Theory of Field Extensions

Master of Science (Mathematics) (DDE)
Semester – II
Paper Code – 20MAT22C1
THEORY OF FIELD EXTENSIONS
DIRECTORATE OF DISTANCE EDUCATION

MAHARSHI DAYANAND UNIVERSITY, ROHTAK
(A State University established under Haryana Act No. XXV of 1975)
NAAC 'A+’ Grade Accredited University
Material Production
Content Writer: Dr Jagbir Singh
Copyright © 2020, Maharshi Dayanand University, ROHTAK

All Rights Reserved. No part of this publication may be reproduced or stored in a retrieval system or transmitted
in any form or by any means; electronic, mechanical, photocopying, recording or otherwise, without the written
permission of the copyright holder.
Maharshi Dayanand University
ROHTAK – 124 001
ISBN :
Price : Rs. 325/-
Publisher: Maharshi Dayanand University Press
Publication Year : 2021
M.Sc. (Mathematics) (DDE)
Paper Code : 20MAT22C1
Theory of Field Extensions
M. Marks = 100
Term End Examination = 80
Time = 3 Hours Assignment = 20
Course Outcomes
Students would be able to:
CO1 Use diverse properties of field extensions in various areas.
CO2 Establish the connection between the concept of field extensions and Galois Theory.
CO3 Describe the concept of automorphism, monomorphism and their linear independence in field
theory.
CO4 Compute the Galois group for several classical situations.
CO5 Solve polynomial equations by radicals along with the understanding of ruler and compass
constructions.
Section - I
Extension of fields: Elementary properties, Simple Extensions, Algebraic and transcendental Extensions.
Factorization of polynomials, Splitting fields, Algebraically closed fields, Separable extensions, Perfect
fields.
Section - II
Galios theory: Automorphism of fields, Monomorphisms and their linear independence, Fixed fields,
Normal extensions, Normal closure of an extension, The fundamental theorem of Galois theory, Norms
and traces.
Section - III
Normal basis, Galios fields, Cyclotomic extensions, Cyclotomic polynomials, Cyclotomic extensions of
rational number field, Cyclic extension, Wedderburn theorem.
Section - IV
Ruler and compasses construction, Solutions by radicals, Extension by radicals, Generic polynomial,
Algebraically independent sets, Insolvability of the general polynomial of degree n ≥ 5 by radicals.
Note :The question paper of each course will consist of five Sections. Each of the sections I to IV will
contain two questions and the students shall be asked to attempt one question from each. Section-V
shall be compulsory and will contain eight short answer type questions without any internal choice
covering the entire syllabus.
Books Recommended:
1. Luther, I.S., Passi, I.B.S., Algebra, Vol. IV-Field Theory, Narosa Publishing House, 2012.
2. Stewart, I., Galios Theory, Chapman and Hall/CRC, 2004.
3. Sahai, V., Bist, V., Algebra, Narosa Publishing House, 1999.
4. Bhattacharya, P.B., Jain, S.K., Nagpaul, S.R., Basic Abstract Algebra (2nd Edition), Cambridge
University Press, Indian Edition, 1997.
5. Lang, S., Algebra, 3rd edition, Addison-Wesley, 1993.
6. Adamson, I. T., Introduction to Field Theory, Cambridge University Press, 1982.
7. Herstein, I.N., Topics in Algebra, Wiley Eastern Ltd., New Delhi, 1975.
Contents
CHAPTER SECTION TITLE OF CHAPTER PAGE NO.
1 1 Extension of Fields 1-27
2 2 Galois Theory 28-43
3 3 Galois Fields 44-60
4 4 Ruler and Compass 61-66

Construction
1
Extension of a Field
Structure
1.1. Introduction.
1.2. Field.
1.3. Extension of a Field.
1.4. Minimal Polynomial.
1.5. Factor Theorem.
1.6. Splitting Field.
1.7. Separable Polynomial.
1.8. Check Your Progress.
1.9. Summary.
1.1. Introduction. In this chapter field theory is discussed in detail. The concept of minimal polynomial,
degree of an extension and their relation is given. Further the results related to the order of a finite field
and its multiplicative group are discussed.
1.1.1. Objective. The objective of these contents is to provide some important results to the reader like:
(i) Algebraic extension and transcendental extension.
(ii) Minimal polynomials and degree of an extension.
(iii) Splitting fields, separable and inseparable extensions.
1.1.2. Keywords. Extension of a Field, Minimal Polynomial, Splitting Fields.
2 Theory of Field Extensions
1.2. Field. A non-empty set with two binary operations denoted as “+” and “*” is called a field if it is
(i) abelian group w.r.t. “+”
(ii) abelian group w.r.t. “*”
(iii) “*” is distributive over “+”.
1.3. Extension of a Field. Let K and F be any two fields and  : F  K be a monomorphism. Then,
F    F   K . Then,  K ,  is called an extension of field F. Since F    F  and   F  is a
subfield of K, so we may regard F as a subfield of K. So, if K and F are two fields such that F is a
subfield of K then K is called an extension of F and we denote it by K \ F or K | F or I KF .
Note. (i) Every field is an extension of itself.

(ii) Every field is an extension of its every subfield, for example, R is a field extension of Q and C is a
field extension of R.
Remark. Let K | F be any extension. Then, F is a subfield of K. we define a mapping  : FxK  K by
setting
   , k    k for all   F , k  K .
We observe that K becomes a vector space over F under this scalar multiplication. Thus, K must have a
basis and dimension over F.
1.3.1. Degree of an extension. The dimension of K as a vector space over F is called degree of K | F ,
that is, degree of K | F = [K : F].
If [K : F] = n <  , then we say that K is a finite extension of F of degree n

and, if [K : F] =  , then we say that K is an infinite extension of F.
Note. Every field is a vector space over itself. Therefore, deg F | F = deg K | K = 1.
Also, we have [K : F] = 1 iff K = F and [K : F] > 1 iff K  F . [F K]

1.3.2. Example. [C : R] = 2, because basis of vector space C over the field R is {1, i}, that is, every
complex number can be generated by this set. Hence [C : R] = 2.
1.3.3. Transcendental Number. A number (real or complex) is said to be transcendental if it does not
satisfy any polynomial over rationals, for example,  , e .Note that every transcendental number is an
irrational number but converse is not true. For example, 2 is an irrational number but it is not
transcendental because it satisfies the polynomial x2-2.
1.3.4. Algebraic Number. Let K | F be any extension. If   K and  satisfies a polynomial over F,
that is, f(  ) = 0, where f  x   0  1 x  2 x 2  ...  n x n ; i  F .Then,  is called algebraic over F.
If  does not satisfy any polynomial over F, then  is called transcendental over F. For example,  is
transcendental over set of rationals but  is not transcendental over set of reals.
Note. Every element of F is always algebraic over F.
Extension of a Field 3
1.3.5. Example. R | Q is an infinite extension of Q, OR, [R : Q] =  .
Solution. We prove it by contradiction. Let, if possible, [R : Q] = n(finite).

Then, any subset of R having atleast (n+1) elements is always linearly dependent. In particular,  is a
real number and we can take the set {1,  ,  2,…,  n} of n+1 elements. Then, there exists scalars
0 , 1 , 2 ,..., n  Q (not all zero) such that
0  1  2 2  ...  n n  0
Thus,  satisfies the polynomial 0  1 x  2 x2  ...  n xn . So,  is not a transcendental number,

which is a contradiction.
Hence our supposition is wrong. Therefore, [R : Q] =  .
1.3.6. Algebraic Extension. The extension K | F is called algebraic extension if every element of K is
algebraic over F. otherwise, K | F is said to be transcendental extension if atleast one element is not
algebraic over F.
1.3.7. Theorem. Every finite extension is an algebraic extension.
Proof. Let K | F be any extension and let [K : F] = n(finite), that is, dim K | F = n.
Every element of F is obviously algebraic. Now,   K be any arbitrary element. Consider the elements
1,  ,  2,…,  n in K.
Either all these elements are distinct, if not, then  i   j for some i  j . Thus,  i   j  0 .
Consider the polynomial f  x   xi  x j  F[ x] and f     i   j  0 .
Thus,  satisfies f  x   F[ x] and hence  is algebraic over F.
If 1,  ,  2,…,  n are all distinct, then these must be linearly dependent over F. so there exists
0 , 1 , 2 ,..., n  F (not all zero) such that
0  1  2 2  ...  n n  0
Thus,  satisfies the polynomial f  x   0  1 x  2 x 2  ...  n x n . So,  is algebraic over F.
Hence every finite extension is an algebraic extension.

Remark. Converse of above theorem is not true, that is, every algebraic extension is not a finite
extension. We shall give an example for this later on.
1.3.8. Exercise. If an element  satisfies one polynomial over F, then it satisfies infinitely many
polynomials over F.
Proof. Let  satisfies f  x   F[ x] .Then f    0 . We define h  x   f  x  g  x  for any g  x   F[ x] .
Then  also satisfies h  x  .
1.4. Minimal Polynomial. If p  x  be a polynomial over F of smallest degree satisfied by  , then

p  x  is called minimal polynomial of  . W.L.O.G., we can assume that leading co-efficient in p  x 
is 1, that is, p  x  is a monic polynomial.
1.4.1. Lemma. If p  x   F[ x] be a minimal polynomial of  and f  x   F[ x] be any other

polynomial such that f    0 , then p  x  f  x  .
Proof. Since F is a field so F[x] must be a unique factorization domain and so division algorithm hold in
F[x]. therefore, there exists polynomial q  x  and r  x  such that f  x   p  x  q  x  +r  x  where
either r  x   0 or deg r  x   deg p  x  .
Now, f    0  p   q   +r    0  r    0 [ p( )  0]
If r  x   F [ x] is a non-zero polynomial, then it is a contradiction to minimality of p(x), since

degr(x) < degp(x). So, we must have r(x) = 0. Thus, f  x   p  x  q  x  .
Hence p  x  f  x  .
1.4.2. Unique Factorization Domain. An integral domain R with unity is called unique factorization
domain if
(i) Every non-zero element in R is either a unit in R or can be written as a product of finite number
of irreducible elements of R.
(ii) The decomposition in (i) above is unique upto the order and the associates of irreducible
elements.
Remark. Let F be any field and F[x] be a ring of polynomials over F, then division algorithm hold in
F[x].
1.4.3. Corollary. Minimal polynomial of an element is unique.
Proof. Let p(x) and q(x) be two minimal polynomials of  . Since p(x) is a minimal polynomial of  ,
so p  x  q  x  . Thus,
deg p  x   deg q  x  ---(1)
Also, q(x) is a minimal polynomial of  , so q  x  p  x  . Thus,
deg q  x   deg p  x  ---(2)
By (1) and (2), degp(x) = degq(x). Hence

p  x   q  x  for some   F
Now, p(x) and q(x) are both monic polynomials, so comparing the co-efficients of leading terms on both
sides, we get   1 . Therefore, p(x) = q(x).
Remark.   F iff deg p( x)  1, where p(x) is minimal polynomial of  . In this case, p( x)  x   .
1.4.4. Irreducible Polynomial. A polynomial f  x   F[ x] is said to be irreducible over F if

f(x) = g(x)h(x) for some polynomial g  x  , h  x   F[ x] imply that either deg g  x   0 or
deg h  x   0 .
1.4.5. Proposition. Minimal polynomial of any element is irreducible over F.

Proof. Let, if possible, minimal polynomial p(x) of   F is reducible over F. Then, we have
p(x) = q(x)t(x) for some q  x  , t  x   F[ x] .
Then, 0  p    q   t    either q   =0 or t    0
which is not possible because deg q  x   deg p  x  and degt  x   deg p  x  and p(x) is an irreducible
polynomial.
1.4.6. Definition. Let S be a subset of a field K, then the subfield K ' of K is said to be generated by S if
(i) S K'
(ii) For any subfield L of K, S  L implies K '  L and we denote the subfield generated by S by
<S>. Essentially the subfield generated by S is the intersection of all subfields of K which
contains S.
1.4.7. Definition. Let K be a field extension of F and S be any subset of K, then the subfield of K
generated by F  S is said to be the subfield of K generated by S over F and this subfield is denoted by
F(S). However, if S is a finite set and its members are a1 , a2 ,..., an , then we write
F ( S )  F  a1 , a2 ,..., an  . Sometimes, F  a1 , a2 ,..., an  is also called adjunction of a1 , a2 ,..., an over F.
1.4.8. Definition. A field K is said to be finitely generated over F if there exists a finite number of
elements a1 , a2 ,..., an in K such that K  F  a1 , a2 ,..., an  .
In particular, if K is generated by a single element ‘a’ over F, that is, K = F(a), then K is called a simple
extension of F.
1.4.9. Definition. Let K | F be any field extension and let F[x] be the ring of polynomials over F. We
define,
F[a]   f (a) : f ( x)  F[ x]
Let f (a)  F[a] where f  x   0  1 x  2 x 2  ...  n x n  F[ x] . Clearly,
f  a   0  1a  2 a 2  ...  n a n  F (a)
Thus, F[a]  F (a) .
Remark. a1  F iff F  a1   F .
1.4.10. Theorem. Let K | F be any field extension. Then, a  K is algebraic over F iff  F (a) : F  is
finite, that is F(a) is a finite extension over F. Moreover,  F (a) : F   n , where n is the degree of
minimal polynomial of ‘a’ over F.
Proof.Let  F (a) : F  is finite and let  F (a) : F   n . Thus, dim F F ( a )  n
Now,Consider the elements 1, a, a2,…, an in F(a).

These are (n+1) distinct elements of F(a), then these must be linearly dependent over F. so there exists
0 , 1 , 2 ,..., n  F (not all zero) such that
0  1a  2 a2  ...  n an  0
Thus, a satisfies the polynomial f  x   0  1 x  2 x 2  ...  n x n . So, a is algebraic over F.
Hence a is algebraic over F.

Conversely, let a  K be algebraic over F.
Let p  x   F[ x] be the minimal polynomial of ‘a’ over F. Further, let deg p  x   n  1 .
We claim that [F(a) : F] = n.

Let p  x   0  1 x  2 x 2  ...  n x n , n  0 is the minimal polynomial of ‘a’ over F, so p(a) = 0 and,
if g  x   F[ x] is any polynomial such that g(a) = 0, then p(x)|g(x).
Consider t  F[a] . Then, t = f(a) for some f  x   F[ x] .
If t  0 , then f  a   0 , that is, f(x) is not satisfied by ‘a’. Thus, p  x  | g  x  .
Since p(x) is irreducible in F[x] and f  x   F[ x] such that p  x  | f  x  .
As F[x] is an Euclidean ring, so we get g.c.d.(p(x), f(x)) = 1. Therefore, there exists polynomials
h  x  , g  x   F[ x] such that
1  f  x g  x  p  x h  x
Put x = a, 1  f  a  g  a   p  a  h  a   1  f a g a
Now, g  x   F[ x]  g  a   F [a ]  f  a  is invertible .
We know that an integral domain in which every non-zero element is invertible is a field. Hence, F[a] is
a field.
But we know that F[a]  F (a) , where F(a) is the field of quotients of F[a]. Therefore,
F[a] = F(a).
Let t  F[a]  F (a)  t  f (a) for some f ( x)  F[ x] .
Now, f ( x)  F[ x] and p( x)  F[ x] , so by division algorithm, we can write
f(x) = p(x)q(x) + r(x) where either r(x) = 0 or degr(x) < degp(x).

So let r  x   0'  1' x  2' x2  ...  n' 1 x n1  F[ x]
Note that we are saying nothing about 0' , 1' , 2' ,..., n' 1 which enables us to take degree of r(x) is equal
to (n-1).
Then, t  f  a   p  a  q  a   r  a   r  a   0'  1'a  2' a 2  ...  n' 1a n1
Thus, t is a linear combination of 1, a, a 2 ,..., a n 1 over F. Thus, the set 1, a, a 2 ,..., a n 1 generates F(a).
Let, if possible, the set 1, a, a 2 ,..., a n 1 is linearly dependent.
Thus, there exists scalars  0 , 1 ,..., n 1  F (not all zero) such that
 0 1a  2a2  ...  n1an1  0
That is, ‘a’ satisfies a polynomial of (n-1) degree, which is a contradiction to minimal polynomial.
Hence 1, a, a 2 ,..., a n 1 is linearly dependent and so it is a basis for F(a) over F.
Therefore,  F (a) : F   n   .
1.4.11. Theorem. Let K / F be a finite extension of degree n and L / K be a finite extension of degree
m, then L / F is a finite extension of degree mn, that is
[L : F] = [L : K][K : F].
-OR- Prove that finite extension of a finite extension is also a finite extension.
Proof. Given that L / K be a finite extension such that [L : K] = m, that is dim K L  m .
Let x1 , x2 ,..., xm  be a basis of L over K. Now, given that K / F is finite extension such that
[K : F] = n, that is dim F K  n .
Let  y1 , y2 ,..., yn  be a basis of K over F.
Let   L . Then,
m
  1 x1   2 x2  ...   m xm    i xi , i  K
i 1
Now,  i  K and  y1 , y2 ,..., yn  be a basis of K over F, so

n
i  i1 y1  i 2 y2  ...  in yn  ij y j , ij  F
j 1
m m  n 
Thus,     i xi      ij y j  xi    ij xi y j ,  ij  F and xi , y j  L .
i 1 i 1  j 1  i, j
Therefore, x1 y1 , x1 y2 ,..., x1 yn , x2 y1, x2 y2 ,..., x2 yn ,..., xm y1, xm y2 ,..., xm yn  spans L over F and have mn
elements in number.
We claim that these mn elements are linearly independent over F.
If   0 , then
m  n  m
0    ij xi y j      ij y j  xi    i xi
i, j i 1  j 1  i 1
Since  i  K and  x1 , x2 ,..., xm  are L.I. over K. Thus,  i  0 for i  1, 2,..., m .
Again, since  y1 , y2 ,..., yn  are L.I. over F. Thus, ij  0 for j  1, 2,..., n .
Thus, ij  0 for i  1, 2,..., m, j  1, 2,..., n .
So x1 y1 , x1 y2 ,..., x1 yn , x2 y1 , x2 y2 ,..., x2 yn ,..., xm y1, xm y2 ,..., xm yn  is L.I. and hence it is basis for L over
F.
Therefore, [L : F] = [L : K][K : F] = mn.
1.4.12. Proposition. If F  E  K and a  K is algebraic over F, then
 E  a  : E    F  a  : F  .
Proof. Let F  E  K and a  K is algebraic over F. Thus, there exists a polynomial
f  x   0  1 x  2 x 2  ...  n x n  F [ x]
such that f  a   0 .
Since f  x   F[ x] and F  E  F  x  E  x  f  x   E  x  and f (a)  0 .
If p(x) is the minimal polynomial of ‘a’ over F and p1(x) be minimal polynomial of ‘a’ over E, then
p1  x  | p  x  , since p(x) may be reducible in E[x], that is deg p1 ( x)  deg p( x) .
Hence  E  a  : E    F  a  : F  .
Remark. Let K / F be any field extension, then

