP AS A SUBGROUP OF

THE SYMMETRIC GROUP
In order to apply the Fundamental Theorem of Galois theory to the solvability
of a polynomial by radicals, we need to understand the group action of the
Galois group on the roots of the polynomial. For this, why don’t you prove a
few basic results about the generators of the symmetric group, which you
would have studied in your undergraduate studies.

E2) Prove that:

i) Sn is generated by the transpositions (1 2) , (1 3) ,K , (1 n ).
ii) The transposition (1 2) and the cycle (1 2 K n ) generate Sn . (In
fact, (1 2 K n ) and any one (i j) generate Sn iff i − j and n are
190
 relatively prime. Hence, (1 2K n ) and (i i + 1) generate Fundamental Theorem
of Galois Theory
Sn ∀ i = 1,K, n − 1.)

Now, consider S3 . You know that S3 ⊇ A 3 ⊇ {I}, and {I} A 3 S3 .

Further, S3 A 3 and A 3 {I} are abelian groups. This kind of chain of
subgroups, with the properties shown, hold true more generally for the type of
group we now define.

Definition: A finite group G is said to be solvable if there exists a sequence of

subgroups
G = G 0 ⊃ G1 ⊃ G 2 K ⊃ G m = {e},
such that G i G i−1 , and G i −1 G i is abelian.

From the definition, you may have noted that every finite abelian group is
solvable. This is because every subgroup is normal in this case, and every
quotient group will be abelian.

Also, there is a famous theorem, proved by the famous mathematicians W. Feit

and J. G. Thompson in the 1960s, that says that every finite group of odd
order is solvable.

Let us consider the question: Is Sn solvable for n ∈ N ?

Example 4: Show that Sn is solvable for n ≤ 4.

Solution: S1 = {I} is trivially solvable.

S2 is abelian, and hence, solvable.
S3 is solvable, as you have seen above.
For S4 , consider the series of subgroups S4 ⊇ A 4 ⊇ V4 ⊇ {e}, where
V4 = {I, (1 2) (3 4), (1 3) (2 4), (1 4) (2 3)}.
Then I V4 A 4 S4 . Also S4 A 4 , A 4 V4 are of prime order, and hence
abelian. Further, V4 I − ~ C 2 × C 2 , and hence, is abelian.
Thus, S4 is solvable.
***

Now, doing the following exercises will help you answer the question raised
before Example 4.

E3) Let G be a group and H be a subgroup of G. Prove the following:

i) If G is solvable, then H is solvable.
ii) Let H be normal in G. Then G is solvable if and only if H and
G H are solvable.

E4) Prove that every p-group is solvable.

E5) i) Check whether or not Sn is solvable for n ≥ 5.

191
Field Theory ii) Prove that A n is solvable for all n ≤ 4, and not solvable for all
n ≥ 5.

E6) A finite group that is not simple must be solvable. True or false? Why, or
why not?

Now, just go back to Unit 13. Here you studied that if L is the splitting field of
an irreducible f ( x ) ∈ K[x ], then L K is a Galois extension and the elements of
G (L K ) permute the roots of f ( x ). This leads us to think about whether every
Galois group is a permutation group. To answer this, let us define a related
term first.

Definition: Let K be a field of any characteristic. Let f ( x ) be a separable

polynomial of degree n over K, with splitting field L, so that L K is a
Galois extension. G (L K ) is called the Galois group of f(x).

In Example 10, Unit 13, you have seen that the Galois group of x 3 − 2 ∈ Q[x ]
is isomorphic to S3 . In fact, more generally, in Unit 13 you have seen that if
σ ∈ G (L K ), the Galois group of f ( x ), then σ maps a root of pi ( x ) to another
root of pi (x ), where p1 ( x ), p 2 ( x ),K, p r ( x ) are the irreducible factors of f ( x ).
Since the roots of the pi ( x ), i = 1,K , r, generate L over K, σ is uniquely
determined by these actions. Hence, if α1 , α 2 ,K, α n be the roots of f ( x ), then
σ defines a unique permutation of {α1 , α 2 , K, α n }. Thus, there is a
homomorphism µ : G (L K ) → Sn : µ(σ) = Pσ , where
 1 2 K n 
Pσ =  .
 σ(1) σ(2) K σ(n ) 
Here is an immediate property of the homomorphism µ.

