.

Lesson' 0 PRELIMINARIES

OBJECTIVES
After reading this lesson, you should be able to
• define the concept of a group and give certain examples
• state and prove the Lagrange's theorem on subgroups of a finite group
• define the notion of a normal subgroup of a group and the corresponding quotient
group
• give certain examples of cyclic groups and their generators.

STRUCTURE
,
0.1 Introduction

0.2 Groups
0.3 Subgroups
0.4 Lagrange's theorem

0.5 Normal subgroups and quotient groups

0.6 Cyciic groups

0.7 Summary
0.8 Technical Terms
0.9 Answers to Self Assessment Questions

0.10 Reference Books

0.1 INTRODUCTION:
In this lesson we study one of the most important alqebric concepts, that of a group. A
group is a nonempty set on which a law of composition is defined such that all elements have
inverses. For example the set of all nonzero real numbers forms a group under multiplication and
the set of all real numbers forms a group under addition. The set of all invertible n x n matrices of
reals is an important example in which the law of composition is' matrix multiplication. Thus the
concept of a group and the axioms which define it have a naturality about them.

0.2 GROUPS:

For any nonempty set A , a mapping f: A x A ~ A is called a binary operation on A . That

.;:.-!.
 Centre for Distance Education 2 Acharya Nagarjuna University
---------------
is, for any two elements a and b in A there is a unique element associated in A and that unique
element will be denoted by f (a, b) or ajb . Usually, binary operations will be denoted by symbols

like +,',0, *, etc. If we say that * is a binary 'operation on A , this means that, for any elements a
and b , in this order, in A , there is another element denoted by a * b in A . The usual addition' + '
and multiplication. On the set IR of real numbers, the composition '0' of mappings on the set of
mappings of a given set into itself, the set intersection or the set union on the set of subsets of a
given set are familiar examples of binary operations.

0.2.1 Definition: A binary operation * on a set A is called associative if a * (b * c) = ( a * b) * c for

all a, b, c EA. A set A together with an associative binary operation on A is called a semi group.

The addition' + ' and the multiplication. On the real number system IR are associative,
while the binary operation defined by a * b=a-lrisnotassociative, since 2-(3 -1)*( 2-3) -1. A
pair (A, *) is a semigroup if A is a nonernptyset and * is ~sociative binary operation on A .
~ ~- ---. -

Now, (IR, +) and (JR,.). are semigroups, while (\R';-) is~migroup.

--=:::::.....

0.2.2 Definition: Let (A, *) be a semigroup. An element e in A is called an identity element if

a*e=a=e*a for all aEA.

If a*e=a for all a E A, then e is called a right identity and, if e v a=a for all a E A, e is /
called a left identity. If e is aright identity and e' is a left i~ntity in a semigroup .(A, *) then
'''',
e'· = e' *e= e
Therefore, it follows that a semigroup can have atmosf one identity. The real number 0 is
/

the identity in (JR, +) and the real number 1 isih~ identity in (JR, .). If JR+ denotes the set of

posinve rear/numbers, then (IR+, +) is a semigrotlp)wlll1out Identity. A semigroup having identity is

called a monoid.

0.2.3 Definition: Let (A, *) . be asemiqroup with identity e. An element a in A is said to be

invertible if there exists an element b in' A such that a*b=e = b*a .
*
If a b = e , then b is called a right inverse of a and a is called a left inverse of b . If b is a
right inverse of a and b' is a left inverseof the same element (that is, a * b = e and b' * a = e), then-

./
~ Lesson O:)I~~~~~~~~~~~~J~~l~-~-~~g~~~~~~~~~~CAlgebra
-' ---- I

and hence a is invertible. This implies that if a is invertible, there exists unique b such that
a * b = e = b * a and this unique b is called the inverse of a and is denoted by a -1 .

For example, in the semigroup (JR, +) , every element is invertible (for any a E JR, -a is the

inverse of a). In the semigroup (JR.,.), 0 is not invertible and every non-zero element is invertible.

In the semigroup (JR.+ U {O}, +) , the identity element 0 is the only lnvertible element. Note that the

identity element, if it exists, in any semigroup is invertible.

0.2.4 Definition: A semigroup (A, *) with identity in which every element is invertible is called a

group. That is, a pair (A, *) is called a ~ if A is a nonempty set and * is a binary operation on

A satisfying the following:

(i) a*(b*,c)=(a*b)~c for all a.b.c eA

(ii) There exists e E A such that 'a*e=a=e*a for all a EA.

(iii) For each aEA, there exists a-I EA such that a*a-1=e=a-1*a

0.2.5 Examples:

(i) (JR, +), (Q, +) and (Z, +) are all groups, where + is the usual addition on the set ]R of
real numbers, on the set Q of rational numbers and on the set Z of integers.

(ii) (JR.,.) is not a group, since 0 is not invertible. But if ]R' is the set of nonzero real numbers,
. Z\r
then (JR.', -) is a group.

(iii) Let X be any set and S (X) bethe set of all bijections of X.Dnto itself. Then (S (X), 0)
is agroup, where 0 is the .composition of mappings.

(iv) For any set X, let M (X) be the set of all mappings of X into itself. Then (M (X), 0)
is a semigroup with identity (where the identity mapping is the identity element), butnot a

group (unless X is a single element set). An element f EM (X) has a right inverse if

and only if f is a surjection (that is, onto map) and f has a left inverse if and only if f
is an injection (that is, one-one map)

(v~ Let A be the set of all 2 x 2 matrices over the real numbers. Then (A, +) is a group,
\
 Centre for Distance Education 4 Acharya Nagarjima Univers~

where + is the usual addition of matrices. If A' is the set of all non-singular matrices,
then (A',.) is a group, where- is the usual multiplication of matrices.

(vi) Let n be any positive integer and Zn = {O,1,2, ..., n -I} .Define

a+b if a+b-cn
a+b= {
a+b-n if a+b> n

Then (Zn' +n) is a qroup, which is called the additive group of integers modulo n.

0.2.6 Theorem (Canellation Laws) : Let (G, *) be a group and a, b, C E G. Then

a* b=a* c=:>b=c

and b * a = c * a =:> b =c
Proof:

=> e * b = e * c, where e is the identity element

=:>b=c
JIJ(\

Similarly, by applyi'ijg a-I from right, we get that

The following two results are easy to prove and are left for you as excercises.

0.2.7 Theorem: l,et (G, * ) be a group. Then the following are true.

(1)/ the identity element in Gis unique.

(2) for any a E G, the inverse of a in G is unique.

0.2.8 Theorem: Let (G, * ) be a group and a,EE G. Then the following hold.
 C Algebra E

(3) a» x =b has a unique solution in G

(that is, there exists unique x E G such that a * x = b )

(4) y=a+b has a unique solution in G.

0.3 SUB GROUPS:

If (G, *) is a group and a, bEG, then we simply write a b for a * b. For simplicity, we

supress the symbol * which denotes the binary operation. Accordingly, we simply say that G is a
group, when there is no ambignity about the binary operation with which G is a group. If G is a
group with respecfto more than one binary operation then we mention the operation also.

0.3,1 Definition: Let G be a group. A nonempty subset H of G is called a subgroup of G if

ab -1 E H for all a, b E H

It can be easily verified that a nonempty subset H of G is a subgroup of G if and only if

As a consequence, we get that every subgroup contains the identity of the group and contains
the inverse of every one of its elements. Therefore a subgroup of.({ji\:is a group on its own with
respect to the operation on the group G .

0.3.2 Theorem: Let H be a nonempty finite subset of a group G . Then H is a subgroup of G if

and only if ab E H for all a, b E H .

Proof: Since H is finite, we can write H ={al,a2, ....,an} for some positive integer n . If H is a
subgroup, then

~eEH,aEH and bEH

=> b-1 =eb-1E H and a E H

I
 \-

w. Centre for Distance Education 5 Acharya Nagarjuna University

Conversely suppose that a b « H for all a, bE H . Consider the set Hal = {aj adl ::;i < n} .
By hypothesis, Hal is a subset of H having the .same number of elements as in H (for

ajal =ajal ~ aj =aj by 0.2.6). Therefore Hal =H and hence al E H = Hal; thatis al =aj al for

some i . Now eal =al =aj al and hence e=ai EH = Hal , which again implies that e=aj al and

hence all = aj E H . Similarly, ail E H for 2::;i s;n . Therefore, a-I E H whenever a E H.

Thus H is a subgroup of G .

The above theorem fails if H is infinite. For, the set Z+ of positive integers is a subset of

the group (Z, +) satisfying the property that a + b e 71'+for a, b « Z+ ; but Z+ is not a subgroup of

(Z,+) ..
0.3.3. Theorem: Let G be a group. Then the intersection of any class of subgroups of G is again
a subgroup of G . The union of a class of subgroups of G may not be a subgroup of G .

Proof: You can easily prove that the intersection of subgroups is again a subgroup. Regarding
unions, consider

2Z = {2ajaEZ} and 3Z = {3ajaEZ}.

Then 2Z and 3Z are subgroups of the group (Z, +). But 2ZU3Z is not a subgroup,

since 3,2E~ZU3Z and 3-2\l2ZU3Z._~

0.3.4 Self Assessment Question :

Let Hand K be subgroups of a group G . Then prove that

(1) H UK is' subgroup of G if and only if either H sK or K sH

(2)The product HI<. = {abja E Hand b E,K} is a subgroup of G if and onlylflfK =KH .

0.3.5 Definition: Let G be a group and X be a subset of G . Then the intersection of all subgroups
of G containing X is the smallest subqroup of G containing X and is denoted by (X). If X
consists of only one element, say a, then we write (a) for ({a}). (X) is called the subgroup

generated by X. zr Is'called a generator for (a) and X is called a generating set for (X)
0.3.6 Theorem:

(1) For any nonempty subset X of a grou~ G,

(X)={al a2 an / for eachi, either aj E X or ail E X n ;:::1 }

(2) For any element a of a group G

(~)={an
------
In Z} E

where an is defined inductively by

e if n=O
I
a~ an-1 a if n>O
-.
- -~n--
--
(a-I). if n<O

Proof:
-I
(1) Let A={al «: a,)foreachf, either ai E X or ai-1 E X, n ;:::l}

It can be easily observed that XYE A whenever x,Y EA. Also, (al a2 an rI
=a~l a~~l ail all. Therefore A is a subgroup of G. Clearly X ~ A and, if H is any

subgroup of G such that X ~ H , then A <:;;;; H . Thus, A is the smallest subgroup of G containing

X and hence A=(X).

(2) This follows from (1) and-the definition of an for any n E 7L.
\

Q x=aE1 aEn, where each Ej =±1

--- :---
QX=~ .... +EnEZ
---- ,--:..-
.. -.
0.4 LAGRANGE'S THEOREM .
In this section, we shall prove one of the most celebrated, simple and most useful theorem
due to Lag:range. IfG is a finite group, then the number of elements in G is called the order of G
and is denoted by O( G). The Lagrange's theorem states that, if H is a subgroup of a finite group
G , then the order of H divides the order of G . Before going to the proof of this, first let us have the
following.

0.4.1 Definition: Let H be a subgroup of a group G. For any a E G, let

aH = {ah/h E H} and Ha={ha/h E H}.

aH is called the left coset of H corresponding to a in G and Ha 'is called the right coset
of H correspondinq to a in G .

0.4.2 Theorem: Let H be a subgroup of a group G . Then any two left(right) cosets of H in G
are either equal or disjoint. If H is finite, then the number of elements in a left (right) coset of H is
equal to O(H).

Proof: Let a.b e G: Suppose aHnbH:t:.rfJ. Then there exists x e ali Csbli , so that

a~ =x=bh2 for some ~,h2 E H. Then a-1b = Iqh21 E Hand b-1a=h2h1-1 E H . Now

YEaH => Y = ah for some h E H

and y.'ieIt'h'H => Y = bh for some h E H

Thus aH = bH . That is, if aH and bH are not disjoint, then aH = bH . On the same lines,
you can prove that Ha and Hb are either equai OJ disjoint. The map h H ah is a bijection of H
onto aH ·and, therefore, when H is finite, aH is also finite and H and'aH have the same number
of elements. Same argument is valid for right cosets also.

0.4':'3 Theorem (Lagrange's Theorem) : Let H be a subgroup of a finite group G. Then O( H)

divides O( G).
/

Proof: Each element a EGis in the corresponding left coset itH (since a = ae E all ). Therefore,
by theorem 0.4.2, the left cosets aH form a partition of G . Since G is finite, the member of left
~ Lesson 0) ~ Aigebra~

eosets of H in G is finite. Let

be all the distinct left co sets of H in G. Then

Therefore

n n
O( G)= ~ 14iHI = L O( H) = n O( H), since laiHI = O( H)
1=1 1=1

O(G)
Thus O(H) divides O(G) and O(H)=-n, the nurnberofleft co sets of H in G.

0.4.4 Corollary: If H is a subgroup of a finite group G , then the number of left cosets of H in G

" O(G)
is equal to the number of right eosets of H in G and this number is equal to O( H) ,

. O(G)
Proof :In the proof of 0.4.3, n is the number of left eosets of H in G and n = o( if:' . The same

arguement is valid, if we consider right cos=ts.

0.4.5 Oefinition : Let H be a subgroup of a group G. Then the lndex ot f! in G !';;'iefined as the
number of left (right) eosets of ;; in G and is denoted by [G: H) or l{, ( H) , if it i~'>inite, If G is a
/'

finite group, then

 Centre for Distance Education 10 Ac!larya Nagarjuna University

[G:Hl= 0(0)
O(H)

This definition may be extended to infinite groups. Let H be a subgroup of a group G (finite
or infinite). If fR is the set,of distinct left cosets of H in G and Z is the set of distinct left cosets of
II in G, then the cardinal number I~Iof the set [R is equal to the cardinal number 1.2'1of the set

9;', i.e. I~I= lo2'\, for the map IR ~ 02' given by Ha ~a-lll is a bijection since Ha=Hb

<;:;>ab-1E li<::::.' (a-1

r 1
b-l E H <:>a-1 H =b-l H . The index'of H in G, denoted by [G :H] is

the cordinal number of the set of distinct left cosets of H in G . It is true that lei = [G : H] [H]. The

index of H in G may be finite without G or H being finite. For, consider the group (Z, +) of

integers. Let n be a positive integer and H =nZ = {na/a E Z}. Then, you can check that

H, 1+ H, 2 + H, , (11-1) +H are all the distinctleft cosets of H and hence H is of index n in

Z·

0.4.6 Self Assessment Question: Let X = {a, b, c} and G be the set of all bijections of X onto

itself. Define f E G by f (a )=b , f( b)=a and f( c )=c. Prove that that H ={e,f} is a subgroup

of G , which is a group under the composition of mappings. Compute all the left and right cosets of
H in G and observe that. though their number is the same, they are different.

0.4.7 Theorem: Let H be a subgroup of a group G. For any a, b e G , define a relation by

Then ~ is an equivalence relation on G whose equivalence classes are preaisely the left
eosets of H in G.
Proof: Since H is a subgroup of G I H should contain the identity of G. For any a E.G , we have
a-1a=e and hence O > a. Therefore ~ is reflexive.

Therefore - is symmetric. Also,

 ( Algebra ~

a= b and b-:c ~a-lb and b--1CEH =>a-lc=(a-lb)(b--lc)EH:::)a~c and hence

., is transitive. Thus ~ is an equivalence relation on G . Further, for any aEG I

a= b <=> a-1b'ER<=> b e all

and hence aH is the equivalence class containing a with respect to the relation r-: •

0.4.8 Self Assessment Question: For any subgroup H of a group G , define a relation ~ on G

by a - b <=> ah-I E H .

Prove that - is 'an equivalence relation on G whose equivalence classes are precisely the right
cosets of H in G.
0.5 NORMAL SUBGROUPS AND QUOTIENT GROUPS :
Normal subgroups are a special kind of subgroups and these fecilitate the construction of
quotient groups. In 0.4.4 and 0.4.6, we have observed that a subgroup H of a group G will have
the same number of left cosets and of right co sets and still a left coset of H may not be a right
coset of H and vice-versa.

Now, we shall prove the following, where for any subsets A and B of a group G , AB stands .

for {aYaEA and bEB} and A-I for {a-IaEA}'

0.5.1 Theorem: Let H be a subgroup of a group G . Then the following are equivalent to
', ..
(1) aH ~ Ha for all a E G

(2) HacaH for all aEG

(3) aHa-1 =H for all a E G

(4) Every left coset of H in G is a right coset of H in G.

(5) Every right coset H in G is a left coset of H in G.

(6) The product of any two left cosets of H in G is again a left coset of H .

(7) The product of any two right cosets of H in G is again a right coset of H .

(8) (Ha )(Hb )==H (ab) for all a, bEG

(9) (aH)(hH)=(ab)JJ ;orall a.b e G

 Centre for Distance Education 12 Acharya Nagarjuna University

(10) For any a, bEG ab-I E H <;::.;> a--I/) E H

Proof:

(1)~(2): ForanYaEG,wehave a-1H~Ha-l (by(1))andhence

H =(Ha)a--1 c aHa-1 and there/ore aHa-1 =H

(3)~(4): Thisistrivial,since aHa i "-.::>a1-!=Ha

(4) ~ (5) : Consider a right coset Ha with a E G. By (4), the left coset aH must be equal to
a right coset; that is, aH= Hb for some bEG. Then a is a common element to
Hb as well as Ha . By theorem 0.4.2, Ha+Hb =aH Therefore, eV8'Y right coset
of H is a left coset of H .

(5)~(6): If Ha = bH, then aE(aH)n(bH) and hence, by 0.4.2, hH-=7H, so that

Ha=aH. Now, for any a.b e G,

Note that NH ~j./=H-l, since H is a subgroup.

(b).:.:...>(7): For any a,bEG,(b-1H)(a-1H)=xH for some xEH (by (6» and hence we

have

(7}~(8): Let a.b e G: The.i, by (7), (Ha)(Hb)=Hx for somo.:EG. Then

ab=( ea)( eb)E (Ha~ (Hb )=Hx and hence

(Ha)(Hb )=Hx=H (ab), again ~y 0.4.2.

 ( Algebra ~

(8) => (9): For any a, bEG, we have (Hh-1)( Ha-1) = H (b-1a-l) (by(8)) and hence

(9) => (10) : Let a.b EG . Suppose ab -1 E H . Then

bEHbH=((ab-I)H)(bH)=((ah-l)h)H =aH and hence a-1bEH . Similarly,

we can prove that a-1bE H =>ab-l E H·

(10)=>(1): ForanYaEG and hEH,wehave (ahrla=h-la-la=h-lEH and hence,

by (10), (ah)a-1 E H and therefore ah E Ha. Thus aH <:;;;; Ha ,

\
. \
0.5.2 Definition: A subgroup H of a group G is said to be normal if anyone (and hence all) of the
above conditions in 0.5.1 is satisfied.
If the operation on a group is commutative, then obviously every subgroup is normal. In
example 0.4.6, a subgroup which is not normal is given.

0.5.3 Theorem: Let N be a normal subgroup of a group G and % be the set of ali right cosets
of N in G; that is, % = {Naja E G}. Under multiplication of sets, % is a group.

Proof: For any subsets X and r or G, we have defined the product X y as

\ .

XY={xyjXE); andj c F].

Since N is a normal subgroup, we have (by 0.5.1 (8)),

(Na)(Nb) = N(ab)

for any a, bEG and hence the multiplication of sets is a binary operation on %, which is
clearly associative. Also, consider the element N =Ne E %. We have
(Na)N =N(aN)=N(Na)=Na, since aN =Na,

and therefore N is the identity element of %.Also, for any a E G ,

 Centre for Distance Education 14, Acharya Nagarjuna University

(Na)(Na-I)=N(aa~l)=Ne=N=Na-la = (Na-I)(Na) an~ hence Na-1 is the

inverse of Na in %. Thus % is a group.

0.5.4 Definition: For any normal subqroup N of a group G, the group % constructed in the

above theorem is called the quotient group of G by N or simply the quotient of G by N

0.5.5 Example: Let n be any positive integer and nZ = {nal a E-tJ ' Then nZ is a subgroup of the

group (Z, +) clearly nZ is a normal subgroup, since the operation + is co~utative,

For simplicity, let us write a to demote the coset nZ+a. Then you can easily check that
the quotient groUp

Z/'
/ nZ = {--0, 1, , -}
n-1
Also, the operation defined on the quotient group is as follows,

ifa+b<n
if a+b > n
for any 0 S a, b S; n -1. Compare this with the example given in O.2.5(vi).

:(L6 CYCLIC GROUPS :

For any element a in group G , we have constructed the subgroup (a) generated by a in
G (see 0.3.6). (a) is called a cyclic subgroup of G.

ttS.1 Definition: A group G is called a cyclic group if there exists a E G such that

In this case, G is said to be generated by a and a is a called a generator of G.

If + is the binary. operation on a group, it is conventional to write na for all ; that is

f 0, theidentity, if n=O
na=) (n-l)a+a if n>O
l(-n)(-a) ifn<O
0.6.2 Exampies :

(1) The group (Z, +) of integers ;s cyclic group, since Z == (1)== (-J). Here, both 1 and
w1 are generators for this cyclic group.
(2) For any positive integer n , consider the additive group Zn of integers modulo n
(see O.2.5(vi)). Recall that

Zn = {0,1,2,.. ,n-l}

Here the operation is addition modulo n , denoted by +n' Note that

1+/11 = 2, 2+nl=3, ,(n-2)+n l=n-l and

(n-l)+nl=O, O+n 1=1, .

1
~·~2
0" Il

n-l.
!. \•
•
•
j
•
.. ~
•

By adding 1to each element 0~a < n-l in Zn ' we are getting the next element a+n 1in

Zn and by adding 1 to n -1 we get O. This is the reason for calling it a cyclic group. Zn is a cyclic
group generated by 'I whose order is n .
0.6.3 Theorem: Let G be an infinite cyclic group. Then the following hold
.:UO ....

(1) x" =e<::>x=e or n=O, for any XEG

(2) For any x in G, x" =xm co n =m or x=e

(3) There are exactly two generators for G.
Proof: Since G is a cyclic group, there exist a E G such that

G=(a)={an In Z}.
E

We shall first prove that an '* e for all n '* O~Suppose, if possible an =e for some n oj:. 0,

We can assume that n is positive (since an =e if and only if a-n =e). Then, by the division algorithm,
any integer m can be written as m == nq + r for some q, r E Z with 0 s r < n and hence
 Centre for Distance Education 16 Acharya Nagarjuna University

-
r
This implies that G ={a /05. r < n} which is a contradiction for the hypothesis that G is

infinite. Thus an oF e for all n oF 0.

(1) - If x=e or n=O, then clearly x" =e . Conversely suppose x EG and x" =e . We can write
m
x=a for some m EZ. Now, amn «x" =e and hence, by the above observation, mn=O
so that m =0 or n = O. Therefore x = e or n = 0 .

(2) This follows from (1) and from the fact that x" =xm if and only if xn-m =e.

(3) We have that a is generator of G. Suppose b is another generator of G . Then

(a) = G =(b)

so that a=bn and b-s a" for some integers nand m . Then a' =a=bn=(amr and

therefore, by (2) mn = 1 which implies that m = 1= n or m = -1 = n . From this it follows that

m::::1 or -1 and hence b = a or a-I. Thus a and a-I are the only generators of G. Also,

note that a * a-I and (a) = \ a -1). Thus G has exactly two generators.

0.6.4 Theorem: Let G be a finite cyclic group of order n > 1 . Then the following hold
:, ,

(1) If a is a generator of G, then G={e,a,a2, ,an-1} and n is the least positive integer

such that an =e .
(2) For any generator a of Gand for any integer m ,

am =e~ n d~\tides m .

(3) The number of generators of G is equal to ¢ (n) , the number of positive integers less than
n and relatively prime to n .

Proof: (1) Let a be a g~nerator of G. The~ a oF e (since O( G» 1) and G=(a)~{am /m z}.

E

Since G is finite and Z is infinite, we should have am =at for some m < t ..Then at-m =e and
 C Algebra ~

t - m is a positive integer. Let s be the smallest positive integer such that as = e . Then, ai ~ aJ for

all 050i~j50s-1 and, G={e,a,a2, ..... ,aS-1} (seethe discussion in the begining of the proof of

0.6.3). Since O( G)=I1, it follows that s=n and G ={e,a, a2 , ..... ,an-1} .

(2) Let a be agenerator of G. Then, by (1), an =e and hence (an t =e for all integers q.

This implies that am = e whenever n divides m . On the other hand, suppose am = e . Then
we can write m=nq+r- for some q,rE7l, with 050r<n. Now,

e = am =anq+r =( an r a" =ear =ar . Since n is the least positive integer such that an =e

and since ar =e and O:s; r < n , it follows that r = 0 and m = nq . Thus n divides m .

(3) We shall prove that for any 0 < r < n , a" is a generator of G if and only if r is relatively

prime to n . Let 0 < r < 11. Suppose ar is a generator of G . Then

and hence a =( ar )S for some integer s , so that ars -1 = e. By (2) n divides rs -1 and

therefore nt = rs -1 so that rs - nt = 1 . From this it follows that

. ,rand n are relatively prime .
)

Conversely, suppose rand S are relatively prime. Then there exist integers sand t such

that rs=nt+) and hence a=ars-nt=(arf(anrt =(arrjwhichimplie~that aE(ar).

Therefore \ a) ~ (at") ~ G = (a) and hence G = ( J) so that at" is a generator of G . ~ISO

at" ~ as for any 0 < r, S < n . Thus, the number of generators of G is equal to the
note that
number of positive integers less than n and relatively prime to n .

0.6.5 Definition: Let G be a group and a E G. If there exists an integer n '* 0 such that an ~ e ,
then we say that the order of a is finite. In this case, if m is the least positive integer such that
am = e , then m is called the order of a, written 0 (a) . If no such integer n exists, then a is said to
be of infinite order.

0.6.6 Corollary: Let G be a group and a E G. Then

 (i) an=e for integer n:f:.O <=> O(a) divides n .

(ii) (a) is of order m <=>O(a)=m'

Proof: (i) folltsws from 0.6.4 (2), while (ii) follows fromO.6A (1).

0.7 SUMMARY:
In this lesson you have learnt about the concept of a group and certain examples of groups
have been given. Also you have been provided '(Viththe concept of a subgroup and elementary
properties of subgroups have been presented. The concept of a coset of a subgroup has been
introduced and famous Lagrange's theorem has been proved. Further you have learnt the concept
of a cyclic group and order of an element of a group and certain important properties of these have
been proved. '

0.8 TECHNICAL TERMS

Semigroup
Group
Subgroup
Subgroup generated by a set
Cosets of subgroups of a group
Index of a subgroup
Normal Subgroups
Cyclic groups
. Order of an element of a group
0.9 .I'J~:::·"VERSTO SELF ASSESSMENT QUESTIONS .
0.3.4 (1) Suppose H UK is a subgroup of G. Suppose also that H q;,Kand K q;,H. Then
three exist elements a E H - K and b e K - H . Now a, hE H UK. By hypothesis,

O,':~; abEHUK. If ab e H then b=a-1abEH, a contradiction. On the other hand if

ab K , then a=bb-1
E E K , again a contradiction. Therefore either H c K or K c;;,H
Converse is trivial.

(2) Let HK =KH. HK isnonempty, since e E HK. Let a,h E HK. Then a=f1tkt> b=h2k:

for some lIj,h2EH and kJ,k2EK. Hence ab-l=hjklki-lh2=hlYlhi', where

YI =kjki1 E K - Now y1hil E KH =HK. Hence y1hil =xIY2 for some h",~ E Hand

K],K2 E K . Now y\hi! E KH =HK. Hence y1h2'1 =xlJl2 for some Xl E H,Y2 E K~
 · )rl
.. :..--

Therefore an - i == hl XJ)~ :... '12Y2. where xl =hlXl E H. Hence ab-1 E HK. Thus HK
is a subgroup,
I
Conversely, suppose that HK is a subgroup. Let a E KH . Then a = kh , for some
I
k E K, hE H . Therefore a-1 :::::.h-1k-1 z; HK =:>a ~ HK . This KH s HK . Next, let

bE HK. Then b-1 E HK. Hence b-l =xy , for some

x E H, y E K ~ b = y- J X -I, E KH . Thus HK r;;;, KH '

0.4.6 If e is the identity mapping on X , then e is the identity element of G. Define o , g GO G by

cr(a)=h, a(h):::::.c & cr(c)=a land g(a)=a,g(b)=c and g(c)=b respectively, We can

verify that G ={e,cr, cr2, f, g .o f}. Since ]2 =e.H is closed under multiplication .. ', His

a subgroup of G. Note that H ={e,f}, aH == {a, af} and o-2H = {a2,g}, since

a2 J=« are the distinct left cosets of H in G. Furhter H =={e,f} ,

He ={(),g}, H cr2 ={cr2 ,af} , since Iv=s and 1cr2 =a f are the distinct right cosets

of H in G.
U.4.8 Proof is similar to that of OA.7

0.10 REFERENCES: :,.'1'

l.N. Herstein, Topics in Algebra, 2nd ed. Wiley, New York 1975
Michael Artin, Algebra, Prentice - Hall of India, New Delhi 1994
Thomas W. Hungerford, Algebra, Holt Rinehart and Winston, Ine 1974
FB.Bhattacharya, S.K. Jain, S,R. Nagpaul, Basic Abstract Algebra, 2nd ed. Cambridge
University Press, 1995.

Lesson Writer
Prof: U.M. Swamy
Andhra University, Visakhapatnam,
Lesson - 1 HOMOMORPHISMS

OBJECTIVES :
After reading this lesson, you should be able to
(i) define the notion of a homomorphism of groups and gives certain examples.

(ii) define the concept of the kernel of a homomorphism and prove certain elementary ./

properties of homomorphisms and their kernels.

(Hi) define the notion of an isomorphism of groups and prove certain elementary properties
of isomorphisms.
(iv) state and prove the fundamental theoram of homomorphisms.
(v) state and prove the Cauchy's and Sylow's theorems fer abelian groups.

STRUCTURE
1.1 Introduction
1.2 Definition and examples of homomorphisms
1.3 the kernel of a homomorphism

1.4 Isomorphisms

1.5 The fundamental theoram of homomorphisms

1.6 Cauchy's theorem and sylow's theorem

1.7 Summary
1.8 Technical terms

1.9 Model examination questions

v ,

1.10 Answers to Self Assessment Questions

I. '
',.J
1.11 Exercises
1.12 References

1.1 INTRODUCTION
A relationship between groups G and G' is gene-ally exhibited in terms of a structure
related mapping f: G -j- G' which are called hornornonhisms. SUCh a map often gives us
information about the structure of G' from known structuraproperties of G, or information about
the structure of G from known structural properties of G' The study of such structure-related
'L2.3 Example: Let G be the group of all real: .urnbers .mder the usual addition and G' be the
grol;lp of all non zero real numbers under usual multiplication. Define 1:G ~ 0' by

f (a) := Za for all aEG

Then I(a+b) = I(a) . j(b) for all a bEG

I and hence j is a homomorphism.

1.2.4 Example: Let (G, .) and (G' , *) be any groups. Define 1:G ~ d by

f (a) :::;e' for all aEG

Where e', is the identify element in G'. Then 1 is a homomorphism which is called the
~triviaihomomorphism.

1.2.5 Example: Consider the group Z of all integers under the usual addition let n be an arbitrarily
fixed integer. Define 1: Z ~ Z by 1(a) = na for all a E Z. Then f is a hornomorphlsm.

1.2.6 Example: Let G' = {I, - J \. Then G' is a group under the usual multiplication. Define

I: Z ~ G' by

- { 1 if a is even
f(a) = -1
if a is odd

Then 1(a + b) = f (a) f (b) for all a bEZ

I and hence f is a homomorphism.

1.2.7 Self Assessment

-
Questions
I
:
,,J

Prove that the mapping f in 1.2.6 above is a homomorphism.

1.2.8 Example: Let n be any positive integer and Zn be the additive group of integers module n .

Define f :Z ~ Z n by f (a) = r , where a :::qn + r, 0 ::; r < n .

Note that r is the remainder obtained by dividing a with n. Then f is a homomorphism.

1.2.9 Self Assessment Question: Prove that f is a homomorphism in 1.2.8.

1,2.10 Example: For any group G, the map f :G~ G. difined by 1(a) :::;
a for all a E G. is
a homomorphism called the identity homomorphism,

1.2~11 Example: let G be the group-of all 2x2 matrices (: !) over the real numbers for
~ Centre for Distance EducatioV=-=='_~E~~~~~~~(~A~C~h~ar2'ya Nagarjuna University);;;;

mappings from one algebraic structure to a similar algebraic structure is an important area in
Algebra. In this lesson, we shall introduce the concept of a homomorphism explicitly and study
certain important elementary properties of homomorphisms.

1.2 DEFINITION AND EXAMPLES OF HOMOMORPHISMS :

Let us recall that a group is a set toqether with a binary operation satisfying certain axioms.
A mapping between the underlying sets of two groups which preserves the operations of the groups
is called a homomorphism. More precisely, we have the formal definition of a homomorphism in
the following.

1.2.1 Definition: Let( G, -) and (d, *) be any two groups. A mapping f :G ~ d is called a

homomorphism of groups if

j(a.b) =: j(a) * j(b) for all a,b E G

In other words, the image of the product a- b is equal to the product of the images j (a)
and f (b) for any elements a and bin G.

(a,b)EC!xG ••
----~) G a,b

I1

I!
j l

ti'v G ------·~G'
l
(/. ' \:
fa)
j' j
f" /fJ \;
\')/
11----*-' ---'7) j' (a) * i (h)
The following exarnpels help you in understanding the concept of a homomorphism,
1.2.2 Example: let G be the group of all positive real numbers under the usual multiplication
,
and G' be the group of all real numbers under the usual addition. Define j :G ~ G by

f(a) =Iog , a foral! aEG,

lhell j(a· b) = log2 (a· b) = log2 a + log2 b = f(a) + f(b) for all a, bEG, Therefore

f is a homomorphism of groups,
 (ii) For any a E G, we have

j (a) j (a-I ) =j (a .a-I) =j (e) = e'

(Ill) For any a, bEG,

j (a) = j (b) <=> j (a) j (b) -1 := e'

<=> f (a) f (b -I ) = e'

and j(a) = j(b) <=> j(afl = j(bfl

c> j (a-I) = j (b-J ) <=> j (a -1b ) = e'

;-

Another important example of a homomorphism is given in the form of a theorem in the following

1.2.14 Theorem: Let N be a normal subgroup of a group G and % the quotient group. Define

j :G ~ % by j (a) = aN for any a E G. Then j is a surjective homomorphism.

Proof: Recall that the bin?,!)' operation in the quotient group is defined as aN . bN = (ab ) N for
! "

any a, b E G and hence'

j(a·b)= (a.b)N = aN ·bN = j(a) . j(b)

Therefore f is a ho~omorphism. Since any element of % is of the form aN = f (a) for

some a E G; j is a surjection too.

1.2.15 Self Assessment Question: If j: G ~ Hand g: H -4 K are homomorphisms of

groups, prove that g(~f :G -4 K is also a homomorphism.
 Centre for Distance Education 4 Acharya Nagarjuna Univer~ -~

which ad --be -:j:. 0 _ Then G is a group under the usual matrix multiplication. Let G' be the group
of non-zero real numbers under the usual mutiplication. Define f :G .---)G' by

j((a
led)
bJi == ad - be

Then f is a homomorphism.

1.2.12 Example; Let (G, .) be may group and a E G. Let (Z • +) be the group of all integers

where + is the usual addition. Define f : 7L ~ G by fen) =an for all n E 7L, where an is
de~~nedinductively by

-~i
r if n==O
i
e
,,'; an == an-l·a if n>O

l(a- fl1
1
if n<O
«:

Here a -I is the inverse of a and e is the identity element in G. Then f is a homomorphism.

In the fOllowing theorem, certain important elementary properties of homomorphisms are

derived.

1.2.13 Theorem: Let f :G -..:;G' be a homomorphism of groups. Then the following hold.

(i) f (e) == e' I where e and e' are the identities in G and G' respectively.

(Hi) For any a, bEG,

f (a) = f (b) <=> f (ab -}) = -e' c> f (11-1b ) = e'

Proof;
(i) Consider,

f (e) f (e) = f (e. e) = f (e) == e'· f (e)

and now, by the cancellation laws in G', we have f (e) = e' .

 Centre for Distance Education 6 Acharya Nagarjuna University

1.3 THE KERNEL OF A HOMOMORPHISM .

In theorem 1.2.13 (i) you have learnt that any homomorphisn: of G into G' carries the
identity element of G to that of G' . There may be several elements in G which are carried to the
identity element of G'. The collefion of such elements in G is called the kernel of G and this plays
an important role in the study the homomorphism itself. Formally, let us have the following:

1.3.1 Definition: Let f :G -+ G' be a homomorphism of groups and e' be the identity element of
G'. The Kernel of f is defined as the set

Ker f = {a E o] f(a) = e'}

!
1.3.2 Self'Assessment Question: Determine the kernels of each of the homomorphisms given in
examples 1.2.2 to 1.2.12.

1.3.3 Theorem: The kernel of any homomorphism f :G -+ G' is normal subgroup of G .

Proof: Let f :G -+ G' be a homomorphism of groups and

K = Ker j = {a E ojf (a) = e}

Where e and e' are the identities in G. By theorem 1.2.13(i), f (e) = e' and hence e E K,
so that K is a nonernpty subset of G. Now

a, hEK =:> j(a) = e' = f(h)

=:> f( ab-1)= e' (by 1.2.13 (iii))

and therefore K is a subgroup of G. Also,

=> xax-1 E K

Thus K is a normal subgroup of G .

1.3.4Note: A convese of 1.3.3 can be stated as follows: Any normal subgroup of G is the kerne
'of some homomorphism of G into a suitable group G'. This state is also true; for, let N be a

normal subgroup ofa group G and considarthe quotient group % and define f :G -+ % by
 \

. ( Algebra ~

f(a)=aN for any aEG(see 1.2.14).

Then f is a homomorphism and

,

Ker f = {aEG/f(a) = N} = {aEG/aN = N}= N

Note that N is the identity element of the quotient group %.

Before taking up the next important property of the kernels of homomorphisms, let us recall
that, for any mapping f ':A -t Band b E B an element a E A is called an inverse image of b
t

under f if f (a) := b. There may be several (or not even one) inverse images of b. However, if

f is a surjection of A onto B , then any b E B has at least one inverse image in A under f. In
the following, we describe an important property of kernels of surjective homomorphisms.

1.3.5 Theorem: Let f be a homomorphism of G onto G' (i.e., f: G -t G' is a surjective

homomorphism) and Y E G'. Let K be the kernel of j. Then the set.of all inverse images of y
in Gunder f is equal to the coset aK, where a is any inverse image of y.

Proof: Let a be an inverse image of y inG under f; i.e., a EG such that f (a) = y. Then, for

any XEG,

X E aK <=> x = az for some Z EK

<=> a-l x E K

<=> f(a-lx) = e'

<=> j(x) = j(a) = y (by 1.2.13 (iii))

Thus aK = {xEG/f(x) =y}

1.4 ISOMORPHISMS :

Consider the group G = {I , -1} under the usual multiplication and the group Z,2 = {O, 1}
~ ~, .r

under the addition modulo 2. These- two groups look like similar, except for the labelting or naming
of the elements. The identity in G is 1 wrtile in Z,2 it is O. Also the other element -1 inG has the

property that (-1) (-1) =1 while the element 1 in Z,2 has a similar property that 1+1 = O. In other

words,there is a bijection j:G -t Z,2 which is a homomorphism too, such an j is defined by

 Centre for Distance Education B Acharya Nagarjuna University

1(1)=0 (which is a necessary property of homomorphisms) and f( -1)=1. Let us formalise this
idea in the following.

1.4.1 Definition: A homomorphism f: G ~ G' is called an isomorphism if j is a bijection (l.e.,

f is both an injection (one - one) and a surjection (onto)). Two groups G and G' are said to be
isomorphic and is expressed as G == G' if there is an isomorphism f:G -+G'.
!
1.4.2 Theorem: A hornomorphtsm f :G ~ G' is an injection if and only if Ker f = {e} , where e is

the identity in G.

Proof: Let f :G ~G' be a homomorphism of groups. Suppose that f is an injection. Recall

that \

Ker f = {aEG/f(a) = e'}

Since it is always true that f (e) = e' , we have e E ker f. On the other hand,

a E Ker f => f (a) = e' = f (e)

=> a =e (SH.ce f is an injection)

Thus Ker f ={e}. conversely, suppose that ker f = {e}. Then, for any elements a and
b in G,

f(a) = f(b) => !(ab-1) =e' (by 1.2..13 (iii))

=> ab-l EKer f = {e} ,,

=> a =b

Thus f is an inlection.

Injective 'nornomorphirns are usually called monomorphisms and surjective

no,· 1ui •• '::l ;::;hisms are called epimorphisms. An isomorphism is both a monomorphism and an
epimorphism. A homomorphism ra ~·G' is an isomorphism if and only if Ker f = {e} and
every element of G' has an (unique) inverse image in G.
<,
~ Lesson 1::

1.4.3 Theorem :

(i) The inverse of an isomorphism is also an isomorphism

(ii) If / : G -+ Hand g : H -+ K are isomorphisms, then so is gof.

Proof: (i) Let / : G -+ H be an isomorphism. Then / is a bijective homomorphism. since / is

'3 bijiection, the inverse map 1-1 : H -+ G exists such that 101-1 and 1-101 are identity

maps of Hand G respectively. Also, 1-1 is clearly a bijection. Further, for any x , Y E H , we
have

and, since 1is an injection, we have

Thus /-1 : H -+ G is a homomorphism also and hence /-1 is an isomorphism.

"ollows from the facts that the compostition of two bijections (homomorphisms) is also
a :.IijeGllun (homomorphism respectively).

1.4.4 Self Assessment Question: Let G, Hand K be groups. Then prove that the following

(i) G:=::G

(ii) G:=::H => H :=::G

. (
(iii) G:=:: Hand H == K => G == K
1.5 THE FUNDAMENTAL ,-THEOREM OF HOMOMORPHSMS ,
.
We shall prove a very crucial and fundamental theorem which states that any homomorphic
image of a group G is isomorphic to a quotient of G. In theorem 1.2.14, we have already proved

that any qutient % of a group G is a homomorphi~ image of G. We prove a converse of this in

the following.
1.5.1 Theorem (Fundamental Theorem of Homomorphisms) :

Let / : G
.
-+ G' be a surjective homomorphism of groups. Then GI
7Ker / ::::G'
- .

Proof: Let K = Ker 1. We have already proved (in 1.3.3) that K is a normal subgroup of C
Now, define
 Centre for Distance Education 10 Acharya Nagarjuna University

g:% ~G' by g(ak) = l(a)

for any ak E o/~ with a E G. Let us recall that, for different elements a and b, aK and

bK may be equal. On the face of it, g( aK) looks like dependting on a. We shall first prove that

g( aK) depends on the coset aK but not on the element a, For any a, bEG,

aK = bK <=> ah-I E K =Ker 1

. <=> 1(a) = 1(b) (by 1.2.13 (iii))

this proves that g is well defined and g is an injection also. further, for any a, bEG,

g(aK .hK) = g((ab)K) = f(ab) = j(a)·f(h) = g(aK)· g(hK)

and hence g is a homomorphism. since 1is given to be surjective, any x E G' is of the

form l(a) for some a E Gand hence x = l(a) = g(aK) and aK E %. Therefore g is a

surjection also. Thus g is an isomorphism of % onto- G' and %=c.

The above theorem is actually an important tool in the development of the structure theory
of groups. Before going to certain applications of the fundamental theorem of homomorphisms, let
us prove the following:

1.5.2 Theorem: Let 1:G ~ G' be a surjective hornomorphism with kernel K.

(i) If H' is a subgroup of G' and H = 1-1 (H') = {a'~G/f (a) E H1' then H is a subgroup

of G containing K.

(ii) The correspondence H' H .1-1(H') is a one-to-one correspondence between the

subgroups of G' and the subgroups of G containing K .

(iii) H' is a normal subgroup of G' if and only if j-l (H') is a normal subgroup of G.

Proof: (i) Let H' be a subgroup of G' and H = 1-1 (H') = {a E G If (a) E H'}

For any a E K , we have f (a) = e' E H' and hence a EO H· Therefore K.~ Hand,

in particular H oF ¢. Also,
 ( Algebra ~

a, b E II ~ I(a) , I (b) E H'

1
~I (a )/( b- ) ElI' (since H' is a subgroup of G')

Thus H is a subgroup of G containing K .

(ii) Let S' and T' be subgroups of G' such that I-I (S') c I-I (T'). Then

XES' ~ I (a) = XES' for some a E G (-,' I is onto)

~ aE/~I(S')c/-I(T')

~ I(a) E T'

~xET'

and, therefore, S' cT'. Thus S'= T' whenever 1-1 (S') = I~~(T'), which implies that
'I

the correspondence H' ~ I-I (lI') is one-one. To prove that this is onto also, let S be a subgroup

of G containing K. Put H' = {j(a}/ dES}. Then clearly H' is a subgroup of G' and

S c I-I (H'). Further,

bE/-I(H') ~ l(b)EH'

~I (b) = I(a) for some aES

- --- »>:

~I (ba -1 ) = e' and a ES

~ ba -1 E Ker I=K <;;:;; S and a E S

 - Centre for Distance Education 12 - Acharya Nagarjuna UniversiiY:E

Therefore S = /-1 (H'). Thus H' H /-1 (H') is a one - to - one correspondence between

the subgroups of G' and subgroups of G containing K .

(iii) Suppose H' is a normal subgroup of G' _ Then xyx -t E H' for all y E H' and x E G' ,

Now,forany bEG and aE/-1(H'),wehave f(a)EH' and f(b)EG' and hence

so that bab -1E /-1 (H'). Thus /-1 (H') is a normal subgroup of G .

Conversely suppose 1-1 (H') is a normal subgroup of G, where H' is a subgroup of G'.
Let Y E H' and x E G' _ Since f is onto, we canchoose a and b in G such that y = f (a) and

x=/(b). Then aE/-1(H') and hEG. Since 1-1(H') is normal in G, we have

bab-1E/-1(H')and therefore f(bab-1)EH'. Now xyx-l = f(b)f(a)f(hrl

== f (bab -I ) E H' .

Thus H' is normal in G'.

1.5.3 Theorem: Let f :G~ G' be a surjective homomorphism with Kernel K.· Let N' be a

normal subgroup of G' and N = 1-1 (N') _Then - (n;

GI G'I
IN = IN'

and, equivalently
GI
-; N
-(%)/
= / ( %)

Proof: Define g :G ~ GIN, by g(a) = f(a)N'. Then, for any a, bEG, we have

g(ab) = f(ab)N'

== f(a)f(b)N' (': f is a homomorphism)

 ( Algebra :E
= f(a)N'· f(b)N'

=g(a).g(b)

Therefore g is a homomorphism. Also,

X E GIN, => x = zN' for some z E G'

=>x=f(a)N'=g(a),forsomeaEG (since f is onto)

Therefore, g :G ~ GiN, is a surjective homomorphism.

By the fundamental theorem of homomorphism (1.5.1),

GI - G'I
/Ker g = IN'

But, Ker g = {a E G / g( a) = N', the identity in GiN,}

= {a E G / f(a)N' = N'}

Also, if we restrict f to N, therr'it becomes a surjective homomorphism of N onto N'

whose Kernel is K again and hence by the fundamental theorem of homomorphisms (1.5.1).

N' == % and G' == %

Thus

1.5.4 Self Assessment Question: Let Nand K be normal subgroups of a group G such that

K <;;; N. Then prove that % is a normal subgroup 01% and that (0/%IK) " %

1.5.5 Self Assessment Question : Let N be a normal subgroup of a group G and K be an)
 Centre for Distance Education 14 Acharya Nagarjuna University '-

subgroup of G. Then provethat KN is a subgroup of G, K nN is a normal subgroup of K and

1.6 CAUCHytS THEOREM AND SYLOW'S THEOREM:

In the previous section, we have learnt that the homomorphic images of a given group G
coincide (upto isomorphism) with the quotient % of G , where N is a normal subgroup of G. A
group is said to be simple if it has no nontrivial homomorphic images or, equivalently, if it has no

nontrival normal subgroups. When we construct the quotient group %, where N is a normal

subgroup of G, knowinq the struture of % help us in knowing the structure of G "upto Nil. We
can ascertain certain information about G by looking at those of a quotient of G. These ideas are
applied in proving the following two theorems. In fact, later we prove these results in a much more
several set up and in easier way. However, the proofs of these two results are important on their
own in view of the use of several group theoritic concepts and illustrations in proving these.

1.6.1 Theorem (Cauchy's theorem for abelian groups) : Let G be a finite abelian group and p
a prime number such that p divides o( G). 'Then G has an element a such that a:j::e and

aP=e.

Proof :We shall use induction on o( G). since p divides o( G), we have p s; o( G). If o( G)= p,

then we can take any element a:j::e in G, for, by the Lagr~nges theorem, o( a)= p and hence

·aP=e.

Next, suppose that o( G) > p and assume that the theorem is true for all abelian groups of

order less than o( G). Choose an element b:j:: e in G. Let n be the order of b; i.e., n is the least

positive integer such that b" =e. We shall distinguish two cases.

Case( 1) : Suppose p divides

element a.
n. Then blr '"e and (blr r~ e and hence blr is the req uired

Case(2) : Suppose p does not divide n. Let H be, the subgroup generated by b in G. Then
 ( AlgebraE'"

o(H)=n>l. since G is abelian group, H is normal in G and hence we can consider the

quotient group % .Also, we have

o(G) = o(H).o(%) = n.o(%)

Since p divides o( G) and p does not divide n, it follows that p divides 0(%). Also

% 'is "group of order less than 0 (G) (since n > 1and 0 ( %) = O~G)). Therefore, by the

induction hypothesis, there exists a non identity element X in % such that X P = H , the identity

in %. Let X=xH forsome XEG. Then Xo(x) = (xH)O(X)= xo(x)H = H . Since,o(X) =p

in %, we get that p divides o( x). Now, as in case(1), xo(x)/ P is the required element a in G.

1.6.2 Theorem (Sylow's theorem for abelian groups) : Let G be a finite abelian group, p a

prime and r a non negitive integer such that p" divides o( G) and pr+l does not divide o( G).
. r
Then. G. has a subgroup of order p .

Proof: If r = 0 , then pr = 1 and {e} is the subgroup of order pro

Suppose r >0. Then p divides o( G) and hence, by the Cauchy's theorem (1.6.1), there
I\ .

exists an element a:f:. e in G such that aP = e. Now, consider

H = {x E G / xpn = e for some n E Z+ }

Then H has at least two elements, namely e and a. We shall prove that H is a subgroup
of G.

n m
X, y E H => xP =e = v" for some n, m E Z+

=> (xy-l r n+m

= xP n+m (
. yP
n+m )-1 = e
=> xy-l EH
 Therefore H is a subgroup of G. Next, we shall prove that p is the only prime dividing

o(H). If q is a prime dividing o( H) and q -:;:.

p, then again by 1.6.1, there exists x -:;:.
e in H such
n
that xq = e. Since x E H, xP = e for some n E Z+. Since q -:;:.
p. q and p~~3-re relatively

prime and hence there exist integers t and s such that tq + sp" = 1 and now consider

which is a contradiction, since, by our choice x -:;:.

e. Therefore no prime other than p

divides o(H). This implies that

o(H) = pm for some m> O.

Since o(H) divides 0 (G) (by the lagranges theorem) and since pr+l does not divide,,;
i r ,

o( G), it f~IIOWSthat

"
\
I
Sincr a~Hand a v-e , o(H»l and hence m i-I), Finally we prove that m=r and

conclude thit H is the required subgroup of G. On the contrary, suppose m < ri. Consider the

quotient gro~p % (note that, sin;~ G is abelian,H is normal in G). Then . -'.
i ~

, if:. ,;

so that pr-m divides 0( %) and r - m > O. Thus p divides 0 (%). Again by the

Cauchy's theorem 1.6.1, there exists an element xH in % such that

xli e H and (xHt =H
pn
Then xP H =H and hence xP EO ,H so that (xp) =e for some n and therefore

I
n+l
xP = e which implies that x E Hand xH =H which is a contradiction to the choice of x.

Thus m=r and o(H) = pr

 Later, we extend both the above theorems for arbitary finite groups. However, if G is a finite
, abeltan group, then the subgroup H described in theorem 1.6.2 is unique. This is proved in the
foH~),rin~.

1,5.:1 Theorem: Let G be a finiteabelian group, p a prime and r a non regitive integer such that

pr di /ides o(G) and pr+l does not divide o(G). Then G has a unique subgroup of order r",
Proot ;The theorem is trivial for r = O. Therefore, we can suppose that r > O. We have proved

the existence of a subgroup of order pr in the Sylow's theorem (1.6.2), we shall now prove the
.. '. - r
uniqueness. Let H .and K be any subgroups of order P in G. Since G is abelian, HK =KH
and hence HK is a subgroup of G. Also,

Therefore o( HK) = pS for some integer s . By the lagrange's theorem, pS divides o( G).
Since H ~ HK,

-qC hence r s, s. But, since pr+l does not divide 0 (G), s cannot be strictly greater than

r. Th: :" r=swhichimpliesthato(H)=o(HK) andhenceH=HK. ThisyieldsthatK<:;H,

since A : r,d H are of the same order, we get that K = H .

T 't~ above theorem fails for non abelian groups; that is, if G is finite non abelian group

pr / o( G and pr+l X o( G), then G may possess more than one subgroups of order pr. The
following example illustrates this.

,1.6.4 Example: Consider the group S3 ' the symmetric group of degree 3. Then S3 is a non abelian

group of order3!.= 6. Take p=2 and r=1. Then pr /0(5]) and pr+lx 0(53), For any

transposition o in S3, {id, ()} is a subgroup of order r' in 5]. There are three distinct
transpositions in 53 ; namely (1 2), (23) and (31) and hence there are three distinct subgroups,
each ot orderz.Jn S3'

1.6.5 Self Assessment Question: How many subgroups are there in 5] , each of order 3?
=!~~t~re~f~o~r~Di~st~a~nc~eJE~d~u~ca~ti~on~~~~~~18~~~~~~~~1A~c~ha~r~ya~N~a~g~arTjU~n~aiu~n~iV~er~s
, , r' r+I "
Later, we shall prove that any two subgroups J-I and K of order P ,where P doesn t
divide o(G), must be conjugate to each other, in the sense that J-I = aKa -I tor some a E G, even
though they may not be equal.

1.6.6 Self Assessment Question: Prove that any two subgroups of order 2 in S3 are' conjugate
to each other.

1.7 SUMMARY:
In this lesson, you have learnt the concepts of a homomorphism, isomorphism and Kernel
of a homomorphism and proved certain important properties of these. We have proved three very
important theorems, namely, the fundamental theorem of homomorphisms, Cauchy's theorem for
abelian groups and sylow's theorem for abelian groups.

1.8 TECHNICAL TERMS

Homomorphism I

Isomorphism (bijective homomorphism)

Monomorphism (injective homomorphism)

Epimorphism (surjective homomorphism)

Kernel of a homomorphism
,

Homomorphic image
Fundamental theorem of Homomorphism
Cauchy's Theorem

Sylow's theorem

1.9 MODEL EXAMINATION QUESTION :

1.9.1 : Define the concepts of a homomorphism of groups and 'its kernel. ; Prove that a
homomorphism f :G ~ d is injective if and only if Ker f = {e} '.
1.9.2 : State and prove the fundamental theorem of homomorphisms.

1.9.3 : State and prove the Cauchy's theorem for abelian groups,
1.9.4 : State and prove the Sylow's theorem for abelian groups.

P
1.9.5: If G is a finite abelian group, P is a prime and r is a non negitive integersuchthat ;
//o(rJ}
,' .. -.
 \ "',.fA
\ if
~ Lesson 1) ,r;{Q . C~!~::> ~p ~

and pr+1 X. o( G) , then prove that there can be atmos: one subgroup of order pr in G. Is this
true for non abelian groups? Justify your answer.

, 1.10 ANSWERS TO SELF ASSESSMENT QUESTIONS:

1.2.7: Let a, bEZ. We have prove that I(a+b) = I(a). I(b). If a and b are both even, then

so is a + b and I (a + b) = 1, j (a) = 1= I (b), If both a and b are odd, then a+ b is even and

j(a+b) =1 = (-1) ·(-1) = j(a). j(b)

Similar argument can be made when one of a and b is even and the other is odd.

1.2.9 Let a, bE Z, write a = qln + rl and b = q2n + r2, where\; q2, rl, r2 E Z, 0 s; rl < nand

o s; r2 < n. Then I (a) = rl and j (b) = ri Also, we have 0 S; rl~2 < 2n we shall distinguish
two cases.

Case(1) Suppos rl +r2 <n. Then a+b = (ql +q2)n +(1"1+r2) and

I(a+b) ='1 +r2 = rl +» r: = I(a) +» j(b) where +» is the addition modulo n.

C a se (2) : Sup p 0 s e rl + r2 ?:. n. The n a +b = (ql + q2 + 1) n + h + r: - n) and

0::;;r1+r2 =n c.n and hence j(a+b) = rl +r2 -n = rl +}1 r2 = j(a)+n I{b) ,

Thus I is a homomorphism,

1.2.15 Forany a, bEG,

(goj)(ab) = g(j (ab))

= g(j(a)j(b)) jsmce/ is a homomorphism)

= g (j (a) ) g tr (b)) (since g is a homomorphism)

= (gqf) (a) . (goj) (b)

Therefore go j is a homomorphism.

1.3.2 : (1.2.2) The identity in G' is 0 and hence

,
 Centre for Distance Education 20 Acharya Nagarjuna Univer"\! -

Ker f = {a E G I f (a) = a}
= {aElR I a> a and log2a = o} ={l}
(1.2.3) The identity in G' is 1 and hence

Ker f = {a E G I f (a) = I}

={aE~/2a =1 }={a}

(1.2.4) Every element of G is mapped to e' and hence Ker f =G

(1.2.5) f :Z ~ Z is defined as 1(a) = na .: If n = a, then 1is the trivial homomorphism

and Ker 1=Z. If n 7= 0, then Ker f = {a E Z / na = a}= {a}.

. I
(1.2.6) Ker f={aEZI f(a)=l}

= {a E Z I a is even} = the set of even integers.

(1.2.8) 0 is the identity in Z n and hence

Ker f = {a E Z Ia = q n for some q E Z} = nZ .

(1.2.10) Ker f = {e}

(1.2:11)

(1.2.12) Ker f = { n EZ J an = e} = {a} if a is of infinite order; that is o" :;c e fc

all n 7= 0 and Ker 1= nZ, if a is of order n ; that is n is the smallest positive integer sur' the

1.4.4 (i) The identity mapping of Gonto G is an isomorphism and hence G ~G.

(ii) If G ~H ,then there exists an isomorphism f :G ~ H and, in this case,

1-1 : H -+ G is an isomorphism (see 1.4.3 (i)) and hence H ~ G ...

(iiij This follows from 1.4.3(ii).

 ( Algebra ::e
1.5.4 Let f:G ~ ~ be defined by j(a)=aK for any aEG. Then feN) = % whichisa
normal subgroup of ~ and, by theorem 1.5.3,

1.5.5 Since N is normal, NK =KN and hence KN is a subgroup of G. If x E K and a E K nN

then xax -1 E K nN and hence K nN is a normal in K. Define f : K ~ K% by f (a) = aN.
Prove that f is an epimorphism and its kernel is K nN and use 1.5.1.

1.6.5 Let a be the 3 - cycle (1 23) in S3' Then a2 =(1 3 2) and a3 =-id . {id , a, a2} is the

only subgroup of order 3 in S3 .

1.6.6 There are three subgroups, each of order 2, in S3 and these are

HI == {id , (1 , 2)}

H2 =- {id , (2, 3)}

and H3 = {id , (3 , I)}

Let j3 = (1,2), r=(2,3) an~~J =- (3,1) then

1.11 EXERCIS~ . ;
.'-

1.11." Determine whether the indicated function f is a homomorphism from the first group into
the second group.

(a) f(a) = -0.'1, (]I{, +), (]I{, +)"" .

(b) f (a) = 2a" (]I{, +) , (.]1{ - {o},"-)

(c) f(a) =l+a, (2,+), (2,+)

• "<.
 Centre for Distance Education 22 Acharya Nagarjuna University

(d) .f (a) = a 2 , (IR - {o} , .) , (IR + , .)

1.11.2 Let / : G ~ G be a homomorphism and

H == {a E G / / (a) = a}
Prove that H is a subgroup of G .

1.11.3 Define / : 28 ~212 by

/(0) = 0 = /(4), /(1) = 3 = /(5)

/(2) = 6 = /(6), /(3) = 9= /(7)

Prove that / is a homomorphism and determine its Kernel.

1.11.4 Let a be an element in a group G. Define / : 2 ~ G by f (n) = an. Provethat f is a

homomorphism and determine its Kernel.

1.11.5 Let G be a finite cyclic group of order nand m E Z+ such that m and n are relatively
.:
prime. Define / :G ~ G by / (x) = xK for all x E G. Prove that / is an isomorphism.

1.11.6 Give an example to show that a group G may be isomorphic to a proper subgroup of G .

1,.11.7 Prove that the group (IR, +) is not isomorphic to the group (IR - {o} , .) .

1.11.8 Prove that any group with p element is isomorphic to Z p , if P is a prime.

1.11.9 Prove that any 4 element group is isomorphic to either 24 or Z2 x Z2 .

1.11.10LetG be a group having no non trivial proper normal subgroups. Let / :G ~G' be a
surjective homomorphism. Prove that either G == G' or / is the trivial (or zero) homomorphism.

1.12 REFERENCES:
1. Topics in Algebra I.N. Herstein, Second edition, John wiley & sons, New
York, 1975.

2. Basic Abstract Algebra P.B. Bhattacharya, S.K. Jain and S.R. Nagpaul, second
edition, Cambridge University Press, 1995.

>.4.
g:: Lesson 1) ( Algebra ~

3. Lectures in Abstract Algebra, .: : : "N. Jacobson, Vol -I, D. Van Nostrand Company, Inc.,
London, 1964.

4. Algebra Serge Lang, Addison-Wesley publishing Company,

London 1980.

5. Algebra T.T. Moh, World Scientific, Singapore, 1992.

6. University Algebra N.S. Gopala Krishnan, Second edition, wiley Eastern

Limited, New
..
'
Delhi, 1986.
7. Abstract & Linear Algebra - David M. Burton, Addison - Wesley publishing company,
London 1972.

8. Algebra M. Artin, Prentice - Hall of India, New Delhi 1994.

9. Algebra J.W. Archbold, Fourth edition, the English language book

society, 1972.

10. A First Course in Abstract Algebra ,- John B. Fraleigh, Addison - Wesley Longman, Inc.,
Fifth edition,1999.

Lesson Writer
Pro}: us« Swamy
Andhra University
Visakhapatnam.
Lesson - 2 AUTOMORPHISMS

Ojectives:
After reading this lesson, you should be able to

(i) define the notion of an automorphism and give certain examples.

(ii) prove that the set of automorphisms of a group G is itself a group under the
composition of mappings.
(iii) define the concept of an inner automorphism and prove that the group of inner

automorphisms of a group G is isomorphic to the quotient % (G), ~here Z(G) is

the centre of G.
(iv) determine all the automorphisms of a given cyclic group.

Structure:
2.1 Introduction

. 2.2 The group of automorphisms

2.3 Inner automorphisms

2.4 Automorphisms of a cyclic group

2.5 Summary

2.6 Technical terms

2.7 Model examination questions

2.8 Answers to Self Assessment Questions
2.9 Exercises

2.10 Reference books /

2.1 INTRODUCTION
I
/
In the previous lesson, you have ~earnt the concepts of a homomorphism and a bijective
homomorphism, which is called an isom6rphism. When an isomorphism is from a group G onto
itself, it is called an automorphism of G. Automorphisms play an important role in the structure
theory of groups. For any group G, the automorphisms of G form a group under the compostition
of mappings. On several occations, the structure of the group o,f automorphisms of G reveal that
,
 (Distance Education 2 Acharya Nagarjuna University )

of the group G itself. In particular, when G is an infinite cyclic group, then there are exactly two

automorphisms and when G is a finite cyclic group of order n, then there are exactly ¢ (n) number
of automorphisms of G which is also equal to the number of generators of G. In this lesson, we
shall have a detailed discussion on these topics.

2.2 "J:HEGROUP OF AUTOMORPHISMS :

Isomorphisms of a group G onto itself are called automorphisms. In the following theorem
we shall prove that the automorphisms of a given group form a group again. First, let us have the
formal definition of an automorphism.

2.2.1 Definition: Let G be a group. Any bijective homomorphism of G onto G it self is called an

automorphism of G. The set of all automorphisms of G will be denoted by Aut (G) .

Note that the homomorphism given in example 1.2.5 is injective but not surjective and that
given in example 1.2.8 is surjective but not injective. This says that surjective (onto) and injectivity
(one - one) are both necesary for a homomorphism to become an isomorphism or an automorphism.

2.2.2 Theorem: Lei: G be a group. Then the set Aut (G) of all automorphisms of G forms a group
under the composition of mappings.

Proof : If/ and G are automorphisms of G, then / : G ~ G and g :G ~ G are isomorphisms

and hence so is /0 g , so that fog is also an automorphism of G. Thus, the composition of

mappings is a binary operation on the set Aut (G). Clearly, we have /0 (g 0 h) = (fo g) 0 h for all
f, G, hE Aut ( G). Also, the identity map ic ' defined by to (a) = a for all a E G, is an automorphism
••
of G and hence tc E Aut (G) and, for any / E Aut (G),

tc 0/ = / = /oia
Therefore to is the identity element of Aut(G). Further, for any / E Aut (G), / being f
bijection, its inverse /-1 exists and /-1 is also an automorphism (see 1.2.3(i)). Since

f /-1 = ia
. 0
. = /-1 f
0.,

it follows that /-1 is the inverse of/in Aut(G). Thus Aut(G) is a group under the composition
of mapping.

Every element of a group G has an inverse in G and hence a!-~a-l can be treated as a
 3 Algebra D
function of G into G. If a-I = b -1 for any elements a, bEG, then a = (a-1 r l
= (b-l r l
= band,

for any YEG, y-1EG and (y_l)-1 =y. That is the map a 1---7 a-I isabijiectionofGontoG.

But, this mapping need not a homomorphism in general. However, G is abelian, if and only if, the

map a 1---7 a-I is a homomorphism and hence an automorphism of G.

2.2.3 Self Assessment Question: Prove that the following are equivalent for any group G.

(1) G is abelian

(2) The map 1:G ---+ G , defined by 1(a) =a -1 for any a EGis an automorphism of G.

(3) ( ab)-1 = a -lb-l for any a, bEG

(5) There exist three concecutive integers n such that (ab r = an bn for all a, bEG.

2.2.4 Theorem Let G be a group and a E G. Let 1 be an automorphism of G. Then

o(1(a)) = o(a).

Proof: If a is of order zero, then an oj:. e for any positive integer n and hence

1(a r = 1(an) oj:. 1(e) = e

So that 1(a) is also of order zero. Now, suppose 0 (a) > 0 and let 0( a) = n. Then n is
the least positive integer such that an = e. We have

f(ar = 1(an) = fee) = e

am
and, for any posi.ive integer m < n , oj:. e and hence

f (at = f (am) * 1(e) = e

Thus n is the least positive integer such that 1(a t = e. Therefore, 0 (1(a)) = 0 (a)
 ••

(Distance Education 4 Acharya Nagarjuna University J

2.2.5 Self Assessment Question: If f: G ~ G' is an isomorphism ofg;oups and a E G, then

prove that the orders of a and f (a) in G and G' respectively are same.

2.3 INNER AUTOMORPHISMS :

In this section we shall introduce a special type of automorphisms known as inner
automorphisms. With each element a in a group G, we shall associate an automorphism of G
as defined in the following:

2.3.1· Theorem: Let G be a group and a E G. Define

fa:G ~G by fa(x)=axa-1 for any XEG.

Then fa is an automorphism of G.
/

Proof: For any x , y E G , we have

and hence fa is a homomorphism. Also,

by the cancellation laws. Therefore fa is one - one. Finally to provethat fa is a surjection,

1
let us take YEG. Then a-lya~? and fa (a-1ya) =a(a-1ya)a- =j'. Therefore fa is a

surjection also. Thus fa is an isomorphism of G onto intself; that is fa is an automorphism of G.

2.3.3 Definition: For any group G, the centre of G is defined as the set·

Z (G) = {a E G I ax = xa for all x E G}

2.3.4 Self Assessment Question: Prove that the centre Z(G) of a group G is a normal sub
group ofG.

2.3.5 Theorem: For any group G, the set I (G) of inner automorphisms of G is a group under the
composition of mappings and
@-esson - 2 5 Algebra D
%(G) == I (G) Where Z (G) is the centre of G.

Proof: Let G be a group and Aut (G) be the set of automorphisms of G. We know that Aut (G)
is a group under the compositions of mapping of (see 2.2.2). We shall prove that the set I (G) is

a sub group of Aut(G). For any a E G, we have fa E I( G).

If fa,fb E I(G) with a, bEG, then

Ie (x) = exe-1 =X = id(x)

and hence faofb = fab and fe = id, the identity map. Therfore

Thus I(G) is a sub group of Aut(G). Now, let us define,

a :G ~ I (G) by a (a) = fa

then a(ab)=fab =faofb =a(a)oa(b) for any a, bEG. Therefore a is a

~\
\., _-..
homomorphism. Also, clearly a is a surjection (since any element of I (G) is of the form fa for

some a E G). Now by the fundamental theorem of homomorphisms (1.5.1), we have

;,

G/
/Kera
= l(G)
Let us compute Ker a . By the definition of the kernel, we have

Ker a = { a E G la (a) = identity in I (G)}

= {aEG I Ja(x) = x for all xEG}

, .
~istance Education 6 Acharya Nagarjuna University }:

= {a E G I axa -I = x jor all x E G}

= {aEG I ax = xa jor all XEG} = Z(G)

Thus Yz (G) is isomorphic to J (G) .

2.3.6 Self Assessment Question:

For any group G, prove the following:

(i) G is abelian, if and only if, J (G) is trivial

(ii) I(G) is a normal sub group of Aut(G)

2.4 AUTO MORPHISMS OF A CYCLIC GROUP :

Let us recall that a group G is said to be cyclic if there exists an element a generating G;
that is, G = < a> = {an In E z}. In this case, a is called a generator of G.

There can be more than one generator of a group G. For example 1 and -1 are both
generators of the group Z of integers under addition. In this section we shall determine all the
automorphisms of a cyclic group and prove that th~re is a one - to - one correspondence between
the automorphisms of a cyclic group G and the generators of G. Let us begin with the following:

2.4.1 Theorem: Let G be a group,fan automorphism of G and a E G. Then a is a generator of

G if and only ifj(a) is a generator of G.

Proof: Suppose a is a generator of G. Then

G = (a) = {an I n Z} E

fis an automorphism and, in particular, it is a surjection, so that./{G) = G. Now, we have

G= j (G) = {j (an) I n E Z} = {f (a r In E IZ} = (f \ a))

Therefore f(a) is a generator of G. Converse is clear, since r: is also an automorphism

ofG.
In the following, we shall determine all the automorphisms of a cyclic group. First, let us
fiLesson -2 7 Alge"brnJ)

recall that for any integer n > 1, the set

_Un = {r E f+ I r < nand rand n are relatively prime}

isa group under multiplication modulo n. For example,

U4 = {I, 3}

U5 = {I, 2, 3, 4}

Ul2 = {I, 5, 7, II}

First, let us takeup the case of a finite cyclic group.

2.4.2 Self Assessment Question: Determine the sets U30 and Un·

2.4.3 Theorem:

Let G be a finite cyclic group of order n> 1 and G = (a).

(1) An element bEG is a generator of G if and only if b = ar for some r E U;

(i.e., O<r<n and r is relatively prime to n).

(2)

Proof: (1) First suppose b =ar for some r E Un' Since r is relatively prime to n, there exist
integers a and f3 such that ar + f3n = 1 and hence

and, since any element of G is of the form am, me 7/." itfollows that any element of G is

of the form bk , k E Z. Thus G = (b) and be is a generator of G_

coriverselysuppose b isageneratorofG. Then h=ar for some O<r<n. Also,

 (Distance Education 8 Acharya Nagarjuna University J

and hence a = b" for some ineger s . Now al-rs = a. (ar fS = ab -s =e

and hence o(a) divides l-rs.· Since o(a) =n, it follows that

nt=l-rs or l=nt+rs

Which implies that r is relatively prime to n ; that is r E Un and b = a" .'

(2) For any r E Un' define gr : G ~ G by

Then clearly gr is a homomorphism. We shall verify that gr is a bijection, so that it

becomes an automorphism of G. since r is relatively prime to n, there are integers sand t such
that rs+nt=l. Now, for any X,YEG,

=>X=Y

Therefore s, is an injection. Further, by (1) above, G = \ -: and we have

Y E G => Y = (ar)m for some integer m

, Therefore gr is a surjection too. Thus gr is an automorphism of G. Now, consider the

ma~ \1': Un ~ Aut( G) defined by 'Y (r) = gr for any r E Un· You can easily verify that 'Y is

homomorphism (for, 'Y (r· s) = Sr-s = gr 0 K~)for any r, s E U; .

 '.-"1'": ": j'

fiLesson.2 9 Alge6ra'n}

~ ar-s =e

~ r - s = 0 C" r - s < nand 0 ( a) = n)

~r=s

Therefore 'I' is an injection. Finally, for any 1 E Aut (G), 1 (a) is a generator of G (by
2.4.1) and hence, by (1), f(a) =ar for some r E Un' If X =am EG, then

and therefore f = gr = 'I' (r). Thus 'I' is an isomorphism of U; onto Aut (G).
2.4.4 Self Assessment Question : Prove that the map 'I' defined in the above proof, is a

homomorphism of Un into Aut(G).

2.4.5 Self Assessment Question: Determine all the automorphisms of the group ZIO of integers
modulo 10

2.4.6 Theorem: Let G be an infinite cyclic group. Then Aut (G) ,cc:lnsistsonly two automorphisms,

namely, the identity map and the map x H x -1 .

Proof: Let G = (a). Since G is infinite, an of e for all non zero integers n. Let 1 be an

automorphism of G. Then f (a)- ~-d~ (a) and hence

f (a) = an for some nEZ

By theorem 2.4.1 1(a) is a generator of G and G = (I (a)) . ','

Therefore
 (Distance Education 10 Acharya Nagarjuna University J

and hence nm = 1 which implies that either n = 1= In or n = -1 = In. Therefore j (a) = a or

If f(a)=a,then j(x)=x for all xt=G and hencef is the identity map. If f(a)=a-1,

then f( x) =x-I for all x E G. Thus, the identity map and the map XH x-I are the only
automorphisms of G.

2.4.7 Self Assessment Question: Determine all the automorphisms of the group Z of integers.

2.4.8 Self Assessment Question: If G is an infinite cyclic group, prove that Aut (G) == Z2.

2.5 SUMMARY :
In this lesson, you have learnt the concept of an automorphism of a group G and proved

that the set Aut (G) of all automorphisms of G is a group under the composition of mappings.
Also, we have defined the notion of an inner automorphism of a group G and proved that these form

a group which is isomorphic to the quotient group Yz (G), where Z( G) is the centre of G.
Finally, we have completely determined all the automorphisms of a cyclic group. In particular, if G

is a cyclic group of order n, we have proved that Aut (G) is isomorphic to the group U; of positive
integers less than n and relatively prime to n.

2.6 TECHNICAL TERMS:

Automorphism

Inner Automorphism

The Centre Z (G)

Cyclic Group

The group ti;

The group Zn
fiLesson - 2 . 11 Algebra D
2.7 MODEL EXAMINATION QUESTIONS

2.7.1 : Prove that the set Aut (G) of all automorphisms of a group G forms a group under
composition of mappings.

2.7.2: Define the notion of the centre Z (G) of a group G and prove that rz (G) is isomorphic to

the group !(G) of all inner automorphisms of G.

2.7.3: For a cyclic group G of order n, prove that Aut (G) == U;

2.7.4: Determine all the automorphisms of an infinite cyclic group.

2.8 : ANSWERS TO SELF ASSESSMENT QUESTIONS:

2.2.3 (1)=>(2): f(ab)=(abrl =b-1a-1 =a-1h-1 =f(a)f(b) foranya,bEG.

Therefore f is a homomorphism. Also f (a -1) = (CJ r

I 1 = a.

for any a E G and hencef is a surjection. Further,

f(a) = f(b) => a-I = b-1 => a =b

Therefore f is an injection too. Thus f is an automorphism of G.

(2) => (3): ( ab )-1 .= f (ah ) = f (a) f (h) = a -_1b -1

(4) => (5): We have (ab)O =e = e·e = aO . bO,

t, (Distance
,~'1:-. ~~~~~~~~~~~~~~~~~~~~~~~~
Education 12 Acharya Nagarjuna University J

Therefore ,.0, 1 and 2 are three concecutive integers n such that (ab r= a" bl1 for any a,

(5) ~ (1): Let n be an integer such that

for all C!, bEG. Then for any a, bEG

and therefore ahl1-1 = hl1-1a. Similarly, ab" = b" a. Now, consider

and therefore ab = ba .

2.2.5 This follows from the fact that, for any integer n,

Recall that the order of a is zero if a" =e for all n 7= 0 and is n if n is the least positive

integer such that an =e .

2.3.4 Z (G) 7= ¢ since e E Z (G)

for any a,b E Z(G) we have

(ab ) x = a ( bX) = (hx) a = (xb) a = x (ba) = x ( ab )

for all x E G and hence ab E Z ( G). Also,

-1
ax = xa ~ xa = a -1x
 Therefore Z (G) is a sub group of G. For any a E Z( G) and g E G, we have

gag -1 = agg -1 = a E Z (G) .

Thus Z (G) is a normal sub group of G.

2.3.6: (i) G is abelian <=> ab =ba for all, bEG

<=>G==Z(G)

<=> % (G)
"\
is trivial

<=> I (G) is trivial (by 2.3.5)

(ii) Let fa E I( G) and g E Aut( G). Then, for any x E G,

(g a fa ~ g -I) (x) = g (fa ( g -I ( X))) = g (ag -I (X ) a-I)

= g(a) g(g-I(x))g(arl

==g(a) X g(arl = Jg(a) (x)

Therefore g 0 fa a g -1 = fg(a) E I (G). Thus I (G) is a normal sub group of Aut (G).

2.4.2 U30 ={1, 7, 11, 13, 17, 19, 23, 29}

U13 = {I, 2,3,4,5, 6, 7, 8, 9,10, 11, 12}

2.4.4 First note that O(a) = n and, for any m e Z, am =e <=> n divides m . For any r , S E U;
let rs =1, where t is the remainder obtained by dividing the usual product rs with n; i.e.

if rs=qn+t, 0 s::t < n , then rs>t .

Also note that x" == e for all x E G.

= ( x n)q x t == ex t =X
t
== gr.s .( X )
 (Distaiice t:ducation 14 Acharya Nagarjuna University J

for all XEG. Therefore gr ogs = gr.s and hence 'I' is a homomorphism of U; into Aut(G).

2.4.5: Since UlO = {I, 3, 7, 9} and hence there are four automorphisms of '£10. These are

Note that x9 = x-I for any x E '£10

2.4.7: By theorem 2.4.6, the only automorphisms of Z are the maps XHX and XH -x (since
\

Z is an infinite cyclic group).

2.4.8 Let G be an infinite cyclic group. Then Aut (G) = {id, f} where id is the identity map

andfisdefinedby f(x)=x-1 for all xEG. Anygroupoforder2mustbeisomorphicto '£2 and

hence Aut (G) == '£2.

2.9 EXERCISES:
2.9.1: Let f be an automorphism of a group G and H ~ G. Prove that

(a) H is a sub group of G ¢::> f(H) is so

(b) H is a normal sub group of G c> f(H) is so.

2.9.2: Determine which ofthe following are automorphisms?

(a) f:;;Z~;;Z,f(x)=:=-x

(b) f: }R.+ ~ lI.{+.: <l(x) = x2

Where IR+ is t.h~ group of postive real numbers under multiplication.

(c) f:'£12~'£12,f(x)=3x

(d) f :S3 ~ S3 ' f(x) =x-1

2.9.3 If G = S3 prove that G == 1 (G) , the group of inner automorphisms of G.

2.9.4 Determine Aut(G) for the group G = {e,.a,h,ab} where a2 =b2 =e and ab=ba .

2.9.5 Letf be an automorphism of a finite group G such that, for any x E G, fx =x <=> x = e.
Provethat any y E G can be reprerented as y = x -If (x) for some x E G.
 \
.:;ace
«lesson -2 15

2.9.6 If j2 =id in the aboveexercise prov'elhat G is abelian.

2.8.7 For any automorptusrnj'ota group G, prove that

J(Z(G))c Z(G)
2.9.8 Let G be a group of order 9 generated by elements a, b where a3 = e = b3. Find all the
automorphisms of G.

2.9.9 Prove that Aut (G) is abelian, if G is cyclic.

2.9.10 Prove that Aut(Zn) isanabelian group isomorphic to Un'

2.10 REFERENCES:
1. Topics in Algebra I.N. Herstein, Second edition, John wiley & sons, New
York,1975.

2. Basic Abstract Algebra PB. Bhattacharya, S.K. Jain and S.R. Nagpaul, second
edition, Cambridge University Press, 1995.
3. Lectures in Abstract Algebra - N.. Jacobson, Vol -l, D. Van Nostrand Company, Inc.,
London, 1964.

4. Algebra Serge Lang, Addison-Wesley publishing Company,

London 1980.

5. Algebra TT Moh~ World Scientific, Singapore, 1992.

6. University Algebra N.S. Gopala Krishnan, Second edition, wiley Eastern
Limited, New Delhi, 1986.
,o\" ,

7. Abstract & Linear Algebra - David M. Burton, Addtson - Wesley publishing company,
London 1972 .. "

8. Algebra M. Artin, Prentice- Hall of India, New Delhi 1994.

9. Algebra J,W. Archbold, Fourth edition, the English language book
society, 1972.
10. A First Course in Abstract Algebra - John B. Fraleigh, Addison - Wesley Longman, Inc.,
Fifth edition, 1999.

Lesson Writer:
Prof. U.M. Swamy
Andhra University
Visakhapatnam.
Lesson - 3 CAYLEY'S THEOREM

Objectives:
After reading this lesson, you should be able to

(i) prove that the bijections of any set onto itself form a group.

(ii) state and prove the Cayley's theorem and

(iii) qudte certain important applications of Cayley's tneorern.

Structure:
3.1 Introduction
3.2 Group of Permutations
3.3 Cayley's Theorem
3.4 Applications of Cayley's Theorem

3.5 Summary

3.6 Technical Terms

3.7 Model examination questions .
3.8 Answer to Self Assessment Questions

3.9 Exercises

3.10 References
~

3.1 INTRODUCTION':
Most of the groups, when they were first identified, were in the form of a set of transformations
of a particular mathematical structure. Most finite groups appeared as groups of bijections of an n

element set onto it self for some positive integer n. It is known that the set S (X) of all bijections of
. . .

a set X onto it self forms a group under the usual composition of mappings. The English
mathematician Cayley first noted that any abstract group can be viewed as a subgroup of the
group S (X) for a suitable set X. In this lesson, we shall prove this theorem of Cayley and derive
certain important consequences.
 (Distance Education 2 Acharya Nagarjuna University J

3.2 GROUP OF PERMUTATIONS:

The Cayley's theorem states that any group can be identified with a subgroup of the group of
permutations on a suitable set. Before taking up the proof of the cayley's theorem, we shall first
prove that the permutations on any set form a group. Let us begin with the following:

3.2.1 Definition: For any non-empty set X, any bijection of X onto itself is called a permutation or

symmetry on X. The set of all permutations on X will be denoted by A (X) .

3.2.2 Theorem: For any non empty set X, A (X) is a group under the composition of mappings.
\ '
Proof: Let us recall that an injective (one - one) and surjective (onto) function is called a bijection
and that the composition of two bijections is again a bijection. Now, let X be any non empty set.

Then the composition 0 is a binary operation on A (X). Clearly 0 is an associative operation.

Also; the map id: X ~ X, defined by id ( x) =x for all x EX, acts as the identity element in

A(X); 'that is,

.~ \

I 0 id = I = id I 0

,
for any f E A(X). further, for any bijection I: X ~ X, we can define 1-1: X ~ X by

I-I (y) = x if and only if I (x) 0;:: y (since I is a bijection, there exists unique x in X such that

I(x)=y). Then, clearly I-I is a permutation onXand

Therefore, each element of A (X) has inverse. Thus A (X) is a group under the
composition 0 of mappings.

4.2.3 Self Assessment Question: List all the elements of A (X), where X = {I, 2, 3}

4.2.4 Self Assessment Question: If X has n elments, provethat A (X) has n! elsments.

4.3 CAYLEY'S THEOREM :

In this section, we shall prove that any abstract group can be identified with a group of

permutations on a suitable set X (that is a subgroup of A (X) ).

fiLesson - 3 3

4.3.1 Theorem (Cayley's Theorem) :

Any group is isomorphic to a group of permutations on a set.

Proof: Let G be a group; that is, (J is a non empty set together with a binary operation satisfying
the axioms of groups. Take X to be the set G, ignoring the binary operation on it. for each U E G,

define t a : X ~ X by t a (x) = ax for all x EX. Note that X = (; and the product a x in (; is

written as ax. Then fa is a bijection; for,

la(X) = ta(Y) => ox = (~F => x = .I'

Therefore I a E A (X) , the group of permutations on _Y. Let

We shall prove the following for any a, hE G :

(1)

(2) Ie = id , the identity map on X.

-I
(3) ta = I -I
a

For any x E G , Consider

and therefore ta r: Ih = lah' Also

Ie (x) = ex = x = id(x) and hence Ie =id.

Now , ta 0 ta -1 = t aa -1 =te =.1'd. = ta -1a = t a -1 (' '-ca'

I and hence ta-I = t -I

These (1), (2) and (3) imply that H is a subgroup of A (x) .

We shall prove that G == H . for this, define

 IDistance Education
'--
4 Acharya Nagarjuna University 1
"

IC; -"t H by f(a) =10 for any aEG.

Clearly f is a surjection, Also, by (1) above

and hence f is a homomorphism. Also, for any a, b e G ,

=> ae = he => a = h.
Therefore f is an injection too. Thus f :G ~ H is an isomorphism and G is isomorphic

to the subgroup H of A (x) .

)

The above theorem enables us to identify any abstract group as a more concrete object,
namely, as a group of permutations. However, there are certain short comings; for if C; is a group

of order II, then the group A (X) , considered in the proof of the above theorem, has n! elements.

Our group (; of order 11 is some what lost in the group A (X) which is huge in comparison to G.

Now, one can ask the question: 'Can we find a more economical X so as to find smaller A (X) in
which C; can be identified'. This is accomplished in the following, which is actually a generalization
of theorem 3.3.1.

4.3.2 Theorem:

Let G be a group and H a subgroup of G. LetX be the set of all left co sets of H in G. Then

there is a homomorphism f of C; into A (X) such that the kernel of f is the largest normal
subgroup of G which is contained in H

Proof :We are given that X = {xH / x E C}. For any a E C;,

define go X ~ X by go (xH) = (ax) H for any xH EX. First notice that, ga is well-
defined, in the sense that g a (xH) does not depend on x, but it depends on the whole coset xH ,
in which x is a particular element, for

xH = yH => x -)y E
H
 5 AlgebraD

=> (ax -1)( aYJ:)=- x -1a -1ay = x -1y E H

=>(ax)H = (ay)H
Therefore ga is well defined. Now, we shall verify that ga is a bijectioh of X onto itself. For

any xH , yH EX,

=> xH == yH

Therefore s; : X ~ X is an injection. Also, for any yH EX, (a -1y ) HEX and

Therefore ga is a surjection too. Thus ga is a permutation on X; that is, ga E A(X) for

each a E G. Now, define

I :G ~ A ( X) by I (a) =:= g a f?~:~ny a E G. For any a, bEG and xH EX, we have

and hence ga ogb ==gab. This says that I(ab) = I(a) I(b). Therefore I is a

homomorphism. Now, let us evaluate the kernel of I.

Ker 1= {aEG If(a) =id}

={aEG/~a=id}
r:

= {a E G / s: (xH) = xH, for all xH EX}

= {a E G /axH = xH for all x E G}

 (Distance Education 6 Acharya Nagarjuna University )

= {a E G / x -1ax E H for all x E G}

Being the kernel of a homomorphism, ker f is a normal sub gorup of G (see 1.3.3) and

clearly Ker f cH (for, a E Ker f ~ x-lax E H for all x E G => e-lae E H => a E H).
Also, if N is a normal subgroup of G which is contained in H, then N <;;;;: Ker f
(for, a E: N => x -lax E N c H for all x E G => a E Ker·f). Thus, Ker f is the largest normal

subgroup of G which is contained in H.

3.3.3 Self Assessment Question: Deduce Cayley's theorem from the abovetheorem 3.3.2

3.4 APPLICATIONS OF CAYLEY'S THEOREM :

We shall apply the Cayley's theorem of the previous section in proving certain results on
the structure of certain finite groups. Let us begin with the following:

3.4.1 Definition: A group G is called simple if it has no non trival normal subgroups; that is, if {e}
and G are the only normal subgroups of G.

3.4.2 Examples: The trivial group {e} is clearly simple. Also, any group of order a prime is simple;
. for if G is of order p and p is a prime and if H is a subgroup of G, theno(H) divides o( G)= p (by
the largrange's theorem) and hence o(H)=1 or p, so that H={e} or H=G.

3.4.3 Self Assessment Question : For any positive integer n> I, prove that the group 7l n is
simple if and only if n is a prime number.

Let us recall that the order of a subgroup H of a '1i;,)"tegroup G divides the order of G and

o( G)/ is equal to the number of distinct left (right)\cosets of Hin G. Also, o( G){(H) is
/o(H) . /o(
known as the index of H in G and is denoted by i(H). We shall use theorem 3.3.2 in proving the
following:
/'

3.4.4 Theorem: Let H be ~ proper subgroup of a finite group G such that o(G) does not divide
i(H)!. Then H contains a n9n trivial normal subgroup of G and, in particular, G is not simple.
. /'
Proof: LetX be the set of an left cosets of H in G. Then IXI = f(H) and o( A( X)) = i(H)!. By

theorem 3.3.2, there exists a homomorphism f: G ~ A (X) such that Ker f is the largest normal
subgroup of G which is contained in H. Since H :F. G and Ker t s: H , it follows that Ker f is a
fiLesson -3 7 Algebra »
proper normal subgroup of G. If Ker j = {e} , then j is a monomorphism and hence G == j (G) /

and j(G) is a subgroup of A(X) and therefore o(G) (= o(j(G))) divides o(A(X)) = i(H)\
which is a contradiction to the hypothesis. Therefore Ker j 7; {e}. Thus H contains a non trivial
normal subgroup of G (namely, Ker j) which implies that G is not simple.
3.4.5 Example: Suppose G is a group of order 36 and G has a subgroup H of order 9. Then

i ( H) = ; iZ~ = 4 and hence o( G) does not divide i ( H) !. Therefore there is a nontrivial normal

subgroup G contained in H (by theorem 3.4.4) and G is not simple. Later, we shall prove that any
group of order 36 has a subgroup of order 9. After proving this, we can state that any group of order
36 is not simple.
3.4.6 Self Assessment Question: Suppose G is a group of order 175 and G has a subgroup of
order 25. Then prove that G is not simple.

3.4.7 Example: Let H be a subgroup of order 11 in a g'roup G of order 99. Then

o(G) 99
i(H) = -(-) =- =9 o(G) does not divide i(H)! .
.> •

and Then, by theorem 3.4.4, there exists a

o H 11 .
non trivial normal subgroup N of G which is contained in H. Since o(H) = 11, which is a prime and
since o(N) is a divisor of o(H), it follows that o(N)=o(H) (note that 0 (N) > 1, since N is non trivial)
=
and hence N H. Thus H is a normal subgroup of G. To summarise, any subgroup of order 11 in
a group of order 99 is normal.
3.4.8 Self Assessment Question: Prove that any subgroup of order 13 in a group of order 65 is
normal.

3.4.9 Theorem: Any non abelian group of order 6 is isomorphic to A (X) , where X is a three
element set.

Proof: Let G be a non-abelian group of order 6. Then Ghas an element of order 2 (see 3.4.10

below). Let a E G be an element of order 2. That is, e 7; a E G and a2 =e.

Put H ={e, a}.

o(G) ..
Then H is a subgroup of order 2 in G and i ( H) = 0(H) = 3. Let X bA the .set of left coset,

ofHinG. Then A(X) is a qroup ot order ji (since X has 3 elements); that is, o(A(X)) =3. By
 (Distance Education 8 Acharya Nagarjuna University )

theorem 3.3.2, there is a homomorphism f :G -4- A(X) such that Ker f is the largest normal

subgroup of G which is contained in H. In particular, Ker f <:;:;;: Hand Ker f is a normal subgroup
of G. Since o(H) = 2, it follows that Ker f = {e} or Ker f = H. We shall argue that Ker f *- H .

Otherwise, suppose Ker f = H. Then H is a normal subgroup of G and hence xax -1 E H

forall XEG. But xax-1:t:e (since a*-e) and therefore xax-1 =a or xa=ax for all XEG.
Therefore a E Z (G), the centre of G and H ~ Z (G) <:;:;;: G. Since o( H)= 2, we have that 2

divides o( Z (G)) and o(z (G)) is a divisor of o( G)=6. From these, itfollows that o( Z (G) )=2 or

6. But o( Z (G))*- 6 (for, since G is non abelian, Z (G) is a proper subset of G). Therefore

o( Z (G) )=2 and H =Z (G). Now, choose bEG such that b ~ H and consider the normalizar

N(b) of s,

i.e., N(b) = {XEG / bx = xb}

Then Z (G) is a proper subgroup of N (b) (since bE N (b) and b ~H = Z(G)) and hence

2 is a proper divisor of o( N(b)} and o( N (b)) is a divisor of o( G)=6. Therefore

o(N(b)) = 6 = o(G) and hence G=N(b) so that xb=bx for all x E G; that is, b e Z(G)= H,

which is a contradiction to the choice of b. Thus Ker f *- H and hence Ker f = {e} .
Therefore, f is a monomorphism of G into A (X) and hence G == f (G) <:;:;;: A (X). But

o(J(G)) = o(G) = 6 = o(A(X))

Therefore, f(G)=A(X) and hence f:G -4- A(X) is a surjection too, Thus f is an

isomorphism of G onto A (X) and G == A (X) .