F  a1 , a2 ,..., an   F  a1 , a2 ,..., an 1  an   F  a1 , a2 ,..., an  2  an 1 , an 
 ...
 F  a1  a2 ,..., an 1 , an 
1.4.13. Theorem. Let K / F be an algebraic extension and L / K is also algebraic extension, then L / F
is an algebraic extension.
-OR- Prove that algebraic extension of an algebraic extension is also a algebraic extension.
Proof. To prove that L / F is algebraic extension, it is sufficient to show that every element of L is
algebraic over F. Equivalently, we have to prove that if a  L , then  F  a  : F    .
Now, ‘a’ satisfies some polynomial f(x) over K[x], say f  x    0  1 x   2 x 2  ...   n x n  K[ x] ,
where  i  K for 0  i  n .
Now,  0 , 1 ,  2 ,...,  n are elements of K and K / F is an algebraic extension. Thus, each  i is algebraic
over F.
Consider the element  0 . Then,  0 is algebraic over F. Thus,
 F  0  : F      F0 : F   , where F0  F  0 
and we have F  F0  K .
Now, 1  K is algebraic over F. So by above remark, we have
 F0 1  : F0    F 1  : F   
Put F0 1   F1 , then  F1 : F0    .
So, we have F  F0  F1  K .
Now, consider F1  2   F1 . Then, as discussed above, we have
 F2 : F1    F1  2  : F1    .
In general similarly, we choose Fi 1 i   Fi , then  Fi : Fi 1    .
Then, by definition, Fn1  n   Fn , then  Fn : Fn1    .
By construction, we get that

Fn  Fn1  n   Fn2  n1 ,  n   ...  F0 1 ,  2 ,...,  n   F 0 , 1 ,  2 ,...,  n  .
Now, by last theorem, we have
 Fn : F    Fn : Fn1  Fn1 : Fn2 ... F1 : F0  F0 : F  .

Thus,  Fn : F  is finite since all the numbers on R.H.S. are finite.
Now, as  0 , 1 ,  2 ,...,  n  Fn , so f  x    0  1 x   2 x 2  ...   n x n  Fn [ x] and since f  a   0 .

Thus, ‘a’ is algebraic over Fn. So
 Fn  a  : Fn   degree of minimal polynomial 'a' over Fn   .
Therefore,  Fn  a  : F    Fn  a  : Fn   Fn : F    .
Thus, Fn ( a ) / F is a finite extension. So Fn (a ) is algebraic extension over F. In turn, ‘a’ is algebraic over
F.
Hence L is algebraic extension of F.
1.4.14. Theorem. Let K / F be any extension and let S   x  K : x is algebraic over F . Then, S is a
subfield of K containing F and S is the largest algebraic extension of F contained in K.
Proof. Let   F  K . Since  satisfies a polynomial f  x   x   in F[x], so  is algebraic over F.
Thus,   S and so F  S . So, S is non-empty.
Let a, b  S . We claim that a  b  S and if b  0 , then ab1  S . Since K is a field, therefore, trivially
a  b  K and if b  0 , then ab1  K .
Now, to prove that a  b  S and if b  0 , then ab1  S it is sufficient to show that a  b and ab 1 are
algebraic over F. We have a  S , that is, ‘a’ is algebraic over F. Thus, [ F (a) : F ]   .
Put F(a) = F1, so [ F1 : F ]   .
Also, b  S , that is, ‘b’ is algebraic over F. Thus, [ F (b) : F ]   .
Now, b is algebraic over F and F  F1  K . So, b is algebraic over F1 and
[ F1 (b) : F1 ]  [ F (b) : F ]  
Now, [ F1 (b) : F ]  [ F1 (b) : F1 ][ F1 : F ]   . Thus, F1(b) is finite extension of F and, thus, F(a,b) is an
algebraic extension of F, as F1(b) = F(a,b). Hence every element F(a,b) is algebraic over F.
Since a, b  F (a, b)  a  b  F (a, b) and ab 1  F (a, b) .
Thus, a-b and ab-1 are algebraic over F.
So, a  b, ab 1  S and, therefore, S is a subfield of K containing F. Hence S is an algebraic extension of

F.
Let E be any other algebraic extension such that F  E  K . Let   E  K    K . Therefore,  is
algebraic over F. Thus,   S  E  S .
So, S is the largest algebraic extension of F contained in K.
1.4.15. Corollary. If K / F is algebraic extension. Then, K = S.
Proof. In above theorem, S is a subfield of K. Therefore, S  K .
Also, S is the largest algebraic extension of F and K is an algebraic extension of F. Therefore, K  S .
Hence S = K.
Note. In above theorem, the field S is called algebraic closure of F in K.
1.4.16. Corollary. If K/F be any extension and a, b  K be algebraic over F. Then,
a  b, a  b, ab and ab 1 (b  0) are also algebraic over F.
Proof. If a and b are algebraic over F, then F(a,b) is algebraic extension of F. So, every element of
F(a,b) is algebraic over F. This implies a  b, a  b, ab and ab 1 (b  0) are also algebraic over F.
1.4.17 Eisenstein Criterion of Irreducibility. Let f  x   0  1 x   2 x 2  ...   n x n where

 i  Z ,  n  0 . Let p be a prime number such that p | 0 , p | 1,..., p | n1, p | n and p2 | 0 , then f(x) is
irreducible over the rationals.
1.4.18. Counter Example. Example to show that every algebraic extension need not be finite.
Let C be the field of complex numbers and Q be the field of rationals. Then z  C is called an algebraic
integer if it is algebraic over Q.
Let E  z  C : z is algebraic integer .
Then, trivially Q  E and so E is a subfield of C containing Q such that E / Q is algebraic extension.
We claim that E / Q is an infinite extension.
Let, if possible, [ E : Q]  n   .
Consider the polynomial f(x) = xn+1-p, where p is some prime.

Then, by Eisenstein criterion of irreducibility, f(x) is irreducible over Q. Let  be any zero of the
polynomial f(x). Then,  will be a complex number such that f    0 . Thus,   E .
Since f(x) = xn+1-p is irreducible monic polynomial satisfied by   E , therefore, f(x) is minimal
polynomial of  over Q. So,
[Q( ) : Q]  n  1
Now,   E and Q  E . So, Q    E , since Q   is the smallest field containing Q and  .

Therefore,
[Q( ) : Q]  [ E : Q]  n 1  n
which is a contradiction. Thus, E / Q is an infinite extension.
1.5. Factor Theorem. Let K / F be any extension and f  x   F[ x] , then the element a  K is a root of
polynomial f(x) iff ( x  a) | f ( x) in K[x], that is, iff there exists some g(x) in K[x] such that f(x) = (x-
a)g(x).
Proof. Let ( x  a) | f ( x) in K[x]. Then, we have f(x) = (x-a)g(x) for some some g(x) in K[x]. Therefore,
f(a) = (a-a)g(a) = 0
Thus, ‘a’ is a root of f(x).
Conversely, let ‘a’ be a root of f(x) where a  K .
Consider thepolynomial x-a in K[x].
Now, f  x   F[ x]  K[ x] . Therefore, by division algorithm in K[x], there exists unique polynomials
q(x) and r(x) in K[x] such that
f  x   ( x  a ) q ( x)  r ( x)
where either r(x) = 0 or degr(x) < deg(x-a) = 1, that is, r(x) = constant.
But f(a) = 0, implies that r(a) = 0. Thus, r(x) = 0.
Hence f(x) = (x-a)g(x). Therefore, ( x  a) | f ( x) in K[x].
Note. We have earlier proved that if ‘a’ is algebraic over F, then F[a] = F(a).
1.5.1. Theorem. Let K / F be any extension and a  K is algebraic over F. Let p( x)  F[ x] be the
minimal polynomial of ‘a’. Then,
F [ x]  p ( x)   F [a]  F (a) .
Proof. Consider the rings F[x] and F[a]. We define the mapping  : F[ x]  F[a] by setting
  f ( x)   f ( a )
We claim that  is an onto ring homomorphism.
Let f ( x), g ( x)  F[ x] . Then,
  f ( x)  g ( x)   f (a)  g (a)    f ( x)     g ( x) 
and   f ( x) g ( x)   f (a) g (a)    f ( x)   g ( x) 
Thus,  is a ring homomorphism.
Again, let   F[a] , then   h(a) for some h( x)  F[ x] .
Then,   h( x)   h(a)   .
Thus,  is onto.
By Fundamental theorem of ring homomorphism

F[ x] Ker  F[a]
Now, we claim that Ker  p( x)  .
Let f  x   Ker    f  x    0  f (a)  0  a satisfies f ( x) .
 p( x) | f ( x) , since p(x) is minimal polynomial.
 f ( x)  p( x)q( x) , for some q( x)  F[ x] .

 f ( x)  p( x)  .
 Ker  p( x)  .
Again, let f ( x)  p( x)  .
 f ( x)  p( x)q( x) , for some q( x)  F[ x] .
 f (a)  p(a)q(a) .
 f ( a)  0 .
   f  x    0  f  x   Ker
  p( x)  Ker .
Thus, Ker  p( x)  and so
F [ x]  p ( x)   F [a ]
Since ‘a’ is algebraic over F, therefore, F[a] = F(a) and hence

F [ x]  p ( x)   F [a]  F (a) .
Note. In the above theorem, preimage of ‘a’ is x+f(x), where f ( x)  p( x)  .
Proof.   x  f ( x)     x  p( x)q( x)    ( x)    p( x)q( x)   a  p(a)q(a)  a .
1.5.2. Conjugates. Let K / F be any extension. Two algebraic elements a, b  K are said to be
conjugates over the field F if they have the same minimal polynomial, that is, we can say that all the
roots of a minimal polynomial are conjugates of each other.
1.5.3. Corollary. If ‘a’ and ‘b’ are two conjugate elements of K over F, where K / F is an extension.
Then, F (a)  F (b) .
Proof. Let p(x) be the minimal polynomial of ‘a’ and ‘b’ both. Then by above theorem
F[ x]  p( x)   F [a] and F[ x]  p( x)   F [b]  F[a]  F[b]
1.5.4. Corollary . If ‘a’ and ‘b’ are any two conjugates over F, then there always exists an isomorphism
 : F[a]  F[b] such that  (a)  b and  ( )   for all   F .
Proof. Given that ‘a’ and ‘b’ are conjugates over F, therefore, they satisfy same minimal polynomial,
say p(x), over F. Then, there exists an isomorphism  1 : F (a)  F [ x]  p ( x)  given by
 1 ( )     p( x)  and  1 ( a)  x   p( x)  . …(1)
Further, p(x) is also minimal polynomial for ‘b’, so there exists an isomorphism
 2 : F (b)  F [ x]  p( x)  given by
 2 ( )     p( x)  and  2 (b)  x   p( x)  . …(2)

1  21
Consider F (a )  F [ x]  p ( x)   F (b) . Take,    211 . Then,
 (a)   211 (a)   21 ( x  p( x) )  b
and  ( )   211 ( )   21 (   p( x) )   .
1.5.5. Definition. Let K / F be any extension and f ( x)  F[ x] be a non-zero polynomial. Then, ‘a’ is
said to be a root of f(x) of multiplicity m  1 if ( x  a) m | f ( x) but ( x  a ) m 1 | f ( x) .
1.5.6. Proposition. Let p( x)  F[ x] be an irreducible polynomial over F. Then, there always exists an
extension E of F which contains atleast one root of p(x) and  E : F   n  deg p  x  .
Proof. Let I = <p(x)> be an ideal of F[x]. Now, we know that a ring of polynomials over a field is a
Euclidean domain and any ideal of Euclidean domain is maximal iff it is generated by some irreducible
element. So, F[x] is a Euclidean domain and I = <p(x)> is a maximal ideal as p(x) is irreducible.
Now, since every Euclidean domain possess unity, therefore, F[x] is a commutative ring with unity. We
further know that if R is a commutative ring with unity and M is a maximal ideal of R, then R/M is a
field. So, F [ x]  p( x)  is a field.
We claim that E is an extension of F.

We define a mapping  : F  E by setting
 ( )      I for all   F .
Then, for 1 , 2  F , we have
 (1  2 )  1  2  I   1  I    2  I    (1 )   (2 )
and  (12 )  12  I   1  I  2  I    (1 ) (2 )
Therefore,  is a homomorphism.
Also, if  (1 )   (2 )  1  I  2  I  1  2  I  I  p( x) 
 1  2  p( x)   p( x) | 1  2  1  2  0  1  2
Therefore,  is monomorphism.
Thus, (E,  ) is an extension of F.
Let p  x   0  1 x  2 x 2  ...  n x n  I  p( x) 
Consider the element x  x  I  E . Then,

p  x   0  1 x  2 x 2  ...  n x n  0  1 x  2 x 2  ...  n x n  I  p( x)  I  I
Thus, p(x) has a root x in E.

We claim that 1, x , x 2 ,..., x n 1 form a basis of E over F. Let us consider a representation
0 1  1 x  2 x 2  ...  n 1 x n 1  0, identity of E
 0  1 x  2 x 2  ...  n 1 x n 1  I  I
 0  1 x  2 x 2  ...  n 1 x n 1  I  p ( x) 
 p( x) | 0  1 x  2 x 2  ...  n 1 x n 1
 0  1  2  ...  n 1  0 ( deg p( x)  n)
Thus, 1, x , x 2 ,..., x n 1 are linearly independent.
Further, let   E  F [ x]  p( x)  , then   f ( x)  I for some f ( x)  F[ x] .
We can write f(x) = p(x)q(x) + r(x), where either r(x) = 0 or degr(x) < degp(x).
Then,
  f ( x)  I   p ( x)q ( x )  r ( x )   I
.
  p( x)q ( x)  I    r ( x)  I   I  r ( x )  I  r ( x )  I
But degr(x) < n, therefore,

  r ( x)  I   0   1 x   2 x 2  ...   n 1 x n 1  I
  0 (1  I )   1 ( x  I )   2 ( x 2  I )  ...   n 1 ( x n 1  I )
2 n 1
  01   1 x   2 x  ...   n 1 x
2 n 1
Thus, 1, x, x ,..., x generates E and so it is a basis for E.
Hence we get [E : F] = n = degp(x).

1.5.7. Theorem. Let f ( x)  F[ x] be any polynomial of degree n  1 , then no extension of F contains
more than n roots of f(x).
Proof. Given that f ( x)  F[ x] and degf(x) = n.
If n = 1, then f ( x)   x   ,  ,  F,  0 .
Consider the element   1  F . Then, f    1   0 . Thus,   1 is a root of f(x).
Let K be any extension of F and let  be any root of f(x) in K, then
f    0      0     1
So, any extension K of F contains the only root   1 of f(x). Therefore, K cannot contain more than
one root of the polynomial f(x).
Since K was an arbitrary extension, so Theorem is true for n = 1.
Let us assume that the result is true for all polynomials of degree less than degree of f(x) over any field.
Now, let E be any extension of F. If E does not contain any root of f(x), then result is trivially true.
So, let E contain atleast one root of the polynomial f(x) say ‘a’. Then, we have to prove that E does not
contain more than n roots. Since a  E and ‘a’ is a root of f(x). suppose the multiplicity of ‘a’ is m.
Then, by definition, we can write
f ( x)  ( x  a) m g ( x), g ( x)  E[ x]
and ( x  a) m | f ( x) but ( x  a ) m 1 | f ( x) .
Now, ( x  a) m | f ( x) , therefore, m  n .
Further, g ( x)  E[ x] and degg(x) = n-m < n.
Therefore, by induction hypothesis, any extension of E does not contain more than n-m roots of g(x). So,
E / E being an extension of E cannot contain more than n-m roots of g(x). Now, any root of g(x) is also
a root of f(x) and a root of f(x) other than ‘a’ is also a root of g(x). Hence f(x) cannot have more than
(n-m)+m, that is, n roots in any extension of F.
1.5.8. Theorem. Let f ( x)  F[ x] be any polynomial of degree n. Then, there exists an extension E of F
containing all the roots of f(x) and [ E : F ]  n!.
Proof. We prove the result by induction on n.
Given that f ( x)  F[ x] be a polynomial of degree n.
If n = 1, then f ( x)   x   ,   0 , with a root   1 . Since
 ,   F    1  F .
Hence F contains all the roots of the given polynomial with [ F : F ]  1  1! .
Thus, result is true for n = 1.
Let n > 1 and suppose that result is true for any polynomial of degree less that n over any field.
Then, f ( x)  F[ x] is either irreducible or f(x) has an irreducible factor over F. Now, let p( x)  F[ x] be
any irreducible factor of f(x). Then, deg p( x)  deg f ( x)  n .
Suppose that degp(x) = m. Then, p( x)  F[ x] is irreducible polynomial over F with degp(x) = m.