Proposition 3: Let f ( x ) be a separable polynomial of degree n in K[x ],

where char K need not be zero. Let L be its splitting field over K. Then
µ : G (L K ) → Sn : µ(σ) = Pσ is an injective map.

Proof: Let Pσ = I. Then σ(α i ) = α i ∀ i = 1,K, n , where the roots of f ( x ) are

α1 ,K, α n . Hence σ = I.
Thus, µ is 1-1.

By Proposition 3, we can treat G (L K ) as a subgroup of Sn . Let us consider a

particular case.

Example 5: Let L be the splitting field of ( x 2 − 2)(x 2 − 3) ∈ Q[x ]. Find the

subgroup of S4 to which G (L Q) is isomorphic.

Solution: L = Q( 2 , 3 ). In Example 9, Unit 13, you have seen that the roots
are α1 = 2 , α 2 = − 2 , α 3 = 3, α 4 = − 3. There you have also seen that
G (L Q) = {τ1 , τ 2 , τ3 , τ 4 }, where τ1 = I, τ 2 = (3 4), τ3 = (1 2), τ = (1 2) (3 4).
Thus, G (L Q) − ~ {I, (1 2), (3 4), (1 2) (3 4)} ≤ S4 .

192 ***
Why don’t you prove some other properties of the homomorphism µ now? Fundamental Theorem
of Galois Theory

E7) i) From Unit 13, recall the definition of the discriminant D of a

polynomial. Let K ⊆ C, and let G be the Galois group of f ( x )
over K. Prove that σ(D) = D ∀ σ ∈ G , and hence D ∈ K.

ii) Prove that D ∈ K if and only if G ⊆ A n , where G = n.

E8) For each of the following polynomials over Q, determine whether its
Galois group is contained in A 3 or not.

i) f ( x ) = x 3 − 3x + 1.
ii) f ( x ) = x 3 + 3x + 1.
iii) f ( x ) = x 3 + x + 5.

And now we shall discuss the two applications that Galois had in mind while
developing the theory you have been studying.

14.4 APPLICATIONS OF GALOIS THEORY

In this section we shall consider two applications of the Fundamental Theorem
of Galois theory, one on classical algebra and one on classical geometry.

14.4.1 Application to Solvability by Radicals

In Unit 13, we had discussed the solutions of quadratic, cubic and quartic
polynomials over a field. You saw that there are algebraic solutions to such
polynomials involving radicals. Over there, we had said that no such solution
exists in general for quintic and higher degree polynomials. This finding is due
to Galois. In fact, the Galois correspondence given in Theorem 1 leads us to the
following condition that must be satisfied by a polynomial solvable by radicals.

Theorem 2: Let f ( x ) be a non-constant polynomial over K ⊆ C, and let G be

its Galois group. Then the polynomial is solvable by radicals if and only if G
is a solvable group.

We shall not prove this theorem here, but shall consider some examples of its
use. The first example is the following proposition.

Proposition 4: A polynomial of degree less than 5 over a field K is solvable

by radicals.

Proof: The Galois group of an irreducible polynomial of degree ≤ 4 is a

subgroup of S4 , and you have seen that S4 is solvable. So the result follows
immediately from E3 and Theorem 2.

We shall now give an example of a polynomial of degree 5 which is not

solvable by radicals.
193
Field Theory Example 6: Show that f ( x ) = x 5 − 16 x + 2 ∈ Q[ x ] is not solvable by radicals.

Solution: Firstly, you can check that f ( x ) is irreducible over Q, using

Eisenstein’s irreducibility criterion.
Next, let α1 ,K , α 5 be the roots of f ( x ). Then Q (α1 ) Q is of degree 5. Hence
5 o(G ), where G is the Galois group of f ( x ). Hence G contains an element
of order 5. Since G O S5 , and the only possible elements of order 5 in S5 are
the 5-cycles, G contains a 5-cycle, say (α1 α 2 α 3 α 4 α 5 ).