3.4.10 Self Assessment Question: Prove that any finite group of even order has an element of
order 2.

Actually, we prove later a more general result than 3.4.10. If G is a finite group and p is a
prime dividing o(G), then G has an element of order p. This is nothing but the Cauchy's theorem
(1.6.1) for a general group-which will be proved later. However, you can try to prove 3.4.10 by
elementary methods.
fiLesson -3 9 Algebra D,

3.5 SUMMARY :
In this lesson, you have learnt that the permutations on any set form a group under the
composition of mappings and the Cayley's theorem which states that any group is isomorphic to a
group of permutations on a suitable set. We have also proved a generalised version of Cayley's
theorem and applied this to prove thatcertain groups are not simple; in particular, we have proved
that any non abelian group of order 6is isomorphic to the group of permutaions on a three element
set.

3.6 TECHNICAL TERMS .

Permutations

Group of permutaions A (X)

Cayley's theorem

Simple group

3.7 MODEL EXAMINATION QUESTIONS

3.7.1 Prove that the set A (X) of permutations on a setX is a groupunderthe composition of
mappings.

3.7.2 State and prove the Cayley's theorem.

3.7.3 LetHbe a subgroup of a group G andXbe the set of all left cosets of Hin G. Then prove
that there is a homomorphism f of G into permutation group A (X) 'such that ker f is
the largest normal subgroup of G that contained in H.

3.7.4 Let H be a proper subgroup of a fintie group G such that o(G) does not divide i(H)!.
Then prove that G is not simple.

3.7.5 Prove that any non abelian group of order 6 is isomorphic to the group of permutations on
a three element set.

3.8 ANSWERS TO SELF ASSESSMENT QUESTION'S :

3.2.3 Give that X = {1, 2, 3} consider the maps id, a, f3 of X into X as defined below.

id(l)=l a (1) =2 P (1) = 2

 (Distance Education 10 Acharya Nagarjuna University )

id(2)=2 a (2) =1 f3 (2) = 3

id(3 )=3 a (3) = 3 f3 (3) = 1

Then id , a, fJ E A (X) and a 0 f3, f3 a0 and p2 are all other elements in A (X). Thus

3.2.4 Apply induction on n.

3.3.3 Take H = {e} in theorem 3.3.2. Then Ker f ~ {e} and hence Ker f = {e} , so that f is an

injective homomorphism of G into A (X) , where X is the set of left cosets of H in G. Note: that,
in this case X is bijective with G.

3.4.3 Suppose n is aprime number. For any subgroup H of Z n ' o(H) is a division of 0 (z n) ::::n
and hence o(H)=1 or n so that H = {O} or H = 7Ln. Therefore 7Ln has no non-trivial (normal)
subgroups and hence 7Ln is simple. Conversely, suppose that n is not a prime. Then n=mk for

some m.k > 1. Consider H ={m,2m, .. · ,(k-l)m}. ThenHis a non trivial normal subgroup

of 7Ln and therefore Zn is not a simple group.

3.4.6 LetH a subgroup of G. o(G) = 175 and 0(11) = 25. Then i(H) = ;~~~ = 7 and o(G) does

not divide i (H)!. By theorem 3.4.4, there exists a non trivial normal subgroup N of G such that

N ~ H . Since H * G , follows that N * G. Thus G is not simple.

3.4.8 Imitate the argument given in 3.4.7.

3.4.10 Let G be a finite group and o(G) be even. If G is an abeiian group, then we can use the
Cauchy's theorem (1.6.1). If the Cauchy's theorem (1.6.1) is proved for an arbitary group, then we
are through (infact we prove 1.6.1 for a general group later). But here is an elementary proof.

For each a E G, let Aa = {a, a-I}. Then the Aa 's form a partition of G; that is, for any a,

bEG, either Aa = Ab or Aa n Ab =¢ and the union of all Aa's is G. Aa may consists of only

oneelment(thishappenswhen a=a-1). For example Ae ={e}. If, for each a'*e, Aa is a two
 11 Algebra DI
element set then 0 (G) = 2m + 1, a contradiction to the hypothesis that o(G) is even. Therefore,

2 == e and
there exists a -:j:. e such that Aa is a sigleton set; that is, a = a -1 . Thus a -:j:. e and a
hence a is of order 2.

3.9 EXERCISES :

3.9.1 For each element a in a group G, define sa :G ~ G by Sa (x) =xa for all x E G.

Then prove that sa is a bijection and that sab =sb 0 sb for any a, bEG (compare with

fa'S defined in the proof of 3.3.1).

3.9.2 For each element a in a group G, define fa : G ~ G by fa (x) = ax for all x E G, as

in the proof of 3.3.1. Then prove that, for any a, bEG, fa oSb = Sb a fa (where sa is as
in 3.9.1 above).

3.9.3 For any bijection f :G ~ G , prove that f 0 sa = Sa f 0 for-all a E G if and only if .l = tb

for some bEG.

3.9.4 For any bijection f :G ~ G, prove that f ota = fa a f for all a E Gif and only if f = sb
for some bEG.

3.9.5 Prove that any group of order p2 , where p isa prime, is not simple.

3.9.6 Let G be a group of order p2, where p is a prime. Then prove the following:

(i) G has a normal subgroup of order p

(ii) Any normal subgroup of G is contained in the centre of G.

(iii) G is abelian

3.9.7 Prove that any group of order 2p , where p is prime, must have subgroup of order p
and that this subgroup is normal.

.3.9.8 Prove that any two non abelian groups of order pq , where p and q are distinct primes,
are isomorphic.
 [Distance Education 12 Acharya Nagarjuna University J

3.10 REFERENCES
-;
! 1. Topics in Algebra I.N. Hersiein, Second edition, John wiley & sons, New
York, 1975.

2. Basic Abstract Algebra PB. Bhattacharya, S.K. Jain andS.R. Nagpaul, second
edition, Cambridge Un-iversity Press, 1995.

3. Lectures in AbstractAlgebra - N, Jacobson, Vol -I, D. Van Nostrand Company, Inc.,

London, 1964.

4. Algebra _Serge Lang, Addison-Wesley publishing Company,

London 1980.
5. Algebra TT. Moh, World SCientific, Singapore, 1992.
6. University Algebra N.S. Gopala Krishnan, Second edition, wiley Eastern
Limiter', New Delhi, 1986.

7. Abstract & Linear Algebra - David M. Burton, Addison -Wesley publishing company,
London 1972.
8. Algebra M. Artin, Prentice - Hall of India, New Delhi 1994.

9. Algebra J.W. Archbold, Fourth edition, the English language book

society, 1972.

10. A First Course in Abstract Algebra - John B. Fraleigh, Addison - Wesley Longman, lnc.,
Fifth edition, 1999.

Course Writer
Prof: UM. Swamy
Andhra University
Visakhapatnam
 (

'Lesson - 4 Permutation Groups

Objectives:
After reading this lesson, you should be able to

(i) define the concept ofthe symmetric group Sn of degree n

(ii) prove that any permutation on a finite set is a product of disjoint cycles and

(iii) define the notion of the alternatinq group An and prove that An is a normal subgroup

of index 2 in Sn'

/
Structure:
4.1 Introduction

4.2 The symmetric groun Sn

4.3 Cycles and transpositions

4.4 The atternating group An

4.5 Summary
4.6 Technical Terms
4.7 model examination questions

4.8 Answers to Self Assessment Questions

4.9 Exercises

4.10 References

4.1 INTRODUCTION :

In the previous lesson, we have proved that the set A (X) of permutations on any setX is a
group under the composition of mappings and also proved the Cayley's theorem which state that
any abstract group is isomorphic to a subgroup of the permutation group A (X) for a suitable set
X; infact, we have taken X to be the underlying set of the group itself. In particular, any finite group
is isomorphic to a group of permutations on a finite set. For this reason, the groups of permutations
on finite sets become prominent in the theory of structure of finite groups. In this lesson, we
discuss these groups and prove certain important properties.
 ijDistal1ce Education 2 Acharya Nagarjuna Universiti]

4.2 THE SYMMETRIC GROUP Sn:

If X and Yare sets and if there is a bijection a :X ~ Y, then the groups A (X) and A (Y) of

permutations on X and Y respectively are isomorphism under the isomorphism f H a 0 I= a-I.

Therefore" if Xis a finite set with n elements, then the permutation groups A(X) and A (In ) are
isomorphic, where In = {1, 2, ....., n} .

4.2.1 Definition: Let n be any positive integer n and In = {I,2, , n}. Then the

permutation group A (In) is called the symmetric group of degree n and is denoted by Sn'

The elements of the symmetric group Sn are permutations on In = {I, 2, , n}. If

f E Sn ,then f(1), f(2), , f(n) are n distinct elements of In and hence

In = {f(I), f(2), , fen)}. For convenience, we describe f e S; by

. For example f 6 S9 defined by f(1) = 4, f(2) = 6, f(3) = 9, f(4) = 5, f(5) = 7, f(6) = 1.

f(7) = 3, /(8) = 2 andf(9) = 8 is denoted by

1 2 3 4 5 6 7 8 89J
f= (4 6 9 5 7 1 3 2

4.2.2 Self Assessment Question : Let

,f E S7 be givAn by f= (1
, 5
what isf(3) andf(6).

Iffandgarepermutationsin Sn' then recall thatfog is defined by (fog)(i) = f(g(i)) and

that Sn is a group under this composition of mappings.

4.2.3 Self Assessment Question: Let f and g E S8 be piven by

f=
'(I 2 3 4 5 6 7 8)
57134682
 ' ,.,,:.

flL9SS0n -4 3 Algebra »

Then compute I g
0 and I-I

4.2.4 Self Assessment Question: Compute (~ : : ~ ! ~)

G ~~~:
0 ~}n S6

4.2.5 Note: Let nand m bel positive integers and n <m . Define () : Sn ~ Sm by

For any f E Sn' Then, as you can easily check, () is a monomorphism. In other words,

Sn is isomorphic to a subgroup of Sm and Hence, we can identify permutations in Sn with those

4.2.6 Example:

The ~ermutation (~ ~. ~ ; ~) in S5 can be identified with (~ ~ ~ ; ~ : ~ !) in

S8'

4.3 CYCLES AND TRANSPQSITIONS :

In this section, we shall discuss a special type of permutations, namely cycles. These
cycles playa very important role in the study of permutations. Infact, we shall prove that any
permutation can be expressed as a product of disjoint cycles in a unique way, in some sense. Let
us begin with the following.
4.3.1 Definition:

Let n be any positive irlteger and al> a2,'" .a; be distinct elements in

In = {I, 2, , n} and 1'>1. Let f :In ~ In be defined by

r
1 if «=« 1::::;;<,..

f(a) =: :1 if «=»-
a ai,
if ;t. l::::;i::::;r
 (lliStance Education 4 Acharya Nagarjuna UniversitYJ 1
Thenf is a permutation on In and is denoted by (al a2 a3 !.. ar). This is called an
r - cycle or a cycle of length r.

For example, the cycle f = (3 4 2 6 8) is a permutation in S8 (or in S; for n ~ 8)

defined by

f(3) = 4, f(4) = 2, f(2) .= 6, f(6) = 8, f(8) = 3 and

/3~'
8 4
f(a) = a for all a (j:. {3, 4, 2, 6, 8}.
\ 6~ )
Note that the cycles (3 4 2 6 8) and (2 6 8 3 4) are same.

4.3.2 Self Assessment Question : Let f = (5 2 7 3 6 8), a cycle in SlO' Then what is

f(i) for any Ls r s Iu ?

, 4.3.3 Self Assessment Question : Is the, permutation f ~ G ~ ~ ~ ! :;~:) a

cycle in S9 ? If so, writef in the cycle notation.

4.3.4 Definition : Two cycles f = (al a2 a3 ar) and g = (bi h2 hs) are said to be
disjoint if aj :t:- bj for any 1~ i ~ rand 1~ j ~ s ; that is, f and g are called disjoint if the sets

{aI, a2, ,ar} and {q, b2, , bs} have no common elements.

4.3.5 Examples : The cycles (4 6 3 5 2) and (7 1 8 9) are disjoint while the cycles

(5 3 4 2 8) and (4 1 6 7) are not disjoint.

4.3.6 Note: If f and g are disjoint cycles in Sn' then fog = g ~f ; for, if f = (aI, a2 ar)
and g = (ht b2 bs) with aj :t:- bj , then

(fog)(a)-= f(a) = a = g(a) = (gof)(a) if a:t:-aj and a:t:-aj

and (fog)(aj) = f(ad = ai+l = g(a;+d = (gof)(a;)

and, similarly (f 0 g) (b j ) = (g oJ) (bj ). Thus fog. This is to say that any two disjoint.
cycles commute.
fiLesson - 4 5 Algebra]

Now we shall prove the. following theorem which expresses any permutatison as a product
of disjoint cycles. Remember that an r - ecycle is defined only when r > 1 .

4.3.7 Theorem':

Let j be any non-identity permutation in Sn : Then j can be expressed as a product of

disjoint cycles each of length> 1. Also, this expression is unique in the sense that, if

j = al 0 a20 0 ak

and j = PI 0 P2 0 0 Pm
where ai's and Pj 's are disjoint cycles, then k=m and there is a permutation (J in Sm

such that Pj = aO"(j) for each 1~ j ~ m .

Proof: Let T = {a E In / j (a)::F-a} . Since j is not the identity map, it follows that ¢::j; T ~ In'

Choose al E T and consider

These are.all elements of -/n and hence these can not all be distinct (since / n is finite).

Therefore, there exists m-c k such that jm (al) = jk (al) and hence r: (ad = al and

k - m > O. Therefore there is a positive integer r such that jr ( al) = al' Now, let rl be the least

positive integer such that jrl (al)' =; al. Then rl > 1, since j (al) ::j; al. Consider the rl - cycle

j = al' Otherwise, we can choose a2 E T -Ii and repeat the above procedure to construct an

r)cycle. '<,

a2 = (~2 f(a2) j2 (a2)····.· .. ········./r2-1 (a2)) , where jr2 (a2) =a2' Againjanda2 are

equal on the set T2 = {a2', j(a2)' >o •••• .r=' (a2)}. Also, Ii and T2 are disjoint and hence
 (Distance Education 6 Acharya Nagarjuna University J
al and a2 are disjoint cycles. If Ii U T2 * T, we continue the procedure to construct cycle a3
and so on. since T is finite, this process should terminate at a finite stage; that is, we get pairwise
disjoint sets 71, T2' , TK and cycles at> , aK such that

Ii U T2 U .: UTk =T
aj = (aj j(aj)J2(aj) j1)-l(aj)), j1) (aj) = aj

T, = {aj, j(aj), j2(aj), ,j1)-1(ai)}

andf and a, are same on Tj.

All these data gives us that

j = al 0 a2 0 0 ak

and al , a2, , ak are pair-wise disjoint cycles (note that each ai is identity on Tj
for t= i).

To'prove uniques, suppose f = /31 0 /32 0 0 /3m be another expression of f as

a product of disjoint cycles pj's. Suppose m ~ K. Suppose a E Jn such that /31 (a) :f.a ( i.e. a
. I

is involved in the cycle PI)' Then P j (a) =,a for all i : 1 and
f(a) = (/310132 0 o/3m)(a) = fh(a):f.a

and hence there exists ti. such that ail (a) :f. a and Cfj (a) =a for all i * il. Then, it

follows that PI = ajl and by the cancellation law,

/32 0 0 Pm = .Jr. aj
l*lI

Now, let us continue the above procedure to get }Z, 13, .i; such that Pi = aji for

2 ~ i ~ '!l. If m <k , we get that the product of certain a j'S is identity, which is a contradiction.

Therefore m=k and iH ii is a permutation of {I, 2, , m} such that /3j = all' This
completes the proof of the theorem .

. 4.3.8 Self Assessment Question: Imitate the procedure given in the above proof to express the
following permutations in S12 as products of disjoint cycles:
filesson -4 7 Algebraj--

(1) f~G 2 "-'

2 7
4
.~~

6
5
10 4
6 7 8 9 10 1112J
1 12 5 9 11 8

2 3 4 5 6 7 8 9 10 11
(2) g ~(:
5 3 7 2 6 9 10 4 12 11 :2J

(3) h~G 2 3 4 5
2 1 11 7
6

4
7 8 9 10 11
10 8 9 6 3 12
12J

4.3.9 Definition: A 2-cycle is called a transposition. That is, for any a 7:. b E In the permutation

wh±::h maps a to band b to a and keeping all the other elements of In fixed is called a
transposition.
For example the 2-cycles (4 5),(3 4), (5 9) are all transpositions. You can easily check

that any r - cycle a = (al a2 c , ar) can be expressed as

which is a product of r -1 number of transpositions.

4.3.10 Self Assessment Question : Express the cycle (6 9 4 2 7 5) as a product of'

transpositions.

4.3.11 Theorem: Any permutation inSn (n>l) can be expressed as a product of transpositions.

Proof: Let n » 1 and f be a permutaion in Sn : If f is identity then

f = ( a b) 0 (a, b) for any a 7:. b e I n .

Suppose f is not the identity. by theorem 4.3.7, we can write

f = al 0 a2 0 0 as

where aJ, a2, .a; are cycles and each cycle a, is a product of transpositions.
Therefore f is a product of transposition.

4.3.12 Self Assessment Question: If a is an r - cycle, then prove that o(a) = r and that the
 (Distance Education 8 Acharya Nagarjuna University J

order of any transposition is two.

4.4 THE ALTERNATING GROUP An:

In the previous section, we have proved that any permutation can be expressed as a product
of transposition. But this expression may not be unique as in theorem 4.3.7. However, we have the
following:

4.4.1 Theorem: Let 1be a-permutation in Sn and

f == al 0 a2 0 oak and f = PI 0 P2 0 0 Pm

be two representations of f as products of transpositions ai's and Pj 'so Then k is even

if and only if m is even.

Proof: We can suppose that n» 2, since the theorem is trivial for n = 1 or 2. Consider the
polynomial in n - variables given by

P = P(xl , x2 , : , xn) = lr(xi - Xj)

t<j

For any permutation 1in Sn' let

Clearly (f g)( p)
0 == f (g (p )) for any f, g E Sn . You can easily check that a (p) = - P

for any transposition a in Sn .

Now, if f= al 0 a2 0 0 ak = PI 0 P2 0 0 Pm' then

k m
(-1) P = 1(p) = (-1) P

. and' hence (_l)k = (_l)m which implies that

k is even <=> m is even.

4.4.2 Self Assessment Question: Let n=5. In this case, what is the polynomial p giveni.n the

above proof. What is j(p) if

(1) 1=(3521)0(24)
 @:sson;4 9 AI9~)

(2) ! = (2 5)
4.4.3 Definition : A permutation ! is called an even permutaion if ! can be expressed as a
product of even number of transpositions. ! is called odd if it is not even.

In theorem 4.4.1 ,we have proved that, if! is a product of even number of transpositions,
then it cannj be a product of odd number of transposition. Therfore, the concept of even permutation
(4.4.3) is well defined. also, in view of theorem 4.3.11, any permutation is either even or odd.

4.4.4 Self Assessment Question : Determine which of the following permutations are even?

(1) f~(: 2
7
3
8
4
6
5
5
6
4· 1
7
~)
(2) g = (3 5 4) 0 (7 6)0(2 8)

(3) h = (3 2)0 (7 4)0 (1 8)0 (2 3 8) 0(4 6)

(4) a = (5 8 2 6 4 7)
:"~

4.4.5 Self Assessment .Question : Prove that an r - cycle is even if and only if r is odd.

4.4.6 Theorem: For any n » 1, the set all even permutations is a normal subgroup of index 2 in Sn.

Proof: Consider the group G = {I, -I} under the usual multiplication of real numbers. Let An be

- ~ot of all even permutations in SI1" n > 1. Define e : Sn ~ G by

o( )= {I if! is even
! -1 if! is odd

ince! 0 g is even if and only if either both! and g are even or both of them are odd, it
.iat e is a homomorphism. Since 1 is the identity element in the group G, we have

Therefore An is a normal subgroup of S; (by theorem 1.3.3). Also, by the fundamental

l. ,_ rem of homomorphisms (1.5.1), we have

Sn/ =Sn/
/ An /Ker e =G
- .
-:
 (Distance Education 10 Acharya Nagarjuna University J

. Sn/
Since G is a two element group, so is the quotient group ;; An . since

ISnl = IAn IIS~n I = IAn I·2 , it follows that An is of index 2.

4.4.7 Definition: The group of even permutation in Sn' is called the alternating group of degree n
/'

and is denoted by An .

n!
4.4.8 Self Assessment Question: For any n» 1, prove that o( An) = -
2

4.4.9 Self Assessment Question: List all the elements of A2 , A3 and A4.

4.5 SUMMARY :
In this lesson, we have introduced the notion of the symmetric group Sn and proved that

any permutation in Sn can be expressed as a product of disjoint cycles. Further, we have introduced
the notion of an even permutation and proved that the set of all even permutations is a normal
subgroup of index 2 in Sn .

4.6 TECHNICAL TERMS .

Permutation

Symmetric group S;
").'
,\
Cycle
Transposition
Even permutation
Odd permutation

Alternatinq group An
Disjoint cycles

4.7 MODEL EXAMINATION QUESTIONS:

4.7.1 Prove that any permutation in Sn can be uniquely expressed as a product of disjoint cycles
d~~sson- 4 11 Algebra])
•

4.7.2.: Define the notion of the alternating group An and prove that it is a normal subgroup of
index 2 in Sn'

4.8 ANSWERS TO SELF ASSESSMENT QUESTION's

4.2.2 1(3)=1 and /(6)=3

4;2.3 (1og)(1) = J(g(I)) = 1(4) = 3, (1og)(i) = 1(g(i)), .

fog:G ~ ~ ~ ~ ~ : :J
r' : G : ~ ; : : ~ ~J
4.2.4

1 2 3 4 5 6
-J, -J,-J, -J, -J, -J,
5 641 3 2
-J, -J, -J, -J, -J, -J,
. 6 5 241 3

t ,yo

--
I
._~ 1
4.3.3 1= (1 5 4 7 9 8 '2)
~:' ...
__
j

4.3.8 (1) Choose a such that / (a) :;t: a . For example / (4) :;t: 4 'and consider 4,1(4), f2 (4) ,

........ so on until you set 4 again. In this example, we have 4,/(4)=6, f2( 4) = f (6) = 4.

Now, you have the cycle (4 6). Take some other a, other than 4 or 6, such that 1(a) :;t: a .
For example, we can take 5, since 1(5):;t: 5. Then

5, f (5) = 10, f2 (5) = 9, f3 (5) = 5 .

 Distance Education 12 \ Acharya NagarjunaUniversity .r

\
. Then we have the cycle (5 10 9)\ N~xt pic 3 'to\set

8, f (8) = 12, f 2 (8) = 8 . ,

-,
\" <,
and consider the cycle (8 12). t~,~,oth~e~m~names 2 and 11 are fixed by f Thu~,
we have . .

(1). f = (4 6) 0 (5 1'0 9) 0 (3 7 1) 0 (8 12)

(2) g=(1 8 10 )2)0(2 5)0(4 7 9)

(3) h = (1 5 7 10 6 4 11 3)

4.3.10 (6 9 4 2 7 5) = (6 5) 0 (6 7) 0 (6 2) 0 (6 4) 0 (6 9)
'-....
4.3.12 Let a = (al a2 ..·..·.ar)' an '<cycle. Then
'- "

For any t s i s; r , a = (aj, ai+l ar al a2 aj) -and hence ar (aj) =aj for every

1:s; i :s; r . Thus tx" = id . Also, a' * id for any i <r and hen,~~0 (a) = r. If a is a transposition,

then it is a 2 - cycle and nence 0 (a) =2 .

4.4.2 Consider the polynomial

p = P(~l' x2, ...~..... ,xs)= ?r(xj -Xj)

"'\ I <j

i.e. P = (XI-X2)(Xl -XZ,)(XI-X4)(Xl-XS)(X2 -X3)(X2 -X4)(X2 -xs)

. (X3 - x4 ) ( x3 - Xs ) ( x4 - Xs )

(1) j ~ G ~ ~ ~ ~) and hence

fiLesson - 4 13

f (p) = (X3 - x4 ) ( x3 - Xs ) (X3 - Xl) (X3 - x2 ) ( x4 - Xs ) ( x4 - xl ) ( x4 - X2) .

(XS -XI)(XS -X2)(XI -X2)

(2) f = (2 5) and hence

f(p) = (x1 -XS)(Xj -x3)(Xj -X4)(xI -X2)(XS -X3)(XS -X4)(XS -X2)

(X3 -X4)(X3 -X2)(X4 -X2)

4.4.4 (1) f = (1 3 8 2 7) 0 (4 6)

= (1 7) (1 2) (1 8) (1 3) (4 6)
0 0 0 0

Therefore f is odd

(2) g is odd (since (3 5 1 4) is odd)

(3) h is even

(4) a is odd, since a is a 6 - cycle.

4.4.5 A~y r - cycle a = (Cl] Cl2 Cl3 ar) can be expressed as

a = (at a )"
r (a] ([r-l) o o(ala2) which is a product of r-1 number of

transpositions.
--
Thus a is even .<=> r - 1 is even

<=> r is odd

4.4.8 An is a normal subgroup of index 2 in S; (by 4.4.6)

and henc~ 2 = o(Sn~ ) = o(Sn) = n! . Therefore o( An) = n!

/ An ()( An ) ()( An ) . 2

4.4.9 A2 = {id}

A3 = {(I 2 3), (2 1 3), id}

id,(1 2 3), (2 1 3), (2 3 4), (3 2 4), (3 4 1),

A4

l ( 4 3 1), (1 2)
(1 2- 4), (2
0

l,~
(3 4), (1

\. ,
3) 0 (2 4), (1 4) 0 (2
(Distance Education . 14 Acharya Nagarjuna University J

4.9 EXERCISES:
4.9.1 Prove that the order of any I E S; is the l.c.m of the lengths of its disjoint cycles.

4.9.2 Prove that two cycles in Sn are conjugate if and only if they are of the same length.

4.9.3 Determine all subgroups of S3 and S4'

4.9.4 Prove that Z(Sn) = {e} for any n23.

4.9.5 What isthe centre of S2?

4.9.6 Prove that 1-'1 is a cycle if I is a cycle.

4.9.7 Compute I g/-1 where I = (1 3) (2 5 4) and g = (5 7 9 1)

4.9.8 Let 1= (2 3) (4 1) and g = (1 3) (4 6). find hES6 such that h-l I h =g

4.9.10 Prove. that there is no I in SlO such that i-I (2 3 6)1 = (7 9) (4 5 1 8)

4.9.11 Prove that the smallest subgroup of Sn containing (12 3 n) and (1 2) is Sn itself.

4.9.12 Find all normal subgroups in S4'

5.10EXERXCISES:
1. Topics in Algebra I.N. Herstein, Second edition, John wiiey & sons, New
York,1975.
2. Basic Abstract Algebra P.B. Bhattacharya, S.K. Jain and S.R. Nagpaul, second
edition, Cambridge University Press, 1995.
;

3. Lectures in Abstract Algebra - N. Jacobson, Vol -I, D. Van Nostrand Company, Inc.,
London, 1964.

4. Algebra Serge Lang, Addison-Wesley publishing Company,

London 1980.
 »
/

~Lesson - 4 15 Algebra

5.' Algebra T.T./Moh, World Scientific, Singapore, 1992.

6. . University Algeora N.S. Gopala Krishnan, Second edition, wiley Eastern

Limited, New Delhi, 1986 .
. 7. .. Abstract & Linear Algebra - David M. Burton, Addison - Wesley publishing company,
London 1972.
8. Algebra M. Artin, Prentice - Hall of India, New Delhi 1994.

9. Algebra J.W. Archbold, Fourth edition, the English language book

society, 1972. /

10. A First Course in Abstract Algebra - John B. Fraleigh, Addison - Wesley Longman, Inc.,
Fifth edition, 1999.

Lesson Writer
Prof tiu. Swamy
Andhra University
Visakhapatnam
Lesson - 5 A COUNTING PRINCIPLE

Objectives:
After reading this....lesson,~. you should be able to
(i) define the cor~cepts of conJugatec1ass and normalizer of an element of a group
(ii) obtain the class equation of a finite group
" "

(iii) prove that the centre of a group of order pr (p is prime, r >0) is nontrivial and that

"any group of order p2 is abelian

(iv) state and prove the Cauchy's theorem for a general finite group

(v) determine the conjugate classes in '~"11'

Stnicture:
5.1 Introduction
5.2 The conjugacy relation
5.3 The class equation of a finite group

5.4 Groups of order r:

5.5 Canchy's theorem

5.6 Conjugate classes in S;

5.7 Summary
5.8 Technical terms
5.9 Model examination questions
5.1-0 Answers to Self Assessment Questions
5.11 Exercises
5.12 References

5.1 INTRODUCTION:
We shall illustrate a counting principle by introducing an equivalence relation on a finite set,
measuring the size of each of the equivalence classes under this relation and then equating the
number of elements in the set to the sum of the orders of these equivalence classes. We shall
 ~nce Education 2 Acharya Nagarjuna University )

introduce an equivalence relation on a given finite group:; and find an element description for the
size of each equivalence class. Using these we shall derive an equation known as 'the class
equation' of any finite group and deduce several beautiful, powerful and celebrated results on the
structure of finite groups. In particular, we extend the Cauchy's theorem for finite abelian groups (in
theorem 1.6.1) to an arbitrary finite group.

5.2 THE CONJUGACY RELATION :

In this section, we shall introduce an equivalence relation on an arbitary group and study the
equivalence classes under this relation. Let us begin with the following:

5.2.1 Definition: Let G be any group and a, bEG. Then b is said to be a conjugate of a if there

exists CE G such that b = eae -I. If b is a conjugate of a, we denote it by a - b. This relation - is

called the conjugacy relation on G.

5.2.2 Theorem: For any group G, the conjugacy relation - is an equivalence relation on G.

Proof: (i) - is reflexive; that is a ~ a for any a E G (since a = eae -I)

(ii) - is symmetri ; that is - b ~ b - a ; for

a - b =;> b = cae -I for some e E G

:::::;.
b=a
(iii) - is transitive; that is, a ~ band b - c ~ a - c; for,
a ~ band b - e ~ b = sas-I and c = tbt-I for some s, t E G
.: &;

=;>c=t(sas--l)rl=(ts)a(tsrl and tSEG

5.2.3 Self Assessment Question If G is an abelian group and a, bEG, prove that

a=b c» a=b

5.2.4 Definition: For any element a in a group G, let C (a) be the equivalence class of a w.r.t.
-; that is,

C (a) is called the conjugate class of a.

~Lesson -5 3

5.2.5 Theorem: For any elements a and b in a group.G, the conjugate classes C (a) and C (b)
are either disjoint or identical and therefore the distinct conjugating classes in G form a partition of
G.
Proof: This is a result on general equivalence relations. Any two equivalence classes are either
disjoint or identical. You are adviced to try to prove this on your own.

5.2.6 Self Assessment Question: Prove 5.2.5 ~-

5.2.,7 Definition: Let a be an element of a group G. Define

N(a) = {XEG /xa = ax}

}II (a'j is called the normalizer of a.

5.2.8 Theorem: For any a E G, N( a) is a subgroup of G.

Proof: Since ea x a=ae , eEN(a) and hence N(a) isanonemptysubsetofG.

x, Y E N(a) =>xa=ax and ya=ay

=> xa =ax anday-1 = y-1a

There :ore, N ( a) is a subgroup of G.

- ~:-~ following, we shall establish relation between the conjugate class C (a) and the

normalizer N :a) , for any element a of a group. Let us recall that, for any subgroup H of a group G,

two lett coset~ aH and bH are equal if and only if a-I b E H and that any two left cosets of Hare
either equal .r disjoint. Note that the set of left cosets of H may not form a group (unless H is
normal in G' .nder the usual generation. /
/
5.2.9 Theore-n : Let a be an element of a group G, C(a) the conjugate class and N(a) the

normaiizer or a in G. Then the set C(a) is bijective with the set of all left co sets of N( a) in G.

Proof: L~t usrecau that C( a) = l= -1 / x E G} and

 Distance Education 4 Acharya Nagarjuna University

N(a) = {XEG I xa =ax}

Let X be the set of all left cosets of N ( a) in G; that is,

X={N(a)lxEG}

Define f :C(a) ~ X by

f( xax-1 )=xN( a) for any xax-1 E C(a) , x E G

First observe that, for any x, y E G,

xax -I = yay -I
~ y -Ixa = ay -Ix

~ y-1xEN(a)
,
" ~ xN(a) = yN(a)
and therefore f is well defined. Also,

f{xax-I) = f(yay-I)=> xN(a) = yN(a)

~y-lxEN(a)

Therefore f is an injectieh. Clearly f is a surjection too.

Thus f :C (a) ~ X is a bijection.

~~2.1-Q~SelfAssessment Question: For any group G, prove that the centre Z (G) = n N (a)
. aEG

5:.3' .~Hli/G.LASSEQUATION OF A FINITE GROUP:

We shall now derive the class equation of a finite group which is a powerful tool in proving
several important results in the structure theory of finite groups. For any finite setX, let us write

IXI to denote the number of elements in X.

 flLesson -5 5 Algebra »
5.3.1 Theorem: Let G be a finite group and a E G. Let C ( a) and N ( a) be the conjugate class
and the normalizer of a respectively.

. o(G)
Then IC(a)1 = o(N(a))

Proof :We have C(a) = {xax-1 / x E G} and N(a) = {x E G / xa = ax}

Let X be the set of all left cosets of N (a) in G. Since G is finite, C (a), N (a) , X are all
finite and it is known that

O(G). .
IXI = O(N(a)) = the Index of N(a) In G.

By theorem 5.2.9, X is bijective with C (a) and hence these have the same number of
elements in each. Thus

jC(a)j = IXI = O(G)

. O(N(a))
5.3.2 Theorem (The Class Equation): Let G be atihite group. Then

Where al> a2, , an are such that C(al)' C(a2), ,C(an) are all the distinct
conjugate classes each with more than one element.

Proof :We know that the conjugate classes form a partition of G, that is each element of G is one
nd only one conjugate class. Hence

Where
O(G)

C(ai),C(a2)'
m
= IGI = 2:IC(ai)1
i=l

,C(am)are
C(aJ
Cray 1::&
all the distinct conjugate classes in G. Note

at, for any a E G , we have a E C (a) and hence

7)
. /
 ~-.----'--' ..'.-'.

lDistance Education

IC(a)l= 1 ¢=;> C(a) = {a}

..;::-'> xax-l =afor all x E G
<=-> xa = ax for ail x E C]

¢' C' f' Z (G), the centre of G

Among the aLe,'€: conjugate class C(a1), "C(am), some of them may be

singleton sets (in wnich case that ai must be in the centre Z(Cl)) and the others are with morethan

one element ineach .. We can assume that C(a1), 'C'(({n) (n~m) are all the conjugate
.classes each with more than one element. Therefore,

!7

o(e;) -z: ~IC(ai)1 + O(Z(G))

i=J

Now, by theorem 5.3.1, we have

. 11 O(G)
O(G) = O(Z(G)) + I (N ())
ai
i=l ()

where oJ, ,. .. " an are such that C (ad, ,(' (an) are all the distinct conjuqate
classes each with more than one element.

5.3.3 Example: If C; is an abelian group, then Z(G)=-=Gand any conjugate class isa singleton set
and therefore the class equation doesn't given any new information.
;
Obviously, the class equation is useful only when we consider non abelian groups. In the
following, let us consider the nonabelian group with least number of elements note that any group
of order less than 6 is abelian and-the only non abelian group of order 6 is S3 ' the symmetri group
of degree 3 (see theorem 3.4.9).

5.3.4 Example: Considerthe group 83, We know that o( S3) = 6. Let e be the identity,

/ =(1 2), g=(2 3), h=(3 1) the transpositions, Then

fg = (1 2 3) and gf = (1 3 2)

We have /-1=/, g-l =s. h-1=h, (fgrl=gl and

S3 = {e,j,g,h,.fi,gf}

Note that fh = (I 3 2)=gfand gh = (1 2 3) = /g

 7 Algebra »
(

Let us determine all the conjugate classes in S3 .

C(e) = {e}
C (i) = {~jX -I / x E S3 }

= { £!fie -I '. fff-

.
l . {.-I
,giX,? Ijh-I ,.Ig../g ( ,) f'( g,j.)-I Jl
'( I' ) f'( F)-I ,gj.

= {i, h, s} = C(h) = C(g)

C(ig) = {gl, ./g}
Thus there are three distinct conjugate classes in S3 .

5.3.5 Note: For any element a in a group G, Note that

Ie (a)1 = 1 Q a E Z (G) Q N (a) = G

Q O(N(a)) = O(G) (if G is finite)

(See the argument given in the proof of 5.3.2)

5.4 GROUPS OF ORDER pI':

Groups of order p,., where p is a prime number, play an important role in the structure
" ,.
theory of finite groups. Later; we shall prove that a finite group is:.~forder P if and only if ever one

of its elements is of order pl7 for some n e O. The following is an important property of groups of
r
order P .

5.4.1 Theorem: Let G be a group of order p" where p is a prime number and r is a positive
integer. Then Z (G) 7; {e} .
Proof: Let us recall the class equation of G as given in theorem 5.3.2 we have

pr =O(G)=O(Z(G))+
/
f O(G)
(:] 0 (N ( Gj )) ®
'(Dis,tance Education 8 Acharya Nagarjuna University J

where C (al), , C (an) are all the distinct conjugate classes each with more than one
element;

i.e., IC(aj)1 > I tor all ] <i <n

By 5.3.5, each normalizer N (aj) is a proper subgroup of G. by the Lagrange's theorem,

o (N (ai)) is a divisor of IGI = pI' since p is prime, we have

atG) r
a(N(aj)) =P I for some tj >0.

Thus, by (2),

n
O(Z(G)) = p' - 'I III
i=1

Since rand " are all positive integers, it follows that () (Z (C;)) is an integral multiple of p
and is a positive integer. Thus

and hen ce Z ( G) :;t: {e}

The above theorem can be rephrased as follows:
Any group of prime power order has non trivial centre.
5.4.2 Self Assessment Ouestion.: Prove that any group of order 625 has nontrivial centre.
-,
Recall that any group of order a prime p is a cyclic group and hence an abelian qroup.rn

the following we prove that any group of order 'p2 (where p is a prime) is abelian, though it need not
be a cyclic group.

5.4.3 Theorem: Any group of order p2 is an abelian group, where p is a prime.

Proof: Let C; be a group of order p2 , where p is a prime. Then, by theorem 5.4.1, Z (G):;t: {e} and

hence o(Z(G))>l. By the Lagranges theorem, o(Z(G)) is a divisor of o(G):;t:1:2 He?~e

o(Z(G))=p or p2. We shall prove that o(Z(G)):;t:p. Suppose, if possible, o(Z.(G))=)J.

«Lesson -5 9 Algebra »
Then Z (G) is a proper subgroup of G. Choose a E G such that a (l Z (G). Then there exists

x E G such that ax 7:xa ; that is x (l N (a) , the normalizor of a. Now,

Z(G)c N(a) c G
7- 7-

(since aEN(a)-Z(G) and xEG - N(a). This implies-that

i.e., p<o(N(a))<p2

But o(N(a)) is a divisor of p2. This leads to a contradiction. Therefore our assumption

that o(Z(G))= p is wrong. This implies that o(Z(G))= p2 = o(G) and hence Z(G)=G. Thus

G is abelian.
6.4.4 Self Assessment Question: Prove that any group of order 169 is abelian.
6.4.5 Self Assessment Question: Prove that any group order 961 is abelian.

5.5 CAUCHY'S THEOREM :

You have already learnt Cauchy's theorem for finite abelian groups in lesson 2. This states
that if p is a prime dividing the order of a finite abelion group G, then G has an element of order p.
We shall extend this to general finite groups in the following. . .:c'.

5.5.1 Theorem (Cauchy's Theorem) : Let G be a finite group an:dp be a prime such thatp divides

o( G). Then G has an element of order p.

Proof: We have to prove the existence of an element a in G such that a 7: e and ap = e. Let us

write o( G)= p.k ,where k is a positive integer. Then

p ~O(G)

We shall use induction on o( G) to prove the theorem. Since O( G) is atleast p, we

can start induction atp. If O(G) = p, then G is a cyclic group of order p and hence any non indentity
element is of order p.
(Distance Edu:~ti~n 10 Acharya Nagarjuna University]

Next, suppose that 0 (G) > p and assume that the theorem holds good for all groups of

order less than that of G (that is, if H is a group such that 0 ( H) < 0 (G) and p divides 0 ( H) ,
then H has an element of order p). If G is an abelian group, we are through (by theorem 1.6.1).

Therefore, we can assume that G is not abelian. The Z (G) is a proper subgroup G and

O(Z(G))< O(G).
IfP divides o( Z ( G) ) , then, by the induction hypothesis, Z (G) has an element of order

p and hence so is G. Now, suppose that p doesn't divide 0 (Z (G)). Let us recall the class
equation of G as given in theorem 5.3.2. That is,

11 O(G)
O(G) = O(Z(G)) +~ O(N(a: ))

pi X O(G) .
Since /0 (G) , and P X °(z (G)), it follows that P 0 (N (ai)) for some ai' Since

and since P /0 ( G) , we g.et thatp divides () (N (ai )). Also, in the class equation, ai's are

suchthat!C(a;)I>l and hence O(/a(N(a;))>1 sothatO(N(aj))<,O(G). Again, by the

induction hypothesis, N( a;) has an element of order p and thus so is G.

5.5.2 Self Assessment Question: Prove that there exists an element a 7= e in a group of order
61
793 such that a =e
5.5.3 Self Assessment Question : Prove that any group of even order has an element a 7= e

such that a2 = e.
5.5.4 Theorem: Let G be a finite group and p be a prime. Then G is of order pr if and only if every

element of G is of order pl1 for some n 2: 0 .

~sson -5 11 Algebra »
Proof: If G is order t/: then, for any a E G, o( a) is a divisor of r' and hence 0 (a) = pl1 for

some O'S; n~ r , conversely, suppose G is not of order pI' ; that is 0 ( (;) is not of the form pI' for

any J':2: O. Then there exists a prime q, other than p, dividing 0((1). Now, By the Cauchy's

theorem 5.5.1, G has an element a of order q and hence 0 (a) is not of the form p" , n>0.
The above theorem 1';;;':: If C; is not finite. In the following example, we have constructed an

infllite group in which every; :'::;ment is of order p" , where p is a fixed prime.

5.5.5 Example: Letp be a fixed prime and Qp be the set of all rational :lumbers of the form ~11 ,

where a is an integer and n i~; a non negitive integer. Then Qp is a group under the addition of

rational numbers. Also, Z is a subgroup of Qp , where 7/.,is the set of integers. Let G be the

quotient group Qp Then, any element of G is of the form ~- + 'j~and

/Z~. pl7

so that 0
(apl1. + J
- Z is a divisor of p" and hence 0 (apn. + z,J') = p
- k
for some k 2': O. Thus

any element of G is of order pk for some k 2': O. Note that G is ,~p,infinite group.

5.6 CONJUGATE CLASSES IN Sn:

In this section, we use several results proved on the conjugate classes to determine all tnf;
conjugate classes in the symmetri group Sn' . Let us recall that the elements of Sn are the

permutations on the set In = {1, 2, , n} and that any non identity permutation is ,a product of
disjoint cycles, each of length >1 (see theorem 4<) 7) Inf2(~~,vve have defined an r .. cycle only
when r > 1 (see definition 4.3.1), now, let us agree, for convenience, that 1 - cycle is defined to be
the identity map (identity element in Sn)' For example, for any 1 s is n , the 1 - cycle (i) denotes
the identity map. With this convention, we can rephrase theorem 4,3.7 as given in the foilowing.
(Distance Education 12 Acharya Nagarjuna University )

5.6.1 Theorem: Any permutation in Sn is a product of disjoint cycles; that is, any f E Sn can be
written as

f= al oa2 o oak

where aj is a cycle of length rj with rl 5 '2 5 r3s" 5 rk and rl + r2 + r3 + + rk =n ,

Proof: Let f E Sn' If f is the identity in Sn' then

f = (1)o(2)o(3)o o(n)

and each (i) is a cycle of length 1 and hence we are through suppose f is not the identity.
then, by theorem 4.3.7,

f = a] o a2 0 0 ak

where aI, a2, , ak are cycles each of length Sj > 1. Note that, for any a E In'

f ( a) :;t: a q aj ( a) :;t: a for some i and a j (a) =a for all j :;t: (

Let T = {a E In / f (a )=a}. Since Tis finite, we can write

T={aj,a2,·········,at}

Then f = (a])o (a2) 0 ....•....... 0 (at) 0 aj 0 a20 ........•••... oak. Also, since aj'sare

disjoint, they commute with each other and hence we can rearrange aI, , ak such that

:heir lengths are in increasing order. Further, for any a E In '

f(a):;t: a q a is involved-in some aj

Thus we have

f = /31 0 /32 0 0 /3t 0 al 0 a2 0 ....••••.... 0 ak

where f3i'~ are 1-cycles and a.:« are .'Ii - cycles, 15 515 .'I} 5 5 sk and

l+ +l+s} + +s« = ITI + Iln - TI = n.

In view of the above theorem, let us introduce the following concept taken from the theory of
numbers.
5.6.2 Definition : Let n be a positive integer. A finite increasing sequence of positive integers
 13 Algebra·»

whose sum is n is called a partition of n; that is, a finite sequence h, ri. ,fk} of positive
",
integers is called a Partition of n if

'1 ~ r: ~ ~ rk and rl + r2 + + rk = n

Let us denote the set of partitions of n by P(n). The number of partitions of n (i.e., the

number of memebers of P (n) ) is denoted by P (n) .

5.6.3 Examples:

p (1) = 1, since {I} is the only partition of 1.

p(2) = 2, since {I, I} and {2} are the only partitions of 2

p(3) = 3, since {I, 1, I}, {I, 2} and {3} are the only partitions of 3.

p(4) = 5, since {I, 1, 1, I}, {I, 1, 2}, {I, 3}, {2, 2} and {4} are the only
partitions of 4.

5.6.4 Self Assessment Question : List all the partitions of 5 and find p(5).

5.6.5 Theorem: The number of conjugate classes in SI1 is equal to the number of partitions of n.

Proof: We shall exhibit a one - to - one correspondence between the conjugate classes in SI1 and
the partitions of n.

Any f E SI1 can be expressed as

-,

f = al 0 a2 0 0 ak

where each a, is a cycle of length r., '1 -::;,r2 ~ ~ ']'; and rl + r: + + rk =n

(see theorem 4.6.1). that is, h ' r2, ·.. ·.··.···.,rk} is a partition of n. Further, by the uniqueness
proved in theorem 4.3.7, any representation of f as product of cycles yields the same partition of
n. Let us call this unique partition as 'the partition induced by f'. We shall prove the following:
)

1. For any f, g E SI1 ,f is a conjugate of g if and only if both f and g induce the same
partition of n.
2. Any part of n is induced by some f E Sn:.

These two give us a bijection (namely, C (f) r~ the partition induced by f) of the set of
 (Distance Education 14 Acharya Nagarjuna Univer~

conjugate classes in Sn onto the set of partitions of n. Thus, it is enough if we can prove (1) and (2)
above.

(1) Letj, g E S,1 suppose f ~g. Then f = h -,g oh-1 . Let

g = al c, (/2 o 0 ak
. (3)

where ai's are cycles of length 1j,'I:S; rz :s; ~iJ..: and.

Ii +'2 + + rk = n. Then frt, '2, ""k} is the partition of n induced by g. We sha!l prove
that the same partition is induced by f also; that is,! has a similar representation as in (8 for g,

which yields the same partition {'I, '2, ,'k}. We have

f=hogoh-1 = (h, 0 al oh-l) 0 (hoa2 oh-1)o O(hoak oh-l)

arid therefore f = f3I c/h () 0 f3k, where f3i = h «a, "h -1. If the cycle aj =( al a2 arj ) , then

you can easily prove that h" aj "h -1 = ( h ( al) h ( a2 ) h (a'l )) which is again a cycle of length
I

1j. Thus h, '2, , rk} is the partition induced by f also.

Conversely, suppose both f and g induce the same partition h, ri; ....'k } . Then choose
two representations.

and g = f31 Ilz o 0 .•.......•....•.. 0 f3k

where ui and f3i are cycles each of length 'i (for each 1:s; i :s; k ),

'1 :s; ''2 :s; < 'i,. and rr + '2 + + rk =n . Let

and f3i = (Od h! 2 hi,! )

Then {aij} = In = {hij} and, for any i:f:. j thesets {ail,

I ai'!} and {aji, ,ajrj}

are disjoint (since a, and aj are disjoint cycles). The s~me is the case with bin 's. Now, define

h : In ~ In by
(§son -5 15 Algebra »

Then h E Sn and h j 0 0 h -·1 = g , so that j ~g .

(2) This is simple; for, let h, r2, ,rd be a partition of n. Then

o<rl ~r2 ~ ~rk and rl +r2 + +r« =n.

Put ai = (ri-l + 1 ri-l + 2 ... ···.···..ri-l + rd for 1'5, i '5, k (where r; = O. Then ai
is a cycle of length 'i and the permutation given by

j=al oa2 o oak

induces the partition h, r2, ·,rk } .

This completes the proof of the theorem.

5.6.6 Self Assessment Question: If a = (a] a2 ar) is a cycle in Sn and hE S; '

prove that hoaoh-1 is the cycle (h(al) h(a2)···············h(ar))

5.6.7 Self Assessment Question: Consider the partition {I, 3, 4, 6}" of 14. Construct (as in

the proof of (2) in the above proof) a permutation in 814 that induces the given partition of 14.

5.6.8 Self Assessment Question : How rrrany conjugate classes are there in each of
SI> S2, S3, S4 and S5·
..
5.7 SUMMARY:
In this lesson, you have learnt the concept of the conjugacy relation on a group C and, for
any a E G, the conjugate class C(a) and the normalizer N(a) and proved that Cia) is bijective with
the set of left cosets of N(a) in G. Using this, we have derived the class equation ofa finite group
which is a crucial tool in proving the Cauchy's theorem for a general finite group. Also, we hve
proved that the centre of a group of prime power order is non trivial and deduced that any
,
group
, of

order p2 is abelian if p is a prime. Further we have determined the conjugate classes of the
I
symmetric group 5'/1 by proving that the number of conjugate class in SIJ is equal to the number of
partition of n.
 ·'"

(Distance Education 16 Acharya Nagarjuna University )

5.8 TECHNICAL TERMS .

Conjugacy relation, conjugate class, normalizer of a, class equation, Cauchy's theorem
for finite groups, patitions of n.

5.9 MODEL EXAMINATION QUESTIONS :

5.9.1 Define the notion of the conjugacy relation on a group G and prove that the conjugate class
of an element a in G is bijective with the set
. , of left cosets of the normalizer N(a) of G.

\5.9.2 State and derive the class equation of a finite group.

\
\

5.9.3 State and prove Cauchy's theorem for finite groups.

5.9.4 Prove that any group of order p2 is abelian, where p is a prime number.

5.9.5 For any positive integer n, prove that the number of conjugate classes in Sn is equal to the
number of partitions of n.

,
5.10 ANSWERS TO SELF ASSESSMENT QUESTIONS

5.2.3 a ~ b => a = ebe -1 = bee -.1 = b , for some eEG.

(
(since Gis abelian)

a =b => a = ebe-1 => a ~ b

5.2.6 Let a, bEG and C (a) n C (b) 7:- ¢. Then choose C E C (a) n c (b) .
Then xa.;c-1 = C = yby-l for some'.- x, y
. .;
E G

r
:

1
Now, (y -1x ) a (y -1X = Y -1 (xax -1) Y = y -1cy = b
and hence a ~ b we have
xEC(a) <=> x=-:a
<=> x ~ b (since a ~ b)

<=> xEC(b)
fiLesson - 5 17 Algebra »
and therefore C (a) = C (b) . Since a E C( a) , we have G= U C (a)
aEG

5.2.10 XEZ(G) <=> xa =ax for all aEG

<=> x E N (a) for all aEG

6XE n N(a)
aEG

5.4.2 Let G be a group ~f order 625. Th13n O( G)= 54 and 5 is a prime. Therefore, by theorem

5.4.1, the centre Z (G) is non trivial.

5.4.4 169= 132 and 13 is a prime. Now use theorem 5.4.3.

5.4.5 S-imilar to 5.4.4 since 961 = 312 and 31 is a prime.

5.5.2 793 == 61 x 13. If G is a group of order 793, then 61 divides o( G) and 61 is a prime. Then,
by the Cauchy's theorem (5.5.1), G ha an element of order 61. Therefore there exists
a 'j:. e in G such that a61 =e·

5.5.3 Similar to 5.5.2; for, if G is a group of even order, then 2 divides 0 (G) and 2 is prime.

5.6.4 The following are all the partitions of 5.

{I, 1, 1, 1, I}
{1, 1, 1, 2}
') i

{I, 1, 3}
{1, 4}
{I, 2, 2}

{2, 3}
{5}
Therefore there seven partiotons of 5 and hence p(5) = 7

5.6.6: Let a = (aI, a2, .....·ar), h « Sn and

 (Distance Education 18 Acharya Nagarjuna Univeri§J

We shall prove that h sa = fJ 0 h and hence h .a oh-1 = fJ. For any a E In

r h( ai+ 1) if a=ai, i < r

(fJoh)(a) = fJo(h(a)) = lh(~i) if a=ar

h(d) ifa:;t:aj,l::;;i::;;n

=(hoa)(a)

Therefore j3 oh = h oa .

5.6.7 : Put al =(1), the 1 - cycle which is the identity map,

a2 =(2 3 4), a 3-cycle,

a3 = (5 6 7 8), a 4 - cycle,

and a4 = (9 10 11 L ]3 14).

Put f =al " a2 0 a3 0 a4· Then tile partition of 14 induced by f is {I, 3, 4, 6}.

5.6.8 : Follows from 5.6.5 and from the facts that p(1)=l, p(2)=2, p(3)=3, p(4)=5 and

. p(5)=7. The number of conjugate classes in Sl,S2,S4, and Ss is 1, 2, 3, 5 and 7

respectively.

5.11 EXERCISES:

5.11.1 : If N is a normal subgroup of G and a EN, provethat C (a) ~ N .

5.11.2 : For any normal subgroup N of G, prove that N = U C (a), where the union is taken
over certain a's.

5.11.3 : Let G be a finite group and a E G such that a has exactily two conjugates in G. Prove
that G has a normal subgroup other than {e} and G.

5.11.4 : Find all the conjugate classes in S3 and verify the class the equation for S3 .
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-~
~~f_e_~::_o_n_-_5 1_9 ~ A~'9~.~'_]-
5.11.5 : Find all the conjugate classes in S4 andthe normalizers of elements of S4 and verify
the class equation.

5.11.6: Find the number of elements in each conjugate class in As, the alternating group of
degree 5.

5.11.7 : Exhibit two elements in As which are conjugate in S5' but not in A5 .

5.11.8 : Prove that As is simple.

5.11.9 : If D( G) = pr where p is a prime, prove that G has a subgroup of order p" for a.:
o s:; n < r (use theorem 5.4.1).

5.11.10: F"vrany group G, prove that G is abelian if and only if Ji( G) is cyclic.

5,11.11 : Prove that any group of order 15 is cyclic.

5.11.12: Prove that any subgroup of order pn-l in a group of order pl1 is normal, where pis ;)
prime.
5.11.13: Prove that any group of order 28 is not simple.

5.1·~.14: If 0 (G) = 28 and G has a normal subgroup of order 4, prove that G is abelian.

5.11.15: Find the number of conjugates of (1 2) (3 4) in Sf!' n ~ 4.

5.11.16 : If H is a proper subgroup of a group G of order p" , where p is a prime, prove that there

exists x E G· such that x e H and xHx -1 =H .

5.11.17: Let G be a group containing an element of finite order n>l. If G has exactly two
conjugate classes, prove that D( G) = 2.
5.11.18 : Prove that in any group the sub set of all elements that have only a finite number of
conjugates is a subgroup.

5.12 REFERENCES .
1. Topics in Algebra I.N. Herstein, Second edition, John wiley & sons, New
York, 1975.
~tance Education 20 Acharya Nagarjuna University J

2. Basic Abstract Algebra P.B. Bhattacharya, S.K. Jain and S.R. Nagpaul, second
edition, Cambridge University Press, 1995.
3. Lectures in Abstract Algebra - N. Jacobson, Vol -I, D. Van Nostrand Company, Inc.,
London, ,1964.
4. Algebra Serge Lang, Addison-Wesley puBfrshing Company,
London 1980. "-,;
\
. -"'.~ .
5. Algebra T.T. Moh, World Scientific, Singapore, 1'99~:
6. University Algebra N.S. Gopala Krishnan, Second edition, wiley Eastern
Limited, New Delhi, .1986. . f
l
7. Abstract & Linear Algebra - David M. Burton, Addison - Wesley publishing company,
condon 1972.
8. Algebra M. Artin, Prentice - Hall of India, New Delhi 1994.
9. Algebra J.W. Archbold, Fourth edition, the English language book
society, 1'972.
10. A First Course in Abstract Algebra - John B. Fraleigh, Addison - Wesley Longman, Inc.,
Fifth edition, 1999.

Lesson Writer
Prof: U.M. Swamy
Andhra University,
Visakhapatnam.
 Lesson - 6 SYLOW'S THEOREM

Objectives:
After reading this lesson, you should be able to

(i) state and prove the Sylow's theorem in three different methods

(ii) define the notion of a p - Sylow subgroup and prove that any two p - Sylow subgroups
of a group are conjugte to each other and

(iii) prove that the number of p - Sylow subgroups of a group G is a divisor of 0 (G) and is
the form 1 + kp, k >a .

Structure:
6.1 Introduction
6.2 The first proof of Sylow's theorem

6.3 The second proof

6.4 The third proof

6.5 Sylow's theorem • II

6.6 Sylow's theorem - III

6.7 Summary

6.8 Technical terms

6.9 Model examination questions.

6.10 Answers to Self Assessment Questions. '
6.11 Exercises
6.12 References

6.1 INTRODUCTI.ON :

It is well known that the order of a subgroup of a finite group G is a divisor of 0 (G) (the

Lagrange's theorem). The converse of this is not true; that is, for any divisor d of o(G) ,there may

not exist any subgroup of order d in G: For example, there are no subgroups of order 6 in A4'

B)
 (Distance Education 2 Acharva Nagarjuna University J

even though 6 is a divisor of 0 ( A4)' There are very few theorems which assert the existence of
subgroups of prescribed order in arbitary finite groups: In this direction, a classic theorem due to
the Norwegian mathematician Sylow is the most basic and widely use one. We present three
proofs which are of completely divese nature and each gives nice application of different things that
we have learned.

6.2 THE FIRST PROOF OF SYLOW'S THEOREM

In this section, we present an elegant and simple elementary proof of Sylow's theorem
which uses certain basic ideas from number theory and combinotories. Let us recall that, for any
positive integers n and k with k s n, the number of ways of picking a subset of k elements from
a set of n elements is equal to

nJ n!
( k - k! (n~k)!

We shall begin with proving the following:

(

6.2.1 Lemma: If n = pa m , where p is a prime, plm and pr+! X m , then

Proof: First, let us consider

(p;: J = (pa)1 i:::)~pal!

_ pam (pam _ 1) (pa m - i) (pa m _ pa + 1)
- pa (pa _ 1) (pa _ i) (pa _ pa + 1)
on the right side of the above equation, you can easily ~ee that, except for tile term m in the

numerator, any power of p dividing (pa m - i) is the same as that dividing pa - i . Therefore, on
the right side of the above equation, all powers of p cancel out except the power which divides m
Thus
(
fiLesson -6 3 \ . Algebra ».

6.2.2 Self Assessment Question: In the above, verify that, for any fJ ~ a,

6.2.3 Theorem ( Sylow theorem - I ):

Let G be a group, p a prime number and a a non negative integer such that pa divides

O( G). Then G has a subgroup of order pa.

Proof: Let 67 be the set of all sub sets of G each with pa elements. We shall prove that 67 has

a number whlch is a subgroup of G. Since pa divides 0 ( G) , we can write 0 (G) = pa m , where

m is a positive integer. The prime p mayor may not divide m. Let fJr be the largest power of p
that divides m. Then, we have

r e O, O(G) pam, plm and'pr+l X m

pam]
The number of elements in 67 is [ pa ., which is divisible by pr and not divisible by
~" \
\ .
pr+l (by lemma 6.2.1).
Let us define an equivalence relation - on 67 as follows:

for any A, B E r;f)',

A ~B ¢:> A = Bg for some g E G.

Then - is an equivalence relation on 67. Let A be the equivalence class containing A in 67

and IAI be the number of elements in A-, Since the equivalence classes form a partition of 67, WP
get that
 (Distance Education 4 Acharya Nagarjuna University J

pam]
Since pr+l does not divide
[
pa ,it follows that pr+l does not divide IAI for some
-~,

equivalence class A, Let such an equivalence class be

Then A - A j for all j and pr+l X n.

put H = {g E G I Ag = A}

Then H is a subgroup of G, We shall prove that O(H)= p'", Since A - Ai, we can write

A; = Ag; for some gi E G and it = {Ag1, ".",,,,Agn}, Let us observe that

Hg1, Hg2,,,,,,,.,,,,,,.,,,,Hgn are all the distinct right cosets of H; for consider

Hg, = Hg , <=> gjg; -I E H

<=> Agigj -1 = A

<=> Ag, = Ag j <=> i = j

Also, given any coset Hg;we have Ag - A and hence Ag = Ag, for some i which implies

that g;g-l E H so that Hg ~ Ji'gj. Thus Hg-, Hg2,,,,,,,.,,,,.,Hgn are all the distinct right co sets

of H in G and, since Hg, 's form a partition of G, we have

n· O(H) = O(G) = pam

Since pr
1mI and pr+l X m, it follows that

a+r I
p In. O(H) and pr+l X n
(~esson -6 Algebra D
pal
and therefore 10 (H). For any g E H and a E A , we have ag E Ag = A,

which implies that A has atleast O(H) number of elements. Since A E Q9

pa =jAj ?O(H). Earlier, we have proved that P/o(H). Thus O(H) = pa and therefore H

is a subgroup of order pa in G.
5.2.4 Self Assessment Question: In any group of order 2500, prove that there is a subgroup of
order 125.

,ml
6.2.5 Corollary: Let G be a finite group and p a prime such that p I (G)
0 and r= X O( G).
Then G has a subgroup of order pm.

Proof: This is an immediate consequence of theorem 6.2.3.

The above corollary has an importance, even though it is a direct consequence of theorem
6.2.3. The reason is that 6.2.5 implies 6.2.3 (see 6.2.6 and 6.2.7 given below). Therefore, in order
to prove the main theorem, the Sylow's theorem I, it is enouh if we prove 6.2.5. This is what we are
going to do in the second proof and the third proof given in the next two sections.

6.2.6 Self Assessment Question : Let G be a group of order pn where p is a prime. Then, for
r
each 0 S; r S; n , prove that G has a subgroup of order p .

6.2.7 Self Assessment Question : Prove that corollary 6.2.5 implies the Sylow's theorem - I.
6.2.8 Self Assessment Question : In any group of order 405, prove that there is a subgroup of
order81. \1.'

m/
6.2.9 Definition: Let G be a finite group and p a prime such that PI0 (G) and pl11+1 X O( G) ..

Then any subgroup of order p'" is called a p-Sylow subgroup of G.

6.2.10 Self Assessment Question: Prove that for any finite group G and any prime p, there
exists a p - Sylow subgroup of G.

6.3 THE SECOND PROOF

As it is mentioned in the introduction to this lesson, we present another proof of the Sylow's
theorem - Iwhich is completely different from the first proof given in the previous section and is
 (Distance Education 6 Acharya Naqarjuna University )

based on induction on the order of the given group .

. 6.3.1 Theorem: Let G be a finite group and p a prime such that P/o(G) and pm+! X O(G).

Then G has a subgroup of order pm.

Proof: We can assume' that m is a positive integer (for, if m = 0, then G has the subgroup {e}
whose order is 1 = pm). ,To prove the theorem, we shall use induction on the order of G. O( G)

must be atleast2 (since P/o(G)' P is a prime and m > 0). If O( G)=2. thenp = 2 and m = 1 and

G itself is a subgroup 0/ order

"-
p", Now, let 0 (G) > 2 and assume that the theorem is valid for all
groups of order less than O(G). Recall that the centre Z (G) of G is defined as

z (G) = {a E G / ax = xa for all x E G}

and.that Z (G) i's a normal subgroup of G. We shall distinguish the followin~o cases.

Case (1 ): Suppose 10 (Z (G)): By the Cauchy's theorem 5.5.1, Z (G) has an element of order

p. Let a E Z( G) be such that a =F e and aP'=e. Consider the subgroup N =(a) generated by a.

Then Nis a normal subgroup of G (since a E Z( G), xaix-I = ai for all x E G) andO(N) = p.

Now, the quotient group % is of order O~) which is less than O(G) and hece, by the induction

, hypotheses, the theorem is valid for the group %. Also, since P/o( G) and pn1+1 X O( G) , it

follows that

Therefore, by the induction hypothesis, % has a subgroup of order pm-I. Let k be a

subgroup of order pm-l in %.

fiLesson - 6 7 Algebra D
Let H = {x E G / xN E k}

Then H is a subgroup of G containing N as a normal subgroup and % = K. Now,

O(H) = 0(%) O(N) = O(K)O(N) = pm-I. p = s"

Thus H is a subgroup of order pm in G.

Case (2): Suppose p X 0 ( Z (G») :

Let us recall the class equation of as given in 5.3.2.

n O(G)
O(G) =O(Z(G»)+ ~ O(N(aj ») ~

where at> a2, ,an are such that C (ad, , C (an) are all the distinct conjugate
classes each with more than one element. Also, recall that

C(a) = {xax-1ixEG},

N(a) = {XEG /xa = ax}

~
~~
..... ~

Since Ie (a;)1 > 1, itfoliows that N( aj );<G for each I. 'Since P/a( G) and m > 0, we

have 1o(G)' Since p X o( Z( G»), we get that

O(G)
p '/. 0 (N (aj ») for some i

that is, pia ( (a;)) N for some i. Also, 0 ( N ( aj ) ) < 0 ( G) and hence, by the induction
 (Distance Education a Acharya Nagarjuna University J

hypothesis, N(ai) has a subgroup of order t/" (note that pnl+l X O(N(ai)) ,since O(N(ai)) is

a divisor of O( G) and pnl+l X O( G)). But any subgroup of N( a) is a subgroup of G also. Thus

G has a subgroup of order pm.

6.3.2 Theorem (Sylow's Theorem - I) :

Let G be a fintie grou p and p a prime such th at Pfo (G). The n G has a su bg rou p of order pa

Proof: Letmbe anon negitive integer, such that P/a(G) and pnt+l X O(G) (such am exists,

since p" cann't divide 0 (G) for all positive integers n and hence we can choose the least 11 such

that p" divides o( G) and take m = n - 1). Then, by theorem 6.3.1, there exists a subgroup K of

order r" in G. Now, if P/o( G)' then a ::;m and P/o( K) and ~ence, by 6.2.6, K has a

subgroup H of order pO . Then H is a subgroup of order pm in G also.

6.4 THE THIRD PROOF:

In this section we present yet another proof of the Sylow's theorem which is completely
different from either of the two earlier proofs. Here, we shall first prove that the Symmetric group
k ...
Sp k of degree p . has a p - Sylow subgroup and later prove. that, if G a·.J M has a p - Sylow

subgroup, then so is G.finally we shall the Cayley's theorem to get a sufficiently large k such that
G is embedded in Spk . To begin with, let us estimate the sige of a p - Sylow subgroup in S pk .

Recall that the order of S pk is (pk)! .

6.4.1 Definition:

Le P be a fixed prime number. For any positive integer k, n(k) is defined to be the positiv=
integer such that
 fiLesson- 6 9 Algebra]]

In other words, pn(k) is the largest power of p that divides (pk)1 .

k
6.4.2 Lemma:
" ;-1 =
n (k) = L.. P
2
1+ P + P + +p
k-1
, for any positive integer k .
i=1

Proof :We shall use induction on k . If k = 1, then (pk)1 = pi = 1.2 (p -1) P and it is clear
I

that % I and p2 X p! and hence n (1) = 1. Now, let k » 1 and assume that the lemma is true for

k-1; that is

n(k-1) = 1+ P + p2 + + pk-2

Let us check that what terms in (pk)! can contribute to powers of p dividing (pk)! clearly,

only the multiples of p, that is, p, 2p, , pk-l P are such terms. This is to say that n(k) must
be power of p which divides

p (2p 3p
)() (P
k-l
P ) -_ P pk-l. (
P k-l), . and therefore we have

(k'
n (k) =p k-J +n-l) = p k-l +p k-2.· +·········+p+l

.)1\.::" .
In the following, we shall prove that Sp k has a p - Sylow subgroup; that is, Sl a subgroup

of order pn(k).

6.4.3 Lemma: For any prime p and positive integer k, the symmetric group Spk has a p - Sylow
subgroup .

. Proof: Let p be a prime and k any positive integer. We have to prove that Spk has a subqrou

n(k)
of order p where
(Distance Education 10 Acharya Nagarjuna University J

n (k) =1+p+p 2
-t +p k-f

(see 6 ..4.1 and 6.4.2) we shall prove this by using induction on k. If k = 1, then the cycle (1
2 '. p) is an element of order p in Sp and hence it generates a subgroup of order

p (= pn(1), sin ce n (1) = 1) in S p . Therefore, the lemma holds good for k = 1. Now, let k > 1 and

assume that the lemma is true for k - 1; that is, S/-1 has a subgroup of order pn(k-1). Let us

divide the integers 1, 2,3, , pk into p clumps, each with pk-l elements as given below:

{ 1,2,3, , P k-l} , { P k-l +1, P k-1 +2, ,2p k-l} ,

............ , {( P -1 ) P k-1 + 1, ( p -1 ) P k-1 + 2, , p k}

Consider the cycles aJ, a2, ,a k-l defined by

p
,

al = (1 pk-l +1 2pk-l +1 (p_l)pk-l +1)

apk-1 = (P
k-J
2p k-l 3p k-l .p k)

a1, a2, , a pk-l are disjoint cycles, each of length p and hence
-
I'

a!'1 =l'd and aoa·

I J =a·oa·
J 1
for all 1<1'
-,-
j'<pk-J.

Put a = al 0 a2
.
0 ...••........ 0 a
P
k-l

Then aP =id and a satisfies the following property:

If g ES k such that g(i) = i for all i » pk-1

P
(8)
{ then a - j ga i i,
(1) _.
- I for all I..tt {JP k-l + 1, JP. k-l + 2, .
(J+ 1)P
k-l
}
fiLesson - 6 11 Algebra D
Now, let A ={g E Spk / g(i) = i for all i > pk-l}. Then A is a subgroup ~f Spk . Also

g g1 H
~lp
k-I gives us an isomorphism of A onto Spk-I and hence A -= S
. P
k-J. By the induction
. \

hypothesis, A has a sub gorup Bl' of order pn(k-l).

each B, -= Bl and hence O(Bi) = O(Bl) = pn(k-l), also, for any i 7= j. BiB) = B)Bi
(since B, (s) 7= S => B) (s) = s). Therefore B is a subgroup of S pk . Also, since B, n Bj = {e} for

i 7= j, it {ollows that
" "'-

a (B') _- 0 (Bl ) O()B2................. O()B P -- P l1(k-l) . P l1(k-1) p l1(k-1) _- p n(k-l)p

Further, since (J"P =e and B, = (J" -(i-I) B1ui-1, it follows that

Now, put H = {a) b / b E Band 0 ::;;j ::;;

P -1}. Since a r;. Band (J" -1B o =B , we get that
-, ~-
H is a subgroup of S pk and

O(H) ,-= p.O(B) = p.pl1(k-1)·P = pn(k-l)p+l = pl1(k)

,

Th~-m~a subgroup of order pn(k) in S k .

--....... p

6.4.4 Self Assesment Ouestion: Prove1h9!!i defined in the above proof is a subgroup of Spk .

6.4.5 Self Assesment Question : Prove that n ( k - 1).~~.

.s-:
n (k )
..--:
.----:: , . .-
. the property ® in the above proof holds.
 (Distance Education 12 Acharya Nagarjuna Univers2!iJ

6.4.7 Self Assesment Question: Let f, g E Sn such THAT the sets {a E In / f (a)"* a} and

{a E II! / g (a) "* a} are disjoint. Then prove t~at fog = go f. This is used in the proof of 6.4.3
where?
6.4.8 Self Assesment Question: Using the algorithm described in the proof 6.4.3, construct a 2
- Sylow subgroup in S4'

Before we reach the third proof of the Sylow's theorem, we need the following terminology.

6.4.9 Definition: Let A and B be subgroups of a group G. For any x E G, let

AxB = {;xb /a E A and b E B} .

Then A x B is called a double coset of A, B in G.

6.4.10 Lemma: Let A and B be subgroups of a group G and define a binary relation - on G as
follows:

For any x, Y E G, x ~ Y <=> x = ayb for some a E A and bE B.

Then - is an equivalence relation on G with respect to which the equivalence class containing
x is equal to the double coset AxB .

Proof: Since A and B are subgroups. e E An B and x = exe and hence x - x for all x E G. Also,

x- y :::::> x = ayb for some a E A and b E B

:::::> y- x \,

and x - y and y - z ~ x +ayh and y =aI z bI, a,aI E A, b, bI E B

Therefore, - is an equivalence relation. If x is the equivalence class containing x, then

x = {y E G /x - y}
= {YEG I y = axbfor some aEA and bEB}
= AxB
6.4.11 Lemma: For any finite subgroups A andB of a group G and x E G, the number of elements
 [Lesson
~~.T
-6 13 Algebra D
in the double coset AxB is given by

(--; of : Let x E G. Then the sets AxB and AxBx -I are bijective (the map f :AxB ~ AxBx -I ,
de .ied by, f (axb) +axbx -1 is a bijection) and hence they have the same number of elements.
, Therefcre,

6.4.12 Self Assessment Question: For any subgroups A and B of a group G, prove that any two
disvict double cosets are disjoint and that, if G is finite,

n , n O(A).O(B)
O(G) = LO(AxjB)=
1=1
L
1=1
( ( .:
0 An xlis
-1))
where AXlE, , AXI1E are all the distinct double cosets of A, E in G.
The following is the last one in the sequence of lemmas which we use in proving the main
theorems of this lesson.
6.4.13 Lemma: Let G be a subgroup of a finite group G' and p a prime. If G' has a p - Sylow
subgroup, then G also has a p.- Sylow subgroup.

Pro of : Let P be a p - Sylow subgroup of G': Then

we shall prove that G n (xPx -I) is a p - Sylow subgroup of G for some x E G'. To pr,ove

this, suppose P/a( G) and r= X o( G). Then we have to prove that o(Gn( xpx-1)) = pn for

some x E G". Consider the double cosets G X P of G, P in G': Since the double cosets form a
parition of G' , we have

O(G') = IO(Gxp)
Where the summation is over some elements x E G' . Also, by 6.4.11, we have,
 (Distance Education 14 Acharya Nagarjuna University )

where 0 (G) = r" .k with P X k. Also, since G n (xPx -1) is a subgroup /

of xPx -I and

o( xpx-I)= a (p) = r", we get that

ABO, srice G n(xPx -1) is a subgroup G, it's order pmx must be a divisor of 0 ( G). But

pn is the largest power of p that divides o( G). Therefore lnx ~ n also. We shall prove that

mx = n for some x E G'. For, if mx < n for all x , then

nkpm
O(G X P) = P
pmx
= P n-mxkpm and n - mx > a

rn+l/
and hence pm=l divides 0 (Gxp) for all x E G' which implies that p /0(07) (since

r=
\

O( G') = LO (Gxp)). This is a contradiction, since x O( G'). Therefore

\'

mx =, nand o( Gn( xpx- 1

)) ~ p"
l
i

~'
Thus G n(xpx-1) is a p - Sylow subgroup of G.

Now, we are well prepared to present athird proof of the Sylow's theorem.

6.4.14 Theorem (Sylov;'s theorem I) :

, ,a/
Let G be a finite group, p a prime and a any positive integer such that p/0 (G) . Then G
a
has a subgroup of order P ,

Proof: Let O( G) = n. Then, by the Cayley's theorem (3.3.1), G can be regarded as a subgroup
 of s., Choose k such that n $ pk . I"en In ell and nence an,! permutation on In can be

regarded as a permutation on 1pk (see 4.2.5). Therefore Sn can be regarded as a subgroup of

"

Spk and hence G is a subgroup of Spk . By lemma 6.4.3, Spk has a p - Sylow subgroup and, by

" lemma 6.4.13, G has a p - Sylow subgroup. Let p be a p - Sylow subgroup of G. Then

. O( p) = r" , where Pla( G) and pm+l X 0 (G) Since plo( G) , it follows that as m. Now,

, by 6.2.6, P has a subgroup of order pa which will be a subgroup of G also.

'6.5 SYLOW'S THEOREM - II :

pm/
If P is a p-Sylow subgroup of a group G, then O( P) = t/" where / o(G) and

pm+l X O( G) and since

O( xpx-1
) = o(P) = r",

it follows that xPx-1 is also a p - Sylow subgroup of G for any x E G. In the following we
: shall prove that any other p - Sylow subgroup of G must be of the form xPx-1 for some x E G. Let

us recall that two subgroups Hand K are said to be conjugate if H = xKx -1 for some x E G.

6.5.1 Theorem (Sylow's Theorem- II):Let G be a finite group and p a prime. Then any two
p - Sylow subgroups of G are conjugate.
Proof: Let P and be p - Sylow subgroups of G. Then

O( P) = O(Q) = r". where Pla( G) and pm+l X O( G)

Let us consider the double cosets PxQ ofP, Q inG. For any xEG,wehave(by6.4.11),
 (Distance Education 16 Acharya Nagarjuna University )

Since P n(xQx -I ) is a subgroup of P, it's order should divide 0 (p) ( = pm) and hence
•

o( pn( xQx -I)) = plnx for some 0 s mx sm. We shall prove that mx = m for some

x E G. On the contrary, suppose m, <m for all x E G. Sinee the double eosets PxQ form a
partition of G, we have

since mx <m, 2111 - I11x 2:: 111 + 1 and hence pm+l divides p2m-mx for all x E G which

implies that r: divides o(G). This is a contradiction, since pl11+1 X O( G). Therefore, our

supposition that mx < m for all x E G leads to a contradiction. Thus, mx =m for some x E G.

But this implies that o(pn(XQx-I)) = O(P) and hence pn( xQx-l) = p so that p ~ xQx-i.

But, P and xQx -1 are of same order pili. Thus P = xQx -I for some x E G and hence P and Q
are conjugates.

6.5.2 Self Assessment Question: Determine all ;. - Sylow subgroups of S3 ,S4 and A4 .
6.5.3 Self Assessment Question: ;:"rove that p - Sylow subgroup is normal if and only if it is the
unique p - Sylow subgroup.

6.5.4 Definition: Let H be.a.subqroup of a group G and define

. N (H) = {x E G / xHx -I = H}
Then N(H) is a subgroup of G and is called the normalizer of H in G.

Note that H is a normal subgroup of N(H) and 0 (N (H)) is a divisor of 0 (G) and

O(G)j (N(H)/J- O(G)j. ..

/O(N(H))·O /H - O(G) and hence j O(N(H)) IS a divisor of O(G) .

. 6.5.5 Theorem: Let G be a finite group andp a prime. Then the number of p - Sylow subgroup of

O(G)j. .,
G is equal to j 0 (N (P))' where P IS a p - sylow subgroup and N(P) IS the normalizer of P,
 ~Lesson - 6. 17 Algebra D
and this number is a divisor of 0 (G) .
Proof: Let P be any p - Sylow subgroup of G and X be the set of all p - Sylow subgroups of
G. Then, by Sylow's theorem II (6.5.1) we have

you can easily check that xPx -1 H xN (P) is a bijection of X onto the set of left cosets of
N(P) in G and hence the number of elements in Xis equal to the number of left cosets of N(P) in G.
Thus, the number of p -Sylow subgroup is equal to the index of N(P) in G which is same as

O( /o(N(P)) and this divides O( C).

6.5.6 Self Assessment Question: Prove that the number of p - Sylow subgroups of a group G is
a divisor of the index of a p - Sylow subgroup in G.

6.6 SYLOW'S THEOREM - III :

In the previous section, we have proved that tile number of p - Sylow subgroups of a group
G is a divisor of 0 (G). The following gives us more information about this number

6.6.1 Theorem (Sylow's theorem - III ) : Let G be a finite group and p a prime. Then the number
of p' - Sylow subgroups is of th~ form 1+ pk for some non negitive integer k.

Proof: Here again we use the technique of double cosets. Let J> be any p - Sylow subgroup of G
and N(P) be the normalizer of Pin G. Then, by theorem 6.5.5, the number of p - Sylow subgroups
J _. O(G) O(G)~
of G is equal to 0 (N (P)) . Therefore it is enough if we can prove that ~ 0 (N (p)) is of the

form 1+ pk.

Consider.the double cosets PxP , x E G. Since these form a partition of G, we have

Where xl, .x; are such that PxIP, .Px.P are all the distinct double

9
 (Distance Education 18 Acharya Nagarjuna University J

cosets in G. We shall divide the above summation into two parts. We can assume that

x}, .x; E1N(P) and xr+l, 'xn $ N(P).

Observe that

X E N(P) => xPx-1 = P

=> xP = Px
=> PxP = (xP)P =xP

=>O(PxP) = O(p) = r", say

and x $ N(P) => xPx-1 =1= P

=> pn( xPx-1) 7: P

=> o(pn( x;Px-1 ))= pn for some n <m

=> O(Pxp) ~ p2m-n and 2m-n ~ m+l

m+l/
=> p /O(PxP)
Now, we have
r r
O(G) = IO(P xi P) + I o(p Xi p)
i=1 i=r+l

-- 'P~
, m
+ p m+l . k c.
lor some k >
_0 to>.
'<Y

Also, for any x E N(P), we have

XP=PXP=PXiP for some l~i~n and, in particular XjEXP or

x -1 xi E P c N (P) which implies that Xi' E N (p) and hence 1~ i :S r . This is to say that every

left coset of P in N(P) must be of the form x.P for some 1 :S i :S r and therefore, there are exactly
r number of distinct left co sets of Pin N(P); that is

r = 'the index of Pin N(P) = O(N(P)) = _O~(N_(_P--,-))

. O(P) pm
fiLesson - 6 19 Algebra »
from the equation 09, we get that

O(G) = O(N(P)) + pm+lk

O(G)
or O(N(P)) = 1+ pk

Thus, the number of p - Sylow subgroups of G is of the form 1+ pk for some k >a .
In the following examples, we shall demonstrate now these three Sylow theorems can be
used to know about the structure of finite groups.

6.6.2 Example : LetGbe a group.?forder 1225. Then O(G) =52 .72. Let np be the number of
p -Sylow subgroups of G. Then.by 6.5.6 and 6.6.1 we have

ns = 1+ 5k and ns divides 72

From this it follows that ns = 1 and hence G has a unique 5 - Sylow subgroup; that is, G

has a unique subgroup of order 52 ="25, which be normal in G (see 6.5.3). Let A be a normal

subgroup B of order 72 = 49. Also, O( A n B) must be a common divisor of 25 and 49 and

hence O(A n B)=l; that is, AnB ={e} and hence",

and therefore AB =G .
Further, for any a E A and b E B

aba-1b-1 = (aba-l )b- l E B (-: B is normal, aba-l E B)

= a(ba-Ib~l ).EA CA i~ normal ba-1b-1 E A)
and hence aba-1b-1 EAnB = {eJ'so·thataba-1b-1 = e. Thus

a .b = b· a for all a E A and b E B

Also, A and B are abelian' qroups (these are groups of order prime square) and any element
of G can be written as ab for some a E A and bE B. All these data' yields that G is an abelian .
group. Thus any group of order 1225 is abelian,
 (Distance Education 20 Acharya Nagarjuna University )

6.6.3 Self Assessment Question: Prove that any group of order 20449 is abelian
6.6.4 Example: We shall prove that any group of order 72 is no. simple; that is, it has a nontrivial

proper normal subgroup. Let G be a group of order 72. Then O( G) = 23 . 32. Let n3 be the
number of p ~Sylow subgroups of G. Then

\ n3/
n3 = 1+3k for some k 2 Oand } O( G)

From these it follows that n3 = 1 or 4. If n3 = 1, then there is a unique 3 - Sylow subgroup

which must be normal and is of order 32 so that it is a nontrivial proper normal subgroup of G.

. O(G)
Next, suppose n3 = 4. Let P be a 3 - Sylow subgroup of G. Then O( N (P)) = 4 (see 6.5.5); that

is, i(N(P)) = 4. Since O(G) X i(N(P))!, it follows, by theorem 4.4.4, that N(P) contains a
nontrivial normal subgroup H of G. Then H is a nontrivial proper normal subgroup of G. Thus G is
not simple

where np' is then umber f}!p.- Sylow subgroups ofG. from these, you can prove that n3 = 1 =ns
Thus these exist normal subgroups A and B of order 3 and 5 respectively. Then, as in 6.6.2,
AnB={e} and AB=G. Further, A:=Z3 and B:=Zs. Also, the map (a,b)f-7ab is an
/
isomorphism of Ax B onto G. Thus G:=A xB:= Z3 xZs == ZlS and hence G is cyclic.

6.6.5 Example: We shall prove that any qroup of order 15 is cyclic. Let G be a group order 15.
ThenO(Gh=3.5.Then ·r.

l
and ns =1+5k

6.6.6 Self Ass;ssment Question: Letp and q be distinct primes such thatp < q and p X q -1.
Then prove that any group of order pq is cyclic. In particular, prove that any~loup of order 35 or
77 or 143 is cyclic.

6.7 SUMMARY:
In this lesson, you have learnt the three Sylow's theorems on the existence of p - Sylow
subgroups of a finite group and, in particular, we have proved that any two p - Sylow subgroups of
~Lesson - 6 21 Algebra ~
~
-
a group G are conjugate to each other Also, we have proved that the number of p - Sylow subgroups

of G is a divisor of o( G) and is of the form 1+ pk for some non negitive integer k. Using these
Sylow's theorem, we have determined the structure of certain groups of given orders.

6.8 Technical Terms:

p - Sylow subgroup
Conjugate subgroups

Sylow's Theorems

Normalizer N(P)

Double Coset A xB
Simple group
Cyclic group

6.9 MODEL EXAMINATION QUESTIONS

6.9.1: State and prove Sylow's theorem - I

6.9.2 : Define the concept of a p-sylow subgroup and prove that any two p - Sylow subgroups of a
finite group are conjugate.
6.9.3: State and prove Sylow's theorem - III

6.9.4 : Prove that the number of p - Sylow subgroups of a finite group G is a divisor of 0 (G) and is

of the form 1+ pk, k e0.

6.10 ANSWERS TO SELF ASSESSfi!lENT QUESTIONS :

-
6.2.2 : Since f3 ::;a , we have
p%- pP/
p
a and /1 p a m and therefore
p% a
p m-[
. <::>
p% p
a ..
-I-

6.2.4: Let o( G) = 2500 = 54 , 22 and hence%( G) . Also 5 is prime. By theorem 6.2.3,_G has

a subgroup of order 53 = 125.

 (Distance Education 22 Acharya ~agarjuna University )

6.2.6: We can assume that O( G) = p" when n E Z+ . Then we shall use induction on n. If n = 1,

then {e} and G are subgroups of orders pO and pI respectively. suppose that n > 1 and that the

result is true for all groups of order p'": m'<n. Then, by theorem 5.4.1, Z (G) :t= {e} . Since Z (Q)

is a subgroup of G, 0 ( Z ( G)) divides 0 (G) = pn and hence 0 ( Z ( G)) = pa , a > O. In particular,

p divides 0 ( Z (G)). Again, by the Cauchy's theorem 5.5.1, Z(G) has an element of order p. Let

a :t= a E Z( G) and aP = e. Then (a) is a normal sub goup of G (since ax = xa for all x E G) and

31a) is of order pn-l. Since pr divides o(G), r' divides O(%,)) and hence 31a) has

a subgroup of order p" -I . If K is a subgroup of 31a) then the subgroup H defined by

H = {XEG / x (a)EK}

is of order p' in G (since '}1a t K and I(a)1= p )

6.2.7: If p/o( G)' then a ~ m , where m is such that f/o( G)-and pm+l X O( G). By 6.2.5,

there exists a subgroup K of order pm in G. Then" by 6.2.6, K has' a subgroup H of order pa.

Then H is a subgroup of order pa in G,I ."

. " . I'

6.2.8: This follows from the facts that 81 =34, 3/{05 and'3 is prime and by theorem 6.2.3, G has

a subgroup of order 34 =' 81, if 0 (G) = 405 .

6.2.10 This is an immediate consequence of corollary 6.2.5.

6.4.4: Since 0'-1BO' =B , we have

-' B
0' 1
'. = B and hence B0' i
0'1 = 0' iB .

some
fiLesson -6 23 Algebra D
o ~ k ~ P -1 (since (JP = e), for any q, b2 E B and a ~ i ,j ~P -1. Thus H is a subgroup of

SP k•.

6.4.5: By lemma 6.4.2, n( K) = 1+ P + p2 + + pk-l

=l+p 2 + k-l)
( l+p+p +p

='l+pn(k-l)

6.4.6 This follows from the fact that, if a = (aI, a2, ,ar) is an r - cycle, r>l then

for any a E 1pk = {I, 2, , pk}

If/(a) =a = g(a), then (fog )(a) =a = (g~f)(a).

Suppose / (a) i-= a. Then / (/ ( a)) i-= / ( a)

and hence g (f (a)) = f (a) = f (s (a)). Thus fog = gof ..

6.4.8: Divide 14 = {I, 2,3, 4} into {I, 2} and {3, 4}'-.1 Let BI be the subgroup generated by

(1 2) and 0- = (I 3)(2 4). Then 0-2 =e and B2 = 0--lB10-:=; ((3 4)), the subgroup

generated by (3 4). Then the subgroup generated by (1 3) (2 4) and

is a 2 - Sylow subgroup of S4'

6.4.12 : This follows from the fact that the double cosets AxE, x E G, form a partition of G and by
lemma 6.4.11.

6.5.2 : 0 (S3) = 3! = 6 and the order any 2 - Sylow subgroup of S3 must be 2. Let f = (1 2),
 (Distance Education 24 Acharya Nagarjuna University )

g=(2 3) and h=(3 1). Then (.t), (g) and (h) are allthe z+ Sylow subqroups of Sj (Note

that the number of 2 - Sylow subgroups in S3 is a divisor of 6 and is of the form 1+ 2k and hence
it 1 or 3; In this case, it is 3).

Therefore, the order of any 2 - Sylow subgroup in S4 is 8.

-
As in the case of S3 you can conclude, using Sylow's theorem - III, that these three 2 -

Sylow subgroups in S4 also. Try to list them.

o (A4) = 12 and hence the order of any 2 - Sylow subgroup is 4 and again there are three

2 - Sylow subgroups in A4.

6.5.3: Let P be a p - Sylow subgroup of G. Then any p - Sylow subgroup must be of the form

xPx-1 for some x E G (by sylow's theorem II). Therefore

P is unique <=> xPx-1 =P for all x E G

<=> P is normal.
6.5.6 : We know that this number is equal to

0(;7;)) . where P is a p - Sylow subgroup and N(P) is the normalizer of P. Since

P ~ N(P), we have

i{P) = ~i~\ = O(~~])) . O~~~)) (since O{P) divides O(N{P)))

O(G)
and hence ~(N(P)) isadivisorofi(P).

6.6.3 20449 = 112 x 132 and now initate the argument given in6.6.2.

6.6.6 Let G be a group of order pq. Ler n p be the number of p - Sylow subgroups in G. Then

I
nv~
pq and n p
.
= 1 + pk , for some k > 0 .
fiLesson - 6 . 25 AlgebraJ)
Since f2. and q are primes, 'p = 1 or P or q or pq.
Since np = 1+ pk, np :;t:. p . If l1p =q, then q = 1+ pk or pk=q-1 which is a

contradiciton since p X q -1. Also, np :;t:. pq (since p Xl + pk). Thus np = 1. Similar argument

can be made to prove that nq = 1. let P and Q be p-Sylow and q - Sylow subgroups respectively
inG. Then, by6.5.3,P and Q are normal. Also O(p) = p, O(Q) = q and hence O(PIlQ) =1
so that O(PQ) = o(p) O(Q) =pq = O(G). Therefore PQ=G and

(a, h) H ab
is an isomorphism of P x Q onto G. Therefore

G=FxQ =.», =;
and hence G is cyclic.
Note that 35 = 5 x 7 .and 5 X 7 -1

77 = 7 x 11 and 7X 11- 1

and 143 = 11x13and 11 X 13-1

6.11 EXERCISES:
6.11.1 : Prove that any group of order 1986 is not simple.

6.11.2 : Prove that any group of order 2p must have a normal subgroup of order p, where pis \
a prime. \

6.11.3 : Let P and q be distinct primes. Prove that any gorup of order p2 q should contain a
normal P - Sylow subgroup.
6.11.4 : Prove that any gorup of order 200 should contain a normal Sylow subgroup

6.11.5 : Let G be a simple group and 1'1p be the number of p - Sylow subgroups inG. Prove
. I '1 i ;

that O( G) divides (np)!. ..

6.11.6 : Prove that a group of order 48 cannot be simple ..

6.11.7 : Let G be a finite group such that each of its p - Sylow subgro/qp is nbnnq:il. F7~p,~~
that
G is the direct product of its p - Sylow subgroups.

6.11.8 : Let G be a group of order PI P2 P» . where Pi's at'EYdistirictpri!v1es. 8howthat G

/

contains a normal Sylow subgroup.

(Distance Education 26 Acharya Nagarjuna University )

6.11.9: Prove that As is the first non abelian simple group.

6.11.10 : Let P be a p-sylow subgroup of a group G and H be a subgroup of G such that

O(H)=pm ,m>O. Prove that HnN(p)=Hnp.

6.12 REFERENCES:

1. Topics in Algebra I.N. Herstein, Second edition, John wiley & sons, New
York, 1975.

2. Basic Abstract Algebra P.B. Bhattacharya,S.K. Jain and S.R. Nagpaul, second
edition, Cambridge University Press, 1995.

3. Lectures in Abstract Algebra - N. Jacobson, Vol -I, D. Van Nostrand Company, Inc.,
London, 1964.

4. Algebra Serge Lang, Addison-Wesley publishing Company,

London 1980.

5. Algebra T.T. Moh, World Scientific, Singapore, 1992.

6. University Algebra N.S. Gopala Krishnan, Second edition, wiley Eastern

Limited, New Delhi, 1986.