Therefore, there exists an extension E ' of F containing atleast one root of p(x) and [ E ' : F ]  m  n .
Let  be a root of p(x) in E ' , then  is also a root of f(x). So, we get that f ( x)  F[ x] is a polynomial
with root   E ' such that [ E ' : F ]  m  n . Since   E ' is a root of f(x) so ( x   ) | f ( x) in E '[ x] .
Hence we can write f ( x)  ( x   ) g ( x) where g ( x)  E '[ x] and degg(x) = n-1. Now, g ( x)  E '[ x] and
degg(x) = n-1 < n.
Therefore, by induction hypothesis, there exists an extension E of E ' such that E contains all the roots
of g(x) and [ E : E ']  n  1! .
Since   E '  E    E also.
Therefore, E is an extension of F which contains all the roots of f(x). Then, we have
[ E : F ]  [ E : E '][ E ' : F ]  n 1!.m  n(n 1)!  n!.
1.5.9. Remark. Let R and R ' be any rings and  : R  R ' is an isomorphism onto. Consider the rings
R[x] and R '[t ] . Then,  can be extended to an isomorphism from R[x] to R '[t ] .
Proof. Let f ( x)  R[ x] and f ( x)  0  1 x  2 x2  ...  n xn .
We define  : R[ x]  R '[t ] by setting
 ( f ( x))   (0 )   (1 )t   (2 )t 2  ...   (n )t n

We claim that  is an extension of  and is an isomorphism also.
Let g ( x)   0   1x   2 x2  ...   m xm  R[ x] . Then, if k = max{m,n}
 ( f ( x)  g ( x))   (0   0 )   (1   1 )t   (2   2 )t 2  ...   (k   k )t k

  (0 )   ( 0 )  [ (1 )   ( 1 )]t  ...  [ (k )   ( k )]t k
  ( f ( x))   ( g ( x))
Similarly, we can show that
 ( f ( x) g ( x))   ( f ( x)) ( g ( x))
Therefore,  is a ring homomorphism.
We claim that  is one-one.
Let f ( x)  ker    ( f ( x))  0, identity of R[x]
  (0 )   (1 )t   (2 )t 2  ...   (n )t n  0   (i )  0 for all 0  i  n
Since  is a monomorphism, so i  0 for all 0  i  n .
Thus, f ( x)  0  ker  {0}
Therefore,  is a monomorphism.
We claim that  is onto.
Let f '(t )  R '[t ] and f '(t )   0'   1't  ...   n' t n where  i'  R ' .
Now, since  : R  R ' is onto, therefore, there exists  i  R such that  ( i )   i' .
Consider f ( x)   0   1 x   2 x2  ...   n xn  R[ x] and we have
 ( f ( x))  f '(t )
Therefore,  is onto.
Remark. If f ( x)  0  1 x  2 x2  ...  n xn . Then, f '(t )  0'  1't  ...  n' t n where  (i )  i' is called
the corresponding polynomial of f(x) in R '[t ] .
Remark. f ( x)  R[ x] is irreducible iff f '(t )  R '[t ] is irreducible, where f '(t ) is corresponding

polynomial of f(x). Also, if A is any ideal in R[x] then  ( A) is also an ideal of R '[t ] . Further, A is
maximal iff  ( A) is maximal. Also, we can find an isomorphism  * such that
 *: R[ x] A  R '[t ]   A given by
 *( f ( x)  A)  f '(t )   ( A) .
1.5.10. Proposition. Let  : F  F ' be an isomorphism onto. Let p(x) be any irreducible polynomial of
degree n in F[x] and p '(t ) be corresponding polynomial in F '(t ) . Let u be any root of p(x) and v be any
root of p '(t ) in some extension of F and F ' respectively. Then, there exists an isomorphism, say
 : F (u)  F '(v) which is onto and is such that  ( )   ( ) for all   F and  (u)  v .
Proof. Given that p( x)  F[ x] is irreducible polynomial over F with root u which is in some extension
of F. Then, we know that there exists an isomorphism onto, say  1 : F [ x]  p ( x)   F (u ) given by
 1 ( f ( x)  p( x) )  f (u )
and [F(u) : F] = degree of minimal polynomial of u over F.

Since p '(t ) is irreducible polynomial over F ' andvis a root of p '(t ) in some extension of F ' , so there
exists an isomorphism onto, say  2 : F '[t ]  p '(t )   F '(v) given by
 2 ( g '(t )  p '(t ) )  g '(v)
Now,  : F  F ' is given to be an isomorphism onto. By last remarks, we have  is also an extension
of  from F ( x)  F '(t ) with  ( p( x))  p '(t ) and correspondingly, we denote the isomorphism for
F[ x]  p( x)   F '[t ]  p '(t )  by  again. Now, we have
11 : F (u)  F[ x]  p( x) 
 : F[ x]  p( x)   F '[t ]  p '(t ) 
 2 : F '[t ]  p '(t )   F '(v)
Consider    211 : F (u)  F '(v) .
Now,  2 , and 11 are all isomorphism onto, therefore,  is also isomorphism onto.
For   F , we have
 ( )   2 11 ( )   2  11 ( )    2 (   p( x) )   2 ( ( )  p '(t ) )   ( )
Now, compute
 (u)   211 (u)   2 ( x  p( x) )   2 (t   p '(t ) )  v .
1.6. Splitting Field. Let F be any field and f ( x)  F[ x] be any polynomial over F. An extension E of F
is called a splitting field of f(x) over F if
(i) f(x)is written as a product of linear factors over E.
(ii) If E ' is any other extension of F such that f(x) is written as product of linear factors over E ' ,
then E  E ' .
Remark. We have proved a theorem that for any polynomial f ( x)  F[ x] , where degf(x) = n, there
always exist an extension E of F such that E contains all the roots of f(x) and [ E : F ]  n!. So, we can
say that splitting field of a polynomial is always a finite extension.
1.6.1. Another Form. Let f ( x)  F[ x] and let 1 ,  2 ,...,  n be roots of f(x). Consider the extension
K  F 1 ,  2 ,...,  n  . By definition, K is the smallest extension of F containing 1 ,  2 ,...,  n . Also, let E
be the splitting field of F.
Now, F  E and also 1 ,  2 ,...,  n  E , therefore, K  E .
Also, E  K , since E is the splitting field. Therefore,

E = K.
Thus, splitting field is always obtained by adjunction of all the roots of f(x) with F. Hence if
f ( x)  F[ x] is a polynomial of degree n and 1 ,  2 ,...,  n are its roots, then splitting field is
F 1 ,  2 ,...,  n  .
1.6.2. Example. Let F be any field and K be its extension. Let a  K be algebraic over F of degree m
and b  K be algebraic over F of degree n such that (m, n) = 1. Then, [ F  a, b  : F ]  mn .
Solution. Let p(x) be minimal polynomial of ‘a’ over F. Then,

degp(x) = m = [F(a) : F].
Let q(x) be the minimal polynomial of ‘b’ over F. Then,
degq(x) = n = [F(b) : F].
Now, [F(a,b) : F] = [F(a,b) : F(a)][F(a) : F] = [F(a,b) : F(b)][F(b) : F] …(*)
Therefore, m  [ F (a) : F ]|[ F (a, b) : F ] and n  [ F (b) : F ]|[ F (a, b) : F ] .
Since (m, n)  1  mn |[ F (a, b) : F ]  [ F (a, b) : F ]  mn …(1)
Now, a  F (a, b) is algebraic over F with minimal polynomial p(x) of degree m.
Since F  F (b)  p( x)  F (b)[ x] . Therefore, ‘a’ is algebraic over F(b).

So, let t(x) be the minimal polynomial of ‘a’ over F(b).
Now, p(a)  0  t ( x) | p( x)  deg p( x)  deg t ( x)  deg t ( x)  m .
 [ F (a, b) : F (b)]  [ F (b)(a) : F (b)]  deg t ( x)  m
Then, by (*),
[ F (a, b) : F ]  [ F (a, b) : F (b)][ F (b) : F ]  mn …(1)
By (1) and (2), we have
[ F  a, b  : F ]  mn .
1.6.3. Definition. A field F is said to be algebraically closed field if it has no algebraic extension.
Thus, a field is called algebraically closed if f(x) has splitting field E, then E = F. For example, field of
complex numbers is algebraically closed.
1.6.4. Remark. Algebraically closed fields are always infinite.
Proof. Let F be any algebraically closed field and, if possible, suppose that F is finite. Then,
F = {a1,a2,…,an}. Consider the polynomial
f(x) = (x-a1)(x-a2)…(x-an)+1
in F, where 1 is unity of F.
This polynomial has no roots in F. So, F cannot be algebraically closed.
Hence our supposition is wrong and so F must be infinite.
1.6.5. Example. Find the splitting field and its degree for the polynomial f(x) = x3 – 2 over Q.
Solution. Let x 3  2  Q[ x ] . Then,   3 2, w, w2 are its roots.
Let E be the splitting field of x3 – 2 over Q. Therefore,  ,  w,  w2  E  w E .
Thus, E  Q( , w)
Consider [E : Q]. Here,   E and   Q . So,
[ E : Q]  [ E : Q( )][Q( ) : Q]
Now,   Q , therefore,
[Q( ) : Q]  degree of minimal polynomial of  over Q  3

since x3 – 2 is monic and irreducible.
Also, w  E and w  Q . Therefore,
[Q(w) : Q] = 2
since basis of Q(w) over Q is {1,w}. Also,
[E : Q] = [E : Q(w)][Q(w) : Q]
Since (2, 3) = 1, so we have [E : Q] = 6 = 3!.
1.6.6. Algebraic Number. A complex number is said to be an algebraic number if it is algebraic over
the field of rational numbers.
1.6.7. Algebraic Integer. An algebraic number is said to be an algebraic integer if it satisfies a monic
polynomial over integers.
Exercise. Find the splitting field and its degree over Q for the polynomials
(a) f(x) = xp-1
(b) f(x) = x4-1
(c) f(x) = x2+3
Exercise. Show that the polynomials x2+3 and x2+x+1 have same splitting field over Q.
Exercise. Show that sinm0 is an algebraic integer for every integer m.
Exercise. Show that 2  3 5 is algebraic over Q of degree 6.

1.6.8. Example. If a  K is algebraic over F of odd degree show that F(a) = F(a2).
Solution. Let K be an extension of F and a  K be algebraic of odd degree. Let p(x) be minimal
polynomial of ‘a’. We can write
p( x)  0  1 x  ...  2n x2n  2n1 x2n1
Now, a  F (a)  a 2  F (a)  F ( a 2 )  F ( a) …(1)
To prove F (a )  F (a 2 ) , it is sufficient to prove that a  F (a 2 ) .
We are given that p(a) = 0, that is,
0  1a  ...  2n a2n  2n1a2n1  0
 a(2n 1a2n  2n 1a2n 1  ...  1 )  2na2n  2n  2a2n  2  ...  0  0
 a(2n1a2n  2n1a2n2  ...  1 )  (2n a2n  2n2 a2n2  ...  0 )
 aX  Y …(2)
where X  2n1a2n  2n1a2n2  ...  1 , Y  2n a2n  2n2a2n2  ...  0 in F(a2).
Now, we prove that X  0 .

If X = 0, then ‘a’ satisfies the polynomial
2n1 x2n  2n1 x2n2  ...  1
which is of degree 2n < degp(x).

But p(x) is minimal polynomial of ‘a’ which is a contradiction. Hence X  0 and so X-1 exists. By (2),
a = -YX-1
But X  F (a 2 ), Y  F (a 2 )  -YX 1  F ( a 2 )  a  F (a 2 ) .
Therefore, F (a)  F (a 2 ) ---(3)
By (1) and (3), we have
F (a)  F (a 2 )
Remark. Let F be a field of characteristic p and let f(x) = xp-1.
Then, f '( x)  px p 1  0 [ p.1 = 0] .
So, degree of f '( x) depends upon the characteristic of field considered.
Again, let F = {0, 1} be the given field and f(x) be a polynomial over F given by
f ( x)  x10  x 9  ...  x  1
Then, f '( x)  10 x9  9 x8  ...  2 x  1  0 x9  x8  ...  1  x8  x 6  ...  1
So, deg f '( x)  8 .
1.6.9. Lemma.Let f ( x)  F[ x] be a non-constant polynomial. Then, an element  of field extension K

of F is a multiple root of f(x) iff  is a common root of f(x) and f '( x) .
Proof. Let  be a root of f(x) of multiplicity m > 1. Then, we can write
f ( x)  ( x   ) m g ( x), g ( x)  K [ x] and g ( )  0
f '( x)  m( x   ) m 1 g ( x)  ( x   ) m g '( x)
f '( )  m(   ) m 1 g ( )  (   ) m g '( )  0
Thus,  is a root f '( x) also.
Conversely, let  is a common root of f(x) and f '( x) . Then, we have to prove that  is a multiple root
of f(x).
Let, if possible,  is not a multiple root of f(x).
Then, f ( x)  ( x   ) g ( x), g ( x)  K[ x] and g ( )  0 .
Therefore, f '( x)  g ( x)  ( x   ) g '( x) and so f '( )  g ( )  0 , a contradiction.
Hence  is a multiple root of f(x).

1.6.10. Lemma. Let f ( x)  F[ x] be irreducible polynomial over F, then f(x) has a multiple root in some
extension of F iff f '( x)  0 identically.
Proof. Let f ( x)  F[ x] has a multiple root of multiplicity m > 1, in some extension K of F where f(x) is
an irreducible polynomial over F.
Let f ( x)  0  1x  ...  n xn  F[ x] be an irreducible polynomial of degree n. Let  be its multiple

root of multiplicity m > 1. Then, by above lemma,  is also a root of f '( x) , that is, f '( )  0 . But
f '( x)  1  22 x  ...  nn xn1  F[ x] and deg f '( x)  n  1 .
W.L.O.G., we can assume that n  1 so that f(x) is monic and irreducible polynomial and hence is
minimal polynomial of  . But  satisfies f '( x) . Therefore, f ( x) | f '( x) .
Thus, f '( x)  0 identically, since deg f '( x)  deg f ( x) .
Conversely, let f '( x)  0 and K the splitting field of f(x) over F. Let deg f ( x)  n .
Let 1 , 2 ,..., n be the roots of f(x) in K. We can write

f ( x)   ( x  1 )( x  2 )...( x  n ) for some   F .
Then, we have
f '( x)   ( x  2 )...( x  n )   ( x  1 )( x  3 )...( x  n )  ...   ( x  1 )( x  2 )...( x  n 1 )
 f '(i )   (i  1 )...(i  i 1 )(i  i 1 )...(i  n )
Now, since f '( x)  0 identically, so f '(i )  0 . But   0  i   j for some i  j .
Therefore, f(x) has multiple roots.

1.6.11. Corollary. Let charF = 0 and f(x) be any irreducible polynomial over F, then f(x) cannot have
multiple roots.
Proof. Let degf(x) = n >1.
Let f ( x)  0  1x  ...  n xn  F[ x] . Here n > 1 and n  0 .
f '( x)  1  22 x  ...  nn xn1
Now, nn  0  f '( )  0  f '( x)  0
Hence by above lemma, f(x) cannot have multiple roots.

Remark. Any irreducible polynomial over field of rationals, field of reals or field of complex numbers
cannot have multiple roots because all these fields are of characteristic zero.
1.7. Separable polynomial. Let f ( x)  F[ x] be any polynomial of degree n > 1, then it is said to be
separable over F if all its irreducible factors are separable. Otherwise f(x) is said to be inseparable.
1.7.1. Separable irreducible polynomial. An irreducible polynomial f ( x)  F[ x] of degree n is said to

be separable over F if it has n distinct roots in its splitting field, that is, it has no multiple roots.
1.7.2. Inseparable irreducible polynomial. An irreducible polynomial which is not separable over F is
called inseparable over F. Equivalently, if f ( x)  F[ x] is irreducible polynomial having multiple roots
of multiplicity n > 1 is called inseparable over F.
Remark. By the corollary of above lemma, we conclude that inseparable implies ch.F  0 and ch.F = 0
implies separable. But converse is not true, that is, if ch.F  0 , then the polynomial may be separable or
inseparable.
1.7.3. Lemma. Let ch.F  p( 0) and f ( x)  F[ x] be an irreducible polynomial over F. Then, f(x) is
inseparable iff f ( x)  F [ x p ] .
Proof. Let f(x) be any irreducible polynomial over F of degree n and is separable. Let
f ( x)  0  1x  ...  n xn , n  0
Therefore, f '( x)  1  22 x  ...  nn xn1
Since f ( x)  F[ x] is irreducible polynomial and is inseparable, so f(x) must have repeated roots.
Therefore,
f '( x)  0  1  22 x  ...  nn xn1  0  1  22  ...  nn  0 ---(*)
Since i  F andch.F p > 0. Therefore, if k i  0  p | k or if p | k , then i  0 .
Therefore, by (*), we get

1  2  ...  p1  0
and pp  0  p may or may not be zero.
Further, ( p  1)p1  0  p1  0 . So
p1  p2  ...  2 p1  0
Again, 2 p2 p  0  2 p may or may not be zero and so on. Therefore,

f ( x)  0   p x p  2 p x2 p  ...  m xmp
where n = mp if m  0 . Thus,
f ( x)  0   p x p  2 p  x p   ...  m  x p   F[ x p ]
2 m
Conversely, if f ( x)  F [ x p ] . Then,
f ( x)  0   p x p  2 p x2 p  ...  k xkp
where 0 , p , 2 p ,..., k  F .
Then, f '( x)  0  p p x p 1  2 p2 p x 2 p 1  ...  kpk x kp 1  0 [ch.F  p] .
Thus, f(x) has multiple roots and hence f(x) is inseparable.

1.7.4. Separable Element. Let K be any extension of F. An algebraic element   K is said to be
separable over F if the minimal polynomial of  is separable over F.
1.7.5. Separable Extension. An algebraic extension K of F is called separable extension if every
element of K is separable.
1.7.6. Proposition. Prove that if ch.F = 0, then any algebraic extension of F is always separable
extension.
Proof. Given that ch.F = 0 and let K be any algebraic extension of F. Let   K . Then,  is algebraic
over F.
So, let p(x) be the minimal polynomial of  over F. Then, p(x) is irreducible polynomial over F and so
p(x) is separable.
Therefore,  is separable. But  was an arbitrary element of K. So, K is separable extension.
1.7.7. Perfect Field. A field F is called perfect if all its finite extensions are separable.
1.7.8. Theorem. Let K be an algebraic extension of F, where F is a perfect field then K is separable
extension of F.
Proof. Let a  K . Since K is algebraic, so ‘a’ is algebraic over F. Therefore,
[F(a) : F] = degree of minimal polynomial of ‘a’ over F = r (say)
Thus, F(a) is finite extension. But F is perfect, therefore, F(a) is separable extension. So, ‘a’ is separable
over F.
Hence K is separable.
1.7.9. Theorem. Let ch.F = p > 0. Prove that the element ‘a’ in some extension of F is separable iff
F(ap) = F(a).
Proof. Let K be some extension of F such that a  K and ‘a’ is separable over F. So, ‘a’ is algebraic
element with its minimal polynomial, say
f ( x)  0  1x  ...  n1xn1  xn
and f(x) has no multiple roots.