Now, f (−3) = −193, f (−2) = 2, f (1) = −13, f (2) = 2. Thus, f ( x ) has a root in
each of the intervals ] − 3, − 2[, ] − 2, 1[, ]1, 2[. Can f ( x ) have any more real
roots? If it had more real roots, then by Rolle’s theorem (of Calculus)
f ′( x ) = 5x 4 − 16 would have three real roots, which it does not. Thus, f ( x ) has
3 real roots and one pair of complex conjugates as roots. Let the complex roots
be α1 , α 2 = α1 (by re-numbering them if necessary), and the real roots be
α3 , α 4 , α5.
So we have shown that G contains the five cycle (α1 α 2 α 3 α 4 α 5 ) and the
transposition (α1 α 2 ) which corresponds to the automorphism of complex
conjugation.
Since (1 2 3 4 5) and (1 2) generate S5 , using the 1-1 map µ : G → S5 , we
conclude that G −~ S5 .
Since you have seen that S5 is not solvable, G is not solvable.
Hence, f ( x ) is not solvable by radicals.
***

Try some exercises now.

E9) i) Let g( x ) = 3x 5 − 15x + 5. Follow the same steps as in Example 6 to

prove that g( x ) is not solvable by radicals.

ii) Let h ( x ) = x 5 − 6x + 3. Imitate the arguments of Example 6 to

prove that the Galois group of h ( x ) is S5 . Is h ( x ) solvable by
radicals?

E10) Show that the Galois group of the polynomial x 5 − 1∈ Q[ x ] is cyclic, and
hence solvable.

Let us now consider a geometric application of the Galois correspondence.

14.4.2 Application to Constructibility of Regular p-gons

We shall now discuss an application of Galois theory to a problem in classical
By ‘construction’ we Greek geometry, namely, the construction of regular n-gons. As you know
mean a construction
using a straightedge and from your school days, we can construct an equilateral triangle using only a
a compass only. straightedge and compass. The question is: Can a square be constructed in this
way? What about a regular pentagon? In general, for which values of n can a
regular n-gon be constructed? The answer to the general question was not
194
known for a long time, though the construction of a regular pentagon (n = 5) Fundamental Theorem
of Galois Theory
was known. The mathematician Gauss, whom you have met in this course
several times, was the first to give an explicit construction of a regular 17-gon.
A ‘straightedge’ means
However, to find all the values of n for which a regular n-gon can be a ruler without any
constructed, we need to apply Galois theory. To do this, first we have to markings.
translate this problem of Euclidean geometry to a problem in coordinate
geometry.

Recall that constructing an equilateral triangle amounts to constructing an

angle of measure π 3 (or, equivalently, the complementary angle 2π 3).
Similarly, constructing a regular n-gon amounts to marking n points on the
unit circle, which divide its circumference into equal parts. The angles between We shall use ‘ruler’ and
‘straightedge’
the radii to these n points is 2π n . Hence, constructing the n-vertices of a interchangeably.
reqular n-gon amounts to constructing an angle of measure 2π n. So the
question is, for which values of n can an angle of measure 2π n be
constructed using a straightedge and compass only. This question, you may be
surprised to know, is answered completely by Galois theory. Let us see how.

First, we translate the problem of constructibility by ruler and compass into a

problem in coordinate geometry. Let us start with the 2-dimensional Euclidean
plane. You know that the real numbers are in one-to-one correspondence with
the points on a line. Let us start with two points A , B and draw a line through
them. This is a ruler construction. We take this as the real line through the
points A, B. So, we can think of A, B as real numbers, A denoting 0 and B
representing 1. Now, we think of these two numbers as constructed. The
distance between A and B is the unit distance, and is also constructed.

Now, try to answer this question: which real numbers can you construct using
only a straightedge and compass, starting with these two given numbers 0 and
1? Using a ruler means drawing a line through two constructed numbers.
Using a compass means drawing an arc of a circle with a constructed See the website
number as centre, and with radius the length of a constructed segment. youtube.com/watch?v=1
Then, as you know from Euclidean geometry, the point of intersection of two 2GbLOzhMng
lines drawn (constructed), or the intersection of a line and a circle drawn to see how a regular
(constructed), or the intersection of two circles drawn (constructed) are all pentagon is constructed
constructible points. We can start with {0, 1}. But we can repeat the with a ruler and
compass.
construction process as many times with the already constructed numbers. For
example, when we start with 0 and 1, the length 1, which is the measure of the
segment AB, is a constructed length. So you can draw a circle with 0 as centre
and radius 1. This will intersect the line through 0 and 1 in two points: 1 and
the other point, which we can identify as − 1. But you can also draw a circle of
radius 1 with centre as 1. This will intersect the real line in two points. One of
them is the already constructed number 0 and the other we can identify as 2.
Now, by the same construction, with centres at − 1, 2, respectively, and
repeating inductively, we can mark or construct all the integers on the real line.
We can also draw a perpendicular to the real line through 0, using the
Euclidean construction.