7. Abstract & Linear Algebra - David M. Burton, Addison - Wesley publishing company,
London 1972.

8. Algebra M. Artin, Prentice - Hall of India; New Delhi 1994.

9. Algebra J.W. Archbold, Fourth edition, the English language book

·;ar.society, 1972.
\ I
,
10. A First Course in Abstract Algebra - John B. Fraleigh, Addison - Wesley Longman, Inc:,
Fifth edition, 1999.

Lesson Writer
Prof: U.M. Swamy
Andhra University,
Visakhapatnam.
Lesson· 7

Objectives:
After reading this lesson, you must be able to
(i) define the concept of an external direct product of groups and quote certain examples
and

(ii) define the notation of an internal direct product and prove that any internal direct
product is isomorphic to an external direct product and vice-versa.

Structure:
7.1 Introduction
7.2 External Direct Products

7.3 Internal Direct Products

7.4 Summary

7.5 Technical Terms

7.6 Model Examination Questions

7.7 Answer to Self Assessment Questions

7.8 Exercises

7.9 References

7.1 INTRODUCTION

When (A, .) and (B, *) are two groups, we can define a binary operation on the Cartesian

product AxB in a natural way as':-' ..

and, with respect to this binary operation, A x B becomes a group whose properties largely depend
on those of A and B. In this lesson we discuss certain necessary and sufficient conditions for a
group G to be represented as A x B where A and B are groups again.

7.2 EXTERNAL DIRECT PRODUCTS :

We are already in the hapit of using the same symbol + to denote the additionof real
numbers, addition of complex numbers, addition of matrices, addition of function and on se-veral
 Centre for Distance Education 2 Acharya Nagarjuna University

other occasions. There is no arnbiqnity if Wfi= are aware of what elements are to be added. Therefore,
we have agreed to denote the binary operation on an abstract group by the symbol, and even)his
is not mentioned explicitely, with this understanding, we are simply saying that' G is a group'

instead of' (G, -) is a group'.

7.2.1 Definition: Let A and B be two groups and define a binary operation, on the cartersian
product AxB.asfoliows: For any (abq), (a2,b2) E AxB define
'. . . '

(aI, q) (a2' b2) = (al a2, bI b2)

Note that al a2 is the product of al and a2 inA while q b2 is the product in A. A x B is called the.
external direct product of A and Band the binary operation defined on A x B is called coordinate
wise operation.

7.2.2 Theorem: LetA and B be two groups. Then A x B is a group under the binary operation as
defined above.

Proof: First we shall verify the associativity of the binary operation on A x B. For any

(al,q), (a2,b2), (a3,b3) EA x B , we have

((al,q)·(a2,b2))·(a3,b3) = (al a2, '1 b2)· (a3,b3)

~ ((al a2) a3, (bI b2) b3)

== (a] (a2 a3)' h] (b2 b3)) (by the associativity in A and B)

Therefore, . is associative on -Ax B . Next;' let us denote the identity elements in A and B
by the same symbol, namely e. Then (e, e) E A x B and

(a,b).(e,e)=(ae,be) = (a,b)=(ea,eb)~(e,e).(a,b) for all (a, b)EAxB and, therefore

(e, e) is the identity element in A x B . Also.Tor -any (Ct, b) E A x B,we have the inverses a-: and

b -I of a and b in A and B respectively arid

~ Lesson 7)

Th-erefore (a -1, b --1) is the inverse of (a, b) in A x B or, equivalently,

(a,b )-1= (-1

a .b -I).: Thus AxE is a group.

The above idea can be extended to the product of any finite number of groups. Explicitely,
we have the following theorem.

7.2.3 Theorem: Let AI, A2, ,An be groups and A =Ar xA2x.: x An' the cartesian product of

A1,A2, .... ,An· Forany a=(aba2,: ...,an) and b=(q,b2, ...;:,bn), define

ThenA is a group under this binary operation.

Proof: Simply imitate the proof of 7.2.2. This is left toyou as .e-xercise.'Prove this in detail.

7.2.4 Self Assessment Question: In the above group A=A1 xA2 x .... xAn' what is the identify
element and, for any a E A , what is the inverse of ~. in A ?

7.2.5 Definition: The group Al xA2 x ... xAn is called the external direct product of AI, A2, .... , An'

7.2.6 Self Assessment Question: Let A and E be groups, aEA and bE B. Suppose o( a )=n

in A and O(b )=m in B. Then prove that

O(a,b)=l'.cm.,{n,m}in AxB

7.3 INTERNAL DIRECT PRODUCTS:

Consider the additive group Z6 of integ~rs modulo 6. Then' '26 = {O,1,2,3, 4, 5}. Put

A = [o, 3} and B = {O, 2, 4} . Then A and B are subgroups of Z6 and hence we can treat A and B

as groups on their own. You can easily verify that every element in Z6 can be uniquely expressed

as a+b with a E A and bE B. In other words, the map

(a,b) H a+b: AxB~ Z6

 Centre for Distance Education 4 Acharya Nagarjuna University

is an isomorphism of the product A x B onto 76, Hene-s Z6 is isomorphic to the external direct
product of the groups A and B. This motivates the following.

7.3.1 Definition: Let G be a group and N}, N2, ... , N; be normal subgroups of G2, such that
every element g E G can be uniquely expressed as

Then G is called the internal direct product of Nl, N2, ...., N; and we denote this by writing
G=N1 EB N2 EB EBNn·

7.3.2 Self Assessment Question Prove that the group Z6 = A ffiB , where A={O,3} and

B={O,2,4} .

We shall prove that, if G is the internal direct product of N1, N2, ... , N; ' then G is isomorphic

to the external direct N} x N2 x .... x Nn. First, let us prove the following.

7.3.3. Lemma: Let NJ, N2, ... , N oe normal subgroups of a group G such that G is the internal

direct product of N1,N2, .... ,Nn. Thenthefollowingholdforany i::f:-j.

(2) ab = ba for any a E N, and b E Nj

Proof: Let 15, i ej <n

(1)
aENirlNj=:'>e e ae e =e e e e a e e
/h /h /h

=:'> a=e by the uniquness in 7.3.1

. (2) '., Let a ~ N, and b E N j . Then consider

(since a E N, and N, is normal)

(since b « Nj and Nj is normal)

 Therefore aba-1 b-1 E N, nNj ={e} and hence aba-1 b-1 =e, so that ab+ba .

7.3.4 Theorem: Let a group G be the internal direct product of the normal subgroups
NI, N2, , N; . Then

Where NI xN2 x xNn is external direct product of NI,N2, ,Nn .

By the lemma 7.3.3(2): we have aiaj =ajaj for all ai ENi and aj ENj and i-:t j. We
'\. ..'

~~:ovethat f is an isomorphism. For any a=(al, ....,an) and b=(q, .....,bn )ENl x ..... xNn,
we have

=al (qa2)b2a3b3····anbn

=al (a2q)b2 a3 b3 an s; (:a2 q =q a2 by 8.3.3(2P

= . J .•

=f(a) f(b).

Therefore f is a homomorphism. By the uniqueness in the definition 7.3.1, f is an injection.

Also, any element of G is of the form al a2 ....an for some ai E N, . Therefore f is a surjection too.
Thus f is an isomorphism and ,

NI xN2 x ..... xNn == G.

In the above theorem, we have proved that any internal direct product is isomorphic to an
external direct product. The converse of this is proved in the following.
 Centre fu;- Distance Education 6

7.3.5 Theorem: Let G}, G2, .... ,Gn be the groups and G=G1 xG2 x ..... xGn, the external direct
product of G}, G2, .... , Gn. For any l:::;i:::;n, let

Ni = {(aI, G2, .... , an ) E G/ a j = e for all J :;:.i}

Then N, is a normal subgroup of G, G is the internal direct product of N1, N2, .... , N; and

each G, is isomorphic to N; .

Proof: First note that N, consists of all n - tuples (e, e, a., e, , e) where, except the /h
component, each other component is the identity in G, . Clearly (e, e, e) E N, and hence N, :;:.¢ .

For any a=(aJ,a2, .... ,an) and b=(ht,b2, .... ,bn)ENi,wehave aj=e=bj for all i=! and

= (alol,-1' h-I
a2 2 , ,an n
h-I)

=(e,e, ..... ,e, aA-I,e, .... ,e)

where', except in the /h component, all are the identitives (in the corresponding groups

Gj ) and hence ab -1 E N, . Th-erefore M~Jisa subgroup of G . Also, for any a = (aI, a2,"" an ) E N,

and X=(Xi,X2, .... ,Xn)E G,

x ax -1 = ( xl, x2, ... , Xn) (a], a2 , ... , an ) ( Xl, X2, .... , XI1 r I

r=( Xl, X2, ... , Xn)( e, .... , e,ai,e, ... ,e)( Xl-I, XlI, .... , x;:;l)
r •
 C Algebra :e
and hence xa x-I E N, Thus N is a normal subgroup of G for each l:=;;i:=;;n. Also, for any

a=(al,a2, ...,an)E G, we can write

a=( aloe, ..., e).( e,a2,e, .... ,e) .... ( e,e, ...,e,"Cin)

Therefore any element can be expressed as an element in the product Nl N 2 .... Nn It can
be easily verified that this representation is unique also. Thus,

the internal direct product of Nl, N2, ..... , N; . Now, for any I:=;; i :=;; n ,define .Ii:G, ~ N, by

Then, you can easily check that .Ii is an isomorphism and therefore G, =:: N,

In the above two theorems, we have established that the external direct product and the
internal direct product are almost the same (up to isomorphism) and, for this reason, we say
simply say direct product in either case. In the next lesson, we shall prove that any finite abelian
)group is a direct product of a finite number of cyclic groups. .

7.3.6 Self Assessment Question: Let N}, N2, ... , N; be normal subgroups of a group G. Then

prove that G = Nl EEl N 2 EEl ..... EEl Nn ' if and only if, the following are satisfied.

In particular, G=N1EElN2 ifandonlyifG=NlN2 and NlnN2 ={e}.

7.4 SUMMARY:
In this lesson, you have learnt the concepts of an external direct product and an internal
direct product and proved that these two are same, upto isrriorphism.

7.5 TECHNICAL TERMS

External direct product
Normal subgroup

10
~~ Centre for Distance Education~;;; ~Jl~~~~~~~(~D.~Ch~a~ry~a~N~a~ga~r~ju~n~a ~U!!!ni~ve:!:r!!sit~YJE~

Internal direct product

Isomorphism

Coordinate wise operation

7.6 MODEL EXAMINATION qUESTIONS'

7.6.1.: If A},A2, .... , An are groups,:) ove that the product A=A}xA2x .... xAn is also a group
under coordinate-wise operation.
7.6.2. : Define the notion of an interna : irect product and prove that it is isomorphic to an external
direct product.

7.6.3. : Let G =G} x G2 x .... x On' wher-« :eh G, is a group, prove that there exist normal subgroups
N}, N2, .... , N; of G such that G is the internal direct product of NJ,N2,"" N; and G,=.Ni for all
l~i~n.

7.7 ANSWERS TO SELF ASSI:SSMENT QUESTIONS

7.2.4 Let us write e to denote identity e ? nent in each Ai Then the element (e, e, ...,e) is the
identity element in AI. since

7.2.6 n is the smallest positive integer such that an =e and similarly m for b. Let k = I c m {ll, m} .
We shall prove that k is thesmallest positive integer such that (a, b )k = (e, e), the identity element'

in A x B . Since n/ k and m] k , we can write nse k and mt=k for some positive integers sand
t. Now, we have
 ( Algebra ::e

Also, for any positie integer u,

~nl and ~ml

lu lu

k
Thus k is the smallest positive integer such that (a,b) =(e,e) and hence

o( a.b )=£.c.m. {Q( a), O(b)}

7.3.2: First note that 3 + 3= 0=2 + 4, 2 + 2 = 4 and. 4 + 2 = 2 if Z6 and hence A and B are (normal)
= =
subgroups of Z6 . Also, 1 3 + 4 and 5 3 + 2 . Therefore, every element of Z6 is of the form a + b

for some a E A and b E B . You can easily observe that it is unique also i for, if

a+b =a' +b', a, a' E A and b.b' E B,

then a-a'=b'-bEAnB={O} and hence a-a'=O=b-b'

.',

7.3.6.: First suppose that G = NI (f; N2 (f; (f; N; ; that is, any element of G can be uniquely

expressed a product as al,a2, ....,an, with aj ENj. Then clearly G=NI N2 ....N; and hence (1).

TOprOve(~), let l~i~n be fixed and a e Nin(N1 ..... Ni-1 Ni+1 ..:·.Nn).

Then a=al ·ai-l ai+l an for some aj ENj. By7.3.3, we have aaj =aj a and hence

al .....ai-l a-I ai+1 ....an = e e ..... e and e E Nj. By the uniqueness, it follows that aj =e

for all j =t.i and a-I =e, so that a =e . Thus

 Centre for Distance Education 10 Acharya Nagarjuna University

7.8 Excercises :
7.8.1. Let G1, G2, ...., Gn be groups and G = G1 x ..... x Gn . Prove that therE~are normal subgroups

G
A1,A2, .... ,An of G such that ~
I
= G, for all l~i=::;'n.
7.8.2. Let NI, N2, .....,Nn be normal subgroups of a group G such that G=NI ffiN2 ffi ..... ffiNn.

Let Ai =N1····Ni-l> Ni+1·····Nn foreach l~i~n. Then prove that Ai is a normal subgroups of G

G
and ~= Ni.
I

7.8.4. LetA be a group and G =A x A . Let H = {(a,a )/a E A}. Then prove thatH is a subgroups
of G and H = A. Also, prove that His normalin G is and only if A is abelian.

7.8.5 Express Z30 as the internal direct product of NI, N2 and N3. where NI, N2, N3 are (normal)

subgroups of 230 with O(NI)=2, O(N2)=3 and O(N3)=5 .

. 8.6 Prove that any finite abelian group is isomorphic to the direct product of its p - Sylow subgroups
7.8.7. Let A and B be cyclic groups of orders nand m respectively. Prove that the direct product
AxB is cyclic if and only if nand m are relatively prime.

7.8.8. Give an example of a group G and normal subgroups NI, N2, ..... , N; such that

G=NI N2 .... Nn and N, nNj ={e} and yet G is not the internal direct product of N, 's.

7.8.9. Let p be a prime. Prove that any group of order p2 is isomorphic to either Z p2 or Z p x Z p .

7.8.10. Let Gb G2, .... .G; be groups and G=GI xG2 x .... xGn. Describe the centre of G in terms
of the centres of G, 's .
 ( Algebra .:::e
7.9 References .
1. Topics in Algebra I.N. Herstein, Second edition, John wiley & sons, New
York,1975.

2. Basic Abstract Algebra P.B. Bhattacharya, S.K. Jain and S.R. Nagpaul, second
edition, Cambridge University Press, 1995.

3. Lectures in Abstract Algebra - N. Jacobson, Vol -I, Q. Van Nostrand Company, Inc.,
London, 1964.
4. Algebra Serge Lang, Addison-Wesley publishing Company,
London 1980.

5. Algebra T.T. Moh, World Scientific, Singapore, 1992.

6. University Algebrc. N.S. Gopala Krishnan, Second edition, wiley Eastern

Limited, New Delhi, 1986.
7. Abstract & Linear Algebra - David M. Burton, Addison - Wesley publishing company,
London 1972.

8. Algebra M. Artin, Prentice - Hall of India, New Delhi 1994.

9. Algebra J.W. Archbold, Fourth edition,the English language book

society, 1972.

'10. A First Course in Abstract Algebra - John'S. Fraleigh, Addison - Wesley Longman, Inc.,
Fifth edition, 1999.

--'I "'.: Lesson Writer

Prof: U.M. Swamy
Andhra University,
Visakhapatnam.
Lesson - 8

Objectives:
After reading this lesson, you should be able to
(i) Prove that any finite abelian group is a direct product of cyclic groups.
(ii) define the notion of invarients of a finite abelian group and prove that any two abelian
n
groups of order P are isomorphic if and only if they have the same invarients.
(iii) derive a formula for the number of non-isomorphic abelian groups of given order.

Structure:
8.1 Introduction
8.2 Fundamental theorem on finite abelian groups
8.3 Invariants
8.4 Abelian groups of a given order
8.5 Summary
8.6 Technical terms
8.7 Model examination questions
8.8 Answers to Self Assessment Questions
8.9 Exercises
8.10 References

8.1 INTRODUCTION :
In this lesson, we pay special attention to finite abelian groups. The reason is that no other
general class of groups has the structure as completely known as easily described. Whe!1 one is
setting up a structure theorey, the overall strategy is to express the complicated algebraic systems
in terms of those better behaved we accomplish this in the present lesson by proving that any finite
abelian group is a direct product of cyclic groups. We also derive at a formula to know toe number
of (non isomorphic) abelian groups of a given order.

8.2 FUNDAMENTAL THEOREM ON FINITE ABELIAN GROUPS:

It is well known that any cyclic group is abelian that any finite cyclic group must be isomorphic
to the additive group 7Ln of integers modulo n, where n is the order of the group. In this section we
 (i51Stiince Education 2 Acharya Nagarjuna University J

prove that any finite abelian group must be a direct product of finite cyclic groups. That is, in a

sense, the finite cyclic groups (or Zn 's) are like building blocks in the theory finite abelian groups.
We shall first prove the following which is given as an exercise in the previous lesson.
8.2 1 Theorem: Any finite ~elian group is the (internal) direct product of its P - Sylow subgroups.

Pro ot : Let G be a finite abelian grcup and o(G) = J!'1J p;2 p:r , where
P.I, !-'2, ·······'Pr are distinct primes and nl, n2, n; are non r.eqitive integers. By Sylow's
theorem ..I, G has a P - Sylow subgroup, for each prime J, end it should be unique (since G is
abelian and by Sylow's theorem - II). Foreach 1::;i s: r , let Pi bi~the unique Pi - Sylow subgroup
of G. Then,

Pi n (11 ··.... ·'~-l ~+I""""Pr,) = {e}

(since its order is a common divisor cf p? and a product of P j 's, j -;;t:. i).
Also, by inducuon w, A. you cen :')rov6' tnat

0(11 ., }-je) :oo: o(fl) 0(P2) · ·O(Pk) = p{l! p~} pZk

for any 1::; k S; r and, in particular,

. and therefore, 1i P2 P; =G, Thus G is the internal direct product of the Pi Sylow

subgroups Pi's. Note that O(Pi) = p;i .

8.2,2 Theorem (Fundamental theorem on finite abelian groups) : Any finite abelian group is a
direct product of cyclic groups.
Proof: Since any finite abelian group is the direct product of its p - Sylow subgroups (by 8.1.1), it

suffices to prove that any abelian group of order pn, where P is a prime, is a direct product of

cyclic groups. Let G be a finite abelian group of order pn. We shall use induction on n. If n = 0 or
n = 1, G is itself a cyclic group and we are through. Now, suppose n > 1 and suppose that the

theorem is valid for any abelian group of order p'", for all m < h.

Since 0 (G) = »" , the order of every element of G must be a power of p. Let a E G be an

element of maximal order, say o( a) = pk (k ::;n) and consider the cyclic subgroup H =(a)
=
generated by a. If k n, then H = G, so that G itself is cyclic in which case the theorem is trivial.
Thus we can assume that k < n and, in this case H is a proper non trivial subgroup of G. Now,
fiLesson - 8 3 Algebra D
consider the quotient group % whose order is pn-k < pn = o( G). By the induction hypothesis,

% is a direct product of cyclic subgroups. Suppose

% = HI ffi H2 ffi : ffi Hr
Where Hi is a cyclic subgroup of order pmi (1::; i ::;r) in %.
We can assue that Hi = H fn where Hi is a subgroup of G containing H. Since Hi is

cyclic, it is generated by element of Hi . Let the set b.H generate Hi . Then, we have

Hi = (biH) and o(biH) = pmi in %.

Now, we shall provethe following:

For each 1 S; i S; r, there exi sts Cj E b, H such that the

Q9 order of Cj in G is equal to the order of b.H in %; that

is ()()
ci = Pm i and c.H = b.H

To prove C8>
,let us write b for b, and m of mi temporarily, for convenience. Since

(bHl
lII

= H ,the identity element in %,

we get that b
pnI
E H = (a) and hence

since a is of maximal order ( by our choice of a ), we have 0( b) ::; 0 ( a) = pk. Since 0 (b)

o(b)~ k
is also a power of p, we get that ~ pk so that bP =e and therefore
(Distance Education 4 Acharya Naqarjuna University J

since o(a) = pk, it follows that Ph k-m and hence Ipk =spk-m which implies that

tp'" = s. Now, put e = ba -t . Then CE bH and hence cH = bH . Also,

m

cpm = (ba-t)P = bpm a -tp'" =a'~ -a -8 =e.

Therefore =:. Further (bHr(c) = (CHr(c) = co(c)H =cH = Hand,

,ml
since o(bH) = pm in %, we get that p la(c)' Thus we have

a(e) = pm = a(bH) and bH =cH .

This completes the proof of @.

Thus, for each 1~ i ~ r , there exists an element ci E G and an integer si = r" such that

Next, let K be the subgroup generated by cl, c2, , cr in G.

Then we shall prove that G = HEEl K ; thatts G =H K and H nK = {e} .

If x E G, then xH E % and hence

xH = (btH)tr (b2Hf2 (brH)tr for some integers 'I, '2" » rr and, since

b.H = c.H we get that

Thus G = HK . Again to prove that H n K = {e} , we have

fiLesson - 8 . 5 . Algebra D

=> (qHtl (brHt2 = (CIHtl (crHtr = Cfi c~r H =xH =H

=> (biHti =H (since any element of % can be uniquely expressed as a

'\ product of b.H 's)

=> o(b.Hr)/
I Iui for each 1:::::i :::::
r

=> o(c)( for each l:::::i:::::r

=> cri =e for each 1::::: r

i :::::
=>x=e.

Thus HnK ={e}, Therefore G =H EBK. Since o(K) < o(G) and 0(K) is a powerofp
(because o(K) divides o( G) = pn), we have, by the induction hypothesis, thatK is a product of

cyclic groups. If K = Kl (fJ K2 EB EBkw and K, 's are cyclic groups, then

G =H EBKl EB l.EBKw .

This completes the proof of the theorem.

8.2.3 Self Assessment Question: LetX be a set of n elements and G be the set of all subsets of
X Then G is an abelian group under the operation, defined by

A * B = (A-B) U (B-A)
Initiate the proof of 8.2.2 to prove that G is a direct product of n copies of Z2 ' the additive
group of integers modulo 2.

8.2.4 Self Assessment Question : Give an example of a non-cyclic abelian group of order pn,
where p is a given prime and n>1.

8.3 INVARIANTS:

With any abelian group of order pn, we shall associate a finite sequence of positive integers
whose sum is n and these integers are called the invariants of the group. This helps us in getting
/

(Distance Education 6 Acharya Nagarjuna University )

a formula for the number of distinct (non isomorphic) abelian groups of order pn, where p is a
given prime and n is a positive integer. We shall begin with the following:

8.3.1 Theorem : Let G be an abelian group of order pn, where p is a prime and n is a positive

integer. Then, there exists a unique list of positive integers mI, m2, , mr such that

and G = Al EB EB Ar ' where Ai is a cyclic subgroup of order pmi; and, in this case

G == Z pmJ x x Z pmr

Proof: By theorem 8.2.2, we get that

G = Al (fl A2 (fl EBAr

'Nhere AI, A2' , Ar are cyclic subgroups of G. Since o( Ad divides o( G) = pn ,

we have o(Ai) = pmi for some mi such that o<mi s;n . We can rearrange Ai's such that

and hence 0 < mI 5, m2 5, 5, mn ' Also,

and therefore n = mI + m2 + + mr . Further, since Ai is a cycilc group of order

Pmi , we have Ai == Z p mi and hence

G == Z p I11J. x Z p m2 x x Z mr
p

Thus we have obtained the required integers ml, m2, , mr.

Now, we shall prove that these are unique. Suppose

G = Al EB A2 EB EBAy = BI (J) B2 (J) ffi Bk .. ·· · (1)

where Ai's and B j 's are cyclic subgro rps of G with

o (A i ) = Pmi , 0 < I',

. , ... .-,."-..
fiLesson -8 7 Algebra

Then, we have to prove that r =k and mi =ni for all 1~ i ~ r .

Since Bk is a cyclic group of order pnk, Bk has an element of order pnk. But every

element of G is of order ~ p": (see 8.3.2 below). Therefore pnk ::::;

pmr. Bya similar argument,

pmr ~ pnk and hence mr = nk. Now, let r":' = s , for simplicity and convenience. Consider

By (1), we have

= B{ E9 E9 Bk-l ffi Bt

Since mi ::::;
mr-l for 1~ i ~ r -1, pmj and therefore XS =e for all x E Ai and hence

At is trivial for 1 < i ~ r -1. Hence o( G S

) = 0 (A,~) = o( Bk) . From this, it follows that o( Bj) =1

for all l s; j ~ k -1 and hence pnk-I divides s. Therefore pnk-I ~ pm,._I. Bya symmetric argument,

pJl1r -l/
/ pn -. and hence pmp_1 = pnk-I so ,t~?t mr-I = nk-I . Proceeding in this manner, we can
k I

prove that mr-i = nk-i for i = 0, 1,2, ·········iiyBut,

ml + m2 + +mr = n= nl +n: + +n«

Thus r =k and mi = ni for all 1~ i ~ r. This completes the proof of the theorem.

8.3.2 Self Assessment Ouestion : Let G be an abelian group and G = Al EEl EEl Ar ' where

each Ai is a cyclic group of order pmj with ml ::::;

m2 ~ ~ mr .

Then prove that every element of G is of order ~ pmr .

8.3.3 Definition: Let G be anabelian group of order pn, where p isa prime and n is a positive

integer. If G = Al EEl EBAr where each Ai is a cyclic subgroup of order pmj in G such
 (Distance Education 8 Acharya Nagarjuna University )

that 0 < ml ~ m2 ~ ~ mr, then the integers ml, m2, , mr are called the invariants of G.

In theorem 8.3.1, we have proved.that every abelian group of order pn has invariants and
that these are uniquely associated with the given group. We shall prove that two groups of order
pn have the same set of invariants. if and only if they are isomorphic. First note that the subgroups

Al ,Az, , Ar and their generators in the definition (8.3.3) of invariants are not unique. For,
consider the following:

8.3.4 Example: Let G = {e, a, b, ab} with a2 = e = b2 and ab = ba. Then G is an abelian

group of order 22. Let A= {e, a}, B = {e, b} and C = fe, ab}. Then G= A EB B,

G = A EEl C and G = B EEl C are three different decompositions of G into products of cyclic subgroups
though the invariants obtained are same.

8.3.5 Self Assessment Question: What are the invariants of the group G given in 8.3.4.

8.3.6 Theorem: Let p be a prime and n a positive integer. Then any two groups of order pn are
isomorphic if and only if they have the same.set of invariants.

Proof: Let G and G' be two groups of order 'pn. Let nil ' 1112, , 111r be the invariants of

G. Then G = Al EEl A2 EEl EEl AI'

Where Ai is a cyclic subgroup of order r" in G,

O<ml ~mz ~ ~ mr and ml +mi + +mr =n

Now, supposerj ~ G' and let .f : G ~ G' be ani'isomorphism. Then, put B, = j(Aj) for

] ~ i ~ r , You can easily check that B, is a cyclic subgroup of order pm, in G' for each 1:S i ~ r
and

G' =Bl EB B2 EB EBBr

By the unique ness of the invariants of G' (see theorem 8.3.1), it follows that

1111> m2, , 111r are the invariants of G'. Thus G and G' have the same invariants.
Conversely suppose that G and G' have the same invarients ml, mi> :',mr . Then

G = Al EB A2 EB \ EB Ar and G' = Ai EB A2 Efj EB A;

where Ai and B, are cyclic subgroups of order p": in G and G' respectively. Recall that
any two cyclic groups of the same order are isomorphic. Therefore there exist isomorphisms
fiLesson -8 9 Algebra D
gj : Aj--) r . Now, define g :G --) G' by
Ai for 1:::::i :::::

for any aj E Aj. You can easily verify that g is an isomorphism. Thus G == G' .

8.3.7 Self Assessment Question: Prove that the map g : G --) G' defined the proof of 8.3.6 is
an isomorphism.

8.3.8 Self Assessment Question: For any positive integer 111, prove that any two cyclic groups of
order 111 are isomorphic.

8.4 ABELIAN GROUPS OF A GIVEN ORDER:

In this section, we shall derive a formula for the number of non isomorphic finite abelian
groups of a given order. In this derivation, we shall mainly use the uniqueness of the invariants of
an abelian group of order pl1,where p is a prime, which is proved in the previous section. First, let
us recall the following form 5.6.2.

For any positive integer. I, by a partitio'n of n we mean a sequence of positive integers

I1IbI1l2, ,l11r such that 1111SI112 s <m; and 1111+1112 + +m. = n. We have

agreed to denote the set of partitions of n by P ( n ) .

8.4.1 Theorem: For any positive integer n, the number of non isomorphic abelian groups of order

pn is equal to the number of partitions of n, where p is a given prime.

Proof: Let n be a fixed positive integer and p a prime, we shall exhibit a bijection between the set

P (n) of partitions of n and the set of non-isomorphic abelian groups of order pn.

For any partition {111b 1112, , 111r} of n, let

G {I11I' 1112, ........... , mr } = Z p ml X Z

p
m2 X xZ
p
m;

Then G {111I' 1112, ........... , 111r} is an abelian group of order

ml m2 mr _ ml +m2 +. +rn ; _ 11
P P .p -p -p

and ml, 1112 , :.,l11r are invariants of this group. If {ml, ,mr} and

{n1' , nk } are partitionsof n such that

-
 (Distance Education 10 Acharya Nagarjuna Univer<;i!ij

G{mI> ,mr} =: G {n], ,nk}

Also, if G is any abelian group of order pn, then by theorem 8.3.1, there exists unique list of

positive integers {mJ, m2, ,mr } such that

and G = Al EB A2 EB EBAr' where Ai is a subgroup of order pmi in G. Then

{mI , m2 , ,mr} is a partition of nand

G{ml,m2,··············,mr}=Z p
11I1 x xZ p mr =:AIEBA2EB EBAr=G

Thus {ml , m2, , mr} HG {ml , m2 , ,mr} is bijection of the set P(n) of

partitions of n onto the set of non isomorphic abelian groups of order p" and hence these two sets
have the same number of elements. (Note that these two sets are both finite).
8.4.2 Self Assessment Question: How many non isomorphic abelian groups of order 81 are
.' there? list all these.

8.4.3 Theorem: Let PI> P2, .Pt: be distinct primes and nI, n2, : ,nk be positive

integers. Then the number of non isomorphic abelian groups of order p{11.p;2 p? is equal

to the product p (nd . p (n2) p (nk) , where p (nj) is the number of partitions of ni'

. . 111 112 11k G

. Proof: Let G be an abehan group of order PI P2 pk . Let i be the unique Pi Sylow

subgroup of G. Then o( Gj) '= p;i for 1 < i :s:;k and, by theorem 8.2.1,

G = GI EB G2 EB EBGk =: G1 x G2 x x Gk

Let p (nj) be the number of partitions of ni ' for each 1 :s:; i < k .

Then, by theorem 8.4.1, we have p (ni) number of choices for the ith place G; in the

above decomposition of G. Thus, the number of non isomorphic abelian groups of order

p~l p~2 pZk is precisely equal to p(nl) . p(n2) : p(nk),

 11 Algebra n
8.4.4 Self Assessment Question : How many non isomorphic abelian groups of order 600 are
there? and make a list of all these.

8.4.5 Self Assessment Ouestion": Describe all the non isomorphic abelian groups of order 1936.

8.4.6 Self Assessment Question: Prove that there is a unique (upto isomorphism) abelian group
of order 49445 and that it is cyclic.

8.5 SUMMARY :
In this lesson, you have learnt the fundamental theorem on finite abelian groups which states
that any finite abelian group is a product of cyclic groups. We have introduced the notion of invariants
of an abelian group of order pn , where p is prime, and proved that two such groups are isomorphc
if and only if they have the same set of invarients. Also, we have proved that the number of non
isomorphic abelian groups of order pn is equal to the number of partitions of n, and using this we
have derived a formula for the number of non isomorphic abelian groups of a given order.

8.6 TECHNICAL TERMS:

Abelian group
Fundamental theorem

Cyclic group

Invariants
Partition

8.7 MODEL EXAMINATION QtJJ~STIONS :

8.7.1 State and prove the fundamental Theorem on finite abelian groups.

8.7.2 For any prime p, prove that two abelian groups of order p" are isomorphic if and only if they
have the same set of invariants.

8.7.3 For any prime p and any positive integer n, prove that the number of non isomorphic abelian

groups of order pn isequal to the number of partitions of n.

8.7.4 State and derive a formula forthe number of non isomorphic finite abelian groups of a given
order.
 (Distance Education 12 Acharya Nagarjuna University )

8.8 ANSWERS TO SELF ASSESSMENT QUESTIONS :

8.2.3 For convenience, we shall denote the elements in G (= p (X), that set of all subsets of X)
by small letters a, b, c, etc., and write a-b for a*b (=(a-b)U(b-a)). Note that

o( G) = 2n and a- a = e (= the empty set ¢) and hence o( a) = 2 for all e:f:: a E G . As in

the proof of theorem 8.2.2, any element a :f:: e is of maximal order (namely 2) and hence we

can pick any element a"* e in G. Put H = (a) .. Then 0(%) = 2'1z = 2 11 1
- < o( G) and

by the induction hypothesis,

o /H(G/) -- HI- (J} CE-

(J}H r

where Hi is a cyclic subgroup of %. Note that, in this case, any cyclic subgroup is order

atmost 2 (since any element is of order 1 or 2). We can assume that Hi is non trivial and
hence Hi = {H, biH} for some b, E G such that b, tt. H. Then take ci = b, . a. Then
c.H = b.H and o(biH) = 2 = O(Ci). Now, you can proceed as in the proof of 8.2.2 and

conclude that G is a direct product of n copies of Z2 (Recall that any group of order 2 is

isomorphic to Z2).

8.2.4 Consider Z p x Z p x x Z p (n - times). This is an abelian group of order pn and is non

cyclic, since any element, except the identity, is of order p.

pmj
8.2.5 Note that a =e for any a E Ai (since o( Ad = pmj) Also, since 1111 < 1112 :-:; :-:; 111r '

we have

and hence apl11r =e for all a E Ai , 1:-:;i :-:;r .

Now, if x E G, then x =al a2 ar, with ai E Ai and

»": =a pmr pmr mr - ') m

X
l . a2 af! = e-e : e =e and hence. :; x :s;p r .

11
fiLesson - 8 . 13 Algebra »
8.3.5 a( G) =4 = 22 and the invarients of G are 1, 1, , since G has a decomposition
I

G =A ffiB, where a(A) = 2 and a(B) =2.·

8.3.7 The map g:G ~G' is defined by

Note that any element G can be uniquely written as al a2 ar with ai E Ai and that of G'
can be uniquely written as ai a2 a~ with at E Ai . Therefore g is well defined. since
each gj is a homomorphism, you can easily verify that g is a homomorphism .

. x' Ed=> x' =a~ a~ a; for some a; E A;

and since gj is an isomorphism, gj (ai ) = a; for aj E Aj .

=> x' =gl (al) g2 (a2)··········gr (ar) =g(ala2········ar)

and therefore g is a surjection. Also,

g(ala2······ar) =g(qb2 br) => gl (ad ······gr (ar) == gl (l)) ..·..·..·gr (br)
=> s, (aj ) = s. (L,) for all 1:s;i :s;r

=> ai = b, for all 1 ~ i .~ r

=> ala2 .....ar = ~ b2 . ·····A

and therefore g is an injection too. Thus g is an isomorphism.

8.3.8 Let G be a cyclic grouR of order m. Then G= (a) for some a whose order is m and

G = {2m-I}
e,a,a , ,a .
'.
You can easily verify that i H ai is an isomorphism of Zm onto G. Therefore G =.71m. If h
is another cyclic group of order m, then

G == Z m == H and hence G =. H .
8.4.2 Note that 81 = 34 and 3 is prime. By theorem 8.4.1, the number of non isomorphic abelian

groups c~ order 3: is equal to the number of partitions of 4 which are five in number (see

5.6.3) and these partitions and the corresponding groups of order 34 are given below.
 [Distance Education 14 Acharya NagarjunaUniversltYJ

{1, 1, 1, I} Z3 x Z3 x Z3 x Z3

{l, 1, 2} Z3 x Z3 x Z32
{I, 3} Z3 x Z33

{2, 2} Z32 x Z32

{4} Z34, which is cyclic of order 81.

8.4.4 We have 600 = 23 . 52 . 31

p (3) =3 and p (2) =2 where p (n) is the number of partitions of n. Therefore, by

theorem 8.4.3, the number of non; isomorphic abelion groups of order 600 is

p (3) . P (2). P (1) = 3·2·1 = 6 . To list all these six groups, we have to look atthe partitions of

3 and 2 and determine groups of order s23 , 52 and 31 and take products as below;

Partitions of 3 Groups of order 23

{I, 1, I} Z2 x Z2 x Z2

{I, 2} Z2 x Z22

{3} Z23

similarly, {I, l} and {2} are partitions of 2 and hence Z 5 x Z 5 and Z 52 are the only

group's of order 52. Now, we can list all the abelian groups of order 23,. 52 . 31 (= 600).
22 x Z2 x Z2 x Zs x Zs x Z3 (== Z2 x ZIO x Z3 ~'Z6 x ZIO x ZIO)
Z2xZ2xZ2xZ52XZ3 (=,Z6 xZ2xZ:SO)

Z2 x Z22 x Zsx Zsx Z3 ( =' Z]Cx Z20x Z3 =,Z30x Z20)

Z2 x Z22 x ZS2 x Z3 ( =' Zso x Z12 =,Z4 x ZIS0)

Z23X ZSxZSxZ3 (=' Z40xZlS =Z24XZSXZS)

Z23 x ZS2 x 113 ( == Z24 x 1l2S =,Z8 x Z7S)

8.4.5 1936 = 24 .112 I

The required number if p( 4)· p(2) = 5·2 = 10

fiLesson - 8 15 Algebra D
Partition of 4 Partition of 2 Group of order 24 . 112

{1, 1, 1, I} {I, I} Z2 x Z2 x Z2 x Z2 x Zu x Zll

{1, 1, 1, I} {2} Z2 x Z2 x Z2 x Z2 x Z121

{1, 1, 2} {I, I} Z2x Z2x Z22 x Zllx Zll

{1, 1, 2} {2} Z2 x Z2 x Z22 x Zl12

{1, 3} {I, I} Z2 x Z23 x Zll x Zll

{1, 3} {2} Z2 x Z23 x Zl12

{2, :) {l. I} Z22 x Z22 x Zll x Zl1

.{2, . 2} {2} Z22 x 222 X 2112

{4} {I, I} 224 x 211 x 211

{4} {2} 224 x Z112

8.4.6 We have 49445 = 51 .111.291.311 and 5,11,29 and 31 are primes. By theorem 8.4.3, the

'number of non isomorphic abelian groups is p (1) . P (1) . P (1) . P (1) = 1. Thus, there is

exactly one abelian group of order 5. 11 .29.31 and this is Z5 x 211 x 229 x 231 == 249445'

Note that Zn x Zm == 2nm if nand m are relatively prime.

8.9 EXERCISES :
8.9.1 Describe all the abelian groups of order
. ':~~,.
..

(d) 72. 5 (e) 210

8.9:> if G is an abelian group of order pn ,p a prime and 0 < ml ::::;

m2 ::::; mr are the invariants
::::;

of G, prove that the maximal order of any element of G is pm,. .

.85}:,j If a finite abelian group G has subgroups of orders m and n, then prove that G has a subgroup
whose order is the least common multiple of m and n.

8.9.4 Let n be a positive integer which is not divisible by m 2 for any integer m> 1. Prove that Z n is
the only group of order n.
 (Distance ~ducation 16 Acharya Nagarjuna University J
8.9.5 Let {ml, m2, : ,mk) be the invariants of an abelian group Goforder pn andHbea

non trivial subgroup of G. Let {nl , n2,'" ,nr } be the invariants of H. Then prove that

k ~ rand ni s mi for 1 :s; i <r .

A

8.9.6 Let G be a finite abelian group and G be the set of all homomorphisms of G into the
multiplication group of non zero complex numbers. Prove that G is an abelian group under
the operation defined by

(j.g)(x) = j(x)g(x) tor all t, gEG and xEG

8.9.7 For any j E G and x E G, prove that f (x) is a root of unity, if G is a finite abeuan group.

8.9.8 If G is a finite cyclic group, prove that G is also a cyclic group and o( G) = o( G) .

8.9.9 If G is a finite abelian group and X:f:. Y E G, prove that there exists j E G such that

j( x) :f:.j(y).

8.9.10 Prove that G == G for any finite abelian group G.

8.9.11 Let G be a finite abelian group and l:f:.

.
f E G, prove that .L' f (x)
XEG .
=0

8.9.12 Prove that 0(% (G)):f:.17 for any gorup G

8.9.13 Let G be a group of order PI P2·· ...····Pr ' where Pi's are distinct primes. Then G is cyclic

8.9.14 Prove that a non-cyclic finite abelian group G should cont~;~ a subgroup of order p2

8.9.15 Let G be a finite abelian group and fa (G). Prove that the number of solutions of x" = e
in G is a multiple of 11.

8.9.16 Prove that an abelian group of order pn, whose invarients are ml , m2, , mr ;

contains P" -1 elements of order p.

8.10 REFERENCES
1. Topics in Algebra I.N: Herstein, Second edition, John wiley & sons, New
York, 1975. !
ULesson - 8 17 Algebra D
2. Basic Abstract Algebra P.B. Bhattacharya, S.K. Jain and S.R. Nagpaul, second
edition, Cambridge University Press, 1995.

3. Lectures in Abstract Algebra - N. Jacobson, Vol -I, D. Van Nostrand Company, Inc.,
London, 1964.

4. Algebra Serge Lang, Addison-Wesley publishing Company,

London 1980.

5. Algebra T.T. Moh, World Scientific, Singapore, 1992.

6. University Algebra N.S. Gopala Krishnan, Second edition, wiley Eastern

Limited, New Delhi, 1986.

7. Abstract & Linear Algebra - David M. Burton, Addison - Wesley publishing company,
London 1972.

8. Algebra M. Artin, Prentice - Hall of India, New Delhi 1994.

9. Algebra J.W. Archbold, Fourth edition, the English language book

society, 1972.

10. A First Course in Abstract Algebra - John B. Fraleigh, Addison - Wesley Longman, Inc..
Fifth edition, 1999.

Lesson Writer
Prof: U.M. Swamy
Andhra University,
Visakhapatnam.
LESSON 9

DEFINITIONS, EXAMPLES AND SOME SIMPLE

RESULTS OF RING THEORY

Objectives

The objectives of this lesson are to:

-- . Introduce different concepts like ring, commutative ring, fields, division ring (or skew field)
• Discuss some examples of different types of rings. ,
• Understand some well known and unknown sets which forms a ring with respect to suita~le
operations: addition and multiplication.
• Introduce the concepts like integral domain, characteristic of an integral domain, n-torsion
property.
• Prove some fundamental results related to the known concepts
~ Understand the relation between the concepts: field and integral domain.

Structure
9.0 Introduction
9.1 Some Definitions and Examples
9.2 Integral Domain
9.3 Boolean ring
9.4 Some prcl iminary results on rings
9.5 Summary
9.6 Technical Terms
9.7 Answers to Self Assessment Questions
').8 Model Questions
9.9 Reference Books

9.0 INTRODUCTION

Ring is a fundamental abstract concept in the study of algebra. A group is endowed with
only one binary operation but a ring is endowed with two binary operations connected by some
inter relations. We shall give an axiomatic definition of a ring and study some .of its elementary
properties.

Despite the differences, the analysis of rings will. follow the pattern already laid out for
groups. Study of rings serves as one of the fundamental building blocks for the abstract algebra.
 jAlgebra 9.2 Definitions, Examples and .... 1

From the study of groups, we shall be able to make the requisite definitions, intertwine
them with meaningful theorems and finally, we end up proving results that are both interesting
and important about mathematical objects with which we have had long acquaintance. To give
one example, using the tools developed here, we can prove that it is impossible to trisect an angle
of 60° using only a straight edge and compass in forth coming lessons.

It is clear that the definition of a ring is an abstraction of the ring Z of integers. Although
rings are a direct generalization of the integers, certain arithmetic facts to which we have become
accustomed in the ring of integers need not hold in general rings. For instance, we know that the
product of two non-zero integers is non-zero, but this may no longer be true in a general ring. In

the ring R of 2 x 2 matrices. we will come across the situation that (lO 0)1(.0
l~ =
,0 1 ,0 0). \,0 0)
(0 0'- Thus

(0 1\ 0 0'\
. even though both ( I I and
I are non-zero, their product is zero in R. This leads to the study
\9 1) \0 OJ
of some special classes of Rings: Integral domains,\ division rings and fields. Also we state the
pigeon hole principle which i." useful in proving the theorem that states that 'a finite integral
domain is a field'.

9.1 < SOME DEFINITIONS AND EXAMPLES

9.1.1 Definition: A non empty set R is said to be a ring (or an associative' ring) if there exists
two .operations + and. on R such that
(i) (R, +) is an Abelian group;

(ii) (R, .) is a semi-group; and

(iii) for any a, b, c E R we hfly;ta (b + c) = ab;+ ac, (a + b)c = 3C + bc.
By spelling out details of the definition of the group (R,+ ), we may define the algebraic SYStCb1
")~j~\

ring as follows:
Let Rbe a nonempty set on which two binary operations + and'·' are defined. Then (R, + .. ) j·s,
\

a ring if it satisfy the following nine conditions:

(i). a + b E R Ji' all a, b E R
(ii). (a + b) + c = a + (b + c) for all a, b, C E R
-, ..
(iii). there exists an element '0' in R such that a + 0 = a = 0 + a for all a E R
I
(iv: for each a E R there exists an element b E R such that a + b = °= b + a (Here, the element '0'

is called as the additive identity element of R)

(Here, the element b is called as the additive inverse of a, and it is denoted by '- a').
JAcharya Nagarjuna University 9.3 Centre for Distance Educatio~

(v). a+b=b+aforalla,bER

(vi). a-b E R for all a, b E R

(vii). (ab)c = a(bc) for all a, b, c E R
(viii). a-rb + c) = a-b + b-e for all a, b, c E R {this is called as left distributive law)
(ix). (a + bj-c = a-c + b-e for all a, b, c E R (this is called as right distributive law)

9.1.2 Definition: Let CR, +, .) be a ring.

If 1 E R such that a.1 = 1.a = a for every a E R, then we say that R is a ring with identity (or unit)
element.
If a.b = b.a \;j a, b E R, then R is said to be a commutative ring.

9.1.3 Example: We write 2 = { ... ,-3, -2, -1, 0, 1,2,3, ... }, the set of integers. Then the
algebraic system (Z, +, .), where + and '. ' are the usual addition and multiplication of numbers, is
a commutative ring with identity.

9.1.4 Example: We write 2Z = {... ,-4, -2, 0, 2, 4, ... }, the set of even integers. Then the

algebraic system (22, +, .), where + and '. ' are the usual addition and multiplication of numbers,
is a commutative ring with out identity.

9.1.5 Example: We write Q = {p / p,q are integers such that q is nonzero}, the set of rational
q
,I ~ \

numbers. Then the algebraic system (Q; +, .), where + and '. ' are the usual addition and
multiplication of numbers, is a commutative ring with identity.

9.1.6 Self Assessment Question

Find the multiplicative inverse of a given non - zero rational number

9.1.7 Example: Let n be a positive integer grater than or equal to 2. We write Z, = {O, 1,2, .

3, ... , n-l}, the set of integers modulo n. Then the algebraic system (ZI1'+, .), where + and '.' are
the addition modulo n; and multiplication modulo n, respectively, is a commutative ring with
identity element. Recall that if a, b are in
IAlgebra 9.4 Definitions, Examples and .... 1

Z, then a + b is the remainder when we divide a+b by n. Similarly a-b is the remainder when
wedivide ab by n.

9.1.8 Definition: Let R be a ring and S be a nonempty subset of R. If S is a ring with respect to
the operations on R, then we say that S is a subring of R.

9.1.9 Example: Consider the ring Z of integers. Consider 2Z the set of even integers. We know
that 2Z and Z are rings with respect to usual addition and multiplication of numbers. Also 2Z is a
nonempty subset of Z. Hence 2Z forms a subring of Z.

9.2 INTEGRAL DOMAINS AND FIELDS

9.2.1 Definition: If R is a commutative ring, then a non-zero element a E R is said to be a zero

divisor ifthere exits a non-zero element b E R such that ab = O.

9.2.2 Definition: (i) A commutative ring is said to be an integral domain if it has no zero
divisors.
(ii) A ring R is said to be a division ring if (R*, .) is a group (here R* = R - {O}).
(iii) A division ring is said to be a field if it is commutative. We can also define the concept 'field'
as follows:
I
An algebraic system (F, +, .) where F is a non-empty set and +,. are binary operations, is said to
be a field if it satisfies the following three conditions:
(i)(F, +) is an Abelian group;
(ii) (F*, .) is a commutative group where F* = F - {O};and
(iii) a(b + c) = ab + ac, (a + b)c = ac + be for all a, b;c in F.

9.2.3 Example: The set of integers forms an integral domain.

9.2.4 Example: The set of rational numbers is a division ring and also a field.

9.2.5 Example: The set of real numbers forms a division ring as well as afield.

9.2.6 Example: Let R be the set of all real valued continuous functions on the closed unit"interval
[0, 1].
 IAcharya Nagarjun;a University 9.5 Centre for Distance EducatiorJ

For f, g E R and x E [0, 1] define (f + g)(x) = f(x) + g(x), and (f.g) (x) = f(x) . g(x).
(i) Define two functions f and g on [0,1] as follows:

f(x) = 1 - 2x if 0 ::;x ::; 1/2 and f(x) =0 if 112::; x ::; 1; and

g(x) = 0 if 0::; x ::; 1/2 and g(x) = 2x - 1 if 112::; x ::; 1.

Now the functions f and g are continuous real valued functions such that f 7= 0, g 7= 0 and fg = O.
Thus we can conclude that the ring.R is not-an integral domain.
(ii) Define h: [0, 1] -7 R by hex) = V2 - x if 0 ::; x ::; Y2 and hex) = 0 if 'h ::; x < 1. Then h is

continuous and so hER.

Now we verify that h do not have its inverse. In a contrary way, suppose hi E R such
that hhl = e =} (hhl) (1/2) = e(I/2) = 1 =} h(1/2). hl(1I2) = 1
=} O. h1(1/2) = I=} 0 = 1, a contradiction. Hence h do not have its inverse.
So we can conclude that the ring R is neither a division ring nor a field.

9.2.7 Self Assessment Question

Find two non - zero elements in Z 8 whose product is zero.

9.2.8 Example: Let F be the field of real numbers and let Rbe the set of all formal square arrays

(: :J where a, b, c, d are any real numbers.

We define (al ~J+ (~ ~J = (al +~ ~ +~J. It is easy to verify that R forms an Abelian group
<; ~ '2 ~ <;+'2 ~+~ . ,

under addition with (~ ~J acting as the zero element and (: ~ J is the negative (inverse) of

. (: :}
We define the multiplication in R by (a dbJ.(f sJ = (ar+bt as+bul.
c tu cr+dt cs+du]

The element (~ ~) acting as multplicative unit element. Clearly R is a ring .

..Since (~ ~} (~ ~) = (~ ~} we have that R is not an integral domain.

 IAlgebra 9.6 Definitions, Examples and .... 1

tnce G ~. (~ ~)~ (~ ~)* (~~)~(~ ~J(~~} we have that R is not a commutative ring.

sp we can conclude that R is not an integral- domain.

9.~.9 Example: Let C be the set of all symbols (a, b) where a, b are real numbers. We define
(a.b) = (c,d) ¢:> a=cand b=d .
./
In C we introduce an operation addition '+' defined by
(a.b) + (c,d) = (a+c, b+d).
Now the set C becomes an additive abelian group with respect to the operation defined above.
The element (0,0) acts as the additive identity; and the element (-a, -b) acts as the additive
inverse of (a,b).
Now we define an operation which serves as a multiplication on C.

The multiplicative operation is defined by

(a,b)'(c,d) = (ac ~bd, ad + be).

Now the system (C,+,. ) becomes a commutative ring.

The element (1,0) acts as the identity element in C with respect to multiplication.

For any nonzero element (a, (3) in C, the element [ 2

a +13
a 2' 2- ~ 2
a +13
J. acts as the multiplicative
inverse.
Thus the system (C,+,' ) satisfies all the axioms offield, and so system (C,+,. ) is a field.

9.2.10 Self Assessmentrstion

In the 'above example;

/' .
verify that (a, 13)' [
a
2a
+W~
,-, 2-
a +13
13 2 J = (1, 0).

().2.11 Example: Let F be the field of real numbers and consider the set of all formal sym,b~~

+- ad + a2.i + a3k, where ao, aI, a2, a3 E F. Equality and addition of these symbols arc defined as/
follows:
IAcharya Nagarjuna University 9.7 Centre for Distance Educationl

We now come to the multiplication. When Hamilton discovered on Oct 6tt, 1843, he cut
the basic rules of this product out with his Penkrure on Isrougnam bndge mlJublm.l he product
base d on 1,2 = J,2 = k2 = - 1"
,1J= k ,Jik = 1,
Lki1= Ji and
an .. = -k , kiJ = -1,i.Ik
J1 '
1 = -].

kOj
1

If we go around the circle Clock wise the product of any two successive ones is the riext one, an
going around counterclockwise we get the negatives. We can write out the product of any tw
quaternions

(ao + ad + aL.i+ a3k) (~o + ~li + ~L.i+ ~3k) = Yo + Yd + Y4i + Y3k where
10 = ao0o - al01 - a202 - a303
II = ao01 + al00 + a203 - a302
\ ----- (1)
12 = a002 - a103 + a200 + a301
13 = ao0o + al02 - cX20l+ a300
If some aj is 0 in x = ao + ad + a4i + a3k, we shall omit it in expressing x; thus 0+ Oi+ OJ +-

Ok will be written as 0; 1 + Oi + OJ + Ok as 1; and 0+ 3i + 4j +Ok as 3i +

4j.
It can be observed that
(ao + ali + aL.i + a3k) (ao - ad - aL.i- a3k) = a02 + al2 + al + al .....(2).
This has a very important consequence.
Suppose thatx = ao + ad + aL,j+ a3k -:;t:. 0 (so some ai -:;t:. 0).

Now the a's are real, 0 = ao2 + al2 + a/ + a/ -:;t:. O.

From (2). we get that (ao + all . + a:?J. + a3 k) (a O

---, a-1--J--
J• a2· a~ k) =v 1
i .
, ' ~ ~ ~ ~

So if x -:;t:. 0, then x has an inverse in the set of all quaternions. Thus the set of all quaternions form
a non-commutative division ring.
So we can conclude that the ring Quaternions forms a division ring which is not a field.

9.2.12 Self Assessment Question

Show that every field is an integral domain.
/Algebra 9.8 Definitions, Examples and .... 1

9.2.13 SelfAssessm it QU'..stion .

Find out two examples for an integral domain which are not fields.

9.2.14 Result: Let D be an integral domain, a, b E D. Suppose an = b" and am = b" for two
relatively prime positive integers m and n. Then prove that a = b.

Proof: We know that every integral domain satisfies the property that xy =0~ x= 0 or y = O.
Now the proof follows from the Result 9.1. 7.

9.2.15 Problem: Let p be a prime number and Q(-Yp) = {a + bvp / a, b E Q} where Q is the set
of all rational numbers. Then show that Q(-Yp) is a field under the usual addition and multiplication

of real numbers.

Solution: It is clear that Q(-Yp) ~ R where R is the set of all real numbers.

Since R is a field, it is enough to show that

(i) (Q(-Yp), +) is a subgroup of (r., +); and

(ii) (Q(-Yp)*, .) is a subgroup of (R*, .),

First we verify (i).

Let a + b-Yp, c + d-Yp E Q(-Yp)
~ (a + bvp) + (c + d-Yp) = (a + c) + (b + d) -YpE Q(-Yp).
Therefore Q(-Yp) is closed under addition.

Let a + b-Yp E QC-Yp).Then a, bEQ ~ -a, -bEQ => (-a) + (-b) -YpE Q(-Yp).

Also (a + b-Yp) + [(-a) + (-b) -Yp] = 0 Hence -(a + b-Yp) = -a -bvp E Q(-Yp).

Therefore every element ofQ(-Yp) has inverse(additive) in Q(-Yp).

This shows that (Q(-Yp), +) is a subgroup of (R, +).

(ii) Let a + b-Vp, c + d-Vp E Q(-VP)*

=> (a + b-Yp) (c + d-Yp) = (ae + pbd) + (ad + be) -YpE Q(-Ypr

Therefore Q(-Yp>*is closed under multiplication.

Now for a + b-YPEQ(-Yp)* ~ a+bvp e 0 and a, b E Q. If b = 0 then a2 - pb2 = a2:1= O.

If b :1= 0 then a/b is a rational number and so aIb :1= -Yp~ a :1= b-Yp ~ a2 :1= b2 P
IAcharya Nagarjuna University 9.9 Centre for Distance EducaiIQQI

2 2.· 2 2
=> a -: pb :;c O. Thus in both cases, we get that a - pb 7:- O.

Now a/(a2 - pb2) and b/(a2 - pb") E Q and so a/(a2 - pb'') - b/(a2'- pb"). -Vp E Q (-Vp). Now it is

easy to verify that (a + bvp) [a/( a2 - pb") - b/( a2 - pb2)-vp] = 1.

Hence (a + bvp)" E Q(-Vp).Therefore (Q(-Vp)*, .) is a subgroup of (R*, .).

This shows that Q(-Vp) is a sub field of R. Hence Q(-Vp) is a field.

9.3 BOOLEAN RING

9.3.1 Definition: A ring R is said to be a Boolean ring ifx2= x for all x E R

9.3.2 Self Assessment Question: If a, b, c, dER, R is a ring, then evaluate (a+b) (c+d).

9.3.3 Self Assessment Question: Prove that if a, b E R, then (a+b)/. = a2 + ab + ba + b2.

9.3.4 Result: If R is a Boolean ring then R is a commutative ring.

Proof: Let a E R. Consider a2 + a2 = a + a = (a + a)2 = a2 + a2 + a2 +a".

Therefore ---- (1).

Now for any a, b E R, consider

a + b = (a + b)2 = a2 + ab + ba + b2 = a + ab + ba + b => 0 = ab + ba

=> ab = -ba = ba (by (1». Hence R is commutative.

9.3.5 Result: Let R be a ring such that x3 = x for all x E R. Then R is a commutative ring.

Proof: For any x E R, we have

2x = (2x)3 = (x + X)3 = (x + X)2 (x + x) = (x2 + 2X2 + X2)(X + x) = (4X2) (2x) = 8x3 = 8x.
Therefore 8x = 2x implies 6x = 0 for all x E R (i)
(x2 _ x) = (x2 _ x)3 = x6 _ 3x5 + 3x4 _ x3 = X2_ 3x3 x2 + 3x3 X _ x3

= x 2 - 3 x + 3 x 2' - x = 4 x 2 - 4 x .. , (iIIi)

=> 3(x2_ x) = 3(4 x2 - 4x) = 12 X2 -12x = 6x(2x -2) = 0 (by (i)

2
=> 3(x - x) = 0 implies 3x2 = 3x for all XE R (iii)

Let S = {3x I x E R}

Now S is a subring of R. [Verification:s, t E S implies s = 3x, t = 3y for some x, y E R. Now s-t

= 3x - 3y = 3(x - y) E S. This shows that S is a subgroup of (R,+).

Also s-t = 3x·3y = 3(3xy) E S.
 !Algebra- 9.10 Definitions, Examples and .... 1

Now it follows that S is a ring, and so S is a subring of R].

Let yE S. Then y = 3x.
Now y2 = (3x)(3x) = 9x2 = 6x2 + 3x2 = 3x2 (by (i) = 3x (bytiii) = y.
Now we proved that S is a ring in which y2 = Y for all YES.
Therefore by result 9.3.2, S is a commutative ring.
So (3x)(3y) = (3y)(3x) irnplies9xy = 9yx implies 3xy = 3yx (since 6xy = 0 by (i)
Now x + Y = (x + y)3 = (x:+ y)2(X+ y) = (x2 + xy + yx + y2) (x + y)
= x3 + x2y ~ xyx + xy2 + yx2 + yxy + y2 X + y3
= X + x2y + xyx + xy2 + yx2 :I- yxy + y2 X + Y
= x2y + xyx + xl + yx2 + yxy + l
~ 0 x (i)A
By the given property x - y = (x _ y)3

=> -x 2y - xyx + xy 2 - yx 2 + yxy + y.2 x· =:0 - (··)A

11

Adding (i)A and (ii)A we get

2xy2 + 2yxy + 2y2x = 0; (iii)A
Multiply (iii)A by Y on right, we get
2xy + 2yxy2 + 2y2xy =0 (iv)A
Multiply (iii)A by y on left, we get
2yxy2 + 2y2xy + 2yx = 0T (v)A
Subtracting (v)A from (iv)A, we get 2xy = 2yx. But3xy = 3yx
Therefore xy = yx. This is true for all x, y E R. Hence R is commutative.

9.4 SOME PRELIl\lINARY)~ -llL'[SON ~INGS

, '"
9.4.1 Lemma: If R is a ring then for all a,b E R we have
(i)aO = 0 = Oa;
(ii) ar-b) = (-a)b = -ab;
(iii) (-a)( -b) = ab.
In addition.itR pa~)dentity 1, then
(iv) (-1) a.= -a, (-1) (-1) = L

Proof: (i) aO = a (0 + 0),= aO + aO. Now 0'+ aO= aO = aO + aO ~ aO = 0 (by Right Cancellation
Law). Similarly Oa + 0 =.Oa = (0 + O)a = Oa +Oa => 0 = Oa.
IAcharya Nagarjuna University 9.11 Centre for Distance Education)

(ii) 0 = aO = a(b+(-b» = ab + a(-b) => -Cab) = a(-b) and 0 = Ob = (a + (-a»b

= ab+(-a)b~-(ab)=(-a)b.

(iii) (-a) (-b) = -(a (-b» = -(-Cab»~= ab. (iv) from (ii), (-l)a = -(1 a) = -a.
(v) from (iii), (-1) (-1) = 1.1 =1.

9.4.2 Noteu'I'he Pigeon Hole Principle) If n objects are distributed over m places and if n > ill

then some place receives at least two objects.

9.4.3 Lemma: A finite integral domain is a field.

Proof: We know that in an integral domain we have ab = 0 => a = 0 or b = O. Now it is enough tc
show that every non-zero element has multiplicative inverse. Let D be an integral domain. Now
we show (i) there exists 1 E D such that a . 1 = a for all a E D, (ii) 0 if; a ED=> there exists bED

such that ab = 1.
Let D = [x., X2, ... ,xn} and O:i:- a E D. Now xia, X2a, ... , x.,a are all distinct (for x.a = xja
=> (Xi- xj)a = 0 => Xi-Xj = 0 => Xi= Xj(since a if; 0». Therefore
= Xkafor some 1 ~ k ~ n (since a E D). Since D is commutative, we have Xka = a = axr. We show

Xkis the identity element. For this, let y E D, then y = x.a for some i. Now consider y.x, = (Xia) x,
:=: Xi (axj) = x.a = y. Thus yXk= Y for all y E D. Therefore Xkis the identity element. Now for XkE
D = {XIa, X2a, ... , x,a} => Xk= xja for some 1 ~ j ~ n. Therefore Xj is the multiplicative inverse of
a. Hence D is a field.

9.4.4 Corollary: If p is a prime number, then Jp (that is Zp), the ring of integers modulo p, is a
field.

Proof: We know that Jp contains exactly the elements 0, 1,2, ... (p-I). Let a, b E Jp such that ab ==
.
o (mod p).

Now ab == 0 (mod p) => ab - 0 is divisible by p => P divides ab

=> P divides a or p divides b (since p is prime)
=> a-O is divisible by p or b - 0 is divisible by p
 /Algebra 9.12 Definitions, Examples and .... 1

=> a == 0 (mod p) or b == 0 (mod p)

o
Therefore Jp is an integral domain, and by above Lemma 9.4.3, J, is a field.

9.4.5 Definition: An integral domain D is said to be of characteristic 0 if the relation ma:;tO

where 0 :;ta E D and m is an integer, can hold only if m = O. An integral domain D is said to be of

finite characteristic if there exists a positive integer m such that ma = 0 V a E D. If D is of finite

characteristic, then we define the characteristic of D to be the smallest positive integer p such that
pa = 0 for all a E D. It will be denoted by Chao R = p.

9.4.6 Examples: (i) (Z, +, .) is an integral domain with characteristic O.

(ii) (Zp, +, .) is an integral domain with characteristic p. Because for x E Zp, we have x = a. Now
px = pa = a + a + ... + a (p times) = ap = a.O = 0 (since p = 0 in Zp).
, (iii) (Z6, +, .) is a commutative ring, but not an integral domain (because 2.3 = 6 = 0, but 2 :;t0 :;t
3).

9.4.7 Result: R is a ring, a, be Rand m, n are integers, then (na)(mb) = (nm)(ab).

Proof: Consider (na)(mb) = (a + a + ... + a) [ n times] (b + b + ... + b) [m times]
= ab + ab + ... + ab (mn times) = (mn)(ab).

9.4.8 Result: Show that the characteristic of an integral domain is either zero or a prime number (
alternative statement: D is an integral domain then if D is of finite characteristic, then prove that
... ( ,...
the characteristic of D is a prim~' number).

Proof: Let D be an integral domain. Suppose the characteristic of D is non-zero (i.e. D is of finite
characteristic) and char D = p. "
Now we have to. show that this p is a prime number.
If p is not prime then p = mn for some integers 1 < m < p and 1 < n < p.
If nx = 0 for all xED then it is a contradiction to the minimality of p (note that p is, the

characteristic means p is the least positive integer such that px = 0 for all x E D). Therefore mxee
 t..'..

IAcharya Nagarjuna University 9.l3 Centre for Distance EducatioIlj

o for some XED. Fix y E D such that ny ;;j:. O.Now for any xED, consider (mx) (ny) = (mn) (xy)

= p(xy) = 0 => rnx = 0 (since ny r 0 and D is an integral domain). Therefore mx = 0 for all xED.
This is a contradiction to the rninimality of p. Hence p is a prime number. Therefore the
characteristic an integral domain is either zero or a prime number.

9.4.9 Result: If D is an integral domain and if na = 0 for some 0 ;;j:. a E D and some integer n ;;j:. 0,
then prove that D is of finite characteristic.
Proof: Without loss of generality, we may assume that n is a positive integer. (otherwise, (-n) a = -
(na) = -(0) = 0).
Now na = 0 =? n(1a) == 0 =? (nl)a = 0 =? n = 0 (O;;j:. a E D and D is an integral domain). For any x
E D we have nx = n(l. x) = (n. l)x = O.x =0.

Therefore there exists a positive integer n such that nx = 0 for all x E R.

This completes the proof.
9.4.10 Result: Suppose D is a commutative ring. Then D is an integral domain if and only if 0 ;;j:.

a, b, C E D such that ab = ac => b = c (cancellation law).

Proof: Suppose D is an integral domain. Suppose ab = ac.

Now ab = ac =? ab-ac = 0 ~ a(b-c) = 0

=? b-e = 0 (since 0 ;;j:. a E D and D is an integral domain) => b = c. Conversely,
suppose that ab = O.
Now ab = 0 => ab = 0 = a.O => b = 0 (by our cancellation law).
Hence D has no zero divisors. This shows that D is an integral domain .
. ~-.'

9.5 SUMMARY:

The abstract algebraic concepts: ring, ring with unit element, commutative ring, integral
domain, division ring, field were introduced. The set of integers & the set of rational numbers
with usual addition and multiplication of numbers f01111commutative rings with unit element. The
set of even integers under the usual operations of addition multiplication forms a commutative ring
without unit element. The set of integers" modulo n ( where n is a positive integer with n ;:::2)
forms a finite commutative ring. Write C = {(x, y) / where x, yare real numbers}. This set forms
IAlgebra 9.14 Definitions, Examples and' .... I

a field with respect to the operations + and· defined by (x, y) + (a, b) = (x + a, y+ b) and (x, y)-
(a, b) = (xa - yb, xb + ya). The ring of real quatemions form a division ring but not a field.

A finite integral domain is a field. For any prime number p, the set of integers modulo p forms a
field. An integral domain D is of finite characteristic if there exist a positive integer 111 such that

ma = 0 for all a E D. If D is of finite characteristic then the smallest positive integer p with pa = 0

for all a E D, is called the characteristic of the integral domain D. This p is a prime number.

9.6 TECHNICAL TERMS

Ring with unit element: Let (R, +, .) be a ring. If 1 E R such that

a.I = l.a = a for every a E R, then we say that R is

a ring with identity (or unit) element.

Commutative ring: If a.b = b.a '\I a, b E R, then R is said to be a

commutative ring.

Division ring: A ring R is said to be a division ring if (R*, .) is a

group (here R=* R - {O}).
Zero divisor: If R is a commutative ring, then a non-zero element
a E R is said to be a zero divisor ifthere exits a
non-zero element b E R such that ab = O.

Integral-domain: A commutative ring is said to be an integral domain

if it has no zero divisors.
Field: A division ring is said to be a.field if it is
commutative.
The pigeonhole principle: If n objects are distributed over m places and if
n > m then some place receives at least two
objects.

Finite characteristic: An integral domain D is said to be of finite

characteristic if there exists a positive integer ill

such that ma = 0 V a ED.

Characteristic: If D is of finite characteristic, then we define the
characteristic of D to be the smallest positive
integer p such that pa = 0 for all a E D.
IAcharya Nagarjuna University 9.15 Centre for Distance Educationl

9.7 ANSWERS TO SELF ASSESSMENT QUESTIONS

9.1.6: Let Q be the set of all rational number and i:- q E Q.

Then q = ~, a, b are integers with b i:- O. Since q i:- 0 we have that a i:- O.
b
·N
1 ow b"Q
IS m
.r: suc Ih b q = -.ba - = 1.
1 t at -.
a a a b
Thus ~ is the inverse of the given non - zero element q = ~ E Q.
a ' b

9.2.7: We know that 8 = {O, 1,2,3,4,5,6, 7}. Write a = 2 and b = 4.

Then a i:- 0 i:- b. a- b = 2· 4 == 8 == O.

Thus a = 2 and b = 4 are two non zero elements of Z8 whose product is zero.

9.2.10: (a, ~)- ( 2 a 2' 2-~ 2 J

Ct +~ a +~

9.2.12: Let F be a field. We have to show that F is an integral domain. For this take two nonzero
elements a, b in F.
In a contrary way, suppose that ab = O. Since 'a' is a nonzero element ofF, inverse of a, that is, a-
I exists. Now 0 = a-\O) = a-I(ab) = (a-1a)b = 1· b = b, a contradiction since the element b is a
nonzero element ofF. Hence ab cannot be equal to zero for any two nonzero elements a, b in F.
This shows that F is an integral domain. Thus we can conclude that every field is an integral
domain.

9.2.13: (i) The set of integers forms an integral domain which is not a field.

(ii) The set of real quaternions forms an integral domain which is not a field.

9.3.2: Consider (a+b) (c+d) = a(c+d)+b(c+d) = ac + ad + be + bd.

9.3.3: Consider (a + bi = (a + b) (a + b) = a(a + b) + b(a + b) = a2 + ab + ba + b2

1A1gebra 9.16 Definitions, Examples and .... 1

9.8 MODEL QUESTIONS:

1. Define ring, commutative ring. Give two examples of each. Give an example of a ring which
is not commutative.

2. What do you mean by ring of real quaternions, explain in detail.

3. Prove that every field is an integral domain

4. IfR is a ring and a, b E R, then show that (-a)(-b) = ab.

5. Show that every finite integral domain is a filed.

6. Show that the characteristic of an integral domain is a prime number if D is of finite

characteristic.

9.9 REFERENCE BOOKS·

1. Gopalakrishnan N.S. 'University Algebra', Wiley Eastern Limited, New Delhi,

1995.
2. Herstein LN. 'Topics in Algebra', Wiley Eastern Limited, New Delhi, 1988.

Dr. Bhavanari Satyanarayana

~r
I'
LESSON 10
HOMOMORPHISMS AND IDEALS

Objectives
The objectives of this lesson are to:

• Introduce the concepts like homomorphism, kernel, isomorphism

• Discuss some examples for homomorphism and kernel
• Develop the skills to verify whether a given mapping is a homomorphism
• Introduce the concepts like ideal, quotient ring
• Discuss some examples of ideals
• Know how to construct a quotient ring RlI for a given ideal I.

Structure
10.0 Introduction
10.1 Ring Homomorphisms
10.2 Kernel of a Homomorphism
10.3 Ideals
10.4 Quotient Rings
10.5 Summary
10.6 Technical Terms
10.7 Answers to Self Assessment Questions
10.8 Model Questions
10.9 Reference Books

10.0 INTRODUCTION
In the beginning of this lesson, we introduce the concepts: homomorphism, isomorphism
and kernel. The notion of homomorphism is one of the central ideas that are common to all aspects
of modern algebra. By this, one means a mapping from one algebraic system to a like algebraic
system which preserves the structure. Recall that, for groups (with operation multiplication) a
homomorphism was defined as a mapping such that <pCab)= <p(a)·<PCb).Since, a ring has two
operations, we add the condition: <p(a+ b) = <»(a)+ <»(b)to it. So in case of rings, we stipulate that
the isomorphism preserves both the additive and multiplicative structures. We studyproperties of
homomorphism, kernel and isomorphism and give some examples for these concepts. '>
\

Next we carryover the concept of normal subgroup to rings. We define ideal of a given
ring R, and construct the quotient ring of R, with respect to a given ideal. in a natural way. As we
know already, the normal subgroups play a vital role in the theory of groups. Like wise, ideals
play an ana' .igous role in the study of rings. The real significance of ideals in a ring is that they
enable us to construct other rings that are associated with the given ring in a natural way.
!Algebra 10.2 Homomorphisms and Idealsl

10.1 RING HOMOMORPHISJ\tIS

10.1.1 Definition: Let (R, +, .), and (RI, +, .) be rings. A mapping <I>: R -) RI is said to be a
homomorphism (or a ring-homomorphism) if
(i) <1>(a+ b) = <1>Ca)+ <1>Cb);and
(ii) <I>(ab) = <I>(a)<I>(b)for all a, b E R.

10.1.2 Lemma: If <1>:R-)R1 is a homomorphism, then

(i) <1>(0)= 0; and
(ii) <1>(
-a) = -<1>(a)for all a E R.

Proof: (i) It is clear that 0 + <1>(0)= <1>(0)= <l>CO

+ 0) = <1>(0)+ <1>(0)~ <1>(0) = O.
(ii) 0 = <1>(0)= <I>(a+ (-a» = <l>Ca)+ <l>C-a)==} <I>(-a) = -<I>Ca).
10.1.3 Note: Let <1>:R -) Rl be a nonzero homomorphism.
(i) If R is an integral domain, then <1>(
1) = 1.
(ii) If R, RI are rings with identity, and <1>is onto, then <1>(1)= 1.

[Verification; (i) Since <I>is non-zero, we have <I>(R):f:. O.

This implies there exits x E R such that <1>(x)"# O.

Now l.<I>(x) = <1>(x)= <1>(1.x) = <1>(1).<1>(x)==} [l-<1>(1)]<1>(x)= 0

=
==} 1- <1>(1) 0 (since <1>(x)"#0 and RI is an integral domain) ~<1>(1) = 1.

(ii) First we show that <I>(1}is the identity in RI. For this, let y E RI. Since <1>is onto, there exists

an element x E R such that <I>(x)= y.

Now <1>(1).y = y = <I>(1).<I>(x)= <I>(l.x) = <1>(x)= y.
Therefore <1>(1).y = y for all y E R I. Similarly Y.<1>(1)= y, for all y E RI.

This shows that <1>(1)is the identity in R 1. Since identity is unique, we have <1>(1)= 1].

10.1.4 Note: Let Rand R 1 be rings. If we define <1>:R -) R I by <1>(x)= 0 for all x E R, then <1>is a
homomorphism. This homomorphism <1>is called as
zero-homomorphism.
jAcharya Nagarjuna University 10.3 Centre for Distance Educationl

10.1.5 Self Assessment Question : Consider Z, the ring of integers, and the ring

J(.J2)={m+n.J2 / m,nE Z}. Define<p:J(.J2)----1J(.J2)by

<l> (rn + n.J2) = m - n..fi. Show that <j)is an isomorphism.

10.2 KERNAL OF A HOMOMORPHISM

10.2.1 Definition: (i) Let <j): R ----1R 1 be a ring homomorphism. Then the set

{x E R / <P(x) = O} is called the Kernal of <pand is denoted by kero or l( <P).

(ii) A homomorphism <1>:R ----1RI is said to be an isomorphism if <pis 1-1, and unto.
(iii) Two rings Rand R i are said to be isomorphic, if there exist an isomorphism

10.2.2 Lemma: Let \~: R ----1RI be a homomorphism. Then

(i) Kero = {O} ¢=> <j) is i-L

(ii) If <j) is onto, then ¢ is an isomorphism <=> Ken» = {O}.

Proof: (i) Suppose Kero = {O}. To see <pis 1-1, suppose x, y E R such that <j)(x) = <P(y) =>

!j)(x) - <j>(y) =0 => <j>(x) + <p(-y) = 0 => <1>(x- y) =0

=> x - Y E Kero = {O} => x- y = 0 => x = y. Therefore <pis 1-1.
Converse: Suppose <pis 1-1. Since <p(0) = 0, we have 0 E Kero => {O} c Kero.

Now let y E Kero => ¢(y) = 0 = ¢(O) (we know that <P(O) = 0)
I
=> y = 0 (since ¢ is ]-1) => Kero c {O}. Therefore Kero = {OJ.

(ii) Suppose <!> is an isomorphism => <I>is a bijection is 1-'1 => Kero
=> </> = {O} (by (ij).

Converse: Suppose Kero = {OJ => <j)is 1-1 (by (i).

Since <l> is onto, we have that <l> is a bijection. Hence <pis an isomorphism.

10.2.3 Result: If R is a ring with identity and <p:R ----1R 1 is a ring epimorphism, then show that

<1>(1) is the identity in RI.

Proof: Let y E RI.

We have to show that $(1 ).y y y. <P(1). = =
Since </>is onto, there exists an element x E R such that <P(x) = y.
Consider </>(l).y = </>(l).</>(x)
= </>(l.x) </>(x)= y. =
Also Y.<I>(l) = <P(x).¢(l) =
</>(x.l) = </>(x)= y. This is true for all y in RI.
Therefore <p(l) is the identity in R 1.
I .
Algebra 10.4 Homomorphisms and Idealsl

10.2.4 Self Assessment Question : Let 1 be the ring of integers, ln, the ring of integers modulo
- -
n. Define <j>: 1 ~ J, by <j>(a) = a, where a is the remainder of a division by a. Then <j>is
an onto homomorphism with Kernel n1 (the set of all multiples of n).

10.3 IDEALS
10.3.1 Definition: Let R be a ring and I be a subgroup of (R, -r). Then I is said to be
(i) a left ideal of R if ra E I for every r E R, a E 1.

(ii) a right ideal of R if ar E I for every r E R, a E I.

(iii) an ideal (or two sided ideal) of R if I is both left ideal and right ideal.

10.3.2 Lemma: If <1>:

R ~ R I is a ring homomorphism, then Kero is an ideal of R.

Proof: To show that Kero is an ideal of R, we have to show

(i) Kere is a subgroup of (R, +)

(ii) a E Kero, r E R:::::> ar, ra E Kero.

To prove (i), let us take a, b E Kero ee <1>(a)= 0 = <1>Cb).

Now<1>(a-b)=<1>(a)-<1>(b)=O-O=O :::::> a-bE Ker<1>.

Hence Kero is a subgroup of CR, +).

To prove (ii), let us take a E Kero ee <j>(a)= O.

Now <1>(ar)= <j>(a).<1>Cr)
= O.<1>(r)
=0 :::::> ar E Kero.

Similarly, we can show that <1>(ra)= 0 :::::> ra E Kero.

Now we can conclude that Kero is an ideal of R.

10.3.3 Result: If U is an ideal of R and lEU, then prove that R = U.

Proof: Since U is an ideal of R, we have that U ~ R.

To show that R ~ U, let us take r E R.

Since r E R, lEU and U is an ideal of R, we have that r = lre U

Therefore R ~ U. Hence we get that R = U.

10.3.4 Result: If F is a field, then only ideals of Fare (0) and F itself.
Proof: We know that (0) and F are ideals of F.
Let I be an ideal of F such that I * (0).
!Acharya Nagarjuna University 10.5 Centre for Distance Educationl

It is enough to shbw that I = F. Let 0 i:- x E 1.

Since x E F - {O}and F is a field, the multiplicative inverse x -lof x exists in F.

So x -I E F. Now since x E I, X -I E F, I is an ideal of F, we can conclude that
1 = X-IX E I. Therefore 1 E I.

Now by the above Result 10.3.3, we get that I = F. This completes the proof.

-
10.3.5 Result: Prove that any homomorphism of a field is either an isomorphism or takes each
element into O.
Proof: Let FI, F2 be two fields and f: FJ --7 F2 be a homomorphism, If f(x) = 0 for all x E Fl, then

f = O. Now suppose f i:- o.

Then there exists a E F, such that f(a) i:- o.
We know that Ker f is an ideal of Fl.
Now by above problem either Ker f = 0 or Ker f = Fl.

If Ker f = FJ then f(x) = 0 for all x E FJ, a contradiction. Therefore Ker f = O.

This means f is one - one (by Lemma 10.2.2).
Now f(F]) is a sub field of F2. Also f: FJ --7 f(F,) is onto.
Hence f: FJ -7 f(FJ) is an isomorphism of rings.

10.3.6 Problem: If R is a ring and a E R, then show that

(i) aR = far / r E R} is a right ideal,
(ii) Ra = {ra / r E R} is a left ideal,

(iii) moreover, if R is a commutative ring, then Ra = aR is an ideal of R.

Solution: (i) Let x,y E aR::::} x = ar, y = as for some r.s E R.

Now x - Y = ar - as = a(r - s) E aR (since r - s E R).

Therefore aR is a subgroup of (R, + ).
Let x EaR, t E R => x = arforsorne r E Rand t E R.

Now xt = (ar)t = a(rt) E aR (since rt E R).

Hence aR is a right ideal. -.
 jAlgebra 10.6 Homomorphisms and rcteaIS)
. (ii). Using similar steps of the proof of (i), we can get that Ra is a left ideal.

(iii). Suppose R is a commutative ring.

Now x E Ra:=} x = ra for some r E R.

Since R is commutative, we have that x = ra = ar EaR.

Therefore Ra c aR. Similarly aR c Ra.
Hence Ra = aR. Now from (i), (ii), we can conclude that Ra = aR is UP ideal.

10.3.7 Problem: If U, V are ideals of R, then the sum of two ideals U and V is defined by U+ V
= {a + b / a E U, b E V}. Show that U + V is an ideal of R

Solution: Let x, y E U + V :=}x == a + b, y = c.+ d for some a, c E U and b, d E V. Now x;-y = (a

+ b) - (c + a) = (a - c) + (b - d) E U + V(since U, V are subgroups we have that a, c E U => a - c

E U; and b, d E V=> b - d e V).

Therefore U + V is a subgroup of (R. +).
Let r e R.Now xr=(a+b)r=ar+'brE U+V.
Therefore U + V is a right ideal.
Similarly, we can prove that U + V is a left ideal. Hence U + Vis an ideal of R.

10,3.8 Problem: If A. B are any right (left, two sided) ideals of R, then show that An B is

also a right (left, two si?ed) idear8~R, respectively .

. Solution: Given that A, B are two right (left, two sided) ideals of R.

Now we have to show that A n B is also a right (left, two sided) ideal of R.
Let a, b, c E A (] Band r be any element of R.
Then a, b, c E A and a, b, c E B.
But A and B are both right (left, two sided) ideals. of R.
Hence a- b, cr E A (r c E A, respectively), a -e- b.cr E B (rc E B, respectively)

=> a - b, cr E A (] B (r c E A (] B, respectively).

Hence A (] B is a right (left. two sided) ideal of R. .

IAcharya Nagarjuna University 10.7 Centre for Distance Educatio~

10.3.9 Note: If A is a right ideal and B is a left ideal of R then A n B cannot be a left or right
ideal of R. we provide an illustration in the following:

Illustration: Let R I = M2(R), the ring of 2X2 matrices over the ring of real numbers.

Write A ~ {[~ ~]I a, bE R} and B ~ {[: ~]/c, d e R} It is a routine verification to

observe that A is a right ideal of R 1 ; and B is a left ideal of R 1, but their intersection is

A B ~ {[~ ~] I
CO aE R}
1 0
l
Now it is clear that r3 4] x· [a
0 0
0] = [3a
a o0] ~ A' n B; and .
a
[o 0
0]
x [3 4]
1 0
= [3a
0
4a]
0
~ AnB.

This shows that An B isneither a left ideal nor a right ideal of R I.

10.3.10 Example: Let Z be ring of integers, A = 2Z, even intezers, B = 5Z, the set of multiples of
5. Then A and B are two ideals of Z.

4, 5 E A u B, but 5 - 4 = 1 ~ Au B. This shows that Au~ is not a subgroup. Hence Au B is

not an ideal of Z.
Thus in general if A and B are right (or left or two sided, respectively) ideal of a ring R, then Au
B may not be right (or left or two sided, respectively) ideal of R.
,;'1'
10.3.11 Theorem: Let A and B be left (respectively, right, t'YPt~i~ed) ideals of a ring R. Then A u
B is a left (respectively, right, two sided) ideal of R <=> A c B or B c A.

Proof: If A k: B or B k: A, then A u B is either equal to. B or equal to A.

So in this case, A u B is an ideal of same type as A and Bare.
To show the converse, suppose that A u B is an ideal of R and A ct. B.

We have to prove that B c A. For this, take an element b e B.

Since Act. B, there is an element a E A such that a ~ R
Now a, b E Au B. Since A u B is an ideal. we have that a - b e Au B.-'
a - b E A u B ~ a - b E A or a - b E B.
 ~gebra 10.8 Homomorphisms and Idealsl

If a - b e B, then (a - b) + b E B (since B is closed under addition)

=} a E B, a contradiction to the choice of a.

Therefore a - b E A. Now a E A, a - b E A =} a - (a - b) = b E A.
Now we proved that every element of B is an element of A. Therefore B C A.
This completes the proof.

10.3.12 Note: We define the sum of left (respectively right, two sided) ideals A and B of a ring R
as follows:
A+B = {a + b / a E A and b E B}.
As in Problem 10.3.7, we can verify the following:
. A, B are left (respectively, right, two sided) ideals of R =} A+B is a left (respectively, right, two
sided) ideal of R.

10.3.13 Problem: If A and B are ideals (of any but same kind) in a ring R, then show that both A
and B are contained in A + B.

Solution: For each a E A, we have that a = a + 0 E' A + B.

This shows that A C A + B.
Similarly, for each b E B, b =0 +b E A + B, and so B ~ A + B.

10.3.14 Problem: If U, V are ideals of R, then the product of ideals is defined as follows:
n

UV = {~>i b, / n is a positi·ve integer, a, E U, and b, E V for 1:$i :$ n} .

;=1
n m
Solution: Let x,y E UV =} x = L a, b, and y = L c, d, for some positive integers n, m and a., c,
;=1 i=1

E U, b, .d, E V for each admissible i.

n m n m
Now x v y « 2:aibi - 22cidi= Laibi + L-CidiE UV.
>; ;=1 i=i ;=1

Hence UV is a subgroup of (R, +).

--
~-

\Acharya Nagarjuna University, 10.9 Centre for Distance Edur''l:iliiril

n n
For any element r E R, we have that xr = (L aj b.) r = L a, (b, r) E UV (sjnce b.r E Vas V is
i=1 ,i=1

an ideal of R). Hence UV is a right ideal.

, Similarly, we can prove that UV is a left ideal. Thus UV is an ideal of R.

10.3.15 Problem: If U and V are as in above problem, then prove that UV c U n V.

n

Solution: Let x E UV ~ X = L a, b., for some positive integer nand a, E U, b, E V for each
;=1

. 1, _.1_
1, <. < n.

Since a, E U and b, E V c R, we have that aj b, E U (by using the right ideal property of U) for
all i.
n
Since U is a subgroup of (R, +), it follows that x = L a, b, E, U.
;=1

Since b, E V and a, E U c R, we have aj b, E V (by using left ideal property of V).

n

Since V is a subgroup of (R, +), we have that x = L a, b, E V.

;=1

Hence x E Un V. This shows that UV C U n V.

10.3.16 Problem: If I is an ideal of a ring R, then show that the set

rei) = {x E R / xa = 0 for all a E I} is an ideal of R. This r(I) is,called the
left annihilator of 1.

Solution: Let x, y E rei) => xa = 0, ya = 0 for all a E I.

Now (x ., y) a = xa - ya = 0 =>x - Y E r(l). Hence r(l) is a subgroup of (R, +).
Let s E R. To show that rei) is an ideal, it is enough to show xs, sx E rei).
Now (sx)a = s(xa) = s(O) = 0 for all a E I~ SX E rei).
Now (xs)a = x(sa) = 0 (since sa E I for all a E I, s E R) => xs E rt l).
This shows that rei) is an ideal of R.

.\
/Algebra 10.10 Homomorphisms and Idealsl
o
10.3.17 Problem: If 1 is an ideal of a ring R, then show that the set
[R : IJ ={ x E R / rx E I for all r E R} is an ideal of R. Also I c [R : IJ.

Solution: Let x, y E [R : IJ => rx, ry E l for all r E R => rex - y) = rx - ry E I for all

r E R => X - Y E [R: I]. Thus [R: I] is a subgroup of (R, +).

First we verify that [R:I] is a left ideal. 0

Let s E R. Consider sx. Now r(sx) = (rs)x E I (since rs E R and x E [R : I]). This is true for all
r E R. SO SX E [R: I]. Therefore [R : I] is a left ideal.

Next we verify that [R : I] is a right ideal. Let s E R.

Consider xs. Now r(xs) = (rx)s E I

(because x E [R: I] => rx E I. Now rx E I, s E R, I is an ideal of R we have (rx)s E I).
Therefore r(xs) E I for all r E R => xs E [R: IJ. Therefore [R: I] is a right ideal ofR.

Finally, we have to verify that I C [R : IJ.

Let x E I. Since I is an ideal, rx E I for all r E R => X E [R: I]. Hence I C [R :1].

10.3.18 Problem: If R is a ring, a E R, .show that the set rea) = {x E R / ax = O}is a right ideal
of R. This rea) is called the right annihilator of a.

Solution: Let x, y E rea) => ax = 0, ay = 0 => a(x - y) = ax - ay = 0 => x - Y E rea). Therefore rea)
is a subgroup of (R, +).
Let x E rea), s E R. Consider a(xs,p,(ax)s = O.S = 0 => xs E rea).
Hence rea) is a right ideal.

10.3.19 Problem: Let R bea ring with 1, R not necessarily commutative, such that the only right
ideals (similarly left ideals) of Rare (0) and R itself. Prove that R is a division ring.

Solution: Suppose R is a ring with 1.

To show that R is a division nng , it is enough to show that every non-zero element has
multiplicative inverse.
Let 0 '* a E R. Since aR i right ideal, 0 '* a = a.1 E aR we have (0) '* aR and
jAcharya Nagarjuna University 10.11 Centr~ for Distance Educatio@
/'

so aR = R. Now IE R = aR ~ there exists b E R such that ab = 1.

Clearly b ;to (If b--'=0 then 1 = ab = aO = 0, a contradiction).

NowbR is a non-zero right ideal and so bR = R ~ 1 E R = bR~ there exists a' E R such that
1 = bal.
Now a = a.l = arba') = (abja' = 1: al = a'.
Therefore ab = 1 = ba. This shows that b is the inverse of a.
Hence every non-zero element has its multiplicative inverse.
So we can conclude that R is a division ring.

10.3.20 Self Assessment Question: Let R be a commutative ring with unit element whose only
ideals are (0) and R itself. Then R is a field.

10.3.21 Problem: Consider R = Z, the ring of integers with respect to usual operations + and.
(0 I is a subgroup of (Z, +) ¢ 1= nZ = {nx / x E Z} for some integer n E Z.
(ii) I is an ideal of the ring (Z, +, .) ¢ 1= nZ = {nx / x E Z} ifor some n E Z.

Solution: (i) Sup~9'8e I is a subgroup of (Z, +). If I = 0 then I = O.Z.

Suppose I :F- O. Let n be the least positive integer such that n E I.
\
\

Let rh E {nx / x E2}

-, = nZ ~ m = nx for some x E Z.
Now = nx = (n + n'4, ... + n) E I (since n E I and I is a subgroup).
re nZ C I. To show that I C nZ, take k E I.

Then b \division algorithm there exists integers a, b such that k = an + band 0 ~ b < n.
Now if b :F- 0, then b = k - an E I (Since k E I and anE I, we have b e I) and b < n, a contradiction
to the selection of n.
\

Therefore b = 0 and hence k:: an + 0 = an E {nx / XE Z}.

Hence I = {nx / x E Z}.

Converse: We can easily show that, for any n E Z, the set {nx / x E Z} is a subgroup of Z. This
completes the proof of (i).
iAlgebra 10.12 Homomorphisms and Idealsl

(ii) Suppose I is an ideal of (Z, +, .).

Then I is a subgroup of (Z, +). By (i), we have that I = nZ for some n E Z.

Converse: Consider nZ for, some n E Z. By(i), we have that nZ is a subgroup of (Z,+)".

To show that nZ is an ideal, let r E R =Z and me nZ.
Then m = nx for some x E Z. It is clear that rm = r (nx) = n(rx) E n2 which implies

mr= rm E nZ.
This shows that nZ is an ideal of Z = R.
10.3.22 Self Assessment Question: Find all the ideals of the ring of-integers (Z, +, . .)

10.3.23 Self Assessment Question: The ring R of 2 x 2 matrices with real coefficientsis not a
skew field as it has zero devisors. Show that R has no proper ideals.