Let g(x) be the polynomial
g ( x)  0p  1p x  ...  np1xn1  xn
Then,
g (a p )  0p  1p a p  ...  np1a( n1) p  anp   0  1a  ...  n1a n1  a n    f (a)   0
p p
Therefore, apsatisfies a polynomial g ( x)  F[ x] .
Now, a  F (a)  a p  F (a)  F ( a p )  F ( a) ---(1)
Further, F(ap) and F(a) both are vector spaces over F and F (a p )  F (a) , therefore,
[ F (a p ) : F ]  [ F (a) : F ]  n
We claim that [ F (a p ) : F ]  n .
We know that [ F (a p ) : F ] = degree of minimal polynomial of apover F.
We shall prove that g(x) is minimal polynomial of ap over F. For this, it is sufficient to prove that g(x) is
an irreducible polynomial.
Let h( x)  F[ x] be a factor of g(x). Then,
g(x) = h(x)t(x)
for some t ( x)  F[ x] . Thus,
g(xp) = h(xp)t(xp)
and so h(xp) is a factor of g(xp) in F[x].
    f ( x) 
p
But g ( x p )  0p  1p x p  ...  np1 x( n1) p  x np  0  1 x  ...  n1 x n 1  x n
p
 h( x p ) |  f ( x )   h( x p )   f ( x)  for some integer k , 0  k  p.

p k
Taking derivatives both sides
h '( x p ) px p 1  k  f ( x)  f '( x)  0  k  f ( x) 
k 1 k 1
f '( x) [ch.F  p]
Since f(x) is separable polynomial so f '( x)  0 . Therefore, either k = 0 or k = p.
If k = p, then h(xp) = (f(x))p = g(xp)  h( x)  g ( x) .
If k = 0, then h(xp) = (f(x))0 = 1  h( x p )  1 , a constant function, so h(x) = 1.
Thus, g(x) is irreducible polynomial of degree n, therefore,

[F(ap) : F] = n.
Hence [F(ap) : F] = [F(a) : F]  F (a p )  F (a ) .
Conversely, suppose F (a p )  F (a ) .
We claim that ‘a’ is separable over F.

Let, if possible, ‘a’ is not separable.
Let f ( x)  F[ x] be the minimal polynomial of ‘a’. Then, by our assumption f(x) is not separable over F.
Since ch.F = p > 0 and f(x) is inseparable over F.
So, f ( x)  F [ x p ] .
Let f ( x)  g ( x p ) for some g ( x)  F [ x]  g (a p )  f (a)  0 .
ap is a root of the polynomial g ( x)  F[ x] . But
deg f ( x) n
deg f ( x)   , where n = deg f(x).
p p
n
Therefore, degree of minimal polynomial of a p  .
p
n
So, we get n  [ F (a) : F ]  [ F (a p ) : F ] 
p
which is a contradiction. Hence ‘a’ is separable over F.
1. Find the splitting field of x5-1 over Q.
2. Find the splitting field of x2-9 over Q.
3. Show that [K : F] = 1 if and only if K = F.
1.9. Summary.
In this chapter, we have defined Extension of a field and derived various results. The result worth
mentioning is that if p(x) is a polynomial of degree n over some field F, then the number of zeros, to be
considered, of this polynomial depends on the extension that we are considering.
Books Suggested:
2
Galois Theory
Structure
2.1. Introduction.
2.2. Normal Extension.
2.3. F-Automorphism.
2.4. Galois Extension.
2.5. Norms and Traces.
2.7. Summary.
2.1. Introduction. In this chapter, we shall discuss about normal extensions, fixed fields, Galois
extensions, norms, traces and the dependence of all these on normal extensions.
(i) Normal Extensions.
(ii) Fixed Fields, Galios Groups
(iii) Norms and Traces.
2.1.2. Keywords. Normal Extensions, Galois Group, Fixed Fields.
2.3. Normal Extension. An algebraic extension K of F is said to be normal extension of F if each
irreducible polynomial f(x) over F having a root in K splits into linear factors over K, that is, if one root
is in K, then all the roots are in K.
If E is the splitting field of f(x) over F such that a root ‘a’ of f(x) is in K, then E  K .
Galois Theory 29
2.3.1. Lemma. Let [K : F] = 2, then K is normal extension of F always.
Proof. Let g ( x)  F[ x] be any irreducible polynomial over F. Let  be a root of f(x) and   K . Now,
we have
[ F ( ) : F ]  [ K : F ]  2  [ F ( ) : F ]  2  deg f ( x)  2 .
If degf(x) = 1, then let
f(x) = ax+b with a, b  F , a  0 .
b
Then, 0  f ( )  a  b     ,a  0 .
a
b
But   F  K   K .
a
If degf(x) = 2, then let f(x) = ax2+bx+c with a  0 . If  be a root of f(x), then,
b
f ( x)  ( x   )( x    ), a  K
a
b
  (  ) is other root of f(x).
a
b b
Since  F  K and   K   (  )  K .
a a
Hence K is a normal extension of F.
2.3.2. Theorem. Let K be a finite algebraic extension of a field F then K is a normal extension of F iff K
is the splitting field of some non-zero polynomial over F.
Proof. Let K = F(a1,a2,…,an) be a finite algebraic extension of F. Suppose K is normal extension of F.
For each ai  K , let fi(x) be the minimal polynomial of ai over F. Since K is normal extension of F, so
fi(x) splits completely into linear factors over K.
Let f(x) = f1(x)f2(x)…fn(x).
Let ‘a’ be any root of f(x), then ‘a’ is also a root of some fi(x) and hence a  K . Let E be the splitting
field of f(x). Then, E  K .
n
Now, F (ai )   f j (ai )  0 . Therefore, ai is a root of f(x), that is, ai  E .
j 1
Therefore, F (a1 , a2 ,..., an )  E  KE.
Thus, K = E.
Hence K is the splitting field of f(x) over F.
Conversely, let K be the splitting field of some non-zero polynomial f(x) over F. Let a1,a2,…,an be the
roots of f(x). Then, K = F(a1,a2,…,an).
By definition, [ K : F ]  n! .
So, K is finite algebraic extension of F. Let p(x) be any irreducible polynomial over F with a root  in
K. p(x) is also a polynomial over K with ( x   ) as a factor in K[x]. So p(x) is not irreducible over K.
Let L be the splitting field of p(x) over K. We claim that L=K.

Let, if possible, L  K . Then, there exists a root  ' of p(x) in L such that  '  K . As  and  ' are
conjugates over F, there exists an isomorphism  : F ( )  F ( ') such that
 ( )   ' and  ( )   for every  in F . Now, F  F ( )  K gives K is a splitting field of f(x) over
F ( ) .
Further, K (  ')  F (a1 , a2 ,..., an )(  ')  F (  ')(a1 , a2 ,..., an ) gives K ( ') is a splitting field of f(x) over
F ( ') . Then, there exists an isomorphism  : K  K ( ') such that
 ( x)   ( x) for every x in F( ) .

But then  ( )   ( )   ' and  ( )   ( )   for every  in F .
Hence  : K  K ( ') is an onto isomorphism, such that  ( )   ' and  ( )   for every  in F . If
f ( x)  0  1x  ...  n1xn1  n xn
in F[x] with  n  0 . Then,
f ( x)   n ( x  a1 )( x  a2 )...( x  an )
Let  ' : K[ x]  K ( ')[ x] be an extension of  such that
 '( f ( x))   '( 0  1 x  ...   n1 x n1   n x n )   '( 0 )   '(1 ) x  ...   '( n 1 ) x n1   '( n ) x n
  ( 0 )   (1 ) x  ...   ( n 1 ) x n 1   ( n ) x n   0  1 x  ...   n 1 x n 1   n x n
 f ( x)
Also,
 '( f ( x))   '( n ( x  a1 )( x  a2 )...( x  an ))   '( n ) '( x  a1 ) '( x  a2 )... '( x  an )
  n ( x   (a1 ))( x   (a2 ))...( x   (an ))
We get that  (a1 ), (a2 ),..., (an ) are also roots of f(x). Since  is one-one, so
 (a1 ), (a2 ),..., (an )  a1, a2 ,..., an 

It implies  permutes the roots of f(x). Therefore,
K  F (a1 , a2 ,..., an )  F ( (a1 ), (a2 ),..., (an ))
Galois Theory 31
However,
K ( ')   ( K )    F (a1 , a2 ,..., an )   F ( (a1 ), (a2 ),..., (an ))  F ( a1, a2 ,..., an )  K
It implies  '  K , which is a contradiction.
Thus, L = K, so p(x) splits completely over K. Hence K is a normal extension of F.

2.3.3. Corollary. Let K be a finite normal extension of F. If E be any subfield of K such that
F  E  K , then K is normal extension of E.
Proof. Since K is a finite normal extension of F, so there exist a polynomial f(x) over F such that K is
splitting field of f(x) over F. Then K is also a splitting field of f(x) over E. Hence by above theorem K is
normal extension of E.
2.3.4. Corollary. Let K be finite normal extension of F. If 1 and  2 be any two elements in K
conjugate over F, then there exists an F automorphism  of K such that  (1 )   2 .
Proof. Let K be the splitting field of the non-zero polynomial f(x) over F. Since 1, 2 are conjugates
over F there exist an isomorphism  such that  : F(1)  F(2) defined by
 (1) = (2) and  () =  for all   F.
Now [F(1) : F] = [F(2) : F] = degree of minimal polynomial of 1 (or 2).
Now , f(x)  F[x]  F(1)[x] and f(x)  F[x]  F(2)[x]
Therefore, K is splitting field of f(x) over F(1) as well as F(2).
Then there exists  : K  K s.t. () =  () for all   F(1) and () =  () =  for all   F.
Then (1) =  (1) = 2. Hence  is an F-automorphism of K such that (1) = 2.
Remark. Converse of Corollary 1 need not be true, for if F  Q, E  Q( 2) and K  Q( 4 2) . Then K is

normal extension of E, E is normal extension of F but K is not a normal extension of F.
2.3.5. M(S, K). Let K be any field and S be any non-empty set.The set of all mappings from S to K is
denoted by M(S, K).
2.3.6. Theorem. If  1 ,  2 ,...,  n be any n monomorphisms in M(E, K), then these are always L.I., where
E and K are fields.
Proof. If n = 1, then consider  1 and let, for a1  K
a11  0  a11 ( )  0 for all   E
Since a1 1 is a homomorphism from E to K and
a1 1 ( )  0 for all   E
In particular,  a11  (1)  0 where 1 E   a1 1 (1)  0 .

Since  1 is a monomorphism so  1 (1)  0 , then a1 = 0.
Hence  1 is linearly independent.
Now, let us assume, as our induction hypothesis, that  1 ,  2 ,...,  n 1 are L.I.
We have to prove that  1 ,  2 ,...,  n are L.I.
Let 1 , 2 ,..., n are scalars such that
11  2 2  ...  n n  0 …(1)
If any of i is zero, then the above relation reduces to a combination of (n - 1)  i ’s and by induction
hypothesis, all i ’s are zero. Hence we assume that 1 , 2 ,..., n are all non-zero.
So, let W.L.O.G., n  0 . Then dividing (1) by n , we have
b11  b2 2  ...  bn1 n1   n  0 …(2)
i
where bi   i n1 .
n
Since  1 and  n are distinct, so there exists an element x1  E such that
 1 ( x1 )   n ( x1 )
Then, clearly x1  0 , since image of 0 is 0 for any homomorphism.
Now, let x  E be any element then xx1  E also. Compute
 b11  b2 2  ...  bn1 n1   n  xx1   0  xx1   0

 b11  xx1   b2 2  xx1   ...  bn1 n1  xx1    n  xx1   0
 b11  x  1  x1   b2 2  x  2  x1   ...  bn1 n1  x  n1  x1    n  x  n  x1   0
Since  n ( x1 )  0 , so dividing above equation by  n ( x1 ) .
 1  x1   x   x 
b1  1  x   b2 2 1  2  x   ...  bn 1 n 1 1  n 1  x    n  x   0 …(*)
 n  x1   n  x1   n  x1 
From (2), we also have

b1 1 ( x)  b2 2 ( x)  ...  bn 1 n 1 ( x)   n ( x)  0 …(**)
Subtracting (**) from (*), we get

Galois Theory 33
  x    x    x  
b1  1 1  1  1  x   b2  2 1  1  2  x   ...  bn1  n1 1  1  n1  x   0 …(3)
  n  x1     n  x1     n  x1  
 1 ( x1 )  1 ( x1 )
Since  1 ( x1 )   n ( x1 )  1  1  0
 n ( x1 )  n ( x1 )
Now as above equation (3) holds for every x  E , so
  x    x    x  
b1  1 1  1  1  b2  2 1  1  2  ...  bn1  n1 1  1  n1  0
  n  x1     n  x1     n  x1  
which is a combination of (n-1)  i ’s. So, we get
  x    x    x  
b1  1 1  1  b2  2 1  1  ...  bn 1  n1 1  1  0
  n  x1     n  x1     n  x1  
 (x ) 
Now, as 1 1  1  0 , so b1 = 0 and so 1  0 , which implies 1  0 , a contradiction.
 n ( x1 ) n
Hence any set of n monomorphism is linearly independent.
2.3.7. Definition. Let K be any field, then the set of all automorphisms on K is denoted by AutK.
2.3.8. Corollary. AutKconsists of linearly independent elements.
Take E = K in above theorem, the result follows.
2.3.9. Exercise. The set of all automorphisms of K form a group under composition of mappings.
2.4. F-Automorphisms. Let F be any field and K be any extension of F. An automorphism  : K  K
is called F-automorphism of K if
 ( x)  x for all x  F .
Notation. G(K, F) will denote the group of all F-automorphisms of K. G(K, F) is called Galio’s group of
K over F and known as group of automorphisms from K to K which fixes F.
2.4.1. Exercise. Prove that G(K, F) is a subfield of AutK.
2.4.2. Theorem. If P is a prime subfield of K, then prove that AutK = G(K, P), that is every
automorphism on K fixes P.
Proof. Let   Aut (K) then  (0) = 0 and  (1) = 1
Case 1. CharK = P for some prime p.
Then P  Zp = {0 , 1 ,......, p  1}. If   Zp then  = 1+1+......+1 ( times)
 () =  (1+1+......+1) =  (1) + (1)+......+  (1) = 1+1+......+1 = 
  () =  for all   Zp.   fixes P.
   G(K , P)  Aut K  G(K , P).

Case 2. CharK = 0.
Then P  Q = {mn1 : m n  Z} and
 (mn1) =  (m)  (n1) =  (m)  ( n)

1
= mn1 for all mn1  Q
  fixes P.    G(K , P)  Aut K  G(K , P).

So, in both cases, we get Aut (K)  G(K , P). But G(K , P)  Aut (K) always.
So Aut (K) = G(K , P).
2.4.3. Theorem. Let K be any extension of F and   G( K , F ) . If ‘a’ is an element which is algebraic
over F then ‘a’ and ‘  (a ) ’ are conjugates over F.
Proof. We know that G(K , F) = {   Aut K : () =  for all   F }.

Let a  K be an algebraic element over F. So let f(x) = 0 + 1x +...+ xn be the minimal polynomial of
‘a’ over F and then 0 = f(a) = 0 +1a +...+ an  K also, since a , 0 , 1 , ...  K.
Now, 0 =  (0) =   f ( a)  =  (0 +1a +...+ an)
=  (0) + (1)  (a) + ...+  (an)
= 0 + 1 (a) +...+  ( a)  = f  ( a) 

n
 f  ( a )  = 0, so (a) is also a root of f(x)
  (a) is conjugate of ‘a’ over F.

2.4.4. Exercise. Let G be a group of automorphisms of a field K. Then, the set
F0  {x  K :  ( x)  x for all   G} is a subfield of K.
Also, this subfield is known as fixed field under G.
2.4.5. Example. Let K  Q  2  . The minimal polynomial of

3 3
2 over Q is x3 – 2. It has only one root,
namely, 3
2 in K. Since K is a field of real numbers. Let  be any Q – automorphisms of K. Then
  2   K is a root of x
3 3
– 2. So,   2 
3 3
2 . Let x be any element of K, then x can be expressed as:
 2  c , where a, b, c  Q .
2
a  3 2b  3
So,  ( x)   (a)   ( 3 2) (b)     2    ( c)  a 

3
2
3
2b   2 c  x
3
2
   I . Thus, AutK = { I }.
Hence in this case K itself is the fixed field under AutK.
Galois Theory 35
2.4.6. Theorem. Let G be a finite subgroup of AutK. If F0 is fixed subfield under G, that is,
F0  {x  K :  ( x)  x for all   G} . Then, [K : F0] = o(G).
Proof. Let [K : F0] = m and o(G) = n.