We are now in a position to introduce coordinates: Treat the real line as the x-
axis, and the line perpendicular to it through 0 as the y-axis. So the point A
will now be called the origin (0 , 0), as in the usual Cartesian system. The
coordinate plane is the same as the Cartesian product R × R. Every point in the 195
Field Theory plane is given by an ordered pair ( x , y), x , y ∈ R. Both the x-axis and the y-axis
are in one-one correspondence with the real numbers. The points on the x-axis
are of the form ( x , 0), x ∈ R and those on the y-axis are of the form
(0 , y), y ∈ R.

With the procedure described above we can construct all the points
( x , 0), x ∈ Z and (0 , y), y ∈ Z. You can also construct 2 . [This is done by
joining (1, 0) and (0, 1) and then taking the intersection of the circle with
radius being the length between (1, 0) and (0, 1) and with (0, 0) as centre,
with the x-axis. The two points you get will represent 2 and − 2 .]

Now, coordinate geometry allows us to use algebra to decide whether a point in

the plane (a complex number) can be constructed or not. Constructing a point
in the plane by ruler and compass amounts to solving simultaneous equations
of the following type:
i) two linear equations (corresponding to the construction of finding the
intersection of two lines), or
ii) a linear equation and a quadratic equation (corresponding to the
intersection of a line and a circle), or
iii) a pair of quadratic equations (corresponding to the intersection of two
circles).

In fact, this construction can be translated into an equivalent statement in field

theory. For this we identify the plane R 2 with the field of complex numbers.
Thus, a complex number z = x + iy is constructible iff the point (x, y) in the
plane is constructible, and this is true if and only if the coordinates x , y are
constructible. In terms of polar coordinates, z = re iθ is constructible if and
only if r and cos θ, sin θ are constructible.

Consider the following comment in this context.

Remark 2: All rational numbers are constructible. It can be proved that the set
of all constructible numbers is a subfield of C containing Q[i]. This means
that the sum, difference, product and quotient of constructible complex
numbers are constructible.

Now consider the following theorem, which we give without proof.

Theorem 3: A complex number α is constructible by ruler and compass if and

only if there exists a tower of fields,
Q = K 0 ⊂ K1 ⊂ K ⊂ K m
such that both the following conditions are satisfied:
i) α ∈ Km ,
ii) K i K i−1 is a quadratic extension ∀ i = 1,K, m.

An immediate, useful, corollary of this theorem is the following.

Corollary 1: If a complex number α is constructible, then α is algebraic over

196 Q of degree 2n , for some n ∈ N ∪ {0}.
Proof: By Theorem 3, α ∈ K m for some m, and [K m : Q] is a power of 2. As Fundamental Theorem
of Galois Theory
Q[α] ⊂ K m , [Q[α] : Q] divides [K m : Q]. Thus, deg α = [Q[α] : Q] = 2 n , for
some n ∈ N ∪ {0}.

From Corollary 1 you can see that a complex number which is transcendental,
or which is algebraic with degree not a power of 2, is not constructible. For
example, e, π, 3 2 are not constructible.

Try an exercise now.

E11) Show that a complex number α is constructible if and only if α is

constructible.

Theorem 3, and its corollary, can be applied to answer the following classical
problems formulated by the ancient Greeks:
1. (Squaring the circle): Can we construct a square whose area is equal to
the area of the unit circle?
2. (Doubling a cube): Can we construct a cube whose volume is double the
volume of a given cube?
3. (Trisecting an angle): Can we trisect a given angle using only ruler and
compass?

These questions have been answered in the following corollaries to Theorem 3.

Corollary 2: A square whose area is equal to the area of a given circle cannot
be constructed (by ruler and compass).

Proof: The area of a circle of radius r is πr 2 . For constructing a square whose

area is πr 2 , we have to construct a square whose side is r π. Now, r is the
radius of the given circle, and hence r is constructible. Thus, r π being
constructible would imply π is constructible, and hence π is constructible,
by E11. Thus, we reach a contradiction. Hence, the corollary follows.