10.4 QUOTIENT RINGS

lil.4.1 Lemma: Let I be an ideal of a ring R. If we consider (1,+) as, a subgroup of (R,+), and
(R/I, +) the quotient additive group of R with respect to I. Define multiplication on R/I as follows:
(a+Iub+I) = ab + I.
Then we have the following:
(i) RlI is a ring;
..). ~ "

(ii) If R is a commutative ring, then RJI is also a commutative ring; and

. • I,: ~~ •

(iii) RlI is a homomorphic image of R (that is there exists an onto homomorphism <1>:
R -7 R/I
.f I t : I . : •• !,

defined by <I>(a)= a + I) , ..L ,

• 1 •.',

Proof: For the convenience of the ~~~der, we provide detailed proof for this lemma.
, !- ">: '.'.J

(i) Define a relation - on R as follows:

a - b <=> a - b e I for a, b E R.
"' t..•.

Since a - a = 0 E I =::} a -a. So the relation - is a reflexive relation.

Suppose a - b =::} a - bEl =::} b- a E I=::} b - a. So the relation - isa symmetric relation.

Suppose a -,.b?;b - c ==>a - bEl, .'

b~cE I =::} a - c,= (a - b) + (b - c)~. I==>a ..-
• < <.< t <
c. Therefor
.-,i};.
the
relation - is a transitive relation. , £~ • ...,!3

Since the relation - is reflexive,

•
symmetric and transitive, it follows that the .relation
.>. • .....,
- is an

equivalence relation on R.
 jAcharya Nagarjuna University . 10.13 Centre for Distance Educatio~

We know that a + I = {a + x / x E I} is the equivalence class containing a E R. This is true for

all a E R.

We write R/I = {a + I / a E R}, the set of all equi valence classes under the relation. -.
We define + and. on RtI as (a + I) + (b + I) = (a + b) + land (a + I).(b + I) = (a.b) + I.
By the discussion in group theory, it is clear that (RtI, +) is an abelian group as (R, +) is an abelian
group. As in group theory, it is easy to verify that 0 + I is the additive identity and (-a) + I is the
additive inverse of a + 1 in the quotient ring Rzl..
In a similar way, we 'can verify that RtI is a semi group with respect to multiplication and
distributive laws hold good. Hence we may conclude that (R/I, +, .) is a ring.

(ii) If R is a commutative ring, then

(a + I).(b + I) = (a.b) + I = (b.a) 4- I =(b + I).(a + I) for any two elements (a + I), (b + I) E RlI. This
shows that RII is a commutative-ring.

R ~ RlI as </>(a)= a + I for all a E R.

(iii) Defined </>:
Now we verify that </>
is a homomorphism.
</>(a+ b) = (a + b) + I = (a + I) + (b + I) = </>(a)+ </>(b);and
</>(ab)= ab + I = (a + I) (b + I) :;;~(a)<p(b)~.

This shows that <pis a homomorphism,

Now for any x + I E RlI, we have that x E Rand </>(x)= x + I.
~ •I ~ " l. ~ ". t

. Therefore </>
is onto. Hence </>:
R ~ RlI is an onto homomorphism.

J0.4.2 Self Assessment Question: If lUs a ring with unity, then show that RlI is also a ring with
. )

unity.
I ~ _.~ •

,;10.4.3 Definition: (i) Consider th~ mapping </>:R ~ WI defined as above. This' </>is called as the
canonical epimorphism from Ranta R/l. '
s . (ii) The ring (RII, +, .) is called the Quotient ring of the ring R by the given ideal I.
IAlgebra 10.14 Homomorphisms and Idealsl

10.4.4 Note: Consider </>:

R -7 R/I as in lemma 10.4.1
Then x E Kero ¢:::> </>(x)= 0 = 0 + I ¢:::> x + I = 0 + I
¢:::> [x] = [0] ¢:::> x '- 0 ¢:::> x - 0 E I ¢:::> X E I.

Therefore Kero = I.

10.4.5 Lemma: Let f: R ---7RI be a ring onto homomorphism.

(i) If 1 is an ideal of R, then f(J) = {f(x) / x E 1} is an ideal of R I;

(ii) If J1, h are ideals of R such that f(1l) = {f(x) / x E Jd and f(Jz) = {f(x) / x E Jz} are equal,

and Ker f ~ h, then 1I c Jz; and

(iii) If WI is an ideal of R I, then CI(WI) = {XER / f(x) E w'j is an ideal of R ..

Proof: (i) a', b' E fO) ==> there exits a, b e J such that f(a) = a', f(b) = bl
=:>al - bl = f(a) - f(b) = f(a - b) E f(1).
Let r' E RI.

Since f: R---7 R I = Im f (as f is an epimorphism), there exits an element r E R such that fer) ~ r'.
Now rial = f(r)f(a) = f(ra) E f(l), and a'r ' = f(a)f(r) = fear) E fO).
Therefore fO) is an ideal of R I.

(ii) Suppose 11 and 12 are ideals of R such that fOI) ~ f(h) and Ker f c 1z.
Now x Ell ==> f(x) E fO I) ~ f(2) = fey) for some y E 12
==> f(x)

==> f(x) - fey) = 0 ==> f(x"~ y) = 0 ==> x - Y E Ker f c J2 ==> X - Y E J2.

·1:
Now X. - Y E 12and y E 12 ==> X = (x ~ y) + Y E 1z ==> x E h Therefore 1I' C Ja.

(iii) Write W = {x E R / f(x) E Wi}

Now we verify that W is an ideal of R. Let a,b E Wand r E R.

f(a - b)= f(a) - f(b) E Wi ==> a - b E W.
Also f(ra) = f(r).f(a) E Wi (since f(a) E WI and fer) E rl and Wi is an ideal of RI) =:>ra E W.

Similarly, fear) = f(a).f(r) E WI =:>ar E W. Hence W is an ideal of R.

jAcharya Nagarjuna University 10.15 Centre for Distance Educatio~

10.5 SUMMARY

A mapping <1>from a ring R into a ring R 1 is said to be a ring homomorphism it; it satisfies
the following two conditions: (i). <1>(a+ b) = <1>(a)+ <1>(b);and. (ii). <1>(ab)= <1>(a)<1>(b),for
: R -+R 1 is a homomorphism,
all a, b E R. If <1> then <1>(0)= 0 and <1>(-a) = - <1>(a),for all a E R,

where '0' is the additive identity in R. The set {a E R I <1>(

a) = O}is called the kernel of <1>and it is

denoted by 1(<1»or Ker <1>.<1>:R ---+ R 1 is said to be an isomorphism if it is a bijection and a

homomorphism. A homomorphism R ---+ R 1 is one - one
<1>: ¢:::> = (0).
Ker <1>

,
A non - empty set I ofa ring R is said to be an ideal of R if it satisfies the following two

conditions: (i). I is a subgroup of (R, +); and (ii). ar, ra E I for all a E I and r E R. For any
ring homomorphism <p,the set Kero is an ideal of the domain of <p. If I is an ideal of R, then the
set R/I = {r + II r E R} forms a ring (called the quotient ring modulo the given ideal I). It is a
homomorphic image of R, where the homomorphism <p of R onto RlI is given by <1>(a)= a + I for

every a E R. whose kernel is exactly I.

10.6 TECHNICAL TERMS

Homomorphism: Let (R, +, .), (R 1, +, .) be two rings. A mapping

<»: R ---+ R 1 is said to bea homomorphism (or a

ring-homomorphism) if
, - .'
(i) <pCa+ b) = <p(a). + <p(b); and //
/
/
(ii) <p(ab) = <pCa)<»(b) for all a, b
/ "

E/R.

Kernel: Let <»:R ---+ R 1 be a homomorphism. Then the

set {x E R I <»(x) = O} is called the Kernal of <p

/
,
and is denoted by kero or I(<»). I ;

Isomorphism: A homomorphism <1>:

R ---+ Rl is said to be an

isomorphism if <I>is 1-1, onto.

/Algebra 10.16 Homomorphisins and Ideal~':'

Left ideal: Let R be a ring and I be a subgroup of (R, +). Then lis said
to be a left ideal of R if ra E I for everyr E R, a E I.

Right ideal: Let R be a ring and I be a subgroup of-(R, +). Then I is said
to be a right ideal of R if ar E I for every r E R, a E I.

Ideal (or two sided ideal): Let R be a ring and I be a subgroup of (R, +). Then I is said
to be an ideal (or two sided ideal) ofR if! is both left ideal'
and right ideal.

10.8 ANSWERS TO SELF ASSESSMENT QUESTIONS

10.1.5: Let m + nJ2, a + bJ2 E J( J2)

<j>«m+ nJ2) + (a + bJ2)) = <j>«rn+ a) + (n + b) J2)'
= (rn + a) - (n + b) .J2 (by definition of <j>
)

=rn+a-n.J2 -b.J2 =(m-n.J2)+(a-b.J2)

+ n J2
= <j>(rn ) + <j>(a+ b J2 ) (by definition of <j»

This proves that <j>(x+ y) = <j>(x)+ <j>(y),for all x, y E 'I( J2)

Now we verify that <j>

«m +.~.
n.J2 )(a + bJ2») = <j>(m+ nJ2) . <j>(a
+bJ2)
-A.Ll I

L.H.S = <j>
(un + n.J2 )(a + b.J2» .; I)~'

= <j>(rna+ rnbJ2 + naJ2 + nbJ2 J2)

= <j>(rna+2 nb + (fib + na) J2)
= (rna +2 nb) - (rnb + na) J2
R.H.S = <j>(m+ nJi) <p(a+b Ji )
= (rn - nJ2 )(a -bJ2)

= rna - mbJ2 - naJ2 - n.fi (-b.fi)

= rna - (mb + na) J2 + 2nb

=(rna+2nb)-(rnb+na) -Ii,
(Acharya Nagarjuna University 10.17 Centre for Distance Educatio~

So L.H.S = R.H.S
This proves that $ is a homomorphism.

Suppose </>(m+ n.J2) =0~ m = 0 and n = 0 ~ m + n.J2 = O. So </>is an one - one mapping.
- n.J2 ) = m + n.J2 , so <I>
For in + n.J2 E J(.J2), the element m - n.J2 ~ J(.J2) and <I>(m is onto.

Hence </>
is an isomorphism.

- -
10.2.4: Let a, b E 1. Then </>(a+b) = a +b = a + b (by the oefinition of addition in the ring of

integers modulo n) = <I>(a)+ <I>(b)

<I>(a·b)= ab = a b (by the definition of product in the ring of integers modulo n)

, = <I>(a)·</>Cb)
This shows that <I>
is a homomorphism.

For any a E =
In with 0 ~ a < n, we have that a E J and <I>(a) a.
This shows that <I>
is onto.

XE Kero <=> </>

(x) = 0 <=> x = 0 <=> x is a multiple of n <=> x E nJ.

Thus Ker</>= nJ.

Hence <1>:
J---7 Jn is an onto homomorphism such that Kero = nJ.

10.3.20 : Since the only ideals of Rare (0) and R itself, by the problem 10.3.19,it follows that R
is a division ring. Given that R is a commutative ring. So we "have that R is a commutative
-.,\

division ring. Hence R is a field. . \"'

10.3.22: By using problem 10.3.21, we conclude that the set of all ideals of Z is given by { nZ / n
is a positive integer}."

10.3.23: Let I be a non - zero ideal in R and choose a non-zero element A E I and-let

A = [: : ]. Then at least one of a, b, c, d is non-zero, say a" O. Consider the elements

EJ = [1 0], E2 = [0 1], E3 = [0 0], and E4 =[0 0] of R. -

00 00 10 01
 , /

IAlgebra 10.18 Homomorphisms and Idealsl

Now EIAEI = [a 0] E I as A E I. Hence EI =[1

o 0 . 0
Therefore EI E I. In a similar way, we can show that E2, E3 and E4 all belong to I. Since any
element

B = [:: :: ] of R can be written as a linear combination

B = aIEl + blE2 + clE3 + dlE4, we have BEL This proves that 1= R.

10.4.2: Let 1 E R be the unit element in R. Consider 1 + I in R/I.

(1 + I)·(a + I) = I-a + I = a + I. This is true for all a+I E R/I. Therefore 1+1 is the unity
element in R/I. Hence R/I is a ring withunity. \

10.8 MODEL QUESTIONS

1. If <\>: R -7 R I is a ring homomorphism, then show that Kero is a subgroup of (R, +).

2. If <\>: R -7 R I is a homomorphism then Kero =0 ¢::} <!> is a one one mapping.

3. Show that intersection of two ideals is an ideal

4. Show that if A and B are ideals in R, A u B is not, in general, an ideal in R. Also show that
A u B is an ideal ¢::} either A c B or B c A.

5. Find all the ideals of the ring Z of integers.

10.9 REFERENCE BOOKS /

1. Go;zaia ishnan N.S. 'University Algebra', Wiley Eastern Limited, New Delhi,
1995.
Hers in LN. 'Topics in Algebra', Wiley Eastern Limited, New Delhi, 1988.

Dr. Bhavanari Satyanarayana

 LESSON 11
MORE ON IDEALS & QUOTIENT RINGS

Objectives:
The objectives of this lesson are to:
• Prove some homomorphism theorems on rings
• Define maximal ideal of R
• Give some examples of maximal ideals in some concrete rings
• Find the conditions for which the homomorphic image of a ring becomes a field
• Get a necessary and sufficient condition for R/I to be a field
• Define embedding over ring, and field of quotients of an integral domain
• Define equivalence relation and equivalence classes
• Prove a theorem namely 'Every integral domain can be imbedded in a field'

Structure
11.0 Introduction
11.1 Homomorphism Theorems
11.2 Maximal Ideals
11.3 Prime Ideals
11.4 Imbeddings
11.5 Summary
. 11.6 Technical Terms
11.7 Answers to Self Assessment Questions
11.8 Model Questions ,.
11.9 Reference Books

11.0 INTRODUCTION
We continue the discussion of ideals and quotient rings of the previous lesson. First we
provide some homomorphism theorems on rings. We have seen that some properties of a ring R
are carried over to the quotient ring RlI, for instance the commutative property, and the existence
of unit element. Also there are some properties which are valid in a ring that are not valid in the
quotient ring. A ring may be an integral domain, but one of its quotient may not be an integral
domain. For instance, Z is an integral domain but Z4 = Z/4Z is not an integral domain.

In this lesson, we provide conditions on an ideal I under which R/I is (i). an integral
domain; (ii). a field. For commutative rings we give a complete answer in this lesson to the
question 'under what conditions is the homomorphic image of a ring a field'. For this we are to
introduce the concept of maximal ideal.
!Algebra 11.2 More on Ideals & Quotient Ringsl

In the previous lessons, we introduced ideals and quotient rings. We recollect that an
integral domain is a commutative ring D with no zero divisors. We can enlarge the ring of
integers to the set of rational numbers which is a field. Now we prove this for any integral
domain. The concept of embedding and extension can be introduced for rings, and we prove that
every integral domain can be imbedded in a field.

11.1 HOMOMORPHISM THEOREMS

11.1.1 Theorem: Let R, R 1 be two rings and f: R ---7 R 1 is an epimorphism such that

I = Ker f. Then

(i) RIKer f = RfI is isomorphic to R 1. (First Homomorphism theorem)

(ii) There exits a one-to-one correspondence between the set of ideals of R 1 and the set of ideals of
R contains I.(Correspondence Theorem)

(iii) This correspondence can be achieved by associating with an ideal WI of R I, the ideal W of R
defined by W = {x E R / f(x) E WI}. With W so defined, R/W is isomorphic to RIIWI. (Third

homomorphism Theorem)

Proof: We know that Im f = {f(x) / x E R} =R 1 (since f is an epimorphism).

Define g: RlI ---7 R 1 by g(x + I) = f(x). J

Let x, y E R such that x + I = Y + 1.

Now x + I = Y + I ¢::} x - Y E 1= Ker f ¢::} f(x - y) = 0 ¢::} f(x) - fey) = 0
¢::} f(x) = fey) ¢::} g(x + I) = g(y + I).
Hence g is well defined and 1-1.
Let x' E RI = Im f ~ there exits x E R such that f(x) = x'.
Now g(x + I) = f(x) = x 1. Hence g is onto.
This shows that g is a bijection.
Now we verify that g is a homomorphism.
g[(x + I) + (y + I)] = g[(x + y) + I] = f(x + y) = f(x) + fey) (since f is homomorphism)
= g(x + I) + g(y + I).
 IAcharya Nagarjuna University. 11.3 Centre for Distance Educationl

Also g[(x + I)(y + I)] = g[(xy + I)] = f(xy) = f(x)f(y) = g(x + I)g(y + I).
Therefore g is a homomorphism.
Hence g: RJI --7 R 1 defined by g(x + I) = f(x) is an isomorphism.

(ii) Let A = {J / J is an ideal of R and I C. J} and B = rw' / W i

is an ideal of R I}.

Define h: A --7 B by h(J) = f(J) = {f(x) / x E J}.

By Lemma 1O.4.5(of Lesson 10), h(J) is an ideal
I
of R 1 and hence h(J) E B.
If h(J 1) = h(J2), then f(J d = f(J2)'
Since Ker f = I c Jz. we get that J 1 c h (by Lemma 10.4.5 of Lesson 10).

Since Ker f = I <;; J1, we get J2 C. h Hence J1 = h

This shows that h is 1-1.
Now we show that h is an onto mapping. For this take WI E B.
Now W = {x E R / f(x) E w': is an ideal of R (by Lemma 10.4.5 of Lesson 10).
Clearly I = ker f = f -1(0) C f -1(Wl) =w and so W E A.

Also heW) = feW) = WI. Hence h is onto. Therefore h: A --7 B is a bijection.

(iii) Given that f: R--7Rl is an epimorphism.

Consider the natural epimorphism rt: RI--7Rl/Wl

Write <I> = nof.

Since nand fare epimorphisms, we have that the composition <I> = nof is also an epimorphism.
Now x E <=> <1>(x)=
Ker <1> 0 <=> (nof)(x)=O <=> n(f(x» = 0 <=> f(x) % w' = 0 + Wi
~ f(x) - 0 E Wi ~ f(x) E Wi <=> XE W.

. Hence Ker <1> == R1/WI.

= W. By (i), it follows that RJKer<1>

Since Ker <I> = W, we can conclude that RIW == R1/Wl.

11.1.2 Theorem: (Second homomorphism theorem) Let A be a subring of a ring R and I an

ideal of R. Then

(i) A+I = {a+i / aE A, ie I} is a subring of R;

(ii) I is an ideal of A+I; and

!Algebra 11.4 More on Ideals & Quotient Ring~

(iii) (A+I)/I == A/CAnI).

Proof: (i) Let Xl, X2E A + 1. Then x, = a, + i1and X2= a2 + i2 for some al, a2 E A and iI, h E 1.

Now XI- X2= (a, + il) - (a2 + b) = (a, - a2) + (it - b) E A + I and

XIX2= (a. + it) (a2 + i2) = ata2 + ali2 + i'la2 + i}i2E A + I (because a}a2 E A, ali2 , ila2 , ilh E I).

Therefore x, - X2,XIX2E A + I for all xi, X2E A + 1.

This shows that A + 1 is a subring of R.

(ii) Now we show that I is an ideal of A + I.

Take i E I and rEA + 1. Then r ~ a + it for some a E A and ilE I.

Since I is an ideal ofR and a E A ~ R, we have ir = i(a + il) = ia} ii, Eland
ri = (a + i})i = ai + i}i E L Hence I is an ideal of A + I.

(iii) Define f: A~ (A + 1)/1 by f(a) = a + 1.

(

Firstwe verify that f is a homomorphism.

Now f(a} + a2) = (a, + a2) + I = (a, + I) + (a2 + I) = f(a,) + f(a2) and
f(ala2) = ala2 + I = (at+ I)(a2 + I) = f(a})f(a2).
So f is a homomorphism.
Now we show that f is onto. Let r + I E (A + 1)/1.

Then r = a + i for some a E A and iE 1.

Since r + I = a + i + I = a + I, we have that f(a) = a + 1= r + I.
.. .< ~ t; .

This shows that f is onto. Weproved that f is an epimorphism.

Now we verify that ker f = A .~. 1. Let X E A ==domain of i
x E ker f ~ f(x) '~' 0 ~ X+ I=:O and X E A. ~ X+ I = 0 + l and x E A ~ X - 0 E I and X E A
. ,. ,
.

~xEAnI.
o ; 1 ;

Hence ker f= A n 1. Thus f: A~ (A + 1)/A is a ring epimorphism with ker f= A n 1.

We know that AI ker f == im f. Hence AI(A n I) =: (A + 1)/1.
This completes the proof.

11.1.3 Note: Let f: R ~ R* be a ring homomorphism. Write Rl = i~ f.

We know that Rl is a subringof R*. So Rl is a ring in its own right.
IAcharya Nagarjuna University 11.5 Centre for Distance Educationl

Clearly f: R ~ Im f = R 1 is the ring homomorphism.

Also it is onto. By theorem .11.1.1, we get that . RiKer f == R 1 = Im f.

So for any homomorphism f, we have that RIker f == Im f.

11.2 MAXIMAL IDEALS

11.2.1 Definition: An ideal I of a ring R is said to be a maximal ideal of R if it satisfies the
following two conditions:
(i) I is not equal to R; and
(ii) M is an ideal of R such that I c M k: R => either 1= M or M = R.
.:'0',

. _'t~·..... <, .

11.2.2 Problem: Let R be the 'set ofhll real valued continuous functions on the closed interval
[0,1]. Then
(i) R is a ring;

(ii) If <!>: = f(1/2), then <!>is a homomon-hism;

R ~ R defined by <!>(f)
(iii) Kero = {f E R / f( 1/2) = O}and this is a maximal ideal of R; and
(iv) l(x) = x for all x E [0,1], is identity in R.

Solution': (i) Define (f + g)(x) = f(x) + g(x) and (fg)(x) = f(x)g(x) for all x E [0, 1] and for all
f, g E R.
Then we can easily verify that (R, +, .) is a ring.
, !;)1 j; •~ '- >

If we define O(x) = 0 for all x E' [O,l}, then

. 0 is the additive
.
identity in ~, '~~ ,

The function '-f' defined by (-f)(x) ~ -(f(x)) is the additive inverse of ftx),
If f, g E R, then (f + g)(x) = f(x) + g(x)
.
= g(x) + f(x) =~.(g + f)(x). This implies that f + g = g + f.
Therefore (R, +) is an abelian group. The rest of the proof is a direct verification.

.
+ g)
(ii) Let f, g E R. Consider <I>(f
~.
= (f + g)(l/2) = f(1/2) + g(1/2) = <I>(t)+ <I>(g). .

Also <I>(fg)= (fg)(1/2) = f(1l2)g(1/2)= <I>(t)<I>(g).

Hence <I>
is a homomorphism.
(iii) f Eo Ker <I>
~<I>(t) = 0 ~ f(1/2) = O. Hence Ker <!>
= {fE R / f(1I2) =O}

Write M = { fER /f(1I2) =O}. "

 ff\lgebra= 11.6 More on Ideals & Qu~tient Ringsi

We know that kernal of a ring homomorphism is an ideal.

l"JOW we have to show that this M is a maximal ideal.

For this, consider an ideal U of R such that M C U c R with M -:j:. U.

Now it is enough to show that U = R.
First we verify that the mapping 1 defined by I(x) = 1 for all x E [0, 1] is the identity element in
. the ring R.

Clearly 1 E R. For any fER, (f. 1)(x) = f(x).l (x) = f(x).l = f(x).
Therefore f.I =f for all fin R. Hence 1 is the identity element in R.
Since Me U, there exists g E U such that g is not in M.
Now g is not in M:::::}g(1I2) -:j:. O:::::}r = g(l/2) -:j:. O.

Now define two constant functions q and qi as follows:

q(x) = r, q'(x) = l!r for all ·x E [0, 1].
Then q, q 1 E M. Write h = g - q.

Now h(1/2) = (g - q)(l/2) = g(1! - q(1I2) =r- r=0~ hEM c U.

Also h, g E U ~ q =g- h E U.
Since ql E Rand U is an ideal, we have that qqI E U.
Clearly qqI = (qqI)(X) = q(X)qI(X) = r.(lIr) = 1 = l(x) ~ qql = 1.
Therefore 1 = qql E U:::::}U = R (by a known result).
Hence M is a maximal ideal of R.

(iv) The proof is a part of (iii).

11.2.3 Example: Consider R as in above problem. Let o ~ a ~ 1, then

I = {f E R I f(a) = O} is a maximal ideal of R. The verification for this statement is similar to the
verification given in the solution of the problem 11.2.2.·
/

11.2.4 Problem: If F is a field, then (0) is a maximal ideal of the ring F.

Solution: Suppose that F isa field. By Result 10.3.4 (of Lesson 10), we know that F has only
two ideals (0) and F.
 !Acharya Nagarjuna University 11.7 Centre for Distance Edu('~

This means that there are no other ideals lies between (0) and F.
Hence we can conclude that (0) is the unique maximal ideal of the ring F.

11.2.5 Problem: Consider R == Z, the ring of integers with respect to usual addition and
multiplication of numbers. Write nZ = {nx !x E Z}, the set of multiples of ~ for any integer n E
Z. Then I is a maximal ideal of Z ¢::} I = pZ for some prime number p E Z. \
\
\
'.
Solution: We know that nZ is an ideal of Z, and every ideal of Z is of this form. Suppose that I
is a maximal ideal of R = Z.
Since I is an ideal, I = pZ for some positive integer pE Z.
Now we show that p is a prime number.
In a contrary way, suppose that p is not a prime number. This implies p = ab, 1< a < p, 1< b < P
for two integers a and b.
Write J = aZ. Now we show that I c J. Now if a E I, then a E pZ
=? a = px for some x =? p = ab = (px)b = p(xb) =? xb = 1 (by canceling p) =? b = 1 or b = -1, a
. contradiction to the selection of b.
Therefore a E I.

Now mE I=pZ
=? m = px for some XE Z =? m = px = (ab)x = a(bx) E J. Hence I c J.
Since I ~ J c R .and I is a maximal ideal, we have that J = R =? 1E R = J = aZ
=? 1 = ay for some integer y =? a = lor a = -1, a contradiction to the selection of a.
Hence p is a prime number.
Converse: Suppose that p is a prime number and I = pZ .
Now we show that I is a maximal ideal.
Suppose K is an ideal such that I ~ K c Rand I::f: K.
Now it is enough to show that K = R.
Since K is an ideal, by a known result, we have that K = nZ for some positive integer n.

Now p E pZ = I ~ K = nZ =? p E nZ =? p = nm for some integer m.

=? p divides n or p divides m (since p is a prime number).
 IAlgebra 11.8 More on Ideals & Quotient Ringsl

If P divides n; then n = ps for some s E Z

=> n E pZ and for any x, nx = (ps)x = p(sx) E pZ = I => nZ c I
=> K ~ I => K = I (since I c K) , a contradiction to I :t K.
Therefore p does not divide n. This shows that p divides m.
This means m = pr for some rE Z.
Now p = nm = n(pr) = p(nr) => 1= nr E nZ = K => IE K => K= R (since any ideal containing 1 is
equal to R).
This completes the required verification.

11.2.6 Theorem: If R is a commutative ring with unit element and M is an ideal of R, then M is
a maximal ideal of R ¢::> RIM is a field.

Proof: M is a maximal ideal of R

¢::> there is no ideal I of R such that M c I ~ R

¢::> there is no ideal I of R' such that (0) = MIM c 11Mc RIM (since thereis al-J correspondence
between the set of ideals of R containing M-and the set of ideals of RIM)

¢::> RIM contains only ideals (0) = MIM and RIM

. ¢::> RIM contains only ideals (0) and itself

¢::> RIM is a field (by the 10.3.20 Self Assessment Question (of Lesson 10)

11.2.7 Self Assessment Question:

Show that a commutative ring with identity is a field ¢::> {O}is a maximal ideal.

11.2.8 Self Assessment Question :

Let R be a commutative ring with 1 (:t 0). Show that the following statements are equivalent
(i). R is a field; (ii). R has no non-trivial ideals; and (iii). {O}is a maximal ideals of R
 iAcharya Nagarjuna University 11.9 Centre for Distance Educationl

11.2.9 Problem: Let J be the ring of integers, p is a prime number and (p) = pJ is the ideal of J
consisting of all multiples of p. Prove (i) J/(p) is isomorphic to Jp (that is Zp, the ring of integers
modulo p ). (ii) Using the theorerri 11.2.6 and part (i) of this problem show that J, is a field.

Solution: (i) Define f: J -7 Jp by f (n) = n where n is the equivalence class containing n(under the
relation modulo p).
fen! + n2) = n, + n2 =. n, + n2 = fen,) + f(n2) and also
f(n1 n-) = n1 n2 = n1 n2 = f(n,)f(n2) .
Therefore f: J -7 Jp is a ring homomorphism.
Now for any mEJp, we have 0:::;m:::;(p-I) and so f(m) = m.
This shows that f is onto .Now by theorem 11.1.1 (i), we have that
J/(Ker f) is isomorphic to Jp. Now we have show that Ker f = pZ .
Now SE Kerf =? f(s) =0~ s = 0 =? s-O is divisible by p ~ s is divisible by p =? S = px for

some x ~ SE pZ.
Hence Kerf c pZ. Let r E pZ =? r = px for some x E Z =? P divides r = r-O =? r = 0 (mod p) =?

fer) =r=0~ r E Ker f.

Hence Ker f = pZ = (p) .
Therefore J/(p) == Jp.
(ii) Since p is a prime number we have (p) = pZ is a maximal ideal of J.
, By the theorem 11.2.6, we have that J/(p) is a field.
_ Hence J/(p) == Jp is a field.

11.3 PRIME IDEALS

3.3.1 Definition: An ideal P of a commutative ring R is said to be a prime ideal if it satisfies the
following two conditions:
(i) P is not equal to R; and
(ii) a, b E R such that ab E P =? either a E P or b s P.

I
I
IAlgebra 11.10 More on Ideals & Quotient Ringsl

11.3.2 Self Assessment Question Let R be a ring. If 1 E R, then R2 = R.

11.3.3 Result: In a commutative ring, show that every maximal ideal is a prime ideal.

Proof: Now we show that every maximal ideal is a prime ideal.

Let M be a maximal ideal,
If M is not a p:ime ideal, then there exists a, b E R such that ab E M ~ a e M and be M
Consider Ra + M and Rb + M.
Since a = l.a + 0 F Ra + M and a ~ M, we have that
M C Ra + M c R. Since M is a maximal ideal of R, we have that Ra + M = R. Similarly, we can
verify that Rb + M = R.
Now R = R2 ( since R contains 1) = (Ra + M)(Rb + M)
~ RaRb + RaM + MRb +MM ~ Rab + M eM + M (since ab E M ~ Rab eM) =M.
This shows that R = M, a contradiction.
Hence M is a prime ideal.

11.3.4 Result: Show that in a commutative ring R, we have that P is a prime ideal ¢:::> RIP is an
integral domain. 'i

Proof: Suppose that P is a prime ideal.

Let a', bl E RIP such that a'. bl = 0 =0+ P in RIP.
Now a' = a + P, bl = b + P for some a, b e R.
Now 0 + P = aibl = (a + P) (b + P) = ab + P ~ ab E P

=} a E Par b E P (since P is prime ideal) ~ a + P = 0 + P or b + P = 0 + P

~ a' = a + P = 0 or bI = b + P = 0 ~ a' = 0 or bl = O.
Thus we have proved that a'b' =0~ a' =; 0 or bl = 0 (in RIP).
Hence RIP is an integral domain.
Converse ~Suppose 'that RIP is an integral domain. Let a, b E R such that ab E P. Write a1= a +P

and b' = b + P.
 (Acharya Nagarjuna University 11.11 Centre for Distance Educatio~

Now a'b' = (a + P) (b + P) = ab + P = 0 + P ~ a' = 0 + P or b ' = 0 + P (since RIP is an integral

domain)

~ a + P = a I = 0 + P or b + P = bi = 0 + P ~ a E P or b E P.
Thus we have proved that ab E P~ a E P or b s P.
This shows that P is a prime ideal.

11.4 IMBEDDINGS
.11.4.1 Definition: A ring R is said to be imbedded in a ring R I if there exists a monomorphism
f: R -7 R 1. Moreover, if Rand R 1 are rings with 1 then we insist that f( 1) = 1.

11.4.2 Theorem: Every integral domain D can be imbedded in a field.

Proof: Write M = {(a, b) / a, bED, and b"* OJ.

Define a relation - on M as (a, b) - (c, d) {::::> ad = be.
First we verify that - is an equivalence relation.
(i) (a, b) E M. Since ab = ba (because integral domain is commutative) we have (a, b) - (a, b).

This shows that - is reflexive.

,
(ii) Suppose (a, b), (c, d) E M such that (a, b) - (c, d) ~ ad = be ~ cb = da
=> (c, d) - (a, b). This shows that - is symmetric. o

(iii) Suppose (a, b), (c, d), (e, f) are in M such that (a, b) - (c, d) and (c, d) - (e, f)

~ ad = be and ef = de. Note that d is not equal to zero. :{

We have to show that (a, b) - (e, f) that is af = be.
Now cf = de ~ b(ef) == btde) ~ (bc)f = (bd)e ~ (ad)f = d(be) ~ deaf) = d(be)
~ af = be (since d is a nonzero element of an integral domain)

~ (a, b) - (e, f).

This shows that - is transitive.

Since the relation - is reflexive, symmetric and transitive, we may conclude that the relation is an
equivalence relation on M.

Let us denote the equivalence class containing (a, b) by [a, b] or a/b.

Write F = {a/b / a/b is the equivalence class containing (a,b) }.
 1A1gebra 11.12 More on Ideals & Quotient Ringsl

We define addition operation + and the multiplication operation· on F as o

aIb + c/d = (ad + bc)/bd and (aIb).(c/d) = (ac)/(bd) for any aIb, c/d E F.

Since b * 0, d * 0 we have bd * 0 (because the ring is an integral domain). o

One can easily verify that + and'·' are binary operations on F.

Also these two operations are associative.
O/b acts as the zero element (that is, additive identity) of (F, +).
Also (-a)/b acts as the additive inverse of aIb.
A routine verification shows that the distributive laws hold good in F.
I ,

Therefore (F, +, .) is a ring.

For any 0 * d E D, did acts as the multiplicative identity

(for this for any aIb, we have a/b-d/d = ad/bd = a/b).
If O:j; a/b EF, then a * O. Consider (a/bj-Ib/a) = (ab)/(ab) = 1 = did.
Hence b/a is the inverse of a/b if a/b * O.
Therefore (F, +,.) is a division ring.

Also (a/bj-tc/d) = (ac)/(bd) = (ca)/(db) = (c/dj-Ia/b).

Hence (F, +, .) is a commutative division ring (that is, a field).
Fix 0 * xED. Define <1>: = (ax)/x.
D ~ F by <1>(a)
It is clear that (ax)/x = (ay)/y for any 0 :j;y E D [because

axy « ayx ~ (ax, x) - (ay, y»).

= ((a + b)x)/x
Now <1>(a+b) = (ax + bx)/* = ax/x + bx/x = <1>(a)+ <1>(b)
.
. Also <j>(ab)= (abx)/x = (abx2)/x2 = (aklx}(bx/x) = <1>(a)·<1>(b).
Therefore <1>
is a ring homomorphism.

Now we have to verify that <pis a one-one function. For this, let a, bED such that <p(a)= <1>(b)

~ (ax)/x = (bx)/x
~ (ax, x) - (bx, x) ~ ax2 = bX2~ ax = bx (by right cancellation law) ~ a = b.Therefore <1>:
D~F is a 1-1 homomorphism.
Thus D can be imbedded in F.
IAcharya Nagarjuna University 11.13 Centre for Distance Educationl

11.4.3 Note: The field F constructed in the above theorem is called the field of quotients of D.
We may verify that if D = Z, then F = Q.

11.4.4 Self Assessment Question: Let R be an integral domain and F is the field of quotients
of R Then every element x of F can be expressed as x = ab' for some elements a, b E R with
b ;t: O.

11.4.5 Self Assessment Question: Let R be an integral domain and F is the field of quotients of
R. Then F is the smallest field containing R

11.5 SUMMARY
The concept of maximal ideal of a ring is introduced. Some maximal ideals in some
concrete rings were given. M = pZ is a maximal ideal of the ring of integers Z where p is prime
number. We verified that M = {f(x) ER / f( 1/2) = O} is a maximal ideal of R where R is the ring
of all the real valued continuous functions on the closed unit interval. We tried to link a general
ring with fields. We found the conditions for which a homomorphic image of a ring becomes a
field. We solved this more explicitly for commutative rings. We proved that if R is a
commutative ring with unit element and M is an ideal of R, then M is a maximal ideal of R ¢::?
RIM if a field.

We learnt that the ring of integers can be enlarged to the set of rational numbers which is a
field. First we defined embedding. We proved the theorem that
"every integral domain can be imbedded in a field". Tevprove this, we defined an equivalence
relation on M, the set of all ordered pairs (a, b) where a, QED, b ;t: 0, where D is an integral
domain; and there by we consider the set of all equivalence classes in M. We defined addition and
multiplication on the set F of all equivalence classes. We proved that ·F, the set of all such
equivalence classes [a, b) where a, bED and b ;t: 0 forms an abelian group. Later we verified that
F is a field. Finally we showed that D can be imbedded in F. This field F constructed here is
called as the field of quotients of D.

11.6 TECHNICAL TERMS

"-
Maximal Ideal: An ideal I of a ring R is said to be a maximal ideal of R if it
satisfies the following two conditions:' - .,
(i) I is not equal to R; and
(ii) M is an ideal of R such that I <;;;;; M <;;;;; R => either I =
Mar M=R.
 !Algebra 11.14 More on Ideals & Quotient Ringsl

Prime Ideal: An ideal P of a commutative ring R is said, to be a prime

ideal if it satisfies the following two conditions:
(i) P is not equal to R; and
(ii) a, b E R such that ab E P => either a E P or b E P.

Imbedded: A ring R is said to be imbedded in a ring R I if there exists 'a

monomorphism f: R ~ R I. Moreover, if Rand R I are rings
with 1 then we insist that f( 1) = 1.

11.7 ANSWERS TO SELF ASSESSMENT QUESTIONS

11.2.7: Let R be a commutative ring with identity. By the Theorem 11.2.6, we know that "if R
is a commutative ring with unit element and M is an ideal of R, then M is a maximal ideal of R <==>
RIM is a field".
Taking M = {a}, we can conclude that {O}is maximal ideal <==> R/{O}= R is a filed.

Therefore we have that "if R is a commutative ring with 1, then R is a field <==> {a} is a maximal
ideal of R.

\"\
11.2.8: (i) ~ (ii): Suppose R is a field. The Result 10.3.4 (of Lesson 1O)! states that "if F is a
field, then only ideals of F are (0) and F itself". So we can conclude that if R is a field, then R
contains no nontrivial ideals.

(ii) ~ (iii): Suppose that R contains no nontrivial ideals. So the ideals of Rare {a} and R. This
implies that there are no ideals in between {a} and R other than {a} and R. This means that {OJ
is the maximal ideal of R.

. (iii) ==> (i): This part follows from the Self Assessment Question 11.2.7.

11.3.2: Let x E R. Now x = x-I E R-R. So R ~ R-R = R2.

Since R is closed under multiplication, we have that R·R ~ R, that is R2 cR. Hence R2 = R.
 !Acharya Nagarjuna University 11.15 Centre for Distance Educatio~

llr4.4: Identify R with its image feR), that is, every element a in R being identified with its
image f(a).
Let x = a/be F.

Then x = a/b = (ab11b1)(b1/bb1) = f(a).f(b) -1 = ab', under the identification.

11.4.5: Let pI be any field containing R. Then for any x E P, X = ab', a, b E R, b 7: O. Since
R c FI, a, b E Fl and since Fl is a field, it follows that x = ab' E Fl. Thus Fe Fl. This shows
that F is the smallest field containing R.

11.8 MODEL QUESTIONS

1. Define the term maximal ideal. If F is a field, then prove that (0) is a maximal ideal of the
ring F.
2. Let R be a commutative ring with 1 (:;t: 0). Show that the following statements are equivalent
(i). R isa field; (ii). R has no non-trivial ideals; and (iii). {O}is a maximal ideals of R
3~ Define the term prime ideal. In a commutative ring, show that every maximal ideal is a prime
ideal.
4. Define the term 'embedding'. Show that every integral domain D can be imbedded in a field.

--''i1.9 REFERENCE BOOKS 'P

, . ".
':..

I. Gopalakrishnan N.S. 'University Algebra', Wiley Eastern Limited, New Delhi,

1995.
2. Herstein LN. 'Topics in Algebra', Wiley Eastern Limited, New Delhi, 1988.

Dr. Bhavanari Satyanarayana

 LESSON 12-
EUCLIDEAN RINGS

Objectives

The objectives of thislesson are to:

• Define Euclidean ring, principal ideal ring, division ring, greatest common divisor, unit,
associates, prime element, relatively prime elements
• Prove some basic lemmas and theorems on the concepts defined
• Deduce that a Euclidean ring is a principal ideal ring
• Understand the difference between a unit and a unit element
• Prove the unique factorization theorem for Euclidean rings
\i Determine all the maximal ideals in an Euclidean ring
,; Define the domain of Gaussian integers J[i]
• Study the ring of Gaussian integers, a particular Euclidean ring
• Prove Fermat's Theorem

Structure

12.0 Introduction
12.1 Euclidean rings' .
12.2 Principal Ideal Rings
12.3 Prime elements
12.·1 The Ring of Gaussian Integers
12.~:Summary
12.6 Technical Terms
12.7 Answers to Self Assessment Questions
12.8 Model Questions
12.9 Reference Books

12.0 INTRODUCTION

We now formulate the concepts like divisibility, factorization, prime elements, greatest
.common divisor, etc., for a general commutative ring. In this lesson, we study some types of
rings which possess the property similar to the property of division algorithm in the ring Z of
integers. We prove the unique factorization theorem. We also prove that-any ideal A in an
Euclidean ring R is of the form A = (ao) where (ao) = {xao / x E R}. We give a simple, precise
answer to the question: What conditions imposed on an ideal A = (ao) insure that A is a maximal
ideal of R? In the last part of this lesson, we are about to particularize the notion of Euclidean
ring to a concrete ring: the ring of Gaussian integers. We defined the set of Gaussian integers and
observed that the set of Gaussian integers forms an Euclidean ring. Out of the observations and
/Algebra 12.2 Euclidean Ringsl

the results developed previously, we prove a classic theorem in number theory due to Fermat,
which states that "every prime number of the form 4n + 1 can be written as the sum of two
squares" .

12.1 EUCLIDEAN RINGS

12.1.1 Definition: An integral domain R is said to be an Euclidean ring (Euclidean domain) if

for every 0 *-a E R, there defined a non negative integer d(a) such that
(i) for all a, b E R both non zero, d(a) ::;d(ab)

(ii) for any a, b E R both non zero, there exists t, r E R such that a = tb + r
where either r = 0 or d(r) < deb).

12.1.2 Example: Let R = Z, the ring of integers. Define d(a) = I a I (that is, the modulus of the
integer a) for 0 * a E Z, then Z becomes an Euclidean ring.
12.1.3 Remark: In a commutative ring R with 1, for any a E R. we know that aR = Ra is an ideal.
If A is an ideal such that a E A, then aR s A.
Therefore Ra = aR is the smallest ideal containing a. If we denote the smallest ideal containing a
by (a), then (a) = aR = Ra.
From this discussion, we may conclude that for any element a in a commutative ring R with 1,
the ideal generated by a is of the form aR. In the following theorem, we prove that any ideal A of
an Euclidean ring R is of the form aR for some element a 'in A.
12.1.4 Theorem: Let R be an Euclidean ring and A is an ideal of R. Then there exists
a E A such that A = {ax / x E R} = aR.
Proof: If A = (0), then a = 0 and hence A = OR = aR.
Suppose A *-(0). Since A *-(0), there exists an element y suchthat 0 *-yEA
So the set { d(x) /0 *- X E A} is a nonempty set which is bounded below by O.
Thus we can find a minimum number from the set { d(x) / ° *- X E A}.
Now we can select an element a E A such that a*-O and

d(a)=min{d(x)/O*-xE A}.
IAcharya Nagarjuna University 12.3 Centre for Distance Educatio~

Since A is an ideal and a E A, we have that aR c A.

Now we verify that A c aR.

For this, take b E A. Since a, b E R, and R is an Euclidean domain, there exists t, r E R such that
b = ta + r where either r = 0 or d(r) < d(a).
Now a E A, A is an ideal =:}ta E A, also b E A =:}r = b - ta E A.
If r -:F 0, then d(r) < d(a) and rEA, a contradiction to the minimality of d(a).

Therefore r = O. This implies b = ta + r = ta + 0 = ta = at E aR.

This shows that A s; aR. Hence A = aR.

12.1.5 Corollary: An Euclidean ring R possesses a unit element

Proof: Since R is an ideal of R, by the theorem 12.1.4, there exists a E R such that R = (a) = aR.
Since a E R = aR, there exists e E R such that 'a = ae.
Let x E R. Since x = aR = Ra, there exists
E R y E R such that x = ya.
Now xe = (ya)e = y(ae) = ya = x=:} xe = x.
This is true for any x E R. Since R is commutative xe = x = ex, for all x E R.
Hence e is the identity in R. Thus R possesses the unit element.

12.2 PRINCIPAL IDEAL RINGS

12.2.1 Definition: (i) Let R be a ring and a E R. Then the smallest ideal containing 'a' is called
the ideal generated by 'a'. The smallest ideal containing 'a' is denoted by (a).

(ii) An integral domain R with unit element is called a principal ideal ring if for every ideal A of
R there exists a E A such that A = (a).

12.2.2 Corollary: Every Euclidean ring R is a principal ideal ring

Proof: By the above Corollary 12.1.5, the given Euclidean ring R contains the identity. By the ,_
Theorem 12.1.4, every ideal A of R is of the form A = Ra for some a E A. By Remark 4.1.3, it
follows that A = Ra = (a) This shows that R is a principal ideal ring.
 12.4 Euclldean~

12.2.3 ' vefinition: Let R be a commutative ring and a, b e R, a:f. a.

We say that a divides b if there exits c E R such that b = ac.
We use a I b to represent the fact 'a divides b'. We use' a Db to mean 'a does not divide b'.

12.2.4 Remark: Let R be commutative ring. Then the following facts can be verified directly

(i) a I b, b I c ~ a I c;
I I => a I (b ± c); and
(ii) a b, a c

(iii) a I b => a I bx
for all a, b, c, x E R.

12.2.5 Definition: Let R be a commutative ring and a, b, c, d E R.

Then d is said to be a greatest common divisor (gcd or g.c.d.) of a and b if:
(i) d I a and d I b (means d is a common divisor of a and d)
o
(ii) C 12" c 1 b => c 1 d.
The gcd of a and b is denoted by' (a.b).

12.2.6 Lemma: Let R be an Euclidean ring. Then any two elements a, b in R have a greatest
common divisor d. Moreover the gcd is of the form
d:=: A-a" ~b for some A,!-l E R.

Proof Write A = {ra + sb / r, s E R}.

First we show that A is an ideal of R.
For this, take x, yEA.

Now x = ria + Sib, Y = r2a + S2b for some rl, r2, SI, S2E R.
o
=> x - y = (ria + sib) - (r2a + S2b) = (r, - r2) a + (s, - S2)b E A.
-If r e R, then rx e r Ir.a+ sib) = (rr.ja « (rsl)bE A. Hence AisanidealofR.

Now by Theorem 12.1.4, there exists d E A such that A = dR = (d), whered E A.

\

Now dB. A and so by the definition of A, there exists A, /-l E R such that d = Aa + ub.
By the Corollary 12.1.5, R possess the identity.
So a = La + a.b E A, and b
I
= a.a + l.b E A.
IAcharya Nagarjuna University 12.5 Centre for Distance Educatio~

Since a, b E A;::: dR, it follows that a;::: da., b;::: da, for some aI, a2 E R. Thus dla and dlb, and so
d is a common divisor of a and b.
To verify that d is a gcd ofa and b, take c E R such that cia and clb.
l

Now clAa and club (by Remark 12.2.4(iii»

=> cl(Aa+J.!b)(by Remark 12.2.4(ii» => cld.

This shows that d;:::Aa + ub is a g.c.d of a and b.

12.2.7 Definition: Let R be a commutative ring with identity (that is, unit element). An element a
E R is said to be a unit in R if there exists an element b E R such that ab ;:::1.

12.2.8 Note: Observe the difference between unit and unit element.

12.2.9 Lemma: Suppose R be an integral domain with identity, and a, b E 'R are two elements
such that both a I band b I a are true. Then a ;:::ub for some unit u in R.

Proof: Now a I b, b I a => b ;:::xa, a ;:::yb for some x, y E R

Observe that l.b;::: b;::: xa;::: x(yb);::: (xyjb?
. .
Recall that cancellation laws hold g?od in integral domains.
Now Lb > (xy)b => ~ ;:::xy (by cancellation laws)
Therefore x, yare units.
Hence a ;:::ub where u ;:::y is a unit in R.

12.2.10 Definition: Let R be a commutative ring with identity. Then a, b E R are said be
associates if b = au for some unit u in R.

12.2.11 Problem: In a commutative ring R with identity 1, prove that the relation a is an
associate of b is an equivalence relation. '> .
. 0

Solution: Write a - b ¢::> a is an associate of b (means b;:::'ua for some unit u).
Since a ;:::1.a, and 1 is a unit we have that a-a. Therefore - is reflexive.

Suppose a - b => b ;:::ua for some unit u.

 !Algebra 12.6 Euclidean Ringsl

Since u is a unit, there exist w E R such that wu = 1.

Now wb = w(ua) = (wu)a = La = a => wb =a and w is a unit => b ~ a.

Hence ~ is symmetric .
•
Suppose a ~ b, b ~ c => b = ua,c = wb for some units u, w in R.
Since u, ware units, there exists u ', w' E R such that uu' = 1, ww(=
I
1.

So (uW)(WIUI)= u(WWI)UI= u.Lu' = uu' = 1.

Hence' uw is also a unit.
Now c = wb = w(ua) = (wu)a and wu is a unit => a ~ c.
Hence ~ is transitive.
Now we proved that the relation ~ is reflexive, symmetric and transitive.
Hence the relation ~ is an equivalence relation.

12.2.12 Problem : If R is~n Euclidean ring, and a, b E R, then prove that any two greatest
common divisors of a and b are associates.

Solution: Let d1, d2 be two g.c.d's of a and b. Since d, is a common divisor of a and b, and d2 is a
g.c.d. Of a and b, we have that d.] d2 (by the definition of g.c.d).
'~ Similarly since d2 is a common divisor and d, is a g.c.d., we have that d21di.
Since dd d2, and d21 db by Lemma 12.2.9, we can conclude that d, = ud- for some unit u in R.
Therefore d., d2 are associates. This proves that any two greatest common divisors of a and b are
associates.

12.2.13 Note: In the proof of the Theorem 12.1.4, we verified that, if A is an ideal and x E A
such thatd(x) = min {d(y) / yEA}, then A = Rx. We use this fact in the next Lemma 12.2.14.
12.2.14 Lenc.aa : Let R be an Euclidean ring and a, b E R. If 0 ::j; b E R is not a unit in R, then

d(a) < d(ab).

Proof: Since R is a commutative ring, we have that A = Ra = { xa / x E 'R: lis an ideal of R, and
A = (a). \

Now yEA ~ Y = xa for s?me x E R.

Since R is an Euclidean ring, we have that d(a)::; d(xa) = d(y) for all yEA.
IAcharya Nagarjuna University 12.7 ,Centre for Distance Educationl

Therefore d(a) s min {d(y) / YEA}. \

Now we show that d(a) < d(ab).
In a contrary way, assume that d(a) = d(ab).
Now d(ab) = d(a) = min {d(y) / YEA}. By Note 12.2.13, we conclude that A = Rab.
Since a E A, there exists c E R such that a = cab =} l.a = cba =} 1 = cb (by cancellation law).
Therefore b is a unit, a contradiction to the hypothesis that b is not a unit.
Hence d(a) < d(ab).

12.2.15 Self Assessment Question: Let R be an Euclidean ring and a an element in R. Sho~
that the element ais a unit ¢:? d(a) = d(l).
12.2.16 Self Assessment Question: Let R be an Euclidean ring. Show that
(i) d(a) 2::d(l) for each non-zero a E R; and

(ii) d(a) = deb) for any two non-zero elements a, b E R that are associates.

12.3 PRIME ELEMENTS

12.3.1 Definition: In an Euclidean ring R, a non-unit 1t is said to be a prime element of R if

whenever a, b E Rand 1t = ab, then either a is a unit or b is a unit in R.

12.3.2 Lemma: Let R be an Euclidean ring and a E R. Then either a is a unit or

a = ql q2 ... qrn where qi, 1 s ism are prime elements of R.
Proof: In this proof, we use the mathematical induction on d(a).
Since d( 1) s d(l.x) = d(x) for all x, we have dt l ) S min {d(x) / x E R}.
If d(a) = d(l), then by Self Assessment Question 12.2.15, we get that 'a' is a unit.
Induction hypothesis: Assume the result for all x E R such that d(x) < d(a).
If 'a' is a prime element, then there is nothing to prove.
So we suppose that a is not a prime element.
Then a= be, where neither b nor c is a unit in R.
By Lemma 12.2.14, we get that deb) < d(bc) = d(a) and also d(c) < d(bc) = d(a).
Thus by our induction hypothesis, we get that b = PIP2 ... P» and c = SlS2 .. Sk where pi, 1 S i S n
and Sj, 1 s j skare prime elements ofR.
I
jAlgebra 12.8 Euclidean Ri~

Now it is clear that a = be = PIP2 '" pn Sr52 .. Sk.

This completes the proof of the Lemma.

12.3.3 Definition: Let R be an Euclidean ring. The elements a, b E R are said to be relatively
prime if the g.c.d of a and b is a unit ofR.

12.3.4 Lemma: Let R be an Euclidean ring, and a, b, c E R such that albc and (a, b) = 1. Then

a' c.

Proof: By Lemma 12.2.6, the g.c.d of a and b can be written as ra + sb for some r, s E R. Given
that the g.c.d of a and b is 1.
Hence 1 = ra + sb
Now c = c.I = c.(ra + sb) = cra + csb.
Since a I be, we hav.e that a I csb. Clearly a I era. Therefore a I (era + csb) => a I c.
This completes the proof of lemma.

12.3.5 Lemma: If R is an Euclidean ring, 'a' is a prime element and a I be where b, c E R, then a
I b or a ~c.

Proof: Suppose a do not divide b. Now we have to show that a divides c.

Write d = (a, b).
Now d I a and a is a prime element => d = a or d is a unit.
If d = a, then we get that a divides b, a contradiction to our supposition.
In the other case, we have that (a, b) = 1. Now {a, b) = 1 and a I be.
By Lemma 12.3.4 we get that a I c.

12.3.6 Corollary: If a is a prime element in the Euclidean ring R and a I b.b, ... b, for some b, E
R, 1 ::; i s n, then a I b, for some 1 s i s n.

Proof: The proof is by induction on n. If n = 1, then a I b I.

Suppose the res~lt is true for n-1.
Given a I (b1b2 ..• bn-j)bn• If a divides b.; then it is clear.
IAcharya Nagarjuna University 12.9 Centre for Distance Educationl

Now consider the case that a does not divide b.,

By Lemma 12.3.5, we can conclude that a I b, ... bn-I.
Now by induction hypothesis, it follows that a I b, for some i.
This completes the proof of the corollary.

12.3.7 Theorem (Unique Factorization Theorem): Let R be an Euclidean ring and O:t a E R
. S uppose t h at a
not a umt. = qlq2 .. , qn = SIS2... Smwere
h qi, 1<'<
_ 1 _ n an d Sj, 1<'< .
_ J _ mare pnme
elements of R. Then m = n and each qi is an associate ofs, for some j (1 ~ j ~ m).
Proof: Given that a = qlq2 ... qn = SIS2·.· sm'
Since q] I a, we have that qll SIS2... Sm.By Corollary 12.3.6, it follows that qll Sj for some 1 :::;j ~
m.
Without loss of generality, we may assume that j = 1.
Then qll Sl ~ Sl = u.qi. Since SI is prime, we have Ul or ql is a unit.
As q, is a prime element, it cannot be a unit and so UI is a unit.
This shows that ql is an associate of s I.
Now qlq2 ... qn = SIS2.. · Sm = UlqlS2 ... Sm~ q2 ... qn = U1S2... Sm (by canceling qj on both sides
(Note that as R is integral domain cancellation laws holds)).
Since q21 q2 ... qn = U1S2"-Sm ~ q21 Sj for some 2 ~j ~ m.
With out laws of generality, we may assume that q21 S2·
Then S2 = U2q2for some unit U2in R.
Therefore q2 .. , qn = U1S2.. Sm = UIU2Q2S3
... Sm~ q3 ... qn = UIU2S3".s., (by canceling q2 on both
sides).
If we repeat this argument up to n steps, the left hand side becomes 1 and the right hand side is

equal to UjU2... UnSn+l... Sm.

Therefore n ~ m and 1 = UIU2... UnSn+1... Sm.
If n < m, then there exists at least Sn+1on right hand side.
Since 1 = Sn+1(UIU2... Un... ), we have that Sn+1is a unit, a contradiction to the fact that Sn+1is a
prime element.
Hence n = m. In the process we proved that every qi is an associate of Sj for some j.
 k\lgebra 12.10 Euclidean Rings!

12.3.8 Result: Every non-zero element of an Euclidean ring R is either a unit or it can be
uniquely written (up to associates) as a product of prime elements of R.

Proof: The proof is the combination of Lemma 12.3.2 and Theorem 12.3.7.

12.3.9 Self Assessment Question: Let R be an Euclidean ring and B an ideal, b E R such that

B = (b) = Rb. Then d(b) = min {d(c) / c E B}.

12.3.10 Self Assessment Question: Let R be an Euclidean ring and B an ideal of Rand
B = (b) = Rb. Then B = R ¢:::} b is a unit.

12.3.11 Lemma: Let R be an Euclidean ring, A an ideal of R and a E R such that

A ~ (a) = Ra. Then A is a maximal ideal of R ¢:::} a is a prime element of R.

Proof: Suppose that A is a maximal ideal of R.
In a contrary way suppose that 'a' is not a prime elemer ' of R.
Then a = be where neither b nor c is a unit.
Write B= (b) = Rb. Clearly B ::f:. R (by the Self Assessment Question 12.3.10).

Let x E A = Ra:=:} x = ra for some r E R:=:}x = ra = r(bc) = (rc)b E Rb = B.

Hence A k B. Now since A is maximal, A c B c Rand B 1= R, it follows that A = B.

Now b E B = A = Ra :=:}b = ta for some t E R
:=:}l.b = b = ta = tbc = (tc)b:=:} 1 = tc (by canceling b on both sides) :=:}c is a unit, a contradiction.
Therefore 'a' is a prime element of R.
Converse: Suppose that 'a' is a prime element of R.
To show that A is a maximal ideal of R, take an ide~l U of R such that A c U c R.
Since R is an Euclidean ring and U is an ideal of R, by Theorem 12.1.4, there exists an element s
E R such that U = Rs.
Now a E A C U = Rs :=:}a = ws for some w E R
==>either w or s is a unit (since 'a' is prime).
Suppose s is a unit.
Then Rs = R (by the Self Assessment Question 12.3.10), and hence U = Rs = R.

15)
[Acharya Nagarjuna University 12.11

Suppose w is a unit.
Then a = ws::=} (w·l)a = s::=} S = w·1a E Ra = A::=}U = Rs c A::=}U c A..
Hence U = A (since we assumed that A C U).
Now we proved that A C U c R, U is an ideal of R ::=}U = A or U = R.
Hence A is a maximal ideal of R.

12.4 .THE RING OF GAUSSIAN INTEGERS

12.4.1 Note: (1) Jr~] = {a + ib / a, b E Z}where Z is the set of integers forms. an integral domain
with respect to usual addition and multiplication of complex numbers.
This (J[i], +, .) is called the domain of Gaussian integers.

(ii) For ° ;:j:. x = a + ib E J[i], define d(x) = a2 + b2. Then

d(x) is a positive integer for each x ~ J[1J
Since x;:j:. 0, we have that either a or b is nonzero, and so d(x) =, a2 + b2 ~ 1.
'iii) For any two Gaussian integers c, = x + iy, C2= U + iw we have
d(CIC2)= d[(x + iy) (u + iw)] = d[(xu - yw) + i(xw + yu)]
= (xu - YW)2+ (xw + yU)2
= (XU)2+ (ywi - 2 (xu)(yw) + (XW)2+ (yui + 2(xw)(yu)
= (XU)2+ (YW)2+ (XW)2+ (YU)2= (x2 + l)(u2 + w2)
= d(x + iy) d(u + iw) = d(Cl)·d(C2).
Therefore d(CIC2)= d(Cl)·d(C2) for any two Gaussian integers. If we define d suitably for complex
numbers, then this condition is-also true for complex numbers c, and C2.

(iv) Let x, y E J[i] such that x;:j:. 0, y ;:j:.0. Then by (ii), we get that

d(x) ~ 1 and d(y) ~ 1.

Also d(x) = d(x).1 ~ d(x)·d(y) = d(xy) ::=}d(x) ~ d(xy).

(v) Let m, n are ifitegers such that n > 0.

Then there exists two integers t and r such that m = tn + r where I r I::;(V2)n
[Verification: Consider t, r E Z such that m = tn + rand ° < r < n (Existence by division

algorithm).
/Algebra 12.12 Euclidean Ringsl

If I r I < (Y2)n, then it is clear. Otherwise, I r I ~ (Y2)n.

Now it is clear that In - rl ~ (Y2)n.

Now m = tn + r = (t + l)n + (r - n),and I r - n I = I n - r I ~ (Y2)n].

12.4.2 Theorem: J[i] is an Euclidean ring.

Proof: For 'each a t:. x ~ a + ib E J[i], define d(x) = a2 + b2•

Then by the above note 12.4.1, we have
(i) d(x) ~ a for each a t:. x E J[i]; .and
(ii) d(x) ~ d(xy) for each a t:. x, a :;:.Y. E J[i].
A direct verification shows that J[i] is an integral domain.
Now it is enough to show the following:
If x, y E J[i] and a =t X, then there 'existst, r E J[i] such that y = tx + r where r = a or d(r) < d(x).
We divide the remaining part of the proof into two cases.
Case (1)~ Suppose x = n = n + 0 ='n + i.O; n is an integer; and y= a + ib.
By the Note 12.4.1, we can find integers u, Ul such that a == un + UI and also integers w, WI such
that b = wn + WI, where Iud < (Y2)n and IwI! < (Y2)n.
Write t = u + wi; and r = UI + w.i. . "
T~en y = a + bi = un + UI + i (wn + WI) =n(u +iw) + (UI + iWI) = tn + r.
If r:;:' 0, then d(r) = d(u] + wji) = U]2 + w/ ~ (n/2)2 + (n/2)2 = n2/4 + n2/4 = n2/2 < n2 = den +
ia) ~ den) = d(x). . c'" •

Hence there exists t, r E J[i] such that y == tn +r ~tx + r with r ='(}o'r d(r) < d(x).
Case (2): Let a t:. x E J[i], y E J[i].

Write n = xx. Then n is a positive integer. Now yx E J[i], and n is a positive integer.
Therefore by case (1)~ there exists t, r E J[i] such that yx = tn+r with r = ° or d(r) < den). If r = 0,
then yx = tn = txx ~ y = tx = tx+O and so we get the conclusion.
Now we suppose that d(r) < den).

yx = tn + r ~ yx - tn = r => d(yx - tn) = d(r) < den)

~ d(yx - txx) < d(xx) (since n = xx)
~ d(y - tx).d(x) < d(x).d(x) (by the Note 12.4.1(iii»
IAcharya Nagarjuna University 12.13 Centre for Distance Educationl

=:} d(y .; tx) < d(x) (since d(x) is a positive integer).

Write rl = y- tx. Then d(rl) = d(y - tx) < d(x).
Finally, we got that y = tx + rl where d(rl) < d(x).
This completes the proof of the result.

12.4.3 Lemma: Suppose that p is a prime number and for some integer c relatively prime to p,
2
we can find integers x and y such that X2+ y2 = cp. Then there exists a and b such that p = a2 + b .
Proof: It is easy to verify that Z is a subring of J[i].
Part (1): In this part, we show that p is not a prime element of J[i].
-,,
In a contrary way suppose p is a prime element of J[i].
Since cp = X2+ y2 = (x + iy)(x - iy), by Lemma 12.3.5~p divides either x + iy or p divides x - iy
in J[i].
If P divides x + iy, then x + iy = p(u + iw) =:} x = pu and y = pw
=:} p divides (pu) - (pw)i = x- yi.
Therefore 0p2 divides (x + iy)(x - iy) = X2 + l = cp =:} p divides c, a contradiction to the
I

assumption that p and c are relatively prime.

In a similar way, if p divides x - iy, then we can get a contradiction.
This shows that p is not a prime element of J[i].

Part (2): Since p is not a prime element in J'[i], we have that p = (a + bi)(g+ di) for some non-
units a + bi, g + di E J[i].
Since a + bi and g + di are non-units, by Note 12.4.1, it follows that a2'+ b2 >1 and g2 + d2 > 1.
Since p = (a + bi)(g + di) is an integer, we can verify easily that p = (a - bi)(g - di). Thus p2 = p.p
= (a + bi)(g + di)(a - bi)(g - di) = (a2 + b2)(g2 + d2)
=:} a2 + b2 divides p2 =:} a2 + b2 = 1 or p or p2 (since p is a priIllY number).

We know that a2 + b2 i:- 1.

If a2 + b2 = p2 = (a2 + b2)(g2 + d2) =:} g2!- d2 = 1, a contradiction.

Hence a2 + b2 = p.
This completes the proof of the Lemma.
~~b~ra~ ~1~2~.1~4~ ~ =E=u~cl~id~e~a~n~R=in~g~sl

12.4.4 Lemma: If p is a prime number of the form 4n + 1, then we can solve the congruence
X2 == -1 modp.

P·roo f : L et x = 1. 2 . 3 ..... p-1

--2- -_ p-1
2 ,
.

p-l
Gi ven p = 4n + 1 ===> P - 1 = 4n ===> --- = 2n.
2

p-1
Hence x = --- ! = (2n)! is a product of even number of terms.
2 ""

Therefore we can write x = (-1) (-2) ... (_ p-1)

2

Also we know that (p - k) = (-k) (mod p).

So -1 = (p -1) (mod p), ,.2= (p - 2) (mod p), ....

2 p-1 p-1
Now x =x.x=[1.2 .... -][(-1)(-2) ... (--)]
2 .' 2

_ p-l p-1
= [1. 2.... -] [(p-I) (p-2) ... (p--)]
. 2 . 2

p-'-l p+1
= [1.2 .... -] [(p-1)(p-2) ... -]
2 2

p-1 p+1
= [1. 2 ..... -][- ... (p-2)·(p-1)] = (p-1)! = (-1)
2 2

[By a known theorem (namely, Willson's theorem), if p is prime numberthen

(p-l)! == - l(mod p)].

12.4.5 Theorem (Fermat's): If p is prime number of the form 4n + 1, then

p = a2 + b2 for some integers a, b.
jAcharya Nagarjuna University 12.15 Centre for Distance Educatio@

Proof: Part (1) : First show that there exists 0 < x ~ P -1 such that X2= -1 mod p.
By lemma 12.4.4, there exists y such that y2 = -1 (mod p).
By division algorithm, we get that y = ap + x with x =0 or 0 < x < p.
Ifx = 0, then y = ap = 0, and which implies 0 = y2 = -1 ~ 0 = -1 ~ 1 = 0 - (-1) is divisible by p,
a contradiction. So x =0 is not true.
Therefore 0 < x < p holds.
Now l = (ap + X)2= a2 p2 + 2apx + x2 ~ y2 - x2 is divisible by p ~ y2 = X2(mod p).
Hence X2 y2= = -1 (mod p).
Thereforethere exists 0 < x ~ P -1 such that X2==-1 (mod p).

Part (2): Now we wish to find a number s such that Isl < p/2 and.82 ==-1 mod p.
If [x] < pl2, then s = x will do.
Otherwise x > p/2 ~ -x < -(pf2). Write s = (p - x).
Now s = P- x < p - (p/2) = p/2.
Consider S2= (p _ x)(p - x) = p2_ 2px + X2~ S2_ X2= pep' - 2x)

~ S2- x2 is divisible by p ~ S2,==

x2 ==-1 (mod p).

Hence there exists s such that Isl < p/2 and S2= -1 (mod p).
Part (3): Now S2=-1 mod p ~,S2 +1 is divisible by p
~ S2+ 1 = cp for some integer c.

Now cp = S2+ 1 ~ (p/2i + 1 ='(p2 + 4)/4 < p2 ~'cp < p2

~ C < P ~ c, P are relatively prime.
Therefore by the Lemma 12.4.3, we have that p = a2 + b2 for some integers a and b.
This completes the proof of the theorem.

12.4.6 Problem: Find all the units in J[i].

Solution: Suppose that O:f. x = a + ib E J[i] is.a unit.
Now either 'a:f. 0 or b :f.0 (if both a e O and b = 0 hold, then x = 0, a contradiction),
Since x is a unit, there exists y = c + id E J[i]. such that xy'= 1.
(Algebra 12.16 Euclidean Ringsl

Now 1 + OJ = 1 = xy = (a+ibj Ic+id) = (ac - bd)+ i(ad+bc)

=> ad + be = 0 and ac - bd = ]

Solving these two equations, one can get that c = (a/(a2+b2» and d= (_b/(a2+b2».

Case(I): Suppose a = O. Then c = 0 and d = (_bJb2) =~IJb.

Since d is an integer we have b = ± 1.
Therefore in this case a + ib = 0 + l i = i or a + ib = 0 - I i = - 1.

Case(2): Suppose a = 1.
Then c = (l/(I+b2» is an integer => 1 + b2 =± 1
=> b2 = 0 ( as b is an integer the case b2 = -2 does not hold) => b = O(b is an integer).

In this case a+bi = 1+Oi = 1.

Case(3): Suppose a = -1.

Then c = (-1/1 +b2) is an integer => 1 + b2 = ± 1 => b2 = 0 => b = 0 (b is an integer).
In this case a+bi = -1+0i = -1.
Therefore the set of units of J[i] is {1,-1 ,i,-i}, which is same as
{a+ib / a.b e Z, a2+b2= I}.

12.4.7 Self Assessment Question: If 0 *- a+bi is not a unit of Jji], then a2+b2 >1.

4.5 SUMMARY

The abstract algebraic concepts like Euclidean ring, principal ideal ring, division, greatest
common divisor, unit, associates, prime element, relatively prime, were introduced. We
established that an Euclidean ring has a unit element. Every Euclidean ring is a principal ideal
ring. The relation of being associates is an equivalence relation. In an Euclidean ring any two
greatest common divisors of two given elements are associates. We proved the unique
factorization theorem. Every non - zero element in an Euclidean ring E. is either a unit jn Ror it
can be uniquely written (upto associates) as a product of prime elements. An ideal A::'=: (ao) of an
Euclidean ring R is a maximal ideal of R ¢::} ao is a prime element of R

Later, we defined the domain of Gaussian integers J[i]. We particularized the notion of a
Eluclidedan ring to a concrete ring, the ring of Gaussian integers. We proved that J[i] is an
!Acharya Nagarjuna University 12.17 Centre for Distance Educatio~

Euclidean ring. The odd prime numbers can be divided into two classes: those which have a
remainder of 1 on division by 4; and those which have a remainder of 3 on division by 4. We
showed that every prime number of the first kind can be written as the sum of two squares.

4.6 TECHNICAL TERMS

Euclidean ring: An integral domain R is said to be an Euclidean ring if for

every 0 # a E R, there defined a nonnegative integer d(a)
such that
(i) for all a, b E R both non zero, d(a) ::::;
d(ab)
(ii) for any a, b E R both non zero, there exists t, r E R such
=
that a tb + r where either r = 0 or d(r) < d(b).

Ideal generated by an element: Let R be a ring and a E R. Then the smal1est ideal containing
'a' is called the ideal generated by 'a'. It is denoted by (a).

Principal ideal ring: An integral domain R with unit element is called a principal
Ideal ring if for every ideal A of R there corresponds an
element a E A such that A = (a)

Greatest common divisor: Let R be a commutative ring and a, b, d E R.

Then d is said to be a greatest common divisor (gcd) of a and
b if:
(i) d I a and d I b (means d is a common divisor of a and d)
(ii) c I a, c I b ~ c I d.

unit: Let R be a commutative ring with identity (that is, unit

element). An element a E R is said to be a unit in R if there
corresponds an element b E R such that ab = 1.

Prime element: In an Euclidean ring R, a non-unit 1t is said to be a prime

element of R if whenever a, b E Rand 1t = ab, then either a
is a unit or b is a unit in R.

Relatively prime: Let R be an Euclidean ring, a, b E R. Then a and b are said

to be relatively prime if the gcd of a and b is a unit of R.

12.7 ANSWERS TO SELF ASSESSMENT QUESTIONS

12.2.15 : If a is a unit, then there exists b E R such that ab = 1.

Now d(a)::::;d(ab) = dt l ). Also dO)::::;d(1.a) = d(a). Hence d(a) = d(l).
~gebra 12.18 Euclidean Ringsl

Converse: Suppose d(a) = d(1). If a is not a unit, then by by Lemma 12.2.14, we have that
d(l) < d(1.a) = d(a), a contradiction. Hence a is a unit.

12.2.16: (i). Let a "*a E R.

Since 1 "*0, a= a-I and R is Euclidean ring, we have that d(a) = d(1.a) ;::::
d(1).
(ii). Suppose a and b are associates ee a = bu, for some invertible element u E R :=} b = au".
Now a = bu :=} deb) :::;d(bu) = d(a).

Also b = au" :=} d(a):::; dtau') = deb).

From the above facts, it follows that d(a) = deb).

12.3.9 Solution: Let c E B :=} C = rb for some r E R

:=} deb) :::;d(rb) = d(c) ~ deb) :::;min {d(c) / c E B} :::;deb) (since b e B)

=> deb) =min {d(c) / c E B}.

12.3.10: Suppose B = R. Then 1 E R = B = Rb :=} 1 = sb for some s E R.

This shows that b is a unit.
Converse: Suppose b is a unit.

Then there exists s E R such that sb = 1.

Clearly B = Rb ~ R. Now let r E R.
'Then r = r.1 = r(sb) = (rs)b E Rb = B.
Therefore R c B. Hence B = R.

12.4.7 : Since a + bi "*0 we have that a "*a or b "*A.

If a "*0, then a2+b2;:::: a2;:::: 1.
If b"* 0, then a2+b2;:::: b2 ~ 1.
Therefore in any case a2+b2;:::: 1.
If a2 +b2 = 1, then a+ib is a unit, a contradiction. Hence a2 +b2 > 1.
IAcharya Nagarjuna University 12.19 Centre for Distance Educatio~

12.8 MODEL QUESTIONS·

1. Let R be an Euclidean ring. Then show that that any two elements a, b in R have a
greatest common divisor d. Moreover d = Aa + ub for some A, I-l E R.

2. State and prove Unique factorization Theorem.

12.9 REFERENCE·BOOKS

1. Gopalakrishnan N.S. 'University Algebra', Wiley Eastern Limited, New Delhi,

1995.

2. Herstein LN. 'Topics in Algebra', Wiley Eastern Limited, New Delhi, 1988.

Dr. Bhavanari Satyanarayana

LESSON 13
POLYNOMIAL RINGS
Objectives

The objectives of this lesson are to:

• . Define polynomials; equality, addition, multiplication, degree, and irreduciability ef
polynomials; and the ring of polynomials F[x] over a field F. I
• Prove F[x] is an integral domain
• Understand and prove the division algorithm for polynomials .
• Apply the division algorithm to solve some problems and further theorems.
• Prove F[x] is a principal ideal ring
i
• Define primitive polynomial, content of a polynomial, integer monic polynomial
• State and prove Gauss lemma.

Structure
13.0 Introduction
13.1 Polynomial Rings
13.2 Irreducible Polynomials
13.3 Polynomials over the Rational Field
13.4 Summary
13.5 Technical Terms
13.6 Answers to Self Assessment Questions
13.7 Model Questions
13.8 Reference Books

13.0 INTRODUCTION

Consider expressions of the type X2 - 4x + 3 or X4 + (l/4)x3 - (l/2)x2 + (l/8). These are

called polynomial expressions. The first expression X2 - 4x + 3 is called as a polynomial with
integer coefficients; and the second expression
3 2
X4 + (l/4)x - (1/2)x + (1/8) is called as a polynomial with rational coefficients. We are familiar
with their properties like factorization, nature of roots, etc., In this lesson, we shall consider the
set R[x] of all polynomial expressions with coefficients from a given commutative ring R with
unit element. We shall define addition and multiplication on R[x], and show that R[x] forms a ring
with respect to these operations. This ring will be a Euclidean ring when R is a field. So we can
apply the results already obtained for Euclidean rings to this ring R[x] when R is a field, We state
and prove the division algorithm in the ring R[x]. Later, we define the concepts: primitive
polynomial, content of a polynomial and integer monic polynomial. We state and prove Gauss
lemma.
/Algebra 13.2 Polynomial Ringsl

13.1 POL YNOMIAL RINGS

13.1.1 Definition: Let F be a field, and x an indeterminate.

Write F[x] = {ao + a.x + ... + a.x" / n is a positive integer, a, E F, 1 ~ i ~ n}. Each element
of F[x] is called a polynomial with coefficients from F (or polynomial over F).

13.1.2 Note: Let F be a field and x an indeterminate.

(i) If p(x) = ao + a.x + ... + am xm, q(x) = bo + b.x + ... + b.x" E F[x], then
, , we define the equality
of polynomials as follows:
p(x) = q(x) ~ aj = b, for all i ~ O.

(ii) We define the addition of two polynomials p(x) and q(x) as follows:
p(x) + q(x) = Co+ CjX+ ... + Ctx'
where each c, = a, + b, for 1 ~ i ~ t, and t = max {m, n}.
Th~n (F[x], +) is an Abelian group.
If we define O(x) =0, then O(x) is the additive identity (or zero element) and - p(x) = (-
ao) + (-at) x + ... + (-am) x'" is the additive inverse of p(x).
(iii) Now we define the binary operation product on F[x]. For any two polynomials p(x) and q(x),
we define the product as follows:
t

p(x).q(x) = Co+ CtX+ ... + Ctxt wher2!ct = L. a Ct-k) .b, = £lobt+ajbt-I+ ...+atbo.
k=O

Then (F[x], .) is a semi group with identity 1 = 1 + Ox + Ox2 + ....

(iv) By a routine verification, we can understand that the distributive laws hold good. Therefore
(F[x], +, .) is a ring. This ring is called the ring of polynomials in the indeterminate x over the
o
given field F.

13.1.3 Self Assessment Question: Consider the ring R = {a + b N/a, b E Z}, two

polynomials f(x) = X2+ (1 + N )x + 2 + 3 Nand

g(x) = x3 - (3 + 7 r-s )x + (2 - 3 r-s )over R. Calculate (f + g)(x) and (fg)(x).

IAcharya Nagarjuna University 13.3 Centre for Distance Educationl

13.1.4 Definition: (i) If f(x) = ao + a.x + ... + an x" and an ::f; 0, then we say that the degree of f(x)
is n. We write it as deg f = n. Some times we say that an is the highest coefficient of f(x) ..
(ii) If g(x) = CE F, then degree of g(x) is O. A polynomial with zero degree is called a constant
polynomial. /
(iii) A polynomial f(x) = ao + a.x + ... + an x" is said to be monic if an = 1.

13.1.5 Note: It is easy to verify that deg.(p(x)+q(x» = max {deg p(x), deg q(x)} and
deg (p(x).q(x» = deg p(x) + deg q(x).

13.1.6 Definition: Suppose f(x), g(x) E F[x] with g(x) ::f; O. We say that g(x) divides f(x), written

as g(x) I f(x), if f(x) = a(x)g(x) for some a(x) E F[x).

13.1.7 Definition: For any two polynomials f(x) and g(x) E F[x] (not both zero), the polynomial
d(x) E F[x] is the greatest common divisor of f(x), g(x) if d(x) is a monic polynomial such that
(i). d(x) I f(x) and d(x) I g(x); and
(ii). If hex) I f(x) and hex) I gfx), 'then hex) I d(x).

13.1.8 Lemma: If f(x), g(x) are, two non-zero polynomials in F[x], then
deg (f(x).g(x» = deg (I(x) + deg (g(x».

Proof: Suppose f(x) = ao + a.x + ... + am x'", g(x) = bo + QIX+ ... + bnxn E F[x] with am ::f; 0 and
. ; ~. ,

bn::f; O.

Then degff(x) = m, and deg(g(x» = n.

Now f(xj-gjx) = Co+ ClX+ ... + cm+nxm+n. and Cm+n = ambn.
Since F is a field and am ::f; 0 ::f; b., we have that ambn::f; O.
, , ,i

So Cm+n= ambn::f; 0, it follows that deg(f(x)'g(x» = m + n = degif'(x) + deg(g(x».

13.1.9 Corollary: If f(x), g(x) are non-zero elements in F]x], then

degtf'(x) .s;deg(f(x).g(x»
Proof: By the Lemma 13.1.8, we have that'

deg(f(x)·g(x» = degtf(x) + degtgtx) ~ degjf(xj).

IAlgebra 13.4 Polynomial Ringsl

13.1.10 Corollary: The polynomial ring F[x] <ver the field F is an integral domain.
Proof : (The proof of this corollary is similar to the proof of the Lemma 13.1.8, but for
completeness, we provide the proof for this corollary).
Suppose f(x) = ao + a.x + + am x'", g(x) ; bo + b.x + ... + bn x" E F[x] with am i: 0 and b, :r=l1:

Now f(x)·g(x) = Co+ CIX+ + Cm+nxm+n and Cm+n= ambn.

Since F is a field and ami: 0 i: b., it follows that ambn·:;tO.
So Cm+n= ambn:;t0 => f(x)'g(x):;t 0.
/
Hence F[x] is anintegral domain.

13.1.11 Note: Let F be a field. By the Corollary 13.1.10, it follows that F[x] is an integral domain.
Let F* be the field of quotients of F[x]. This F* is called the field of rational functions in x over
F.

13.1.12 Lemma (I)ivisionalgori~hm): Given two polynomials f(x) and g(x) i: 0 in F[x]. Then

there exists two polynomials teA) and rex) in F[x] such that f(x) = t(x)'g(x) + rex)
where rex) = ° or degfrtx) < deg(gfxj).

Proof: If f(x) = 0, then we write t(x) = 0, rex) = O. Thenthe conclusion f(x) = t(x)·g(x)+ rex) IS

clear.
If deg(ftx) < deg(g(x», then write t(~) = 0 and rex) = f(x).
·N
In this case also, the conclusion f(x) =_.t(x)·g(x) + rex) is clear since
degmx) = ~eg(f(x) < deg(g(x».
Suppose m = deg f(x) ~ degg(x) = nand
, .'X,· : . .

m
f(x) = ao + alX + ... + amx , g(x) = bo + b.x + ... + b.x".
Induction Hypothesis: Now we suppose the result for all the polynomials of degree (m-I)
Write fleX) = f(x) -:-:(ambn-l)xm-n.g(x)
_.' ; '".-"7 ~ ~( ,

= (ao + alx + + amxm)- (ambn-l)xm-n)(bo + b1x + + b.x")

= (ao + alX.+ + amxm)- [((ambn-l)xm-n)(bo+ b.x + + bn_Ixn-1)] - amxm

= ao + alX + ... + am-Ixm-I - [ amb n,I Xm-n (b 0 + b IX+ ... + b n-\Xn-l)].

Hence deg(f1(x» ~ m-I.
 '~;.\ ---..,......-,. ---

jAcharya Nagarjuna University 13.5 Centre for Distance Edu,o";Qri!

By induction hypothesis, there exists t,(x) and nx) such that f,(x) ~ t,(x).g(x) + r(x) where'nx) ~ a
or deg rex) < deg g(x).
Now f(x) - (ambn-')xm-n.g(x)= flex) = t,(x).g(x) + rex) .

o =} = (am bn-' xm-n+ t,(x» g(x) + rex).

f(x)
Also we have that rex) = 0 or deg rex) < deg g(:'\
This completes the proof of the theorem.

13.1.13 Theorem: If I :f. (0) is an ideal of F[x], then there exist gee) E F[x] such that I =
{f(x)·g(x) / f(x) E F[x]}; that is, I consists of all multiples of the fixed polynomial g(x) by the
elements of F[x].

Proof: First we need to produce the fixed polynomial g(x). Note that I is a nonzero ideal of F[x].
Take a nonzero element hex) in I. If hex) is a non-constant polynomial, .then hex) is a polynomial
of degree not equal to zero. If hex) is a constant polynomial, then (h(x».x is a polynomial 01
degree one and it is in I. Hence there exist elements in I having degree not equal to zero;
So we can select a polynomial g(x) i:- 0 in I of minimal degree.
Suppose g(x) :j:: 0 in I, and deg t(x) ~ deg g(x) if 0 i:- t(x) E I.

Let O:j:: t(x) E I. Now by the division algorithm (lemma 13.1.12),

t(x) = q(x)g(x) + rex), where rex) = 0 or deg rex) < deg g(x).
Since g(x) E I, and I is an ideal of F[x], it follows that q(x)g(x) E I.
By assumption, t(x) E I, and so t(x) - q(x)g(x) E I =} rex) ~-t(x) - qtx) g(x) E I.

Since g(x) has minimal degr~e among the elements of I, and rex) E I, it follows that deg rex)
cannot be less than deg g(x). Thus we can conclude that .rtx) = 0, which implies that t(x) = q(x)
g(x).

Now we proved that every element in I is a multiple of g(x).

On the other hand, since g(x) E I, and I is an ideal of F[x], we get that f(x) g(x) E I for all f(x) E

F[x]. Therefore 1= {f(x)g(x) / f(x) E F[x]}.

13.1.14 Theorem: F[x] is an Euclidean ring.

Proof: By the Corollary 13.1.10, we have that F[x] is an integral domain.
/Algebra 13.6 Polynomial Rings/

For any O:f: f(x) E F[x] , define d(f(x» = deg (f(x». Then d(f(x» ~ O.

Now (i) d(f(x» ~ 0 for all 0 :f: f(x) E F[x].

(ii) d(f(x» ::;d(f(x).g(x» for all 0 :f: g(x) E F[x] (by the Corollary 13.1.9).
(iii) Suppose f(x), g(x) E F[x] such that g(x) :f: O.
Then by the division algorithm there exists two polynomials t(x) and rex) in F[x] such that f(x) =

t(x)·g(x) + r(x) where rex) = 0 or d(r(x» < d(g(x».

Hence we can conclude that F[x] is an Euclidean ring.

13.1.15 Lemma: F[x] is a principal ideal ring.

Proof: By the Theorem 13.1.14, we have that F[x] is an Euclidean ring.
'Now by the Corollary 12.2.2 (of Lesson 12), we have that F[x] is a principal ideal ring.

13.1.16 Lemma: Given two polynomials f(x), g(x) E F[x]. Then they have a greatest common

divisor d(x) = A(x).f(x) + /--l(x).g(x)for some A(X), /--lex)E F[x].

Proof: By the Theorem 13.1.14, F[x] is an Euclidean ring.

Now by the lemma 12.2.6 (of Lesson 12), the g.c.dof f(x), g(x) is of the form
d(x) = A(x).f(x) + /l(x).g(x) for some A(X),/lex) E F[x].

13.2 IRREDUCIBLE POLYNOMIALS

13.2.1 Definition: An element ~(~) E F[x] is said to be irreducible over F if it satisfies the
following condition: a(x), b(x) E Flx] and p{x) = a(x)·b(x)

=} degtatx) = 0 or degtbtx) = o.

13.202 Example: (i) If F = R, where R is the set of real numbers, then X2 + 1 is an irreducible
polynomial over F = R.
Verification: Suppose X2 + 1 = (x - a)(x - b) =} x =a orx = b is a root of X2+ 1
=} a2 + 1 = 0 or b2 + 1 = 0 =} a2 = -lor b2 = -1. This cannot be true for any real numbers a, b
R. This shows that X2+ 1is irreducible over R.
(ii) If F = C, the set of complex numbers, then X2+ 1= (x + i)(x - i).
IAcharya Nagarjuna University 13.7 Centre for Distance Educationl

Also degtx + i) = 1 :f. 0 and deg(x - i) = 1 :f. O. Theref~re x2 + 1 is not an irreducible polynomial
over C.

13.2.3 Theorem: If p(x) E F[x), then the ideal generated by p(x), that is, (p(x) in F[x) is a

maximal ideal of F]x] <=> p(x) is irreducible in F[x).

Proof: First we prove that if p(x) is irreducible in F[x], then the ideal M = (ptx) is a maximal
ideal of F[x).
For this, suppose that p(x) is irreducible in F]x], and N is an ideal of F[x] such that N => M.
By Lemma 13.1.15, N = (ftx) for some f(x) E F[x).

Now p(x) E MeN:=} p(x) = a(x)·f(x) for some a(x) in F[x].

Since p(x) is irreducible in F[x), it follows that either a(x) is a constant or f(x) is a constant.
Case (i): Suppose that a(x) is a constant. Then a(x) =aE F and so p(x) = af(x) which implies
tnat fix)= a-I ptx) E M.
Therefore N = (ftx) eM. and hence N =M.
Case (ii): Suppose that f(x) is a constant. Then f(x) =bE F, and so 1 = b-1b E N (since N is an
ideal of F(x]).
Thus g(x).l E N for all g(x) E F[x].
This shows that N = F[x]. Hence M is a maximal ideal of F[x].
Converse: Suppose M = (ptx) is a maximal ideal of F[x).
Now we have to prove that p(x) is irreducible.
In a contrary way, suppose that p(x) is not irreducible.
Then p(x) = a(x)·b(x), where degratx) ~ 1, degibtx) ~ 1.

Write N = (a(x», the ideal generated by a(x).

Then p(x) = a(x)·b(x) E N, that is p(x) E N. Therefore MeN.
Since degtatx) ~ 1, N = (a(x» :f. F]x], since every element in (atx) has degree at least that of
a(x).
Now by the maximality of M, we conclude that M = N.
But a(x) E N
,
= M, implies that a(x) = f(x)·p(x).
1f\
rAlgebra
t_. --'_"'-- --:.- -=-.~
138 -",-,.'(,'---"a1
,_'/':':..:'.'1., RI'ngs)
_ I

Now p(x) = a(x)·b(x) = f(x)·p(x).b(x) and so by cancellation law, we get that

1 = b(x)·f(x).
It is clear that deg(1) = 0< degtbrx) ::5:deg(b(x)·f(x» = deg(1) = 0, a contradictio:i.
Thus p(x) is an irreducible polynomial.

13.2.4 Self Assessment Question: (i). If ptx) is an irreducible element in FIx]. then show that
p(x) is either a unit or a prime element in the Euclidean domain F[x].
(ii) Observe that every prime element is an ireducible element.

13.2.5 Lemma: Any polynomial F[x] C an be written in a unique manner as a product of

irreducible polynomials in F[x].

Proof: LeI a(x) E F[x]. If a(x) is a unit, then it is an irreducible element, and so in this case, the
conclusion is clear.
If a(x) i. not a unit, then by theorem (Unique factorization theorem) 12.3.7 (or Chapter 12), a(x)
can be written as a product of prime elements in a unique way. Now by Ii:<' Self Assessment
Question 13.2.4, we can conclude that a(x) can be written as a product of irreducible elements in a
unique manner.

13.3 POLYNOMIALS OVER' THE RATIONAL FIELD

13.3.1 Definitions: (i) If f(x) =ao + alx +; .. +.an x" where 30, aI, ... , an are integers, then the ,"
g.c.d of 30, aI, ... '; a, is called the content of f(x}.
(ii) A polynomial f(x) = 30 + aix + ." + an x" where 30, a., ... , an are integers is said to be
primitive if the g.c.d of 30, a., ... , an is 1. In other words, a polynomial f(x) == ao + a.x + ... + an
x" where 30, a., ... , an are integers is said to be primitive if the content of f(x) ,is 1. •!

(iii) A polynomial f(x) = 30 + aix + ... + an x" is said to be integer monic if ao, ai, ... , an are
integers arid/an = 1. Some times we say that lln is the highest
, '
coefficient of f(~).
IAcharya Nagarjuna University 13.9 Centre for Distance Education]

13.3.2 Note: A straight forward verification shows that the product of two monic polynomials is

always monic.

13.3.3 Lemma.: If f(x) and g(x) are primitive polynomials, then f(x)·g(x) is also a primitive
polynomial.

Proof: Let f(x) = a, + a.x + ... + an x", and g(x) =: bo + b.x + ... + bm x" are two primj~ive
polynomials.

In a contrary way, suppose that the coefficients of f(x)·g(x) would be divisible by an integer k with
k> 1.
Let p be a prime number that divides k.

Then p divides all the coefficients co, ci, ... , C"+mof

f(x).g(x) = Co + CIX + ... + Cn+mx"+m.

Since f(x) is primitive, p can not divide all ai's.(1 $; i $; n).

Let t (1 $; t $; n) be the least positive integer such that p does not divide at.

Similarly since g(x) is primitive, there exists a least positive integer s (1 ~ s ~ m) such that p does

not divide b., '

Nowp divides a, for 1$; i $; t - 1and p divides bj for 1$; j $; s -1.

Consider the known relation

=
Ct+s at+s bo + a(t+s-l)b, + ... + a.b, + at-Ibs+1 + ...+ aobt+s ..... (1).
For 1 $; i ~ t - 1, we have that p divides a., and so p divides ai~(t+S)-~11I
" -~. " • J ' .'.

For 1 $;j $; s - 1, we have that p divides bj,'and sop divides a(l+s)_jbj.

Therefore p divides all the terms on the right side sum of (1) except atbs.

Since p divides Ct+~,we have that p divides a.b., which. implies that either p divides at or p divides

bs (since p.is prime), a contradiction.

o
Hence f(x)~g(x) is a primitive polynomial. .

This completes the proof the Lemma.

13.3.4 Theorem (Gauss Lemma) If the primitive polynomial f(x) can be factored as the
., , ., .

product of two polynomials having rational coefficients, then it can be factored as the product of

two polynomials having integer coefficients.

iAl-g-eb-r-a----------~----------------13-.-10------------------~P-o-ly-n-o-m-i-al~-

Proof: Suppose that f(x) = u(x)·v(x) where u(x) and vex) have rational coefficients.

By clearing of denominators we can write f(x) = .!.. (A1(X).J..I\x» where b is an integer, and both
b .
,,,l(X), /-I\X) have integer coefficients;

If al is the g.c.d of the coefficients of ,,,l(x), and a2 is the g.c.d. of the coefficients of /-Il(X),then

).}(x) = al.A(x) and I..I\x) = a2./-I(x) where. both A(X) and /-I(x) have integer coefficients and are
primitive.
Write a = a,a2.
Then f(x) = .!.. (Al(X)'/-I1(x» = .!.. ala2(A(x)/-I(X» 9 ~ A(X)/l(X)
b b b
=> b.f(x) = a.A(x)/l(X).
o
Since f(x) is primitive, the g.c.d of its coefficients is 1, and so the content (that is the g.c.d. of the
coefficients) of b.I(x) is equal to b.
Since A(X)and /lex) are primitive, by the Lemma 13.33, A(X)./l(x) is also primitive.