Let, if possible, m < n.
Let  1 ,  2 ,...,  n are elements of G and let {x1, x2, . . . , xm} be a basis of K over F0.
Consider a system of m linear homogeneous equations, 1  j  m
 1  x j  u1   2  x j  u2  ...   n  x j  un  0 …(1)
Note that  1  x j  ,  2  x j  ,...,  n  x j  are elements of K and u1, u2, . . . , un are variables.
Since the number of equations is less that the number of variables, so the system (1) has a non-trivial
solution, say, y1, y2, . . . , yn, here not all yi’s are zero.
 1  x j  y1   2  x j  y2  ...   n  x j  yn  0 …(2)
for j = 1, 2, … , m.
Now, if x  K , then
x  1 x1   2 x2  ...   n xn , where  i  F0 .
Multiplying jth equation of (2) by  j , we get
 1  x j  y1 j   2  x j  y2 j  ...   n  x j  yn j  0

  1  x j   1  j  y1   2  x j   2  j  y2  ...   n  x j   n  j  yn  0
because  j  F0 and  j  G and F0 is fixed under G.
  1  j x j  y1   2  j x j  y2  ...   n  j x j  yn  0 for j = 1, 2, …, m.
Thus, we have the system of equations,
1 1 x1  y1   2 1 x1  y2  ...   n 1 x1  yn  0
1  2 x2  y1   2  2 x2  y2  ...   n  2 x2  yn  0
…………………………………………………..
1  m xm  y1   2  m xm  y2  ...   n  m xm  yn  0
Adding all these equations, we get
 1 1 x1   2 x2  ...   m xm  y1   2 1 x1   2 x2  ...   m xm  y2
...   n 1 x1   2 x2  ...   m xm  yn  0
 1  x  y1   2  x  y2  ...   n  x  yn  0 for all x  E

  y11  y2 2  ...  yn n  x   0 for all x  E
 y11  y2 2  ...  yn n  0
where atleast one of y j  0 .
Hence  1 ,  2 ,...,  n are L.D. over K, a contradiction.
Thus, m  n .
Now, if possible, suppose that m > n.
Then, there exist (n + 1) L.I. elements, say x1, x2, . . . , xn+1 in K over F0. Consider the system of n linear
homogeneous equations in (n+1) variables
 j  x1  u1   j  x2  u2  ...   j  xn1  un1  0 …(3)
for j = 1,2,…n.
Since the number of variables is again greater than the number of equations, so these homogeneous
equations have a non-trivial solution. Let z1, z2, . . . , zn+1 be a non-trivial solution of the system (3). Let r
be the smallest non-zero integer such that zj = 0 for all j  r  1 .
Then, the system (3) reduces to

 j  x1  z1   j  x2  z2  ...   j  xr  zr  0 …(4)
zi
Since zr  0 and z r  K . Consider, zil  . Then, from (4), we get
zr
 j  x1  z1l   j  x2  z2l  ...   j  xr 1  zrl 1   j  xr   0 …(5)
for j = 1,2,…n.
Let for j = 1,  1  I , we get from (5), that
x1 z1l  x2 z2l  ...  xr 1zrl 1  xr  0 …(6)
If all z1l , z2l ,..., zrl 1 are in F0, then from (6), we get that x1, x2, . . . , xr are L.D. over F0, which is not
possible.
Hence atleast one of zil is not in F0, say z1l  F0 .

Further, we get that r  1 , because if r = 1, then we get that z1l  1 and so z1l  F0 .
Since z1l  F0 , so there exists some  i  G such that  i  z1l   z1l .
Applying  i  G to (5), to get
 i  j  x1  z1l    i  j  x2  z2l   ...   i  j  xr 1  zrl 1    i  j  xr    0
  i j  x1   i  z1l    i j  x2   i  z2l   ...   i j  xr 1   i  zrl 1    i j  xr   0

Galois Theory 37
Since G is a group, the set  i1 ,  i 2 ,...,  i n  coincide with the set 1 ,  2 ,...,  n  , though the order
of elements will be different. So, we get
 j  x1   i  z1l    j  x2   i  z2l   ...   j  xr 1   i  zrl 1    j  xr   0 …(7)
Subtracting (7) from (5), we have
 j  x1   z1l   i  z1l    j  x2   z2l   i  z2l   ...   j  xr 1   zrl 1   i  zrl 1   0
Put tk  zkl   i  zkl  . Then, the above system becomes
 j  x1  t1   j  x2  t2  ...   j  xr 1  tr 1  0
where t1  0 . Thus, (t1, t2, …, tr-1, 0, 0, …, 0) is a non-trivial solution of given system, which is a
contradiction to the choice of r. Therefore, n  m
So, m = n. Hence the proof.
2.5. Galois Extension. A finite extension K of a field F is said to be Galoi’s extension of F if F is the
fixed subfield of K under the group G(K, F) of all F-automorphisms of K i.e. K/F is Galoi’s extension if
KG  K , F   F .
2.5.1. Simple Extension. An extension K/F is said to be simple extension if K is generated by a single
element over F.
2.5.2. Corollary. Let K  F ( ) be a simple finite separable extension of F. Then, K is the splitting field
of the minimal polynomial of  over F iff F is the fixed field under the group of all F-automorphisms of
K, that is K is Galoi’s extension of F.
Proof : Let f(x) be the minimal polynomial of  over F and let degree f(x) = m.
Then [K : F] = m. Let 1 = , 2 , 3 ,…, r be the distinct conjugates of  in K.
Then K = F (i) for all i = 1, 2, …, r. Since  and i are conjugates over F, so  an isomorphism, say
i : F(1)  F(i) given by i(1) = i and i() =  for all   F. But K = F(i) for all i, so we
have that
i : K  K s.t. i(1) = i and i() =  for all   F.
Since 1 generates K over F, each i is uniquely determined. Further, we know for any F-automorphism
 of K , i(1) is a conjugate of 1 and hence i(1) = i for some i.
From this, it follows that  = i for some i.
Hence the group G(K , F) consists of 1 , 2 ,…, r . Let F0 be the fixed field under G(K, F). Then by
theorem 2.4.6.,
[K : F0] = o[G(K , F)] = r.
So , F = F0 if and only if r = m. Hence F is the fixed field under G if and only iff f(x) has all m roots in
K, that is, if and only if K is the splitting field of f(x) over F.
2.5.3. Theorem. Let K be a finite extension of F and ch.F = 0. Then, K is normal extension of F iff the
fixed field under G(K, F) is F itself, that is, K is Galoi’s extension of F.
Proof. We know that any finite field extension of a field of characteristic zero is simple extension so
K F is a simple extension. So , let K = F() for some   K.
Now, suppose that K is a normal extension of F. Then, by definition, every irreducible polynomial over
F having one root in K splits into linear factors over K. Since [K : F] is finite , so  is algebraic over F.
Let f(x) be minimal polynomial of  over F and K be its splitting field over F. Then K  K. Also,
  K , F  K
 K  K.
So K = K i.e. K is splitting field of f(x) over F. Hence, by corollary 2.5.2., F is itself fixed subfield under
G(K , F), that is, K F is Galois extension.
Conversely, suppose that F is itself the fixed subfield under G(K, F). Again, by corollary 2.5.2., K is the
splitting field of the minimal polynomial of  over F. Further we know that if K is a finite algebraic
extension of a field F iff K is the splitting field of some non-zero polynomial over F. Hence K is a
normal extension of F.
2.5.4. Fundamental Theorem of Galoi’s Theory.
Given any subfield E of K containing F and subgroup H of G(K, F)
(i) E  KG ( K , E )
(ii) H  G( K , K H )
(iii) [K : E] = o(G(K, E)) and [E : F] = index of G(K, E) in G(K, F)
(iv) E is a normal extension of F iff G(K, E) is a normal subgroup of G(K, F)
(v) when E is a normal extension of F, then
G( E, F )  G( K , F ) .
G( K , E )
Proof. (i) Since K is a finite normal extension of F and F  E  K , we must have that K is a finite
normal extension of E. so, by above theorem fixed field under G(K, E) is E itself, that is E = G(K, E).
(ii) By definition, K H  {x  K :  ( x)  x    H } , that is each element of KH remains invariant
under every automorphisms of H. So, clearly, we have
H  G(K , K H )
Now, we know that if F0 is fixed subfield under subgroup G, then [K : F0] = o(G).
Here KH is fixed subfield under H, so we must have [K: KH] = o(H) …(1)
Galois Theory 39
Now, K is normal extension of KH, so KH is fixed subfield under G(K, KH), by above theorem. So again
we have
[K : KH] = o(G(K, KH)) …(2)
By (1) and (2), we obtain
O(H) = o(G(K, KH))
So, H = G(K, KH)
(iii) Since K|F and K|E both are finite normal extensions, so by above theorem fixed field under G(K,
F) and G(K, E) are F and E respectively.
Hence [K : E] = o(G(K, E)) and [K : F] = o(G(K, F))
Now, [K : F] = [K : E][E : F]
[ K : F ] o(G ( K : F ))
So [E : F ]    index of G(K, E) in G(K, F)
[ K : E ] o(G ( K : E ))
(iv) Let E be a normal extension of F. Then, E is algebraic extension of F. Let a  E , then ‘a’ is
algebraic over F. Let p(x) be the minimal polynomial of ‘a’ over F. Then, E|F being normal and E
contains a root of p(x), then all roots of p(x) are in F.
Hence E contains all the conjugates of ‘a’ over F. Let   G( K , F ) , then  (a ) is a conjugate of ‘a’ and
hence  (a)  E .
Let   G( K , E ) then  : K  K such that  ( )   for all   E . In particular,
 ( (a))   (a) [ (a)  E ]
  1 ( ( ( a)))   1 ( a)  a  ( 1 )( a)  a   1  G( K , E )
Hence G( K , E )  G( K , F ) .
Conversely, let G( K , E )  G( K , F ) .
We shall prove that E is a normal extension of F.
Let a  E  K  a  K and K is normal extension of F.
Therefore, K contains all the roots of minimal polynomial p(x) of ‘a’ over F. Equivalently, if L is the
splitting field of p(x) over F, then L  K .
Let b be any other root of p(x), then b  L  K and b is a conjugate of ‘a’ over F. Hence there exists an
isomorphism  : K  K such that
 (a)  b and  ( )   for all   F
Let   G( K , E ) , then  1  G ( K , E ) . Therefore,
 1 (a)  a   ( (a))   ( a)   (b)  b for all   G ( K , E )
But E is fixed under G(K, E), therefore, we get
b   (a)  E  b  E  L  E
Thus, E is normal extension of F.
(v) Let E be a normal extension of F. Then, E = F(a) for some a  E . For any   G( K , F ) , let  E
denotes the restriction of  to E. Since  (a)  E , we get  ( E)  E .
But [ ( E) : F ]  [ E : F ] . Therefore, we get  ( E )  E . Hence  E is an F-automorphism of E and so

 E  G(E, F ) .
Define a mapping  : G( K , F )  G( E, F ) by setting
 ( )   E for all   G ( K , F )
Clearly, for any  ,  G( K , F ) , we have
 ( )  ( ) E = E E   ( ) ( )
Hence  is a group homomorphism.
Consider any   G( E, F ) . Now,  (a) is a conjugate of ‘a’ over F. Thus, there exists an
F-automorphism  on K such that  (a)   (a) .
Further, as  and  are both identity of F and E is generated by ‘a’ over F. We get
 ( x)   ( x) for all x  F (a)  E     E   ( )
This proves  is onto mapping. Hence
G( E , F )  G( K , F ) Ker
Now, if   Ker  iff  E is identity on E iff  ( x)  x for all x  E iff   G( K , E) .
Hence Ker  G( K , E) and we obtain
G( E, F )  G( K , F ) G( K , E ) .
2.5.5. Example. Determining Galois group of splitting field of x4+1 over Q.

Solution. Roots of x4+1 over Q are
( 2 m 1)  i
x= e 4 , m=0,1,2,3
i 3 i 5 i 7i
= e4 , e 4 , e 4 , e 4
i
Let a= e4 ,
Then roots are x = a , a3 , a5 , a7
Therefore, splitting field K of x4+1 over Q is given by
K = Q(a , a3 , a5 , a7) = Q(a).
Clearly, x4+1 is irreducible over Q , so it is minimal polynomial of x4+1 over Q.
Now , [K : Q] = [Q(a) : Q]
= degree of minimal polynomial of ‘a’ over Q
= degree (x4+1) = 4
Since K is splitting field of some non-zero polynomial over Q, so K must be normal extension of Q.
Also, charQ = 0, so we must have that the fixed field under the Galois group G(K, Q) is Q itself.
So, we must have o  G( K , Q )  = [K : Q] = 4
Galois Theory 41
Now , K = Q(a) and [K : Q] = 4
so {1, a, a2, a3} must be a basis of K over Q. If y  K be any arbitrary element, then
y = 0.1 + 1. a + 2 . a2 + 3 . a3 , i  Q , 0  i  3.
and  (y) =  (0.1) +  (1. a) +  (2 . a2) +  (3 . a3)
= 0 + 1  (a) + 2  ( a)  + 3  ( a)
2 3
Hence any   G(K , Q) is determined by its effect on ‘a’.

Now, (a) must be a conjugate of ‘a’ and G(K, Q) contains four elements, so we must have
G(K , Q) = {1 , 2 , 3 , 4}, where 1(a) = a, 2(a) = a3, 3(a) = a5, 4(a) = a7.
Now, G(K, Q) is a group of order four means that either it is a cyclic group of order 4 or it is isomorphic
to Klein’s group.
We observe that 1(a) = a  1 = I and 22 (a) = 2  2 ( a)  = a9 = a
32 (a) = 3  3 ( a )  = a25 = a and 24 (a) = 4  4 ( a)  = a49 = a
Hence, 22 = 32 = 24 = I.
So, the Galois group G(K, Q) contains no element of order 4 which in turn implies that
G(K, Q) is isomorphic to Klein’s four group.
2.6. Norms and Traces.
Let E be a finite separable extension of degree n over the subfield F and K be a normal closure of E over
F. Then, there are exactly n distinct F-monomorphisms, say,  i ,1  i  n , of E into K. Consider the
mappings NE/F and SE/F of E into K as:
n n
N E / F ( x)   i ( x) , S E / F ( x)   i ( x) ,
i 1 i 1
for every x  E and 1 ≤ i ≤ n.

Then, NE/F(x) and SE/F(x) are known as norm and trace respectively of x from E to F.
The next theorem, indicates why to use “of x from E to F” in the definition of norm and trace.
2.6.1. Theorem. Norm, NE/F(x) is a homomorphism of the group E* = E-{0} of the field E into the group
F* = F-{0} of the field F. Also, the trace SE/F is a non-zero homomorphism of the additive group E of the
field E into the additive group F of F.
Proof. For justifying that these mappings are homomorphisms on the said structures, consider x, y  E ,
then
n n n n
N E / F ( xy )   i ( xy )   i ( x) i ( y )   i ( x) i ( y )  N E / F ( x) N E / F ( y )
i 1 i 1 i 1 i 1
and,
n n n n
S E / F ( x  y )   i ( x  y )    i ( x)   i ( y )    i ( x)   i ( y )  S E / F ( x)  S E / F ( y )
i 1 i 1 i 1 i 1
Further, if  is any F-automorphism of K, then, for x  E , the mappings ρi, 1 ≤ i ≤ n, of E into K

defined by i ( x)   ( i ( x)) are clearly n distinct F- monomorphisms of E into K and so
{ 1 ,  2 ,...,  n } = {  1 , 2 ,..., n }, might be with different order. Let x be any arbitrary element of E, then
 n  n n
  N E / F ( x)      i ( x)    i ( x)   i ( x)  N E / F ( x)
 i 1  i 1 i 1
 n
 n n
and   S E / F ( x)      i ( x)    i ( x)   i ( x)  S E / F ( x) .
 i 1  i 1 i 1
Therefore, norm and trace of x belong to the fixed field under G(K,F). Since K is a normal closure of a
seperable extension, so it is finite separable normal extension of F. Hence it follows that the fixed field
under G(K,F) is F itself. Hence N E / F ( x), S E / F ( x)  F .
Now, we only need to prove that SE/F is not the zero homomorphism. On the contrary assume that
n
S E / F ( x)   i ( x)  0, for all x  E
i 1
However, it concludes that the set {  1 , 2 ,..., n } of distinct monomorphisms of E into K is linearly
dependent over K, which in turn contradicts as we already have proved the result “If E and K be any two
fields, then every set of distinct monomorphisms of E into K is linearly independent”. Hence the proof.
Now consider two possibilities:
1. Let D be a finite separable extension of subfield F and E be a subfield of D, containing F. Then
D is a separable extension of E and E is a separable extension of F. Thus if x is any element of D,
define the norm N D / E ( x) of x from D to E, which is an element of E as obtained in Theorem 1,
and then define the norm of N D / E ( x) from E to F, which is an element of F.
2. Also, define the norm of x from D to F.

The next theorem shows that these two procedures lead to the same element of F.
2.6.2. Theorem. Let D be a finite separable extension of a subfield F and E be a subfield of D
containing F. Then, for every x  D,
i) NE/F(ND/E(x)) = ND/F(x)
ii) SE/F(SD/E(x)) = SD/F(x).
Galois Theory 43
Proof. Let K be a normal closure of D over F and [E : F] = n, [D : E] = m, then due to tower law,
[D : F] = mn. Thus, there are exactly n distinct F-monomorphisms σ1,…..,σn (say) of E into K and m
distinct E-monomorphisms ԏ1, …… , ԏm (say) of D into K. Extending σ1,…..,σn from E to K, we can
obtain n distinct F-automorphisms 1' ,  2' ,...,  n' of K which act like σ1,…..,σn on E.
Let φij(i = 1,…, n; j = 1,…., m) be the mappings of D into K defined by
φij(x) =  i' (ԏj(x)) for all x  D.
These mn mappings are distinct F-monomorphisms of D into K and hence they form a complete set of
F-monomorphisms of D into K. If x  D, then we have
 
N D / F ( x)    ( x)    
ij i
'
j ( x)     i'    j ( x) 
1i  n
1 j  m
1i  n
1 j  m
1i  n  1 j  m 
   i'  N D / E ( x)     i  N D / E ( x)   N E / F  N D / E ( x) 
1i  n 1i  n
Similarly, we can derive the result for traces also.