Corollary 3: A cube whose volume is twice the volume of a given cube cannot
be constructed.

Proof: Consider a cube of side 1 unit. Its volume is 1 cubic unit. The new cube
will have volume 2 cubic units. So, constructing the new cube amounts to
constructing 3 2 , the side of the new cube. But 3 2 cannot be constructed as
its degree is 3, which is not a power of 2.
Hence such a cube cannot be constructed.

In your school days you would have often bisected a given angle using a ruler
and compass. But, were you ever asked to trisect an angle? Maybe you could
do it for 90o , or 180o. But it cannot be done for all angles, as we prove in the
following corollary.

Corollary 4: An angle of 60o cannot be trisected by a ruler and compass. 197

Field Theory Proof: Suppose a given angle θ can be trisected. This means, φ = θ / 3 is
constructible. This means cos φ is constructible. Using the formula
1
cos (3φ) = 4 cos3 φ − 3 cos φ, we see that cos φ is a root of 4x 3 − 3x − , for
2
o
θ = 60 . (Observe that cos θ = 1 / 2 is constructible.)
Now, this polynomial is irreducible over Q (see E12). It follows that the
degree of cos φ is 3. Therefore, it is not constructible. Hence an angle of 60o
cannot be trisected.

Try some exercises now.

1
E12) Prove that 4x 3 − 3x − is irreducible over Q.
2

E13) Prove that 90o can be trisected by a ruler and compass.

E14) Prove that any given angle can be bisected using a ruler and compass.

We shall now apply Theorem 3 to answer the question we had raised about the
constructibility of a regular p-gon.

Theorem 4 (Gauss): Let p be an odd prime. A regular p-gon is constructible

by ruler and compass if and only if p − 1 is a power of 2.

Proof: As mentioned earlier, a regular p-gon is constructible if and only if the

angle 2π p is constructible. This is possible if and only if the complex number
ζ = e 2 π i / p is constructible, that is, if and only if there is a tower of successive
quadratic extensions
Q = K 0 ⊂ K1 ⊂ K ⊂ K m , …(1)
with ζ ∈ K m , by Theorem 3.

First, let us assume that p − 1 is a power of 2. You know, from Sec.13.5, that
the Galois group G of the minimal polynomial of e 2 π i / p is cyclic of order
p − 1 = 2 r , for some r. So there exists a sequence of subgroups of
See the site G , G = G 0 ⊃ G1 ⊃ K ⊃ G m = {0} such that [G i : G i+1 ] = 2. Using the
youtube.com/watch?v=8 Fundamental Theorem of Galois Theory, we get a tower of successive
7uo2TPrsI8
for the construction of a
quadratic extensions Q = K 0 ⊂ K1 ⊂ K ⊂ K m , satisfying the conditions of
regular 17-gon. Theorem 3. Hence a regular p-gon is constructible by a ruler and compass.

Conversely, assume that a regular p-gon is constructible by a ruler and

compass. Then there is a tower of successive quadratic extension fields, as
given in (1).
So [K m : Q] = 2 m for some m. Also ζ ∈ K m implies Q(ζ ) ⊂ K m . Therefore,
[Q(ζ ) : Q] is a power of 2. But [Q(ζ ) : Q] = p − 1. Thus, p − 1 is a power of 2.

Using Theorem 4, you can see that a regular pentagon (5-gon) is constructible,
but a regular 7-gon is not.
198
In Theorem 4, the primes covered are of the form 2n + 1. Such a prime number Fundamental Theorem
of Galois Theory
is called a Fermat prime. Note that if 2n + 1 is a prime, then n must be a
power of 2 (see E15). It is not known whether there exist only finitely many, or
infinitely many, Fermat primes.

Why don’t you try some exercises now?

E15) If 2n + 1 is a prime, then show that n must be a power of 2. (The

converse is not true.)

E16) Prove, or disprove, the following statements:

i) A regular 11-gon is constructible.
ii) A regular 257-gon is constructible.

Now let us look at an example of constructibilty of a regular n-gon for some

composite values of n.

Example 7: Check whether or not a regular 9-gon is constructible.