Hence the content of A(X)./l(x) is 1 => the content a(A(x)./l(x» is equal to 'a'.
~
.

Therefore a = b. This shows that f(x) = ~b. A(X)/-I(X)= A(X)/-I(x).

t"

Ience f(x) = A(X)/l(X)where both A(X)and /lex) are having integer coefficients.

13.3.5 Corollary: If an integer monic polynomial factored as the product of two non-constant
"
polynomials having rational coefficients, then it can be factored as the product of two integer
fI'lQ)nicpolynomials.

proof: Let f(x) be an integer monic polynomial.

Gi vcn that f(x) can be factored as the product of two polynomials having rational coefficients.
Th~p.by the theorem 13.3.4, the polynomial f(x) can be written as f(x) = A(X)/-I(X)where A(X) and
IAt:.::) have integer coefficients.
I/y'<& know that if a, b are highest coefficients of A(X) and /-I(x) respectively, then ab is the highest

~o€fficient of f(x).
8i]1)ce f(x) is monic, its highest coefficient is l.
IAcharya Nagarjuna University 13.11 Centre for Distance Educatio~

Hence ab = 1. This implies that a = 1 = b, or a = -1 = b.

. : .
If a = 1 = b, then f(x) = A(X»)l.(X)where A(x),)l.(x) are both integer monic polynomials.
If a = -1 = b, then f(x) = (-.A(x»(-)l.(x» = .A(x)J...l(x)where A(X), !lex) are both integer monic
polynomials.

13.3.6 Problem: Prove that x2 + x + 1 is an irreducible polynomial over the field of the integers
modulo 2.
Solution : If X2+ X + 1 is not irreducible, then x2 + x + 1 = a(x)b(x) for some polynomials of
degree lover Z2, the field of integers modulo 2...
The only polynomials of degree lover Z2 are x and x+l ,
Case (i): Suppose a(x) = x. Then x.b(x) = x2 + X+ 1
=> x divides X2+ x + 1, a contradiction.
Therefore a(x) :j:. x. So this case does not arise.
Similarly we can show that b(x) :j:. x.

Case (ii): Suppose a(x) = x+l, and b(x) = x+I.

Now X2+ X + 1 = a(x)b(x) = (x+l)(x+l) = X2+ X + x +1 = X2+ 2~ +1 = X2+1 (since 1+1 = 0 in Z
2).

So x2 +x + 1 = X2+ 1 => x = 0, a contradiction. This case does not arise.

Hence we can conclude that X2 + x + 1 is an irreducible polynomial over Z2 .

13.3.7 Self Assessment Question: Prove that the polynomial x3 - 9 is an irreducible polynomial
over the field Z31 = to, 1, ..., 30} of integers modulo 31.

13.4 SUMMARY

The abstract algebraic concepts and operations like polynomials, equality of polynomials,
multiplication, degree, irreduciability and the ring of polynomials are introduced. We proved that
F[x], the ring of polynomials over a field F, is an integral domain. We deduced the division
algorithm. F[x] is a Euclidean ring. F[x] is also a principal ideal ring. The ideal A = (p(x) in
F[x] is a maximal ideal ¢=> p(x) is irreducible over F.
In the next part of this lesson, we introduced some concepts.like primitive polynomial, content of a
polynomial, integer monic polynomial. We proved that the product of two primitive polynomials
is also a primitive. Gauss lemma was proved.
jAlgebra 13.12 Polynomial Ringsl

13.5 TECHNICAL TERMS

Polynomial: Let F be a field. x an indeterminate. Write

F[x] = {a, + a,x + ... + a.x" / n is a positive integer, a, E

F; 1 :::;i :::;n }. Each element of F[x] is called a polynomial
with coefficients from F.

Constant polynomial: A polynomial with zero degree is called a constant

polynomial.

Division Algorithm: Given two polynomials f(x) and g(x) '* a in F[x], there exists

two polynomials t(x) and rex) in F[x] such that f(x) = t(x)
g(x) + rex) where rex) :=;: 0 or deg rex) < deg g(x).

In-educible polynomial: A polynomial p(x) E F[x] is irreducible if p(x) is of positive

degree and given any polynomial f(x) in F[x], then either

i,
p(x) I f(x) or ptxj.is relatively prime to f(x).

Primitive polynomial: A polynomial f(x) = £10 + a.x + .,. + an x" where an, aI, .. -, an
are integers is said to be primitive if the g.c.d of £10, a., _-.,
an is 1.

Content: If f(x) = ao + a.x + _._+ an ·xn where ao, ai, .... an are integers,
then the g.c.d of ao, a., ... , an is called the content of f(x).

"

Monic polynomial: A polynomial.ftx) = ao + a.x + "" + an x" is said to be monic

if an = 1.
. .

Integer monic: A polynomial f(x) =

ao + a.x + .__+ an x" is said to be integer
monic if ao, a., .. "' an are integers .~'
and an 1. =
lAcharya N<l:g_~:!:.iuna
University 13.13 C_en_t_re_£_o_r
Distance EducatIOTIj

13.6 ANSVVERS TO SELl' ASSESSME:";T QUESTIONS

13.1.3: (i) (f + g)(x)

= {x'+(1 +N )x+2+3Nl +f x: (3+7N)x+(2-3.J-S)}

= x; + X2 - (2 + 6 N )x + 4.

(ii). (fg)(x):= {x2 + (1 + N )x + 2 + 3.J - 5) ,<.3 - (3 + 7 r-s )x + (2 - 3 r-s )i

:= x5 + (1 + ..J-5 )x4 - (1 + 4..J-5) + (34 - 13..J-5 )x2
~
+ (116 - 2",J-5)x + 49.

13.2.4 (i) .i.ppose that p(x) is an irreducible e. -rnent in F[x] which is not a unit
We have ~" .now that p(x) is a prime element j. F]x]. For this suppose that
p(x) = a()c, b(x), Since p(x) is irreducible, we nave that either degiatx) = OJr deg(b(x) = O.
This implirs that either a(x) is a constant or b.c) is a constant, and so either a(x) is a unit or bex)
is a unit. ihis shows that every irreducible element in F[x] is either a unit or a prime element.

(ii). Fro"! the definitions of prirne element and irreducible element, it is clear that every prime
element i.:
an irreducible element.

13.3.7: "a contrary way, suppose that the polynomial x3 - 9 is not an irreducible polynomial
over the ,··ld Z31 = to, 1, ..., 30}.
Then x3 '.,) = f(x).g(x) for some polynomials [(x) and g(x) over Z31

Now either f(x) or g(x) is of degree 1.

Suppose deg (ftx) = 1.
Since f(x) = x - a is a factor of x3 - 9, the polynomial x3 - 9 has a root in Z31.

So for some 0 < x < 31, we have x3 - 9= 0 ~ x3 == 9 (mod 31) for some 0 < x < 31.
Since 31 is a prime number, the number of integers lies between 0 and 31, which are relatively
': •. 0

prime to 31 is 30.

/
 !Algebra 13.14 Polynomial Ringsl

= 30 (Euler's function <I>

So <1>(31) is used here). (Recall the statement of the Euler's theorem: If a
is the number of non-negative integers <
is relatively prime to n, then a$(n)== 1 (mod n) where <I>(n)
n that are relatively prime to n].
Now 31 is a prime number and 0 < x < 31 ~ x is relatively prime to 31
~ X$(31)== l(mod 31) ~ x30== 1 (mod 31).

Already we have x3 == 9 (mod 31) ~ x30;'"910(mod 31).

By transitive property we get 910 == 1 (mod 31) ~ 320== 1 (mod 31).

Since 3 is relatively prime to 31 we have 330== 1 (mod 31) (by Eulers theorem).
Now 320 == 1, 330 == 1 ~ 330 == 320 (mod 31) (by transitivity) ~ 31 1330- 320 ~ 31 1320 (310- 1) ~
31 1320or 311 (310- 1).
. Since 31 cannot divide 320 (otherwise 3'113 since 31 is prime), it follows that
31 divides (310- 1) = 58049 - 1 = 58048, a contradiction (since
58048 = (1904)(31) + 24).
This shows that x3 - 9 is irreducible over Z31.

13.7 MODEL QUESTIONS

1. Given two polynomials f(x) and g(x) ::f; 0 in F[x] where F is a field. Then prove that
there exists two polynomials t(x) and rex) in F[x] such that f(x) = t(x)·g(x) + rex)
where rex) = 0 or degrrtx) < deg(g(x».
2. Suppose that F is a field and p(x) E F[x]. Then the ideal generated by p(x), that is,
(p(x» in F[x] is a maximal ideal ofF[x] <=> p(x) is irreducible in F[x].
3. Define the term 'primitive polynomial'. If f(x) and g(x) are primitive polynomials,
then show that f(x)'g(x) is also a primitive polynomial.
4. State and prove Gauss Lemma.
5. Prove that X2+ x + 1 is an irreducible polynomial over the field of integers modulo 2.
6. Prove that the polynomial x3 - 9 is. an irreducible polynomial over the field Z31of
integers modulo 31.

13.8 REFERENCE BOOKS

1. Gopalakrishnan N.S. 'University Algebra', Wiley Eastern Limited, New Delhi, 1995 .
. 2. Herstein LN. 'Topics in Algebra', Wiley Eastern Limited, New Delhi, 1988.

Dr .Bhavanari Satyanarayana
 LESSON 14

POLYNOMIAL RINGS OVER COMMUTATIVE RINGS

Objectives

The objectives of this lesson are to:

• State and prove the Eisenstein criterion
• Apply the Eisenstein criterion to find the irreducibility of a given polynomial.
• Define the concepts like: polynomial ring R[XI.X2 xn] over a given ring R in n-variables,
field of rational functions, unique factorization domain.
• Find the influence of the structure of R on that of R[xJ. X2 xn]
• Apply the Gauss lemma to the ring R[x], where R is a unique factorization domain
• Prove some basic theorems and lemmas

Structure
14.0 Introduction
14.1 The Eisenstein Criterion principle
14.2 Polynomial Rings over Commutative Rings
14.3 Summary
14.4 Technical Terms'
14.5 Answers to Self Assessment Questions
'14.6 Model Questions
14.7 Reference Books

14.0 INTRODUCTION

We continue the study of polynomials. We state and prove Eisenstein criterion. We use
the Eisenstein criterion in- verifying whether a given polynomial is irreducible. We define R[x],
the polynomial ring in x over R; R[xJ, X2, xn], the ring of polynomials in the n -variables XI,
X2, ",Xn over R. We study the influence of the structure of R on that of R[XI, X2, xn]. We
define unique factorization domain. If R is an unique factorization domain, then so is R[x]. Then,
we are able extend this to R[xJ, X2, xn] by using mathematical induction. Also we prove that
if F is a field, then F[Xl, X2, xn] is a unique factorization domain.

14.1 THE EISENSTEIN CRITERION PRINCIPLE

Now we state and prove an important well known theorem.

 r-------
lAlgebra ------_._ ... _-----------
14.2 . Polynomial Rings ~.::::

14.1.1 Theorem (The Eisen .tein Criterion principle): Let I(x) = ao + a.x +...+ a.x" he
polynomial with integer coefficients. Suppose that some prime number p divides a, for a :s:; i~::(;:
1); p does not divide an; and l does not divide ao. Then f(x) jc an irreducible polynomial over \:

field of rational numbers.

Proof: Without loss of generality we may assume that f(x) is primitive [ If not, we take out L··,;

g.c.d. of its coefficients. It does not disturb the hypothesis, bee .l1l0C p does not divide an].

In a contrary way, we assume that f(x) is not irreducible.

Then f(x) = (bo + b.x + ... + b.x' )(co + CIX + ... + c.x' ) where hr * 0 * c.; and b, for O:S:;i 5 r : ,lei
Cj for 0 sj :::;
s are rational numbers.

Without loss of generality (Please refer Theorem 13.3.4(Gau:s Lemma of Lesson 13)), we .. </
assume that b,' sand c,' s are integers.

Since ao + alx + ... + a.x" = (bo + b.x + ... + b.x' )(CO+CIX+

,. ... + t ,xs
"
), it follows that
.

ao = boco and ak = bkCo+ bk-ICI + ... + bOCk.

Given that p I ao :::::>p I bo Co:::::>p I bo or p I Co.

I I
If P bo and p Co, then p21 boco :::::>p21 ao, a contradiction.

Therefore p divides only one of bo and co.

Case (i): Suppose p divides bo, and p does. not divide Co.

lf all the coefficients b., 0 :::;i :::;rare di visible by p, then all. (if:, 1 ;:~ k ::; n are divisible by p .::,:;,
',~k -.

a contradiction. Therefore p dqyp ppt divide all the h,' s.

Let m be the least number such that p does not divide bm•

This implies p divides bo, b.. ... , bCm-I).

Now am = bmco + bm-I'~l + ... + bocm :::::>bmco'= al)l- (bm-1Co+,'" + bacm),.. ._..

Since p divides am, bm-I, bm-2, ... , bo, it follows that p I b., (note that p does not divide co), a

contradiction to the fact that 'p does not divide bm'.

Hence f(x) is irreducible.

C51se hi)':s~}~<p divides Coand p does not divide boo '. . '.: ~.'.

The p;oof iSi~i~i'~~.the proof in case(i) .. This co~pletes the proof of the Theorem:

. 6.1,~.,pr.~f.e~" If p~rime number, then prove that the polynomial ~"- pis irreducible over
, the flel,g. of rqt~al num~~~.~ '1 .

< '\ 'I f < ' .:' "

IAcharya Nagarjuna University 14.3 Centre for Distance Educationl

Solution : Suppose x" - p ~ ao + a,lx + .:. + an x", where a, = 0 for 1 ::;i ::;(n-l) and an= 1, ao = p.
Now P di vjc/es 3; for 0 ~ i'S n:- 1; P do not divide an; and p" does not divide ao·
By applying the Theorem 14.1.1 (Eisenstein Criterion), we can conclude that the polynomial xn_
p is irreducible over the field of rational numbers .
. I • •

14.1.3 Problem: Prove that the polynomial 1 + x + ... + x(p-l) where p is a prime number, is
irreducible over the field of rational numbers.

Solution: Here we consider the polynomial 1 + (x + 1) + (x + 1)2 + ... + (x + 1)p-1and use

Eisenstein Criterion.
We know that 1 + (x + 1) + (x + 1)2 + ... + (x + 1)p-1
= 1 + (x + 1) + (x2 + 2x + 1) + ... + (x + 1)(p-1)
= P + [1 + 2 + ... + (p -1)]x + [1 + 3 + ... + (p _2)]x2 + ... + X(p-I)
=P + [(p -1) P /2]x + ... + X(p-I)
= ao + a.x + ... + ap_Ix(P-I)
where ao = p, a, = pep - 1) /2, ... a(p-l)= 1.
Clearly pi (10, p I ai, ... and p does not divides ap-l. Also p2 does not divide ao·
Hence by Theorem 14.1. 1(Eisenstein Criterion), conclude that
1 + (x + 1) + (x + 1)2 + ... + (x + I)P-1 is irreducible over the field of rational numbers.
Hence 1 + x + ... + X(p-l)where p is a prime number, is 'irreducible over the field of rational

numbers.

14.2 POLYNOMIAL RINGS OVER COMMUTATIVE RINGS

14.2.1 Note: Let R be a commutative ring with unit element 1.
(i). Write R[x] = {ao + a.x + ... + amxm / a, E R, 0::; i ::;m} is called the set of all polynomials in
one determinant x over R.
The equality, addition (+), and multiplication (.) are same as in polynomials over fields. Then it is
easy to verify that (R[x], +, .) is a commutative ring with unit element.
(ii) Let xi, X2,... , x, are nvariables.
IAlgebra 14.4 Polynomial Rings .... 1

Write R, = Rlx.], R2 = R1[X2],R3 = R2[X3], ... ,Rn = Rn-l[xn].

Then R, is denoted by Rlx, X2,... , xn], and it is called the ring of polynomials in n variables xi,
X2,... , x, over R.
We can understand that

terms. and each a. I"

. E R}.
~ •.• ;.

Here equality and addition are defined coefficient wise; and the multiplication is defined by use of
the distributive law and the rule of exponents

14.2.2 Lemma: (i) If R is a commutative ring with identity, then so is R[x].

(ii) If R is an integral domain, then so is R[x].

Solution: (i) we know that R[x] is a ring.

Now lex) = l+Ox + --Ox" E R[x] is the identity in R]x],
If f(x) = an + a.x + +anxn, and g(x) = bo + b.x .+... + bmxm,then .by tile definition of product in
i
R]x], we have that f(x).g(x) = Co+ c.x + .. :+Cm+nxm+nwhere c, = I,a I bi-t, and g(x)f(x) = do +dlx
1=0

. •• dm+n h
Xm+n were d·-
I -
~b
L.J I
a·',
i-t- fin r ,
1=0

i i
Now d, = I. b, ai-t = I. a
1=0 1=0
I bi-I (since R is commutative) = c, for each
.
l~ i s m+~ ..

So f(x).g(x) = Co+ C1X+ ... +Cm+nxm+n = do +djx+ ... dm+n xm+n = g(x)f(x).
Hence f(x)g(x) = g(x)f(x). This is true for all f(x), g(x) in R[x].
Therefore R[x] is a commutative ring with identity.
(ii). Let 0 '* f(x), 0,* g(x) E R[x].

Then f(x) = an + aix '+ ... + ~m x'", am * 0; g(x) = bo + b.x + ... + bnx
n
, bn * 0; and all the
coefficients .ai, 1 ~ i sm and bj, 1 ~j ~ n are in R. . 'f.~
Now f(x).g(x) = (ao + alX + ... + amxm)(bo+ b.x + ... + b.x")
 /Acharya NagarjunaUniversity 14.5 Centre for Distance Educationl

= aobo + (aObl + a1bo)x + ... + ambnxm+n.

Since am :f:. 0 and bn :f:. 0 and R is an integral domain, we have that ambn:f:.O.

Hence f(x)g(x) :f:. O. This shows that R[x] is an integral domain.

-, 14.2.3 Corollary: (i) If R is a commutati ve ring with identity, then

so is R[Xl,X2, ... , xn].

(ii) If R is an integral domain, then so is R[Xl,X2, ... , xnl

Proof: (i) We prove this by induction. If n = 1, then by Lemma 14.2.2(i), we have that Rlx.] is a

commutative ring with identity 1.

Suppose the induction hypothesis forn - L

This means Rn-I = Rjx., X2, ... , xn-d is a commutative ring with identity element 1.

Now R, = Rn-l[Xn] is 'a commutative ring with identity 1 (by using the Lemma 14.2.2(i) again).

Hence R[Xj,X2, ... , xn] = Rn is a commutative ring with identity 1.

(ii) We prove this part also by induction. If n = L then by Lemma 14.2.2(ii), we have that Rjxj] is
an integral domain (since R is an integral domain). -:

Suppose the induction hypothesis for n - L·

This means Rn-j = R[XI, X2, ., xn-ll is an integral domain when ever the ring R is an integral

domain.
Now R, = Rn-I [xnJ is an integral domain (by using the Lemma J4.2.2(ii) again).

Hence R[XI,X2, ... , xnJ = R, is an integral domain.

14.2.4 Definition: If F is a field, then F[Xl,X2, ... , xn] is an integral domain. Let F(Xl,X2, ... , x,

be the field of quotients of F[xJ,xz, ... , xnl Then F(XJ,X2, ... , xn) is called the field of rationa
functions in xi, X2, ... , Xnover F.

14.2.5 Definition: Let R be an integral domain with unit element I -, Then

(i) an element x E R is said to be a unit in R if there exists y E R such that xy = 1.

(ii) two elements a, b are said to be associates if a = ub for some unit u in R:

/Algebra 14.6 , Polynomial Rings .... 1

(iii) an element a E R which is a non-unit is called irreducible (or a prime element) if whenever a

= be with b, c E R, then either b is unit in R or c is unit in R.

14.2.6 Definition: An integral domain R with unit element is called a unique factorization
domain if it satisfies the following two conditions

(i) 0 * a E R => a is a unit, or a = PI P2 .. pn where Pi E R, 1 :::;i:::;'n are

irreducible elements; and
(ii) if a = PI P2 .. pn = ql q2 ... qm where each Pi and qi are irreducible elements of
R, then n = m and for each i (1 :::;i :::;n) there corresponds j (1 :::;j :::;m) such

that Pi and 'Ii are associates.

14.2.7 Lemma: If R is a unique factorization domain, and a, b e R, then a and b have a greatest
common divisor (a, b) in R. Moreover, if a and b are relatively prime (that is, (a, b) = 1),.and a I
be, then a I c.

Proof: Suppose a, b E R.
Since R is unique factorization domain, we can write a = PIP2 ... pn and b = qlq2 ... qm where Pi, 1
s i :::;
n, and 'Ii, 1 :::;j s m are irreducible elements.
Without loss of generality we may assume that n :::;m.
We prove this result by induction on n.
Suppose n = 1. Then a = PI.
If there exists j (1 :::;j :::;
m) such that PI is an associate of 'Ii, then PI is the g.c.d. of a and b.
If there is no j (1 :::;j :::;
m) such that PI is an associate of 'Ii, then the g.c.d of a and b is 1. Now
we proved that if n = 1, the g.c.d of a and b exists.
Suppose the induction hypothesis that the result is true for n = k -1.
Now we prove it for n = k.
S uppose a = PIP2 ... Pk. N ow a = PIP2 ... h
Pk = a IPk were a I = PIP2 ... Pk-I.
By the induction hypothesis, the g.c.d. of a' and b exists (say d).
If there is j (1 :::;j :::;
m) such that Pk is an associate of CJ.j. then d-p, is the' g.c.d of a and b. Hence

the g.c.d. of a and b exists.

 ---------_.- _._--- _.-.-'
·:\,.:harya Nagarjuna University 14.7 Centre for Distance Educati'--.~l

Second Part: Suppose a, b E R such that a and bare relatively prime and a I be.

Suppose a =PIP2 ... P» and b == q Iq2 .. : gill and c = r1r2 ... rs where Pi (l ::;; i ::;;
n),

m) and ~i (1 S is s) are irreducible elements.

Since a [be, there exists XE R such that be = ax.

Now be = ax== (PIP2 ... Pn)x :~ {QICj2 ... qm )(1'; [2 ... r, ).
Consider Pi (1::;; i :s:;n).

Since R is unique factorization domain, we have that Pi is an associate of either (]j or rk for some 1
:s-; j :s-; m or 1 :s-; k :s-; s.

If Pi is an associate of CJj, then Pi divides both a and b. a contradiction (since a, b are relatively
prime).
Therefore Pi is not an associate of qi for any j (1 ::;;j ::;;
m), Therefore Pi is an associate of rk. for
J

some 1 :s-;kj :s-;s.

Now a = PIP2 ... Pn divides r, ' rk

1 2
' ••• rk II
' and rk J
ik
2
••• rk divides c ::::rl
0
f2 ... r.. Hence a I c..

14.2.8 Corollary: If,aE R is an irreducible element, and a I be then a I b or a I c.

Proof: If a I b, then the result is true.
Suppose a does not divide b.
Since a is irreducible, and a does not divide b, itfollows that the g.c.d.of a and.b is 1.
Now by Lemma 14.2.7, we get that a I c. , He

14.2.9 Definitions: Suppose R is a unique factorization domain; and

f(x) = ao + ajx + ... + amxmE R[x].
(i) The content of f(x) is defined to be the g.c.d. ofa-, a., ... , am.
We denote the content of f(x) by c(f).
(ii) f(x) is said to be primitive if cif) = 1.
,
'"

14.2.10 Self Assessment Question: Let f(x) E R[x]. Then there exist a primitive , polynomial . fi(x)

E R[x] such that f(x) = a.f (x) where a = c(f).

I
IAlgeb ra 14.8 Polynomial Rings .... 1

14.2.11 Lemma: If R is a unique factorization domain. then the product of two primitive
polynomials in R[x] is a primitive polynomial in R[x].

Proof: The proof is parallel to the proof of the Lemma 13.3.3 (of lesson 13).

14.2.Q Corollary: If R is a unique factorization domain, and f(x), g(x) are in R]x], then c(fg)
= c(f)c(g) (up to units).

Proof: Given two polynomials f(x), g(x) E R[x].

Write f(x) = afl(x), and ~(x..)= bgr(x) where a = c(f), b = c(g) and fleX), gl(X) are primitive
polynomials.
Now by Lemma '13.2.1 1, we get that fl,(x).g)(x) is also a primitive polynomial.
Hence c(fj(x).gj(x)) = 1.
This shows that c(f(x)g(x») = c(a~.fj(x)gl(X» = abc(f\(x)gt(x» = ab.l = ab =
c( f( x»·c(g(x».

14.2.13 Self Assessment Ques,ti9P= If R is aunique factorization domain and Ux) E R[x], 1

~i ~k, then c(f\f2 ••• fk) = c(f\).q(f2) ••• c(fd.

14.2.14 Notation: Henceforth, we consider R to be a unique factorization domain. Since it is an

integral. domain by.Theorem p!'~t,R can be embedded in a field F (called the field of
quotients of R).

(i) If we take a polynomial f(X)ERlx], then the coefficients of f(x) are from R.
Since R c F. we may consider the coefficients of f(x) are fromP.
Hence f(x) E F[x]. Therefore R[x] c F[x].

OJ). ~[xl, FDcl are rings and R[Aj ~ F[x] ~ Rlx] is a subring of F[x].

14.2.15 Self Assessment Question: If f(x) E F[x], then show that f(x)
. .
= .!..a fo(x) where fo(x) E

R]x] and aE R.
IAcharya Nagarjuna University 14.9 Centre for Distance Educatio~

14.2.16 Lemma: Let f(x) E R[x] be a primitive polynomial. Then f(x) is irreducible in R[x] {::}
f(x) is irreducible in F[x].

Proof: Suppose that f(x) E R[x] is irreducible.

We have to show that f(x) is irreducible in F[x].
-.
In a contrary way, suppose that f(x) is not irreducible inF[x].
Then f(x) = g(x)h(x) where g(x), hex) E F[x] and these are non-units (equivalently, degrgtx) > 1
and degthrx) > 1).

Now by the Self Assessment Question 14.2.15, g(x) = .!. .go(x) and hex) = .!. ·ho(x)
a b
where go(x), ho(x) E R[x] and a, b E R.

Since go(x), hoex) E R[x], it follows that go(x) = ag](x) and ho(x) = /3h](x) where a = c(go) and
/3 = c(ho), and hj(x), gj(x) are primitive polynomials in R[x] (by the Self Assessment Question
14.2.10).
Note that degth, (x) = deg(ho(x», and deg(gl(x» = deg(go(x».
Therefore f(x) = g(x)h(x) = 1:. go(x}! ho(x) = 1:. agl (x)- ! /3h](x) = a/3 g, (xjh, (x)
I . a b a b, ab

=> abf(x) = a~gj(x) hj(x) (i).

Since g](x) and h\(x) are primitive we have that gj(x)hj(x) is also primitive.
So the content of g\(x)h1(x) = 1
=> the content of a/3g](x)h\(x) = a/3·(content of gl(x)h\(x» =·a13.1 = a/3.
Since f(x) is primitive in R]x], we have the content of f(x) is 1 and so the content of abfrx) = ab.
Hence by using (i), one can conclude that ab ~ a~, which implies f(x) = gj(x)hj(x) where gj(x),
h.tx) E R[x], a contradiction (Since f(x) is irreducible in R[x]).
Hence f(x) is irreducible in F[x].
Converse: Suppose f(x) is irreducible in F[x].
In a contrary way, suppose that f(x) is not irreducible in R(x).
This implies f(x) = f](x)f2(x) and deg(fj(x» > 1, deg(f2(x» > 1,
where fleX), f2(x) E R[x].

!7
jAlgebra 14.10 Polynomial'Rings . .'.\.1

Since R[x] C F[x], we have that f(x),= fj(x)f2(x) and deg(f1(x» 6. 1, deg(f2(x» > 1, where. fj(~),
f2(x) E F[x), a contradiction to the fact that f(x) is irreducible in Flx]. "
Hence f(x) is irreducible in R[x].
This completes the proof of the Lemma. .. , '

14.2.17 Lemma: If R is a unique factorization domain, and p(x) is primitive polynomial in R[x],
then it can be factored in a unique way as the product of irredticible'eI~men'is'in R[x]. ,

Proof: Consider p(x) as an element of F[x]. ,\

By the Lemma 13'.2.5 (of lesson 13), we get thacptx) = Pj(X).P2(X)..... Pk(X) where Pi(X) is an
irreducible polynomial in F[x] for L;S; i ~ k.
Now for 1 ~ i ~k (by using the Self Assessment Question '14.2.'15),; .we get that,' '.piCx)~=
',f, -: ;

Moreover, fi(x) = Ciqi(X) where Ci= c(O, and qi(X) E R[xjis primitive forI- ~,i~·k (by using
the Self Assessment Question 14.2.10). .. ;,
Now Pi(x) = (l/ai).fi(X) = (c/aD.qi(x).,.·" !'i it.

Therefore p(x) = PI(X)P2(X) Pk(X)= [(cj/a)ql(x)] [(C2/a2)q2(X)] [(Ck/ak)qk(X)] ': "J

= [(CjC2... ck)/(aj a2 ... ak)] qj(X)q2(X) ','.' qk(X) ::;" ,i

~ (aja2 ... ak)p(x) = (CjC2... Ck)(qj(X)q2(X)... qk(X».. ' '"

Since p(x), qlx)" 1,~ i ~ k are primitive,
,
,we have that the content
-
of LHS. is,
.

the content of RHS is c) C2... Ck.

Therefore (aja2 ... ak) = (CIC2... Ck),which implies that p(x) = qj(X)q2(X) ... Qk(X).
Hence p(x) can be factored as the product of irreducible polynomials. ,qj(x),,J :5.i,~ kin R[~J. 'i

Uniqueness: Suppose p(x) = qj(X)q2(X) ... qk(X) = fl(x)r2(x) ... rn(x) where each qi(X) and each
rix) are irreducible in R[x].
Since p(x) is primitive, we have eachq.(x) and each rj(x) are also primitive. " '." ,< . J

Since these are irreducible in R[x], by Lemma 14.2.16, these are irreducible in F[x]. Ther~fQre
ptx) = Ql(X)Q2(X).. , Qk(X)= rj(x)r2(x) rn(x) (i)

is a factorization in F[x].
 !Acharya Nagarjuna University 14.11 Centre for Distance Education'

Since F is a field, by Theorem 13.1.'14 (of Lesson 13), we get that F[x] is an Euclidean domain,

.and by Lemma 13.2.5 (of Lesson 13), the factorization given in (i) is unique.

This means that k::;: m and for a given 1 ~ i ~ k, there exists 1 ~ j ~ k such that q.ix) is an
associate of rlx). This completes the proof.

',"

14.2.18 Self Assessment . Question:

'.
Suppose that R is an unique factorization
. ", "
domain.
If a E: R is an irreducible element of R, then the constant polynomial defined by at x) ::;:a is
irreducible in R]x].

14.2.19 Theorem: If R is a unique factorization domain, then so is R[x].

Proof: Let f(X)E R[x]. Bythe Self .Assessment Question 14.2.10, we get that

f(x)::;: c.f.tx) where c::;: ctf) and f1(x) E R[x] is primitive.

~';By the'Lemma.Ia.LTe.we.can deeompose fux) uniquely as fl(x)::;: Pl(X)P2(X) ... Pk(X) where each
Pi(X) E R[x] is irreducible in R]x].

Since c E Rand R is a unique factorization domain. we get that c can be uniquely written as

c ::;:1Wl:2 .•. n;n where each l1j:e:: R'is ~l prime element.

Hence ftx) can be uniquely represented as

f(x)::;: c.fux) ::;:1[\112 ... 1[111 .p.: X)P2(X),.; P,JX,l

'~'iwhere each l1i (l~ ism) lsapdifte' elernenr of Rand eacli' Pi(X)' is an irreducible polynomial in

R[x].
~'. ; .: /'.'

14.2.20. Self'Assessment Question:' If Rts a unique factorization" domain then so is R, = Rlx, .x-.
xn,],~
;. _____ .

14.2.21 Self Assessment Question: U· F is a field. then F[XI,Xz x!)] is' a unique factorization

domain,
 ..
iA1!!ebra 14.12 Polynomial Ringy

14.3 SUMMARY

We stated and proved the Eisenstein criterion. We used the Eisenstein criterion to find out whether
a given polynomial is irreducible. Some abstract algebraic concepts like the ring of polynomials
in one indeterminant x over a given ring R, the ring of polynomials in the n variables over R, field
of rational functions, associates, prime element, unique factorization domain, content, primitive
are introduced. We proved that if R is an integral domain (respectively, commutative ring with
identity 1), then so is R[x], and we also extended this to R[xI, X2, x.]. If R is a unique
factorization domain, then so is R[x], and we also extended it to R[Xl, X2, xn]. If F is a field,
then F[Xl, X2, x.] is a unique factorization domain.

14.4 TECHNICAL TERMS

Field of rational functions: If F is a field. then F[Xl,X2, ... , xn] is an integral domain. Let
F(Xl,X2, ... , xn) be the field of quotients of F[Xl,X2, ... , x.].
Then F(Xl,X2, .. ~,xn) is called the field of rational
functions in xi, X2,... , x, over F.

Unit element: Let R be an integral domain with unit element 1. Then an

element x E R is said to be a unit in R if there corresponds
an element y E R such that xy = 1.

Associates: Two elements a, b are said to be associates if a = ub for

some unit u in R.

----Irreducibleelement: An element a E R which is not a unit is called irreducible (or

a prime element) if whenever a = be with b, c E R, then
either b or c is unit in R.
\

Unique factorization domain: An integral domain R with unit element is called a unique
factorization domain if
"*
(i) for 0 a E R =} a is a unit. or a = PI P2 .. Pll where Pi E R,
1 ~ i ~ n are irreducible elements; and (ii) if a = PI P2 .. PIl =
ql q2 ... qmwhere each Pi and qi are irreducible elements of
R, then i1 = m and for each i (1 ~ i ~ n) there-corresponds

j (1 ~ j ~ m) such that Pi and CJj are associates.

Content: Suppose R is a unique factorization domain.

Let f(x) = ao + a.x + ... + amxmE R[x].
The content of f(x) is defined to be a g.c.d. of ao, a., ... , am·

Primitive: Suppose R is a unique factorization domain. A polynomial

f(x) over R is said to be.primitive if c(i) = 1.
 !Acharya Nagarjuna University 14.13 Centre for Distance Educatio~

14.5 ANSWERS TO SELF ASSESSMENT QUESTIONS

14.2.10: Suppose f(x) = ao + a.x + ... + am x'".
Since c(f) is a g.c.d. of aa, aI, ... , am, we have that
a, = c(f).bi for some b, E R for O:s; i :s; m, and the g.c.d. of bo, b., ... , bm is 1.
Now f(x) = ao + ajX + ... + amxm
= c(f).bo + c(f).blx + ... + c(f).bmxm
= c(f)[bo+ b.x + ... + bmxm]
= a.fjfx) where a = c(f) and fl(x) = bo + b.x +...+ bmxm.

14.2.13: We prove this result by using the principle of mathematical induction on k.

Ifk = 2, then the result follows from the Corollary 14.2.12.
Now suppose the induction hypothesis that the result is true for k = n-l.
. Now consider c(fjf2 .. fn) = c«(fjf2 fn-1)(fn»
= c(fjf2 fn_I).c(fn) (by the Corollary 14.2.12)
= c(fj).c(f2) ... ·c(fn_j)·c(fn) (by the induction hypothesis).
This shows that c(f!f2 ... fk) = c(fj).c(f2) ... c(fk).
This completes the proof of the Corollary.

14.2.15: Suppose f(x) = ao + ajx + .,. + a.x" E F[x].

We know that F = {p / p, q E Rand q :f. O}.

q

Therefore a, =~Pi for some pi, qi E R for 1 :s; i :s; n.

qj

Now f(x) = £.9..+ £Lx + ... + fuxn

qo ql qn
1
=----

= .!. fo(x) where a = qag! ... gn E Rand

a
n
fo(x) = POqlg2... gn + PIgOg2... gn'x + ... + Pngoqj··· gn_Ix E R[x].
This completes the proof.
 [Algebra 14.14 Polynomial Rings .... 1

14.2.18: Suppose that a(x) = p(x)q(x) for two polynomials p(x), q(x) over R. .
,;..;
'; l, t ;.~
..:_ t; . ")I~ !i.,:', ;~. 'iJ.,,'/!~;~ :..~.'f., / >~

Now 0 = deg(a(x)) = deg(p(x)r +:deg(q(x)) ~ deg(p(x))· d 0 = deg(q(x)) ~ p(x) and q(x) are
constant polynomials.
If p(x) = p and q(x) = q, then a = a(x) = p(x)q(x) = pq, a contradiction to the fact that a is
irreducible in R. Hence a(x) E R[x] is also irreducible.

14.2.20: We prove this corollary by mathematical induction on n.

Ifn =1, then by the theorem 14.2.19, we. get that Rjxj] is an unique factorization domain:. "
Suppose the induction hypothesis that if R is an unique factorization domain, then Rn-1is also an:
unique factorization domain '.
" Now by the theorem 14.2.19, we get that R[Xl,X2 ,xnl=f,.~ncr[x~Ji~an uniquefactorization
, ..domain " i

14,2.21: Since F is a field, it is an integral domain.

• f - -~: :-

Since every non zero element of F is a unit, we have that F is a unique factorization domain.
Therefore by the Self Assessment Question 14.2.20, we get that F[Xl,X2..... xn] IS. a unique
factorization domain.

14.6 lYlODEL QUESTIONS· ; j

';
1. State and Prove Eisenstein criterion principle.
.:, .

2. Define the concepts 'primitive polynomial' and 'irreducible polynomial'.

I

;. t
If f(x) E R[x] be a primitive polynomial, then prove that f(x) is irreducible in j •

R[x] ¢::} f(x) is irreducible in F[x].

t'

14.7 REFERENCE BOOKS' iI'

'- •• >

1. Gopalakrishnan N.S. 'Uni~~rsity Algebra', Wil:~y Eastern Limited, New Delhi, 19~5..
2. Herstein LN. 'Topics in Algebra', Wiley EasternLimited, New Delhi, 1988.
"-J

Dr.Bhavanari Satyanarayana
 LESSON~
,<' .• ' • t
15 FIELD EXTENSIONS
.," " .•...

CONTENTS:
-'OBJECTIVES
STRUCTURE
15.1 Introduction
15.2 Preliminaries
15.3 E,xtension fields ': (.
15.4 Finite extensions
TS.5· Summa~~ .'. .. .L
15.6 Technical terms
'15.7 'Model Examination' Questions
15.8 Answers to Self Assessment Questions
15.9 Exercises'
15.10 Reference BooliS
..; ,I;
' .
OBJECTIVES
# ~ \ .. , . ~ . _ '''i .'. '-; ,i : : , '!
. ,
._ "
j\ttel' reading this unit you should be able to know about
• Irreducible pol) :l()i11i;l!~;

• Eisensteins criterion for irreducibility of a polynomial over rationals.

• Another criterion for irreducibility of a polynomial over rationals.
• Field extensions. , tori,

• Finite Extensions.
, ..." :;

15.1 INTRODUCTiON
Field extensions playa central role in field theory. Usually given an algebraic
structure we seek for an extension of it ( an algebraic structure, which contains the given
algebraic structure as a sub-structure ) possessing certain desired properties, which are
not satisfied by the original one. In ring theory we have seen that, if ,R is a ring, not
necessarily containing identity element then (l ring R l' with identity cap. be constructed
such thafR., contairisa subring , R' : whi~h is isomorphic to R. This i~ an instance of
extension of an algebraic structure ( Here R 'isextended to R 1, or, R is imbedded in R 1)
. In the same way we have seen that an integral domain can be extended to a field. In this
section we study the extension<?f fields, particularly finite extensions and algebraic
~:Jo. " •• " •• 1 >; ~
••• j... . ••..: ':"5:;,. .1'. '.- i,:
 Distance Fr+ucatlo n 2 fo~~C Acharya Nagarjuna University

extensions. This section forms a basis for the discussions that follow in this unit and also
ill units 16 and 17.

I 5.2 : PRELIMINAR[ES
In this section we briefly outline some results about polynomial rings, particularly about
irreducible polynomials. We do not 'give proofs as there are elaborately discussed in
lessons relating to Ring Theory.
Let F be a field. By the ring of polynomials, or, simply' the polynomial ring in the
indeterminate , x, over F, we mean the set of all symbols Xo+ al x + ..... + an x'' , where n
is any nonnegative integer and where the coefficients a, , a2 , ... , an are in F and i('1S
denoted by F[x]. We say that p(x) = ao + a.x + ... + am x'" and q(x) ~ bo + b.x + .:... + b, x''
in F[x] are equal if and only if a, = b, for i ~ O.
The addition + and multiplication. are defined on F[ x] as follows. For p(x) = ao /
+ al x + .... am x 111 an d q (x)
x = b 0 + b 1 X + .... + b J1 XJ1. ,
2 t
f(x) + g(x) = Co+ CIX+ C2 X + ..... + Ct X ,wbere
\
Ci= ai + b, for each i.
2 R
f(x) . g(x) = Co + CIX + C2X + ..... + CR X ,

where c, = .aj bo + ai-I b, + aObi for each i.

One can easily verify that ( F[x] , +, . ) is a ring with identity.

Definition 15.2.1: lff(x) = ao + a.x + .... + an x'' * ° and an * ° then n is defined as the
degree off(x) and it is denoted by deg f(x).
We say a polynomial is a constant if its degree is O. This degree function defined on
nonzero elements of F[x] will provide us with the d-function needed in order to make
F[x] a Euclidean ring. In ring theory we have seen that

Theorem 15.2.2 : F[x] is a Euclidean ring with d(f) = deg f(x) for f(x) E F[x] .

Theorem 15.2.3: F[x] is a principle ideal ring.

Theorem 15.2.4: A polynomial p(x) in F[x] is said to be irreducible over F if whenever-
P(x) = a(x) b(x) ,
With a(x) , b(x) E F[x] then one of a(x) or b(x) has degree 0, or, a constant.
If f (x) is not irreducible over F, then it is called reducible over F.
Irreducibility depends on the field. For example, the polynomial X2+I , treated as
polynomial over the field c of complex numbers it is not irreducible , since X2+ 1 = ( x+i)
(x-i) , where i2= -1. .
In ring theory we have seen that
Theorem 15.2.5: Any .polynornial in F[x] can be written in a unique manner as a
product of irreducible polynomials in F[x].
Theorem 15.2.6: The principal ideal A = ( p(x) ) = { p(x) q(x) I q(x) E F[x] } generated
by p(x) is a maximal ideal if and only if p(x) is irreducible over F.
An immediate consequence of the above theorem is that F[x] I (p(x)) is .a field if
p(x) is an irreducible polynomial.
Definition 15.2.7: Let E and F be fields such that F c E. Let f(x) E'F[x] . for aEE
, f(a) E E, is called the value of f(x) at a. if f(a) = a , then a is called a root
off(x). In this case we also say that a satisfies f(x), or, f(x) is satisfied by a.
POLYNOMIALS OVER RATIONAL NUMBER FIELD:
In our discussion, we are mostly concerned with polynomials and their
irreducibility, over the field of rational numbers. Most of the time the coefficients will
actually be integers.
Definition 1~.2.8: Let f(x) = 3{) + alx + a2 X2 + + an x" , where 3{) , al , .., an are
integers. The greatest common divisor of ( ao , a., , an ) of ao , a., ... , an is called the
'" atent off(x). If the content f(x) is 1, that is, if (ao , a., ... , an) = 1, then f(x) is called
a primitive polynomial.
Definition 15.2.9: Let f(x) = ao + alx , ... , + an x" , where ao , ai, ... , an are integers. If
the coefficients of the highest degree term is 1, that is if an = 1 , then f(x) is called a monic
polynomial.
Evidently every monic polynomial is priliminary.
In this connection we have two results which have practical applications in our
discussion. We state them without proof.
Theorem ( Gauss Lemma) 15.2.10: Let f(x) be a primitive polynomial with integer
coefficients. Then f(x) is reducible over Q if and 0I11y if f(x) is reducible over Z.
Theorem 15.2.11 : ( Eisenstein's Criterion): Let f(x) = ao + a, x + ... ,+ anx" E Z[x] ,
n z 1 . If P isa prime such that p divides 3{), ai, a2, ..... an-Iand p does not divide an and
p2 does not divide aD, then f(x) is irreducible over Q, the rationals.
As an illustration of the above results let us consider the following example.
Example "15.2.12: Show that the polynomial f(x) = X4 + 2x + 2 is irreducible over the
field of rational numbers.
Solution: Here ao = 2 , al = 2 , a2 = a , a3 = a and ~ = 1. The prime number 2 is such
that 2 divides ao, a, , a2 , a3 but 2 does hot divide a, , namely, 1 and 22 = 4 does not
divide 3{) , namely, 2 , So by Eisenstein's Criterion, they given polynomial is irreducible
over the field of rational numbers.
Eisenstein'S criterion can be applied to a polynomial f(x) E Z[x] only when it
coefficients satisfy the condition of theorem 15.2.11 for some prime number p and hence
 Distance Education 4

it can not be applied for all polynomials in Z[x]. We present below a result which -gives
another criterion for the irreducibility of a polynomial in Z[x] over Q.
Theorem 15.2.13: Let f(x) = an + a, x + .... an-l xn-1 + x" be a monic polynomial in Z[x]
If f(x) has a root a in Q, then a E Z and a divides an .
Let us illustrate this by means of an example.
Example 15.2.14: Show that x3 - x -1 E Z[x] is irreducible over Q.
Solution: Suppose that f(x) = x3 - x-I is reducible over Q. Then f(x) is a product of
two non - constant polynomials in Q[x]. Since deg f(x) = 3 , one of the factors of f(x)
must be linear or f(x) has a root say a in Q, Since f(x) is a monic, by Gauss' lemma ~ E
Z.
But by theorem 15.2.13, a must divide the constant term off(x). That is, a divides
1. So a = ± 1.
But neither I nor - 1 is a root off(x). So f(x) must be irreducible over Q.

15.3 EXTENSION FIELDS

Definitions 15.3.1 :
Let F be a field. A field K is said to be an extension of F, if there is a subfield of
F of K, which is isomorphic to F. '.
K F

Definition 15.3.2 :
If K is an extension of F" we also say that F is imbedded, or , embedded in K.
When two algebraic structures are isomorphic we know that they have the. same
structures possibly differing in the labeling of the elements. So we can identify
isomorphic algebraic structures and also the corresponding elements under the
isomorphism.
Thus if the field K is an extension of F, and if F is the subfield of K which is
isomorphic to F u~er the isomorphism c , then we identify F with F and the element a
E F with o(a) in F and treat F itself as a subfield of K. Alternatively, we say that K is
an extension of F if F is a sub-field of K. Throughout our discussion block F denotes a
field and K an extension of F.
Example 15.3.3 : Trivially every field F is a subfield of itself, so that F can be viewed as an
extension of itself.
Example 15.3.4: The field C of complex numbers is an extension of the field of real
numbers R.
Solution : Consider R, the field of real numbers, C the field of complex numbers, Then
C = { ( a, b ) I a, b E R } ,
and the addition and multiplication on C are given by
( a, b ) + ( c, d ) = ( a+c, b+d )
( a, b ) . ( c, d ) = ( ac - bd , ad + be ) .
The mapping c : R ~ C given cr(a) = ( a, 0) is an isomorphism of R into C and the sub-
field
R={(a,O)laE R}
'of C is isomorphic to R under cr. So C is an extension of R.
Example 15.3.5 : Let Q be the field of rational numbers and
Q( fi) = {a + fib I a, bE Q}. Then Q( fi) is an extension ofQ.
Solution: It is easy to see that Q( .fi) is a field under usual addition and multiplication.
The map ~ : Q ~ Q( 12) given by ~(a) = a + 12 0 is easily seen to be an isomorphism
of Q into Q( .J2) . SO Q( .J2) is an extension of F.
-
Self Assessment Question 15.3.6 : Show that the set R = { ( a, 0 ) ! a E R } is a sub-
field of C.
Self Assessment Question 15.3.7 : Show that the mapping o : R -7 C given by cr(a) =
(a.O) is an isomorphism of R into C.

15.4 FINITE EXTENSIONS:

Suppose K is an extension of F so that F is a sub-field of K. In vector spaces we
have seen that every field can be viewed as vector space over a sub-field. So K is a
vector space over F. As a vector space we can talk of linear independence, basis;
dimension, and so on, in Kover F. HI
.-, -,

Definition 15.4.1: Let F be a field and K an extension of F. The dimension of K as a

vector space over F is called the degree of the extension Kover F and it is denoted by
[ K : F ]. Thus [ K: F] = dim- K.
If [ K : F ] is finite, that is, if K is a fmite dimensional vector space over F, then we
say that Kis a finite extension of F.
In our study we confine over attention to only finite extension of fields.
Example 15.4.2: Show that every field F is a finite extension of itself.
Solutiun : Let F be a field. Then F is an extension of itself. What is [ F : F ] ? For the
identity element 1 E F , a 1 = a , for all a E F. So {I} spans F over F. Also {I} is
linearly independent over F. (Why?) So the singleton set {I} in F is a basis of F over
F, so that dim rF =1 . Thus [ F : F] = land F is a finite extension of F. ,-
 ~----------
Distance Ed~u~ca~t~io;n--==~~~~~i. 6

Example 15.4.3: Consider the extension C of R. The sub set { 1, i} of C is a basis of C

over R. So dim RC = 2 and [ C : R] = 2.
Self Assessment Question 15.4.4 : Show that the subset { 1, i} of C is a basis of Cover R.
Self Assessment Question 15.4.5 : Show that Q( .fi) is a finite extension of Q. What is [
Q(.fi):Q]?

Example 15.4.6 : Suppose K is a finite field having q elements. Then characteristic of K

is finite and it is a prime number, say p. That is , p is the least positive integer such that
°
pa = for all a E K. If e is the identity element of K then F = { 0, e, 2e, ... , (p-l) e } is a
sub-field of K containing p elements. So K is an extension of F. Further K being finite,
it is finite dimensional over F, so that K is a finite extension of F. If [ K : F] = n then it is
see that q = pn .
Suppose that L is a fmite extension of K and K is a fmite extension of F. The
natural questions that arise are: Is L a finite extension of F? If so, what is [ L : F ] ? The
following theorem answers these questions. .
\
Theorem 15.4.7: If L is a fmite extension ofK and K a finite extension of F than 1, is a
finite extension ofF. Further [ L : F] = [L : K] [ K : F]
Proof: Suppose that [ L : K ] = m and [ K : F ] = n. The best way to go about is to
exhibit a basis of Lover F containing mnelernents. Now L is a m-dirnensional vector
space over K and K is a n-dimensional vector space over F. So let { II , h, .."., 1m} be a
basis of Lover K and { k, , k2 , .. '" k, } a basis of Kover F. Consider the elements l, kj ,
j ::::;
1 :::;i::::;m, 1 ::::; n in K. These are mn in number, which is equal to [ L : K ] [ K : F]. If
we show that the set { l, , kj } is a basis of Lover F then we are done. For this we have to
show that { l, k, } spans Lover F and it is linearly independent overF.
Now let 1 E.L. Then { h_.kh ' ... ",1m} being a basis of Lover K, I has a unique
representation as .tu.:
I =alII +a2h+ +an;Im 15.3.1
. . ,
for some ai, a2, .. "., am E K. Again since { k, ,12 , , k, } is a basis of Kover F,
each cq , 1 :::;i :::;m in K has a unique representation as
o..= ~il KI + ~j2 K2+ + ~il1 KI1 15.3.2
for some ~jj E F. Substituting the values of aj from 15.3.2 in 15.3.1 we get
1 = WII KI + ~12 K2+ + ~In KI1) 11+
W21 K, + ~22 K2 + + ~2 11K, ) h+ +
(~MI KJ + ~M2K2 + + ~mn KI1) 1m
L f3ij Ii ~.i' ,
1!S:is m
]!S:j!S:n
where ~ij E F. This shown that I is a linear combination of { Ii, kj } span Lover F.
We still need to show that the vectors I; ~, 1 :::;i :::;m , 1 :::;j :::;
n are linearly
independent over F. For this we have to show that if for (Xij E F,
"al·k·=O
L.J 1.1 1 .I 15.3.3
Isisrn
l:<;;j:<;;n

then (Xij = 0 for all i, j .

Let us rewrite expression 15.3.3 as
In

L(ailkj +ai2k2 + +ainkn)li = q 15.3.4

i=l

Now { k, , k2 , , k, } is a basis ofK over F, so that

ail k, + (Xi2 k2 + + (Xin k, E K for 1 :::;I :::;m. Since 11 , b , , 1m are linearly
independent over K, 15.3.4 implies. that
(Xiik, + (Xi2 k2+ + (Xink~ = 0 15.3.5
for 1:::; i :::;m. But k, , k2" , ... , k, are linearly independent over F, so that 15,3.5 implies
that (Xij = 0 for i :::;
j:::; ill , 1 :::;j :::;n . '
These establish that the vectors I, k.', 1 :::;I :::;m , 1 :::;j :::;
n in L form a basis of L
over F. So [ L : F] = mn = [ L : K [ K : F] .
Corollary 15.4.8:
If L is finite extension of F and K is a sub-field of L which contains F, then
[ K : F] I[ L :.F ].
Proof: Suppose [ L : F] isfinite. Then L is a finite dimensional vector space over F, so
that the maximum number of linearly independent elements in Lover F is [L :F] ,
Since K c L, any set of linearly independent lelements in J<
over F, is also a set of
linearly independent elements in Lover F, So this number is less than or equal to [L: F].
Thus [ K : F ] = dim- K = maximum number of linearly independent elements in Kover
F :::;[ L : F ] and this is finite, This shows that K is a finite extension of F.
Again F c K implies that any set of linearly independent elements in Lover K is
also linearly independent over F. So [ L : K] :::;[ L :i F ] , and thus [ L : K] is finite.
Thus L is a finite extension of K and K is a finite extension of F. By theorem
15.4.8, [ L : F] = [ L : K] [ K : F] , so that [ K : F] I [ L : F J .

15.5; SUMMARY:
In this lesson we gave a brief account of polynomial rings and Eisenstein's criterion for
irreducibility of a polynomial over-a field of rationals. We also defineg field extensions
:~c;e Education = ~~~~~~ 8 ;e~~-: Acharya Nagarjuna University

and finite extensions. Finally we have shown tha. if L is a finite extension of K and K is
a finite extension of F then L is a finite extension of F and [ L : F ] = [ L: K ] [ K : F ]

15.6 : TECHNICAL TERMS:

In this lesson we have become familiar with the following technical terms.
Polynomial ring
Degree of a polynomial
Content of a polynomial
Primitive polynomial
Monic polynomial
Irreducible polynomial
Root of a polynomial
Extension of a field
Finite extension
Degree of a finite extension

15.7: MODEL EXAMINATION QUESTIONS:

PART A: Short Answer Questions:
(1) Show that the polynomial X2 + 2x + 2 m Z[x] is irreducible over the
field of rationals.
(2) Show that the polynomial x3 - x-I in Z[ x] IS irreducible over the field
of rationals.
(3) Show that the field C? of complex numbers is an extension of the field R
of real numbers. Find [ C :'R ] .
(4) Show that Q(.J2) = [ { a + b -Ji I a, b E Q } is an extension of the field Q of
rational numbers. Find [Q( .J2) : Q ].
PART B: Long Answer Questions
(I) If L is a finite extension of K and K is finite extension of F, show that K is a
finite extension of F and
[L:F]={L:K][K:F]

15. 8 : ANSWERS TO SELF ASSESSMENT QUESTIONS

Self Assessment Question 15.3.6: Show that R = { ( a, 0) I a E R } is a subfield of C.
Solution: Let ( a, 0 ) , ( b, 0 ) E R. Then a, b E Rand
( a, 0 ) - ( b, 0 ) = ( a - b , 0 )
 -
Since a, b E R ~ a- b E R we have ( a - b , 0 ) E R. Since a, b E R ~ a- b E

R we have ( a - b , 0 ) E R. -
Also for ( a, 0 ) , ( b, 0 ) E R, ( b, 0 ) "* ( 0 , 0 ). Then b "* 0 and
-I 1 1 -
( b, 0) = b2 (b, 0 ) = ( b '0) E R.

So
-I 1 a -
( a, 0 ) . ( b, 0) = ( a, 0) ( - , 0 ) = ( -, 0 ) E R.
b b
These show that R is a subfield of C.
Self Assessment Question 15.3.7: Show that the mapping c : R -7 C given by cr(a) =
(a, 0) is an isomorphism of R into C.
Solution: For a, b E R we have
c(a) = (a, 0) and cr(b) = (b, 0). So
o(a+b) = ( a+b, 0) = ( a, 0) + (b,O ) = o(a) + o( b) and
cr(ab) = (ab, 0) = (a, 0) (b,O) = er(a) cr( b)
Hence o isa field homomorphism. It is easy to see that c is onto.

Self Assessment Question 15.4.4: Show that the subset { 1, i} of C is a basis of C

over R.
Solution: Let a + ib E C. Then a, b E Rand
a + ib = a. 1 + b. i .
So { 1, i} spans Cover R.
'Further for a, b E R , if a 1 + b i= 0 , then a = b = 0 , so that { 1, i} is linearly independent
over R. So { 1, i} is a basis ofC over R.
Self Assessment Question 15.4.5: Show that Q( Ji) is a finite extension of Q . What is
[Q( Ji) : Q] ?
Solution: For a E Q , we have a = a + 0.J2 . So Q c Q( .J2) and thus Q is a sub field
ofQ( -fi).
It is easy to see that the subset { 1, Ji } in Q( Ji) is a basis of Q( Ji) over Q. SO
[Q( Ji) : Q] = 2

15.9: EXERCISES:
1. Prove that the polynomial 2X4 - 3x2 + 6x + 12 is irreducible over the field of rational
numbers.
2. Prove that the following polynomials are irreducible over the field of rational numbers.
(a) 8x3 - 6x-1
. - (b) x3 - 2
(c) x3 + X2 - 2x - 1.
3. Show that Q( 12) = { a +b 12 I a, b E Q } is a field under usual addition and
multiplication.
4. Show that { 1, -Ii } in Q( 12) is a basis of Q( 12) over Q and [Q( ../2): Q] = 2.

15.10: REFERENCE BOOKS:

1. ARTIN, EMIL, Galois Theory, 2nd ed Notne Dame, Ind : University of Name
Dame Press, 1955.
2. FRALEIGH, lB., A First Course in Abstract Algebra. 2nd ed., Addision Wesley
Publishing Company.
3. HERSTERIN, IN., Topics in AlgebraVikas.Publishing House.
4. JACOBSON, N., Lectures in abstract algebra. Vo1.lI, Nostnand, 1953.
5. BHATTACHARYA, P.B., JAIN. S.K., NAGPAUL.S.R., Basic Algebra,
(2nd Edition) , Cambridge University Press.

Lesson Writer
Prof L.Nagaml;ni Reddy
S.V.Universiry, Tirupati
 LESSON - 16 : ALGEBRAIC EXTENSIONS

CONTENTS
OBJECTIVES
STRUCTURE
16.1 : Introduction
16.2 Algebraic elements
16.3 Field got by adjoining an element to a field ..
16.4 Algebraic Extensions
16.5 Summary
16.6 Technical terms
16.7 ModeLExamination Questions
16.8 Answers to Self Assessment Questions
16.9 Exercises
16.10 Reference Books

Objectives:
After reading this lesson you should be able to :
• Define an algebraic element
• Obtain a criterion for an element to be algebraic over a field.
• Check whether a given element is algebraic over a field and find its degree when it
is algebraic
• Know about the algebraic extensions
• Establish that the set of algebraic elements in an extension field K of a field F
form a subfield of K.

16.1 INTRODUCTION:
There is a close connection between algebraic elements in a field F and its finite
extensions. Let F be a field and K an extension of F. An element a in K called algebraic
over F if there is a nonzero polynomial f(x) in F[x] such that f(a) = O. For a in K and
a not in F, we construct the sub field F(a) of K, called the field got by adjoining a to F. It
is our aim to establish that a in K is algebraic over F if and only if F(a) is a finite
extension of F.

,8
 Distance Education 2 F~==r:= Acharya Nagarjuna University

16.2 ALGEBRAIC ELEMENTS

Consider an extension K of a field F. Let f(x) = ao + alx + U2X2+ + an x"
. E F[x] , be the polynomial ring over F. For a E K, we call ao + ala + a2a2 + + an an
as the value of f(x) at x = a and denote it by f(a). We say that a satisfies f(x), or, a is a
root or zero off(x), iff(a) = 0 .
Definition 16.2.1 : An. element a in K is called an algebraic element over F if there is a
non-zero polynomial f(x) E F[x] such that a satisfies f(x). Further a is said to be
algebraic of degree n over. F if it satisfies non-zero polynomial of degree n over F but no
non-zero polynomial of degree less than n.

Definition 16.2.2: Let a in K be an algebraic element over F. The monic polynomial of

least degree in F[x], which is satisfied by a is called the minimal polynomial of a over F.
Example 16.2.3 : Consider the extension C of R and the element a = 1 + -J3 i in C. Now
a-I = -J3 i , or, ( a-I i
= -3. That is a2 - 2a + .4 = 0 . This shows that 1 + -J3 i
satisfies the non-zero polynomial X2 - 2x + 4 =0 , over R, so that 1 + -J3 i in C is
algebraic over R.
Example 16.2.4 : Show that ~1- 3.J2
in Q( .J2) is algebraic over Q.
Solution: I - 3 .J2 satisfies the polynomial x4 - 2x2 - 2, so that a is algebraic over Q.
Further this polynomial is irreducible over Q ( by Eisenstein's Criterion withp = 2 ).
Therefore X4- 2X2- 2 is the monic minimal polynomial of ~l + -3J2 and it is algebric of
degree 4 over Q.
Example 16.2.5 : Consider the real number .J2 +-J3. If a = .J2 + .J3 , then a2 = 5 +
216, or, a2 - 5 = 216., .This giv~~n(;:t2- 5 )2 = 24 ? or, a4 - IOa2 + I = 0 , Thus a satisfies
, 4 2 . rr ,~

the polynomial f(x) = x - lOx + 1 = O. We claim that f(x) is irreducible over Q, To

prove our claim we use the following result on the irreducibility of a polynomial with
integer coefficients over the rationals ( Observe that we cannot apply Eisenstein Criterion
here). It states that: If f(x) = x" + al xn-I + a2 xn-2 + + an , an *- 0 , is in Z[x] ,
where Z is the
,
ring of integers, and if, f(x) has a zero in Q then it has a zero a in Z and a
~ust divide an. Now let us use this to show that f(x) has no linear factors in Q[x].
Suppose f(x) has a linear factor in Q[x], then it must have a zero a in Z and a must
divide an. Now Jet us use this to show that f(x) has no linear factors in Q[x]. Suppose
; ,t • .

f(x) has a linear factor in Q[x], then it must have a zero a in Z and a must divide 1. So a
= ± 1. But f(1) ;, f(-I) = -8 so that in either case a cannot be a zero of f(x). So f(x)
cannot have linear factors in Q[x]. '
Next suppose that f(x) can be factored in the following way in Q[x] .
x 4 - lOx2 + 1 == ( X2+ ax + b ) ( x2 + ex + d ).
_ Equating the coefficients of powers of x, we have a + c = 0 , b+ d + ac = -10 , be + ad
'" = 0 and bd = 1 . Now bd = 1 gives b = d = ± 1. This together with b+d+ac = -10 gives
- ac = -12 or -8. Using a+c=O, or, a = -c in this we get a2 = 12 or 8. From this we get a =
± 2 13 or ± 2 J2 so a ~ Q and the above factorization is not possible in Q[ x]. Therefore
a is irreducible over Q, so that f(x) is the monic minimal polynomial of J2 +J3 over Q.
Since deg f(x) = 4, Ji +J3 is algebraic of degree 4 over Q.
Self Assessment Question 16.2.6 : 'Show that ( 2 + J3) /2 is algebraic over Q. Find its
(i) minimal polynomial over Q.
(ii) monic minimal polynomial over Q.
(iii) degree over Q.

16.3 FIELD GOT BY ADJOINING AN ELEMENT TO A FIELD F

Let K be an extension uf a field F. For a E K, let F(a) denote the intersection of
all sub-fields Fa of K which contain both F and a.That isF(a) = na { Fa.! F,].is a
sub field of K containing F and a }. Since the intersection of any class of sub-fields of a
field is again sub-field, it follows that F(a) is a sub-field of K. Further F(a) contains both
F and a, since each Fa contains F and a.
Further F( a) being the intersection of all Fa's containing F and a , it is the
smallest sub-field of K which contains Fand a. We call F(a), the field got by adjoining a
to F. Since F c F(a), F(a) is an extension of F.
This description of F(a) is more abstract. Let us now give an explicit description,
or, express the elements of F(a) explicitly. For a[ E F and any positive integer r, consider
the set
,
E of all elements of the fortn ao +. a[ a +., a2 a)2 .+ ..... + ar a". Since a E K arid F .

c K, we have that E c K. So we can define the quotients of the elements of E. Let

i . .
. -.< .-
"

u_{cxo+.cxla+
- .
..... +cxrarl(l·(l·
s 1-10 + 1-11a +
. (l
+ I-1sa
so}*
~() 1- ~[a + + ~sa'
• . \1

It can be easily seen that X - Y , XlY E U whenever X, Y E, D , so that U is a sub- \

field of K. We claim that U = F(a). Since F(a) is a field and F c F(a) ; a E F(a), it
follows that U c F(a) " Also, for every c. E F, a,= all E. U and a ~ a/I E U , so that fe'
U and a E U. So by the definition ofF(a) we have that F(a) c U. Thus F(a)~,~,
. '. i"..·· ,

Definition 16.3.1 :Let K be an extension of a field F. For a E K, a ~ F, the field F(a) ,

which is the intersection of all subfields of K containing F and a '., is called tn'p field got.
by adjoining a to F.
. ,

Example 16.3.2: Find Q(lfi), the field got by adjoining Vi to Q,

Solution :It is easy to see that
 Distance Education 4 )=~~= Acharya Nagarjuna University

113 213
.
Q(V2) =
{a 1
+a 2
2 +a,2 "2/3Iaj,/3jEQ,lsis3 }
113
/31 + /322 + /332
So, if x E Q(V2) , then for some ai'~i E Q, Is is 3 ,
a +a 21/3 +a 22/3
x= I 2 3 1631
~1+~22113+~322/3 .

Multiplying the numerator and denominator of the R.H.S. of 16.3.1 by the rationalizing
113
factor of ~ I + fh 2 +: ~3 22/3 , namely
~/ + (~2 2
1/3
l 2 1/3
+ ( ~3 2 /3)2 + ~l ~2 2 - ~2 ~3 2 - ~3 ~l 2
213
,

its denominator becomes /3?+ 2/3~+ 4/3~- 6/31/32/33,and this is a rational number. Further,
the numerator simplifies to YI + Y2 2 [,'3 + yo, 22/3 for some YI, Y2, Y." E Q, so that .
.
Q( V2) = { a + 2 II">, ~ + 22/3 Y I a , ~ , Y E Q}
Self Assessment Question 16.3.3: Show that C = R(i)
There is a close connection between the algebraic nature of an element a in K and
F(a) as an extension field of F. The following theorem given this:
Theorem 16.3.4 :
Let F be a field and K an extension of F. The element a in K is algebraic over F if
and only if F(a) is a finite extension of F. ',
Proof: Let F be a field, K an extension of F and a E K. Assume that F(a) is a finite
> extension 'of F and let [ F(a) : F ] = n. Then F(a) is a n-dimensional vector space over F,
so that the maximum number of linearly independent elements in F(a) over F is n. Since
a E F(a), the elements l..e, a2 , ,.,.,., all are all in F(a) and their number is n + 1. So these
n+ 1 elements must be linearly dependent over F. This means that there exist elements ao
. ,'-\f;·
, (XI , .,.,.., an in F, not all zero ana
ao 1+ a I a + a2 a 2+ . . . . . ..+ all a 11-0
-
This shows that a is algebraic over F.
Conversely let us assume that a is algebraic over F and let p(x) be the minimal
polynomial of 'a' over F. That is , p(x) is the smallest degree monic polynomial in F[x]
such that pea) = O. Then p(x) must be irreducible over F. For, if p(x) = q(x) t(x), for
some q(x) , t(x) E F[x] such that deg q(x) < deg p(x) and deg t(x) < deg p(x), then 0,=
pea) = q(a) t(a). Here q(a) , tea) E F(a). Further F(a) being a field, it has no zero
divisors. So either q(a) =0 or tea) = 0, which is a contradiction to the minimality of p(x).
Now that p(x) is irreducible, then the principal ideal V = (p(x)) = {p(x) t(x) I t(x)
E F[x] } is a maximal ideal of F[ x]. So, F[ x] / V is a field.
Consider the map \jf : F[x] -7 F(a) given by
\1' (f(x)) = f(a) , _ 16.3.2 or, equivalently given by
 \V ( ao + a Ix + + all .x" ) = ao + a I a +. + an an
for all f(x) = ao + alx + ..... + an x" E F[x]
It is easy to see that \1' is a ring homomorphism. What is ker \IJ ? By the definition
Ker \IJ = {g(x) E F[x] I \v (g(x)) = O}
= { g(x) E F[x] I g(a) = 0 }.
Since pea) = 0 , p(x) E ker q/ , so that V =( p(x) ) c ker \v .
Further if g(x) E ker \IJ , then g(a) = O. But p(x) is the smallest degree polynomial
such that pea) = O. So p(x) is divides g(x) in F[x], so that g(x) = p(x) t(x) for some t(x) E
F[ x]. This implies that g(x) E V. Thus kef \IJ c V. These show that ker \v = V.
Since Fe F(a) and a E F(a), \IJ ( F[x] ) = { \IJ (f(x) I f(x) E F[x] } = { f(a) I
[(x) E F[x] } = F(a).
Also, for a E F, a = a +Ox2 + Ox2 + .... + Ox" E F[x] and \I'(a) = a + Oa + Oa2 +
...+ Oa'" = a. Thus a E \IJ (F[x]), or, F c \11 (F[x]). Further \1' (x) = \IJ (0 + l x

+ Ox2 -+ ..... + OXIl) = 0 + la + Oa2 + ..... + Oa" = a, so that a E \V (F[x]). Thus F(a) c \IJ (
F[x] ) and \1/ ( F[x] ) = F(a) and \1' is onto.
Thus we have seen that the mapping \IJ : F[x] -7 F(a) given by 16.3.2 is an onto
ring homomorphism with kernal V. So by the fundamental theorem of ring
homomorphism we have
F[x] i V ~ F(a) 16.3.3
Let deg p(x) = n. Suppose we show that dim! ( F[x] ) I V ) = ~, then because of the
isomorphism in 16.3 .3,it follows that dim, F(a) = 11 and [ F(a) : F ] = n, which is finite.
We shall now show that dim, ( F[x] ) I V) = n, For this, it isenough if we exhibit
a basis of. F[x] I V over F containing n elements. 1~~,
.,,---
set of elem'ents we seek for this
purpose IS
S = {V + I, V+x, V +X2, ll I
, V+x - } 16.3.4

These elements being cosets of V by the polynomials l,.x, X2 , ... , Xll-I , belong to
F[x] I v. Now let V + f(x) E F[x] I V. Then f(x) E F[x]. So, by the division algorithm in
F[x], we can find polynomials q(x), rex) E F]x], such that f(x) = p(x) q(x) + rex),
where deg rex) < deg p(x). So .
V + f(x) = V + p(x) q(x) + rex) = V + rex) ,
( since V = (p(x) ) , p(x) q(x) E V) .
Suppose that rex) an + alx + .....+ an-I xn-I (since deg r(x)<deg p(x)
= =n )
Then V + f(x) = V + rex) ~ V + (cxo+ CXIX+ ..... + all-I xn-I )
=( V + ao ) + ( V + CXIX) + ..... + ( V + an-I xn-1 )
- CXo(V + 1 ) + CXI( V + X ) + . . . ..+ CXn-l
( V + x,n-l)
 Distance Education 6

which is a linear combination of elements S. This shows that S spans F[x] I V over F.
nI
To show that V + 1 , , V+ x - are linearly independent over F, let-us assume
that for ao , ai, ..... , an-I E F ,
ao ( V + 1 ) + al ( V + x) + ..... + an-l ( V + xn-I ) = V,
where V is the zero element of F[ x] I V. Then
V + (+
ao alx + ..... + an-I Xn-I ) -- V ,or, ao + alx + ..... + an-I x n-I E V . N ow th e °

degree of this polynomial is n - 1, and this is less than n, the degree of p(x). This
n I
contradicts the minimality of p(x). So ao + cqx + ..... + Un-I x - must be the zero
polynomial, so that ao = al = ..... an-I = O. This implies that the elements V+ I , V+x,
V+X2 , .... , V +xn-I are linearly independing over F 'and thus form a basis of F[x] I V over
F.
Thus dim- (F[x] I V) = n , or, equivalently dim, F(a) = n . So
[ F(a) : F ] = n .
This proves our theorem.
In proving the converse of the above theorem, we incidentally proved a more
sharper result, namely,
Theorem 16.3.5: If a E K is algebraic of degree n over F then [ F(a) : F ] = n .
Example 16.3.6: Show that [ Q (Vi ) : Q] = 3 .
113
Solution: Consider Q (Vi). We have seen in example 15.4.3, that Q(Vi) = { a + 2
23
~ + 22/3 y I a, ~,y E Q } . So the sub-set { 1,21/3 ,2 / } ofQ(Zi2) spans Q(Zi2) over Q.
Further, for a, ~, Y E Q , let a + 2'/3 ~ + 22/3 Y = O. If a i::- 0 , then a =- ( 21/3 + 22/3 ) and
this is not a rational number. So a = O. Then 2113 ~ + 22/3 Y = 0 , or, ~ + 2'/3 y = 0 .
Again, if f3 i::- 0 , then f3 = _21/3 y which is not a rational number. So f3 = 0 and thus y = 0 .
So a + i/3 f3 + 22/3 Y = 0 with a, f3, y E Q implies a = f3 = y = O,so that { 1 , 21/3, 22/3} is
linearly independent over Q. Thus { 1, 21/3 , 22/3 } is a basis of Q( Vi) over Q, so that [ Q
(Zi2): Q ]= 3. By theorem ] 5.4.2 , Vi is algebraic of degree 3 over Q. This cat! be
seen in much simpler way also. V2 satisfies the polynomial x' - 2 and this is irreducible
over Q,by Eisenstein's criterion with p = 2. So x ' - 2 is the minimal polynomial of
Vi over Q so that Vi is algebraic of degree 3 over QO.
Self Assessment Question 16.3.7: Find Q( -J3 ) over Q and [ Q( -J3) : Q ].
Theorem 16.3.8 : The set of all algebraic elements in Kover F form a sub-field of K.
Proof: Suppose E is the set of all algebraic elements in Kover F and let a, bEE. If we
show that a ± b, ab and a/b, ( bi::- 0 ) are also algebraic over F then they belong to E and E
o
is a sub-field of K ..
Put F(a) = T and T(b) = W.
 Since a is algebraic over F, [ T : F 1 is finite, say, m. Let b be algebraic
of degree n
over F. Then the degree of the minimal polynomial of b over F is n. Since F c T, the
degree of the minimal polynomial ofb over Tis::; n. By the theorem 16.3.4, this gives [
w rr j s ».
But, by the Theorem 15.4.7 [W: F] = [W : T] [ T : F] sm n .
So, W is a fmite extension of F.
Since the elements a, b belong to Wand W is a sub-field of K, a ± b, ab and aIb, (
b :/:.0 ) belong to W. So by the corollary 15.4.8, [ F ( a ± b ) : F ] :::;;[ W : F] :::;;m n and
similarly [ F(a) :' F ]:::;; m n ,[ F ( a/b) : F] :::;;mn. So again by Theorem 15.5.12, a ± b,
ab and alb, b '*
0 are all algebraic over F, so that they belong to E. So E is a subfield of
K. From the theorem we have
Corollary 16.3.9 :
If a, b in K are algebraic over F of degree m and n over F respectively, then a ±b ,
'*
ab and a/b, b 0 are algebraic of degree atmost mn over F.
Example 16.3.10: Show that Q (J3 +:J5): Q ] = 4 and Q (.J3 .:J5 : Q) = 2.
Solution: Consider the numbers .J3 and :J5 . Since X2-
3 = 0 and X2 - 5 = 0 , both .J3
and :J5 are algebraic over Q. Let a = a2 = 8 + 2 J15, or, a2 - 8 = 2
J3 +:J5 . Then
.J15 'From this we get a" - 16 a2 + 4 = 0 . So a satisfies the polynomial X4 - 16x2 + 4·
over Q, so that -fj + -f5 is algebraic over Q. As in example 16.2.5 we can see that X4-
16x2 + 4 is irreducible over Q. So.J3 +15 is algebraic of degree 4 over Q. SO [ Q (-fj
+ Fs) : Q ] = 4. Also J3 Fs satisfies x2 - 15 and this is irreducible over Q, by
Eisenstein's criterion with p = 3 or 5. So J3:J5 is algebraic of degree 2 over Q and [
r:; . r;: ,,/,
Q\l3\1S ):Q)=2. '. . .q.
Self Assessment Question 16.3.11: Find the degree of i·'"'i.-fi
., ,
and i-fi over Q .
Suppose that a, b E K. Consider.T = F(a) . Then T isa sub-field of K. So we can
consider T(b) = ( F(a) ) (b). We denote this extension by F(a,b). In the same way we
can find F(b,a) . Since F c F(b) , wehave Fea) c F(b,a). But b E F(b,a) so that (
F(a) ) (b) ) c F(b,a) , or, F(a,b) c F(b,a). By a similar argument we can show that F(b,a)
c F(a,b) . So F(a,b) = F(b,a) .
If ai , a2 , , an are any finite number of elements in K, we can similarly define
the extension F( a) , a2 , ... , an) . ., .'."
f3
Example 16.3.12: Show that [ Q (21/2, zI ) : Q] = 6 .
Solution :Consider 2112 and 21/3 : There is no element in Q ( .J2) which satisfies x3 - 2 .
This is because the degree of .J2
over Q is 2, while a zero of x' - 2 is of degree 3 ( by
Eisenstein's criterion with p = 2 ), but 3 does not divide 2. So 21/3 ~ Q ( .fi) . Thus
2
[ Q ( 2[/2, 2[/3) : Q ( zI/ ) ] = 3 .
 Distance Education 8 .:e~~-:Acharya Nagarjuna University

Also { 1, 21/2 } is a basis of Q(21/2 ) over Q and { t", 21/3 , 22/3 } is a basis of Q(21/2 ,
21/3) over Q(21/2 ). Further by the theorem 15.3.1
{ 1. 1 , 1.2 lI3 , 1. 22/3 , 2112.1 , il2 21/3, 2112 22./3 },

or, { 1" 2112 2113 , 22/3 , 25/6 , 27/6}

is a basis of Q( 2112 , 2l13) over Q. SO
[ Q ( 2112 , 21/3 ) : Q ] = 6.
Self Assessment Question 16.3.13: Find the degree and a basis for Q( -Ii, -fj ) over Q.
Self Assessment Question 16.3.14 : Show that Q( -Ii -+- -fj ) = Q (-Ii ,-fj ) .

16.4 ALGEBRAIC EXTENSIONS:

An extension K of a field F is called an algebraic extension if every element of K
is algebraic over F.
Analogous to theorem 15.4.7, we have
Theorem 16.4.1 :
If L is an algebraic extension of K and if K is an algebraic extension of F then Lis
an algebraic extension of F.
Proof: It is enough if we show that every element in L is algebraic over F. Let I E L .
We are given that L is an algebraic extension of K. So every element of L ( and
hence I ) is algebraic over K. So, for some al , a2 , ... , an in K not all zero,
In+ a I In-l + a2 In-2+ . . . . . .+ an-l I + an -- 0 16.4 . I
But it is given that K is algebraic over F. SO each of al , a2 , .... ,an is algebraic over F,
so that, (by the repeated application of the theorem 16.3.4) M = F (cq , a2
, .... ,an) is a finite extension ofF;' or , [ M : F ] is finite. As al , (X2 , .... ,an E M, we
conclude from equation 16.4.1 that I satisfies the non-zero polynomial x" + an xn-1 + a2
xn-2 + ..... +un over M. So 1 is a:ig"ebraicover M. Now by the theorem 16.3.4, [ M (1) : M
] is finite and consequently,
[ M (1) : F ] = [ M(1) : M ] [ M : F I. .--
is also finite. Again by theorem 16.3.4 this implies that I is algebraic over F. Thus every
element in L is algebraic over F· so that L is an algebraic extension of F.
Definition 16.4.2 : A complex numbers is called an algebraic number if it is algebraic
over the field of rational numbers. A complex number which is not algebraic is called
transcendental.
An application of the theorem 16.3.8 to algebraic numbers yields that the set of
algebraic numbers form a sub-field of the complex numbers.
Example 16.4.3: Show that the element -Ii -i in C is of degree 4 over Q.
Solution: Consider the element 12 - i in C. If a = 12 - i then ( a + i)2 = 2, or , a2 + 2i
- 1 = 2 . This gives ( a2 - 3 i = -2i ,'or, a4 - 6a2 + 13 = 0 . That is , 12 -i satisfies the
polynomial f(x) == X4 - 6x2 + 13 = O. Also, as in example 15.4.2 we can show that f(x) is
in-educible over Q. SO 12 -i is an algebraic number and it is of degree 4 over Q.

16.5 SUMMARY:
In this lesson we have studied the following:
1. An element a in K, an extension of F, is algebraic over F if it satisfies a non - zero
polynomial over F.
2. The minimal polynomial of an element a in Kover F is the smallest degree
polynomial p(x) over F such that pea) = a and the degree of an algebraic element a
over F is the degree of its minimal. -
3. The ficl:' F(a) is the intersection of all the sub-field of K containing both F and a
and it is the smallest sub-field of K containing both F and a. The field F(a) is
called the field got by adjoining a to F.
4. The element a in K is algebraic over F if and only if F(a) is a finite extension of F.
5. If a is algebraic over F then [ F(a) : F ] is equal to the degree of the minimal
polynomial satisfied by a over F.
6. The set of all algebraic elements in K forms a sub-field of K.
7. AI' extension K of F is an algebraic extension if every element in is algebraic over
F.
8. If L is an algebraic extension of K and K is an algebraic extension of F then Lis
an algebraic extension of F.
9. An algebraic number is a complex number which is algebraic over the field of
rational numbers. . '1.

10. The set· of all algebraic numbers forms a sub-field of the field of complex
numbers.

16.6 TECHNICAL TERMS:

In this lesson we have become familiar with the following technical terms.
Algebraic element
Algebraic extension
Minimal polynomial of an algebraic element
Field got by adjoining an element to a field.
Algebraic number.
 -Distance Education

16.7: MODEL EXAMINATION QUESTIONS

PART - A ( short answer questions)
1. Consider the extension Q( -fi) of Q. Show -fi and I - 3 -fi in Q( -fi) are
algebraic over Q.
2. Find the minimal polynomial satisfied by 1 + Vi over Q. Find also its degree
over Q.
3. Find the degree and basis for Q( -fi ,fj) over Q.
2
4. Show that C = R(i) where i = -1
5. Show that [ Q(V2 ): Q ] = 3
6. Show Q( 12 ,J] ) ~ Q (12+J3).
PART - B ( long answer Questions)
I. Let K be an extension of F. Show that an element a in K is algebraic over F if and
only if F( a) is a finite extension of F.
2. If a E K is algebraic of degree n over F show that [ F(a) : F ] = n ..
3. Let K be an extension of F. Show that the set of all algebraic elements in Kover F
form a sub-field of K.

16.8: ANSWERS TO SELF ASSESSMENT QUESTIONS:

Self Assessment Question 16.2.6: Show that ( 2 + .Jj) / 2 is algebraic over Q.
Find its
(i) minimal polynomial over Q.
(ii) monic
. minimaLpolynomial
( ,.
over Q
(iii) degree over Q. (
Solution: Let a = ( 2 + .Jj) -'it.'
Then 2a - 2 = J3, or, 4a2 - Sa + I = 0 . So a satisfies
f(x) = 4X2 - 8x + 1 . As in example 16.2.5 one can see that f(x) is irreducible over Q.
Hence f(x) is the minimal polynomial of a over Q and X2 - 2x + Y4 is the monic minimal
polynomial of a over Q. Since degree of the above polynomial is 2, a is algebraic of.
degree 2 over Q. .
Self Assessment Question 16.3.3: Show that C = R(i) where i2 = -1.
Solution: Let a = i . Then a2 = -1 or a satisfies X2 + lover R. Clearly X2 + 1 is
irreducible over R. SO X2 + 1 is the monic minimal of iver i. This shows that i is
algebraic of degree 2 over Rand {I, i} is a basis ofR(i) over R. Hence
R(i) = { a 1 + bi / a, b E R } = C.
Self Assessment Question 16.3.7: Find Q( fj) over Q and [Q(.Jj) : Q] .
Solution: -13 satisfies the polynomial x2 - 3 over Q and this is irreducible over Q, by
Eisenstein's criterion with p = 3 . SO X2- 3 is' the minimal polynomial of -13 over Q so
that J3 is algebraic of degree 2 over Q. SO
Q( -13) = { a + b -13/ a, b E Q },
and { 1, -13} is a basis of Q( -13) over Q. Hence
[Q (-13): Q ] = 2.
Self Assessment Question 16.3.11: Find the degree of i - J2 and i J2 over Q.
Solution: Let a = i - J2 . Then a4 - 2a2 + 9 = 0 , or, a satisfies the polynomial f(x) = X4 -
2X2 + 9. As in example 16.2.5, one can see that f(x) is irreducible over Q. SO i-J2 is
algebraic of degree 4 over Q.
In a similar way one can see that i J2 is algebraic of degree 2 over Q.
Self Assessment Ques~on 16.3.13: Find the degree and a basis for Q( J2 ,-13) over Q.
Solution: Consider J2 and J3 . It is easy to see that [ Q( J2) : Q ] = 2 and {I, J2}
is a basis of Q( J2) over Q. Further there is no element in Q( J2) which satisfies X2 - 3 .
So -13~ Q( J2) . Thus
[ Q( -.fi , -J3) : Q( J2) ] = 2.
Also { 1, J3} is a basis of Q( J2 ,-13 ) over Q( J2).
So { 1.1 , 1. J2, -13. I , J2 J3} , or, { 1, J2, J3 , J2 13} , is a basis of Q( J2 ,J3 )
over Q. This shows
[Q( J2, -13 ) : Q] = 4.
Self Assessment Question 16.3.14 : Show that Q( J2 +--/3 ) = Q( J2, -J3 ).
Solution: Since J2, 13 E Q( J2 ,13) and since this is a field,
Q( J2 + 13 ) c Q( J2 ,13 ).
Again since Q( J2 +.fj ) is a field and (J2 + -13) E Q( J2 + --.)3 ) we have J2 1 13
2+ 3
E Q( J2+-13 ) , or, -13 - Ji is in Q( J2 +-13 ), Now -13 + J2 and -13 - J2 are in
Q( J2 +13 ) imply that J2, -13 E Q( Ji +-13 ) ,or ,
Q( J2 ,J3 ) c Q( J2 + 13 ).
Hence Q(.fi +-13 ) = Q( .fi ,-13 ).
 Distance Education 12

16.9: EXERCISES:
1. For each of the numbers in R given below show thatit is algebraic over Q, fmd its
minimal polynomial and its degree over Q.
(i) 1 + .Ji . (ii) 1+ Vi (ii) ~3 - 16 (iii) ( 1 + ViY2
2. For each of the numbers in C given below, show that it is algebraic over Q, find
its minimal polynomial and its degree over Q.
(i) 1 + i (ii) ~3fi- i (iii) ..fi + i
3 . Find the degree and a basis for each of the given field extensions -
(a) Q (.Ji) over Q, (b) Q(.Ji ,J) ,fs) over Q.
4. Find the degree of each of the following field extensions:
(a) Q (J) + J7 ) over Q (b) Q (-fi , V5) over Q.
(c) Q (-fi, J6 ) over Q CJ3 ), (d)'Q( -fi + J3 ) over Q( J3);
(e) Q ( -fi ,J3 ) over Q (-fi + J3). (f) Q( 1+i) over Q.
(g) Q(Vi, J) over Q.
. 5. Prove that X2 - 3 is irreducible over Q ( Vi ) .
6. Prove that Q (J3 + J7 r= Q (J3 ,J7).
7. Show that { 1, Ji ;J3 , 16 ) is a basis of Q ( -fi ,J)) over Q.
8. Show that Q ( 2113 23
, 2 / ) = Q( 21/6 ) .

9. Show that Q(.J3 + -!5 ) = Q(.J3 , J5 )

10. Show that 3 - J3 iis an algebraic element. Find its degree over Q.
16.10REFERENCE BOO~
1. ARTIN, EMIL, Galois Theory, 2nd ed Notne Dame, Ind : University of
Natne Dame Press, 1955.
2. FRALEIGH, lB., A First Course in Abstract Algebra. 2nd ed., Addision Wesley
Publishing Company.
3. HERSTERlN, IN., Topics in Algebra, Vikas Publishing House.
4. JACOBSON, N., Lectures in abstract algebra. Vo1.lI, Nostnand, 1953.
5.: BHATTACHARYA, P.B., JAIN. S.K., NAGPAUL.S.R., Basic Algebra,
(2nd Edition) , Cambridge University Press.

Lesson Writer
Prof. L.N agamuni Reddy
S.V.University, Tirupati
 LESSON - 17 : ROOTS OF POLYNOMIALS

CONTENTS
Objectives
Structure
17.1 Introduction
17.2 Roots of Polynomials
17.3 Splitting fields
17.4 Basic Isomorphisms - Uniqueness of splitting fields
17.5 Summary
17.6 Technical terms
17.7 Model Examination Questions
17.8 Answers to Self Assessment Questions
17.9 Exercises
17.10 Reference Books

OBJECTIVES
After reading this unit you should be able to :
• Know about the remainder theorem and its consequences ;
• Establish that given a polynomial f(x) over a field F, there always exists an
extension of F which contains a root of f(x) and also an extension which contains
all the roots of f(x) ;
• Define the splitting field of a polynomial over a £kId and prove its uniqueness;

17.1 INTRODUCTION
Consider the polynomial X2 - 2 over Q, the field of rational numbers. Its roots ±
12 do not lie in Q. But they lie in R, the field of real numbers, which is an extension of
Q. In the same way the roots of the polynomial X2 + lover Q lie neither in Q nor in R but
in C, the field of complex numbers, which is again viewed as an extensiol1 of Q. Thus
given a polynomial f(x) over afield F, it is not guaranteed that F, or, any other extension
of F contains a root of f(x). In this unit it is our primary goal to establish that there exists
an extension of F which contains a root of f(;, > and also an extension of F which contain
all the roots of f(x). Once this is achieved, we will be able to fmd an extension of
minimum degree. over F, which contains all the roots of f(x). Such an extension is called
a splitting field off(x) over F.

-,
 Distance Education 2

17.2 ROOTS OF POLYNOMIALS

Let f(x) = ao + at x + a2 X2 + + an x" be a polynomial over F. In Lesson
15 we have seen that an element a in some extension K of F is called a root of f(x) if f(a)
= 0 , or, equivalently if ao + at a + a2 a2 + .... + an an = O. As indicated earlier our first
task is to establish that there is an extension K of F, which contains a root of f(x). In
doing so, we incidentally obtain an upper bound for the degree of such extension.
Definition 17.2.1: Let F be a field and f(x) E F[x]. All element a in some extension of
F is called a root off(x) iff(a) = 0 . ,
The following theorem, which is familiarly known as the remainder theorem, is of
great use in our future discussion.
Theorem 17.2.2.: Let f(x) E F[x] and K an extension of F. Then for any a E K,
f(x) = ( x - a) q(x) + f(a) , 17.2.1
where q(x) E K[x] and deg q(x) = deg f(x) - 1.
Proof: F c K implies that F [x] c K[x]. Further x -a E K[x] since a E K. So, applying
division algorithm to f(x) and x -a in K[x], there exists polynomials q(x), rex) E K[x]
such that
f(x) = (x - a) q(x) + rex) 17.2.2
where either rex) = 0 , or deg rex) < deg (x - a). That is deg rex) = O. In either case rex)
E K. Let rex) = c E K. To find the value of c we substitute x = a in 17.2.2. Then

f(a) = ( a - a) q(a) + c, or, f(a) = c .

Thus we have f(x) = (x - a) q(x) + f(a) ,
where q(x) EO: K[x]. The deg q(x) = deg f(x) - 1 is immediate if we compare the degrees
of the polynomials on either side of 17.2.1.
Corollary 17.2.3: If a E K is a root of f(x) E F[x] where F c K, then x -a divides f(x)
in K[x].
Proof: If a E K, is a root of fix) then f(a) = O. So 16.3.1 gives that
f(x) =(x- a) q(x) ,
where q(x) E K[x]. This means that x - a divides f(x) in K[x].
Definition 17.2.4: Let f(x) E F[x] and a E K. If (x - a )" divides f(x) and (x - a )m+l
does not divide f(x), then a is called a root of multiplicity m of f(x).
One natural question that arises in this regard is: Given a polynomial f(x) of
degree n over a field F, what is the maximum number of roots -of f(x) that may lie in any
extension of F? The following theorem answers this.
Theorem 17.2.S: A polynomial of degree n over a field can have atmost n roots in any
extension of F.
 Proof: Let f(x) E F[x] and deg f(x) = n. We prove the theorem by using induction on n.
If deg f(x) = 1, then f(x) = a x + ~ , for some a, ~ E F and a =f. 0 and - a-l ~ is the
only root off(x) in F, So the theorem is true in this case.
Let K be any extension of F. If K does not contain any root of f(x), then the
theorem is vacuously hue, as 0 < n. So, let us assume that K contains a root a of f(x) and
this is of multiplicity m. Then
f(x) = ( x - a )" q(x), . 17.2.3
where q(x) E K[x] , deg q(x) = n - m < nand x- a
does not divide q(x). So a is not a
root of q(x). From the equation 17.2.3 it follows that any root of q(x) is also a root of
f(x). Further ifb *' a is a root of f(x) , then f(b) = 0, so that we get (b - a tq(b) =
o , or q(b) = 0, since b - a*,O. Thus the roots of f(x) other than a are given by the roots
of q(x). Since deg q(x) = n - m < n, we have by the induction hypothesis that K contains
atmost n - m roots of q(x). So K contains at most m + n - m c-= 11 roots of f(x).
We are now in a position to achieve our basic goal, which when loosely stated 1S
to show that every non-constant polynomial over a field F has a root in some extension
of F.'
Theorem 17.2.6: Let F be a field and p(x) an irreducible polynomial in F[x] of degree
n ~ 1. Then there is an extension E of F such that [ E : F ] = n , in which p(x) has a root.
Proof: Suppose p(x) E F[x] is irreducible and deg p(x) = n ~ 1 . Then the principal ideal
'V = ( p(x) ), generated by p(x) is maximal in F[x] , so that ' E = F [x] IV is a
field. The map \If : F]x] -7 F[x] IV given by
\If ( f(x) ) = f(x) + V.
is an onto homomorphism with kernel V. If

,
+'
F~ { a V I a E F } ,
-
0";.
eb .. ,
then the restriction of \v to F ( that is \If : F -7 F, defined by \v( a) = a + V ) is an
isomorphism of F onto F ( see SAQ 17.2.10).' So E 'llkn extension of F. Because of
this isomorphism, we can identify F with F and the element a + V in F with a in F.