1. Consider F = Q and E = Q(i), define norm and trace for this structure.
2. Find the Galois group of x3  2 over Q.
2.8. Summary.
In this chapter, we have derived results related to normal extensions and observed that finite algebraic
extension is normal if it becomes splitting field of a non-zero polynomial
Books Suggested:
5. Lang, S., Algebra, 3rd editioin, Addison-Wesley, 1993.
3
Galois Fields
Structure
3.1. Introduction.
3.2. Galois Field.
3.3. Normal Bases.
3.4. Cyclotomic Extensions.
3.5. Cyclotomic Polynomial.
3.6. Cyclotomic Extensions of the Rational Number Field.
3.7. Cyclic Extensions.
3.9. Summary.
3.1. Introduction. In this chapter, we shall discuss about finite fields, cyclic and cyclotomic extensions.
Also it will be derived that a field of composite order does not exist. Further, the relation between finite
division rings and finite fields is obtained.
(i) Normal bases.
(ii) Cyclic and Cyclotomic Extensions.
(iii) Cyclotomic Polynomials.
3.1.2. Keywords. Galois Field, Normal Extensions, Splitting Fields.
Galois Field 45
3.2. Galois Field. A field is said to be Galois field if it is finite.
3.2.1. Theorem. Let F be a field having q elements and ch.F = p, where p is a prime number. Then,
q = pn for some integer n  1 .
Proof. Let P be the prime subfield of F. Now, we know that upto isomorphism there are only two
prime fields, one is Q and other is Z p. Since P is finite prime field. So, P must be isomorphic to Z p.
Hence P must have p elements. Now, F is a finite field and P  F so F is a finite dimensional
vector space over P.
Let [F : P] = n(say) and let {a1, a2, …, an} be a basis of F over P. Then, each element of F can be written
uniquely as
1a1  2 a2  ...  n an where i  P .
As each i can be choosen in p ways, the total number of elements of F is pn.
So, we have q = pn for some integer n  1 .

Remark. In the other direction of above theorem, we shall show that for every prime p and integer n  1 ,
there exists a field having pn elements. First we prove a lemma:
3.2.2. Lemma. If a field F has q elements, then F is the splitting field of f ( x)  x q  x  P[ x] , where P
is the prime subfield of F.
Proof. We know that the set of all non-zero elements of a field form an abelian group w.r.t.
multiplication. So, F* = F – {0} is a multiplicative abelian group. Now, we are given that o(F) = q.
Therefore, o(F*) = q-1.
Now, let  be an arbitrary element of F*. Then,
 q1  1
where 1 is the multiplicative identity of F. Thus,
 q1    q    q    0
That is,  satisfies the polynomial f ( x)  x q  x . Therefore, all the elements of F* are root of
f ( x)  x q  x . Also, f(0) = 0 and so
f ( )  0 for all   F
Since f(x) is of degree q, so it cannot have more than q roots in any extension of P. Thus, F is the
smallest extension of P containing all the roots of f(x).
Hence F is the splitting field of f(x) over P.
Remark. In above lemma, we have proved that every finite field is splitting field of some non-zero
polynomial.
3.2.3. Theorem. For every prime p and integer n  1 , there exists a field having pn elements.
Proof. Since p is a prime number. Therefore, Zp = {0, 1, …, p-1} is a field w.r.t. +p and xp and is also a
prime field. Consider the polynomial
f ( x)  x p  x  Z p [ x]
n
Let K be the splitting field of f(x). Then, K contain all the roots of f(x).
Since degree of f(x) is pn, so f(x) has pn roots in K. Let these roots be a1 , a2 ,..., a pn . Then, we can write
pn
x  x   ( x  ai )
pn
where ai  K .
i 1
Let T  {a  K : a p  a} . Then, T  0 , because 0  T as 0 p  0 and 0  K .

n n
Now, 1 K and 1p  1  1  T .
n
Let k  Z p be any arbitrary element. Then, k = 1+1+…+1 (k-times). Therefore,
k p  (1  1  ...  1) p  1p  1p  ...  1p  1  1  ...  1  k [ch.F  p]

n n n n n
 k T
So, every element of Zp is in T, that is, T contains prime field Zp of K. Further, consider ai any root of
f(x). Then,
f  ai   0  aip  ai  0  aip  ai  ai  T
n n
Thus, T also contains all the roots of f(x).

We claim that T is a subfield of f(x).
Let  ,  T . Then,  p   and  p   . Now,

n n
(   ) p   p   p  0         T
n n n
( ) p   p  p      T .
n n n
and
Thus, T is a subfield of K. So, T  K .

So, we have T is a field which contains all the roots of f(x). But K is splitting field of f(x). So, K  T .
Thus, we have K = T.
Now, if   T , then  p     p    0  f ( )  0
n n
Thus, every element of T is a root of f(x).

Therefore, T  {a1 , a2 ,..., a pn } .
Now, we claim that all these elements are distinct.

Galois Field 47
We have f ( x)  x p  x . Any root ai of f(x) is a multiple root of f ( x) iff ai is a root of f '( x) . But
n
1
f '( x)  p n x p  1  1 ch.Z p  p
n
So, ai is not a root of f '( x) . Therefore, no root of f(x) is a multiple root. So, all elements of T are
distinct. Hence
o(T) = pn = o(K).
Thus, we have obtained a field of order pn.
3.2.4. Theorem. Finite fields having same number of elements are isomorphic.
Proof. Let K1 and K2 be finite fields such that o(K1) = o(K2).
Let ch.K1 = p1 and ch.K2 = p2, where p1 and p2 are primes. Then, we have
Then, we have o  K1   p1n1 and o  K2   p2n2 for some integers n1 and n2. So, we have
p1n1  p2n2  p1  p2  p(say) and n1  n2  n(say)
Let P1 and P2 are prime subfields of K1 and K2 respectively. Then,

P1  Z  p   P2 . So, P1  P2
By previous lemma, K1 is the splitting field of the polynomial f ( x)  x p  x  P1[ x] .

n
Now, P1  P2 so P1[ x]  P2 [ x] .
Let f '(t ) be the corresponding polynomial of f(x) and f '(t )  t p  t  P2 [t ] .

n
Again, by previous lemma, K2 is the splitting field of the polynomial f '(t )  P2 [t ] .
But P1  P2 . Therefore, splitting field will also be isomorphic, that is, K1  K 2 .
3.2.5. Theorem. A field is finite iff F* = F – {0} is a multiplicative cyclic group.

Proof. Let F be a finite field with q elements. Then, F* = F –{0} is a multiplicative group with (q – 1)
elements.
We claim that F* contains elements having order (q - 1).
Since F* is a finite group, so if   F * , then by Lagrange’s theorem
 o( F *)  1 for all   F *
That is, multiplicative order of each element is finite, so let ‘n’ be the least positive integer such that
 n  1 for all   F *
Then, n  q  1 .
Now, consider the polynomial f(x) = xn – 1.

Then, f ( )   n  1  0   satisfies f(x) for all   F * .

But f(x) is of degree n, it can have atmost n roots. Also, all elements of F* are roots of f(x). Therefore,
o( F *)  n  q  1  n .
Hence there exists atleast one element   F * such that o( )  o( F*)  q  1 .
Therefore, F* is cyclic.
Conversely, suppose that F* is cyclic. Let F* = < a >.
If a = 1, then o(F*) = o(a) = o(1) = 1. So, F = {0, 1} is finite.
So, let us assume that a  1 .
Case I. ch.F  0
Since 1  F *   1  F * . Therefore, 1  an for some integer n.
W.L.O.G., let n  1 , then
a 2 n  1  o(a)  2n  o(a) is finite  o( F *) is finite  o( F ) is finite.
Since Ch.F = 0, then prime subfield P of F is such that P  F and P  Q , a contradiction, as

o(Q)   and o( P)   .
Hence this case is not possible.

Case II. ch.F  0
Then, we must have ch.F = p for some prime p.
Let P be the subfield of F, then P  Z p and o(P) = p. Since a  1, a 1 F
 a  1 F *  a   a  1  a n for some integer n  an  a 1  0 .
Thus, ‘a’ satisfies the polynomial f(x) = xn – x – 1 over P[x] and hence ‘a’ is algebraic over P.
Then, [P(a) : P] = degree of minimal polynomial of ‘a’ over P = r (say)
Therefore, P(a) is a vector space over P of dimension r. Thus, P(a)  P( r )  {(1 ,2 ,...,r ) : i  P} .
But o( P)  p  o( P ( r ) )  p r  o( P (a ))  p r . Now, F* = < a > and a  P(a) .
 F*  P(a)  o(F*)  o( P(a))  o(F*)   .
Therefore, o(F*) is finite.

Remark. The above theorem may not be true when a field F is infinite. We give an example of field of
rational numbers. Let Q*  {  Q :   0} .
We shall prove that the multiplicative group Q* is not cyclic.
Let, if possible, Q* is cyclic. So, let g be its generator, that is, Q* = < g >, where
Galois Field 49
p11 p22 ... prr

g  1 2 t
q1 q2 ...qt
where pi’s and qi’s are distinct primes.
Now since 1 Q * , so there must exist a positive integer n such that
n
 p11 p22 ... prr 
1  g   1 2 t 
n
 p1n1 p2n2 ... prnr  q1n1 q2n2 ...qtnt
 q1 q2 ...qt 
which is a contradiction, since pi’s and qi’s are distinct primes. Hence Q* is not cyclic.
Remark. In view of the above remark, we can say that R* and C* are not cyclic because every subgroup
of a cyclic group is cyclic and Q* is not cyclic.
3.3. Normal Bases. Let K be a finite separable normal extension of a subfield F and
G( K , F )  1 , 2 ,..., n 
be the Galois group of K over F. If x  K , then a basis of the form 1 ( x), 2 ( x),..., n ( x) for K over F is
called a normal basis of K over F.
3.3.1. Theorem. Let K be a finite separable normal extension of degree n over a subfield F with Galois
group G( K , F )  1 , 2 ,..., n  . The subset  x1 , x2 ,..., xn  of K is a basis for K over F if and only if the
matrix
  1 ( x1 )  1 ( x2 )  1 ( xn ) 
 
 ( x )  ( x ) ...  2 ( xn ) 
 i ( x j )    2 1 2 2 
 
 n ( x1 )  n ( x2 )  n ( xn ) 
is non-singular.
Proof. Suppose first that the matrix  i ( x j )  is non-singular.
Since [K : F] = n, so it is enough to show that the set  x1 , x2 ,..., xn  is linearly independent over F. For
this, consider
a1 x1  a2 x2  ...  an xn  0
where ai ,1  i  n, are elements of F.
Applying the F-automorphisms  1 , 2 ,..., n , to obtain
a1 1 ( x1 )  a2 1 ( x2 )  ...  an 1 ( xn )  0

a1 2 ( x1 )  a2 2 ( x2 )  ...  an 2 ( xn )  0
. . .
. . .
. . .
a1 n ( x1 )  a2 n ( x2 )  ...  an n ( xn )  0,
which is a homogeneous system of equations in unknowns ai ,1  i  n, with non-singular matrix of

coefficients  ( x )  .
i j It follows from the theory of homogeneous linear equations that
a1  a2  ...  an  0 . Thus  x1 , x2 ,..., xn  is linearly independent and so forms a basis, as required.
Next, suppose that the matrix  i ( x j )  is singular.
Again, due to the theory of homogeneous linear equations, it follows that there exist a non-trivial
solution for the system
a1 1 ( x1 )  a2 1 ( x2 )  ...  an 1 ( xn )  0
a1 2 ( x1 )  a2 2 ( x2 )  ...  an 2 ( xn )  0
. . .
. . .
. . .
a1 n ( x1 )  a2 n ( x2 )  ...  an n ( xn )  0,
in K, say, 1 ,  2 ,...,  n . Since trace is a non-zero homomorphism, so there exists an element α of K such
that SK/F(α) is non-zero. If αk is non-zero, we multiply the above system of equations by ααk-1 to obtain:
1 1 ( x1 )   2 1 ( x2 )  ...   n 1 ( xn )  0
1 2 ( x1 )   2 2 ( x2 )  ...   n 2 ( xn )  0
. . .
. . .
. . .
1 n ( x1 )   2 n ( x2 )  ...   n n ( xn )  0,
where  j   k1 j (j = 1, …, n). Applying the F-automorphisms 11 , 21 ,..., n1 to the above equations
respectively, to obtain
 11 ( 1 ) x1   11 (  2 ) x2  ...   11 (  n ) xn  0
 21 ( 1 ) x1   21 (  2 ) x2  ...   21 (  n ) xn  0
. . .
. . .
. . .
 n1 ( 1 ) x1   n1 (  2 ) x2  ...   n1 (  n ) xn  0,
Adding all these equations, as  i runs through the group G, so does  i1 . we deduce that
SK/F(β1)x1 + … + SK/F(βn)xn = 0.
Galois Field 51
As SK/F(βk) is a member of F and k   k1 k   , so SK/F(βk) = SK/F(α) is non zero, hence the set
x1 , x2 ,..., xn  is linearly dependent over F and so it does not form a basis, a contradiction to the
assumption. Hence the result follows.
3.3.2. Corollary. The collection 1 ( x), 2 ( x),..., n ( x) , images of an element x under the automorphisms
in the Galois group G( K , F )  1 , 2 ,..., n  , form a normal basis if and only if the matrix  i j ( x)  is
non-singular.
Next result proves that every separable normal extension of finite degree has a normal basis. However,
we will prove the result for an infinite field first.
Before starting the main result we are defining some terms:
1. If K is any field, then Pn(K) represents the collection of all polynomials in n indeterminates with
scalars from the field K.
2. If K is any field and f(x) is a polynomial over F, for   K , we define   ( f )  f ( ) . Further, if
f  Pn ( F ) , means it is a polynomial in n inderminates, say x1 , x2 ,..., xn , then for any n-tuple
  1 ,  2 ,...,  n  we can obtain   ( f ) by replacing xi with  i for 1  i  n .
3.3.3. Theorem. Let K be some extension of an infinite subfield F and f be a non-zero polynomial in
Pn(K). Then there are infinitely many ordered n-tuples   1 ,  2 ,...,  n  of elements of F such that
 ( f )  0 .
Proof. Mathematical induction on n is applied to obtain the required result.

For n = 1, let f(x) be a polynomial of degree d in P(K) = K[x]. Then f can have at most d roots in F (as
obtained earlier in Section - I), and so there are infinitely many elements in F which does not satisfy
f(x), that is, f ( )  0 or   ( f )  0 for infinitely many  in F.
Now assume that result holds for n = k, that is, if g is any polynomial in P k(K) then there are infinitely
many ordered k-tuples    1 , 2 ,..., k  of elements of F such that   ( g )  0 .
Consider n = k+1, and let f be any non-zero polynomial in Pk+1(K) = P(Pk(K)), so we may express f in
the form
f  g0  g1xk 1  g2 xk21  ...  gt xkt 1 ,
where g 0 , g1 , g 2 ,..., g t are polynomials in Pk(K). Since f is a non-zero polynomial, at least one of the
polynomials g 0 , g1 , g 2 ,..., g t must be non-zero, say, g i . According to the induction hypothesis, there are
infinitely many ordered k-tuples    1 , 2 ,..., k  of elements of F such that   ( gi )  0 . For each of
these k-tuples    1 , 2 ,..., k  , the polynomial
f     ( g0 )    ( g1 ) xk 1    ( g2 ) xk21  ...    ( gt ) xkt 1

is a non-zero polynomial in P(K). Now following the similar lines as for n = 1, we conclude that there
are infinitely many elements  of F such that   ( f  )  0 . But if we set    1 ,  2 ,...,  k ,   it is clear
that   ( f )    ( f  ) .
Hence we see that the result is true for n = k+1. This completes the induction.
3.3.4. Theorem. Let K be a finite separable normal extension of degree n over an infinite subfield F.
Let G( K , F )  1 , 2 ,..., n  be the Galois group of K over F. If f is a polynomial in Pn(K)
with indeterminates X 1 , X 2 ,..., X n such that, for every  K ,  ( ) ( f )  0, where,
 ( )  ( 1 ( ), 2 ( ),..., n ( )) then f is the zero polynomial.
Proof. Let  x1 , x2 ,..., xn  be a basis for K over F. Then, due to Theorem 1, the matrix  i ( x j )  is non-
singular, and so is invertible with inverse, say,  pij  . Thus,  i ( x j )  pij   I n and so the (i , r)th entry of
this matrix are
n
1, if i  r
 ( x ) p 
0, if i  r
i j jr
j 1
n
Let i   i ( x j ) X j   i ( x1 ) X1   i ( x2 ) X 2  ...   i ( xn ) X n and    1 , 2 ,..., n  . Then, define the
j 1
polynomial g in Pn(K) as
g ( X1 , X 2 ,..., X n )    ( f ) .
If a   a1 , a2 ,..., an  is any ordered n-tuple of elements of F and   a1 x1  a2 x2  ...  an xn , then
 n n n 
 a ( g )  g (a1 , a2 ,..., an )  f   1 ( x j )a j ,  2 ( x j )a j ,...,  n ( x j )a j 
 j 1 j 1 j 1 
 n n n 
 f   1 (a j x j ),  2 (a j x j ),...,  n (a j x j ) 
 j 1 j 1 j 1 
 f  1 ( ), 2 ( ),..., n ( ) 
0
by given hypothesis.
n
Now, if b   b1 , b2 ,..., bn  be any ordered n-tuple of elements of F and c j   p jr br , for 1  j  n . Then,
r 1
b    i ( x j ) p jr br  bi ,
n n n n n
 ( x )c   ( x ) p
j 1
i j j
j 1 r 1
i j jr r
r 1 j 1
n
1, if i  r
since  ( x ) p  .
0, if i  r
i j jr
j 1
Galois Field 53
Hence if c   c1 , c2 ,..., cn  , then
 n n n 
 c ( g )  g (c1 , c2 ,..., cn )  f   1 ( x j )c j ,  2 ( x j )c j ,...,  n ( x j )c j 
 j 1 j 1 j 1 
 f  b1 , b2 ,..., bn 
 b ( f )
However,  c ( g )  0 as obtained above, so  b ( f )  0 for any ordered n-tuple b   b1 , b2 ,..., bn  of

elements of F. Thus f is the zero polynomial, otherwise it will contradict Theorem 2.
Remark. Let G( K , F )  1 , 2 ,..., n  be a Galois group of K over F. If  i , j  G( K , F ) , then
 i j  G( K , F ) and so it must be an element of 1 , 2 ,..., n  . We consider  i j   p (i , j ) . Since
G( K , F )  1 , 2 ,..., n  is a group so due to left and right cancellation laws,  i j   i k if and only if
j = k, that is,  p (i , j )   p (i ,k ) if and only if j = k, it follows that p(i, j) = p(i, k) if and only if j = k.
Similarly, p(h, j) = p(i, j) if and only if h = i.
We can now prove the Normal Basis Theorem for the case of infinite fields.
3.3.5. Theorem. Let K be a finite separable normal extension of on infinite subfield F. Then there exists
a normal basis for K over F.
Proof. Consider now the polynomial f in Pn(K) obtained by
 X p (1,1) X p (1,2) X p (1, n ) 

 
 X p (2,1) X p (2,2) ... X p (2, n ) 
f  det  .
 