Solution: A 9-gon is constructible if and only if the angle 2π 9 is
constructible. This is possible if and only if e 2 πi / 9 is constructible, which is
possible if and only if its square root e π i / 9 is constructible. But that is possible
if and only if cos(π / 9) is constructible, which is the same as saying an angle
of 20o is constructible. But this is not possible, in view of Corollary 4.
Therefore, a regular 9-gon is not constructible.
***

There is a more general version of Gauss’ theorem, proved by Pierre Wantzel

in 1837, which we now state.

Theorem 5 (Gauss-Wantzel theorem): A regular n-gon is constructible by You are familiar with
ruler and compass if and only if φ(n ) is a power of 2, where φ is the Euler-phi φ, from Unit 10.
function.

We shall not prove this theorem here. However, note that φ(n) is a power of 2
if and only if n = 2r p1p 2 Kp t , where the p i are distinct odd primes such
that p i − 1 is a power of 2.

Using Theorem 5, you can immediately see that a regular 9-gon is not
constructible. What about a 20-gon? By Theorem 5, it is constructible.
Here is a word of warning, in this context.

Remark 3: In some books, or on the internet, you may come across methods to
construct various regular n-gons, where n is not of the form given in Theorem
5. But these are approximate constructions, not exact ones. Or they have been
done by using instruments other than a straightedge and compass. The
constructibility discussed here is about exact constructions, not approximate
ones, and not using any other instrument.

We end our discussion on Galois theory here. Let us summarise what you have
studied in this unit. 199
Field Theory
14.5 SUMMARY
In this unit, we have discussed the following points.

1) The statement of the Fundamental Theorem of Galois Theory: Let

L K be a Galois extension. Let Γ be the set of all subfields F of L
containing K, and Σ be the set of all subgroups of G (L K ). Then the
following hold:
i) The maps µ : Γ → Σ : µ(F) = G (L F) , and ν : Σ → Γ : ν(H ) = LH ,
are inverses of each other.
ii) µ and ν are inclusion reversing, that is, for F1, F2 ∈ Γ, F1 ⊆ F2 if
and only if G (L F2 ) ⊆ G (L F1 ).

iii) For F ∈ Γ, F K is a Galois extension if and only if G (L F) is a

normal subgroup of G (L K ) , and in that case,
G (L K )
G (F K ) ~
− G ( L F)
.

2) Illustrations of the Fundamental Theorem of Galois Theory, using

diagrams for the lattices of subgroups and lattices of the corresponding
subfields, in some cases.

3) The definition, and examples, of the Galois group of a polynomial.

4) The definition, and examples, of a solvable group.

5) A polynomial is solvable by radicals if and only if its Galois group is a

solvable group.

6) A polynomial of degree ≤ 4 is solvable by radicals.

7) Some examples of polynomials of degree 5 which are not solvable by

radicals.

8) The application of Galois theory to constructibility of a regular p-gon,

where p is an odd prime number. In particular, the statement of the
Gauss-Wantzel theorem: A regular n-gon is constructible by ruler and
compass if and only if φ(n ) is a power of 2, where φ is the Euler-phi
function.

9) The following constructions are impossible:

i) squaring a circle,
ii) doubling a cube,
iii) trisecting an angle of 60o.

14.6 SOLUTIONS / ANSWERS

E1) From Unit 13, you know that L Q is a Galois extension, with
G (L Q) = {I, σ1 , σ 2 , σ3 , σ1σ 2 , σ1σ3 , σ 2σ 3 , σ1σ 2σ3 }, isomorphic to
200
 Fundamental Theorem
C1 × C 2 × C3 , where C1 = G (Q( 2 ) Q), C 2 = G (Q( 3 ) Q),
of Galois Theory
C3 = G (Q( 5 ) Q) . The Ci are cyclic groups of order 2, generated by
σi , where σ1 ( 2 ) = − 2 , σ 2 ( 3 ) = − 3, σ3 ( 5 ) = − 5.
G (L Q) is abelian. It can have subgroups of order 1, 2, 4 and 8 as it is of
order 8. We write down the subgroups and the corresponding subfields.
The unique subgroup of order 1 is {1}, which corresponds to the fixed
field L.
Similarly, the fixed field of the unique subgroup of order 8, i.e., the
whole group, is the field Q.
There are seven subgroups of index 4 (and hence order 2):
H1 = < σ1 >, H 2 = < σ2 >, H 3 = < σ3 >, H 4 = < σ1σ 2 >, H 5 = < σ1σ3 >,
H 6 = < σ 2 σ3 >, H 7 = < σ1σ 2σ 3 > .
The corresponding fixed fields are of degree 4 over Q, and are:
F1 = Q( 3, 5 ), F2 = Q( 2 , 5 ), F3 = Q( 2 , 3 ), F4 = Q( 5 , 6 ),
F5 = Q( 3 , 10 ), F6 = Q( 2 , 15 ), F7 = Q( 15 , 10 ), respectively.