F[x] E = F[xll V

f(x) + V
'----..
-
F
F
 Distance Education 4

'We now show that [E: F] = nand E contains a root ofp(x) .

Consider the subset S = { 1 + V, x + V, X2 + V , .... , xn-I + V } of F[x]/V. We
claim that S forms a basis of F[x] IV over F. To see this, let f(x) + V E F[x] IV. ,Then
f(x) E F[x] and by division algorithm in F[x], we can find q(x), rex) E F[x] such that.
f(x) = p(x) q(x) + rex) , where either rex) = 0 , or , deg rex) < deg p(x) = n. Then
p(x) q(x) E V and thus f(x) + V . Since either rex) = 0 , or, deg rex) < n, we can write .
.rex) in the form
n-I
rex) = ao + al x + a2 x2 + + an-I x
for some ao, aI, a2, , an-I E F.
Then f(x) + V = (ao + al x + a2 X2+ . . . . . + an-I xn-I ) + V
=(ao+V)+(alx+ V)+ + (an_Ixll-1 +V)
nI
= ao ( I + V) + al ( x + V) + + an-I ( x - + V)
This show that f(x) + V is in the linear span of Saver F, or, S spans F[ x] I V over F.
Further, for ~o, ~I, ~2' .... , ~n-I E F, if ~o ( 1 + V) + ~l (x + V) + ..... -!- ~11-1(
l1 1
x"" + V) = V, the zero ofF[x] I V, then (~o+ ~I X + ~2 X2+ ... " + ~11-1x - ) +
V= V , or, (f30, ~I x , ~2 X2+ ..... + ~n-I X"-I) E V.
So p(x) divides ~o + PI X + ~2 X2 + ..... + ~1l-1 xn-I . Since the degree of this
polynomial is n - I and this is less than deg p( x), it follows that Po + PI X + P2 X2+ ... ,.
nI
+ Pn-I x - = 0 or, Po = PI =[P2 = ..... = PI1-1 = 0 . This shows that S is linearly
independent over F. Thus S forms a basis of F[x] / V over F. But S contains n elements,
so that [ E : F ] = n.
It remains to be shown that E = F[x] / V contains a root ofp(x). Put a=x
+ V. Then a E E, Ifp(x) = Yo+ YIX+ Y2X2+ + YI1x" ,YI1i- 0, then
\}f [ p(x) ] = p(x) + V =;l'i~io+ YIX+ Y2X2+ + YI1x" ) + V
2 ll
= (yo + V) ~tl (X + V) + Y2(X + V) " .. + YI1(X + V)
=( Yo+ V) + YI( X + V) + Y2( X + vi + YI1( X + vt
2 n
= Yo+ YI a + Y2a + + '[« a .:

(under the isomorphism \}f : F ~ F , we identify Yo with Yo+ V ). But p(x) E V = ker
. .. 2
\}f , so that \}f [ p(x) ] = 0 , the zero of E. So Yo+ YI a + Y2a +. .. . . . + Ynan = 0 .
That.is.a = x + V is a root ofp(x) in E = F [x] / V.
Corollary'I-7.2.7: If f(x) is a polynomial in F[x] of degree n ~ 1, then there exists an
extension E of F such that [ E : F] ::; nand f(x) has a root in E.
Proof: Let p(x) be an irreducible factor of f(x) in F[x). Then f(x) = p(x) q(x) for some
q(x) E F[x] and deg p(x) < deg f(x) = n .
If a is a root ofp(x) in some extension E of F then pea) = 0 , and
 f(a) = pea) q(a) = 0 ,
so that a is also a root of f(x).
Since p(x) is irreducible, by the Theorem 17.2.6,t here exists an extension E of F,
which contains a root of p(x) and thus a root of f(x) also. Further [ E : F ] = deg p(x) ~ n.
As a consequence of this Corollary we now prove a theorem which guarantees the
existence of an extension which contains all the roots of a given polynomial and also
gives an upper bound for the degree of such an extension.
Theorem 17.2.8: Let f(x) E F[x] be of degree n ~ 1 . Then there is an extension E of F
degree atmost n! in which f(x) has n roots. I

Proof : We prove the theorem by using induction on n.

If n = 1, then f(x) = a x + ~ for some a,~· E F, a::j:. O. In this case - a-I b is the
only root of f(x) and this lies in F. Then the trivial extension E = F satisfies all the
conditions of the theorem.
So let n > 1 and assume that the theorem holds for all polynomials g(x), whose
degree is less than n. By the Corollary 16.3.7, there is an extension Eo of F with [
Eo : F ] ~ n, in which f(x) has a root say a. Then f(x) factors in Eo [x] as
f(x) = (x - a) g(x) , for some g(x) E Eo[x].
Also deg g(x) = n - 1 < n. So by the induction hypothesis there is an extension E
of Eo with [E: Eo] ~ (n-I)! ,in which g(x) has n - 1 roots. But any root off(x) is either
a, or, a root of g(x). So E contains all the roots of f(x) and
[ E : F ] = [ E : Eo ] [ Eo : F] ~(n -I)! . n = n! .
Self Assessment Question 17.2-.9 : Show that the map 'I': F[x]~F[x]/ (p(x)) defined by
'I' ( f(x) ) = f(x) + ( ptx) is an onto homomorphism with kernel ( p(x) ).
Self Assessment Question 17.2.10: (Show that the map 'I' : F ~ F given in the
Theorem 17.2.6, which is defined by 'I' (a) = a + V , for all a E F is an isomorphism.
Example 17.2.11 :Find an extension of R in which X2+ 1 has a root. Find its degree over
R.
Solution: Let f(x) = X2+ 1. Then f(x) has no roots in R and thus it is irreducible over R.
So the ideal V = ( f(x) ) , generated by X2+ 1 is maximal in R[x] so that R[x] / V is a
field. Identifying r E R with r + V, we can view R as a sub field of E = R[x] I V. Let a =
x + V. Computing in R]x] / V we find
a2 + 1 = ( x + V)2 + 1 + V = ( X2+ V ) + ( 1 + V)
= ( x2 + 1 ) + V = V, since x2 + 1 E V.
Since V is the zero of R[x] / V, it follows that a2 + I = O. Thus a is a root of X2
+ 1 in R[x] / V. Here [E: R] = 2 = deg f(x).
ExampJe 17.2.12 : Obtain two extensions of R such that each of them contains a root of
2
x4 - 5x + 6. Find the degree of each of them over R. o

\9'
 Distance Education 6

Solution: Let F = Q and f(x) = X4- 5x2 + 6. Here f(x) factors in Q[x] as f(x) =
( X2- 2 ) ( x2 - 3 ). Since ± 12 , ± Jj ~ Q, both the factors are irreducible over Q.
. Starting with the irreducible polynomial X2- 2 we construct the extension field E = Q [x]
/ V, where V is the ideal generated by X2- 2 and it contains the root a = x + V of X2-
2. In the same way we can as well start with the irreducible polynomialxi :...-3 over Q
and obtain the extension EI = Q[x] / VI ,where VI is the ideal generated by X2- 3 , such
that it contains the root b = x + VI of X2- 3. Thus we have two extensions E and EI
such that each contains a root off(x). Also note that [ E : Q ] = 2 <4 and [E1 : Q] = 2<
4 , and deg f(x) = 4.

17.3 SPLITTING FIELDS:

Let F be a field and f(x) E F[x] is of degree n ~ 1 . The Theorem 16.3.8
guarantees the existence of an extension E of F, which contains all the n roots of f(x) and
gives n!, as an upper bound for [ E : F] . Iff(x) = 0.0 + 0.1X + 0.2X2+ + an x" , an t-
O and al , a2 , .... , an are its roots in E, then by the Corollary 16.3.3, f(x) can be factored
~s the product of linear factors in E[x] as f(x)= (x -~a\ ) (~ -:-a2 ) ( x - an ). So
there exists a finite extension E of F, over which f(x) splits into linear factors. Thus there
exists a finite extension of minimal degree over F, which has the property that f(x) factors
/

into a product of linear factors over it. For such an extension, no subfield of it contains all
the roots of f(x). This leads to the following definition of a splitting field of a
polynomial.
Definition 17.3.1: F be a field and f(x) E F[x]. A [mite extension E of F is called a
splitting field of f(x), if f(x) factors as a product of linear factors over E, but not over any
subfield of E.
In other words E is a,~.Jl>littingof f(x) over F if E is an extension of F .with
minimum degree, ( such an extension is called a minimal extension) such that f(x) has n
roots in a, where n = deg f(x). ~,~
Examele 17.3.2: Find the splitting field and its degree of extension of the polynomial x3
- 2 over the field Q of rational numbers.
Solution: Consider the field Q of rational numbers and the polynomial x3 - 2 in Q[x].
The roots off(x) ~,e ~2, ~2
.r: .r: (-I+i-J3J
.' 2' and
«r:
~2
(-I-i.J3J
.7 . . ~ong these only ~2
«r: ..
IS

, ".
real. SO QI = Q (V2) is a subfield of R. Since ifi. (-1 . i.J3'J
+
2 . and
., ifi-·(- 1- i.J3 J "are.
2

complex, they do not lie in Q\ , so that Q\ is not a splitting .field of x3 - 2.

Since ifi satisfies an irreducible polynomial x3 - 2 over Q, we have I QI : Q ] = 3
Let Q2 = QI ( i-J3). Then V2 , i -J3 E Q2 and Q2 is a field .. These imply that
 V2 [-1 ~iJ3). V2 ( -1 ~iJ3) E Q, . Thus Q, contains all the roots of x3 - 2 . Now

i -J3 satisfies the irreducible polynomial X2+ 3 over Q (Vi). So [ Q2 : Ql ] = 2. Thus

[ Q2 : Q ] = [ Q2 : QI ] [ QI : Q 1= 2 . 3 = 6 .
IfE is the splitting field ofx - 2, then [ E : Q] ::;L3 = 6 . But[ E : Q] =
3
'" [
E: o, ] [ Q\ : Q ] = 3 [ E : QI ] , so that 3 divides [ E : Q ]. This forces that [E :
3
Q] = 6. Therefore Q2 = Ql (i Jj) is the splitting field of x - 2 and its degree over Q is
6.
Example 17.3.3: Find the splitting field of the polynomial X4+ x" + 1 in Q[x] over the
field Q of rational numbers and also its degree over Q.
Solution: Consider the field Q of rational numbers and the polynomial f(x) = X4+ X2+ 1
in Q[xl Then
f(x) = X4+ 2X2/+ I - X2= (x2 + I )2 _ X2
~ [ ( X2+ 1 ) + x ] [ ( X2+ 1) - x l.
. - - 1 ± i-J3 1 ± i.J3
Thus the roots of f(x) are given by . and . Let QI = Q (i -fj). Since
2 -2
" ' a : -l+iJj , d 1±iJj b 1
Q c QI, we hav~ Yz E QI . Now Yz, i,,3 E QI imply that ~ an e ong
2
to QI . Thus Q\ contains all the roots of f(x). Since i 13 satisfies X2+ 3 over Q and
since X2+ 3 is irreducible over Q, it follows that [ QI : Q ] = 2.
'If QI is any subfield of QI' which contains all the roots of f(x) then [ QI : Q ] =
[ QI : QI ] [ QI : Q] . So [ QI : QI ] divides [ QI : Q l Since [ QI:Q ] = 2, this forces that
either [-Q\ : QI ] = lor [ QI : Q ] = 1 . If [QI : Q ] = I'tEen QI = Q arid this leads to the
contradiction that Q contains all the roots of f(x). SO F~>[': Q\ ] = I, or Q\ = QI. Thus
Q\ is the splitting field of f(x). Further the degree of Q\ over Q is 2 which is much less
than its.upper bound L4., .
Self Assessment Question 17.3.4: Find the splitting fields of x" + 3 and X2+ x + lover
Q and show that they are the same. . \'

17.4 BASIC ISOMORPHISMS ~ UNIQUENESS OF SPLITTING FIELDS

In the construction of the splitting fields of a polynomial f(x) over a ijeld F, we
observed that it is possible that one may obtain more than one splitting field for f(x). So
our immediate concern is to see whether any two splitting fields of a polynomial are
related to each other in any way, if so, what is that relation ? In fact we shall show that
any two splitting fields of a polynomial are isomorphic under an isomorphism-which
fixes every element of F. To achieve this we need some basic isomorphism theorems on
polynomial rings and their quotient rings.
 Distance Education 8 )E~~== Acharya Nagarjuna University

Let F and FI be two fields and ~ and isomorphism between them. For
convenience, we shall denote the image ~ ( a ) of an element a in F under ~ by a I.
Suppose F[x] and FI [t] are respectively the polynomial rings over F and Fl and
f(x)=ao+alx+a2x2+ +anxnE F[x]. Then fl(t) = u~ +u~t+u~t2+ +u~tn E F[t].
This suggests a natural map Z* : F[x] ~ FI[t] , which is defined by ~* ( f(x) ) = fl(t). In
fact ( is an isomorphism, called the isomorphism induced by ~ and we prove this in the
following theorem.
Theorem 17.4.1: The map s* : F[x]
~ FI[t] given by
((ao+al x+a2x2+ +anxn)= a~+a:t+u~e+ +a~tn
is an isomorphism such that ~* ( a ) = al for all a E F.
Proof: Let f(x) = ao + al x + a2 X2+ + an x" and g(x) = So + SI X+ S2 X2+ +
13mx'" be any 'two polynomials in F[x]. Suppose that m ~ n. Then
f(x) + g(x) = ( ao + 130) + ( al + 131)x + ( U2+ 132)x2 + ..... ; + ( am + 13m)Xm +
am+l X m+l + ..... ... + an Xn 74 1
]...

and
_ 2 s
f(x) g(x) - Yo+ YIX + Y2X + + YsX , 17.4.2
r
where Yr= LUj f3r-i and s = m + n .
i=O

Since (ai+ f3i)1= s( ai+ f3i) = S(ai) + S(f3i) = u} +p}

s* (f(x) + g(x)) =.( ao + (30)1 + (al + 131)1 t + (U2 + (32)1 t2 +. + (am + f3m)1t'" +
. a Im+l tm+1+ ..... ... + anI t"

*
=~ (f(x) + ~* (g(x» .
Also, since ~ is a ring isomorphism,

y; ~ C(y,) ~ C(to a; p,-~~ to C(a;)C(P,-l)

17.4)

From 17.4.2 and 17.4.3 we get

. S*(~()
Il X g (»=
X 1 1
Yo+y1t+ ,12
12t + ....·....·....Ysts
 = f''(t) gl(t) .
So (: F[x] -7 FI[t] is a ring homomorphism.
If f''(t) = 0!)+ Bit+ B1t2 + + B~nenE pl[tJ then B}E p1 for 1 ~ i ~m. Since S:
F -7 FI is onto there exists 0i, 1 ~ i ~m, in F such that S ( 0i ) = 0iI . Now the polynomial
00 + ~l X + .....+ 0m x" E F[x] is such that ( (00 + 01x + 02X2 +....+ Smxm ) =
1 1 I 2 n.1 m
00 + 01t + 02t + + Prn t .
So s* is onto.
For f(x) = ao + al x + a2 X2+ + an x" ,
g(x) = 00 + 01x + 02 x2 + + 0m x" in F[x] and m ~ n, let
( (f(x) ) = ( ( g(x) ). Then
1 1 12 1 n _ n.1 Al n.12 n.1 m
an +a1t+a2t + +ant - fJO+fJ1t+t-'2t + · ·+fJmt ,

or, ah = '0~), a; = 0L , a~ = 0~, a~+1 = a~+2 = = a~ = a .

Since S is one - to - one, these imply that ao= 00 , al = 01 , .... , am = 0m , am+l = 0 ,
........ , an = a . This implies that f(x) = g(x) , so that s* is one - to - one.
These establish that ( : F[x] -7 FI [x] is an isomorphism.
Thus we have seen that an isomorphism ~ : F -7 pI induces an isomorphism ~*:
F[x] -7 r' [t]. Suppose p(x) = ao + al x + a2 X2+ + an x" is an element ofF[x]
and p'(t) = ail +a)t+a1t2 + +a:/1 is its image of under (. If( p(x)) and (pI (t)) are
the principal ideals generated by p(x) and pl(t) respectively, we shall show that ( (and
hence S ) induces an isomorphism (* from F[x] / (p(X) ) onto FI[t] / (pl(t) ) .
Theorem 17.4.2: The mapping (* : F[x] / ( p(x) } -7 F! [t] / ( pl(t) ) defined by ~**(f(x)
+ (p(X))=fl(t) + (pi (t)) is an isomorphism such that (* (Ir-) = al and ~** ( x + (P(x))=t+(
pl(t)). .
Proof: One- can easily see that s** is well defined. Let f(x) + ( p(x) ), g(x) + (p(x) ) E

F[x] I (p(x) . Since s* (f(x) ).= f''(t) , (* is also defined by,

(* (f(x) + (P(x)) = ( (f(x)) + (pl-(t) ).
So (* { [ (f(x) ) + (p(x)) + g(x) + (p(x) ) ] }
= s** [f(x) + g(x) + (p(x) ) ]
= s* ( f(x) + g(x) ) + ( pi (t) )
= ( ( f(x) ) + r., * (g( x) ) + ( pI
(t) )
( since ( is a ring homomorphism)
= f I (t) + gl (t) + ( pi (t) )

= f I (t) + ( pl(t) ) + s' (t) + ( pi (t) )

 Distance Education 10 .,e~~::Acharya Nagarjuna University

= c,**(f(x) + ( p(x) ) + c,**(g(x) + (p(x) )

In the same way, using th~ fact that~* is a ring homomorphism we can show that
~**[(f(x) + (p(x) ) (g(x) + (p(x) ) ]
= c,**(f(x) + (p(x) ). (* (g(x) + (p(x) ) )
Thus l:,** is a ring homomorphism. One - to - one and onto ness of (* follow from the
. corresponding properties of c,*.
The following theorem is a first step towards proving that any two. splitting fields
of a polynomial are isomorphic.
Theorem 17.4.3: If p(x) is irreducible in F[x] and if v is a root of p(x) then F(v) is
isomorphic to F' (w) where w is a root of p' (t). Moreover this isomorphism cr can be
chosen such that o(v) = wand o(o.) = a' for every a E F.
Proof: Let p(x) be an irreducible polynomial over F and vbe a root of p(x) in some
extension K of F. Let
M = { f(x) E F[ x] / f(v)
0} =

Since p(v) = 0 , p(x) E M and M -t o . It is easy to verify that M is an ideal of F[x].

Further p(x) is irreducible implies that M = (p(x) ), the idea! generated. by (prx) is a
maximal ideal of fIx] , and thus F[x] / M is a field.
Consider the map \1' : F[x] -7 F[v] is given by
\v (f(x) ) = f(v) .
It can be seen that \v is an onto homomorphism with kerneIM. So, by the fundamental
theorem of ring homomorphism F[x]/M ~ F(v) under an isomorphism \IJ * where
\j/ ( f (x) + M ) = f(v) .

From this we get \/f *( x + M ) = v .

Since p(x) is ilTeducihld\in F[x] , the polynomial.p' (t) is also irreducible In F' [t].
So, if w is a root of p' (t) in some extension K' of F', then, as in the above there exists
an isomorphism 8 : Fl [ t ] / MI -7 FI(W) given by 8 ( fl(t) = fl(W), with kernel MI, where
MI = ( p'(t) ), the ideal generated by p'(t). There exists an isomorphism 8*: F' [t] / M'
-7 F' (w) , given by
.8*,(f (t) + M
r "
) = f (w), where, M " = (p (t)). Further 8 *,(t+ M ) = w.
Also, by the theorem 17.4.2 , there exists an isomorphism
c,**.:F[x]/M-7F'[t]/M' definedby c,**(f(x)+M)=f'(t)+M'.
 F'(W)

f' (w)

e*

F[x] / M
\

Consider the map rr == e * ~ ** ( \If * )-1 .

r
Since (\If* l : F(V) ~ F(x) / M; ~** : F[x] / M ~ F I [t] / M I and o' : F I [t] / M' ~
F I (w), itfollows thato : F(Y) ~ F I (w). Also since \If* , ~** , o' are all isomorphisms, a
is also an isomorphism. Further -
cr(v) = [ e* ~** ( \If* r 1
( l) )

= e* (* ( ( \If * r1 ( u ) )
= e* ( (* (x + M ) ) , (sirice \V * ( x + M) = u)
•• ' . I
=8 (t+ M ) = w.
Also, for any a E F, ,
cr(a) = e * ( (* ({ \If * r 1
( a ))) = e* ( (* ( a + M),,)
=B *( a I +M i) =a I
;'.
Thus o is the required isomorphism.
Corollary 17.4.4: Ifp(x) E F[x] is irreducible and if a, b are roots ofp(x) then F(a) is'
isomorphic to f(b) by an isomorphism which takes a onto b and which leaves every
element of F fixed. '
Proof: In the Theorem 17.4.2 take F I = F and ~: F ~ F as the identity map i. Then
F I [t] = F[x]. If a and b are the roots ofp(x), then 6- : F(a) ~ F(b), where cr =
r'
o' u ( \If* is required isomorphic such that
o(a) = b
and o(cc) = a I = ~(a)
= i(a) = a. So.o fixes every element of F.
 Distance Education 12

We have every thing ready to prove that any two splitting fields of a polynomial
over a field F are isomorphic.
Theorem 17.4.5: Any two splitting fields E and E' of the polynomials f(x) E F[x] and
f" (t) E F' [t] , respectively, are isomorphic by an isomorphism ~ with the property that
~(a) = a' for every aE F. In particular, any two splitting fields of the same polynomial
over a given field F are isomorphic by an isomorphism leaving every element of F fixed.
Proof: Suppose [ E : F] = n. We prove the theorem by using inducting on n.
If [E:F] = 1 , then E = F and f(x) splits into the product of linear factors over F' .
So E' = F' and the isomorphism S : F -7 F' serves the purpose in this case.
Assume that the theorem holds for all fields Fo and the splitting field Eo of f(x) in
Fo[x] such that [Eo: Fo] < n .
Let p(x) be an irreducible factor of f(x) over F, which is of degree r > 1. Since
f(x) splits over E into the product of linear factors, p(x) also splits into the product of
linear factors over E.
Let v be a root of p(x) in E. Consider F(v). Since deg p(x) = r, [ F(v), F] = r Put
F(v) = Fo. Then
[E: Fo] = [E : F] = ~ < n .
[Fo : F] r
Since F c Fo it follows that f(x) E Fo [xl So E is also a splitting field of f(x) considered
as a polynomial over F0 .
Let w be a root of p' (t) in E'. If we put F' (w) = F' 0 , then by the Theorem
17.4.2 , there is an isomorphism c : F(v) -7 F' (w) such that cr(v) = w and c (a) = a' .
Thus we have two fields F0 and F' 0 such that F 0 ~ F ' 0 under the isomorphism o :
E , E I are respectively the splitting fields of f(x) E Fo(x) and f' (t) E F' o[t] and [ E : Fo ]
< n. So by the induction hypothesis E is isomorphic to E 'under an isomorphism, say <l> ,
such that ~(a) = o(a) for all a E Fo. Since F c Fo we have ~(a) = cr(a) = a' for all a E
F. This ~ is the required isomorphism between E and E' .
In particular if EI and E2 are two splitting fields of the same polynomial f(x) E
F[x], then taking F' = F , s = i, the identity map on F in the above, we get ~ as the
isomorphism between E\ and E2 such that <l>(a)= a I = sea) = i(a) = a , for all a E F.
Example 17.4.5: Let p(x) be an irreducible polynomial in F[x] and V a root of p(x) in
some extension of F. Show that
(1) M = { f(x) . f(v) = 0 } is an ideal of F[x] , and
(2) M = ( p(x) ) .
Solution: (1) f(x), g(x) E M and rex) E F[x]. Then f(v) = 0 and g(v) = O. So
f(v) + g(v) = 0 and rev) fey) = rev). 0 = 0
This shows that f(x) +g(x) E M and rex) {(x) E M ...Similarly f(x) rex) E M.
So M is an ideal of f(x).
(2) since p(v) = 0 we have p(x) E M, or, (p(x)) c M.
On the other hand M being an ideal of F[x] and F[x] is a PID,
M = (g(x))
For some g(x) E F[x]. So p(x) E ker ~ => p(x) = g(x) hex) for some hex) E F[x]. Since
p(x) is irreducible over F, hex) must be a constant, so that
(, p(x) ) = ( g(x) ) = M.

·17.5 : SUMMARY: In this lesson we have studied the following:

1. If f E F[x] and K is an extension ofF, then for any a K
f(x) = ( x- a) q(x) + f(x) ,
where q(x) E K[x] and deg q(x) = deg f(x) - 1.
( This is called the r&hainder theorem ).
2. A polynomial of degree n over a field F can have utmost n roots in any
.extension of F.
3. If p(x) is an irreducible polynomial in F[x], of degree n ;:::1 , then there is an
extension E of F such that [ E : F] = nand p(x) has a root in E.
4. If f(x) is a polynomial of degree n 2: 1 , over a field F, then there exists an
extension E of F, of degree atmost n! in which f(x) has Ln roots.
5. Any two splitting fields of a polynomial in F[x] are isomorphic.

17.6: TECHNICAL TERMS:

Root of a polynomial
Splitting field of a polynomial

17.7 MODEL EXAMINATION OUESTIONS :

Part A : Short answers
4
1. Obtain two extensions of R such that each of them contains a root of X -
5x2 + 6. Find the degree of each of them over R.
2. Find the splitting field and its degree of extension of the polynomial x3 - 2 over 0
the field Q of rational numbers.
Part B : Long answers
1. Show that a polynomial of degree n over a field can have atmost n roots in any
extension of F.
 ~D~i~st;an;c~e~Ed;u;ca~ti~on~==~======~
14

2. Ifp(x) is an irreducible polynomial in F[x], of degree n ~ 1 , show that there is an

extension E ofF such that [ E : F] = n andprx) has a root in E.
3. If f(x) is a polynomial in F[x] of degree n ~ 1 , show that there exists an extension
. E of F such that [ E : F] ~ n arid f(x) has a root in E.
4. Let f(x) E F[x] be of degree n ~ 1 . Show that there is an extension E of F degree
atmost n! in which f(x) has n roots.
5. If p(x) is irreducible in F[x] and if v is a root of p(x) then show that F(v) is
isomorphic to F r (w) where w is a root of p I (t), and this isomorphism o can be
chosen such that cr(v) = w and cr(a) = a I for every a E F.
6. Show that any two splitting fields of a polynomial in F[x] ( where F is a field ),
are isomorphic. .

17.8 ANSWERS TO SELF ASSESSMENT QUESTIONS:

Self Assessment Question 17.2.9 :Show that the map \f': F[x] -7 F[x] / ( p(x) ) defined
by
\f' (f(x)) = f(x) + (p(x)) is an onto homomorphism with kernel Lp(x)).
Solution: Denote V = ( p(x) . Let f(x) , g(x) E F[x] / V. Then \jf(f(x) = f(x) + V and
\jf(g(x) ) = g(x) + V.
So by coset addition and coset multiplication we have
\jf[f(x) + g(x) ] = [ f(x) + g(x) ] + V
= [ f(x) + V ] + [g(x) + V ]
= \If ( f(x) ) + \If ( g(x) )

and \jf[f(x) g(x)] =[ f(x) g(x) ] + V

= [ f(x) + V ] + [g(x) + V ]

= \jf Cf(-x) ) . \v ( g(x) )

So \jf is a homomorphism. It is evident that \If is onto.
f(x) E ker \If ¢:> \If (f(x)) = V, the zero element ofF[x] / V .
¢:>f(x) + V = V
¢:> f(x) E V
So ker \jf = V = (P(x) ) .
Self Assessment Question 17.2.10: Show that the map \f' : F -7 F given in the
Theorem 17.2.6, which is defined by I.f'(a) = a + V ,for all a E F is an isomorphism.
'Solution : Here F = { a + V / a E F } .
\ .As in\ \ i'S.A.Q. 17.2.9 one see that \jf is an onto homomorphism.
. .
" '\\ \ For a, ~ E F suppose \jf(a} = \jf(~} . Then
 0,+ V = ~ +V so that a - ~ E V = ( p(X) ). So
a - f3 = f(x) p(x) for some f(x) E F[x].
Since p(x) is not a constant and since a - f3 is a constant, we must have f(x) = O,so that a .
°
- f3 = or a = f3 .-
°
Se must have f(x) = 0, so that ex - f3 = or a = f3 .
So \1' is one - to - one and hence an isomorphism.

Self Assessment Question 17.3.4 : . Find the splitting fields of X2+ 3 and X2+ x + lover
Q and show that they are the same.
Solution: The root of x2. + 3 are .fj i and -.fj i. Consider Q(.fj i). Since.fj i
satisfies the irreducible polynomial X2+ 3.
[ Q( .fj i) : Q ] = 2.

Further Q( .fj i) is a field and .fj i E Q(.J3 i) implies that -.fj i E Q(.fj i) . This shows
that Q( .fj i) is the splitting field of X2+ 3 over Q.
. 2 . '-1 +.fji -1-.fji
Next consider x + x + 1. Its roots me a = . and ~ = consider
2 2
Q(a) . Since X2+ x + 1 is irreducible over Q, [ Q(a) : Q] = 2.
Further Q(a) is a field and a E Q(a) implies that a-I E Q(a) , or, f3 E Q(a). So Q(a) is
the splitting field of X2+ x + 1.
Since ~ E Q, J3 i E Q(a) , we have a E Q( J3 i), or Q(o') c Q( J3 i).
Also a, f3 E Q(a) implies that a - f3 E Q(a) , or, .fj i E Q(a) , so that Q( .J3 i)
c Q(a).
So Q(a) = Q(.J3 i) , (x
Or the splitting fields of X2+ 3 and X2+ x + lover "Q1areone and the same.

17.9 EXERCISES:
1. For the polynomial f(x) = X4+ X2+ 1, obtain an extension of R which contains a root
off(x). What is its degree over R?

2. For the polynomial f(x) = x5 + 2x3 - 3x2 - 6, obtain two extensions of R

. .'.
SUGh
(1'
that
each of them contains a root off(x). Find the degree of each of them over R.
3. Find the degree of splitting field of the polynomial x5 - 3x3 + 'X2over Q.
 Distance i=ducation 16 ~~~:: Acharya Nagarjuna University

(Hint: f(x) = x5 - 3x3 + X2 - 3=( X2 - 3 ) ( x3 + 1 ), . So the roots of f(x) over Care ± -fi ,
_ I , 1 + j-fi . The splitting field of f(x) over Q is Q (± -fi , _ 1, 1 + i-fi ). Observe that
2 2
K = Q ( .fj , i) and [ K : Q ] = 4.
'4. For each of the following polynomials in Q[x] find the splitting field and its degree
over Q.
(i) X4 - 2 (iv) x5 - 1
3
(ii) x - I (v) XO +1
(iii) X4 + 1 (vi)x6-1
2rri

( Ans : (i) Q( 1J2 ,i), 8 . (ii) Q ( e 3 ), 2

hi

(iii) Q( 12, i) ,4
".
(iv) Q ( e 4 ), 4 (v) Q( J3 , i) , 4
(vi) Q( 13, i), 2.)
5. Show that splitting field E of x" - lover Q is the same as that ofx4 + x + lover Q and
[E: Q] = 2 .
6. If P is a prime number, find the splitting field of XO - lover Q.
2 n:i
p1
(Hint: In C , the roots of x" - 1 are 1, w, WW , ... , w · , where w = e P So the splitting
field K of x" - lover Q is given by K = Q( 1, W, w , .: .... , wp-I
2
).Observe that K = Q(w)
xP -1
. Next, observe that w is a root of -- = ] + x + X2 ..+ xp-1 which is irreducible
+ ...--:
x-I
over Q.,So [ Q(w) : q ] = p - l.

17.10: REFERENCE BOOKS:

1. ARTIN, EMIL, Galois Theory, 2nd ed Notne Dame, Ind : University of Natne Dame
Press, 1955.
2. FRALEIGH, lB., A First Course in Abstract Algebra. 2nd ed., Addision Wesley
Publishing Company.
3. HERSTERIN, IN., Topics in Algebra, Vikas Publishing House. JACOBSON, N.,
Lectures in abstract algebra. Vol.II, Nostnand, 1953.
4. BHATTACHARYA, P.B., JAIN. S.K., NAGPAUL.S.R .. Basic Algebra, (2nd
Edition), Cambridge University Press.
Lesson Writer
Prof. L. N agamuni Reddy
S. VUniversiry, Tiruptti
 LESSON 18 : MULTIPLE ROOTS AND SIMPLE
EXTENSIONS
CONTENTS
Objectives
Structure
18.1 Introduction
18.2 Multiple Roots
18.3 Simple Extensions
18.4 Summary
18.5 Technical terms
18.6 Model Examination Questions
18.7 Answers to Self Assessment Questions
18.8 Exercises
18.9 Reference Books

OBJECTIVES:
After reading this lesson you should be able to :
• Know about the derivative of a polynomial.
• Know that the derivative of a polynomial f(x) over a field F depends on the
characteristic of F.
• Obtain the properties of derivaties.
" 2S .'' :
• Know about simple and multiple roots of a polynomial.
• Obtain criterion for a polynomial to have simple and multiple roots.
• Know about simple extensions.
• Know about separable extensions.

18.1 INTRODUCTION:
In this lesson we continue the study about roots of polynomials particularly about
multiple roots and apply them to obtain some more properties of extensions. For this
purpose we introduce the concept of the derivative 'of a polynomial and discuss some of
its well known properties, from the algebraic points of view. At the end of this section
 Distance Education 2

the notion of simple extension is introduced and it is established that every. finite
extension of a field of characteristic zero is a simple extension.

18.2 MULTIPLE ROOTS:

Definition 18.2.1: Let f(x) = ao + al x + a2 X2 + ..... + an x" be a polynomial in F[x].
The derivative off(x), written as f" (x), is the polynomial in F[x] given by
f" (x) = al + 2a2 x + 3a3 X2 + ..... +n an xn-1 .

Some strange things happen here depending on the characteristic of F. Suppose

the characteristic of the field F is P f O. Then p a = 0 for all a in F. So the derivative of
x" is p xp-l = O. That is, it is possible that a non - constant polynomial can have
derivative zero depending on the characteristic of F. This also suggests that the fact that
the derivative of a polynomial is zero need not imply that the polynomial is a constant.
The following example illustrates this point.
Example 18.2.2 : If the characteristic ofF is zero and f(x) E F[x] is such that f" (x) = 0,
prove that f(x) =.a for some a E F, that is a is a constant.
Solution: Characteristic of F is zero implies that for any positive integer n and a: E F if n
a = 0 then either n = 0 or a =
Let f(x) = ao + al x + a2 X2 + ..... + an x" E F[x] , n ;:::1 and f" (x) = O. Then
2
al + 2 a2 x + 3 a3 x + ..... +n an X
n-I
=0.
This implies that,
r ar = 0 for 1 .:::;;
r .:::;;
n .
Since r f 0 for 1 .:::;;
r .:::;;
n , it follows that ar = 0 .
Thus we have f(x) = ao E F.
'*
Example 18.2.3: If the characteristic of F is P 0 and f(x) E F[x] is such that f" (x)
= 0, prove that f(x) = g(xP ) for some polynomial g(x) E F[x].

Solution: Let the characteristic of F is p f 0 . Then p is it prime number such that p« = 0

, for all a E F and if n a = 0 for any positive integer n then either a = 0 or pin.
Now f(x) = ao + a! x + a2 X2 + ..... + an x" E F[x] ,n ~ 1 and f' (x) = O. Then
al + 2 a2 x + 3 a3 x 2 + ..... +n an x n-I = 0
or, r ar = 0 ; for 1 .:::;;r':::;;n.
So either o, = 0 or plr , for 1 .:::;;
r .:::;;
n. This shows that f(x) = g(xP ) for some g(x) E F[x] .
Let us now establish some familiar results about derivatives .
. t,

Theorem 18.i.~: Let f(x) , g(x) E F[x] and a E F. Then

(1) ( f(x) + g(x))' = f" (x) + g' (x)
 (2) (a f(x») , = aft (x)
(3) (f(x) g(x) )' = f(x) g' (x) + f' (x) g(x)
Proof: We prove (1) and (3) and leave (2) as an exercise.
Let f(x) = ao + al x + a2 X2 + + an x"
g(x) = ~o + PI X + P2 X2 + + Pm x" ,
m < n , say, and a E F .
(1) f(x) + g(x) =
2
(ao + Po) + (al + P2)X + (a2+ P2 )x + + (am + Pm )x
m
+ am+1 x m+I + +. an x n .
mI
So (f(x) + g(X»1 = (al + PI) + 2( a2 + P2)X + + m( am + Pm )x -
m· ~I
+ ( m+ 1) am+I X + + an x n
-- al + 2 a2 X + . . ... + n an Xn-I + I-'A I + 2 1-'2X
A + . . . . . .. + ml-'m
A Xm-I ,
.= f" (X) + g' (X)

(2) See SAQ 18.2.5 0 .

(3) In view of (1) and (2) it is enough if we prove the result for f(x) = x' and g(x) = xi
for some positive integers i, j . 1 S i s n, 1 s j sm. In this case
(f(x) g(x»' = ( xi xi )' = (xi+j ) ,
j
= (i+j) Xi+ -I 18.2.1
But f" (x) g(x) = ( xi )1 xj = i xi-j xj = i Xl+j-I
and f(x) g' (x) = x ' ( xi )' :;= xi (j xj-I ) = j Xi+j-1 ,
So that
j
f" (x) g(x) + f(x) g' (x) = ( i+j) Xi+ -1 16.6.2

From 18.2.1 and 18.2.2 we get result .,T1F . ,

SAP 18.2.5: Iff(x) E F]x] and a is a scalar, show that ·:t

',' ( a f(x) )' = aft (x) .
Definition 18.2.6: Let K be the splitting field off(x) E F[x] . Let a be a root of f(x} in
K. Then x - a divides f(x) in K[x] .,)f (x - a y
divides f(x), but ( x - ay+l'does not divide
f(x) , that is, ( x - ay
is the highest power of x - a that divides f(x) , then r is called the
multiplicity of the root a. If r = 1 , then a is called a simple root and if r > 1 , then a is
called a multiple root.
Theorem 18.2.7: The polynomial f(x) E F[x] has a multiple root if and only iff(x) and
f' (x) have a non - trivial common factor.
Proof: Let f(x) F[x] and deg f(x) = n> 1 . We may assume that all the roots off(x) lie
E

in F. (Otherwise, we may replace F by an extension K of P, which contains all the roots

of flx) ). Assume that f(x) has a root a of multip licity m > 1. Then
 Distance Education 4

f(X) =(X- a t g(x) ,

for some g(x) E F[x] and (x - a) does not divide g(x) .
Then by the Theorem 18.2.5 (3), we have
f' (x) = m (x - a)m-lg(x) + (x - at g I (x),.
: = ( X - a ) r(x) , (since m > 1 ),
where rex) = m (x - a )m-2 g(x) + ( x - a )m-l g I (x) E F[x] . This shows that (x - a) is a
common factor of f(x) and f" (x) .
On the other hand, suppose that f(x) and f' (x) have a common factor. If possible
assume that all the roots aI, a2 , , an off(x) are distinct. Then
f(x) = ( x - a 1 ) ( X - a2 ) ( x - an ) .
so
n A

f' (x) = L (x - a, ) (x - a2) ... (x= a.) ... (x - an) ,

i=l

where the symbol /\ over a particular term denotes that particular term is omitted. Now
fior any j,·1<·<
-J-n,
f'(aj)=O+O + Il(aj-a,J+O +0 to,
A.+j

since a" a2 ~ ... , an are all distinct. So none of a, , a2 , .... , an is a root of f' (x) and this
shows that f(x) and f" (x) have no common factors. This is a contradiction to the
hypothesis. Thus f(x) must have a multiple root. I
This result has some useful corollaries.
SAQ 18.2.8 : Let f(x) be an irreducible polynomial over F. Then f(x) has a multiple root
if and only if'f''(x) = o.
Corollarv 18.2.9: Suppos~ Jf) E F[x] is irreducible over F.
(1) If the characteristic-of F is 0, then f(x) has no multiple roots.
(2) If the .characteristic of F is p to, then f(x) has a multiple root only if it is of
the form f(x) = g(xP ) .
n

Proof: Let f(x) = L:aiXi be an irreducible polynomial over a field F, Then f(x) can not
i=O

be a constant.
Further (by SAQ 18.2.8) f(x) has multiple roots
<=> flex) = 0,
n
<=> z)ai Xi-I =0.
;=1

<=> i a, = 0 for 1 ~ i ~ n . 18.2.1

 I

(1) Suppose F is a field of characteristic zero. If f(x) has multiple roots then 18.2.1
gives a, = 0, for i = 1, 2, , n. Then f(x) = ao , or, f(x) is a constant. This is a
contradiction to the fact that f is not a constant.
So f(x) cannot have multiple roots and thusall the roots off(x) are simple.
(2) Suppose characteristic of F is a prime p =t O. Again by 18.2.1 f(x) has multiple
n

roots ¢::> i ai = 0 ¢::> either a, = 0 or p divides i ¢::> f(x) = I>i i X where either a, = 0
i=O

or p divides i. ¢::> f(x) = g(xP) for a suitable g(x) in F[x].

, ' n
Corollary 18.2.10: If F is a field of characteristic p f. 0 , then the polynomial xP - x E
F[x] has distinct roots for n ?: 1 .
n
Proof: Let f(x) = xP - x,n > 1. Then
n n 1
f" (x) = pn xP -1 - 1 = -1 , since pH xP - = 0, as the characteristic ofF is p. This
shows that f(x) and f' (x) has only the trivial cornman factor 1. So ( by the Theorem (
n
18.2.7) xP - x has distinct roots.

18.3 SIMPLE EXTENSIONS

Definition 18.3.1: An extension K of F is called a simple extension if K=F(a) for some
a.e K. If the characteristic of F is zero,' then we shall show that every finite extension can
be realized as a simple extension.
Theorem 18.3.2: If the characteristic ofF is 0 and if a, b are algebraic over F, then there
is an element c E F(a,b) such that F(c) = F(a,b) .
Proof :Let a, b be algebraic over F. Then there exist non-zero irreducible polynomials
f(x) , g(x) in F[x] such that f(a) = 0 and g(b) = O. Let deg f(x) = m and deg g(x) = n. Let
E be an extension of F in which both f(x) and g(x) split completely into the product of
linear factors. Since f(x) and g(x) are irreducible over F and the characteristic of F is 0,
the roots of f(x) as well as those of g(x) are all distinct ( see Corollary 18.2.9).
Let a = ar , a2, ..... , amand b = b., b2, ..... , b, be the distinct roots off(x) and g(x)
respectively. Put a] = a.and b, = b. Then a, f. a for i f. I and bj f. b, for j t- 1.
Since the characteristic of F is 0, it has infinite number of elements. So there is an
element rEF such that
a, + r b, f. a + r b / 18.2.3
c ]1··
lor a. 1, J, 1 < .< d ,... .
_ L n an L::;; J ::;;m.
for, if there is no such r in F, then for every r in F, there exists i , j , 1 ::;;i::;;n
and 2 ::;;j ::;;m such that
 a -a
r = -' - 18.2.4
b-b J

Since 1 :S i:S nand 2 :Sj :S m, 18.2.4 gives only a finite number elements r in F.
This forces F to be finite. This is contrary to the fact that F is infinite.
For r satisfying 18.2.3, put c = a + r b . Our aim is to show that
F(c) =- F(a,b) .
Now rEF and F c F(a,b) imply that r E F(a,b). Evidently a, bE F(a,b) .
So c E F(a,b) and thus
F(C') c F(a,b) 18.2.5
It remains to be shown that F(a,b) c F(c). Let K = F(c) . Since F c F(c) and b
satisfies g(x) over F, it follows that b satisfies g(x) over K also. From rEF c K and c
E F(c) = K we have that hex) = f(c - IX) E K[x]. Now h(b) = f(c-rb) = f(a) =, 0, so that
hex) is also satisfied by b over K. Thus b is a common root of g(x) and hex) over K, or, x
- b is a common factor of g(x) and hex) in K[x]. Further, for j t 1 , h(bj ) = f ( c
- r bj ) = f(a + rb - r bj ) to, since a, t a + r b - r b, ' by 18.2'~3. This implies that none of
(x - bj), j j 1 is a factor ofh(x) over K. Thus x - b is the g.c.d. of g(x) and hex) over K.
So x - b E K[x] so that b e K = F(c) . From b.c , r E FCc)we get that c - rb E F(c) , or, a
c F(c). Again from a, b E F(c) we get that
F(a,b) c F(c) 18.2.6
From 18.2.5 and 18.2.6 it follows that
F(c) = F(a,b)
Corollary 18.3:3: Any finite extension of a field F of characteristic 0 is a simple
extension.
Proof: Let the characteristic of F be 0 and let K be a finite extension of F. Then (by
SAQ 18.3.4) there exist a finite-number of elements a!, a2 , .. '" an in K such that
K = F(a, , a2 , .. '" an ) .
. c

We prove the corollary by using induction on n. lfn = 2, then by theTheorem 18.3.2, the
corollary follows .. 'So let us assume that the result is true Jar all finite extensions E ( /3, ,
/32, , /3t ) of E, which is of characteristic 0 and r < n . Now F (a, , a2
, .. '" an) = F' (an) , where F' = F(ai , a2 , .. '" an-i) and characteristic of F is O. So by
the induction hypothesis
F' = F(b) .
for some b E F' . So, by the Theorem 18.3.2,
K = F(a" a2, .. '" an) = F' (an) = F(b, an) = F(c)
for some c E K .
Self Assessment Question 18.3.4: Let K be a finite extension of F. Then for some
algebraic elements 0,1, 0,2, .... , an in K,
K = F(al, 0,2, .... , an).
Definition 18.3.5 :
(1) An irreducible polynomial f(x) E F[x] is called a separable polynomial if all of its
roots are simple, or, if it has no multiple roots.
(2) A polynomial f(x) E F[x] is called separable if all its irreducible factors are
separable.
(3) Let E be an extension of a field F. An element a in E, that is algebraic over F, is
called a separable element over F if its minimal polynomial over F is separable .
.(4) An algebraic extension E of F is called a separable extension of F if each of its
elements is separable over F.
(5) A field is called perfect if all finite extensions of F are separable.

Example 18.3.6: Show that any field of characteristic zero is perfect.

Solution: Let F be a field of characteristic zero and let E be a finite extension of F. Then
E is an algebraic extension of F.
Since F is a field of characteristic zero, by Corollary 18.2.9, every polynomial
over F has no multiple roots, or, separable. This shows that every element of E is
separable over F, or, E is a separable extension ofF. SO,.Fis perfect.
Self Assessment Question 18.3.7: Let E be a finite extension of a finite field F. Then E
is simple extension ( that is, E =:= F(a) for some ex, in E).
Example 18.3.8: If K is a finite, separable extension of F prove that K is, a simple
extension of F.
Solution: Suppose F is a finite field. So by SAQ 1&'i~,,7, each finite extension of F is
simple. So K is a simple extension of F.
So let us assume that F is infinite. Since K is a finite extension of F , by SAQ
18.3.4, K = F(al, a2, , an ) for some aI, a2, , an E K. Further K is a separable
extension of F implies that the elements a;, a2, .... , an are separable elements of K. So the
minimal polynomials of a/, a2, "" '" an have simple roots.
As in Theorem 18.3.2, first show that, for separable elements a, b over F, F(a,b) =
F(c) for some c E F(a,b) . Then by induction ( as in corollary 18.3 ..3) one can see that K
= F(al, a2, , an) = F(c) for some c E K.

18.4 SUMMARY
In this lesson we have studied the following :
 Distance Education 8

1. The polynomial f(x) in F[x] has a multiple root if and only if f(x) and f' (x) have a
non - trivial common factor.
2. If f(x) in F[x] is irreducible and if characteristic of F is ° then f(x) has no
multiple roots.
'*
3. Iff(x) kF[x] is irreducible and if characteristic ofF is p 0, then f(x) has a
multiple root <=> it is of the form f(x) = g(xP).
4. If the characteristic of F is 0 and if a and b are algebraic over F, then there is an
element c in F(a,b) such that F(c) = F(a,b) .
5. Any finite extension of a field F of characteristic 0, is a simple extension.
6. A finite separable extension of a field F is a simple extension of F.

18.5 : TECHNICAL TERMS

Derivative of a polynomial
Simple root
Multiple root
Simple extension
Separable polynomial
Separable element
Separable extension
Perfect field

18.6 MODEL EXAMINATION QUESTIONS

Part A : Short Answer Questions
°
1. If the characteristic ofF is Ib~ and f(x) E F[x] is such that f" (x) = 0, prove
that f(x) = g(xP ) for some polynomial g(x) E F[x].
2. Show that a polynomial f(x) in F(x) has a multiple root if and only iff(x) and
f" (x) have a non - trivial common factor.
. n
3. IfF is a field of characteristic p to, then the polynomial xP - x E F[x] has
distinct roots for n ~ I .
4. Show that any field of characteristic °
is perfect.
5. Show that a finite extension of a finite field is a simple extension.
Part B : Long Answer Questions
1. Suppose f(x) E F[x] is irreducible over F show that
(i) if the characteristic of F is 0, then f(x) has no multiple roots, and
 (ii) if the characteristic of F. is p to, then f(x) has a multiple root only if itis of
the form f(x) = g(xP ) .
2. If the characteristic of F is 0 and if a, b are algebraic over F, then show that
there is an element c E F(a,b) such that F(c) = F(a,b) .
3. Show that any finite extension of a field F of characteristic 0 is a simple
extension of F.
4. IfK is a finite, separable extension ofF, prove that K is a simple extension.

18.7: ANSWERS TO SELF ASSESSMENT QUESTIONS:

Self Assessment Question 18.2.8 : Let f(x) be an irreducible polynomial over F. Then
f(x) has a multiple root if and only if f''{x) = O.
Solution: Let f(x) be an irreducible polynomial over F. A root a of f(x) is a multiple
root off(x) if and only if it is a root of'F'(x).
Because f(x) is irreducible, a-I f(x) is the minimal polynomial of a over F, where a
is the coefficient of highest degree term of f(x).
Now f(x) has a multiple root a q fl(a) = 0 q a is a root of f''(x).
Since deg f''(x) < deg ( a-I f(x) ) and since a-I f(x) is the minimal polynomial of a,
f''(«) can not be zero unless f''(x) = o. So f(x) has a multiple root Q fI(x) = O.
Self Assessment Question 18.3.4: Let K be a finite extension of F. Then for some
algebraic elements aI, a2 , .... , an in K,
K -=.~ F( a I , a2 , .... , an ).
Solution: Let K be a finite extension of F and let [ K : F ] = n. Let {a[ , a2 ,.. an} be a
basis of Kover F. Then every a E K is uniquely written a~
a = al a[ + a2 a2 + + an an,
for some a., a2 , ... , an in F.
Consider the field F(a], a2, ..... , an ) got by adjoining at, a2, ... , an to F.
Evidently
F(a], a2, , an) c K .
Also ai, a2, , an E F (a[, a2, ..... , an) and F c F(at, a2, ..... , an) imply that
a, al + a2 a2 + + an an E F(at, a2,· , an), or,
a E F(at, a2, ..... , an) , for all a E K.
So K c F(al' a2, , an ).
Hence K = F(a[, a2, , an ).
Further K is finite extension of F implies that K is an algebraic extension of F. So
ai, a2, ..... , an are algebraic over F.
 Distance Education 10

Self Assessment Question 18.3.7: Let E be a finite.

Solution: Let F be a finite field and let E be a finite extension of F. Let [ E:F] = n and
let { aI, CJ:2, ..... , an } be a basis of E over F. Then every element in E is linear
combination of aI, a2, ..... , an over F. Since F is finite, this implies that E is also finite.
So the multiplicative group E* = E - {O} of E is cyclic. Let a be a generator of E*
We claim that E = F(a) .
Since F c E and a E E, we have F( a) c E. Further, if a E E, a '* 0, then a E E*
Since a is a generator of E* , a = ar for some positive integer r. N ow a E F( a) and F( a)
is a field af E F(a) , or, a E F(a). So E c F(a). Hence E ~ F(a).

18.8 : EXERCISES
1 For each of the following fields K, find a in K such that K = Q(a).
(i) K = -J2 , J3 )
Q( (ii) K = Q( i, -J2)
(iii) K = Q( 12', V5 ) (iv) K = Q( w, 12), where w is the cube root of unity.
,

18.9 REFERENCE BOOKS

1. ARTIN, EMIL, Galois Theory, 2nd ed Notne Dame,lnd : University of Natne
Dame Press, 1955.
2.' FRALEIGH, lB., A First Course in Abstract Algebra. 2nd ed., Addision Wesley
Publishing Company.
3. HERSTERIN, IN., Topics in Algebra, Vikas Publishing House.
4. JACOBSoN, N., Lectures in abstract algebra. Vo1.lI, Nostnand, 1953.
5. BHATTACHARYA, P.B., JAIN. S.K., NAGPAUL.S.R., Basic Algebra,
nd
(2 Edition) , Cambridge University Press.

Lesson Writer"
Prof L.Nagamuni Reddy
, S.V.University, Tirupati
 LESSON - 19 : GEOMETRIC CONSTRUCTION
OBJECTIVES
STRUCTURE
19.1 Introduction
19.2 Constructible Points
19.3 Constructible Numbers
19.4 Criteria for Constructibility
19.5 Impossibility of Some Geometrical Construction
19.6 Summary
19.7 Technical terms
19.8 Model Examination Questions
19.9 Answers to Self Assessment Questions
19.10 Exercises
19.11 Reference Books

Objectives:
After reading this lesson you should be able to :
• Know about constructible points in a plane.
• Obtain criteria for constructability
• Establish that it is impossible to
""J \
(i) trisect an angle. d ..

(ii) To double a cube ( that is to constfu:tt cube whose volume is double

the volume of a given cube ).
(iii) To square a circle ( that is, to construct a square having the same
area as that of a given circle ).
(iv) Construct regular polygons with n sides for certain n.

19.1 INTRODUCTION:
Construction problems in geometry have been a favourite subject for
mathematicians since antiquity. From our high school geometry, we are all familiar
about a variety of geometrical constructions that can be performed by ruler and compass
alone. For example a line segment or an angle may be bisected, a line may be drawn
from a point perpendicular to a given line, a regular hexagon may be inscribed in a circle,
by using ruler and compass alone.
 Distance Education 2

Of all the construction problems in geometry, that of constructing a regular

polygon of n sides with ruler and compass alone, has evoked greatest interest in
mathematicians. Over the centuries, it has been established the possibility of the
construction for n = 3 , 4, 5 and 6. But the solution was eluding for n = 7, that is, for a
regular septagon. There are three other famous Greek problems in geometry for which
the solutions have been eluding:
(i) To trisect an arbitrary angle
(ii) To double agiven cube ( that is to volume is double the volume
of a given cube. )
(i) To square a circle ( that is, to construct a square having the same area as
that of a given circle ).
In all these constructions we are allowed to use the ruler and a compass alone.
The attempt to solve these problems lead to the most remarkable development in
mathematics: After a futile search to solve these problems, a suspicion grew in the minds
of mathematicians whether these problems are led to the question : How to solve that
these problems can not be solved?
The question of proving the impossibility of these geometric constructions is
solved by the application of the algebraic results. Using results obtained so far in field
theory, we shall be able to prove the impossibility of trisecting an angle, doubling a cube,
squaring a circle and constructing a regular septagon, by ruler and compass only.
Before actually investigating these problems, let us first understand what we mean
by construction with ruler ( straight edge) and compass. In Euclidean geometry we use
ruler to construct a line passing through two given points. In the same way we use
compass to construct a circle with a given point as centre and the length between two
given points as radius. Basing 'en these facts let us make some formal definitions.

19.2 : CONSTRUCTIBLE POINTS:

Definition 19.2.1 : Let S be a set of points in the Euclidean plane containing atleast two
points. A line is said to be constructible by ruler and compass, or simply constructible
from S if it passes through two distinct points of S. A circle is said to be constructible
from S, if its centre is a point of S and its radius is the distance between two points of S.
Definition 19.2.2: A point in the plane is said to be constnictible from S if
(i) .It is the point of intersection of a pair of distinct lines which are constructible
from S, or,
(ii) It is the point of intersection of two distinct circles which are constructible from
S; or, .
(iii) It is the point of intersection of a line and a circ.c which are constructible from S.
Lemma 19.2.3: Let s(S) denote the set of all constructible points from S. Then S c
s(S).
Proof: Let PES. Since S contains atleast two points, there exists a point Q in S,
different from P. So the circle with as centre and P Q as radius is constructible, as also
the line I passing through P and Q. SOthe points of intersection of this circle and the line
I are constructible and P is one of these two points. So P is constructible, so that P E
s(S). That is, S c s(S). ---
Let us take a rectangular coordinates system for the plane and let X' OX and
y' OY respectively denote the x - and y-axes. In the plane let us take the set Eo =
{ 0 = ( 0, 0 ) , I = ( 1, 0 ) } . We call 0, the origin and I, the unit point. Let EI = s(~ ) .
From Lemma 19.2.4, we have that Eo c EI . Further we observe the following:
(i) The line X' OX passes through the points 0 and I of Eo . So X' OX , or x - axis, is
constructible from Eo. Then length of 01 is 1. So the circle with centre 0 and radius
1 is constructible from Eo . Since this circle intersects X' OX at I' = (-1, 0 ) , the
point I' E EO
(ii) The circle with centre I and radius 1 is constructible from Eo. This circle cuts the x
- axis, in A = (2, 0). So the point A is constructible from Eo. In this way we can
see that the point (n,O) is constructible from Eo , for every integer n.
Inductively define En+I = seEn ) for every positive integer n. Then En c En+I. Let H =
U En , where the union is taken over all positive integers. We say that a point P of the
n

plane is constructible if P E H ..
Lemma 19.2.5: S(H) = H.
Proof: Let 1.be a line constructible from H. Then I passes through two distinct points,
say, P, Q ofH. Since H = U En, P E Em, Q E En·(or some m andn. Since m and n are
n

positive integers, either m::;; n or n ::;;m. For definiteness, let m ::;;n. Then Em c r, and
P, Q E En. So I is constructible from En-I. Next suppose that C is a circle constructible
from H. Then the centre R of C, and the points A., B such that AB is the radius of C, are
in H. So for some positive integers m, n , r, A = En , B E Em and R E Er. If m ~ n ~ r,
then A, B, R E Em and the circle C is constructible from Em-I. From this, it follows that
every line constructible from H and every circle constructible from H, are constructible
from some Em .
Let P be a point of the plane constructible from H, that is, P E s(H) . Then ibythe I
Definition 19.2.2, P is the point of intersection of G1 and G2 , where G1 and G2 are \
d
constructible lines or circles from H. So G1 is constructible from some Em and 2 from \
some En. Again either Em ~ En , then both G] andG, are constructible from En. Then P ~
E En+1c H, and s(H) c H. By Lemma 19.2.3 , H c s(H) . Thus H = s(H).

'\ ji
 Distance Education 4

Definition 19.2.6: We say that a point P of the (Euclidean)plane is constructible ifP E H.

Definition 19.2.7 : A line or a circle in the plane is said to be constructible if it is
constructible from H.
Theorem 19.2.8: If P is a constructible point lying on a constructible line 1, then the line
through P and perpendicular to 1 is constructible.
Proof: 1 is a constructible line through the constructible point P. So there exits a
constructible point Q =F P on 1. The circle with P as centre and PQ as radius, intersects 1 in
another constructible point T. Now the circles with Q andT as centres and radius QT are
constructible . So their point of intersection R is constructible. Finally the line PR IS
constructible and this is the line which passes through P and perpendicular to 1.
Theorem 19.2.9 : If a constructible point P does not lie ona constructible line 1, then the
line through P and perpendicular to Iis constructible.
Proof: Let us take a constructible point Q on 1. The circle with P as centre and PQ as
radius is constructible . In general this circle cuts I in two points, and Q is one of these
points. Let T be the other point. Then T is constructible. If T = Q, then the line passing
through P, Q is the required line. If T i- Q, then the circles with Q and T as' centers and
TQ as ra+ius is constructible. So their point of intersection R is constructible. Thenthe
line through P and R is constructible and this is the required line.
o
Theorem 19 2.10 : If the constructible point P does not lie on the constructible line I then
is
the line throuj.h P and parallel to 1 constructible.
Proof: By the· theorem 19.2.9, the line I' through P and perpendicular to 1 is
constructible. Again by the Theorem 19.2.8, the line m through P and perpendicular to
I' is constructible m is the required line.
Self AS!'1..._",t Questions 192.11 : Y-axis is constructible.

19.3 CC~fi.:'.~_:CTIBLE NUMBERS:

n!~~r\!l.iorL19:~~/\ real number a is said to be constructible if (a,O) is constructible.
L T denote the set of all constructible real numbers. Since 0 = ( 0, 0) and 1=
(1,0) are ccuuructibl. , 0, lET.
Them t;ill 19."~.2: .If the real number a is constructible -a is also constructible.
Proof: Suppose that the real number a is constructible. Then, by the Definition 19.3.2,
the point A = (a,O) isconstructible. So the circle with centre 0 and radius A ( = lal ) is
constructible. This circle intersects x _. axis ( besides A ) at the point ( -a, 0) . So (-
a,O) is a constructible point and thus -a is constructible.
Theorem 19.3.3: If the real numbers a, b are constructible then a ± b are constructible.
Proof: Since a, b are constructible, the points A = ( a, 0 ) B = ( b, 0 ) are constructible.
Sot the circle with centre A and radius AB is constructible and this circle outs the x - axis
in the constructible points ( a + b, 0 ) and ( a-b, 0). So a ± b are constructible.
 Theorem 19.3.4 : IF a is a real number, then the point (a,O) is constructible if and only if
(O,a) constructible.
Proof: Suppose that A = ( a, 0) is constructible. Since 0 = ( 0, 0 ) is constructible, the
circle with 0 as centre and the line segment OA as radius is constructible. This circle
cuts the Y-axis ( which is constructible) in the constructible point ( O,a ).
Conversely , if we assume that the point A' = ( 0, a ) is constructible then the
circle centre at 0 and radius, the line segment OA' cuts the x-axis in the constructible
point ( a,O).
Theorem 19.3.5: If a, b are positive real numbers and if they are constructible, then ab
and 1/a are constructible.
Proof: The line passing through A = (0,1 ) and B = ( a, 0 ) is constructible . Also the
line passing through ( 0, b ) and parallel to AB is constructible. This line outs the X -
axis in (a b, 0). So a b is constructible.
Further, the line passing through the constructible point (1,0) and parallel to the
line through the constructible points ( a,O) and (0,1) , is constructible. This lines cuts the
Y-axis in the constructible point (0, 1/a) . So 1/a is constructible.
The following Corollary is immediate from the Theorems 19.3.2, 19.3.5.
Corollary 19.3.6 :The product of two constructible real numbers is constructible. If a is
a non - zero real number, which is constructible, then 1/a is constructible.
In Theorems 19.3.3 and Corollary 19.3.6 we bave established .
. Theorem 19.3.7: The set of constructible numbers is a sub-field of the field of real
numbers.
We have seen that (n, 0) is constructible for every integer n. So every integer is
constructible and by Corollary 19.3.6, every rational number is contractible. Thus we
have proved.
Theorem 19.3.8 : The field of rational numbers is c-9Hstructibte.
We prove some useful lemmas.
Self Assessment Questions 19.3.9: If a is constructible positive real number, then ± Fa
is constructible:
Lemma 19.3.10: The point ( a, b ) is constructible if and only if a and bare
constructible.
Proof: Assume that the real numbers a and b are constructible. Then the points ( a,
0) and (O,b) are constructible. (see Definition 19.3.1 and Theorem 19.3.4). So theline
through (O,b) and parallel to the x - axis and the line through (a,O) and parallel to th'~-
axis are constructible . From this we get that their point of intersection (a,b) is
constructible.
Conversely, let us assume that the point P(a,b) is constructible. Then the line
through P and parallel to y-axis is constructible. This line outs x - axis in the
 Distance Education 6

constructible point (a,O). So the real number a is constructible. In the same way the
constructible line through P and parallel to x .. axis cuts the y-axis in the constIUctible
point (O,b), so that the real number b is constructible.
We now prove a lemma, which plays a crucial role in the coming discussion.

19.4 : CRITERION FOR CONSTRUCTIBILITY :

Lemma 19.4.1: Let F be a sub field of the field of real numbers and let G, and G2 be
such that G, t G2 and G, and G2 are lines or circles whose equations have coefficients
from F. If (x.y) is a point of intersection of G, and G2 then either both x and yare in F,
or, in some quadratic extension of F.
Proof: Case (i). First let us assume that G, and G2 are both lines. Then their equations
are given by
ax + by + C = 0
a I x + b I v+ C I= 0 ,
where a,b, c , a I , b I , CI E F and either a, h #- 0 or b, b I#-O .
If ( Xi , Yl ) is the point of intersection of these lines ( we assume that they are not
parallel) , then
bcl-ble a1c-ae'
x - and y ----
I - ab I _ a I b I -! ab 1 - a I b
Since F is a field, we have x, , y, E F .
Case (ii) Let one of 0, and O2 is a line and other, a circle, say 01 is a line and O2 , a
circle. Then equations of 01 and G2 are given by
ax + by + C = 0 ,it (3)
~ 2 .~?
x" + Y + ex + fy +%'-=0 (4)
where a, b, c, e, f, g E F and atleast one of a and b say a is non - zero. To find the point
, of intersection of G, and G2 let us eliminate x from the equations (I) and (2) and get

l( b: a
+ 1J + y( a
2~C _ bea + f) _eca + ac~ + g = 0 .
That is, al + f3y + Y = 0
where,
'J
b-
C/.., = 2+1,
a
Then

or,
 Y = ~2:
d
,where d =.J~2., 4ay
Here d2 = 132 - 4 ay E F since a,13, y E F.

Thus the two values of y are given by

Yl = -~ +~ and Y2 = -~ -~ and thus Yl , Y2 E F(d) , the field got by

2a 2a 2a 2a
adjoining d to F. Further we observe that

'. [ I if dE F .
[F(d):F ] = 2 if d ~F . .
Substituting lUC::;C varue f Y in (3) we get the corresponding values of x as
c b c b
Xl = -- - - YI and Y2 X2 = -- - -
and these also belong to F(d).
a a a a
Case om: Let G, and G2 be both circles. Then their equations are, given by
X2 + l + ex + fy + g = 0 (5)
X2 + l+e 1
x + f" 4 + g' = 0 (6)
where e, f, g, e' ,f' E F .
Subtracting (3) and (4) we get the line of intersection ofGl and G2 as .
(e-e')x+(f-f')y+(g-g') =0 (7)
Now the point of intersection of the circles G, and G2 are same the points of
. intersection of the circle Gi ( or G2 ) and the line given by (7). Thus this cases reduces to
the case (ii).
Let us pause for a moment and take stock of the results that are established so far.
First let us see what the Lemma 19.4.1 actually c:?~ys. Suppose that we are able to
construct all the numbers of a some number field F. -Then by case (i) of Lemma 19.4.1,
the point ( x,y) of intersectio~ of any two lines' in th~ plane of F, again lies in the plane of
F, or , x, yare again in F. Observe that the lines are obtained with the help of the ruler
alone. This means that the use of the ruler alone will never lead us out of the field F' . To
break the walls of F we have to use compass, In cases (ii) and (iii) of the Lemma 19.4.1
we observed that the points of intersection ofa line and a circle, or, two circles
constructed from .the plane of F lies either in the plane of F or in the plane of some
quadratic extension F(.Jd) of F. (Remember that a compass is needed to construct a
circle), This means that to break the walls of F we have to use compass.
What does this actually mean? Suppose we start with a sub set E of real numbers,
consisting of atleast one number called the 'fundamental unit' 0 and generate a field F by
means of the rational operations of addition, subtraction, multiplication and division. In
 Distance Education 8 fo~S= Acharya Nagarjuna UniVers~

other words the elements of F are constructible Lorn E by means of the above rational
operations. (For example, the field Q of rational numbers is generated by a set which
contains atleast the real number 1). Then Lemma 19.4.1 says that, using ruler alone we
can construct the elements of F only. But using the compass we can extend the field F to
a wider extension field by selecting a number d of F, extracting its square root Jd and
constructing the field F( Jd), consisting of all,the numbers of the form a + b Jd, where
a and b are in F. In effect we have shown that, any step in a geometrical construction (
drawing a line through two known points, drawing a circle with known centre and radius,
or, find the intersection of two known lines or circles) will either produce a new number
already lying in the field F, or, by the construction of a square root, will open up a new
extension field F.
The totality of constructible numbers can now be described with precision. We
start with a given field F0 generated ( by rational operations ) by a set of real numbers
containing atleast one element, which we call the fundamental unit ( for example ; the
field of rational numbers, is generated by a set consisting of atleast 1). Next by
adjoining Fa ' where do is in Fo , but Fa
is not, we construct an extension field FI of
constructible numbers, consisting of all numbers of the form ao + bo Fa, where ao and
bo are in Fa. That is F I = F(, '-Fa) . Then F2 , a new extension field of F I is defined by
the numbers a, + bl ..}k; where a, and b, are in F I , k, is some element of F I and ..}k; is
not in Fl. That i~;F2 = F I (..}k;) = Fa (Fa,.Jk;) . _Repeating this process, or,
inductively, we shall reach a field F n after adjoining n square root Fa, Jk;, ,
-Jkn-1 to Fo , namely,
r, = Fo (Fa, -Jk; ,,"" '" jk::; ),
\.;

where ki-i E Fi-i but .Jki-I ~ Fi-l and F, = Fi-1 ( .Jki-I).

Constructible numbers are those"{Vhich can be reached by such a sequence of extension
fields ; that is, which lie in a field F, of the type described. In other words U Fn gives
n

the set of all constructible numbers got starting with the number field Fa .
The above discussion leads to the following:
Lemma 19.4.2: Let F be a subfield of real numbers. A point P is constructible from F if
and only if there is finite number of real 'numbers d, , d2 , ""., d, such that
(1) [ F( dl) : F ] = 1 or 2 ;
(2) [ F( d1 , d2 , "'" d, ) : F( d, , d2 '" '" di-1 ) ] = I or 2,
fori=1,2,,, '" nand
(3) P lies in the pJane of F(d, , d2 '" '" d, ).
Proof: Suppose that P is constructible form F. Then the discussion preceding the
Lemma shows that we can fmd real numbers dj, d2 , ..... , d, such that
dfEF, d~EF(dj), d~E F(d},d2) , , d~ E F (dj , d2 , .... , dn-i ), and such thatthe point
P lies in the plane ofF(dj , d2 , , dn).

This proves the first part of the Theorems.

Conversely, if kEF is such that .Jk is real, then by SAQ 19.3.9, .Jk is
constructible and therefore every real number in F( .Jk) is constructible. Thus the points
in the plane of F(.Jk) are constructible and this establishes the second part of the
theorem.
As a consequence of Lemma 19.4.2, we now our final theorem, before going to the
applications.
Theorem 19.4.3: Let Q be the field of rational numbers. A real number a is
constructible if and only if we can fmd a finite sequence of real numbers d, , d2 , ..•.. , d,
such that
o
') .
(1) d1 E Q,
(2) df E Q (dj ,d2 , , dj_j), for i = 1,2,.... , n, and
(3) a E Q ( d, , d2, , dn) .

Proof: By the Defmition 19.3. 1, a real number a is, constructible if and only if the point
(a,O) is constructible. ,c

In Lemma 19.4.2, let take F = Q. Then by the Lemma 19.4.2, the point (a,
0) is constructible if and only if it belongs to the plane of the field K, which is obtained
from Q by a finite number of quadratic extensions. This is the contension of the theorem
.Corollar~19.4.4: If a is a constructible real number, then a lies in some extension of
rationals 0 degree 2f .
Proof: / If the real number a is constructible
.
then
. by the
.
Theorem 19.4.3,
a E Q ( d, , d2 , ..... , dn) ,

where [ Q ( d, ) , Q( dj-i ) ] = 1 or 2 , for 1 S; i S; n .

we have
[Q (dt , d2 , .... , d,.') : Q ] = [ Q ( d, ) :.Q ] [ Q(d2) : Q(di) ]. ...[ Q (dn ) : Q(dn-i ) ]

= 2f, for some positive r S; n .

The following Corollary provides us with an important criterion forothe non-
constructability of a real number.
Corollary 19.4.5: If the real number a satisfies an irreducible polynomial of degree m
over the field of rational numbers and if m is not a power of 2, then a is not constructible.
 Distance Education 10
-------- ----
Proof: Suppose that m is the degree of the irreducible polynomial satisfied by the real
number a and m is not a power of 2. Then
[ Q(a) : Q ] = m .
If possible , assume that a is constructible. Then by the Corollary 16.7.23 , we can
find an extension K of Q such that a E K and
[ K : Q ] = 2f , for some positive integer r.
But Q(a) c K. So [ Q(a) : Q ] divides [ K : Q ] , or, m divides 2f. This cannot
happen since m is not a power of a.
So a is not constructible.
Definition 19.4.6: A line segment is said to be constructible if its end points are
constructible. A polygon is said to be constructible if each of its sides is constructible.
An angle is said to be constructible if its vertex and sides are constructible
Lemma 19.4.7: If a line segment is constructible then its length is a constructible
number.
Proof: Let AB be the given line segment with end points A = ( a, , b, ) and B = (a-, b2)
By the Definition 19.4.6, the points A, B are constructible. So the numbers a, , a2 , b, ,
b2 are constructible. Since constructible numbers form a field and since square root of a
constructible number is constructible, it follows that
AB =)(al
- a2)2 + (b] - b2)2 is a constructible number.
We now give a criterion for the constructability of a regular polygon which is
dependent on the following result, which we assume without proof.

R esu It: Let w = cos-2n +1..sin -2n b e a pnrrutrve

.. . nt h root 0f uni
unity. Th en [Q')
(u) :Q] =--=
n n
~(n), where ~(n), denotes the number of numbers less than n and relatively prime to n.

Theorem 19.4.8: A regular n -.gon ( or equivalently an angle 2n ) is constructible if

n· n
and only if ~(n) is a power of 2.

Proof: Let w = cos 2n +i sin 2n, where w IS a primitive nth root of unity. Then
n n
- 21C
w +w = 2cos-.
n
2n
Set u= CDS -
n
_ We know that u is constructible if, and only if,
[ Q(u) : Q ] = 2k for some positive integer k.
 Consider the following tower of extensions :
Q(w)

I
Q(u)

I
Q
Now
2n 2 . 2 2n
(w - COS-) = -sm -
n n
and so
2. 2n"
w - 2cos- w + 1 = 0 ..
n
This shows that w satisfies the polynomial.
. 2 -. 2n
x - (2cos-)x + 1,
n
over Q(u). This is clearly irreducible ovcr'Qiu). So
[ Q(w) : Q(u) ] = 2.
But [Q(w): Q ] = ~(n). That is,
~(n) ='[ Q(w) : Q] =[ Q(w) : Q(u J [ Q(u) : Q] = 2 [ Q(u) : Q]

SO [ Q(u) : Q] = ~(n) .
2
Thus u and hence a regular n - gon is constructible if and only if ~(n) is a power of 2.

19.5 : IMPOSSIBILITY OF SOME GEOMETRIC~L CONSTRUCTION:

We have everything ready to solve the problems which we stated at the beginning
of this section. First let us consider the problem of trisecting an angle by ruler and
compass. ,
Theorem 19.5.1 : It is impossible to trisect an angle by ruler and compass only.
Proof: It is enough if we show that there is a constructible angle, which cannot be
trisected by ruler and compass.
The point 0 and the x - axis (OX) are constructible. Also the point A = (1, J3) is
constructible. So the line through 0 and A is constructible. By the definition of the
constructible angle, AOX = 60° is constructible. \Ve now show that AOX is not
trisectable.
If we could trisect 60° by ruler and compass, then we can construct the line OP
such that POX = 20°. Let this line intersect the circle with centre 0 and radius 1 ( note

21 ~
 Distance Education 12
J

that this Circle is constructible )at D. The pointD and line through D perpendicular to x -
axis.are c~nshuctible. If this line cuts x - axis at E, then DE is constructible.
\\ J So oe 0
= Cos 20 , is constructible. Putting

X = Cos 200 ,

. In the identity
C\o~3 e = 4 Cos ' 8 - ) Cos 8
W,\ <\btal~ \
\ 8 x3 - 6x - I = 0
But 8x3 - 6x - I is irreducible over Q ( prove). Its degree is 3 and it is not a
power of 2. So by Corollary 19.4.5, x is not constructible. 'So our assumption is wrong
and 600 can be trisected by ruler andcompass.· .
The impossibility of 'doubling the cube'
t,
is proved by the use Corollary
I
1914.5 ..
I

Theorem 19.5.2:Give a side of a/cube, it is not always possible to construct with ruler
. and compass the side of a cube which has double the volume of the original cube.
Proof: Let us consider a cube having a side of length 1. Then its volume is 1. The cube
being sought would have to have a volume of 2, and hence a side of length 'Vi. But if
x=V2 t then x3 - 2 = 0 and x3 - 2 is irreducible of over Q. Since 3 is not a power of2, by
Corollary, V2 is not constructible. So the required cube is not constructible .....
We shall show another impossibility of a geometrical construction, namely,
squaring a circle. .
.. -- --
I .

Theorem 19.5.3: Given a circle, it is not always possible to construct with ruler and
'compass, a square having area equal to the area of the given circle.
Proof: Let the given circle have a radius I, and hence an area of 'IT. If a s8.u~re of area 'IT
is constructible, then each: «\f...its sides of length -J;; is constructible.;' ~,}:hen -J;; is
constructible and thus 'IT = ( -J;;)2 is constructible. ~~<
, '. "-
-··------+hi-S--mea1l£...ihat'ITis an algebraic element over Q. But 'IT is transcendental over Q.
This proves the theorem. ----- --- --. .--- ", .
We close this section by establishing another impossibility of geometrical
construction regarding regular polygons. Note that the regular .n-gon 'is' constructible for
n ~ 3 if and only if the angle 2n is constructible, which is the case if-and only if a line
n . "
segment of length Cos (2'IT) is constructible.
n
Theorem 19.5.4 It IS impossible.to construct
'.---,~-
a regul~' .septagon by the ruler and
"<,
compass.
 Proof: A regular septagon ( polygon with 7 sides )is constructible if and only if a=2
Cos ( 2n) is cons~ctible. But. a satisfies the irreducible polynomial
7 . \
~~?+ X2 - 2x - 1
over Q the field of rational numbers and its degree, namely, 3 is not a power of 2. So a is
not constructible and hence a regular septagon is not constructible.

Aliter: Here ~(7) = 6 and 6 is not a power of2. So by Theorem 19.4.8, a regular
septagon is not constructible.

19.6': SUMMARY:
In this lessonwe have studied the following
1. Let Q be the field of rational numbers. A real number a is constructible if and only if
we can find a finite sequence of real numbers' d 1 , d2 , ..... , dn such that
(i) d? EQ,
, .
(ii) dt E Q ( d, , dz, ,.di-I ) , for 1= 1,2, .... , n,'
and (iii) aE Q (dl, d2 ; :,.dn).
- -~ .

2. If a real number a is-constructible then a lies in some extension of rational number of

degree 2f .
3. The following problems are proved:
r. (i) It is impossible to trisect an angle by ruler and compass.
(ii) It is impossible to duplicate acuoe by ruler and compass'.
(iii) It is impossible to' square a circle by ruler and compass' .
(iv) It is impossible to construct a regular septagon by ruler and compass.

19.7 TECHNICAL TERMS :

Constructible points
Constructible numbers

19.8 MODEL EXAMINATION QUESTIONS:

Part - A Short Answers:
1. Prove that if a real number a is constructible then a fies in some extension of.
rational numbers of degree 2.
2. Show that it is possible to construct an angle of 30° by ruler and compass alone.
 Distance Education 14 ..s~~= Acharya Nagarjuna University

3. Show that it is impossible to trisect an angle by ruler and compass alone.

4. Show that a regular 17 - gon and 15 - gon are constructible by ruler and compass.
5. Show that it is impossible to construct a regular 7 - gon and 9 - gon by ruler and
compass.
Part - B Long Answers:
1. Let Q be the field of rational numbers. Show that a real number a is constructible
if and only if there exists a finite a sequence of real numbers d. , d2 , .... , dn such
that,
(i) d? EQ,
(ii) d: E Q ( d, , d2 '" "., di-l ) , for I = 1, 2, "'" n,
and (iii) a E Q ( d, , d2 '" '" dn) .

19.9: ANSWERS TO SELF ASSESSMENT QUESTIONS:

Self Assessment Question 19.2.11 : Y-axis is constructible.
Solution: The points I = ( 1, °)
and I' = ( -1, 0) are constructible . So the circles with
centres I and I' and radius II' are constructible. Thus the points of intersection of these
circles is constructible. So the line joining these points are constructible and this is
nothing but Y-axis.
Self Assessment Question 19.3.9: If a is constructible positive real number, then ± Fa
is constructible.
Solution: The circle (x2 - 1)2 + l = 1 + a, which has the line segment joining (-a,O)
and (1,0) as radius and (1,0) as centre is constructible. This circle cuts the Y-axis in
the constructible points (0, Fa ) and (0,- Fa) . So ± Fa are constructible.

19.10 : EXERCISES:
L Show that a regularpolygon of n sides is constructible if and only if the angle of
2n
radians is constructible.
n
2. Show algebraically that it is possible to construct an angle of 30° .
3. Prove that a regular hexagon is constructible ( Hint: Use exercises 6 and 7 ).
4. Prove that a regular polygon of 9 sides is constructible
(Hint: Use Theorem 16.7.27 ).
5. Prove that it is possible to trisect 72° .
6. Prove that a regular pentagon is constructible.
7. Prove that a regular polygon of IS sides is constructible.
 8. Show that a regular 17 - gon is constructible.

9. Prove that 1t is a constructible, angle and Cos 1t is a constructible number.

5. 5
10. Prove that an angle of e radians is constructible if and only if the real
number r'l)S e ( Sin e ) is constructible.
19.11 REFERENCE BOOKS:

l. ARTIN, EMIL, Galois Theory, 2nd ed Notne Dame, Ind : University of Natne
Dame Press, 1955.
2. FRALEIGH, 1.B., A First Course in Abstract Algebra. 2nd ed., Addision Wesley
Publishing Company. <,

3. HERSTERN, IN., Topics in Algebra, Vikas Publishing House.

4. JACOBSON, N., Lectures in Abstract Algebra. VoUI, Nostnand, 1953.
5. B~TTACHARYA, P.B., JAIN. S.K., NAGPAUL.S.R., Basic Algebra,
(2nd Edition) , Cambridge University Press.
6. COURANT, R. and ROBBINS, H., What is mathematics? Chapter III,
Oxford University Press, London, 1953.

Lesson Writer
Prof L.Nagamuni Reddy
S.V.University, Tirupati
CONTENTS:
OBJECTIVES
STRUCTURE
20.1 Introduction
20.2 Groups of automorphisms and fixed fields
20.3 Symmetric functions
20.4 Summary
20.5 Technical terms
20.6 Model Examination Questions
20.7 Answers to Self Assessment Questions
20.8 Exercises
20.9 Reference Books

OBJECTIVES -c,

'After reading this lesson you should be able to : " )

• Know about the fixed field of a- group of automorphisms of a field l( ; ,.

• Obtain the properties of the ele~ciit-ary. symmetric functions and use them to
derive certain properties of normal extensions; .
• Introduce the notion of Galois group G(K,F) of extension K of a field F..
• Obtain a bound for the order of G(K,F).

20.1 INTRODUCTION
Galois theory gives a beautiful interplay of groups and field' theory. To each
polynomial f(x) in F[x] one can associate a group called the Galois group of f(x). There
is a close connection between theroot of a polynomial f(x} and' its Galois group. This
can be achieved through the splitting field of the polynomial f(x) and infact, the Galois
group of f(x) is the group of automorphisms of the splitting field of f(x). The
fundamental theorem of Galois theory, sets up a one - to - one correspondence between
the subfields of the splitting field of f(x) and the subgroups of its Galois group. ~ing
this duality we derive a criterion for the solvability of a polynomial by means. of radicals,
which in turn establishes the classical result of Abel that a polynomial of degree 5 is not
solvable by radicals.
In this lesson we discuss the automorphisms of a field. K, which fix every element
of a subfield F of K and observe that these form a subgroup of the group of the
 Distance Education 2

automorphisms of K, as also the field fixed by every element of a certain automorphism

group of K. This forms a prelude to the much stronger correspondence between the
subgroups of the Galois group of a polynomial f(x) and the sufields of the splitting field
of f(x), which we are going to establish later in the fundamental theorem of Galois theory
in the next lesson. Incidentally we obtain an upper bound for the order of G(K,F), the
automorphism group of K, which fixes every element of F.

20.2 GROUPS OF AUTOMORPHISMS AND FIXED FIELDS

Definition 20.2.1: Let K be a field. A one - to - one and onto mapping a of K to itself
is called an automorphism of K if
a(a+b) = a(a) + a(b) ,
a(ab) = a(a) o(b) , for all a, bin K.
Two automorphisms a and 't on K are distinct if a(a) "*'tea) for some a in K.
It is easy to see that Aut (K), the set of all automorphisms of a field K is a group
under the product of maps.
By a group ofautornorphisms of K we mean a subgroup of Aut (K).
Theorem 20.2.2: If K is a field and if al , a2 , .... , an are distinct automorphisms of K,
then it is impossible to find elements a, , a2 , ..... , an in K, not all zero such that al al (u)
+ a2 a2(u) + + an an(u) = 0 , for all u in K.
Proof: If possible let us assume that there exist a, , a2 , .: .., an in K, not all zero and
a, a](u) + a2 <Y2(U)+ + an <Yn(u) = 0 (1)
a, <YI(U)+ a2 <Y2(U) + + am <Ym(u)= 0 (2)
for every u E K, be the minimal relation with this property ..
, .
If m = 1, then a] aj(u) =0 , for all u E K , so that a, = 0 and this is contrary to our
assumption. So let m > 1.
N ow the automorphisms are distinct. So there exists an element c E K such that
<YI(C)"*<Ym(c). Since K is a field CUE K for all u E K . Replacing u by c u in (2) and
observing the fact that ai's are automorphisms, we get
al <YI(C)al(u) + a2aic) <Y2(U)+ + am <Ym(c)<Ym(u), 'if u E K (3)
L •

Multiplying (2) by al(c) and subtracting it from (3) we get

a2 <Y2(U)( <Y2(C)-al(c) ) a, <Y3(U)( <Y3(C)- <YI(C)) .... + am <Ym(u)( <Ym(c)-al(c)) = 0
for all U E K.
That is, b2 a2 (u) + b3<Y3(u)+ + bm<Ym(u)= 0, 'if u E K, (4)
where b, = a, ( <YI(c) - al(c)) E K for I = 1, 2, .., m. Since <Ym(c) "* <YI(C), bm "*0.' So (4)
contradicts the minimal property of relation (2) and our assumption wrong and the lemma
is true.
 Subfields of K which are fixed element-wise by every automorhism of an
automorphism group play an important role in our future discussion. Let us study some
relevant aspects about them.
Definition 20.2.3: Let G be a group of automorphisms of K. The fixed field of G is the
set of all elements a in K such that CY(a) = a, for all CYE G.
Even though we called the set of all elements in K which are fixed under every a
in G as the fixed field of G we are yet to check that it is a field. We do it in the
following.
SAQ 20.3.4: Let G be a group of automorphisms of K. Then the fixed field of G is a
subfield of K.
Having defined the fixed field of a group of automorphisms of K, it is natural to
study the set of automorphisms of K which fix every element of a subfield F of K.
Definition 20.2.5 : Let K be a field and F, a subfield of K. The set G(K, F) denotes the
set of all automorphisms of K, which fix every element of F.. An element of G(K,F) is
called a F-automorphism of K.
Just as we have seen that the fixed field of a group of automorphisms of K is a
subfield of K, we shall prove that the set of G(K,F) of F-automorphisms of K is a
subgroup of the group of automorphism of K.
SAQ 20.2.6: Let K be a field and F a subfield of K. Then the set G(K,F) of F-
automorphism of K is a subgroup of the group of automorphisrns of K.
Definition 20.2.7 : G(K,F) is called the Galois group ofK over F .
We close this section by obtaining a bound to the order of G(K,F).
Theorem 20.2.8 : If K is a finite extension of F, then G(K,F) is a finite group and its
order satisfies, O(G(K,F))::;;. [ K: F ] .
Proof: Suppose that [ K :' F ] = nand { e! , e2 , ..... , en } , a basis of Kover F. Then
every k E K is uniquely expressed as
K = a I e I + a2 e2+ + an en (l)
For some a I , a2 , .... , an E F
If possible assume that G(K,F) contains more than n elements and that CYI, CY2,
... , CYn+l
are any n+l distinct elements ofG(K,F). Consider the following n homogeneous
equations in n+ 1 unknowns XI , X2, , x, ..i . namely
IT!(el) X] + IT2(el) X2+ + O"n+l(el) Xn+!= 0
ITI(e2) x. + IT2(e2) X2+ + ITn+1(e2) Xn+l= 0

ITl(en) XI+ IT2(en) X2+ + CYn+1

(en) Xn+l= 0
 Distance Education 4

Because n + 1 > n., these equations have a non .- trivial solution , say, x, -= a, , X2= a2 .
........" Xn+1= an+!.
Then, for 1 s is n ,
a, crl (ei)+a2cr2(ei)+ +an+! crn+l(ej) =0 (2)
Since crjE G(K,F), for 1 s is n+ 1 , we have
criC
a) = a , "if a E F.
So, for K = al el + + an en E K and for 1 sj S n+ 1 we have
crj(k) = crj( a I e I + + an en )
= crj( al ) crj (el) + + crj(an) crj( en )
= al crj (el) + + an crj( en) .
Then, for all k E K,
a! crl (k) + a2 cr2(k) + + an+I crn+1(k)
= al (a, crl (el ) + a2 cr2(el ) + + an+1crn+1(el ) )
+ a2 (a, crl (e2) + a2 cr2(e2) + + an+1crn+1(e2'))
·········0············································ .

+ an (a, crl (en') + a2 cr2(en) + + an+1crn+!(en))

=0
This is a contradiction to the Theorem 17.3.3 . So G(K,F)<tannot contain more than n
elements and this establishes that
O( G ( K, F) ) sn= [ K : F ] .
Let us discuss some examples which make clear some of the coricepts introduced
so far.
Example 20.2.9: Let K be a field. It is easy to see that G(K,K) = {I}, 1 being the
identity automorphism of K.
Example 20.2.10: For the field C of complex numbers and its subfield R of real
numbers. Find G(C,R) and O(G (C, R)).
Solution: An element of C is of the form a + ib where a, b E Rand i = "f-l . So if
a E G(C,R), then c is automorphism of C for which cr(x) = x, "if X E R. So
o(a+ib) = o(a) + cr(ib) =cr(a) + cr(i) cr(b)
= a + cr(i) b
If we determine c(i), then c is known. Since i2= -1 and cr(-l) = -1 we have
o
-1 = cr(-l) = cr(i2)=(cr(i) )2 .
So cr(i) ,= ± F-i ::=' ± i.
Thus
cr(a+ib) = a ± ib .
SO G(C,R) contains only two 'automorphisms of K, namely, the identity automorphism
I on Crand the automorphism c ofC'which maps every elements z of C to its complex
conjugate ~ in C. So G( C, R) ={ I, c }and O(G(C,R» = 2~\
Observe that O( G( C,R) ) = 2 = [ C : R] . .
Example 20.2.11 : For the fie.ld Q of rational numbers consider Q(!3), the field got by
adjoining J2 to Q fmd G( Q( -fi) , Q). Show also that the fixedf\eld of G( Q( -fi ),Q»
is Q. .
Solution: We know th~t Q( -fi) = (a + -fib / a, b E Q }.
\
If c E G(Q(.fi) , Q), theno(x) = x, V X E Q , so that, for a + J2 \ E Q( J2) , we
have'- ....
a ( a + 12 b) = cr(a) + cr( -fi) cr(b) = a + cr( 12) b .
Let us see what is c (:fi) ?
Since 2 = ( 12 i and c (2) = 2 we have,
.
2 = cr(2) = c ( (.fi i )= c (.fi)2
'---....