X X p ( n ,2) X p ( n , n ) 
 p ( n ,1)
Then as discussed in the remark above Xi occurs exactly once in each row and exactly once in each
column of this matrix. If we replace ordered n-tuple ( X 1 , X 2 ,..., X n ) by (1, 0, …, 0) in f, we obtain the
determinant of a matrix in which the identity element 1 of F occurs exactly once in each row and exactly
once in each column; the determinant of such matrix is either 1 or –1. Hence f is a non-zero polynomial.
Due to Theorem 3, there is at least one element x of K such that
f 1 ( x), 2 ( x),..., n ( x)   0 .
By the definition of the polynomial f, this in term becomes
det  i j ( x)   0 .
Hence, by corollary to Theorem 1, 1 ( x), 2 ( x),..., n ( x) is a normal basis for K over F.
3.4. Cyclotomic Extensions. Let 𝐹 be a field, for every positive integer 𝑚 define
k m  X m 1
in 𝐹[𝑋]. If an extension K of 𝐹, is a splitting field of one of the polynomials 𝑘𝑚 , then it is called a

cyclotomic extension.
3.4.1. Theorem. Let F be a field with non-zero characteristic, then the cyclotomic extension is both
separable and normal.
Proof. Suppose that 𝐹 has non-zero characteristic 𝑝, then every positive integer 𝑚 can be expressed in
the form 𝑚 = 𝑝𝑟 𝑚1 , where 𝑟 ≥ 0 and 𝑝 does not divide 𝑚1 . Then we have
k m  X m  1   X m1  1  (km1 ) p , and so roots of 𝑘𝑚 are similar to those 𝑘𝑚1 . Thus splitting field of
pr r
𝑘𝑚1 over 𝐹 is also a splitting field for 𝑘𝑚 over 𝐹. Thus in this case we consider only those polynomials
𝑘𝑚 for which 𝑚 is not divisible by the characteristic. Then,
dk m
 mX m1
dX
The only non-zero factor of this polynomial are powers of 𝑋, none of which is a factor of 𝑘𝑚 . Thus, no
roots of 𝑘𝑚 are repeated and so 𝑘𝑚 is a separable polynomial. Also being a splitting field of some non-
zero polynomial this extension is normal too. Hence all cyclotomic extensions of 𝐹 are separable and
normal.
Remark. Let 𝐾𝑚 be a splitting field for 𝑘𝑚 over 𝐹 , where 𝑚 is not divisible by the characteristic of 𝐹.
Also assume that 𝐹 is contained in 𝐾𝑚 . As the 𝑚 roots of 𝑘𝑚 in 𝐾𝑚 are all distinct, we call them the
𝑚𝑡ℎ roots of unity in 𝐾𝑚 and denote them by 𝜉1 , … , 𝜉𝑚 . Now if 𝜉𝑖 and 𝜉𝑗 are 𝑚𝑡ℎ roots of unity in 𝐾𝑚 ,
we have (𝜉𝑖 𝜉𝑗 )𝑚 = 𝜉𝑖 𝑚 𝜉𝑗 𝑚 = 1 so 𝜉𝑖 𝜉𝑗 is also 𝑚𝑡ℎ roots of unity, therefore the collection of 𝑚𝑡ℎ roots
of unity form a subgroup of the multiplicative group on non-zero elements of 𝐾𝑚 . Further, being a finite
multiplicative subgroup of non-zero elements of a group this subgroup must be a cyclic group. Any
generator of this group is called a primitive 𝑚𝑡ℎ root of unity in 𝐾𝑚 . If 𝜉 is a primitive 𝑚𝑡ℎ root of unity,
then 𝜉 𝑟 is also a primitive 𝑚𝑡ℎ root of unity for each 𝑟, relatively prime to 𝑚.
If 𝑚 is a prime number, then every 𝑚𝑡ℎ root of unity, except the identity element, is a primitive 𝑚𝑡ℎ root
of unity. It is clear that any primitive 𝑚𝑡ℎ root of unity 𝜉 may be taken as a primitive element for 𝐾𝑚
over 𝐹, that is to say, 𝐾𝑚 = 𝐹(𝜉).
First we are to define the group 𝑹𝒎 .
The elements of 𝑹𝑚 are the residue classes modulo 𝑚 consisting of integers which are relatively prime
to 𝑚, with the product of two relatively prime residue classes 𝐶1 , 𝐶2 is defined to be the residue class
containing 𝑛1 𝑛2 , where 𝑛1 , 𝑛2 are members from 𝐶1 , 𝐶2 respectively. The order of 𝑹𝑚 by ∅(𝑚).
In the next theorem we will obtain the Galois group of a cyclotomic extension.
Galois Field 55
3.4.2. Theorem. Let F be a field, m a positive integer which is not divisible by the characteristic of F, if
ch.F is non-zero. Let K m be a splitting field for k m over F including F. Then the Galois group G(K m , F)
is isomorphic to a subgroup of 𝐑 m .
Proof. Let 𝜉 be a primitive 𝑚𝑡ℎ root of unity in 𝐾𝑚 . If 𝜏 is any element of 𝐺(𝐾𝑚 , 𝐹), then 𝜏(𝜉) is also a
primitive 𝑚𝑡ℎ root of unity. Hence 𝜏(𝜉) = 𝜉 𝑛𝜏 , where g.c.d.( 𝑛𝜏 , 𝑚) = 1. Define a mapping : 𝐺 → 𝑹𝑚
as follows:
𝜃(𝜏) = the residue class of 𝑛𝜏 modulo 𝑚.
If 𝜏 and 𝜌 are elements of 𝐺, then
𝜉 𝑛𝜏𝜌 = (𝜏𝜌)(𝜉) = 𝜏(𝜌(𝜉)) = 𝜏(𝜉 𝑛𝜌 ) = (𝜏(𝜉))𝑛𝜌 = 𝜉 𝑛𝜏 𝑛𝜌 ,
so 𝑛𝜏𝜌 ≡ 𝑛𝜏 𝑛𝜌 (mod 𝑚), and therefore 𝜃(𝜏𝜌) = 𝜃(𝜏)𝜃(𝜌). Hence 𝜃 is a homomorphism.
Further, 𝜃 is one-to-one, as if 𝜏 ≠ 𝜌 then 𝜏(𝜉) ≠ 𝜏(𝜉), that is, 𝜉 𝑛𝜏 ≠ 𝜉 𝑛𝜌 and hence 𝑛𝜏 and 𝑛𝜌 are
members of different residue classes modulo 𝑚.
Hence, 𝐺 is isomorphic to the subgroup 𝜃(𝐺) of 𝑹𝑚 .
3.5. Cyclotomic Polynomial. Let 𝐹 be an arbitrary field and 𝐾𝑚 a splitting field for 𝑘𝑚 over 𝐹
containing 𝐹, we assume that 𝑚 is not divisible by the characteristic of 𝐹 if ch.F is non-zero. If 𝑑/𝑚, the
polynomial 𝑘𝑑 = 𝑋 𝑑 − 1 divides 𝑘𝑚 = 𝑋 𝑚 − 1 and hence roots of 𝑘𝑑 are included among the 𝑚𝑡ℎ roots
of unity in 𝐾𝑚 , that is, there are 𝑑 distinct 𝑑 𝑡ℎ roots of unity among the 𝑚𝑡ℎ roots of unity and, in
particular, ɸ(𝑑) primitive 𝑑 𝑡ℎ roots of unity. Thus, for each divisor 𝑑 of 𝑚 we may define the
polynomial ɸ𝑑 in 𝑃(𝐾𝑚 ) as
ɸ𝑑 = ∏(𝑋 − 𝜉𝑑 ),
where the product is taken over all the primitive 𝑑 𝑡ℎ roots of unity 𝜉𝑑 in 𝐾𝑚 , then 𝑑𝑒𝑔ɸ𝑑 = ∅(𝑑). Since
every 𝑚𝑡ℎ root of unity 𝜉 is a primitive 𝑑𝑡ℎ root of unity for some 𝑑/𝑚, it follows that
𝑘𝑚 = 𝑋 𝑚 − 1 = ∏𝑑/𝑚 ɸ𝑑 .
The polynomial ɸ𝑚 is called the 𝑚𝑡ℎ cyclotomic polynomial.
3.5.1. Theorem. For every positive integer m, the coefficients of the mth cyclotomic polynomial belong
to the prime subfield of F. In case if ch.F = 0, and the prime field is 𝐐, then these coefficients are
integers.
Proof. Mathematical induction on 𝑚 is sued to obtain the result.
For m = 1, result is obvious as ɸ1 = 𝑋 − 1 has coefficients in the prime field.
Suppose now that the result holds for all factors 𝑑 of 𝑚 such that 𝑑 < 𝑚.
Then we have
𝑋 𝑚 − 1 = ɸ𝑚 ∏1≤𝑑<𝑚 ɸ𝑑 .
𝑑/𝑚
By hypothesis, all the factors in the product have coefficients in the prime field; 𝑋 𝑚 − 1 has coefficients
in the prime field. Hence so does ɸ𝑚 . In the case, when the prime field is 𝑸, every factor in the product
has integer coefficients with leading coefficient 1, when we divide a polynomial with integer
coefficients by a polynomial with integer coefficients and leading coefficient 1 the quotient has integer
coefficients. Thus ɸ𝑚 have integer coefficients.
3.5.2. Example. Compute ɸ20 .

Since the divisors of 20 are 1, 2, 4, 5, 10 and 20, so we have
𝑋 20 − 1 = ɸ1 ɸ2 ɸ4 ɸ5 ɸ10 ɸ20.
Similarly, the divisors of 10 are 1, 2, 5 and 10, so we have
𝑋10 − 1 = ɸ1 ɸ2 ɸ5 ɸ10 .
Hence 𝑋10 + 1 = ɸ4 ɸ20.
Now we need to calculate ɸ4 . For this, the divisors of 4 are 1, 2 and 4, so we have
𝑋 4 − 1 = ɸ1 ɸ2 ɸ4 .
Also, 𝑋 2 − 1 = ɸ1 ɸ2 .
So, we have ɸ4 = 𝑋 2 + 1.
𝑋 10 +1
Hence ɸ20 = .
𝑋 2 +1
3.6. Cyclotomic Extensions of the Rational Number Field.

In this section, we will consider that the field F = Q, field of rational numbers, and prove that the Galois
group G ( K m , Q ) is isomorphic to the multiplicative group Rm of residue classes modulo m relatively
prime to m.
3.6.1. Content of a Polynomial. Let f ( x)  0  1 x  2 x2  ...  n xn  Z[ x] be a polynomial over Z,
then the content ‘t’ of f is defined as t  g .c.d .(0 , 1 , 2 ,..., n ) .
3.6.2. Primitive Polynomial. A polynomial f ( x)  Z [ x] is said to be primitive polynomial if its content

is 1.
It should be noted that if f ( x)  Z [ x] , we may write f ( x)  cf1 ( x) , where c is the content of f ( x) and
f1 ( x ) is a primitive polynomial in Z [ x] .
3.6.3. Theorem. If a polynomial f ( x)  Z [ x] can be expressed as a product of two polynomials over Q ,

the rational field, then it can be expressed as a product of two polynomials over Z .
Proof. Let f ( x)  Z [ x] and g1 ( x), g 2 ( x)  Q[ x] such that f ( x)  g1 ( x).g 2 ( x) . Let d1, d2 be the least
common multiples of the denominators of the coefficients of g1 ( x), g 2 ( x) respectively. Then
Galois Field 57
p1 ( x)  d1 g1 ( x) and p2 ( x)  d 2 g 2 ( x) are polynomials in Z [ x] . Let t1 and t2 be the content of

p1 ( x) and p2 ( x) and write p1 ( x)  t1k1 ( x) and p2 ( x)  t2 k2 ( x) , where k1 ( x) and k2 ( x) are primitive
polynomials in Z [ x] . Then we have
d1d 2 f ( x)  t1t2 k1 ( x)k2 ( x) .
We claim that k1 ( x) k2 ( x) is a primitive polynomial.
Let p be any prime number. Since k1 ( x)  a0  a1 x  a2 x2  ... and k2 ( x)  b0  b1 x  b2 x2  ... are
primitive polynomials so each polynomial has at least one coefficient which is not divisible by p. Let ai
and b j be the first coefficients of k1 ( x) and k2 ( x) respectively, which are not divisible by p. Then the
coefficients of Xi+j in k1 ( x).k2 ( x) is

u  v i  j
au .bv .
If v ≠ i, u ≠ j and u + v = i + j, then either u ˂ i or v ˂ j and hence either au is divisible by p or bv is

divisible by p. Thus, all the terms, except for aibj, in the summation are divisible by p and so the sum is
not divisible by p. It follows that for every prime number p, k1 ( x).k2 ( x) has at least one coefficient
which is not divisible p, which implies that the g.c.d. of the coefficients of k1 ( x).k2 ( x) is 1. Hence
k1 ( x).k2 ( x) is a primitive polynomial.
t1t2
Thus, t1t2 is the content of (d1d2)f(x). However, d1d2 is a divisor of the content of (d1d2)f(x). Hence
d1d 2
is an integer, say, l. Then f ( x)  (lk1 ( x))k2 ( x) is a factorisation of f ( x) in Z [ x] .
3.6.4. Corollary. If f ( x)  Q[ x] is a monic polynomial dividing x m  1 , then f ( x)  Z [ x] .

3.6.5. Definition. If f ( x)  0  1 x  2 x2  ...  n xn  F[ x] and k is any positive integer, then we
denote by f k ( x) the polynomial obtained as
fk ( x)  0  1 xk  2 x2k  ...  n xnk  F[ x]
3.6.6. Theorem. Let f ( x)  Z [ x] divides x m  1 and k is any positive integer such that g.c.d.(k,m)=1,
then f ( x) divides f k ( x) in Z [ x] .
Now we will prove that the Galois group G ( K m , Q ) is isomorphic to the multiplicative group Rm of
residue classes modulo m relatively prime to m.
3.6.7. Theorem. Let Km be a splitting field of km over Q. Then G ( K m , Q)  Rm .
Proof. Let ζ be a primitive mth root of unity in Km. Define a monomorphism : 𝐺(𝐾𝑚 , 𝑄) → 𝑅𝑚 as
follows:
𝜃(𝜏) = the residue class of 𝑛𝜏 modulo 𝑚,
for each automorphism ԏ in 𝐺(𝐾𝑚 , 𝑄), we defined ԏ(ζ) = ζnԏ where nԏ is relatively prime to m.
This mapping is onto as well. Hence the required result holds.
3.6.8. Corollary. The cyclotomic polynomials ɸ𝑚 are all irreducible in Q[x].
3.7. Cyclic Extension. Let F be a field. A finite separable normal extension K of F is said to be cyclic
extension of F if G(K,F) is cyclic. We are considering that F  K .
3.7.1. Theorem. Let K be a cyclic extension of a subfield F and G( K , F )    . If x  K , then
y
N K / F ( x)  1 if and only if there is an element y  K such that x  , and S K / F ( x)  0 if and only if
 ( y)
there is an element z in K such that x  z   ( z ).
Proof. Since K is a finite extension of F so let [ K : F ]  n; then | G(K,F)|=n and so  n  I , the identity
automorphism.
y
First, suppose that x  . Then
 ( y)
y  ( y)  2 ( y)  n1 ( y)
N K / F ( x)  I ( x) ( x) ( x) 
2 n 1
( x)   1.
 ( x)  2 ( y )  3 ( y )  n ( y)
Similarly, if x  z   ( z ), we have
S K / F ( x )  I ( x )   ( x)   2 ( x)    n 1 ( x)
 z   ( z)   ( z)   2 ( z)   2 ( z)   3 ( z)    n 1 ( z )   n ( z )  0.
Conversely, suppose that
NK / F ( x)  I ( x) ( x) 2 ( x)  n1 ( x)  x ( x) 2 ( x)  n1 ( x)  1.
Then x is clearly non-zero and so is invertible with x 1   ( x) 2 ( x)  n 1 ( x).
Next, since the set of automorphisms {I , , , , n1} is linearly independent over K , the mapping
2
  x  x ( x) 2   x ( x)  n  2 ( x) n 1
is non-zero mapping of K into itself. That is to say, there is an element t of K such that
y  t  x (t )  x ( x) (t )   x ( x)  n2 ( x) n1 (t )