Finally, we write down subgroups of index 2, i.e., of order 4. There are 7

of them too, namely, H8 = < σ1 , σ 2 >, H 9 = < σ1 , σ3 >, H10 = < σ 2 , σ 3 >,
H11 = < σ1 , σ 2σ 3 >, H12 = < σ 2 , σ1σ3 >, H13 = < σ3 , σ1σ 2 >,
H14 = < σ1σ 2 , σ1σ3 > .
The corresponding fixed fields are of degree 2 over Q, and are:
F8 = Q( 5 ), F9 = Q( 3 ), F10 = Q( 2 ), F11 = Q( 15 ), F12 = Q( 10 ),
F13 = Q( 6 ), F14 = Q( 30 ).

E2) i) You know that every permutation can be written as a product of

transpositions. Also, the transposition (i j) = (1 i) (1 j) (1 i). Hence
Sn is generated by (1 2), (1 3),K, (1 n ).

ii) (1 n ) = (1 2K n ) (1 2) (1 2K n )−1.
Similarly, by conjugating (1 n ) by (1 2K n ), and so on, we can
obtain (1 i) ∀ i = 3,K, n.
Hence, (1 2K n ) and (1 2) generate Sn .

E3) i) If N G , then N ∩ H H.
Thus, let G = G 0 ⊇ G1 ⊇ K ⊇ G n = {e} be a solvable series for G.
Now, as G i−1 G i is abelian ∀ i = 1,K , n,
xyx −1y −1 ∈ G i ∩ H ∀ x, y ∈ G i−1 ∩ H. Hence (G i −1 ∩ H )
(G i ∩ H )
is also abelian. Thus,
H = (G 0 ∩ H) ⊇ (G1 ∩ H) ⊇ K ⊇ (G n ∩ H ) = {e} is a solvable
series for H.
ii) If G is solvable, you can show that H and G H are solvable.
For the converse, let H = H 0 ⊇ H1 ⊇ K ⊇ H n = {e} be a solvable
series for H and G = G 0 ⊇ G1 ⊇ K ⊇ G n = H be a solvable
H
201
Field Theory series for G H, where G i = G i H with H G i ≤ G.
Now consider the series
G = G 0 ⊇ G1 ⊇ K ⊇ G n ⊇ H1 ⊇ H 2 ⊇ K ⊇ H n = {e} …(2)
Here each H i H i −1 and H i−1 H i is abelian.
Also, G i G i −1 ⇒ G i G i −1. Further,
− ( G i −1 H )
G i −1 G i ~ = G i −1 G i , which is abelian.
(G i H )
Thus, (2) is a solvable series for G.

E4) If G = p, or G = p 2 , then G is abelian, and hence solvable.

Assume that any group of order p r is solvable for r ≤ m.
Now let G be a group of order pm +1. If G is abelian, it is solvable.
Suppose G is non-abelian.
Then Z(G ) ≤ G , Z(G ) ≠ {e} and Z(G ) ≠ G.
Also, Z(G ) is abelian, and hence solvable.
Consider G . This is a p-group of order p r for some r ≤ m. Hence
Z(G )
G is solvable.
Z(G )
Thus, by E3, G is solvable.
Hence, by the principle of induction, any p-group is solvable.

E5) i) Regarding S5 , from Unit 2 you know that A 5 is simple and non-
abelian. Hence it is not solvable. You also know that if S5 were
solvable, then, by E3(i), A 5 would be solvable, which is not true.
Hence, S5 is not solvable.
Since Sn O Sn +1 , Sn can be considered a subgroup of Sm∀ m ≥ n.
Further, since a subgroup of a solvable group is solvable, and S5 is
not solvable, Sn cannot be solvable for m ≥ 5.

ii) Since A n ≤ Sn ∀ n , and Sn is solvable for n ≤ 4, A n is solvable for

n ≤ 4.
Since A 5 is simple and non-abelian, it is not solvable.
Also A n ≤ A m∀ n ≤ m.
Hence A m is not solvable for m ≥ 5.