.
So o( .fi) = ± .fi .
From this we get
cr(a+ -J2b)=a+-J2b
or
cr(a+ .fib)=a-.fib
Thus, the identity automorphism cr/ say on Q(.fi) and the automorphism cr2 on
Q( .fi) which takes every element a + -J2 b of Q( -J2 ) to its conjugate, namely, a-.fi b,
are the only elements ofG(QCJ2), Q». So O(G (Q( J2),Q») =~2. . .. . .
Observe that O(G (Q( -fi), Q» = 2= [ Q( -fi) : Q] .
Let us see what is the fixed field of G(Q(.fi), Q). Evidently Q is contained in the
fixed fieldof G{Q( .fi), Q). Suppose a + J2 b is fixed by every element in G(Q( J2), Q).
Then '" ~ ~~. /
cr2 (a + J2b) ~ a.+--J2b.!
But cr2(a +~@£l=a - .fi b .- .
\Then a ~c~a~~b a~d this forces .that'b = 0 - So a ~ -fi b = a E Q. Thus
the eleme~~ of Q are th~ments of Q( J2) that are fixed by every automorphism
in G(Q(:J2), Q ). So the fixed field of G(Q( J2), Q ) is Q itself. -,
 - ",:,,~ .. '

Distance Education 6

Example :20.2.13 Find G(Q({I2), Q). Show that O( G(Q({I2); Q;)<[Q({12 ) : Q].
Solution: Let Q be the rational nwnber field and h£K=~{Vi.)~Then
K = { a + Vi b + (Vi)2 C/ a,b, C E Q} .
Let us find G(K,Q). If 0" E G(K,Q) , then 0" is an automorphism of K and otx) =
x, '\f x E Q. Now 2 E Q implies that 0"(2) = 2 . So
2 = 0"(2) = 0"( (lfi)3 ) = { 0" (Vi) }3
Thus 0" ({12) is also a cube root of 2. Since {12 is the only real cube root of 2 we
have
O"(Vi)=Vi.
Thus, for any a + {12 b + ({12)2 C in K,
0" ( a+ Vi b+ (Vi)2 C ) = a( a) + 0"( Vi) o(b) + cr( Vi i c (c)
=a+Vib+Vic.
So c is identity automorphism on K and G(K,Q) consists of only the identity map.
- "." ~ -
So O(G( Q(Vi), Q)) = 1 (1)
Now Vi is a root of x3 - 2 over Q:- Since ~~. - 2 is irreducible over Q ( by .
Eisenstein's criterion with p = 2). So -
[Q({I2) : Q] = deg (x3 - 2') = 3. (2)
From (1) and (2) we get
O(G(Q(Vi), Q)) < [Q({I2) : Q ]:-.
Further O(G(Q(Vi), Q )) = 1 show that the fixed field.of G(K, Q) is K itself, which
is larger than Q.
.:\j.'.1

20.3 SYMMETRIC FfTNCTIONS =>:

Let F be any field. Then the polynomial ring F [ x, , X2 , .... , x, ] in n - variables
x, , X2 , , x, is an integral domain. So we can find the field of quotients of
F [ Xl , X2, , x, ] and this is called the field of rational functions iri x, , X2 , .... , xn. We
shall denote it by Fe-Xl, X2 , ... , xn ) . . .
Let Sn be the symmetric group of degree n. In fact S, is the set of all permutations
(that is, one - to - one and onto maps) of the set S= { 1,2, .... , n }. For rr E Sn we write
symbolically

a= C,;!; a~2) a;n) J

and the set { 0"(1) ,-0"(2), , cr(n) } is the same as the set S in some order.
 Each c E Sn induces a one - to - one and onto map on F( XI , X2 , .... , xn) . We
shall also denote this map by c and it is defined in the following way :
For r( XI, X2 , ... , x,") E F(XI , X2 , ... , xn), we defme
cr (r (Xl, X2 , ... , Xn)) = r (Xcr(l), Xcr(2), .... , Xcr(n)) (1)
We can say something more about o, namely, it is an automorphism on
F( Xj , X2 , ... , xn). To see this, let r (Xl, X2 , .... , x,.') , S (Xl, X2 , .... , Xn) E F( X]
,... , x, ). From (1) we observe that the effect cr on any rational function is to replace
suffix of each variable by its image under cr. So
c ( r (x. , X2 , .... , x, ) + S (x, X2 , .... , Xn) )
=r ( Xcr(l), Xcr(2), .. .., Xcr(n)) + S( Xcr(l), Xcr(2), .... , Xcr(n))
= o ( r (XI, X2 , .... , Xn) ) + cr ( S (x. , X2," ... , x, ) )
In the same way we can see that
cr (r (xj , X2 , .. ""' Xn) . S (xj , X2 , .... , Xn)) =
= cr (r (xj , X2 , .... , Xn)) + cr (S (x, ,)(2 , .... , Xn))
Thus each cr E Sn induces an automorphism on the field F ( Xl , X2 , ... , x, ) of
rational functions. In view of this we can treat S; as an automorphism group of
F ( x, , X2 , ... , x, ).
What is the fixed field of Sn ? It consists of all rational functions r( x, ,
X2 , ... , x, ) which are fixed by each c E Sn. In other words a rational function r ( x, , X2
,... , x, ) lies in the fixed field of Sn if
c (r (x. , X2 , .... , x, ) ) = c (r (x. , X2, .... , Xn) ) .
or, equivalently,
r (Xcr(l), Xcr(2), .... , xcr(n») == r (x. , X2 , .... , Xn) .
The elements of the fixed field of Sn are called the symmetric rational functions.
This discussion leads to the following definition.
Definition 20.3.1: A rational polynomial r ( x, , X2 , , xn) is called a symmetric
rational function if r ( XI , X2 , ... , xn) = r (xcr(l) , Xcr(2), ; xcr(n)) for all c E Sn. We
shall denote this set by S.
Actually we are interested in a special class of symmetric rational functions caned
elementary symmetric functions which are given below. Consider the following
functions in Xj , X2 , , Xn·
al = XI + X2 + + Xn
a2 = L Xi Xj
icj

a3 = L Xi Xj Xk
i-cj-ck
 Distance Education 8 fo~~~ Acharya Nagarjuna University

an = x, X2 ... x,
We claim that a, , a2 , ... , an are symmetric rational functions. For, if c E Sn ,
smce
{ cr(L), cr(Z) , , o(n)'} = { 1,2, ,n }
we have c(a. ) = xcr(l) + Xcr(2)+ ....+ xcj(n) = XI + X2 + ...+ Xn = al .
and in same way we can see that o'( a, ) = a, for 2 :::;i :::;
n .
So al , a2 , ... , an are symmetric functions.
DefinWon 20.3.2: The function. a, , a2 >an defined above ate called elementary
•••• ,

symmetric functions.
Let us list the elementary symmetric functions for n= 2, 3 and 4.
n= 2.
al = XI + X2
a2 = x, X2
n=3
al = Xj + X2+ X3
a2 = :X:IX2 + XI X3+ X2x,
as = x, X2x,
n=4
al = XI + X2+ X3 + XL!
a2 = x, X2 + XI X3+ x, XL! + X2X3 + X2 XL! + X3 XL!
as = Xl X2Xs + XI X2 XL! + x, x, XL! + X2 X3.X4
a4 = Xj X2X3 X4 .
The following observations are worth noticing:
·2
1. When n = 2, x, , X2 are the roots of the polynomial t" - al t + a2
2
2. When n = 3 , xi, X2 ,X3 are the roots of the polynomial e - at t + a2 t - as
3. When n = 4, Xl ,X2 , X3 ,XL! are roots of the polynomial t4 - ale + a2 t2 - as t + a,
4. In general, XI , X2, ... , x, are the roots of the polynomial
t n - al t n-I + a2 t n-2- ..... + ( - l)". an, were
hIt al , a2 , ... , an are e emen ary
.
symmetric functions.
We now prove a useful theorem on symmetric rational functions.
Theorem 20.3.3: Let F be a field and let F( X; , X2 , .... , x, ) be the field of rational
functions in XI , X2 , .... , x, over F. Suppose that S is the field of symmetric rational
functions. Then
(i) [F ( XI , X2 , , x, ) : S ] = n!
(ii) G ( F( XI , X2 , , x, ) , S ) = Sn , the symmetric group of degree n.
(iii) If al , a2 , ... , an are elementary symmetric functions then
S = F( al , a2 ,..... , an)
(iv) F( XI , X2, .... , x, ) is the splitting field over F ( al , a2 , ... , an) = S of
n
the polynomial t" - a, t -I + a2 tn-2 + .....+ (-1 t an .

Proof: In view of the above discussion, the elementary symmetric functions al , a2 ,.., an
belong to S, the set of symmetric rational functions in XI , X2 .. .. , x, . So the field
F( al , a2 , , an) got by adjoining. a, ,a2, , an to F, is contained in S. That is
, F ( a I , a2 ...". ... , an ) c S (1)
Major part of the theorem is established if one show that
(i) [ F( XI , X2 , , x, ) : S ] = n!
(ii) S = F ( (' a2 , , an ) .
We observe that S, is a group of automorphisms of F( XI , X2 , .... , x, ) , leaving
every element of S fixed. So
S, c G (F( XI , X2 , ..... , xn) , S ) (2)
From the Theorem 20.2.2 and (1) it follows that
[F( XI , X2 , ..... , x, ) : S ] ~ 0 (G F(F( XI , X2, ..... , x, ) , S)
~ 0 (Sn ) = n! (3)
So, if we show that [F( XI , X2 , ..... , x, ) : S ]~ n! ,then (i) follows. To see this let
us consider the polynomial
pCt) = t" - a I tn-I + a2 tn-2 + (-1 an r
This is a polynomial over F( al , a2 , ..... , an) and its roots namely, XI , X2 , ..... , Xn
lie in F( XI , X2 , ..... , xn). So pet) splits into linear factor in F( XI , X2 , ..... , x, ). We
claim that pet) cannot split into linear factors in any suhfield K of F( x, , X2 , ..... , x, ),
which contains F( a, , a2 , ..... , an) . For, such a field K'would than contain F as well as
xj , X2 , ..... , Xn and thus contain F( x. , X2 , ..... , x,"). Thus F( Xl , X2 , ..... , xn) is itself
the splitting field ofp(t) over F( a, , a2 , ..... , an). This proves part (4).
-: Incidentally this proves part (i) of the theorem. Since degree p(x) is n, it follows
that
[F( x, , X2 , ..... , x, ) : F( a, , a2 , ..... , an ) ] ~ nl (4)
 Distance Education 10 )E~~: Acharya Nagarjuna Universltv

Now F( a. , ai", .. ...', an) is a subfield ofS and S is a subfield ofF( XI, X2 , ..... , xn). So
fr6m (3) and (4) we have . .
n! ~ [F( XI·, X2, ..... , xn): F( ar , a2 , ..... , a;rj', ,-, .
.------

= [F( XI, X2 ".' , xn) ~ ] [S, F( a. , a2 ,.~ .., an) ] ~ n! (5)

[F( XI, X2,.· , x.) : F( a. , a2,·· ... , an)J = n! (6)
Again from (3) , (5) and (6) we have
n! = [F( XI, X2 , ..... , xn):-S ]l S : F( a. , a2 , ..... , a;;)]
.~ n! [ S: F( a, , a2 , ..... , an ) ]
So [S:F(al,a2, ,an)]:s:1
But [ S : F( a, , a2 , , an) ] 0 1 , so that
[S : F( a. , a2 , .. ~:== __1.
This forces that
S = F( al , a2 , ..... , an) .
which is the contention (ii). This, together with (6) gives
[F( x, , X2 , ..... , xn) : S ] = n! .
This combined with (3) yield
O(G(F( XI, X2 , ..... , xn) : S = n! =0 (Sn).
That is Sn is a subgroup of G(F( XI , x~,....., x, )., S ) and their orders are equal. So
Sn=G(F(XI,X2, ..... ,xn), S),
'which proves part of the Theorem.

"
Example 20.3.4: Express the following symmetric functions as a rational functions of
the elementary symmetric functions -::
(i) 2 2 2 ( .. ) 2 2 2 2 2 2' . \', . <,
1 Xj+X2+X3 11 XjX2+X2X3+X3Xj '-'
-,
Solution: (i) we have
xl + xi + xl = (Xl + X2 + X3)2 - 2 ( x, X2 + X2X3 + X3x, )

-- aj2 - 2a2
where a, = XI + X2 + X3 and a2 = x, X2 + X2X3 + X3X; .
(ii) Put a = xl xi + xi xl + xl xl
2 2 2 2 2 .2 2 . . 2 22 . I 2' ., -
Then (x, X2 + X2X3 +X3 xj ) = Xl X2 + x2 X3 + X3 Xl + 2( XjX2X3 + X2X3 Xl + X3XI X2)

That is aJ = a + 2 XI X2X3 (XI + ~ + X3), ,.

or,
/
 a§ = a + 2 a, a, .
So a = a§ - 2a3 al ,
Where aI = XI + X2+ X3, a2 = XI X2+ X2X3+ X3XI and a, = XI X2X3.

20.4 SUMMARY:
In this lesson we have studied the following;
I. The fixed field of a group of automorphism of a field K is a subfield of K.
2. Let F be a field and K an extension of F. The set G(K,F) of all automorphisms of
K which fix every element of F is a subgroup of the automorphism group of K.
3. Let F be a, field and let F(xl , X2 ,... , x, ) be the field of rational functions in
XI , X2, .... , x, over F. Suppose S·is the field of symmetric rational functions then
(1) [F (x., X2, , xn) : S ] = Ln.
(2) G(F(xl, X2, , xn) ,S ) = Sn, the symmetric group ofn symbols.
(3) Ifal,a2 , .... , an we elementary symmetric functions in XI,X2,... , x, then S =
F(a],a2,'" ..,an ).
(4) F(X!,;X2,"" x, ) in the splitting field of the polynomial t" - a, tn-I + a2 tn-2
(-Ir an over S = F(al, a2 , .. " an).
...........
4. If K is a finite extensionof F, then G(K,F) is a finite group and its order satisfies
O(G(K,F) ) :;; [ K :'F ]

20.5 :TECHNICffL TERMS

Galois group -of a polynomial
Automorphism of a fi,~tp
Fixed field of a group.ef automorphism
Symmetric functions
Symmetric rational functions
Elementary symmetric functions

20.6 MODEL EXAMINATION QUESTIONS

PART - A.( short answer questions»
1.Let G be' a group of automorphisms of K. Show that the fixed field of G is a subfield of
K·
•
.,.
.<.

2. For thefielci, C of complex numbers and its subfield R of real numbers. Find G(C,R)
. and O(G (C, R»,
 Distance Education 12 ~~s:::: Acharya Nagarjuna University

3. For the field Q of rational numbers caii'sider Q( J2) , the field got by adjoining J2 to
Q find G( Q( J2) , Q ). Show also that the fixed field of G( Q( J2 ),Q» is Q.
4. Find G(Q(V2), Q). Show that O( G(Q(V2), Q) < [Q(V2: Q] .
2nj

5. Find G( Q(w), Q) , where w = e 5, is a 5th root of units.

PART - B (Long answer questions)

1. If K is a field and if a1 , a2 , .... , an are distinctautomorphisms of K, show that
it is impossible to find elements a, , a2, ..... , an in K, not all zero such that
al al(u) + a7.a2(u) + + an an(u) = 0 , for all u in K.
2. If K is a finite extension of F, then show that G(K,F) is a finite group and its
order satisfies, O(G(K,F» ~ [K: F] .
3. Let F be a field and let F( x, , X2, .... , xn) be the field of rational functions in x, ,
X2, .... , x, over F. Suppose that S is the field of symmetric rational functions. Then
prove that
(i) [ F (Xl, X2, , x, ) : S ] = n!
(ii) G ( F( x, , X2, , Xn) , S ) = S, , the symmetric group of degree n.
(iii) If a, , a2 , ... , an are elementary symmetric functions then
S=F(a"a2, ... ,an)
---- (iv) F( x, , X2, .... , xn) is the splitting field over F (ai, a2 , ... , an) = S of
the polynomial t" - a, tn-1 + a2 tn-2+ .....+ (-1)" an .

2~.7 ANSWERS TO SELF ASSESSMENT QUESTIONS:

Self Assessment Question 20.3.4: Let G be a group of.automorphisms of K. Thenthe
fixed field of G.is a subfield of K.
Proof: Let us set
F = { a E K / a( a) , a \7' a E G }.
Then, for a, b E F we have o(a) = a and cr(b) = b. So c being an automorphism ofK,
a (a ± b ) = a( a) ± a(b) = a±b
and cr(ab) ~ cr(a) a(b) = ab
Thus a ± b F and ab F.
E E
.
Also, if a E F and a*
0 , then ( again c being an automorphism of K ),
/~ o(a' ) = (c (a) l = a-I. r
So a-I E F. Thus F is a subfield of K.

2L.
Self AssessmentOuestion 20.2.6: Let K be a field and F a subfield of K. Then the set
G(K,F) of F-automorphism of K is a subgroup of the group of automorphisms of K.
Proof: Let 0"1, 0"2be any two F-automorphisms of K. Then evidently 0"10"2-I is an
automorphism of K. Further O"I(a)= a, 0"2(a) = a , V a E F. Further 0"2(a) = a implies
that 0"2-1(a) = a, v a E F. So (0"10"2-1)('(1) = 0"1(0"2-(a))
1 = 0"1(a) = a. Thus 0"10"2-is
1 also
a F-automorphism and ITI0"2-1 E G(K,F). This shows that G(K,F) is a sub-group of the
group of automorphisms of K.

20.8 EXERCISES:
1. Let Q be the rational number field and K = Q-( 3 + .J2) . Find G(K,Q) and also the
fixed field of G(K,Q).
(Hint : Observe that Q (3 + .J2) = Q( .J2) )
2. Express the following as polynomi~ls in the elementary. symmetric function in XI , X2
and X3
.) 2
( 1 (Xl + X22 )(X2
2
+ X32 )
(2
X3 + Xl
2 )
o
.. ) 3 3 3
( 11 Xl + X2 +X3

(iii)( XI - X2 i (XI - X3i (X2- X3)2

20.9 REFERENCE BOOKS :
1. ARTIN, EMIL, Galois Theory; 2nd ed Notne Dame, Ind : University of
Natne Dame Press, 1955.
2. FRALEIGH, lB., A First Course in Abstract Algebra. 2nd ed., Addision Wesley'
Publishing Company. ;::J •.
3. HERSTEIN, I.N., Topics in Algebra, Vikas Publishing House.
4. JACOBSON, N., Lectures in abstract algebra. Vo1.II, Nostnand, 1953.'
5. BHATTACHARYA, P.B., JAIN. S.K., N,AGPAUL.S.R., Basic Algebra,
(2nd Edition) , Cambridge University Press. \"" . i

/ \'.:?
- l: ;·-1
; \'!,
" 1

! Lesson Writer I'

Prof L.Nagamuni Reddy
S.V.l!nivcrsity, Tirup~i '
 LESSON - 21 : FUNDAMENTAL THEOREM
OF GALOIS THEORY
CONTENTS:
OBJECTIVES
STRUCTURE
21.1 Introduction
21.2 Normal extensions
21.3 Fundamental Theorem of Galois Theory
21.4 Summary
21.5 Technical terms
21.1L Model Examination Questions
21.7 Answers to Self Assessment Questions
21.8 Exercises
21.9 "Reference Books

OBJECTIVES
After reading this lesson you should be able to
• Know about normal extensions;
• Introduce the notion of Galois group of a polynomial;
• Establish the fundamental theorem of Galois theory ;

21.1 : INTRODUCTION:
In the examples on the fixed fields of automorphism groups, .studied in
lesson 20, we have seen that the fixed field of G(K,F) is usually larger than F
itself. Always F is contained the fixed field of G(K,F). But the extensions K of
F for which the fixed field of G(K,F) coincides with F itself, playa crucial role
in the field theory. Such extensions are called normal extensions. For these
extensions we can over obtain the order of G(K,F) by sharpening the Theorem
20.2.8, in which the inequality can be replaced by equality. In addition to this
we obtain some results on normal extensions which are useful in proving the
fundamental theorem of Galois theory in the next section.
I
 Distance Education 2

21.2 NORMAL EXTENSIONS

Definition 21.2.1 : A finite extension K of F is called a normal extension if F
is the fixed field of G(K,F).
Example: 21.2.2: C is a normal extension of R.
Solution: We have seen that G(C,R) = { I, a}, where I is the identity
automorphism on C, that is,
I(a + i b) = a + i b
and c is the automorphism of c given
a ( a + ib ) = a - ib
So, if a + ib is in the fixed field of G(C,R), then a + ib must be fixed by I and a

That is, a ( a + ib ) = a + ib ,
or, a - ib = a + ib
This gives b = 0 and this a + ib = a .
So R is the fixed field of G(C,R) , so that c is a normal extension of R.
We now prove the result we promised above.

Theorem 21.2.3 : Let K be a normal extension of F and let H be a subgroup of

G(K,F). Let KH = { X E K / cr(x) = x, 'If c E H} be the fixed field ofH. Then
(i) [K : KH ] = O(H)
(ii) H = G(K,Kr·J ) .

In particular ifH = G(K,F) then H = G(K,F) and [ K : F] = O( G(K,F» .

Proof: By the definition of H, every element of H fixes every element of KH ,
so that H c G(KH, K ) . '
So O(H) ::; O(G(K,KH ) ) .
But by Theorem 20.2.2,
[ K : KH ] ~ O(G(K,KH ) ) .
So [K : KH] ~ 0 (G(K,KH » ~ O(H) (1)
Our aim is to show that [ K : KH ] = O(H) . Since K is a finite extension
of F and F c KH C K, K is also a finite extension of KH . So K is simple
extension of KH and thus there is an element a E K, such that K = KH(a). Let
m = [K : KH ] then m is the degree of minimal polynomial of a over KH .
Let H = { al , a2 , ... , an } , where al is the identity map, so that O(H) =
n. Consider the elements crl(a) = a, cr2(a) , ..... , crn(a) and the elementary
symmetric functions a 1 ; a2 , ... , an given by :
 al = crj(a) + cr2(a) + ~ crn(a)
a2= L cri(a) crla)
i<j

an = crl(a) cr2(a) crn(a)

We claim that al , a2 ,.., an are fixed by every o E H. For, H being a group,
we have, for every crE H,
Hcr = { criO'I 1 :S; i:S; n } = H .
So crcrj , crcr2, ... , crcrn are the same as crl , 0'2, .... , O'nin some order. Hence
cr(al) = crcrJ (a)+ 0'0'2(a) + + O'O'n(a)
= O'l(a) + cr2(a) +.. : + crn(a) = al
In the same way we can see that o(o, ) = ai for I ::s; i ::s; n . So aj , a2 , .... , an lie
in the fixed field of H, namely KH .'
Further a = O'J(a) , 0'2(a) , , O'n(a) are the roots of the polynomial
(x - O'j(a)) (x - 0'2(a) , , (x - O'n(a)),
which is the same as
x n - al x n-I + a2 x n-2 . . . . . .. + (- l)" an ,
whose coefficients al , a2 , ... , an lie in KH. That is, a = crl(a) satisfies a
polynomial of degree n over KH. But we have already seen that m = [ K : KH ]
is the degree of the minimal polynomial of a over KH .
So n 2: m or O(H) 2: [ K : KH ] ..... (2)
From (7) and (2) we have .(
[ K : KH ] = O(H) (3)
This proves the first part of the theorem.
Again from (1) and (3) we have
O(H) = O(G(K : KH ) ) .
This means that H is a subgroup of G(K,KH ) , having the same order as that of
G(K,KH)' So,
H = G(K: KH)
and this proves the second part of the theorem.
If we take H = G(K,F), the normality of Kover F implies that KH = F.
So, by (3) we have
O(G(K,F) = [K : F] .
 Distance Education 4

We want to investigate, how the normal extensions are connected with

the splitting fields. As a first step in this direction we have
2m
Example 21.2.4: Find G(Q(2), Q), where w = e 5, is s" root of unity. Show
that Q(w) is a normal extension of Q.
2m

Solution: Put w = e 5 . Then w5 = 1. So w satisfies the polynomial x5 - 1.

Since
x5 -1 = (x-I) (x" + x3 + X2 + X + I )and w is not real, it follows that w
satisfies' the polynomial f(x) = X4 + x3 + X2 + X + I. By the Eisenstein's
criterion one can see that f(x) is irreducible over Q. Thus
[ Q(w) : Q ] = 4,
and Q(w) ~ { ao + a] w + a2 w2 + a3 w3 I ao , aj , a2, a3 E Q }.
Let c E G ( Q(w) , Q). Then o is an automorphism of Q(w) are c fixes
every element of Q, that is cr(a) = a for every a E Q. Further w -:1= I (since it
is not real), cr(w)-:I=1. Since cr(a) = a for every a E F , we have
cr(ao + a] w + a2 w2 + a3 w3 ) = ao + aj cr(w) + a2 cr(w") + a3 cr(w'')
= ao + a] cr(w) + a2 cr(W)2 + a3 cr(W)3 .
So o is determined if o'(w) is known.
Now w = 1. This shows cr(w) is also a 5th root of unity. So cr(w) can only
5

be one ofw, W2 , w3 , or w4 .
Thus
G(Q(w) , Q) = { cr] , cr2, cr3,cr4} , where
2' 3 2 3
cr](ao + a] w + a2 w + a3 w ) = ao + a] w + a2 w + a3 w
'- so that cr] is the identity map on Q(w) ;
2 3 2 2 2
cr2(aO + aj w + a2 w + a3 w ) = ao + a] w + a2 (w )2 + a3 (W )3
-.- - - 2 4· 5
---- =-ao + a3 w + u] w + a2 w (slllce w = I)
2 3 3 4
cr3(aO + a] w + a2 w + a3 w ) =ao + a2 w + a] w + a3 w
4
cr4(aO + a] w + a2 w:- + a3 w 3
)
3
= ao + a3 ~+ a2 w + a] w .
Here [Q(w): Q ]= 4,
and O(G(Q(w), Q) ) =4 .
So O(G(Q(w); Q) ) = [ Q(w) : Q]
Hence Q(w) is a normal extension of Q. ,
Lemma 21.2.5: Let K be the splitting field of f(x) E F[x] and let p(x) be an
irreducible factor of f(x). If the roots of p(x) are aj , a2 , ..... , am, then for
each i, there is an automorphism cr; in G(K,F) such that cri (a] ) = aj . ..-
 Proof: Let a\ , ai be any two roots ofp(x). Then they lie in K. Put
F \ = ~(a\) and Fl\. = F( ai ) .
. Since a\ , aI are any two roots of p(x), there exists isomorphism C(see Cor.
16.5.7) from F] to Fl] such that
_'t ( a I and 't(a) = a, 't:j a E F.
) = o, ( I)
It is given that K is the splitting field of f(x) over F. So we can view K as the
splitting field of f(x) over F 1 as well as over Fl1. But any two splitting fields
of a polynomial are isomorphic. So there exists an isomorphism
ai : K -7 K(and so an automorphism of K ) such that
ai(a])='t(ad =ai
That is, ai ( a ) = 't (a ) = a , for all a E F.
and
That is ai is the required automorphism of K.
Ol'he following theorem gives the relation between normal extensions
and splitting fields.
Theorem 21.2.6: K is a normal extension of F if and only if K is the splitting
field of some polynomial over F.
Proof': First let us assume that K is a normal extension of F. Then K is a finite
extension of F such that F is the fixed field of G(K,F). Since K is a finite
extension of F, K = F(a) for some a E K. Further the normality of Kover F
implies that ( by Theorem 2l.2.3 ).
O(G(K,F) ) = [ K : F ]
and this is finite. Let G(K,F)~ a2, , an }, where al is the identity
automorphism. +
Consjdcr-theelementary symmetric functions a) , cii ; ..... , an in the variables
. il = al (a) , a2 (a) , , an (a) given by
a\ = a} (a) + a2 (a) + + an (a)
a2= L ai (a) aj (a)
i<j

l (1)
an = al (a) a2 (a) an (a) .
Since G(J(,F) is a group, for any a E G(K,F) ,
{ a], a2, ... , an} = { o c. , o c-, ...., o o.,}
So (as in Theorem 21.2.3) we can see that a) , a2 , ... , an are fixedby every a
E G(K,F).
 Distance Education 6

Since K is a normal extension of F, we have ( bythe definition of

normal extension) that F is the fixed field of G(K,F) . So aI, a2 '" ..., an,
which are fixed under each of a in G(K,F), must be in F. That is, aI, a2, ..... ,
an E F.
Consider the polynomial p(x) =( x - al(a) ) (x - a2(a) ) (x - an(a) )
= x'" - al x"" + "" ".+ (-lr an , (by (1) )
Since aI, a2 '" ... , an E F , we have p(x) E F[x].
Further a E K, al , a2 , "., an E G(K,F) imply that al(a) , a2(a), .... ,
O'n(a)E K.
Thus the polynomial p(x) splits into a product oflinear factors in K[x].
Next suppose that p(x) splits into a product of linear factors over some
subfield E of K which contains F. Then, a being a root of p(x), must be in E.
That is F c E and a E E, so that F(a) c E, or, K c E. This forces E = K.
These establish that K is the splitting field of the follow&g p(x) over F.
Conversely assume that K is the splitting field of a polynomial f(x) E
F[x] . We prove the converse by using induction on [ K : F ]. Let [ K : F ] = n .
If n = 1 then K = F and evidently K is a normal extension of F. 0

So let us assume that if KI is a splitting field of a polynomial over F

such that [ KI : F] < [ K : F] , then KI is a normal extension ofF.
Let p(x) E F[x] be an irreducible factor of f(x) F[x] and al , a2 , ... , ar
are roots of p(x) in K. Evidently K is the splitting field of p(x) over F. So we
can consider K as the splitting field ofp(x) over F(al)' Since deg p(x) = r.
[ F(al ) : F] = r.
let<. : F] n
So [K:F(al)]= -,} :- =-<n
[F(a1) : F] r
That is, K is the splitting field ofp(x) over F(al)such that [ K : F(al) ] < n .
By induction hypothesis F(al) is a normal extension of K, so that the
fixed field of G(K,F(al) is the same as F(at): Using this fact we shall show
that F is the fixed field of G(K,F), so that K becomes a normal extension of F.
By the definition of G(K,F) ,.F c G(K,F) . If we show that G(K,F) c F,
then G(K,F) equals F.
Let 8 beany element ofK which is fixed by every a in G(K,F). We shall
show that 8 E F. Let 't E G(K,F(al )). Then 't is an automorphism of K
fixing every element of F(al) and hence fixes every element of F. So 't E »:

G(K,F). Since 8 is in F we have 't(8) = 8. Thus

't(8) = 8 , 'yf r EOG (K, F(Ul))
~esson : 21 ==== CD GLGEBRAY

By the normality of Kover F(al) , F(al) is the fixed field of G(K,F(al) ), so

that e E F(al)' Thus I

e = '\/\'0
+ /\'1
'\ al + /1.,2 al 2 +
'I + /\,a.-I
'\ al r-I . (1)
for some Al , tv2, ", " /\,r-IE F
By the Lemma 21.2,5, for each 1 s is
n there exists an automorphism o, of K " '

such that cri E G(K,F) and cri(o, ) = ai ' Since this c, leaves fixed e and each
Aifixed, we have from (1),
8 = cri (6 ) = cr/Ao) + cri(Ad a/ al) + cr/A2) cri((12) + .... ,+ cri(Ar-l)
cr/aia.-I)
2
= Ao + Al ai + A2ai + .....+ Ar-Iat-
I
(2)
from (1) and (2) we get
2
8 = Ao+ AI ai + A2ai + ., ...+ Ar-I at I
for I' = l , 2, .... , r.

This shows that al , a2, ... , urroots of , \

p(x) = (Ao- e ) + Al X + A2X2+ + Ar-lxr-I ,

where p(x) E K[x] , since e , Ao, AI, , Ar-I E K.
That is the polynomial p(x) of degree r - I has r distinct roots, which is not
possible unless p(x) = 0, So
Ao- 8 = 0 , or, 8 = 1..0 .
..
Since 1..0 E F we have 8E F.
That is every 8 fixed under every crj E G(K,F) , is in F, or, F is the fixed field
of G(K,F), So by the definition of normality, K is a normal extension of F.
Definition 21.2.7: Let f(x) be a polynomial in F[~] and let K be its splitting
field over F. The Galois group of f(x) is thergroup G(K,F) of all the
automorphisms of K leaving every element of F fixed.
Remark 21.2.8: Let f(x) = ao + al a + a2 a2 + + an an = 0
Let 0' E G(K,F). Then c(«) = a for every a E F. Further 0' being an
automorphism of K, f'
,. 1

o = 0'(0) = 0' (~~ + a I a + + an an)

ao + al cr(a) +
= + an cr(at
This means that c(a) is also a root of f(x). That is every 0' E G(K,F) takes a . t

root of f(x) into another root of f(x) and hence, it can be viewed as a
permutation on the set of roots of f(x). Thus we can treat the Galois group of
f(x) as a permutation group of the set of roots of f(x) , .
Example 21.2.9: If w = CDS Tr + i sin Tr , show that Q(w) is a normal extension of
4 4
Q.
....----

Solution: Now w = cos Tr + i sin Tr is -a root of X4+ 1 . We shall show that

4 4
Q(a) is the splitting field of x" + lover Q.
Since ( x + 1)4 + 1 = X4+ 4x3 + 6x2 +-:.4x_t2 and this is irreducible over Q
by Eisenstein's criterion with p =2, the polynomial X4+ ] is also irreducible
over Q.
SO [ Q(2) : Q ] = deg ( X4+ 1) = 4.
i
Now (w3l = (w4 = -I; (W5)4 = ( W4)5 = -I and (W7)4'= (w4f = -1.
Implies that w3 , W
S
, w7 are the other roots of X4+ -1.
Since W E Q(w) and Q(w) is a field we have that w3 , WS , w7 E Qtw).
So Q(w) contains al the roots of X4 + I, or,~_+. t...
'spltts into linear factors in
Q(w). ._-;-:-. - _ -
Further, ifE is any field such that-Q ~jtG:>~J-and-iJx4 + 1 splits into.'.
linear factors in E[x] , then WEE. B~ E E implies that Q(w) c
E, so that E = Q(w). These show that Q(w) is the splitting field of x.4+ lover
Q.
SO Q(w) is a normal extension of Q.':'
Example 21.2.10: Show that R is not a normal extension of R.
Solution: Consider the polynomial x3 - 2 E Q[x]. Since
x3-2 =(X_2113)(X2+2113X+22/3),.
1/3 -
only one root, namely, 2 is real and the other two ~!.oots are complex. So R
does not contain all the three roots of x3 - 2, ( conjains only one root Vi) .
That is , R is not the" splitting field of x3 - 2 over-~So R is not a normal
extension of Q. _.
Self Assessment Question 21.2.11: Show that Q(~"';2) is a normal extension
of Q.
Self Assessment Question 21.2.12 : Showuhat fft.4-iJ.) is_ not a normal
extension of Q.

21.3 FUNDAMENTAL THEOREM OF GALOIS THEORY:

We have now everything ready to establish the famous theorem, namely, '
The fundamental Theorem of Galois Theory' It sets up a one-to-one
correspondence between the subfields of the splitting field of a polynomial f(x)
 E F[x] and the subgroups of its Galois group. Furhter it gives a criterion that a
sub field of a normal extension of F is again a normal extension of F.
Theorem 21.3.1: Let f(x) be a polynomial in F[x], K its splitting field over F
and G(K,F) its Galois group. For any subfield T of K which contains F, let
G(K,T) = { a E G(K,F) / aCt) = t for every t E T } and for any subgroup H of
. G(K,F) let KH = {x E KI o'(x) = x for every aEH}. Then the association ofT
with G(K,T) sets of a one-to-one correspondence of the set of subfields of K
which contain F onto the set of s~bgroups of G(K,F) such that
(1) T = KG(K,T) ,
(2) H = G(K,KH ) ,
(3) [K:T] = 0 ( G(K, T) ) ; [ T : F ] = index of G(K, T) in G(K,F) ,
(4) T is a normal extension of F if and only if G(K, T) is a normal
.subgroup of G(K,F) , and . I
(5) when T is a normal extension of F, then G(T,F) is isomorphic
to G(K,F) / G(K, T).
Proof: (I) Let f(x) E F[x] and let K be its splitting field over F. Then by the
Theorem 21.2.6, K is a normal extension ofF. Let T be any subfield ofKsuch
that F c T. Then we can view K as a splitting field of fx) over T. So again by
Theorem 21.2.6, K is a normal extension of T. By the definition of normality
of an extension, it follows that T is the fixed field of G(K, T). So T = KG(K,T)
and this proves (1) .
(2) Now K is a-normal extension ofF. Let H be a subgroup ofG(K,F) and let
KH = { X E K / o(x) = x, \;j a E H }
be the fixed field of H, then H c G(K, KH). Also by the Theorem 21.2.3,
y
[K: KH] = O(H). )rl~
But O(G (K, KH) ) = [ K : KH ] , so that O(G(K,KH» = O(H). Hence
H=G(K,KH) .
This proves part (2) .
Further any subgroup of G(K,F) arises in the form G(K, T) for some
subfield T of K. So the association of T with G(K, T) maps the set of all
subfields of K containing F onto the set of all subfields of K containing F onto
the set of all subgroups of G(K,F).
Let G(K,T!) = G(K,T2 ) , for some subfields T\ , T2 of K containing F.
Then KO(K,T]) = KO(K,T2)'
and by part (1) this gives T\ = T2•
(3) From part (1) we have that K is a normal extension ofT.
So by the Theorem 21.2.3 ,
 Distance Education 10

[ K : T ] = 0 (G(K, T) ) .
Further O(G(K,F) = [ K :/F] = [ K:T] [T:F]
= O(G(K,T) ) [ T :F] .
So [ T : F] = 0 (G(K,F) 10(G(K,T))
= index of G(K, T) in G(K,F)
This proves part (3)
(4) To prove (4) we first prove:
T is a normal extension of F if and only if o'(T) c T, for every c E
G(K,F) .
To see this, let T be a field such that F c T c K and let rr(T) c T , for
every
c E G(K,F). Evidently T is a finite extension ofF. So T= F(a) for some a E T.
Then as in Theorem 21.2.6 we can see that T is the splitting field of the
polynomial
p(x) = IT (x - a(a».
O'EG(K,F)

SO(by Theorem 21.2.6), T is a normal extension of F.

Conversely, assume that T is a normal extension of F, where F c T c K.
Then T is a finite extension of F. So that T = F(a) for some a E T. Let p(x) be
the minimal polynomial satisfied by a. So (by Theorem 21.2.6), T is the
splitting field of p(x) and all the roots of p(x) lie in T. Since a is a root of p(x)
- 0,0+ alx + + as x S ,say, th en
2 s
ao + a Ia + a2a + + as a = 0 .
So cr(ao + al~ + a2a2+ + as as) = 0,
or, 2
0,0+ a]cr(a) + 0,2cr(a ) + + as cr(aS)= 0 .
(Since o'(o.) = a, V a E F) . So cr(a) is also root ofp(x) , V o E G(K,F).
So o(a) E T for every crEG(K,F). Since T = F(a), for any t E T, we have
t= ~o + ~Ia+ + ~s as
for some ~i E F .
So c(t) =r3o + r31cr(a) + + r3scr(aS) E T. That is cr(t) c T.
Thus T is a normal extension ofF, if and only if cr(T) c T.
We now have:
T is a normal extension of F , Q O'(T) c T, 'i/ 0' E G(K,F)
<> O'(t) E T, V t E T and V 0' E G(K,F)
o 1(0'(t)) = c(t) , V 1 E G(K,T), Vcr E G(K,F) and V t E T
 ( .: T e: G(K, T) -r;(t) = t, V' t E T)
1 1
<=> cr- 1(a(t» = a- (a(t» = t, V' a E G(K,F)
V' 1 E G(K,F) and V' t E T
<=> a-11; a E G(K, T), V' a E G(l);:j'E),V' 1 E G(K, T)
<=> G(K,T) is normal in G(K,F).
This proves part (4) ..
(5) Let T be a normal extension of F. Then aCT) c T \/ C' .\

Corresponding to each a E G(K,T) define o- : T -7 T by

c- (t) = o'(t).
Then it is easy to see that a* is an automorphism of T. Further f:;r:.
a*(a) = a(a) = a ( since F c T and a G(K,F) ).
E

So o- fixes &very element of F. So (J* E G(T,F).

\If
G(j' .
G(K,F) ~cr \. -
~ e--+-----f------e

G(K,F) / G(K, T)

For c , 1 E G(K,F) and t E T we have

(o 'C)*(t) ~ o't (t) = a(1(t) = cr(1*(t)) = cr*(1*(t))
= cr*'C*(t) .

So (CY 1)* = CY* 1* (1)

Now consider the map \If : G(K,F) -7 G(T,F) defined by
= (J*
\If ( c)

Then (1) showsthat \If is a homomorphism. What is kef \If ?

Ker \If = { a E G(K,F) / \If ( o) = i, the identity of G(T,F) }
= { a E G(K,F) / CY* is the identity automorphism of T }
= { CY E G(K,F) / CY* fixes every element in T }
=G(K,T).
So, by the fundamental theorem of homomorphism for groups we have
 Distance Education 12

G(K,F) I G(K,T) ~ \f' (G(K,F) ) (2)

But from part (3)
[ T : F] = O(G(K,F))
O(G(K,T))
Further since T is normal normal F,
[ T : F ] = O(G(T ,F) )
That is,
O(G(T,F)) = O(G(K.F)) I O(G(K,T)) (3)
From equations (2) and (3) we obtain
O( \jJ(G(K,F) ) ) = 0 (G(T,F) ),
so that-
(G(K,F)) = G(T,F).
\jf (4)
Now part (5) of the theorem follows from equations (2) and (4). .
Example 21.3.2 : Show that the Galois group of X4 - X2 + I is the~ien 4-
group.
Solution :Let f(x) = X4 - X2 + 1. Then
f(x) = (x+ Ii - 3x2 = (x2 + F3 x + 1)( X2 - F3 x + 1 ),
so that ± F3 ±
2
i are the roots of f(x). So the splitting field K of X2 - X2 + 1

over Q. IS got by acjoimng

dioi th e fiour roots (±13±i]
2. to Q . It can b e east'1y seen

that K = Q( J1, i). Since f(x) = X4 - X2 + 1 is irreducible over Q we have

[ K : Qjl. = deg f(x) = 4.
SO O(G(K,Q)) = [ K : Q] = 4.
By Lemma 21.2.5 there exists (J! , (J2, (J3 , (J4 in G(K,Q), such that

..0"1 _[F3+iJ-
- 2 -
F3+i
2 ' 0"2-
_[lY+iJ-
-- ---
-J3-i
2 2

. 0"3 = (_13_+i] = _-13_3_-i and a4 = (_13_+_) _-_~_+_i

\. 2 2 4

Since O(G(K,Q)) = 4, we have G(K,Q) = { aI, a2 , a3 , a4 } . That is G(K,Q)

is a group of order 4.
 'Nen~e G(K,Q) is a.cyclic group of order 4 or a Klein's group.
, '

.
SInce
(.J3+i]
0'1 -- =--.
.J3+i
.22

"r~~
So

0'1/:!~--'
.J3-i
=-2-
'I :
, -'

In th~l:sayhe!wi'aywe can check that

:' (~J3,3-iJ
/i 2
0'1 = --li-=-iand
(--li+iJ = -J3+i 0'1
• Jd 2 2 2 ~

So :

and ;
"

,/

tA(i)
,
= 0-1(J'J +i _
2.
J3 ~i~
2
~ o-/7I'+IJ-O-/(13
\ \ .2' 2
-iJ =i .

/ ,That is 0'1 fixe,S J3/ ' i and,alsCfevery elem~nt of Q. SO 0'] fixes every
element of Q( Jij ,i) = K. \
I .. i \
" In the sarryeway e ca~ see that : vr I'

\ \(J2CV3)= -Ii, O'~(i) = ~iISO that 0'; (13) = -Ii and -13'; (i.)l~li,'·
'--------/-" - \ \ '\

h)
\~, ' , , ,

(J3( Ji,.)'= -13, O'3(!) = so that c;j,(~J\; '¥3 andcj .J~ = i'~

"q4(:J3)~-:J3'''4(i2\o i so that ,,~(j])~ J\ and ,,~ (i) ~ i ;

So O'~ ~ O'~ = O'~ =O'j the identity automorphism. ,-.
\ , , --- . ::'" \' '

Hence none \e~3e ;-elemenfs'.ofGC~,Q) is agen.,~rator or G(K,Q) and thus

G(~~Q) .~, n~~,:'ccydiC group." T~is f~rc,es \that9~,~ .;Q) is isomorphi;c to ~h'e
, KlemA,,' :iMrroup. " ','" ,'~. ~, . .-;
''r'~ J
J

'\ ''-fam~le 2~.3.3: Show that

I

g!OUP_,?~'X4' _~,~(?
",''.-\
the octic ~'~-~2:L1
. .

is
,grpupJ.~at IS, the groupof sy e, 1\0 efa; squ~£e.,.: ~", ' ,\ .. -. . '
\ 'Sol1;ltion ;,~~~I,-2,i~";irreducible.ot.~i Q,
',' ",':-,',',~~
d};t~
,tk]','Therefo.!~~I:'E?'q(1i!
f( x +~) , f.~'li\V2 \,1\( x-:~~~:,{~~\~~*,
)~'ir~JhesPhtiiP-gheld'~~x4\-'2\'~',er\0/S~~\~"
i~wr"'Y<-, ,/. '\/ .. .. . , .~~ ~~-..
:.~)\
..
, E:Q-'l ~p'/ -/ -irc: :
Distance Educa~ti~o;;n
--::::~~~~~"" 14 fo:~~CAcharya Nagarjuna University

Also, E is a normal separable extension of Q. This gives

O(G(E,Q) ) = [E:Q] = 8.
Put a = 1./2 . If a
G(E,Q) and ~ E E, then
E
2 3 2 3
~ = ao + a, a + a2 a + a, a + a, i + a, i« + a, (ia )+ a7 (ia ).

So we have
aW) = ao + al a(o,) + a2 a(o,i + a, a(o,)3 + a, g(i) + as o(i) a(o,)
+ a, a (i) o'(0,2 )+ a7 a (i) ae0,3 ).
Therefore, a is determined by its effect on i and on a . Because a(i) must be i
or -i ; because a( a) must be a, -a ,ia, or - in ; and because there are eight
elements in G(E, Q) , it follows that
G(E,Q) = { 0"1.0"2, 0"3, 0"4, 0"5, a6 .e-i , 0"8} ,

where crt 0,-+0" 1-+ 1 cr3 : a -+ -cc , i-+ 1

as : a-+ a, i-+ -1 cr7 : a -+ -a, i -+ - i

cr2 : 0,-+ la, 1 -+ 1 a4 : a ~ -io , i-+ 1

a6 : a -+ -a, 1 -+ -1 ~:, ex -+ -io, i -+ -i

Let us take
VI =a, V2 = la,

h
V3= -a, V4 = -la.

d2
V2
-,

11
-.
-.,

dl V3 VI

4
Then the elements of G(E,Q) permute the roots VI , V2, V3 and V4 ofx -
2 as follows.
0 .
crt: 0 rotation ; 0"5 : ref1e~tion about d.
a2 : 90 ° rotation
. ; a6 : reflection about II
~esson: 21 ~J~~~~~~~~GD:S~~~~~~~~~LGEB~
a3 : 1800 rotation; a7 : reflection about d2
0
a4 : 270 rotation; as : reflection about lz .

21.4 : SUMMARY:
In this lesson we have studied the following:
1. Let K be a normal extension of F and let H be a subgroup of G(K,F). Let
KH ~ { X E K / a(x) = x, for all 0' E H } , be the fixed field of H. Then
(i) . [ K : KH ] = O(H)
(ii) H = G(K,KI-!). In particular H = G(K,F) , [ K : F] = 0 ( G(K,F».
2. Let K be the splitting field of f(x) E F[ x] and let p(x) be an irreducible factor
off(x). If the roots ofp(x) are ai , a2, ..... , am, then for each i, there is an
automorphism o, in G(K,F) such that O'i(al ) = ai .
3. K is a normal extension of F, if and only if K is a splitting field of some
polynomial over F.
4. Fundamental theorem of Galois theory.

21.5 TECHNICAL TERMS:

Normal Extension
Galois group of a polynomial
Klein 4 - group

21.6: MODEL EXAMINATIONS:

Part - A ( Short Answer·.(fpestions )
1. Show that Q( Ji, J3 ) is
a normal extension of Q.
2. Prove that the Galois group of x3 - 2 over Q, the field of rational
numbers, is isomorphic to the symmetric group S3 .
3. Show that C is a normal extension of R.
2ni

4. Find G( Q(w), Q), where w = e 5, is a 5th root of unity. Show that

Q(w) is a normal extension of Q.
5. Show that R is not a normal extension of Q.
Part - B ( Long Answer Questions)
1. Show that K is a normal extension of if and only if K is a splitting field
of some polynomial over F.
2. State and prove the fundamental theorem of Galois theory.

23
 Distance Education 16

21.7 ANSWERS TO SELF ASSESSMENT QUESTIONS:

Self Assessment Question 21.2.11: Show that Q( N) is a normal
extension of Q.
Solution: It is evident that - N is a root of X2+ 2. By Eisenstein's criterion
with p = 2, X2 + 2 is irreducible over Q and its roots are N, -N.
Since N E Q( N) and Q( N} is a field ---J-2 E Q( --J-2). SO
Q( N) is splitting field of X2 + 2 over Q. Hence Q( --J-2) is a normal
extension of Q.

Self Assessment Question 21.2.12 : Show that Q( 1J2) is not a normal

extension of Q.
Solution: It is easy to see that a = 1[2 is a root of X4- 2 over Q and this is
irreducible over Q. Since
x4 - 2 = ( x2 - 12 ) (x2 + 12 )
x - 1J2) ( x +1J2) ( x - i1J2) ( x + i 1J2) , the roots of x4 - 2
= (
are ±a, ±ai Out of these four roots ±a E Q( 1[2) , so that Q( 1[2) is the splitting
field of X4- 2. Hence Q( 1[2) is not a normal extension of Q.

21.8 EXERCISES:
(1) Show that Q(5 J7) is a normal extension of Q.
(2) Show that Q( ~ ) is a normal extension of Q.
(3) Show that the field generated by a root of x3 - x-lover Q is not
normal over Q.
(4) Show that Q( 12) is normal extension of Q, while Q(1J2) is not a
normal extension of Q.
(5) Prove that the Galois group of x3 - 2 over Q, the field of rational
numbers, is isomorphic to the symmetric group S3 .
(6) Prove that the Galois group X4 - 2 over Q is isomorphic to the
Dihedral group of order.
(7) Prove that the Galois group of X4 - X2 = 6 over Q is isomorphic to
Klein's 4 - group .
(8) Prove that the Galois group x3 - 3 over Q isomorphic to S3 .
(9) Prove that the Galois group of X4 + 5x + 6 over Q is isomorphic
Klein's 4 - group.
 Q

~sson; 21 ~""~~===~~~~GD.s~~~~~~~!(G-L-G"-EB-RA-""'~
(10) Show that Q( Ji,.J3) is a normal extension ofQ.
(11) Consider Q, Q( Ji), Q(~). Show that Q(.J3) is a normal
.;",

extension of Q ; Q( ~) is not a normal extension of Q.

21.9' REFERENCE BOOKS:

1. ARTIN, EMIL, Galois Theory, 2nd ed Notne Dame, Ind : University of
Natne Dame Press, 1955.
2. FRALEIGH, lB., A First Course in Abstract Algebra. 2nd ed., Addision
Wesley Publishing Company.
3. HERSTERIN, IN., Topics in Algebra, Vikas Publishing House.
4. JACOBSON, N., Lectures in abstract algebra. Vol.II, Nostnand, 1953.
5. BHATTACHARYA, P.B., JAIN. S.K., NAGPAUL.S.R., Basic Algebra"
~nd Edition) , Cambridge University Press.

Lesson Writer
Prof. L.NAGAMUNI REDDY
S.Y.University, Tirupati.
 LESSON - 22 : SOLVABILITY BY RADICALS
CONTENTS
OBJECTIVES
STRUCTURE

22.1 INTRODUCTION
22.2 SOLVABLE GROUPS
22.3 THE SYMMETRIC GROUP Sn
22.4 SOLVABILITY BY RADICALS
22.5 GALOIS GROUPS OVER THE RATIONALS
22.6 SUMMARY
22.7 TECHNICAL TERMS
22.8 MODEL EXAMINATION QUESTIONS
22.9 ANSWERS TO SELF ASSESSMENT QUESTIONS
22.10 EXERCISES
22.11 REFERENCE BOOKS

OBJECTIVES:
After reading this lesson you should be able to :
• Know about solvable groups
• Know about commutator sl!pgroup , or, derived sub group of a group.
• Obtain a criterion for solvability of a group interrns of commutator subgroups.
A"\"
• Know about some propertiesof the symmetric group S, .
• Establish that Sn is not solvable for n ~ 5 .
• Know about solvability of a polynomial by radicals.
• Obtain a criterion for solvability of a polynomial by radicals
• Establish that a general polynomial of degree n ~ 5 is not solvable by radicals.
• Give a criterion for solvability of polynomials of prime degree over rationals.

22.1 INTRODDCTION :
We are ail familiar with the fact that the quadratic polynomial ax2 + bx + c , a 1:- 0
with real coefficients has ( -b ± ~b2 - 4ac ) / 2a as zeros in C. For example, the
 Distance Education 2 ~~~= Acharya Nagarjuna University

polynomial X2 + 2x + 3 over the field Q of rational numbers has 1 ± F-2 as zeros and it
is easy to see that they lie in Q( F-2). In other words Q( F-2) is the splitting field of X2
+ 2x + 3 and its zeros are expressed in terms of radicals. So one gets the natural
question, whether the zeros of polynomials of degree > 2 over Q, can be expressed by
radicals. We know that the answer is 'Yes' for cubic and biquadratic polynomials. In
fact there are well known methods, which express the zeros of cubic and biquadratic
polynomial by radicals. Mathematicians over the centuries tried in vain to prove the same
for a quintic ( a polynomial of degree 5) over Q. It was a triumph when Abel proved that
a quintic over Q need not be solved by radicals. Our job in this section is to establish this
and it should be a treat to watch the application of the sophisticated algebraic tools that
are developed so far. We start with some group theoretic notions about solvable groups
and commutator sub-groups.

22.2 SOLVABLE GROUPS:

Definition 22.2.1: A group G is said to be solvable if there is a finite chain of
subgroups.
G=No::::lNj::::lN2::::l ::::lNr={e} (1)
rrf G such that each N, is a normal subgroupof Ni., and such that every factor group Ni-I /
N, is abelian. The series 1 is called a solvable series.
Example 22.2.2: Every abelian group is solvable.
·Solution: Trivially No ::::lN I , where No = G and N I = {e} is a solvable series.
Example 22.2.3 : The symmetric group S3 is solvable.
Solution: We know that the alternating group A3 is the only normal subgroup of S3 Let
us take No = G, N I = A3 and N2 = {e} Then N I is normal in No and N2 is normal in N I .
_ . O(N) O(S) 6 lIds
Further o(NolN I) - 0 = __ 3- =- =2 and O(N IINZ)
. O(Nj) 0(A3) 3 ';~J'

= O(Nj) = 0(A3) = ~ = 3. But the groups of order 2 and 3 are abelian, so No/N I
0(N2) O(e) 1
and N1/Nz are abelian. Hence S3 is solvable.
Self Assessment Question 22.2.4: Show that S4 is solvable
We shall give a criterion for the solvability of a group in terms of commutator
subgroups of G. For this we need the definition of a commutator subgroup of a group.
Definition 22.2.5 :Let G be a group and a, bEG. The element a-I b-!,.ab of G is called
the commutator of the elements a b in that order and it is denoted by rti;b] .
Example 22.2.6: For a = (1231322 3J , 8 = (1 2 3J'r III
.
.~.
S1, find [ «., 8 ].
Solution: It is easy to check a-I = (1 21 23J, 0-1 = <5. So
. 3

[a
, 8] = a-I 8-1a 8 = (1312321231321
2 3J (1 2 3J (1 2 3J (1 2 3J = a
Self Assessment Ouestion 22.2.7 : For a, bEG show that [ a, b = [ b,a ] <, r
I I
Ramark 22.2.8: For the identity element e of a group G,[e,e] = e- e- e e = e. So the
identity e itselfcan be viewed as a commutator in a group G. Further SAQ 22.2.7 shows
that the inverse of a commutator is again a commutator. So, if the product of two
commutators is again a commutator, then the set of all commutators in a group will form
a sub group of G. Unfortunately, the product of two commutators is not a commutator in
general (check). Then how to form a subgroup out of the set of commutators? To
achieve this we seek the help of the notion of the subgroup generated by a set of element
ofa group.
Definition 22.2.9 : Let G be a group and C, the set of all commutators of G. Then the
subgroup of G generated by the set C is called the commutator subgroup or derived
subgroup of G and it is denoted by G I .
Let us recall that the subgroup <S> , generated by the set S is defined by
<S> = { XI , X2, .... , x, I Xi E S U S-I and r is a positive integer} ,
where S-I = { S-II s E S }. That is < s > is the set of all products of finite number of
elements in S u S-I . Since the inverse of a commutator is again a commutator, we have
c' = { c- I
I CE C } = C,
l
so that C U C = C. Thus any X E G I is given by
X = c] C2 cr,
where each c, is a commutator. The following theorem completely characterizes a
commutator subgroup of a group G.
Theorem 22.2.10: Let G be a group and G I its commutator subgroup. Then
(i) G I is a normal subgroup of G ;
(ii) GIG I is abelian, and
If N is a normal subgroup of G such GIN is abelian, then
(iii) G I c N ( In
other words G I is the smallest normal subgroup of G such that GIG I is
abelian ).
Proof: (i) That G I is a subgroup of G follows from the fact that G I is the subgroup
generated by the set C, the set of commutators of G. To see that G is normal in G, let g I

E G and x E G I. Then

X = c] C2 C3 Cn ,

for some c, and C and some positive integer n. So

 Distance Education 4
-I -I
g X g = g C1 C2 c, g
= gcg -I gC2g-I gC3g-I gcng
-I

For 1 :::;;
i :::;;
n , consider g Cig-I . Since c, is a commutator,
c, = a-I b-I ab,
for some a,b in G. So
gc, g--I = g a-I b-I ab g-I
= ga-Ig-I gb-Ig-I gag" gbg"

= u-I V-Iuv, where u = g ag' and v = gbg".

Thus g x g-I is a product of finite number of commutators, so that g x g-I E G I.
Hence G I is normal in G.
(ii) Let X = Gig, Y = G I h E GIG I then X-I = G I g-I , V-I = G I h-I . So
XV X-I V-I = GI g-I'h-\ gh = GI, since g-I g-I gh E G I •
But G I is the identity element of GIG I. SO this implies that and GIG I is abelian.
(iii) Let N be a normal subgroup of G such that GIN is abelian. Let a, bEG.
Then Na Nb = Nb Na
On simplication this gives N a-I b-I ab = N
so that a-I b-I ab E N or [ a, b ] EN. That is , N contains every commutator of G,
so that G eN.
I

We now define higher corrimutators. Treating G I as a group, we can consider its

commutator subgroup (G I)'. For facility of notation we shall denote it by G(2) . In
general ( or inductively ) we shall define G(k) = (G(k-I» I and call G(k) as the k th
commutator subgroup of G. Thus the sequence of commutator subgroups G(k), k = 1, 2,
....... Satisfy the following properties:
(i) G(k)is a normal subgroup of G(k-I)
.1:..
(ii) G(k-I)I G(k)is abelian.
, .
Let us prove some useful lemmas.
Lemma 22.2.11 : If N is a normal subgroup of G, then N' , the commutator subgroup of
N, is also normal in G.
Proof: Letg E G and x E NI . Then
x == CI C2 ... Cr
for some commutators Cj' C2 ... c, in N. So
-I _. -I _ -I -I -I
g X g - g CI C2 Cr g - gCI g C2g gc, g .
Now, for any i, 1:::;; i :::;n,
g c, g-I = g n -I m-I n m g-I ,lor
c
some n,.' mill. N
= gn -\ -I . -1 -I
g g m g ,g ng gmg -
-1 -\ _ -1-1
n] ill] nl ill] ,
where g n i] = n] and gmg' = m, . Since N is a normal subgroup of G, n] , m, EN. So
g c, g-l E N I and thus g x g-l E N I. This proves that N I is normal in G.
This lemma has an important corolloary.
CorollarY'22.2.12: Every G(k)is a normal subgroup of G.
Proof: We have seen that the commutator subgroup G' is a normal subgroup of G. So
by the Lemma 22.2.11, G(2) is a normal subgroup of G and hence G(3) is a normal
subgroup of G. Proceeding in this way we observe that G(k)is a normal subgroup of G
for every positive integer K.
The following theorem gives a critierion for solvability of a group in terms of its
commutator subgroups.
Theorem 22.2.13: G is solvable if and only ifG(k) = {e} for some positive integer k.
Proof: Let us assume that G(k)= {e} for some positive integer k. Then trivially
. I (2) (k) _
G :::lG :::lG :::l :::lG - {e},
is a solvable series for G. So G is solvable.
Conversely, assume that G is solvable. Then there exits a solvable series
G = No:::l N, :::lN2:::l :::lN, = {e},
such that each N, is a normal subgroup of Ni., and N-l / N is abelian.
By (iii) of the Theorem 17.7.13 , G' C N, . Now
GleN
_ I ~ G(2) _e N11 .

Since N2 is a normal subgroup of N, such that NJ/N2 is abelian, again by (iii) of

the Theorem 17.7. 13 we have
NJ eN2 (2)
From (1) and (2) we get G(2) c N2 .
Proceedings in this we obtain ( ;ff~~
k stages)
G(r) c NT = {e}

This forces that G(e) = {e} and the converse is established.

The following Corollary is immediate.

'-,
Self Assessment Question 22.2.14: Prove that every subgroup of a solvable group is
solvable. ,T,'

Self Assessment Question 22.2.15: If G is solvable and if H is a homomorphic image

of G, then H is solvable.
 Distance Edu~ca~ti~o;n --==~~~~S: 6

22.3 THE SYMMETRIC GROUP Sn.1

The group theoretic results obtained so far are of general nature. The next few
results are slightly in different vein. They are concerned about the symmetric group S, ,
on n symbols. .
Lemma 22.3.1: If N is a normal subgroup of Sn , n 2: 5 and if N contains every 3-
cycle in S, , then Nr also contains every 3 - cycle.
Proof: Let us assume that N is a normal subgroup, of S, , n 2: 5, and that N contains
~ery 3 - cycle in N. For i.j.k in 1,2, .... , n , let (i,j,k) be any 3 - cycle in Sn .
Consider 3 - cycles a = ( 1, 2, 3 ) and b = ( 1, 4, 5 ) in N. (This is possible since n
2: 5 ). Then the commutator a-I b-I ab EN' . But
a-I b-I ab = (3,2, 1) (5,4, 1) ( 1,2, 3) ( 1, 4, 5) = ( 1,4, 2)
So ( 1,4,2) EN'. By Lemma 22.2.11, Nr is also a normal subgroup of Sn . So
d-I ( 1, 4, 2) 0' EN' , (1)
for every c E S. Consider ss E S, such that
cr(I)=i, cr(4)=j, cr(2)=k.
Then
[ cr-I (1,4,2) cr] (i) = 0' { ( 1, 4, 2) [ 0'-1 (i) ] } = cr{ (1,4,2) (i) }
= 0'(4) = i
In the same way , we can see that
[ cr-I (1,4,2) 0' ] CD = k and l c' (1,4,2) (o) ] (k) = i
Further if r =1= 1, 4, 2 , so that r is fixed by o, ( 1, 4, 2) and e'. So r is fixed byo " (1, 4,
2) cr. Thus
0'-1 ( 1, 4, 2) 0' = ( i, j, k) ,'''-:
r V'!
So by (I), (i, j, k) EN'. That is N I contains ev~ry 3 - cycles. i!J,"

The following Corollary, which is a consequence of the Lemma 22.3.1, plays a

central role in establishing that Sn is not solvable for n 2: 5.
Corollary 22.3.2: If G = S, and n 2: 5, then G(k) contains every 3 -cycle for every
positive integer k .
Proof: But G = S, . Trivially G is normal in G and it contains all 3 - cycles. Taking N
= G in the Lemma a 22.3.1, we observe that G I contains all 3:- cycles. Since G I is
normal in G, again by the Lemma 22.3.1, G(2), contains all 3 - cycles; since G(2) is
normal in G, G(3) contains all 3 - cycles. Proceedings in this way we get that G(k)
contains all 3 - cycles, for every positive integer k . ) \'
Let us now prove the much awaited result, which plays a key role in proving that a
polynomial of degree n 2: 5 is not solvable by radicals.
Theorem 22.3.3: Sn is not solvable for n ;:::5.
Proof: Let n ;:::5 and G = Sn. Then by the Corollary 22.3.2, d(k) contains all 3 - cycles
in S, , for k = 1, 2, .. " So O(k):;t: {e} for every positive integer k. By Theorem 22.2.13 G
is not solvable.

22.4 : SOLVABILITY BY RADICALS:

We have gathered the necessary group theoretic information to tackle our problem.
Let us now see what we mean by , soJvability by radicals' of a polynomial.
Definition 22.4.1: Let F be a field. An extension K of F is called an extension by
radicals, or, radical extension, if there are elements a, , a2 , .... , a, E K and positive
integers nj , n2 '''' nr such that Fr = F(al) , .
F2= FI(a2) ,_.,', K = Fr = Fr-I (a"}; a~l EF, a~2 E F1 , ".,a~r EFr~l . A polynomial
f(x) E F[x] is solvable by radicals over F if the splitting field K of f(x) over F is an
extension of F by radicals.
Thus, i:Q( is a radical extension of F, then there are elements al , a2 , .... , a, in K
arid positive integers n I, n2 " , ,., nr such that
K = F(aJ, a2 '''., a.}, afE F and afi E F(al , a2 '" '" ai-I) for I ~ i ~ r .
One can see that
(I) every radical extension of F is a finite algebraic extension of F, and
(2) every radical extension ofF is a normal extension ofF.
iExample22~·4.2: Every quadratic polynomial over Q is solvable by radicals.
Solution: - Without less of generality we can consider a monic quadratic polynomial over
Q, say, "r.
f (x) = X2 + ax + b, a:>b'E Q.
The roots <Xland <X2off(x) are given by
-a+.Ja2-4b -a-.Ja2-4b
a1 = 2 a2 =
2

So, if we put d = ~a2 -4b and QI = Q(d), then it is easy to see that <Xl, <X2E Q(d).
Further
d2 = a2 - 4b E Q
Thus, the fields Q7'~d Ql are such that Ql = Q(d) , d2 E Q and QI contains all the roots of
f(x). So f(x) is solvable by radicals. .

Definition 22.4.3 : An element w of a field of F is called an n th primitive root of unity

if w" = I and wn--;t I if a < m < n.
 Distance Education 8

Example 22.4.4: In ( Zs , +, . ) , find 4th primitive roots of unity.

Solution: In ( Z, , +, . ) , the 4th roots of unity are 2 , "3 and 4 ( check). Let us see
which of these are primitive 4th roots of unity. Here
( 2:)1 = 2::1= 1 ; ( 2: i;::: 4:1= 1 ; ( 2:)3 = "8= 3 "* 1 ; ( 2:)4 = 16 = 1
( "3)1 = "3:1= 1 ; ( "3)2 = "9= 4:1= 1; ( "3)3 = 27= 2:1= 1 ;
( 4)1 = 4 :1= 1 ; ( 4)2 = 16= 1.
So 2 and "3 are primitive 4th roots of unity, while 4 is not.
Theorem 22.4.5: Let n be a positive integer. Suppose the field F has all the n th roots
*
of unity and a E F, a 0 . Let K be the splitting field of x" - a E F [x] over F. Then
(i) K = F (u) where u is any root of x" - a, and
(ii) The Galois group of x" - a over F is abelian.
2ni
2 n1
Proof :(i) The nth roots of unity are given by L W, w , , w -,where w = e n
Observe that
wl1 = 1 but wm:l= 1 if a < m < n ( 1)
. a root a f x n - a , then
If u IS en iIt IS
. easy to see that u uw, UW 2 , ..... ,U ware
at u, n-I aIso
the roots of x" - a . ( Verify? ) Further these all distinct, For, if uw' = u W , i < j ,
i
then wi- = I and 0 < j - i < n. This is a contradiction to (1). Thus u, u w, , U wn-l
give all the roots of x" - a . Since w' E F for ..0:::;;i:::;;n - I , we have u, uw , ,U
wl1-1 are all in F(u). That is, x" - a splits over F(u) . Further if x" - a splits in a field K,
which is such that Fe K c F(u) , then u E K. This together with Fe K implies that
F(u) c K, or, F(u) = K. Thus F(u) is the splitting field of x" - a over F. So G(K,F) is the
Galois group of x" - a over F. .
(ii) Since K = F(u) is the splitting field of x" - a over F, G(K,F) is the Galois group
of x" - a over F. Let a , 'T E G(K,F). Then ojand 'Tare automorphisms of K, fixing
every element of F. So
a (wi) = wi and 't(wi )- = Wi for 0:::;;i :::;;
n - 1 . Further since u is a root of x" - a,
a(u) and 't(u) are also roots of x" - a and thus
cr(u) 't(u) E { u, uw, uw' , , U wl1-1 }

So o(u) = uand 'T(u) = u W for some i,j , 0:::;;i,j :::;;

Wi n - 1.
i
Hence a 't(u) = u w +i = 't a(u)
Further a 't (a) ='t a(a) for all a in F. ) .-
Thus a"( (x) = "( o(x) for all x in F(u) and this means that "h~= 'to" for all rrt E

G(F(u),F). That is, the Galois group x" - a is abelian. '.

 Remark 22.4.6: Suppose F has all the nth roots of unity. Then adjoining one root u of x"
- a to F, we obtain the splitting field feu) of x" - a over F. So, F(u) is a normal
extension of F.
We now prove final theorem in this direction ..
Theorem 22.4.7 : Let F be a field which contains all the nth roots of unity for every
positive integer n. If f(x) E F[x] is solvable by radicals over F, then the Galois group of F
is a solvable group. .
Proof: Let f(x) Er F[x] and let K be the splitting field of f(x) over F. Asume that f(x) is
solvable by radicals over F. Then there exist a finite number of elements WI, W2, ..... , Wr
in K and positive integers n I , n2 , , n, such that
FI = F(Wl), F2 = F1(W2), , F, = Fr-I (w.};

and all the roots off(x) lie in Fr: Since F, contains all the roots of f(x) , K c Fr.
We may assume that F, is a normal extension of F.
Let z, be a primitive n, th root of unity in F. Then z, E Fj_1 and the n, roots of
x"' - W!lj , namely,
w, , Wiz, , w, z? , .... , w, z;"-I are all in Fi. So F, is the splitting field of x"! - w;"
over Fi-1• So by the Theorem 22.4.5, G(Fi , Fi-l ) is abelian and so F, is a normal
extension of Fi-1 . So F, is also a normal extension of Fi-I for i=I,2, ... , r (Here Fo= F).
Since F, is a normal extension of Fj-1 , by the fundamental theorem of Galois
theory G(F r , F, ) is a normal subgroup of G(F r , Fi-1) for i= I,2, ... , r. Consider the chain
G(Fr, F):J G(Fr, F1):J :J G(Fr, Fr-I ) (1)
Since each F, is a normal extension of(Fi-1 , again by the Fundamental theorem of Galois
theory we have _ '
= G(Fr,
J ,

G(Fj, Fi-1 ) Fi-1 ) / G(Fr, Fi ).

But G(Fi , Fi-1 ) is abelian, so that each of G(Fr , Fj-1 ) / G(Fr , F, ) is abelian and (1) is a
solvable series for G(Fj-, F). So G(Fr , F) is a solvable group.
Further K being a splitting field of f(x) over F, it is a normal extension over F and K
c F, . Again by the fundamental theorem of Galois theory G(Fr , K) is a normal
. subgroup of G(F~, F) and G(K,F) is isomorphic image of G(Fr, F) / G(Fr,K). That is
G(K,F) is a homomorphic image of G(Fr, F) (under the above isomorphism ). Since G( F,
, F) is solvable, G(K,F) is solvable. Now K being the splitting field of f(x) over F,
G(K,F) is the Galois group off(x) over F and the theorem is proved. •
We now prove the classic theorem of Abel.
 Distance Education 10 ~~~:: Acharya Nagarjuna University

Theorem 22.4.9: The general polynomial of degree n ~ 5 is not solvable by radicals.

Proof: Let x" + al xn-l + + an be the general polynomial of degree n ~ 5 . Consider
the field of'Fra, , a2 , .... , an). We know that the Galois group of x" + al xn-] +
....... + an over F(al, a2 , , an ) is the symmetric group S, .. In Theorem 22.3.3, we
have seen that S, is not solvable for n ~ 5. So x" + a, xn-l + .....+ an is not solvable by
radicals over F for n ~ 5 .

22.5 GALOIS GROUPS OVER THE RA110NALS :

In this section we give a criterion for solvability of an irreducible polynomial of prime
degree p over rationals, which has exactly two non real roots. For such a polynomial we
shall show that its Galois group over Q is the symmetrical group S,
Theorem 22.5.1 : Let q(x) be an irreducible polynomial of degree p, p a prime, over the
field Q of rational numbers. Suppose that q(x) has exactly two nonreal roots in the field
of complex numbers. The Galois group of q(x) over Q is S, , the symmetric group of
degree p. Thus the splitting field of q(x) over Q has degree p! over Q. .
Proof: Let K be the splitting field of the polynomial q(x) over Q and let a be a root of
q(x) in K. Then, since q(x) is irreducible over Q,
[ Q(a) : Q ] = p .
NowQcQ(a)cK and
[K: Q ] = [ K: Q(a) ] [ Q(a) : Q] = [K :.Q(a) ] p .
So p divides [ K : Q ].
Suppose G is the Galois group ofK over Q, that is, G = G(K,Q) .
Then O(G) = [ K : Q], so that p divides O(G). Hence by Cauchy theorem, G has an
element (J of orderp.
Let al , a2 be the two nonreal roots of q(x)!o Then ( since complex roots occur in
pairs),_ al = (;,2 and a2 = (;,1. Let a3, a4, .... , ap be the remaining real roots of q(x) ,
Then ai = ai, for 3 ::; i ::;
p.
Consider the map r : K ~ K given by
'tea) = a for all a E K.
It is easy to see that 't is an automorphism of K such that
- -
'teal) = a] = a2; t(a2) = a2 = al
and t( ai) = «. : 3:$; i :$;p. ,I:,
Further the elements of G = G(K,Q), take the roots of q(x) into roots of q(X), and
this induce permutations of al , a2 , , ap , In this way we imbed G in S, , the symmetric
group on al , a2 , ... , ap , or 1,2, , P . That is , G c Sp. Also the' automorphism 't
described above is transportation (1,2) since t(al)= a2,'t(a2) = (Xland 't( (Xi)=(Xi,3 ::;;i::; p.
 Mow the element a E G has order p. So o as .m element S, ( since G is imbedded in
Sp) has order p. But the elements in Sp, which have order p are the cycles. Soc is a p
- cycle. Thus cr and 't generate Sp. But a and 't are in G. So the group generate by o
and 1: must be a subgroup of G. So S, is a subgroup of G. Thus G = Sp. This shows that
the Galois group of q(x) over Q is indeed Sp. This proves the theorem.
Remark 22.5.2: In determining that an irreducible polynomial of prime degree over Q
has exactly two non real roots, we use the following two criteria, one from the theory of
numbers and other from calculus :
(1) Descante' s Rule of Signs
(i) The number of positive real roots of q(x) ~ the number of changes of sign in
q(x).
(ii) The number of negative real roots of q(x) ~ the number of changes of sign in
q(-x).
(2) Intennediate value Theorem of Calculus
If f(x) changes sign ( from positive to negative or negative to positive) in an
interval (a,b), then f(x) has a real root in (a,b).
As an application of this 'th' lrem let us consider the following example.
Example 22.5.3 :Show that tne polynomial q(x) = 2x5 - lOx + 5 is not solvable by
radicals over Q.
Solution: Here deg q(x) = 5, a prime and by Eisenstein's criterion it is irreducible over
Q ( with p = 5). To establish that q(x) has exactly two non real roots let us apply
Descarte's Rule of signs to q(x).
The signs of q(x) are: + +
So there are two changes of sign in q(x) and these number of positive real roots of
q(x) is ~ 2 . :n2-
The signs of q(-x) are : + if,- +
So there is only one change of sign in q(-x) and thus the number of negative real roots
of q(x) is ~ 1 .
So the number of real roots of q(x) is ~ 3.
Since q(x) changes sign (from -ve to +ve ) in the interval (-1,
theorem of calculus there is a real root in it.
°), by intermediate value

Similarly q(x) changes sign in the intervals (0,1) and (1,2). So q(x) has a real root in
each of these intervals. Thus q(x) has exactly three real roots. .
So q(x) has exactly two nonreal roots. Hence the Galois group of q(x) over Q is
isomorphic to the symmetric group S5 .
But we know that the symmetric group S, is not solvable for n ;:::5. So S5 is not a
solvable group and thus q(x) is not solvable by radicals over Q.
 Distance Education 12 )5~~==- AcharyaNagarjuna Universi< "E5

7
Self Assessment Question 22.5.4 : Show that the polynomial f(x) =X - IOx5 +15x + 5
is not solvable by radicals over Q.

'22.6 : SUMMARY:
In this lesson we studied the following:
1. Properties of solvable groups.
2. Properties of commutator subgroups :
Let G be a group and G' its commu~ator subgroup . Then
(i) G' is a normal subgroup of G ;
(ii) GIG' is abelian, and
(iii) If N AS a normal subgroup of G such that GIN is abelian, then G' c N (
In oi'fier words G' is the smallest normal subgroup of G such that GIG' is
abelian ).
3. A group is solvable if and only if G(k)= {e} for some positive integer k ,where G(k)is
the commutator subgroup of order k.
4. Sn is not solvable for n ~ 5 .
5. Let n be a positive integer. Suppose the field F has all the n th roots of unity and a E
F, a"* 0 . Let K be the splitting field of x" - a E F [xl over F. Then
(i) K = F (u) where u is any root of x" - a , and
:)

(ii) The Galois group of x" - a over F is abelian.

6. Let F be a field which contains all the nth roots of unity for every positive integern. If
f(X) E F[x] is solvable by radicals over F, then the Galois group of F is a solvable
group.
7. Let q(x) be an irreducible polynomial of degree p, p a prime, over 'the field Q of
rational numbers. Suppose that q(x) has exactly two nonreal roots in the field of
complex numbers. The Galois group of q(xtover Q is S, , the symmetric group of
degree p. Thus the splitting field of q(x) over Q has degree p! over Q.

22.7 : TECHNICAL TERMS:

Solvable group
Commutator
Commutator subgroup or derived subgroup
Solvability by radicals
Primitive nth root of unity .
 22.8 : MODEL EXAMINATION QUESTIONS:
Part - A Short Answer Questions
1. Show that S4 is solvable .
2. If N is a normal subgroup of S, , n ~ 5 and if N contains every 3 - cycle in
Sn , then N I also contains every 3 - cycle.
3. Show that the polynomial X1 - IOx5 + 15x + 5 is not solvable by radicals over Q.
Part B - Long Answer Questions
1. Show that G is solvable if and only if G(k)= {e} for some positive integer k.
2. Show that Sn is not solvable for n ~ 5 .
3. Let n be a positive integer. Suppose the field F has all the n th roots of unity and
*
a E F, a 0 . Let K be the splitting field of x" - a E F [x] over Then show that
(i) K = F (u) where u is any root of x" - a, and
(ii) The Galois group of x" - a over F is abelian.

4. Iff(x) E F[x] is solvable by radicals over F, show that the Galois group of fix) is a
solvable group. Hence or otherwise. That a general polynomial of degree n ~ 5 is not
solvable by radicals.

22.9 : ANSWERS TO SELF ASSES~MENT QUESTIONS:

. Self Assessment Question 22.2.4: Show that S4 is solvable .
Solution: Take No = G , N, = ~, N2 = V4 , N3 == {i} , where An is the alternating group
on n symbols and V4 = { (1,2) (3,4), (1,3) (2,4), (1,4) (2,3)}. Observe that N is normal

Ni-t for i=O,I,2,3. Since O(No/Nt)17) 24 = 2,

12
O(Nt / N2) = 12 = 3, O(N2
4 N3
J = !-.
1
= 4,

the group No/Nt , Nt/N2 , N2/N3 are-abelian .

r
Self Assessment Ouestion 22.2.7 : For a, bEG show that [ a, b t = [ b.a]
Solution: For a, bEG,
[a, b r'
= ( ab a-I b-tyl = ba b-1 a-I = [ b, a]
Self Assessment Question 22.2.14: . show that every subgroup of a solvable group is
solvable.""
Solution: Let 8(1)e a solvable group. Then G(k)= { e} for some positive integer k.
Suppose ab a-h'b-I be a commutator in H. Then a, b E H . Since H is a subgroup of G,
a, bEG and ab a-~b-1 is also a commutator in G. So
H1cG1.
. '.
 Centre for Distance Education 14 Acharya Nagarjuna University

23.7.3 Define the notions of a partially ordered set and a lattice and give an example of a partially
ordered set which is not a lattice.

23.7.4 State the principle of Duality in lattices.

23.8 ANSWERS TO SELF ASSESSMENT QUESTIONS .
23.2.4 Clearly, :s: is reflexive and antisymmetric.

a:S:l band b:S:1 c rsb s;a and c:s:b=:>c:s:b:s:a=:>c:s:a=:>a:S:l c and hence :S:1 is

transitive also. Thuz :S:1 is a partial order.

23.2.6 Excspt in (1), all others are not chains, in generalln(2), we have 2 doesn't divide 3 and 3
doesn't devipe2 and hence 21,3 and 3 1,2 .
,
In (3), if IXI::::; 1, then P(X) has atmosttwo elements, namely ¢ and X. If IXI> 1, choose

x 7= yE X. Then {x}g;{y} and {y}g;{x}.

In (4), if G ls any. cyclic g~oup of order pn, where p is prime, then the subgroups of G form

a chain. In·g~,8eral, cllfG) is not a chain. So is the case n( G) of normal subgroups of G.

23.2.S Let (p,:s:)p.~a finite p.o. set. Then the dual poset (p,::::) of (p,:s:) is also a finite p.o. set
and hence has a maximal element (by the argument given just above 23.2.8) which becomes
9 minimal element in (p,:::;).
2~.2.10 W.e h~ye P={v/nis~po~itive.int~g~randn/2q}

={1,2,4,5,lO,20}

There is pnlyone minimal element, namely 1. 2 al)~p are the minimal elements of P - {1}.
Next, 4 and ~.pare minimal elements in P - {1 , 2, 5h~nd.20 is the greatest element in P. Join
1 and 2, and \1,and 5 by straiqht lives and join 2 to 461:2to 10 and 5 to 10. Finally join 4 to 20
and 1Q to 20, as, given i,nthe diagram. Thisis the Hassas diagram of P.

20
10

23.2: 11 Let X::: {a, b, c} Then P (X) will have eight ( 23) elements and

p(X)={¢,{a}, {b},{c}, {a,b}, {b,c}, {a,c},X} . The Hassae diagram of P(X) is given bellow.
24
"
 8~ G ~ ~) in S3. find
(i) [ a , 0] (ii) [0, a] (iii) [ a , y] (tv) [ y, a] (v) [ y, 8] (vi) [ 8 , y]
check whether [ a, 0 ] = [ 0, a] , [ a, y ] = [ y, a]
and [ y, 8 ] = [ 8, y ]
2. Show that the following polynomials are not solvable by radicals over Q.
(i) x5 - 9x + 3 (ii) 2x5 - 5x4 + 3
r
(iii) x5 - 8x 6 (iv) x5 - 4x + 2

22.11 : REFERENCE BOOKS:

1. ART!N, EMIL, Galois Theory, 2nd ed Notne Dame, Ind : University of
Natne Dame Press, 1955.
2. FRALEIGH, J.B., A First Course in Abstract Algebra. 2nd ed., Addision Wesley
Publishing Company.
3. HERSTEIN, IN., Topics in Algebra, Vikas Publishing House.
4. JACOBSON, N., Lectures in abstract algebra. VoUI, Nostnand, 1953.
5. BHATTACHARYA, P.B., JAIN. S.K., NAGPAUL.S.R., Basic
Algebra, (2nd Edition) , Cambridge University Press.

Course. Writer
Prof. L. Nagamuni Reddy
S.V.University, Tirupati

i (, •

,. • f. • ~'.
Lesson· 23 LATTICES

OBJECTIVES
After reading this lesson, you should be able to
• define the concept of a partially ordered set and those of an upper bound, lower
bound, least upper bound and greatest lower bound of a subset of a partially ordered
set,
• define the notion of a lattice, prove the equivalence of the two types of definitions a
lattice and
• define the concept of a modular lattice and prove certain important properties of
modular lattices.

STRUCTURE
23.1 Introduction

23.2 Partially ordered sets

23.3 Lattices
23.4 Homomorphisms of lattices

23.5 Summary

23.6 Technical Terms

·23.7 Model Examination Questions

23.8 Answers to Self Assessment Questions

23.9 Exercises

23.10 Reference Books

23.1 INTRODUCTION :
George Boole's attempt to formalize propositional lbgic led to the concept of Boolean
algebras. While investigating the axiomatics of the Boolean algebras, the introduction of the concept
of lattices was found to be more useful and, in particular, Dedikind introduced the modularity which
is a weaker form of distributivity. It was Garret Birkhoff's work in the mid-thirties of the twentieth
century that started the general development of lattices as (universal) algebraic structures. He
demonstrated the importance of lattice theory by showing that it provides a unifying framework for
eitherto unrelated developments in many mathematical disciplines and, in particular, in the theory
of groups and rings where we are concerned primarily with certain distinguished subsets like normal
subgroups, ideals, etc.,
 Centre for Distance Education 2 Acharya Nagarjuna University

In the four lessons to come, including the present one, we present a brief discussion of
certain essential portions of lattice theory which are applicable to the theory of groups and rings.

23.2 PARTIALLY ORDERED SETS :

We shall begin with the following which is essential and fundamental for the study of lattices
and is a generalization of the usual ordering on the set of real numbers and of the inclusion ordering
among subsets of a set.

23.2.1 Definition: LetXbe a nonempty set. A binary relation R onX(that is R~XxX) is called
a partial order on X if it satisfies the following:

(1) (x,x)ER for all XEX (thatisRisreflexive)

(2) (x,y) E Rand (y, z) E R => (x, Z)E R (that is, R is transitive)

(3) (x, y) E Rand (y, x) E R => x = y (that is, R is antisymmetric)

Instead of denoting a partial order by R, it is customary to denote it by s and, we write a <b

if ( a, b) E ~ , Further, if a ~ b and a 1= b , we write a < b . Also we write b ? a and b > a for a ~ band
a < b respectively.

23.2.2 Definition: Apair (P,s)is 'called a partially ordered set (or, in short, p.o. set) if P is a
nonempty set and s is a partial order on P.
23.2.3 Examples:

(1) (~, s) is a p.o. set, where, IR is the set of real numbers and s is the usual ordering on ~

(2) Let P be the set of positive integers. For any a, bE P , define

/

as bqa divides b. (that is, ac=b for some CE p.D. Then (p,s) is a p.o. set.

(3) LetXbe a set and P(X) be the class of all subsets of X. For any A,BEP(X), define',

Th~n (p (.1:', ~)) is a po set, this order is called the inclusion order.

(4) The set of all subgroups of a group G together with the partial order as defined in (3)
above, is a p.o. set.
 (5) (n( G), ~) is a part, where n( G) is the set of all normal subgrooups-of G.

23.2.4 Self Asssessment Questions:

Let ::;:be a partial order on a set X. For any a, b EX, define a::;:j b <=> b ~ a

Then prove that ::;:j isa partial order on X. ::;:1 is called the dual order of ::;:. ::;:1 is usually
denoted by ~.

23.2.5 Definition: Let (p,::;:) be a partially ordered set such that fo~ any a, bE P , either a s: b or

b s: a- Then s is called a linear order or total order and (p,::;:) is called a .linearly ordered set or
totally ordered set or a chain.

23.2.6 Setf Asssessment Questions:

Whic~of the p.o. sets given in 23.2.3 are chains?

23.2.7 Definition: Let (p, s) be a p.o. set and a E P Then a is called maximal if there is. no
x E P such that a < x . Similarly, a is called minimal if there is no x E P such that x < a .
If a is maximal and a ::;:x . then we can immediately say that a = x . Similarly, if a is minimal
and x::;: a , then x = a . Note that there can be more than one maximal (minimai) element or there
can be no maximal (minimal) elements in a p.o. set. However, every finite p.o. set must have at
least one maximal element and at least one minimal element. For, if a E P, is not maximal, then
there exists a2 E P such that a <a2. 'If a2 is not maximal, then there exists a3 E P such that
j

a1 <a2 <a3 . This process cannot be continued indefinitely it: J! is finite and hence we arrive at a
maximal element

23.2.8 Self Asssessment Questions:

Prove that every finite p.o. set has a minimal element.
23.2.9 Definition:

Let a and b be elements of a p. o. set P . Then b is called a cover for a or a is said to be

.covered by b if a < b and there is no element x in P such that a < x < b and, in this case, we
write a-« b to say that a is covered by b.

In otherwords, we say that a is covered by b if, for any x E P: a s x s b implies either

a = x or x=b. In a finite partially ordered set, the partial order e can be expressed in terms of the
covering relation -< ; for, we have

as b ~ a = b or there exists a sequence

 Centre for Distance Education 4 Actlarya Nagarjuna University

This observation enables us to represent any finite partially ordered set by a diagram, known
as Hassae diagram, 'as given below. . .

We shall use induction on the number of elements in the p.o. set. If P has only one element,
> 1 and assume that any p.o. set with fewer
then represent it by writing a small circle or dot. Let n
than n elements can be represented by a Hassae diagram. Let P be a p.o. set with n elements.
Let M be the set of all minimal elements in P. Then M is a nonempty finite set. Represent each

. element of Mby a small circle, all in one row. Since M ;;t:.¢, Ip-MI<lpl=n and by the induction

hypotehsis, P - M can be represented by a Hassae diagram and now write , this diagram
.
above ,

the row representing the elements of M and join each element m EM with an element x E P-M
such that a < x bya straight line. This way, you get a diagram for P. Note that, in this diagram,
a<b if and only if there is an ascending broken line connecting a to b.

23.2.10 Self Assessment Questions :

Let P be the set of all positive divisors of 20. For any a, b P define
e a s; b if and only if a
divides b. Then (p,:s:) is a p.o. set. Write the Hassae diagram for this poset

23.2.11 Self Assessment Questions:

Write the Hassae diagram for the p.o. set (P( X),~) where .ris a t~rEJeelement sat.

23.2.12 Self Assessment Questions:

Describe the partial order on the p.o. set whose Hassae diagram is given below. How many
maximal elements are there?

23.3 Lattices :
In this section, we shall introduce the concept of a lattice and discuss two equivalent alternate
methods of defininq a lattice'. First, we have the following ..

23.3.1 Definition:
/
Let (p,:s:) be a p.o. set, S ~P and a E P:
·~--:-- -----
~ Lesson 2~ ~J. ( Algebra ~

(1) a is said to be an upper bound of 8 if s 'S, a for a'il S E 8 .

(2) a is said to be a least upper bound (Iub) of 8 if a is an upper bound of 8 and a ~ b for all
upper bounds b of 8.

(3) a is said to be a lower bound of 8 if a 'S, s for all SE 8.

(4) a is said to be a greatest lower bound of 8 if a is a lower bound of 8 and b 'S, a for all lower
bounds of b of 8.

Least upper bound of a subset 8 of a p.o. set P is unique, if it exists, and is also called
supremum of 8 and is denoted by

lub8 or sup8 or V8

Like wise, the greatest lower bound of 8 is unique, if it exists, and isalso called the infimum
of 8 and is denoted by

glbS' or inf 8 or 1\8.

-
23.3.2. Self Assessment Questions: : In the p.o. set given in 23.2.12, what are lub {a,b} and

glb{c,d}? Does glb{e,g} exist?

23.3.3 Definition: A partially ordered set (p, 'S,) is said to be a lattice if every two element subset

of P has lub and glb in P.

23.3.4 Examples:

(1) If 'S, is the usual order on the set of real numbers IR, then the p.o. set (IR, 'S,) is a

lattice, where, for any a, b E ~ ,

lub {a, b} = Maximum of a and. b

and glb{a,b}=. Minimum of a and b

(2) Let P be the set of positive integers. For any a, b e P , define a e b <=> a devides b .

Then (p,::;) is a lattice in which, for any a.b e P .

. .'

lub { a, b} = least common multiple of a and b

and glb {a, b l-. greatest common divisor of a and b .

 Centre for Distance Education 6 Acharya Nagarjuna University

(3) The p.o. set given in 23.2.12 is not a lattice since {a,b} has no lower bound.

Before going to the next theorem, let us recall that a binary operation on a set S is a
mapping of S x S in to S. In the following, we describe a lattice as a set together with two binary
operations satisfying certain conditions.
23.3.5 Theorem:

Let (L, ~) be a lattice. For any a, bEL, define

avb=lub{a,b} and a r.b = glb{a,b}_

Then v and 1\ are binary operations on L satisfying the following, for any a,b, CE L .

(1) av b+b v a and a r.b = b r. a (commutative laws)

(2) (avb)vc=av(bvc) and (al\b)l\c =al\(bl\c) (associative laws)

(3) a v a = a.=a 1\ a (idempotent laws)

(4) av(bl\a) =a=al\(bva) (absorption laws)

)

(a v b is to be read as ' a join" b ' and a 1\ b as ' a meet b ')

Proof: Let a,b, c be any elements of L . Then

av b= lub {a, b}=lub {b,a}=bva

and a 1\ b = glb {a, b} = glb {b, a} = b 1\ a

and therefore (1) is satisfied

(avb)ve = lub {a vb, e}

" "1 = lub {lub {a, b}, e}

=lub {a,b,c}

\ =lub {a,lub {b, en

=lub {a,bv e}'

=av(bvc)
 C Algebra ~

and, similarly, (a /\ b) /\ C = glb {a, b, c} = a /\ (b /\ c) and therefore (2) is satisfied, (3) is trivial.

To prove (4), note that b /\ a = glb {b, a} S; a and hence

av (b /\ a )=lub {a, b r. a} =a

Similarly a S; b v a and hence

a /\ (b v a) = glb {a, bv a}=a

In above, you can easily observe that, for any a, b EO L,

This is a hint to prove the following theorem which is the converse of theorem 23.3.5.

23.3.6 Theorem:
Let L be a set together with two binary operations v and /\ satisfying the conditions (1)
through (4) given in 23.3.5, for any a, b, CE L . Define a relation s; on L by

as; b if a vb =b
Then S; is a partial order on L such that (L, s;) is a lattice and, for any a, u E L .

Jub{a,b}=avb and glb{a,b}=a/\b

Proof: As per our definition, as; b iff a v b = b. We shall observe that this is also equivalent to
a=a /\ b ; for,

(by (4))

and a=a/\b ~ avb=(a/\b)vb=bv(a/\b)=b (by (4»

Now, we shall first prove that S; is a partial order on L. For any a E L, a v a = a (by (3» and
hence a S; a . Therefore S; is reflexive.

as; band b :;;c => a vb = band bv c=c

~av c=av(b vc)=(av b)v c=b vc=c

and hence ~ is transitive. Also.

 Centre for Distance Education 8 Acharya Nagarjuna University
, :.

a v ( a vb) = ( a va) v b = a v b an~ hence a ~ a v b .