2
is non-zero. Applying the automorphism  , we obtain

 ( y )   (t )   ( x) 2 (t )   ( x) 2 ( x) 3 (t )   ( x) 2 ( x)  n 1 ( x)t  x 1 y.
Thus x  y /  ( y). Similarly suppose
SK / F ( x)  x   ( x)   2 ( x)    n1 ( x)  0.
Then of course  ( x)   2 ( x)    n 1 ( x)   x.
Galois Field 59
Since S K / F is not the zero mapping; so let t be an element of K such that S K / F (t ) is non-zero, and
consider the element
z1  x (t )  ( x   ( x)) 2 (t )   ( x   ( x)    n2 ( x)) n1 (t ).
Applying the automorphism  we obtain
 ( z1 )   ( x) 2 (t )  ( ( x)   2 ( x)) 3 (t )   ( ( x)   2 ( x)    n1 ( x))t
  ( x) 2 (t )  ( ( x)   2 ( x)) 3 (t )   xt.
Hence we have
z1  ( z1 )  x(t   (t )   2 (t )    n1 (t ))  xSK / F (t ).
Since S K / F (t ) lies in F and hence is left fixed by  , it follows that if we write z  z1 / S K / F (t ), then
x  z   ( z ).
3.7.2. Definition. Let a be any element of a division ring D. Then the normaliser of a in D is the set
N(a) consisting of elements of D which commute with a:
so n belongs to N(a) if and only if an = na.
3.7.3. Exercise. Let D be a division ring. Then the centre Z of D is a subfield of D and the normalizer of
each element of D is a division subring of D including Z.
3.7.4. Wedderburn theorem. Every finite division ring is a field.
Proof. Let D be a finite division ring, with centre Z. Suppose Z has q elements and D has qn elements.
We claim that D = Z and n = 1.
The multiplicative group D* can be expressed as a union of finitely many conjugate classes, say
qn 1
C1,…,Ck, w.r.t. the subgroup Z*. Then, |Ci| = where ti < n . Thus,
q ti  1
qn 1
k
q  1  q  1   ti
n
.
i 1 q  1
X n 1
Now the nth cyclotomic polynomial Φn in P(Q) is a factor of both the polynomials Xn - 1 and .
X ti  1
qn 1
Let a = Φn(q). Then a divides qn - 1 and . Hence a divides q – 1.
q ti  1
If n > 1, then for every primitive nth root of unity ζ in the field of complex numbers C we have
|q - ζ| > q – 1. Hence |a| = ∏| q - ζ |> q – 1, and hence a cannot be a factor of q – 1.
It follows that there is no conjugate class Ci containing more than one element. Hence n = 1 and D = Z,
as required.
3.7.5. Corollary. If F is a finite set, then it is a division ring if and only if it is a field.
1. Design fields of order 27, 16, 25, 49.
2. Compute ɸ30 .
3.9. Summary.
In this chapter, we have derived results related to cyclotomic extensions and cyclic extensions. Also It
was proved that a finite division ring is a field, therefore we can say that a division ring which is not a
field is always infinite.
Books Suggested:
Ruler and Compass Construction
4
Structure
4.1. Introduction.
4.2. Ruler-and-compasses constructions.
4.3. Solution by radicals.
4.4. Solvable Group.
4.5. Solution of Polynomial Equations by Radicals.
4.7. Summary.
4.1. Introduction. In this chapter, possibility to construct some geometrical figures using ruler and
compass are discussed by the help of some algebraic structures. Also the solvability by radicals of
generic polynomial is discussed
(i) Normal Extensions.
(ii) Fixed Fields, Galios Groups
(iii) Norms and Traces.
4.1.2. Keywords. Normal Extensions, Galois Group, Fixed Fields.
4.2. Ruler-and-compasses constructions.
Three main problem of Geometry are:
Using the traditional geometrical instruments ruler and compasses can we
1. Trisect an arbitrary given angle.
2. Construct a cube having volume double to that of a given cube.
3. Construct a square with area equal to that of a given circle.
We shall show that these three problems are insolvable.
Consider the Euclidean plane and two straight lines intersecting at right angles in this plane meeting at a
point O . Assume I is an arbitrary point on one of those lines. Then, by taking O as origin and I to be
the point (1,0) , we can set up a Cartesian coordinate system in the plane. Let B be a collection of
points in this plane, including O and I . With the points in B we can start our construction and so these
points will be called basic points.
By ruler-and-compasses construction based on B we mean a finite sequence of operations of the
following types:
(1) Drawing a straight line through two points which are either basic points or points previously
constructed in the sequence of operations.
(2) Drawing a circle with center at a basic point or a point previously constructed with radius equal
to the distance between two points, each of which is either a basic point or a point previously
constructed.
(3) Obtaining points of intersection of any two obtained in (1) and (2), which are (a) points of
straight lines, (b) pairs of circles, (c) straight lines and circles.
Any point P which is obtained by (3) based on B is said to be constructible from B. If B consists of
the points O and I and no others, we simply say that B is constructible.
Let P be any point of the plane with coordinates ( ,  ) determined by O and I . The subfield of R
obtained by adjoining  and  to B will be denoted by B( P ).
4.2.1. Theorem. If the point P is constructible from B , then the [B( P ) : B] = 2n for some non-
negative integer n.
Proof. To obtain P from B in ruler-and-compasses construction let the sequence is P1 , P2 , , Pn  P of
operations of type (3). Suppose that P1 is one of the basic points and the co-ordinates of Pi (i  1, , n)
be ( i ,  i ) .
Let K = B( P1 , , Pn ). We claim that [K : B] = 2n. Then the result follows directly as B( P ) is a subfield
of K and hence [B( P ) : B] is a factor of [K : B].
We prove by induction on n.
If n  1 , then K  B( P1 ) = B, thus [ K : B]  1  20.
Now assume result holds for n = k-1, that is, if L is the subfield of R obtained by adjoining to B the
coordinates of P1 , , Pk 1 then [ L : B]  2 s for some s.
Ruler and Compass Construction 63
If Pi and Pj are distinct points (1  i, j  k  1) then the equation of straight line ij joining them is
( j  i )( y  i )  ( j  i )( x  i ).
Similarly, if Pr and Ps are distinct points and Pt is any point (1  r, s, t  k  1), then the equation of
circle trs , with center Pt and radius equal to the distance between Pr and Ps is
( x  t )2  ( y  t )2  (r  s )2  (r  s )2 . (1)
Let T = B ( P1 , , Pk ) =L( Pk ). If Pk is obtained from P1 , , Pk 1 by intersection of two lines like ij ,

then its coordinates are obtained by solving two linear equations with coefficients in L and so its
coordinates lie in L Thus, T = L and so [ L : B]  [T : B]  2 s .
Similarly, in other cases [T : B]  2t for some t (Left as an exercise to the reader).
This completes the Proof.

4.2.2. Theorem. Let P be a point in the plane and B( P) has a sequence of subfields,
B( P)  K n , K n 1 , , K1 , K 0  B such that K i includes K i 1 and [ Ki : Ki 1 ]  2(i  1, , n), then P is
constructible from B.
Proof. We proceed by induction on n.
If n  0 then B( P)  B. Hence, P is constructible from B. Now, let result holds for n = k-1.
Assume that K has a sequence of subfields K  K k , K k 1 , , K1 , K 0  B . Since [ K k : K k 1 ]  2, it

follows that K k is a normal extension of K k 1 . If   K k such that   K k 1 , then K k  K k 1 (  ) . If
1 1 1
minimum polynomial of  over K k 1 is X 2  aX  b  ( X  a)2  (b  a 2 ) . Considering     a ,
2 4 2
1
we have  2  a 2  b  0; thus  2 is a positive element of K k 1 and clearly K k  K k 1 (  )  K k 1 ( ) .
4
Now, since ( 2 , 0) has coordinates in K k 1 , it is constructible from B, by the induction hypothesis.

Hence every point with coordinates in K k is constructible from B. This completes the induction.
4.2.3. Corollary. Let P be a point in the plane. If the field B( P) is a normal extension of B such that
[B( P) : B] is a power of 2 , then the point P is constructible from B.
Proof. Let G be the Galois group of B( P) over B , Then |G| = [ B( P) : B ] = 2 s. Then, G has a
sequence of subgroups, G  A0 , A1 , A2 ,..., An  {e} each of index 2 in the preceding. Thus, B( P) has a
sequence of subfields B( P)  K 0 , K1 , K 2 ,..., K n  B each of degree 2 over the next. Hence P is
constructible from B.
4.3. Solution by radicals.
Let F be a field of characteristic zero and E is an extension of F, then E is said to be an extension of F
by radicals if there exists a sequence of subfields F = E0 , E1 , … , Er-1, Er = E such that
Ei+1 = Ei(αi),
for i = 0 , …, r -1, where αi is a root of an irreducible polynomial in P(Ei) of the form Xni - ai. A
polynomial f(x) in F[x] is said to be solvable by radicals if the splitting field of f(x) over F is contained
in an extension of F by radicals.
4.3.1. Theorem. Let F be a field of characteristic zero, K a normal extension of F with G(K,F) is
abelian. If [K : F] = n and the polynomial kn = Xn - 1 splits completely in F[X], then K is an extension of
F by radicals.
Proof. Let G = G(K,F). Then, G may be expressed as a direct product of cyclic groups, say
G = C1 x … x Cr.
Define, Gi = C1 x C2 x … x Cr - i, for i = 0, …, r – 1, and Gr =< I >, where I is the identity element of G.
Then Gi+1 is a normal subgroup of Gi and
Gi
 Ci for i = 0, …, r - 1.
Gi 1
Let Ei be the subfield of K left fixed by Gi for i = 0, …, r. Then, Ei + 1 is a normal extension of Ei with
cyclic Galois group, isomorphic to Cr-1 for i = 0, …, r-1. Since the degree ni of Ei + 1 over Ei is a factor of
n and kn splits completely in F[X] and hence in Ei[X], it follows that kn splits completely in Ei[X]. So
Ei+1 = Ei(αi) where αi is a root of an irreducible polynomial in Ei[X] of the form Xni - ai for i = 0, …, r -1.
Thus K is an extension of F by radicals, as asserted.
4.3.2. Theorem. Let F be a field of characteristic zero. For every positive integer n, the polynomial
kn = Xn – 1 in F[X] is solvable by radicals.
Proof. We prove the result by induction on n.
If n = 1, then the splitting field for kn over F is F itself, which is an extension of itself by radicals.
Now, suppose that every polynomial kl with l < m is solvable by radicals.
Let Km be a splitting field of km over F containing F. If [Km : F] = r, then r ≤ φ(m) < m. According to
induction hypothesis, kr is solvable by radicals and so there is a splitting field Kr of kr over F which is
contained in an extension E of F by radicals. Without loss of generality assume that E and Km are
contained in the same algebraic closure C of F, then consider L = E(Km)  C.
Then, L is a separable normal extension of E and the Galois group G(L, E) of L over E is isomorphic to
a subgroup of the Galois group G(Km, F) of Km over F. Hence G(L, E) is Abelian. It follows that
s = [L : E] is a factor of r = [Km : F]. Since kr splits completely in E[X], so too does ks. Thus L is an
extension of E by radicals. Since E is also an extension of F by radicals it follows that L is also an
extension of F by radicals and hence km is solvable by radicals.
This completes the induction.
Ruler and Compass Construction 65
Before proceeding further, we discuss some results of solvable groups.
4.4. Solvable Group. A group G is said to be solvable if there exists a sequence of subgroups
G = G0  G1  G 2  . . .  G n =  e 
such that (i) G i1  G i for 0  i  n – 1
(ii) G i G i1 is abelian for 0  i  n – 1.
Results.
1. Every subgroup of a solvable group is solvable.
2. Every quotient group of a solvable group is solvable.
3. Let G be a group and H be a normal subgroup of G. Then if H and G H both are solvable, then
prove that G is also a solvable group.
4. A finite p-group is solvable.
5. Direct product of two solvable groups is solvable .
6. Let H and K are solvable subgroups of G and H  G then HK is also solvable.
7. Show that every group of order pq is solvable where p , q are prime numbers not
necessarily distinct.
8. Prove that every group of order p2q , p and q are primes , is solvable .
9. Sn is solvable for n  4.
10. Sn is not solvable for n > 4.
11. If a subgroup G of Sn (n > 4) contains every 3 – cycle and H be any normal subgroup of G
such that G H is abelian then H contains all the 3 – cycles.
12. Homomorphic image of a solvable group is solvable.
13. A finite group G is solvable iff there exist a sequence of subgroups
G = G0  G1  . . .  Gn = < e >
such that Gi+1  Gi and G i G i+1 is cyclic group of prime order for 0  i  n.
14. A group G in is solvable iff G (n) = < e > for some n  0.
15. An is not solvable for n  5 and hence Sn is also not solvable for n  5.
We now state a criterion for a polynomial to be solvable by radicals.
4.4.1. Exercise. Let F be a field of characteristic zero. A polynomial f(x) in F[x] has splitting field over
F with a solvable Galois group iff f(x) is solvable by radicals.
4.5. Solution of Polynomial Equations by Radicals.
An extension field K of F is called a radical extension of F if there exist elements 1 ,  2 ,...,  m  K such
that
1. K  F (1 ,  2 ,...,  m )
2. 1n1  F and ini  F (1,2 ,...,i 1 ) for i  1, 2,..., m and integers n1 , n2 ,..., nm
For f ( x)  F[ x] the polynomial equation f(x) = 0 is said to be solvable by radicals if there exists a
radical extension K of F that contains all roots of f(x).
If now {x1 , , xn } is asubset of a field E algebraically independent over the subfield F of E , the
polynomial
gn  X n  x1 X n1  x2 X n2   (1)n xn
in P( F ( x)) is called a generic polynomial of degree n over F . So a generic polynomial over F is one
which has no polynomial relations with coefficients in F connecting its coefficients
4.5.1. Theorem. Let gn  X n  x1 X n1   (1)n xn be a generic polynomial of degree n over a field
F of characteristic zero. Then the Galois group of any splitting field of g n over F ( x1 , , xn )  F ( x) is
isomorphic to the symmetric group on n digits. (Left as an exercise for students)
4.5.2. Theorem. The generic polynomial of degree n  5 is not solvable by radicals.
Proof. Since the Galois group of any splitting field of g n over F ( x1 , , xn )  F ( x) is isomorphic to the
symmetric group Sn,. But Sn is not solvable group when n  5 . Hence f(x) is not solvable by radicals
over F ( x1 , , xn )  F ( x) when n  5 .

1. Design fields of order 27, 16, 25, 49.
2. Compute ɸ30 .
4.7. Summary.
Constructing a cube having volume double to that of a given cube is equivalent to the construction from
the basic points O and I of the point ( , 0), where  is the real number such that  3  2. Since the
polynomial X 3  2 is irreducible in P(Q) , the field Q( ) has degree 3 over Q and hence, since 3 is
not a power of 2, the point ( ,0) is not constructible from O and I . Constructing a square with area
equal to that of a given circle is equivalent to the construction of the point (  ,0). However,  is not
algebraic over the field of rational numbers. Hence (Q( ) : Q) is infinite and hence cannot a power of 2.
Books Suggested:

Theory of Field Extensions

Uploaded by

Copyright:

Available Formats

Theory of Field Extensions

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Theory of Field Extensions

Uploaded by

Copyright:

Available Formats

Master of Science (Mathematics) (DDE)

THEORY OF FIELD EXTENSIONS

DIRECTORATE OF DISTANCE EDUCATION

Copyright © 2020, Maharshi Dayanand University, ROHTAK

CHAPTER SECTION TITLE OF CHAPTER PAGE NO.

1 1 Extension of Fields 1-27

2 2 Galois Theory 28-43

3 3 Galois Fields 44-60

4 4 Ruler and Compass 61-66

1.3. Extension of a Field.

1.4. Minimal Polynomial.

1.5. Factor Theorem.

1.6. Splitting Field.

1.7. Separable Polynomial.

1.8. Check Your Progress.

Note. (i) Every field is an extension of itself.

If [K : F] = n <  , then we say that K is a finite extension of F of degree n

Also, we have [K : F] = 1 iff K = F and [K : F] > 1 iff K  F . [F K]

Solution. We prove it by contradiction. Let, if possible, [R : Q] = n(finite).

0  1  2 2  ...  n n  0

Thus,  satisfies the polynomial 0  1 x  2 x2  ...  n xn . So,  is not a transcendental number,

Consider the polynomial f  x   xi  x j  F[ x] and f     i   j  0 .

Thus,  satisfies f  x   F[ x] and hence  is algebraic over F.

0  1  2 2  ...  n n  0

Thus,  satisfies the polynomial f  x   0  1 x  2 x 2  ...  n x n . So,  is algebraic over F.

Hence every finite extension is an algebraic extension.

1.4. Minimal Polynomial. If p  x  be a polynomial over F of smallest degree satisfied by  , then

1.4.1. Lemma. If p  x   F[ x] be a minimal polynomial of  and f  x   F[ x] be any other

If r  x   F [ x] is a non-zero polynomial, then it is a contradiction to minimality of p(x), since

deg p  x   deg q  x  ---(1)

Also, q(x) is a minimal polynomial of  , so q  x  p  x  . Thus,

deg q  x   deg p  x  ---(2)

By (1) and (2), degp(x) = degq(x). Hence

1.4.4. Irreducible Polynomial. A polynomial f  x   F[ x] is said to be irreducible over F if

1.4.5. Proposition. Minimal polynomial of any element is irreducible over F.

Let f (a)  F[a] where f  x   0  1 x  2 x 2  ...  n x n  F[ x] . Clearly,

f  a   0  1a  2 a 2  ...  n a n  F (a)

Thus, F[a]  F (a) .

Now,Consider the elements 1, a, a2,…, an in F(a).

Thus, a satisfies the polynomial f  x   0  1 x  2 x 2  ...  n x n . So, a is algebraic over F.

Hence a is algebraic over F.

We claim that [F(a) : F] = n.

Consider t  F[a] . Then, t = f(a) for some f  x   F[ x] .

If t  0 , then f  a   0 , that is, f(x) is not satisfied by ‘a’. Thus, p  x  | g  x  .

Since p(x) is irreducible in F[x] and f  x   F[ x] such that p  x  | f  x  .

Put x = a, 1  f  a  g  a   p  a  h  a   1  f a g a

f(x) = p(x)q(x) + r(x) where either r(x) = 0 or degr(x) < degp(x).

Let, if possible, the set 1, a, a 2 ,..., a n 1 is linearly dependent.

 0 1a  2a2  ...  n1an1  0

Let  y1 , y2 ,..., yn  be a basis of K over F.

Now,  i  K and  y1 , y2 ,..., yn  be a basis of K over F, so

Since  i  K and  x1 , x2 ,..., xm  are L.I. over K. Thus,  i  0 for i  1, 2,..., m .

Thus, ij  0 for i  1, 2,..., m, j  1, 2,..., n .

Proof. Let F  E  K and a  K is algebraic over F. Thus, there exists a polynomial

Since f  x   F[ x] and F  E  F  x  E  x  f  x   E  x  and f (a)  0 .

Remark. Let K / F be any field extension, then

Now, 1  K is algebraic over F. So by above remark, we have

Put F0 1   F1 , then  F1 : F0    .