E6) You have seen that S5 is finite and not simple. Also S5 is not solvable.
Hence the statement is false.

E7) i) D = ∏ (α i − α j ) 2 , where αi are the roots of f ( x ).

i< j

For any σ ∈ G , σ permutes the αi . Hence, σ(D) = D. Since this is

true for any σ ∈ G , D is in the fixed field of G , which is K.

ii) According to the Fundamental theorem of Galois theory, D ∈ K

if and only if σ( D ) = D for all σ ∈ G. Now, a transposition
(i, j) changes αi − α j to α j − αi = −(αi − α j ) and leaves all other

202 terms in D unchanged. Thus, if σ is a transposition,

 Fundamental Theorem
σ( D ) = − D . Therefore, σ( D ) = D if and only if σ is an
of Galois Theory
even permutation. This proves that D is fixed by every element
of G if and only if G ⊂ A n .

E8) We apply E7 (ii) to check this. Also, from Unit 13, Section 2, you know
that if f ( x ) = x 3 + px 2 + qx + r, then D = − (27B 2 + 4A 3 ), where
p2 2p 3 pq
A=q− and B = − + r.
3 27 3
i) Here A = −3, B = 1. So D = 81. Hence D ∈ Q. Hence G ⊆ A 3.

ii) Here A = 3, B = 1. So D = −135. Thus, D ∉ Q. Hence G ⊆/ A 3.

~ S3 .
Hence G −
iii) Here A = 1, B = 5. So D = −679. Thus, D ∉ Q. Hence G ⊆/ A 3.
~ S3 .
Hence G −

E9) i) We know that g( x ) is irreducible over Q, using Eisenstein’s

criterion. Now, g( x ) has five distinct roots. As in Example 6,
G~ − S5 . Since S5 is not solvable, g( x ) is not solvable by radicals.
ii) Use the same process to show that h ( x ) is not solvable by radicals.

E10) You know that the splitting field of x 5 − 1 is Q(ζ ), where ζ is a

primitive 5th root of unity. From Sec.13.5, you know that the Galois
group of x 5 − 1 is of order 4, and is cyclic. Thus, it is solvable.

E11) Suppose α is constructible. Then α ∈ K m for some m, where the fields

K i are as described in Theorem 3. Now take K m +1 = K m ( α ). With m
replaced by m + 1, the conditions of Theorem 3 are satisfied for α.
Hence α is constructible.
Conversely, if α is constructible, then α = ( α ) 2 is constructible,
since the product of constructible numbers is constructible (by Remark
2).

E12) To prove a cubic polynomial is reducible over any field K, it is

necessary and sufficient to prove that it has a linear factor over K. Here
K = Q. On simplification, the given polynomial becomes 8x 3 − 6 x − 1.
This is a primitive polynomial over Z, and if it has a rational root it must
have an integer root. But it does not have any integer root, because if it
has one, say α, then 2α will be a root of x 3 − 3x − 1. But this
polynomial is irreducible over Z, as the only possible integers satisfying
it are factors of the constant term, which is 1, i.e., ± 1. Neither of them is
a root. Thus, the given polynomial is irreducible over Z, and hence, over
Q.

E13) cos (θ / 3) is 3 2 when θ is 90o , and 3 2 is constructible as 3 and

1
are constructible.
2 203
Field Theory E14) This follows since α is constructible if and only if α is constructible.
Here α = eiθ .

E15) Suppose n is not a power of 2. Let n = 2 r.d where d > 1 is odd, and
r r
r ≥ 1. Then 2n + 1 = (22 )d + 1. Let x = 2 2 . Then
2n + 1 = x d + 1 = ( x + 1)(x d −1 − x d − 2 + x d − 3 − L + 1).
r
Hence x + 1 = 2 2 + 1 is a proper factor of 2n + 1, a contradiction.
Thus, n is a power of 2.

E16) i) p = 11 is not a Fermat prime. Therefore, a regular 11-gon is not

constructible.
3
ii) p = 257 = 2 2 + 1 is a Fermat prime. By Theorem 4, a regular p-gon
is constructible in this case.

204