Since a v b = b va, it follows that b ~ a v b also. Therefore a v b is an upper bound of

{a,b}. If cis any other upper bound of {a,b} , then a ~ c and b ~ c and hence av c=c=bv c and
hence

(avb)vc = av(bvc)=avc=c
so thatav b s c . Thus

a v b = lub { a, b}

Sirntlarly, we can prove that

a A b = glb { a, b}

The above two theorems illustrate that a lattice can be defined in two different equivalent
forms; one as a partially ordered set in which any two elements have lub and glb and the other as
an alqebric system in which there are two binary operations v and 1\ which are commutative,
associative and idempotent and satisfy the absorption laws. In view of the associativity, we write

av bv c hr ( av b) v c and a A b AC for (a 1\ b) A c. In general, for any elements al,a2, ....,an in

a lattice, v..'2 write

.:,,-.'
al va2 v ..... van or : aj for lub {aJ,a2, ...,an}.
1 =1
ItSI

n
ar·j al A a2 A ....Aan or ~ai for glb{al>~2,.....,an}
1=1

In otherwords, a partially ordered set is a lattice if and ony if every nonempty finite subset
has lub ard glb.

23.3.7 D::finition: Ap.o. set (p,~) is said to be a complete lattice if every subset of P has lub and
glbinP.

23.3.8 C:xample : (JR, ~) is a lattice but not a complete lattice, because the whole set IR has
neitbe: a lower bound nor an upper bound.

23.~ .; Definition: Let (p, ~) be a p.o. set and a E P . Ther. a is called the lea~t element or zero
/
 ( Algebra ~

element if a ::;x for a\l x E P (that is, a is a lower bound of P) and a is called the greatest element
or unit element if x:::; a for all x E P (that is, a is an upper bound ofP), The least element and
greatest element, if they exist, are usually denoted by 0 and 1 respectively.

23.3.10 Examples:
(1) Every complete lattice has both the least and greatest elements.

(2) The lattice (JR.,::;) has neither least nor greatest element, where ::; is the usual order.

(3) The set P of positive integers under the division order (that is, a-: b c> a devides b) has
least element, namely the intger 1, and has no greatest element.

(4) Let P be the set of all non-positive real numbers together with the usual order. Then P has
greatest element, namely the real number 0 and has no least element.

23.3.11 Definition: Let (L,:::;) be a lattice. A non-empty subsetMof L is said to be a sublattice if,

for any a, bE M , the lub and glb of {a, b} in L and are in M.

In otherwords, if (L, A, v) is a lattice, a nonempty subset of M of L is called a sublattice of

L if a v b and a A b belong to M for any a, b e M Note that a subset of a lattice L may not be a
sublattice of L, but still it may be a lattice on its own. For, consider the following.

23.3.12 Example: Let G be an abelian group P (G) be the set all subsets of G. Then (p (G), ~) is .
a lattice, where ~ is the inclusion order. Let c:?(G) be the set of all sub groups of G. Then df\G) is
,
a subset of P(G) and c:?(G) is not a sub-lattice of P ( G); for, if A, B E &'(G) , then lub {A, B} in P(G)
is AU B , which is not in 8(G) in general. However, 8(G) is a lattice on its own in which, for any
A,BE c:?(G)

lub{A,B} =AB and glb{A,B}~AnB.

Note that the partial orders on both P(G) and c:?(G) are the same, namely the inclusion
order.
23.3.13 Example:

Let (L,::;) be a lattice and a, bEL. Then the sets

[a) = {xEL/a:::;x} "

(b] ={xELlxsb}

are sublattices of L. If a ::;b , then the set

 Centre for Distance Education 10 Acharya Nagarjuna University

[a, b] = {X E L / a ~ x ~ b}

is also a sublattice of L and is called a closed interval in L.

23.3.14 Self Assessment Question:

If P is the lattice of all positive integers under the division ordering, what is the interval

[5,100] ?

23.4 HOMOMORPHISMS OF LATTICES

As usual, homomorphism from one algebraic structure into another of the same kind is a
function preserving the operations. So is the case with lattices. However, there is an ordering
relation also on a lattice and hence there is a need to relate the ordering with homomorphisms.

23.4.1 Definition: Let (p,::;;) and (p',::;;) be p.o. sets. A mapping j:PHP' is called an order

homomorphism if, for any a, b E P

23.4.2 Definition: Let Land L' be lattices. A mapping f :L~L' is called a lattice homomorphism
if, for any a, b e P ,

j(avb)=j(a)v j(b) and j(aAb)=j(a)Aj(b)

If j: L ~ L' is a lattice homomorphism, then it is an order homomorphism for, if a ~ b in L,

then a v b = b and hence

j(a)v j(b)= j(av b)= j(b)

So that j (a) ~ f (b) in L' . However, an order homomorphism from a lattice L into a lattice
L' need not be a lattice homomorphism, For, consider the following example.
23.4.3 Example: Let L be the lattice of all positive integers under the usual order (hence, for any
a.b e L, avb=Max{a,b} and aAb=Min{a,b}). Let L' be the same set of all positive integers

under the division order (here, a z: b <;:::> a divides b , a vb =lc.m. {a, b} and a A b = gc.d. {a, b} ).

Define j:L' ~L' by f (a )=a for all a EL'. Then

a sb in L'=>a divides b=>a::;;b in L=>f(a)~j(b) and hence j is an order

homomorphism. But j is not a lattice homomorphism, since
 ) (Algebra ~

/(2/\3)=/(1)=1 (since g.c.d. {2,3}= 1)

(since Min{2,3}=2)

In the above example, we have seen that a function from a lattice L into another lattice L'
may be an order homomorphism without being a lattice homomorphism. However, in the case of
bijective functions, we have a much satisfactory result in the following.

23.4.4 Theorem: Let Land L' be lattices and /:L~L' be a bijective function. Then / is a

lattice homomorphism if and only if both / both /-1 are order homomorphisms, that is, for. any

a.b e L,

a s b ~/(a)~/(b)

Proof: First note that, since /:L~L' is a bijection, the inverse /-1 :L' ~ L exists and it is such

that /(/-9(a'))=a' for any a'EL' and /-I(j(a))=a for any aEL. Now, suppose / is a

lattice homomorphism. Let a, bEL. Then

a S b~av b=b~/(av b)= /(b)

~ j(a)v /(b)= /(b)~ /(a)~/(b)

On the other hand, for any a', b' E L' , choose a, bEL such that f (a) = a' and f (b)::=b'
and, now

a's b' ~ j (a) ~ / (b) ~ j (a) v / (b) = / (b)

~ /(avb)=/(b)
~ avb=b

~a S b, ~ j-l (a') sj-1 (b')

Therefore both j and j' are order-homomorphisms.

Let a.b« L. Then a s av band b <a v b and hence

j(a)sj(av b) and /(b )s/(av b)

 Centre for Distance Education 12 Acharya Nagarjuna University

which imply that f (a vb) is an upper bound of {f (a), f (b)} : Let x be any other upp~~

bound of {f (a), f (b)} in L' .

Choose eEL such that f(c)=x. Then, since f(a)~x=f(e) and f(b)~x=(f(a), we

get that a s;c and b~e;thatand f(b)~x=f(c),wegetthat a s:c and b s c; that is, e isan

upper bound of {a, b} in L and hence a v b < e which implies that f (a vb) ~ f ( c ) = x . Therefore

f (a vb) is the least upper bound of {f (a), f (b)} in L'; tilis is to say that

f(avb)=f(a)v f(b)

Similarly, we can prove that f (a 1\ b) = f (a) 1\ f (b) . Thus f is a lattice homomorphism.

23.4.5 Definition: Let Land L' be lattices. A bijective lattice homomorphism of L intoL'is
called a lattice isomorphism.
Therefore 23.4.4 can be rephrased as follows: A bijective function of a lattice L into another

lattice L' is a lattice iscmorpr =rn if and only if both f and f-I are order homomorphisms.

23.4.6 Self Assessment Question .:

Let L ar:1~L' be lattices and f: L ~ L' be a bijection. Prove that the following are equivalent
<

to each other.

(1) "I is a lattice isomorphism.

(2) f(avb)=f(a)vf(b) forall.a,bEL.

(3) For any a, bEL, a ~ b <;:::> f (a) ~ f (b)

,
(4) i(al\b)=f(a)l\f(b) for all a.b e L
Finally, beu.. .~dosing this lesson, let us recall that while defining a lattice as a set together
with two binary operations v and A we have made precisely the same assumptions on the
operations v ,:,nd /\ . Hence, we have thefoll6wing veryirnportant principle.
23.4.7 The Principle of Duality: If Sis a statement which can be deduced from our axioms (1)
liitht'ougtf(4) of 23.3.5, then the dual. statement S' obtained by interchanging v and 1\ in Scan also
be dedueed.

23.4.8 Self Assessment Ouestion :. Let L be a lattice and a, b, eEL. Prove systematically that
 ~gebra :::s

23.4.9 Self Assessment Question:

Write the dual of the statement given in 23.4.8 and prove it.
23.4.10 Self Assessment Question :
What is the dual of the statement

Is It or its dual correct in any lattice?

23.5 SUMMARY :
In this lesson, you have learnt the concepts of a partially ordered set and a lattice. We have
established two different equivalent forms of a lattice. Also, we have defined the notions of a lattice
homomorphism and an order homomorphism and proved that every lattice homomorphism is an
order homomorphism, but not vice-versa. Further, we haveproved that a bijection I between two

lattices is a lattice isomorphism if and only if both I and 1-1 are order homomorphisms.

23.6 TECHNICAL TERMS

Partial order Partially ordered set Annsymmetric relation
Inclusion order Division order Chain
Totally ordered set Cover Minimal element
Maximal element Hassae diagram. Upper bound
Least Upper bound Lower bound Greatest lower bound
Supremum Infimum Join
Meet Lattice Commutative Law
Associative Law Idempotent Law Absorption Law
Complete lattice Least element . .~\ Greatest element
Sublattice Latnce homomorphism Lattice isomorphism
Order homomorphism

. .
23.7 MODEL EXAMINATION QUESTIONS
23.7.1 State the two different ways of defining a lattice and prove that theSftare equivalent'--

23.7.2 Define the concepts of a lattice homomorphism and an order hornomorphtsm and-prove
that a bijective function f of a lattice 'L into a Lattice L' is a lattice hcmomerphism if and

only if both I and I-I are order homomorphisms.

 Distance Educ~a;'tio:n;---=::E~~~~'- 14 F~~= Acharya Nagarjuna Univ~~
-------------~
Applying the above argument to HI and GI , we get H(2) c: G(2)and, so on, H(2)c G(2),
for every positive integer r. So
H(k) c G(k) = {e} ,

or, H(k)= {e}. So H is solvable.

Self Assessment Question 22.2.15: If G is solvable and if H is a homomorphic image of
G, show thatH'is solvable.
Solution: Let ~ : G -7 H be an onto homomorphism.
for any a, bEG,
~ (ab a-I b-1 ) = ~(a) ~(b) ~(a-I) ~(b-I)
This shows that the image of a commutator in G, undero, is again a commutator in Hand
vice versa.
So HI = ~(GI) . By induction we have
H(n) = ~( G(n) ),

for every positive integer n. So

H(k) = ~ ( G(k) ) =.<1>( { e },) = { e } ,
and- H is solvable.

Self Assessment Question 22.5.4 : Show that the polynomial f(x) = x7 - lOx5 + 15x + 5
isnot solvable by radicals over Q.
Solution: . Here degree f(x) = 7 a prime. By Eisenstein's criterion (with prime 5), f(x) is
irreducible over.Q.
The signs of f(x) and f(-x) are as follows:
f(x): + + _ + (2 changes)
f(-x):.- + - d- (3 changes)
So f(x) can have atmost five-real roots ( two positive and three negative).
Also by intermediate value theorem there are five reals, one in each of the intervals (-
4,-3), (-2,-1), (-1,0), (1,2) and (3,4) (as f(x) changes sign in these intervals).
.So f(x) has exactly five real roots, and hence it has exactly two non real roots.
Hence the Galois group of f(x) isS, .
. Since 7 > 5, ,S7 is not also solvable group. So the Galois group of f(x) is not solvable
,-,~n~l(~)isnot solvable by radicals over Q.

22.10 E~ROIS.ES:
.. .. ' "'

1. For a =
(12 23. 3J1
 ~ Lesson 2~) ( Algebra ::E

.i->: b {b}

23.2.12: We have X ={a,b,e,d,e,f,g}. a and b are minimal elements and e,f and g are
maximal elements. We have

a < d < e, a < e < t. a < e < g ,

b < e < f and b < e < g and hence

s;={(a,a ),(b,b ),( e, e), (d, d), (e,e), (f,f), (s.s)

(a, d), (a,e), (d, e),( a, e), (a, f), (e, f),( a, g), ( e, g)

(b,e), (b,f), (e,f), (b, g)}

23.3.2: Iub{a,b}=e

glb{e,d}=a

glb {e, g} doesn't exist, since a and b are the only lower1'bounds of {e, g} and a and b
- are incomparable.

23.3.14: [5, 100]={a EO pISs a s: 100}

= {a EO P / 5 divides a and adivides 1OO}

={5, 10, 20, 25, 50; 100}

23.4.6: (1)~(2) is clear

(2) =>( 3) use the first part of the proof of 23.4.4

(3) ~ ( 4) use the second part of the proof of 23.4.4

'25)
 Centre for Distance Education 16 Acharya NagarjUna University

(4)=T>(3)=>( 2)=>(1) Arque in a reverse wa~

y if and only if x = x A Y .
23.4.8 : Recall that x ::;; ./

Put x = ( a A b) v ( a A c) and y = a A( b v c) .
Since a A b s. a and a A c s a, it follows that x::;;a.

Also, a r.b cb s bv c and ar.c s c cb v c and

therefora. x =( a A b) v (a A c) ::;;
bvc

From x::;; a and x ::;;b v c ,it follows that

x ::;;a A ( b v c) = y and hence x A y = X .

23.4.9 Dual of the statement given in 23.4.8 is

[(avb)A(avc)]v[ aV(bAc)]
To prove this, use the dual argument of that given to prove 23.4.8

23.4.10 The dual is [a A(b v c)Jv (a Ab)v (a A c) = (a A b)v (a A c)

Both of these are not correct, for, consider the lattice whose diagram is given below, in
which
1

L
~d avb=bvc=cva=l

~('a vb) A( a v ~1~!

1 ,b

av(bAc)=avO=a
I
o
23.9 EXERCISES:
23.9.1 Let X be any nonempty set and, for any x, Y E X define x::;;y if and only if x = y .
I

Prove that ::;;is a partial order on X

,23:9.2 LetP'be the set of all real-valued functions defined on a given nonempty setX For any
If gEP, qef~rle j::;;g~ j(x)::;;g(x) for all XEX I

where::;; on the right side is the usual order on IR . Prove that (p,::;;) is a lattice.
a:: Lesson 2:D ( Algebra'::E

23.9.3 Prove that a p.o. set (p,::;;) is a lattice if and only if every nonempty finite subset of P has
lub and glb in P.
23.9.4 Prove that a p.o. set is a complete lattice if and only if every subset has lub

23.9.5 Given an example of a partially ordered set in which lub {a, b} exists for all elements a
and b and which is 'not a lattice,

23.9.6 Let L be the set of all continuous real valued maps defined on X = [0,1], the closed unit
interval in IR. Prove that L is a sublattice of P, which is given in 23.9.2. Are P and L complete
lattices?

23.9.7 Prove that the set cll'(G) of all subgroups of a group G is a complete lattice under the
inclusion order and that the set of all normal subgroups of G is a sublattice of Q9'{G).

23.9.8 Prove that the set of all ideals of a ring forms a complete lattice under the inclusion order.

23.~.9 Let P and P' be p.o. sets and f:P~ P' such that, for any X,YE P,

x::;;y<=> f (x)::;; f (y)

Then prove that f is an injection.

Z3.10 REFERENCES .
1. Lattice Theory, Garret Birkhoff, revised edition (1949),

AmerMath Soc. Colloq. publication (1949)

2. General Lattice Theory, George Gratzer, Academic Press, New York (1.978)

3. Lectures in Abstract Algebra, N. Jacobson, Vol. 1,') "

D. Van Nostrand Company, Inc, London, (1964)
4. Algebra, M. Artin, Prentice - Hall of India, New Delhi (1994)

.',

Lesson Writer,' .
Prof: us« Swamy
Andhra University;
Visakhapatnam:.
Lesson 24 MODULAR LATTICES

OBJECTIVES
After reading this lesson, you should be able to .
• define the notions of distributive lattices and modular lattices and obtain certain
necessary and sufficient conditions for a lattice to be modular.
• prove that the normal subgroups of a group form a modular lattice under inclusion
order,
• state and prove the Schreier's theorem and
• define the concepts of ascending chain condition, descending chaincondition and
composition series and obtain the fundamental dimensionality relation.

STRUCTURE
24.1 Ir.troduction

24.2 Distributive lattices

24.3 Modular lattices
24.4 Projective intervals '

24.5 The Schreier's theorem

24.6 The chain conditions

24.7 Summary
24.8 Technical Terms

24.9 Model Examination Questions

24.10 Answers to Self Assessment Questions

24.11 Exercises
24.12 Reference Books

24.1 INTRODUCTION
The lattice of normal subgroups of a group and thelatice of ideals of a ring both satisfy a
weak form of distributitivity of the operation v over the operation /\, namely for any elements
a, b, c, a v (b /\ c) = (a vb) /\ (a /\ c) if a s; c. Abstract lattices satisfying this property are called
modular lattices. It is known that the subalgebras of several types of algebras torm' a modular
~ Lesson 2£J ('~3 )' (. Algebra ~

forall a.b.c e L or,equivalently, a/\(bvc)=(a/\b)v(a/\c) for all a.b.c e L

24.2.3 Examples:

(1) Any chain is a distributive lattice, for if b :s:;c, we have b /\ C =b and a v b-: a v c so
that

av (b /\ C )::;::avb=( av b)/\( av c)

(2) The set of positive integers together with the division order: is a distributive lattice, \
since, for any positive integers a, b, c, we have

av (b /\ c)= l.c.m. {a, g.c.d. {b, c}}

=g.c.d.{l.c.m.{a,b},l.c.m{a,c}}

=(avb)/\(avc)

(3) The lattice, whose Hassae diagram is below, is not distributive

since av(b'/\c)=avO=a

and ( a vb) /\ (a v c) = 1/\ 1 = 1

a c

o
24.2.4 Self Assessment Question :
Prove that any lattice with atmost four elements is distributive.

24.2.5 Self Assessment Question :

"

If a, b, c are elements of a distributive lattice L such that

a vb =a v c and a r.b = a /\ c ,
then prove that b = c .
 ~entre for Distance Education 2 Acharya Nagarjuna University

lattice and hence an abstract study of modular lattices is essential in order to set certain
comprehensive results like the Schreier's theorem. In this lesson, we discuss certain important
aspects of moduiar lattices and, in particular, we prove Schreier's theorem which states that any
two finite chains connecting two elements in a modular lattice have equivalent refinements and
deduce the Jordan-Holder theorem which facilitates the introduction of a dimension function on a
modular lattice of finite length:

24.2 DISTRIBUTIVE LATTICES .

As in the case of a ring, we have two binary operations in a lattice. One of the operations
can be regarded as the analogue of addition in a ring and the other can be taken as the analogue
of multiplication. It is therefore natural to think of the distributivity of one operation over the other in
lattices. Unlike the case of a ring where multiplication distributes over addition and addition doesn't
distribute over multiplication, in a lattice the distributivity of one operation over the other implies the
distributivity of the other operation also. This is proved in the following.
, -~
24.2.1 Theorem: Let L=(L, V,I\) be a lattice. Then the following are equivalent to each other.

(1) av(bl\c)=(avb)l\(avc) for all a.b.c eL

(2) al\(bvc)=(al\b)v(al\c) for all a.b.c eL

Proof: .(1) =:>( 2) : Suppose a 1\ (bv c)=( a I\b) v (a 1\ c) for ali'"'a, b, CE L .

Let a, b, CE L. Then consider

(a 1\ b)v (a 1\ c)=(( a I\b)v a) 1\ ((a r.b )vc) (by (1»

\.....
~,.
C-, •

'.:lit. :(by the absorption law)

(again, by (1»

(by the associativity)

(by the absorption law)

Thus (2) holds good. To prove (2)=:>(1), simply dualise the above proof by interchanging
v and 1\.

24.2.2 Definition: A lattice (L, v, 1\) is said to be a distributive lattice if

 - Centre for Distance Education 4 Acharya Nagarjuna University

24.2.6 Self Assessment Question :

Prove that, in 24.2.3(2),

l.c.m. {a, g.c.d {b,c}} =g.c.d {/.c.m {a,b}, l.c.m.{a,c}}

24.3 MODULAR LATTICES:

The lattice of ideals of a ring or the lattice of normal subgroups of a group do not, in general,
form a distributive lattice. However, these lattices satisfy a weak form of distributivity, namely
modularity. In this, we discuss modular lattices and prove certain important properties of these.

24.3.1 Definition: A lattice L is said to be a modular lattice if, for any a, b, c E L ,

a S; c zo a v (b/\c)=( av b) /\ C
24~3.2 Theorem: Every distributive lattice is modular and the converse is not true.

Proof: Let L be a distributive lattice and a, b, c E L . Suppose a s; c . Then

a v ( b /\ c) = ( a vb) /\ ( a v c) (by the distributivity)

=(avb)/\c (since as; c)

Thus L is a modular lattice. The converse is not ture, because the lattice, given by the
following Hassae diagram is modular but not distributive.

a c

o . . ~
-The following theorem is a motivation for undertaking the study of abstract modular tattioes" ,

24.3.3 Theorem: Let G be a group. Then the set of all normal subgroups of G forms a modular
lattice under the inclusion order.

Proof: Let n( G) be the set of all normal subgroups of G . Then (n (G), c) is a lattice, in which

the glb and lub of any A, B are given by

 ( Algebra ~

and AvB=AB (={ab/aEA and bEB})

To prove that n(G) is modular, let A,B,CEn(G). Suppose A ~ C.

We have A c AB and BnC~ AB and hence AB is an upper bound of {A,BnC} so

that Av(BnC)c AB. Also A c C and BnC~ C and hence A v(BnC)~ C . Therefore

on the other hand, suppose x E (AB) /\ C . Then x E C and x = abE C for some a E A
and bE B·

Also, a-1x=bE BnC (since a E C and x

Ac;:;;; E C) and

Therefore (AB)nCc Av(BnC). Thus

A v (B /\C) = A(BnC)=(AB)nC = (A v B)/\C ,

so that n( G) is a modular lattice.

24.3.4 Corollary: The lattice of ideals (left, right, two sided) of any ring and the lattice of submodules
of module are modular lattices.

Proof: You can easily observe that any sublatttice of a modular lattice is modular. The lattice of

ideals of any ring (R, +, .) is a sub lattice of the lattice of normal subgroups of the group (R, +) and
hence it is modular. So is the case of submodules of a module.
Unlike the case of normal subgroups, the subgroups of a group may not form a modular
lattice. To have an example to illustrate this point, necessarily we have to seek a non-abelian group,
since in an abelian group every subgroup is normal. An example of such a group is given in the
following.

24.3.5 Self Assessment Question :

Consider the aJternating group A4 of degree 4. Prove that the lattice of subgroups of A4 is
not modular.
In the next theorem, -we shall obtain an alternative useful definition of a modular lattice.
 Centre for Distance Education 6 Acharya Nagarjuna UniveJsity

24.3.6 Theorem:

A lattice L is modular if and only if, for any a, b, eEL,

avb = avc, a r.b = a r.c and b s. c zz-b=c .

Proof, : Let L be a lattice. Suppose that L is modular and a, b, eEL such that a vb = av c,
a /\ b = a /\ c and b :,;;c . Then

b=bv(a/\b) = bv(a/\c)

=(bva)/\c (since L is modular and b sc;

=(avb)/\c

=(avc)/\c=c

Conversely, suppose that the given condition is satisfied. We have to prove that L is modular.
Let a, b, c E L such that b < C . Then, put

x = b v (a /\ c) and y = (b va) /\ c

Since h :,;;c and a /\ c ::;:c , we get that x = b v (a /\ c) ::;:c .

Also, from x s (b va) and x::; c we get that x -::;,
(b va) /\ c = y.
We shall verify that a v x = a v y and a /\ x = a /\ y . Consider
.U

Since a/\c-::;,bv(a/\c)=x and ar.c s;a , itfollowsthat a r.c s a »;x, Thus we have

and hence a /\ x==a /\ y. Also

$avb (since a s av b and (bva)/\c-::;,avb)

"=av(a/\c)vb =av y

and hence a v x = a v y, a /\ x = a 1\ y and x::;:y .

 Then, by the given condition, it follows that X= y so that

hv(al\c)=(hva)l\c

Thus L is modular.

24.4 PROJECTIVE INTERVALS :

In lesson 2 (see 2.5.5), we have proved that, for any normal subgroups A and B of a group

G, the quotients fA rtB- and A% are isomorphic and recall, from 24.3.3, that the normal
subgroups of a group form a modular lattice. The following result is angalogous to the above result
in group theory.
24.4.1 Theorem:

Let L be a modular lattice and a, bEL. Then the intervals [a 1\ b, a] and [b,a v b] are
isomorphic.

Proof: Define f:[a 1\ b, a ]-+[h, av h] by

f(x)=xvb forall x~[a/\b,a]

Note thatc x e s x s c implies h=(al\h)vh~xvh~avb and hence f is well defined.

For any YE[b,avb], we have bcy sav b and hence al\b~al\y~a(avb)=a so that

aI\YE[al\h,a] and, also

! (a 1\ y) = (a 1\ y}\Y'b

= hv(aI\Y)
(since L is modular and b ~ y)

=Y

and hence f is a surjective map. Also, for any x, y in [a 1\ h, a]

and f(x) ~ fey) ~xv b s;yv b

..
.:
 ·Centre for Distance Education 8 Acharya Nagarjuna University

=>xv (b I\a )::;y V (b I\a) (since L is modular)

=>x::;y (since a 1\ b ::;x, y)

From this, it follows that f is a bijection and

x::; y <=> f (x)::; f (y)

for any x,Y E [a 1\ b, a]. Thus by 24.4.4, f is a lattice isomorphism.

24.4.2 Definition: Let [x, y] and [z, t] be two intervals in a lattice L . These two intervals are said

to be transposes or similar if there exist elements a, b in L such that one of the intervals can be

expressed as [a 1\b, a] and the other can be expressed as [b, a vb] ,

24.4.3 Definition: The intervals. [ x, y] and [z, f] in a lattice L are said to be projective if there
exists a finite sequence.

of intervals in L begining with [x, y] and ending with [z, t] such that any pair of consecutive
intervals are transposes.
24.4.4 Theorem: In a modular lattice, any two projective intervals are isomorphic.

Proof: Let L be a modular lattice. By theorem 24.4.1, any two transpose intervals in L are

isomorphic. Let [x, y] and [z, f] be projective intervals in L. ·;Then these exists a sequence of
intervals.

; each 0::; i s n Now, we have o

[x,y]=[xo,Yo] == [xbyd=···········=[xn'Yn]=[z' f].

Thus [x,y]= [z,t].

24.4.5 Corollary: Let a and b be any elements in a modular lattice. Then

 ( Algebra ~

a 1\ b-:a => b-cav b

24.4.7 Theorem: Every modular lattice is semi-modular. The converse is not true.

Proof: Let L be a modular lattice and a, b, E L . Then, by 24.4.5, a covers a 1\ b if and only if
a v b covers b . Therefore L is a semi-modular lattice. The converse is not ture, for consider the
following.

24.4.8 Example: Let L = {O,a, b, C, d, e, I} be the lattice whose Hassae diagram is given below.

e d

a c

This is a semi modular lattice which is not modular. Note that e -< e v d = 1 even though.

e 1\ d (= 0) is covered by d. Therefore, this is a semimolecular lattice in which y--< x v y need not

imply that x 1\ y--< X . This says that the conditions.

XI\Y--<X=>y-<xv y

and y -< x v y => X 1\ Y -< x

are independent from each other. Also, note that the above lattice is not modular, since

[e 1\ d, d] i [e, e v d]

24.5 THE SCHREIER'S THEOREM .

In this section, we shall generalise two important theorems, the Schreier's theorem and the
Jordan-Holder theorem, on the normal subgroups of a group to the case of a general modular
lattice.Let us begin with the following.

24.5.1 Definition: Let L be a modular lattice and a, bEL. Consider two finite ascending chains

(1)

and (2)

connecting a and b. We say that the chain (1) is a refinement of the chain (2) if every
 Centre for Distance Education 10 Acharya Nagarjuna University

element in (2) is included in (1); that is, each b, is equal to some aj. The chains (1) and (2) are

said to be equivalent if there is a one to one correspondance between the intervals [ ai, ai + 1] of
the chain (1) and the intervals [b j' bj + 1] of the chain (2) such that the .correspondinq intervals
are projective.

24.5.2 Lemma (Zassenhaus' Lemma) : Let L be a modular lattice and aI, a2, q,b2 E L such that
aT· I and a2 :::;;b2 . Then the following intervals.

[ ( a2 /\ q) v al , (q /\ b2 ) v al ] -------- (1)
/
[(al /\b2)v(a2 /\q), ~ ;b2], -------- (2)

are projective to each other.

Proof: Note that the interval (2) is symmetric in 1 and 2 (that is it is not changed if we interchange
the subscripts 1 and 2). and interval (3) is obtained from the interval (1) by interchanging 1 and 2.
Therefore, it suffiices to prove that the intervals (1) and (2) are projective.

I'
and

Therefore, the interval (1) is [a, a vb] and the interval (2) is [a /\ b, b] and hence the intervals
(1) and (2) are transposes. Thus (1), (2) and (3) are projective to each other. >
24.5.3 Theorem: (The Schreier's theorem)

Let L be a modular lattice and a, bEL such that a s; b . Then any two finite ascending
chains connecting a and b have equivalent refinements.
 C Algebra :E
Proof: Let a = a1 <
- a2 <
- <
- an = b ---------- (1)
and a = Vi
h. <
- b2 <
- - bm = b ----..:----- (2)
<

Put aij = ( ai + 1 /\ bj) v ai for 1:::;i < nand 1:::;j :::;

m
I '

and bji=(ai/\bj+I)vbj for lS;iS;n and l:::;j<m.

-'a >
- 31 _ >a
- n-1,m - b

and a-h. <

<h. -
-vll-vi2 <h. -b'21->b22 ->
-vin- ->b-2n-

=b31 ~ ~bmn =b

are refinements of (1) and (2) respectively. Also, by the Zassenhaus' lemma, the intervals

[aij,aij+1} and [bji,bji+11 are projective. Therefore, the correspondence

[ aij' aij + 1] H [ bji , bji + 1] gives us the equivalence of the chains (1) and (2).

We shall use the above Schreier's refinement theorem in proving. the Jordan-Holder theorem.
a
First, let us introduce the notion of composition chain.
,
24.5.4 Definition : Let a and b be elements of a lattice L such that a < b . A finite ascenomq
chain.

connecting a and b is called a composition chain if each aj is covered by aj +1.

24.5.5 Self Assessment Question : Prove that a finite ascending chain connecting a and b is a
composition chain if'and only if it has no proper refinements.

24.5.6 Theorem (Jordan - Holder Theorem) : Let L be a modular lattice and a, b e L such that
a < b. If a=al <vi < <an =b

are two composition chains connecting' a and b, then n = m and there is a one-to-one

correspondence between the intervals [ai' ai + 1] and [b j' bj + 1] such that the correspondinq
intervals are projective.
 .' ~

Centre for Distance Education 12 Acharya Nagarjuna Univer ••"v

Proof: By the Schreier's theorem 24.5.3, the given two composition chains should have squiv.sent
reftnements. But a composition chain has no proper refinements and therefore any two composition
chains connecting a and b must be·equivalent (see 24.5.1)
24.6 THE CHAIN CONDITIONS :
We shall introduce two types of chain conditions, one is the ascending chain condition and
the other is the descending one, and prove that any bounded lattice satisfying these conditions is of
finite length. First, we have the following.

24.6.1 Definition: Let L be a modular lattice with least element 0 and greatest element 1. L is
said to be of finite length if there is a composition chain connecting 0 and 1 and, the length of this
chain (or the number of intervals in this chain) is called the length of L or dimension of L .

By the theorem 24.5.6, the length 'of L is independent of the composition chain conectinq 0
and 1.

24.6.2 Definition: A lattice L is said to satisfy the Ascending Chain Condition(ACC) if there is no
infinite properly ascending chain al < a2 < L. Dually, L is said to satisfy the Descending
o
Chain Condition (DCC) ifthere is no infinite properly descending chain al > a2 > in L .

In other words, L satisfies ACC if and only if, for any asceridjnq sequence

PI:::; a2 ::;;a3 :::; in L, there exists n E Z+ such that an..=an + 1=ai1+2 = Dually, L
satisfies DCC if and only if descending sequence terminates at a finite stage .:
. . .'. . .' ". • I

24.6.3 Self Assessment Question: Prove that a lattice L satisfies ACC (DCC) if and ontYii,~vE.!ry
nonempty subset of L has a maximal (respectively minimal) element.

24.6.4 Theorem: Let L be a modular lattice with 0 and 1. Then L 'is of finite length if and only if L
satisfies both ACC and DCC.

Proof: Suppose L satisfies both ACC and DCC. Let al be ~ tninimal nonzero element in L , if

L :t:{0,1}. Next, if L-{O,al,l}:t:¢, then let a2 be minimal in L-{O,al,l}. Then we get

, 0 -< al -< a2 -< a3 -< if the above process conti~ues infinitely. But, since L satisfies
ACC, this process should terminate at a firiite stage. Then we get a finite sequence,

which is a composition chain connecting 0. and 1 and hence L is of finite length.

/'

. Conversely, suppose that L is of finite length and let

0-< al -<a2 -< ..... -< an -< 1

be a composition chain. By the Schreier's theorem, any two chains connectivity 0 arid 1
 C Algebra ~

should have equivalent refinements. But any composition chain has no proper refinement. Therefore
L has no infinite chains. Thus L satisfies both ACC andDCC.

24.6.5 Definitioh : Let L be a modular lattice with 0 and 1 and L be of finite length. Then, for any
a E L , the interval [0, a] is also a modular lattice of finite length. Define

d(a)= the length of [O,a]

d ( a) is called the rank or dimensio'nality of a.

24.6.6 Theorem: Let L. be a modular lattice of finite length with 0 and 1. Then, for any a, h E L ,

d(av h)+d(a 1\ h) = d(a)+d(h)

This formula is called the fundamental dimensionality relation.

Proof: If X,YE L such that xs Y , then we can easily observe that

d(y)=d(x)+ length of [x,y]

Now, let a, hE L. Then, by theorem 24.4.1, the intervals [a 1\ b, a] and [h, a vb] are
isomorphic and hence they have same lengths. Now,

d(avh)=d(h)+ length of [h,avh]

and d(a)=d(al\b)+ length of [al\b, a]

and therefore,

d (a v h) - d (h )':t=(i ( a) - d (a 1\ h)
~.

Thus d( av h) +d( a 1\ h)=d( a)+d(b)

24.7 SUMMARY :
In this lesson, you have learnt the concepts of a distributive lattice and a modular lattice and
obtained certain necessary and sufficient conditions for a lattice to be modular. We have proved
that every distributive lattice' is modular and an example is given to prove that the converse is not
true. Further, we have proved that the normal subgroups of a group form a modular lattice under
the inclusion order. Also, we have proved the Schreier's theorem and Jordan-Holder theorem for
modular lattices.· Finally, we have defined the notions of the chain conditions ACC and DCC, the
composition chain and ofthe length of a lattice with 0 and 1 and derived the fundamental dimensionality
relation.
 Centre for Distance Education 14 Acharya Naqarjuna University

24.8 TECHNICAL TERMS

Distributive lattice

Modular lattice

Transpose intervals

Projective intervals

Normal subgroups

Ideals

Alternating group A4

Semimodular lattice
Schreier's theorem

Refinement
Equivalent chains

Zassenhaus' lemma

Composition chain

Jordan-Holder theorem

Ascending chain condition

Descending chain condition

Maximal element

Minimal element

LengU I of a lattice
Rank
Dimensionality
Fundamental dimensionality relation

24.9 MODEL EXAMINATION QUESTIONS :

24.9.1 : Define the notions of distributive lattice and modular lattice. Prove that every distributive
lattice is modular. Is the converse true? Justify your answer.

24.9.2: Prove that a lattice L is modular if and only if, for any a,b,c E L,

av b=av c, a /\b=a /\ C and b s; c=>b=c

24.9.3 : In a modular lattice, prove that any two projective intervals are isomrphic.

26
 ( Algebra ~.

24.9.4: State and prove Schreier's theorem for modular lattices.

24.1 o ANSWER's TO SELF ASSESSMENT QUESTIONS :

24.2.4 : It is trivial when ILl 53 , since, in this case L is a chain and every chain is t! istrlbutlve.

Suppose L has exactly 4 element's. Then there are two elements a *- b .other than 0 and 1.
If a 5 b or b 5 a , then again L is a chain and hence distributive. Otherwise, we get a v b = 1 and
a Ab=O. Its Hassae diagram will be as shown below. In this case, we can easily see that L is
distributive.

a C

\ o
24.2.5 : Suppose a v b = a v c and a A b = a A C. Then

=(a Ac)v(b AC )=(a:-- b) Ac=(avc) AC=C

'1 ~ rn n if s'<.' .

24.2.6: Let a = PI P2 Pn = i=Irc P' I

Sl s2 sn n s,
= rc Pi
b= PI P2 ·.. ··-::~P~
i=I .

and

where PI> P2,,··.·,Pn are distinct primes and fj, si, ti are non-negitive integers. Then

_ n max{r;,min{si,ti}}
Iem .{a, g.c.d{b ,c }} - rc Pi
i=I
 Centre for Distance Education 16 Acharya Nagarjuna University

= g.c.d {l.c.m {a, b}, l.cm. {a, c}}

24.3.5 : Construct subgroups A, B, C of A4 such that An B = An C, A vB =Av C and Bee.

*
Then, you can see that

Bv(AnC) =B and (Bv A)nC=C

Thus the lattice of subgroups of A4 is not modular.

24.5.5 : This follows from the fact that in any composition chain we cannot insert new elements in
between any-consecutive elements and that any chain with this property is a composition chain.

24.6.3 : Suppose L satisfies ACC. Let A be a nonempty subset of L . Choose al EA. If al is

maximal in AI' we are though. Suppose al is not maximal. Then there exists a2 E A such that

al <a2' Repeating this argument with i:J2, we get a3 and so on. This way we get a chain

This should terminate at a finite stage and hence there is a maximal element an in A . The
converse is trivial. Dualise this argument for DCC.

24.11 EXCERCISES
24.11.1 : Prove that a lattice L is not distributive if the lattice given in 24.2.3(3) is a sublattice of L .

24.11.2 : Prove that a lattice L .is distributive if and only if, for any a, b, C E L ,

av b = avc and al\b=al\c =::::>b=c.

24.11.3: Prove that a modular lattice L is distributive if the lattice given in 24.2.3 (3) is not isomorphic
to any sublattice of L .

24.11.4 : For any prime p and a positive integer n, prove that the set of all subgroups of a group of
n
order P form a distributive lattice.
24.11.5 : Give an example of a lattice in which

a-cav b:::::;,a
r.b-cb for any a,b ELand in which there exist elements x and y
such that
 o

Ii Co: Algebra ::e

xl\y-<y and x-« xv y

24.11.6 : Prove that a lattice L is distributive if and only if any sublattice of L ls isomorphic to
neither of the following.

24.11.7 : Let L be a lattice. A subset A of L is called an ideal (dual ideal) of L if, for any G, JJ E L ,
av bE A<=> a E A and b e A (a 1\ bE A cs a E A and b e A). An ideal (dual ideal) is called principal
if A = {x E L/x :S;a} ( {x E L/a :S;x} , respectively) for some a EA. Prove that L satisfies ACC if

and only if every'1deal of L is principal.

24.11.8 : Prove that at a lattice L satisfies ACC if and only if its dual LT satisfies DCC (where LT
is obtained by interchanging v and /\ in L ).

24.12 REFERENCES :
1.· Lattice Theory, Garret Birkhoff, revised edition (1949),

Amer Math Soc. Colloq. publication (1949)

2. General Lattice Theory, George Gratzer, Academic Press, New York (1978)
3. Lectures in Abstract Algebra, N. Jacobson, Vol. 1,
D. Van Nostrand Company, Inc, London, (1964)

4. Algebra, M. Artin, Prentice - Hall of India, New Delhi (1994)

Lesson ':Jriter
Prof: U.M. Swamy
Andhra University,
Visakhapatnarn.
Lesson 25 DECOMPOSITION THEORY FOR
LATTICES WITH ACC

OBJECTIVES
After reading this lesson, you should be able to
• define the notions of an irreducible element and an irredundant representation of an
element.
• prove that any two irredundant representations of an element in a modular lattice
have same number of irreducible elments and
• define the notion of independence and state the Kurosch-Ore theorem.

STRUCTURE
25.1 Introduction
25.2 Irredundant representations

25.3 Independence
25.4 Summary

25.5 Technical Terms

25.6 Model Examination Questions . j~:" . ".

25.7 Answers to Self Assessment Ouestions.

25.8 Exercises
,'.
25.9 Reference Books

25.1 INTRODUCTION:
In this lesson, we make an abstraction of the theory of ideals of a Noetherian ring. Let us
recall that a commutative ring is called noetherian if its lattice of ideals satisfies the ascending
chain condition (or equivalently, every ideal is finitely generated). Therefore, the lattice of ideals of a
Noetherian ring is a modular lattice satisfying the ascending chain condition. Any result we prove
for modular lattices with ACC can be applied to ideals of a Noetherian ring. For this reason, we
concentrate our attention on the modular lattices satisfying ACC.
 Centre for Distance Education 2 Acharya Nagarjuna University

25.2 IRREDUNDANT REPRESENTATIONS .

Recall that the ring of integers is a Noetherian ring and, in this, an ideal 1 is primary (or,
equivalently, generated by a power of a prime) if and only if, for any ideals A and B,
1 = An B => 1 = A or 1 = Band such an ideal is called irreducible also; in the sense that, 1 cannot
be expressed as the intersection two ideals which properly contain 1 . This is abstracted in the
following.

25.2.1 Definition : An element a in a lattice L is called (meet) irreducible if, for any elements
b,CEL,

a=b 1\ c=>a=b or a =c.

25.2.2 Self Assessment Question:

(1) Let p be the lattice of positive integers under the division order. Then, prove that
every element of p is reducible (that is, not irreducible).
(2) In a chain, prove that every element is irreducible.

(3) Prove that the greatest element in any lattice is irreducible.

In otherwords, a is called irreducible if it cannot be expressed as a glb of two elements

which are strictly greater than a . The following is a first step towards our decomposition theory.
25.2.3 Theorem: Let L be a lattice satisfying the ascending chain condition. Then any element of
L can be expressed as a glb of a finite number of irreducible elements.

Proof: Let X be the set of all elements which cann't be expressed as a g.l.b of a finite number of
irreducible elements in L . We shall prove that X is the empty set. Suppose, if possible, X is not
empty. Since L satisfies the ascending chain condition, X should have a maximal element, say

Xo . Then, since Xo EX, Xo can not be irreducible and therefore

xo=al\b for some a.b e L with xo<a and xo<b. 8ythemaximalityof xo, itfollowsthat
a, b (j:. X and therefore a and b can be expressed es g.Lb 's of finite number of irreducible elements
and hence so is xo, which is a contradiction to the fact that Xo EX.

25.2.4 Example: Let L be the set of all positive integers and s be defined by

as b<=>a is a multiple of h.

Then L is a lattice in which, for any a, bEL,

a v b = gc.d {a; b}
~ Lesson 2~ ( Algebra .E

and a A b+Lc.m. {a, h}

you can easily reck that L is a distributive (and hence modular) lattice satisfying the
ascending chain cone.Jon (Infact, L is dual to the lattice given in 24.2.3(2)). Note that this lattice L
has largest element (namely 1) and has no least element.
25.2.5 Self Assessment ,Question:

In the lattice L given in 25.2.4, prove that an element a -+ 1 is irreducible if and only if a= p"
for some prime number P and positive integer n . Verify theorem 25.2.3 in this lattice.

25.2.6 Definition: A representation a = PI /\ P2 /\ .... /\ Pn is said to be an irredundant representation

if a -+ PI /\ /\ Pi -I /\Pi + I /\ /\ Pn for any 1 ~ i ~ n .

In otherwords, a representation a= PI /\ P2 /\ ..... /\ Pn is called irredundant if no Pi can be

dropped from the representation a= PI/\ P2 /\ .... P» or a cann't be represented as the g'/'h of

fewer Pi's.

25.2.7 Theorem: Let a be an element of a modular lattice Land

where Pi'S and e,' s are irreducible elements in L. Then, for each Pi' ther exists qj
such that

a= PI /\ .... /\ Pi-I/\ qj /\ Pi+1 /\ ..... /\ Pn

(that is, each Pi can be replaced by a suitable qj).

. ..j.-."
J f .~I

Proof: If suffices to prove the theorem for PI . For eactTT::; j ::;m , let

Put x = P2 /\ .... /\ Pn .

Then PI ;\ x = as; rj ::; x for all 1::;j ::;m ; that is 'i, r2, ....,rm are all elements of the interval

[PI /\ x, x] which is isomorphic to the interval [PI, PI v x]. Since these two intervals are isomorphic,

irreducible elements in one interval correspond to those in the other. Since PI is irreducible in L ,

PI is irreducible in [PI> PI v x] also. Since

 Centre for Distance Education 4 Acharya Nagarjuna University

it follows that a is irreducible in [a, x]. Now, since a = rl/\ r2 /\ .... /\ rm and rj E [a, x], itfollows that

a=rj for some j, 15 j 5 m . Thus, there exists qj such that

a=qj/\P2/\···/\Pn.

25.2.8 Theorem: a be an element of a modular lattice L . Then the number of terms in any
Let
two irredundant representation of a as glb of irreducible elements is the same.

Proof: Let a = PI /\ P2 /\ .... /\ Pn = q1 /\ q2 /\ ..... /\ qm be two irredundant representations of a, where

Pi'S and qj'S are irreducible elements in L. Without loss of generality, we can suppose that

n 5 m . Then, by theorem 25.2.7, each Pi can be replaced by some qjj , 15 i, 5 m. Therefore, we

have

Since a= q1/\ q2 /\ ..... /\ qm is an irredundant representation, a cannot be equal to the glb

of fewer «,' and hence all of q], q2, .... , qm shedd qppear among qji' qh' ..... , qjn . This amounts

to saying that n = m .

25.3 INDEPENDENCE :
Let us recall that a group G is called the (internal) direct product of its normal subgroups

NI,N2, ..... .N; if G=NI N2, ..... N, and

for all 1:::; i :::;n ; That is, the intersection of N, with the lub of other Nj' s is equal to the zero

_ subgroup {e} . In this section, we shall generaiise this situation to a general modular lattice.

25.3.1 Definition: Let L be a modular lattice with the smallest element 0 and the largest element
1. A finite subset {a1, a2,.· ..., an} of L is said to be in9.§pendent if
 ( Algebra ~

25.3.2 Theorem: Let {aJ>a2, .... , an} be an independent set of elements in a modular lattice L
and 1::; r < s < t ~ n . Then
/

r s

i.e., (
Vai v v a,
i=l i=r+l

Proof: We shall first prove that, for any 1::; m < n ,

The statement (1) is true for m = 1, by the independency of {aI, , an}. Now, let m > 1
and assume that(1) is true for 111 -1; that is,

(a1 Y ..... v am -1) r; (am v am + 1 v ....V an ) = 0 .

Consider

(a1 v .....v am ) !\ ( am + 1 V ..... V an) ::;(a1 V ..... v am) r; (am V am + 1 V ..... v an )

= ( am V (a1 V ..... v am -1 ) ) !\ (am V .. , . V an )

and hence

== (al V ..... V am ) !\ 0 = 0
Thus (1) is true for all 1::; m < n . Now, for any 1::; r < s < t ::;n,

J
 Centre for Distance Education 6 Acharya Nagarjuna University

J]
= (r V'a, J vo = r
Va i
i=1 i=1

(here, we have used the modularity of L and the independency of {aI, a2,·····, an}) .

25.3.3 Definition: Let a be an element of a lattice L . Then a is said to be (ioin) decomposable if

a=a1 v ~ for some at> a2 E L with al < a, a2 < a and al /\ a2 =0.

25.3.4 Self Assessment Question: Prove that a is decompsable if and only if

a = al va2 v ....van for some independent set {al> "" '" an } with al < a .

25.3.5 Theorem: Let L be a lattice satisfying the descending chain condition. Then any element
of L can be expressed as the lub of a finite number of independent indecomposable elements.

Proof: Let X be the set of all elements of L which cann't be expressed as the lub of any finite
set of indepehdent indecomposable elements. We have to prove that X is empty. Suppose, if
possible, X is not empty. Since L satisfies the descending chain condition, X contains a minimal
element, say Xo . The~o must be decomposable (since Xo EX) and hence

Xo = a vb, with a /\ b = 0, a < Xo and b < Xo .

By the minimality of xo, a and b are not elements in X and hence a and b can be
expressed as lub's of finite sets of indepen?ent indecomposable elements and hence so is xo,
which is a contradiction, since xQ EX. Thus X is empty.

25.3.6 Definition: Two elements c and d of a lattice L are said to be directly projective if there
exists bEL such that

bvc=bvd and b/\c=b/\d=O

25.3.7 Self Assessment Question : If c and. d are directly projective in a modular lattice L , then
prove that the intervals [0, c] and [0, d] are projective.

25.3.8 Self Assessment Question:

In a distributive lattice, prove that two elements c and d are directly projective if and onl} if c=d.
 25.3.9 Self Assessment Question: In the lattice given below, prove that c and d are directly

projective. Note here that the lattice is not modular and the intervals [0, cJ and [0, dJ are not
isomorphic and gence not projective.

c
h

d
o
The concept of directly projective elements is used in the proof of the lattice theoretic form
of the Krull-Schmidt theorem given below. It's proof is beyond the scope of your syllabus and hence
we are not giving the proof.

25.3.10 Theorem (Krull - Schmidt Theorem) : Let L be a modular lattice with 0 and 1 that
satisfies both the chain conditions. Let a ELand

a1va2 v van=a=bj vb2 v vhm

o
where a,' sand b/s are indecomposable elements in L such that the sets {a1,a2, ....,an}

and {q) b2, .... , bn} are independent Then n == m and there is a one-to-one correspondence between

ai's and bj' S such that the corresponding elements are directly projective.

25.4 SUMMARY :
In this lesson, you have learnt the concepts of an irreducible element and an irredundant
representation of an element and proved that any two irredundant representations of an element in
a modular lattice the same number of irreducible elements. Also, we have defined the notions of
independance and indecomposability.

<. 25.5 TECHNICAL TERMS

Irreducible element
Irredundant representation
. Independent set
o
Decomposable element
o
Indecomposable element
\ Directly projective elements
Krull-Schmidt ~eorem ,0

o
 /

Centre for Distance Education 8 Acharya Nagarjuna University

25.6 MODEL EXAMINATION QUESTIONS .

26.6.1 : Prove that every element of a lattice satisfying the ACC can be expressed as the gtb
of finite number of irreducible elements.

26.6.2 : Define the concept of an irredundant representation and prove that the member of
terms in any two irredundant represntations of an element in a modular lattice as gtb
of irreducible elements is the same.

26.6.3 : Define the notion of an indecomposable element and prove that any element in a lattice
satisfying DCC can be expressed as the Rub of a finite number of independent
indecomposable elements.

25.7 ANSWERS TO SELF ASSESSMENT QUESTIONS

25.2.2 (1) : Let a E P. Choose two primes p =1=q which are not divisors of a. Put
b =av p=l.c.m{a, p} =ap and c=aq . Then a < b, a < c and a=b /\ c. Therefore
a is reducible.

(2) : Let a, b, c be in a chain and a = b /\ c. Then either b s: c or c s; b and hence

b /\ c =h or b /\ C = c, so that a=b or a =c . Thus a is irreducible.

(3) : Let 1 be the greatest element in a lattice L. Then

l=b/\c => l~b and l~c=>l=h and l=c.

25.2.5 : Suppose a =1=1 is irreducible. Let P be a prime dividing a and n a positive integer such
that

»" divides a and pn+l doesn't divide a.

If p" oj:. a , then take b = ~n . Then

and a =1=b and a =1=p" . This is a contradiction since C! is irreducible. Thus a= pn .

Conversely a= pn , where p is a prime and n E Z+ . If

a= b /\ c= l.c.m{ b, c} ,
 then p n /"D, P n / c and b=p r and c= p S for some r,s and hence Ii.c.m {h},c = p max{r.s},

Therefore a= p" or a= r', Thus a is irreducible,

25.3.4 : Suppose a is decomposable. Then a = al va2 for some al, a2 < a with al 1\ a2 = 0, Then

{(7i' a2} is independent. Conversely, suppose a = al va2 v"" v an' for some independent set

faj,G2,"'Cln} with al < a. Then, by dropping certain a.' e, we can assume that

'el2 v (/3 v ...v an < a . The n {al ,a2 v "'"'v an } is in d e pen den t a = a1 v (a2 v ""' v a3) ,

a1 <a, a2 v"'"' v an <a and al 1\ (a2 v ....v all )=0.

25.3.7: Let c and d be directly projective, Then b v c=b v d and b 1\ c=b 1\ d =0 for some b e L·

Put a = h v c . Then the intervals [b, a] and [0, c] are transposes and the intervals [b, a] and [0, d]

are transposes, Thus [0, c] and [0, d] are projective,

25.3.8 : If c = d , then °v = °v d
c and 01\ C = 01\ d = ° and hence c and d are directly projective.
Conversely suppose c and d are directly projective. Then there IS b E L such that

hvc=bvd and h/\c=bl\d=O

Then c = c 1\ (b v c) = C 1\ (h v d) = (c 1\ h) v (c 1\ d)

= (b 1\ d) v (c 1\ d) = (h v c) 1\ d

=(b vd) f\ d =d
•• ,';1 ,~"

25.3.9 Since b v c = 1= b v d and b 1\ C = b /\ d=ti , c and d are directly projective.

25.8 EXCERCISES :

25.8.1 : Let {a},a2' ","",an} be an independent set in a modular lattice L such that

al v a2 v "'"'v an "" 1. For each 1 '5{, i '5{, n , let

Then prove the following.

~~c~en~t~re~f~o~r~Di~st~an~c~eJE~d~uc~a~ti~on~~~~~~10~~~~~~~~1A~c~ha~r~ya~N~a~g~ar~jU~n~a~U~n~iV~er~si~t

(4) ail\bi=O and aivbi=l

25.8.2 : Prove that any subset of an independent set is independent.

25.8.3 : Let {aI, a2, , an } be an independent set in a modular lattice Land 1 :<::;; 'i :<::;; r2 :<::;; < rk =n .
Let

q = al v v a'1 ' b2 = all + I v v a~ ' , bk = aIR:-1 + 1 V v aIR:.

Then prove that {b],b2, ..... ,b,J is independent.

25.8.4 : Let L be a modular lattice satisfying the chain conditions and ill, a2, , an E L. Prove

that {aJ,a2, ,an} is independent if and only if

(see 24.6.5)

25.8.5 : Let L be a modular lattice and al,~,······,an,an+l EL such that {al,a2, ,a,,} is

independent and (a] v v an) 1\ an+] = O. Then prove that {al"':"" an + 1} is independent.
25.8.6 : Prove that {a1, a2, .... ··., an} is independent in a modular lattice L if and only if

25.9 REFERENCES :
1. Lattice Theory, Garret Birkhoff, revised edition (1949),
Amer Math Soc. Colloq. publication (1949)
2. General Lattice Theory, George Gratzer, Academic Press, New York (1\378)
3. Lectures in Abstract Algebra, N. Jacobson, Vol. 1,
D. Van Nostrand Company, Inc, London, (1964)
4. Algebr.a, M. Artin, Prentice - Hall of India, New Delhi (1994)

Lesson Writer
Prof: U.M. Swamy
Andhra University, Visakhapatnam.
Lesson : 26 BOOLEAN ALGEBRAS

OBJECTIVES
After reading this lesson, you should be able to
• define the notion of a complemented lattice and prove that a complemented modular
lattice satisfies both the chain conditions if and only if the largest element is the lub
of finite number of independent points and
• define the concepts of a Boolean algebra and a Boolean ring with identity and
establish equivalence between these two abstract systems.

STRUCTURE
26.1 Introduction

26.2 Complemented modular lattices

26.3 Boolean algebras
26.4 Boolean rings

26.5 Summary

26.6 Technical Terms

26.7 Model Examination Questions

26.8 Answers to Self Assessment Questions
26.9 Exercises

26.10 Reference Books

26.1 INTRODUCTION'

For any set X, the set lP'( X) of all subsets of X form a distributive lattice (under the
usual set operations) in which each element has a complement. If each element of a lattice with 0
and 1 has a complement, it is called a complemented lattice. Complemented distributive lattices,
known as Boolean algebras, are very rich in structure forthe simle reason that any' Boolean algebra
is isomorphic to a sublattice of TID( X) for a suitable set X. In this, lesson, \file shall discuss certain
important properties of Boolean algebras and of complemented modular lattices which are slightly
weaker than Boolean algebras.

/
 Centre tcr Distance Education 2 Acharya NagarjunaUniversity

26.2 COMPLEMENTED MODULAR LATICES .

The lattices of subspaces of a vector space over a field is modular lattice in which, for each
A, there exists another subspace A' such that A v A' = 1 and A 1\ A' = O. In particularly if the
vector space is finite dimensional, then the one-dimensional subspaces are covers of 0 in the
lattice of subspaces and the whole space is the lub of finite numbers of one-dimensional subspaces.
Also, the lattice of subspaces of any vector space satisfies either of the chain conditions if and only
if the vector space is finite dimensional. These ideas are abstracted in the section.

26.2.1 Definition: Let L be a lattice with 0 and 1 and a E L. a is said to be complemented if

there exists dEL such that

a r.o' =0 and av d=l.

In this case a' is called a complement of a. We say that L is a complemented lattice if

every element of L is complemented.
26.2.2 Examples:

(1) The lattice JP>( X) of all subsets of a set X is a complemented distributive lattice in

which, for any A E JP>( x), the subset A' =X - A = {x EX / x E A} is a complement of

A.
(2) The lattice giVen in fig. (i) below-is a complemented modular lattice which is not
distributive.

,
i c
I
a
a
fb c

l
r
o
o
Fig (i) Fig (ii)
(3) TheIattice given in Fig (ii) above is a complemented lattice which is not modular.

26.2.3 Definition: For any element a of a lattice L, let us write La to denote the set {x E L / x ~ a} .
Then La is a sublattice of L , 0 is the least element in La and a is the greatest element in La . If
b E La and bl is a complement of b in the lattice La' that is, if b 1\ bl =0 and b v bj = a, then 11j is
called a complement of b relative to a.
 ~ Lesson 2~) c: Algebra ::B
26.2.4 Self Assessment Question

Give an example of a complemented lattice L in which there exists a E L such that Lq is

not complemented.

If L is a complemented lattice then, for any a E L, La may not be complemented in genAral;

However, the case of modular lattices is different as given inthe following ..

26.2.5 Theorem : Let L be complemented modular lattice. Then La is complemented for all
\

aEL.
Proof: Let a ELand b E La . Then b S; a . Let b' be a complement of b in L . Then h v b' = 1 and
b r.b'<t): Put b1=b"l\a. Then

bv~-chv(b'l\a)=(bvb')l\a (c;ince L is modular and b s;a)

=lI\CI=a

,
Therefore h; is a complement of b in La' Thus La is complemented.
o
26.2 15 Definltlon ; Let L be a lattice with O.An element of L which covers 0 is called a point in L ;
that is, p is a poi~t in L if 0-< P .

26.2.7 Self Assessment Question : Prove that any nontrivial lattice satisfying the descending
chain condition has the least element 0 and has atleast one point.

26.2.8 Self Assessment Question: Prove that any two distinct points are incomparable in any
lattice.

26.2.9 Theorem: Let L be a nontri~ial modularlattic~Jth 0 and 1. Then L is complemented and

satisfies both the chain conditions if and only if the lar~est element 1 is a lub of finite number of
independent points.

Proof: Suppose that L is complementedand that L satisfies both the chain conditions ACC and

DCC. Since L is nontrivial, L :;t:{0} and hence L-{O}:;t:¢. Since L satisfies DCC, L-{O} has

minimal elements which are covers of Oor points in L . Consider the set

A ~ {PI \I P2 'v :..:..~ PII In > 0 and each Pi is a point in L}

By the above argument, A is a nonernpty subset of L . Since L satisfies.ACC, A has a maximal

element, say aa. Then we shall prove thatao = 1. Otherwise, suppose ao:;t: 1. Let aa be a
complement of CIa in L. Then O<ao. Again, by using the DCC in L, wecan get a point p in L

27;
 Centre for Distance Education 4 ACliarya .Nagarjuna University

such that ° < P ::;aO' Now, ao v P E :;1-and hence, by the maximality of ao in A, it follows that

Go v p=ao· This implies that P::; ao and hence P::; ao A ao =0, which is a contradiction. Thus

ao = 1 and, since 1= ao E A, it follows that

1= PI V P2 v .....V Pn

for some pointS-PI' P2, ...., P« in L . Also, by dropping some Pi s if necessary, we can assume
I

that

l}'"-=Pl v .... V Pi-I V Pi+I v .... V Pn for all 1::;i ::;n

Now, for any i , '~O::;Pi A (PI v .....V Pi-I V Pi+l v: ....V P« )::;Pi and, since Pi is a point, and

Pi t PI v .....V Pi-l V Pi+l v ....V Pn , it follows that

1 nus {PI, ...., Pn} is an independent set of points whose lub is 1.

Conversely, suppose that the largest element 1 is a lub of finite number of independent points.
We have to prove that L is complemented and satisfies-both the chain conditions. Let

1= PI V V P«
P2 v .....

where {PI>" ...., Pn} is an independent set of points in L . Then

Put ai = PI v P2 v ....V Pi for 1::; i <n

,
. Then we have ai < ai+l . Also, since ai v Pi +1 =ai+I
'

and ai APi+I =0, the intervals [ai, ai+lJ and [O,Pi+I] are transpose intervals and hence are

isomorphic. Since Pi+l i,s,a point, we have 0-< Pi+I and hence ai -<ai+I. Therefore, w~ have the
composition chain.

0-< PI = al -<a2 -<a3 -< -<an = 1

The existance Of such a composition chain implies that L satisfies both the chain conditions
ACC and DCC. Finally we are left with proving that L is a complemented lattice. Let a E L. For
each 1 ~ i ~ n , 0::; a A Pi ::;Pi and, since Pi is a point, either a /\ Pi = 0 or a 1\ Pi = Pi; ; that is, for
any i ,
..~.
~ Lesson 26) ( Aigebra~

either Pi ::s; a or a 1\ Pi =0
Since 0 and1arecomplements to each other, we can assume that -a:;z:t:Fuf

Since Prv P2 v .---V Pn =1 ":j::. a, there exist atleast one Pi such that Pi i a (if Pi::S; a for
each i , then 1-=PI v ....V P« ::;a). As in the proof of theorem 25.3.2, we can-prove that

a 1\ b = 0 and a v b = 1

(since {PI> P2, ..., Pn} is an independent set) and hence b is a complement of a in L.
Thus L is a complemented lattice.

26.3 BOOLEAN ALGEBRAS

In this section, we shall discuss a class of rich lattice theoritic structures, namely Boolean
algebras named after George Bool who first proposed the algebra of proposionallogic.
26.3.1 Definition: A lattice WItnoanEl-t-WbiGA-is complemented and distributive is called a Boolean
algebra.
26.3.2 Examples:

(1) The two element lattice {O, I} is a Boolean algebra, where 0 < 1 and 0 and 1 are
complements of each other.

(2) For any set X , the lattice lP'( X) of all subsets of X is a Boolean algebra under the
lattice theoritic operations.

(3) Let L be a class of subsets of a set X such that

Then L is a Boolean algebra under the usual set theoritic operations. Such an L is called
a field of sets. -

26.3.3 Theorem : Let B be a Boolean algebra. Then the following hold ,

(1) For any a

.
E B , the complement of a is unique in Band , .
is.denoted
. ~.
by a' .

(2) the map af-7a' is of period 2.

~-

(3) ~The map af-7a' is a bijection of B onto itself.

.~

-._---- -::.:: "'-~

 Centre for Distance Education 6 Acharya Nagarjuna University

(4) Foranya,hEB, (avh) ~a'l\h'and(al\h)' =a'vh'.

Proof:

(1) Let, a E B . Suppose hand carecornplernentsof a, in B . Then a v b = 1 = a v C

and a 1\ h=O=a I\C.

Consider h=hl\l=hl\(av.c}=(hl\a)v(hl\c)=Ov(hl\c)=hl\c and hence b s.c .

b and hence b = c .
Similarly, c -::;,

Thus a has a unique complement in B .

(2) Let a E B and a' be thecomplement of a in B . Then

and hence, by the uniqueness of the complement, a is the complement of a' and therefore
. I
(a') = a . Thus the map a H a' is of period 2.
1. ••

(3) This is an immediate consequence of (2).

(4) Let a,h E B . Then

.
( av h) 1\ (a' 1\ h')=( a 1\ a' 1\ h')v(h Afl' 1\ h') ~OvO=O

and (avh)v(a'l\h')=(avhva')I\(avhvh')=ll\l=l
,
and hence a' 1\ h' is the complement of a v h ; is that is.. (a v h) = a' 1\ b'

·.11 :-
, I "
Now, (al\.b) = (a" I\b") =(a'vb') =a'vb'

26.3.4SAQ: I.:

For any elements a and b in a Boolean algebra, prove the following.

(1) a -::;,
b' cs a r.b=t)
.
b 6'crv h=l
"

"(2) a' -::;,

,
(3) a s.b cs b'<o'

26.3.5 Theorem: The following are equivalent for any Boolean algebra B .
 ( Algebra "-E
(1) B satisfies ACC

(2) B satisfies DCC

(3) B is isomorphic to lP( X) for some finite set X. "

(4) B is finite.
. .

Proof: If al:::::;a2 :::::; is an ascending chain, then by 26.3.4(3), ai;::: a2, ;::: is a descending
. .

chain and vice-versa From this we get (1)<=:>(2).

i ",'

(2)<=:>(3)supPOSE? B s~tisfi.e.s DCC and hence ACC too. Then, by 26.2.9,

1= PI v P2 v ....V Pn , where PI, P2, .... , Pn are distinct. points.

. .
Note that P /\ q = 0 for any two
distinct points P and q in B , then

i . Thus Pl,P2, ..·...Pn are all the points in B. Put

x = The set of all points in B .

Then X={Pl,P2, .... ,PI1} and hence X is finite. Define f:B~ J1D(X) by

j(a)={p EX / P :::::;a}. Then, clearly ,

b z:»
a :::::; f (a) ~ f (b)

on the other hand, supposenno.cse . Then a/\h'=f;O. Since

a /\ b' = (a /\ b' /\ PI) v v (a /\ .',' /\ Pn), it follows that a /\ b' /\ Pi :;t: 0 for some i and hence
. /.

a /\ b' /\ Pi = Pi (since Pi is a point),' so that Pi:::::;a /\ b'; Then Pi:::::; a and p; i b and hence

Pi E f (a) and Pi rt f (b ) . Therefore f (a) c;£ f (b) . Thus we have

,

Also, for any A ~ X , put a = lub of A and then you can easily p~?~~ethat j (a) = A . Thus .

f is surjection too. Therefore f:A ~lP'(x) is an isomorphism.

11 ., I;";.
(3) ~ (4) is trivial, since Ip (X) = 2 1.. if IXI=n .. ", ',"
 Centre for Distance Education 8 Acharya Nagarjuna University

( 4) => (1) is al=o trivial, since.any chain in a finite B is also finite. ()'

26.3.6 Self Assessment Question: Prove that any finite noa-trivial Boolean algebra' has exactly

2n elements for some positive 'integer 11 .

26.4 BOOLEAN RINGS .

In this section, we shall establish that the theory of Boolean algebras is equivalent to the
theory of a special class of rings. We shall first prove that any Booleanjalqebra can be considered
as a ring with respect to suitably defined operations.

26.4.1 Definition: A ring R =( R, +,.) is called.a Boolean ring if a- a =a for a~ka E R. A ring R is

said to be with identity if there exists a multiplicative identity in R ; that is, there is an element 1 E R
such that l·a=a for all a E R.

26.4.2 Theorem: Let R be a Boolean ring. Then a + a = 0 and ab = ba for all a, be R and hence
R is a commutative ring.
o
Proof: For any. a, b E R, consider
.:._"' •...
a + b = ( a + b) ( a + b) = aa + ab + ba + bb
\.

.'

+a+ab+ba+b

and hence ab + ba = O. In particular, by taking a = b , we have a + a = a , for all a E R

Therefore, ab = ab + (ba + ba) =( ab +: ba) + ba = a + ba = ba

- )1\ I,

Thus R is a commutative ring.

26.4.3 Theorem: Let B be a Boolean , algebra. For any a, b e B , defi~_ '

a+b=( a 1\ b')v (a' 1\ b)

---
and a-b=a r.b

Then (B, +,.) is a Boolean ring with identity.

. ~ ... ~--"
Proof: Since a+b=(al\b')v(a'l\b}=(bl\a')v(b'l\a)=b+a, + is commutative. To prove
associativity, c,?nsider
 (a +b) + C = (( a + b) A C') v ((a + b)' r-c )_~

which is symmetric in a,b, C and hence »<

(a+b )+c=( b+c )+a=a+(b+c).

Therefore + is associative. Also, for any a E B ,

----
a+O=(a A O')v(a' A O)=(a A l)v0=a

and hence 0 is the identity with respect to +. Further

,
a+a=(al\a')v(a' J\a)=OvO= 0

for any a E B ,s that a is the inverse of it5e1rwith respect to +.

Thus (B, +) is an abelian group. Now, since ~ is associative, commutative and idempotent,

so is • (since a- b=a A b). Finally,

a- b+a- c=[ (a A b) A(a A c)'Jv[ (a A,b)' 'A (a AC) ]

=[ a r.b A(a' Vc')Jv[(a'v b')Aa AC ]

=( a r.b J\a')v( a r.b AC')V( a' rea AC) v (b' A a AC)

=aA((bAC') v (b' AC))

=a.(b+c)

Thus. distributes over +. Since. is commutative, we have (a+b)-c=a.c+b.c also.

 .Centre for Distance Education 10 Acharya Nagarjuna University

Thus (B,+,.) is a Boolean ring. Also l·a = }i\a=a, where 1 is the multiplicative identity. Thus

(B, +, .) is a Boolean ring with identity.

26.4.4 Theorem: Let (B, +,,) be a Boolean ring with identity. For any a, bE B, define

a v b e.a+b=ab

and a i\ b = a . b

Then (B, v, i\) is a Boolean algebra.

Proof: Since a i\ b = a band . is associative, commutative and idempotent, so is i\. For any
a.b e B:

avb=a+b-ab=b+a-ba=bva.
Therefore v is commutative, Also,

ava=a+a-aa=a+a-a=a
and hence v is idempotent. To prove the associativity of v consider J

(av b)v e=( a+b= ab)+e.-(a+b-ab)e

.= a + b + e - ab - ae - be + abe
which is symmetric in a, b, e . Therefore

(avb)ve =(bve)!va=av(bve)
!
,
and hence v is associative. Further, to verify the absorption laws,

a i\( b v a)=a(b+a-ba )=ab+aa-:-aba=a .

and a v (b r. a) = a + ba - a (ba) = a

Therefore (B, v, i\) is a lattice. To verify the distributivity,

a i\ (b v e) = a (b + e - be) = ab +ae - abe

=ab +ac-( ab)( ac)

=(ab) v( ae)
 ( Algebra :E

Therefore, (E, v, /\) is a distributive lattice. Finally, for any aEB,

av (l-a )=a+(l-a) -a(l- a)= 1

and a /\ (1- a) = a ( 1- a) =0

and hence 1- a is the complement of a. Thus (B, v, /\) is a complemented distributive

lattice; that is, it is a Boolean algebra.

26.4.5 Theorem:

(1) Let (B, v, /\) be a Boolean algebra and (B, +,.) be the Boolean ring obtained from

(B, v, /\) as in 26.4.3. Let (B, v', /\') be the Boolean algebra obtained from (B, +,.) as in 26.4.4.
Then the operations v and /\ coincide with v' and /\' respectively.

(2) Let (B, +,' )be a Boolean ring, with identity and (B, v, /\) be the Boolean algebra

obtained from (B, +,.) as in 26.3.4. Let (B, EfJ,*) be the Boolean ring obtained from (B, v, /\) as in
26.4.3. Then the operations + and; coincide with EBand * respectively.

Pro~f: (1) a v' b = a+b-ab

=a+b(l-a)

= (a' /\b/\a')v(a/\(b/\a')}':i'
Iji:

=(a' /\b)v(a/\(b'va))

=(a'/\b)va

= (d va) /\ (b va)

=l/\(avb)=avh

and a /\' b = a- b = a r. b for any a, bE B

(2) a!f>b= (a/\b')v(a' /\b)

 Centre for Distance Education 12 Acharya Nagarjuna University

=( a(l-b ))v((l-a)b)

=a(l-b )+b(l-a )-a(l-b)b(l-a)

=a-ab+b-ab-O

.=(-a+b)-(ab +ab )=( a+b )-O=a+b

and a * b = a 1\ b = a .b , for any a, b E B

26.4.6 Theorem:

The class of Boolean algebras and the class of Boolean rings are equivalent.
Proof: Follows from theorems 26.4.3,26.4.4 and 26.4.5.

»:

26.5 SUMMARY :
In this lesson, we have proved that a complemented modular lattice satisfies both the chain
conditions if and only if the largest element is the lub of a finite number of independent points. Also,
we have introduced the notions of a Boolean algebra and a Boolean ring and proved that these two
systems are equivalent.

26.6 TECHNICAL TERMS .

Complemented lattice

Complement of an element
Point .-
nt::.'
Independent set

Boolean algebra

Boolean ring

26.7 MODEL EXAMINATION QUESTIONS

26. 7~1 : Prove that a complemented modular lattice satisfies both the chain conditions if and only if
1 is the lub of a finite number of independent points.
26.7.2: Show that a Boolean ring with identity can be constructed from a given Boolean algebra.

26.7.3: Prove that a Boolean algebra can be constructed from a given Boolean ring with identity.
26.7.4 : Show that the classes of Boolean algebras and Boolean rings with identity are equivalent.
/
 ( Algebra ~

26.8 ANSWERS TO SELF ASSESSMENT QUESTIONS

26.2.4: Consider the latticeL given in Fig (ii) of 26.2.2(2). Then L is a complemented lattice and

Le ={O,b,c}, which is not complemented (since b has no complement in Le)

26:2.7: Let L be a nontrivial lattice satisfying the descending chain condition. Then, by 24.6.3, L
has a minimal element, say m , For any a E L, m /\ as; m and hence m /\ a=m so that m ~ a .
Therefore m is the least element in L. Now, consider L \{m}, which is nonempty since L is

nontrivial. Again by 24.6.3, L \{m} has a minimal element, say p Then p is a point in L.

26.2.8 : Let p oF q be points in a lattice L . Then 0 -< p and 0 -< q and hence p i q and q i p . That
is P and q are incomparable.

26.3.4: (1)

a /\ b=O=>a=a /\( b v b')=( a /\ b) v (a /\ b')=Ov (a /\ b')=a 1\ b' ~ b'

(2) b ~ 1=·a v a' ~ a v b => a v b = 1

a' ::::;

av h=l=>b=h v O=h v (a /\ d)=( bv a) /\ (b v d)= 1/\ (h v d)=h v a'> a'

a s b <:::> dvb=l (by (2»

<=> b'< a' (by (2), again)

26.3.6: Let B be a nontrivial finite Boolean algebra. Then, by 26.3.5, B ::::JPl( X) for some finite set

X. Since B is nontrivial, X::j; ¢. Let IXI=n >0. Then IBI= IJtD(X) 1=2/1.

26.9 EXERCISES :
26.9.1 : Let L be a distributive lattice and a, h, E L with a ~ b . Prove that every element in the

interval [a, h] has atmost one complement in [a, h]

26.9.2: Prove that a lattice L is distributive if, for any a, bEL with a s: b·,cevery element of [a, b]
has atmost one complement in [a, b] .
26.9.3 : Prove that a complemented modular lattice satisfies Acc if and, ."only if it satisfies DCC. \"
 Centre for Distance Education 14 Acharya Nagarjuna University

,
26.9.4 : Let B be a Boolean algebra and, for any a, b e B, define a + b = (a vb') 1\ (a' vb) and

1.b=av b. Prove that (B, +,.) is a~Boolean ring ..

26.9.5 : Let (R, +, .) be a ring with identity and

B (R) = {a E R / a2 = a and xa = ax for all x E R}

Forany a.b e B(R), define

avb=a+b-ab and a r.b+ab,

Then prove that (B (R) , v, 1\) is a Boolean algebra.

26.9.6 : Let Band B' be Boolean algebras and Rand R' be the Boolean rings obtained from B
~nd B' respectively as<in theorem 26.4.3. Let f:B~B' be a map. Prove that f is a lattice

homomorphism such that f (0) = 0 and f (1) = 1 if and only if f: R ~ R' is a homomorphism of
rings.

26.9.7: Let B be a Boolean algebra and X be the set of all homomorphisms of B onto the two
element Boolean algebra. For any a E B , let

(J (a) = {f E X /f (a) = 1}
.-
Then prove that (J:B ~JlD( x) is an injective homomorphism.
~.~
,

erio.
26.9.8 : Prove that any Boolean alqebrajsornorphic to a field of subsets of a suitable set X .

26.10 REFERENCES :
1. Lattice Theory, Garret Birkhoff, revised edition (1949), Amer Math Soc. Colloq. publication
(1949) .

2. General Lattice Theory, George Gratzer, Academic Press, New York (1978)
3. Lectures in Abstract Algebra, N. Jacobson, Vol. 1, D. Van Nostrand Company, Inc,
London, (1964)

4. Algebra, M. Artin, Prentice - Hal! of India, New Delhi (1994)

Lesson Writer:
Prof: U.M. Swamy
Andhra University, Visakhapatnam.
Lesson' 27 APPENDIX

OBJECTIVES

The objective of this appendix is to let you know about certain fundamental aspects of
set theory which are necessary to be known by any student of mccern mathematics
and which are already used freely in the present course, or, for that matter, in any
branch of mathematics.

STRUCTURE
27.1 Introduction

27.2 Relations and functions

27.3 Schroeder - Bernstein theorem

27.4 Finite sets and countable sets

27.5 Axiom of choice

27.6 Well-ordering theorem

27.7 Zorn's lemma

27.1 INTRODUCTION:
There are certain fundamental concepts in set theory which are necessarily to be known by
any student of mathematics and which are already freely used in the present course or, for that
matter, in any branch of mathematics. On several occations, we made use of Zorn's lemma directly
and the axiom of choice indirectly. Just to say that the'·'@ictrtesianproduct of a family of groups is
again a group, we should first make sure that it is nonempty set. Here, the use of choice axion is
necessary. It is obvious that the Cartesian product of a finite number of nonempty sets is again
nonempty. But when we consider an arbitrary class of nonempty sets, it is nontrivial that the Cartesian
product is nonempty. In this case, infinitely many choices are involved. For the systematicstudy of
set theory and, through this all branches of mathematics, the inclusion of the existence of a choice
function was necessiated by Zermelo when he proved his calebrated theorem on~the eXis.tenceOf
a well-ordererinq on any given set. Even today, almost a centuary after provi g the Zsimelo's
theorem we do not know any constructible method to define a well-order on an uncountable set.
But, Zermelo proved that there exists a well-order on any set. This arose a considerable debate on
the validity of Zermelo's theorem. After a careful scruitiny of the proof by severall mathematicians
and logicians, it was found that the only doubtful part of the proof is the use 0\ the existence of
choice functions. Then people started doubting the axiom of choice itself. After several years,now
it isunanimous that the axiom of choice, and hence the Zermelo's theorem, is a necessary one to
be included in the list of axiom of settheory.
 Centre for Distance Education 2 Acharya Nagarjuna University

In this lesson, an-attempt is made to briefly expl un all the essential elementary concepts
that.are used in the earlier part of the book.

27.2 RELATIONS AND FUNCTIONS .

Right from the begining of this course, or any other course in your Master's programme,
you are using the names 'relation' and 'function' and several of their properties. In this section, we
shall present the formal definitions and certain important properties of these for the use of persons
who are new to all these.

27.2.1 Definition: Let A and B be any nonempty sets and A x B be the Cartesian product of
A and B . Any subset R of Ax B is called a relation fr,.·m A to B . We write aRb to-denote that
(a, b) is an element of R . -,

Functions can be described as a special type of relations as you see in the following.

27.2.2 Definition: Let A and B be two nonempty sets. A relation R from A to B is called a
function from A in to B (or a mapping from A into B ) if, for each a E A , there exists unique b E B
such that (a, b) E R .

In other words, a function R is a subset of Ax B such that, for each a E A , there exists·
.one and only one bE B such thai. .•.Rb . If R is a function ana 'Rb, we write R( a) or Ra for b;

that is, R (a) =b if and only if aRb. It is conventional to use the letters f, g, h, etc ... to denote a

function. Also, we write f: A ~ B to say that f is a function of A into B . For any a E A , f (a) is

uniquely determined element in B a~ is.called the image of a under f.

27.2.3 Definition: Let f: A ~ B be a flfnction. Then f is called an injection or one - one function

if distinct elements of A have distinct images; that is x 7= YEA => f (x) 7= f (y); or, equivalently

for any x,Y E A, f(x)= f(y )=>x= y.

27.2.4 Definition: A function f: A ~ B is called a surjection or an onto function if each element of

B is the image of some element in A ; that is, for each b E B , there exists a E A such that f (a)= b .
A function f: A ~ B,)s called a bijection or one-to-one correspondence if it is both an injection and
a surjection. If f: A ~ B is a surjection, then we say that f is a function of A onto B .

27.2.5 Exercise: Which of the functions given below are injections, surjections or bijections?

(1) f :IR. ~ IR.; f (x) = 2x

 -,
~ L~~son 27)
-- ;~)

\
(2) j :7L--+ 7L; j (n ) = 2n

(3) j:7L--+7L+,j'(x)=x2 +1

(4) j:C--+R+ U{O}, l(z)=lzl

(5) I :~ --+ ~, I ( x) = sin x

(6) I :~ --+ ~, I ( x) = tan x

27.2.6 Definition : Let R be a relation from A to Band S a relation from B to C. We shall
define a relation SoR from A to C by

Sol'. = {( a, c) E A xC / three exists.b EB such that aRb and bSc}

)

In otherwords; a (SoR) c if and jnlY· if, for some b E B, aRb and IiSc . In particular, if
----
./

j: A --+ 13 and g: B --+ C are functions, we shall define the relation

··goz~/r~c by (goj)(a)=i(j(a))

for any a EA. This is same as saying that there exists b e B such that aft (i.e., f(a)=b) and

bgc (i.e., g (b) = c). It can be easily verified that gof is a function of A into C , whenever f: A ~ B

and g: B --+ C are functions. If j and g are injections (surjections, bijections), then you can easily
verify that gol is also an injection (surjection, bijection respectively). Also, if j: A --+ B is a bijection,
then, for each bE B , there exists unique aE A such that f (a) = b ; This unique a may be denoted

by f-l(b). This defines a function f-l:B~A such that I(f-l(b))=b for all bEB and

f-l(f(a))~forall aEA;thatis,

where 1A and 1B are identity functions defined from A into itself and B into itself

respectively by j~{a)~a and 1B (b )=h for all a EA and b « B.

27.2.7 Definition: Let A be a nonempty set. Any relation from A to itself is called a binary
relation on A . Let R be any binary relation on A . Then R is said to be
I
 Centre for Distance Education 4 Acharya Nagarjuna Univer~~

(1) reflexive if aRafor all a E A

(2) transitive if aRb and bRc imply aRc

(3) symmetric if aRb implies bRa.

(4) anti-symmetric if aRb and bRa imply a=b

(5) an equivalence relation or, simply, an equivalence if R is reflexive, transitive and

symmetric.

If R is an equivalence relation on a set A and a E A, the subset R (a) defined by

R-( 0 = {b E A / aRb}

is called the equivalence class of a with respect tOR. You can easily verify the following
for any elements a and b of A .

(1) R(a)=R(b)<::>aRb

(2) Either R (a ) nR (b) = ¢ or R ( a ) = R (b)

(3) R ( a) is a nonernpty subset of A .

(4) 0 A is the union of all R( a), a EA.

In other words, R (a) 's form a pairvvJsedisjoi~J n~em>pty subsets of A whose union is A
,r. "~ _
Such a class of subsets of A is called a ~rtit~biTOf 'A .
~nth~ other hand, if P is any partition of
A , and if we define a binary relation R .14 by

aRb <::> both a and b belong to' ~he same member of P, then R is an equivalence
relation A and the partition P coincides with the class.of-all equivalence classes R (a), a EA.
. - .! .
27.3 SCHROEDER-BERNSTEIN 'THEORE·M :
In this section, we prove one of the very important theorems in set theory. This theorem
enables us to introduce an ordering on cardinal numbers. First, we shall have the following notation:'

If f: A ~ B is a function, Al <;;; A and BI ~ B , then we write f (Ad

for the set

 ~ Lesson 2~') (Algebra ~

j (AI) is called the image of Al under j and j- I (BI) is called the inverse image of BI

under f. Note that /-1 is not a function (unless / is a birection). If / is a birection, then /-1 (Bd
is the image qf B~ under the function /-1 :B -) A . Now, we prove the following:

27.3.1 : Theorem : Let B be a subset of A and suppose there is an injection I: A -) B. Then

there exists a bijection of A onto B .
\

Proof: Let us define sequences {An} and {En} of sets inductively as follows:

Then, we have

Now, define g: A -DB by

g(a)={f(a) if a E A~-Bn forsomen

a otherwise

Then, you can easily check that g is a bijection of A onto B .

/
27.3.2 Theorem (Schroeder - Bernstein theorem)/: Let j: A -) Band g: B -) A be injections.
Then there exists a bijection h: A -) B .

Proof: Let X = g (B). Then X is a subset of A . Now, define

a:A-)X

by a(a)=g(f(a)); that is, a+ gof , Since both g and f are injections, a is also an injection.

By theorem 27.3.1 , there exists a bijection of A onto X . Let a: A -) X be a bijection. Note that,

.since g:E -+A is an injection, we have, for each x E X =g(B) ,there exists unique b.e B such

that x = g (b) . In otherwords, g can be treated as a bijection of B onto X and hence there exists

a bije~tion g - r :X -) B such that

28)
 Acharya Nagarjuna University
Centre for Distance Education 6

gog-l = IX and g-log=IB

~
Now, define h:A~B by h=g-loa. Since ao.'A~X and g-l:X~B are bijections, h is a
bijection of A onto B .

27.3.3 Example: Let Z+ denote the set of positive integers. Define

..--'
by f(nj~(n,l) and g(n,m)=2n.3m. Then f and g are injections. Note that neither of
these is a bijection. But, by the Schroeder - Bernstein theorem (27.3.2), there is a bijection

h: Z+ ~ Z+ X ~+ .

27.3.4 Example: Let m be any positive integer and PI, P2, ..... , Pm be distinct prime numbers.
, JJefine

Then f and ~,:a~~jectives. By theorem 27.3.2, there is a bijection

,

n.z: ~ (i+ r.
27.3.5 Example: Let Q + denote the set of positive notional number. Then, any a E Q + can be
/' ~ti::;~
uniquely written as !!...-, where nand m are positive integers which are relatively prime. Define
m

~ by
/
T-.and
f (n) ::::
.
g (: ) = (n, m) , when~ver nand m are relatively prime positive integers.'

J '
Then JLJ?_Jm injection. Composing g with the bijection h-1 in 27.3.3, we get an injection
-
gl >Q + ~Z+ and hence, again by 27.3.2, we get a bijection of Z+ onto Q+

c
,27.3.6 Theorem: For any set X, there is an inrection of X into the set lP'~X) of all'sUbset}pf
X ~ut there is-no'~n of-Jlll( X ) into X .
~ Lesson 2D C Atgel1ra ~
\
Proof: We can assume that X is a non-empty set. Define f: X ~ JfD(X ) b~ x) = {x2 }. for all

x E X Then clearly f is an injection. Now suppose, if possible, there is an injection g: J1D(X) ~ X .

Then, by the Schroeder-Bernstein theorem (27.3.2), there is a bijection h:X ~JID(X). For each

x EX, h( x) is a subset of X. Consider the set

A={x E X/x~h(x)}

Then A E J1D(X) and A:;t:h ( x) for all x E X (for, if x E X is such that A E J1D(X) and

A = h (x) , then x E A <=> x ~ h ( x) <=> x ~ Ai). This is a contraction, since h is a bijection and hence
a surjection.

We often write A=::B to say that.there is a bijection of A onto B,Gm-d-in this case, we also
say that A is bijective to B . Theorem 27.3.6 implies that no set is bijective to its power set or
-
* p(X)
\

X for any set X.

27.4 FINITE SETS AND COUNTABLE SETS

A set X is called a finite set if either X is empty or bijective with theset In = {I, 2, ...., n} for

some positive integer n. If X=:: In and X =::IIn ' then In =::1In and hence ~ = m . Therefore, if X is
/

a nonempty finite set, then there exists unique positive integer n such that I In =:: X and, in this
case, X is said to be an n-element set and the elements in X can belisted as
.s f\

where i H xi is a bijection of In onto X . You can try top rove the followi[lg simple theorem which
gives us a tool to check whether a given set is finite.
\ .

27.4.1 Theorem: Let X be a nonempty set and m a positive- integer. Then the following are
equivalent.
-
(1) X is a finite set with atmost m elements.
-'.
(2) There is an injection of X into 1m,

(3) There is a surjection of 1m onto X .

I

27.4.2 Corollary :
 Centre for Distance Education 8 Acharya Nagarjuna University

(1) Every subset of a finite set is finite.

(2) A finite set is not bijective with any of its proper subsets.

(3) The set Z+ of positive integers is not finite.

(4) Any finite union of finite sets is finite.

(5) Any finite product of finite sets is finite .

./
~27.4.3 Definition:
~-
A set X is said to be ---
infinite if it is not finite. X is called countably infinite if it
is bijective with, Z+, the set of positive integers.

By 27.3.3 and 27.3.5, the sets Z+ x Z+ and Q + are countably infinite. The set Z of all

integers is countably infinite; for the map f: Z ----+ Z+ defined by

2n ifn> 0
f 1'1 ={
() -2n+ 1 if n ~ 0

is a bijection. The set Q of rational numbers is also countably infinite. The set lP'(Z+) of all

subsets of Z+ is an infinite set which is not countably infinite, since Q there is no bijection of Z+

onto lP'(Z+) (see 27.3.6).

27.4.4 Definition: A set X is called countable if i!is either finite or countably infinite. A set which
is not countable is called uncountable.
Here again, we shall prove a theorem which is analogous to theorem 27.4.1 and which
provides a technique to check whether a given set is countable. The proof of this also give us a
technique to prove such similar theorems, in particular 27.4.1.
27.4.4 Theorem:

The following are equivalent for any nonempty set X

(1) X is countable

(2) There exists a surjection f: Z+ ----+ X

(3) There exists an injection g: X ~ Z+

Proof: (1) => (2). Suppose that X is countable. If X is countably infinite, then there will be a

bijection of Z+ onto X . Next, suppose that X is finite. Then, for some positive integer m , there is
a surjection h:lm ----)X. Now, let us define r.z: ----)X. by--,

hen) ii n s m
f (n)= {h(l) ifn>m
,
Then clearly f is a surjection.

(2) => (3) : Let f:71.+----)X be a surjection. Define g:X ----)71.+by

g( x) = The smallest integ'er in f-1 ( x )

That is, g (x) = n , where n is the smallest positive integer such that f (n) = x. Since f is a

surjection, there must be atleast one n E 71.+such that f(n)= x and hence g is well-defined. If

X:t:YEX, then f-1(x)nf-1(y)=¢ and hence g(x):t:g(y), since g(x)Ef~l(x)and

g(y) E f-1 (y). Thus g is an injection.

(3) => (1) : Suppose we are giveh that g: X ----)71.+is an injection. Put A = g (X). Then g: X ----)A is
'.
a bijection and hence it is enough if we prove that A is countable; that is, A is finite or countably
infinite. Suppose A is not finite. Then we shall prove that A is bijective with Z+ . Note that A is an

infinite subset of 71.+. Define a:71.+ ----)A as follows, inductively.

a(l)= The smallest integer in A and, for n > 1, a(~)l~ The smallestInteqerIn

A-{a(l), .....,a(n-l)}. Note that, since A is not finite, A-{a(1), .....,a(n-l)} is a nonempty

subset of 71.+and hence has a smallest member (why!). Therefore, a is well-defined. If n < m ,
Then arm )EA -{a(l), ,a(n), a(n+ 1), ,a (m -I)} and therefore a(m):t: a (n) . Thus a
is an injection. To prove that a is a surjection too, let a EA· Then, there is atleasf one n E 71.+

such that a S a (n) (sinc~ Z+ is infinite and a is an injection). Let 110 be the least in Z+""forwhich

a s a (no) . Then, for any 11 < no we have a (ll ) < a and hence
I .,,

/
a E A -{a(l), .....,a(no -I)}
 Centre for Distance EducatiOft~ Acharya Nagarjuna University
_~_...oo.

SO that a (~~ a~~_;;;~thedefinition~of:~ ( no) . Already, we have a ::;a (no) . Therefore a (noJ=a -.
Thus a: z+ ~ A is a bijection and hence Z+ :::A:::X. Thus X is countably infinite. This completes--
the proof of the theorem.

27.4.5 Exercise: In the above proof, we made use of the fact that any nonempty subset.of Z+
has least member. Prove this.

27.4.6 Corollary : \

(1) Any subset of a countable set is countable.

(2) Any countable union of countable sets is countable.

(3) Any finite product of countable sets is countable.

Proof:

(1) This follows from the fact that, if g: A ~ Z+ is an injection and B ~ A, then the restriction
of g to B is an injection of B into z+ .

(2) Let {Ai liE! be a countable collection of countable SAtS; that is, I is countable and each
Ai' i E I, is countable. We can assume that each Ai is nonempty. By theorem 27.4.4, there

exists surjections. /:z+ ~ I and gi :Z+ ~Ai for each i E I. Now, define

n.z: xZ+ ~ U Ai.

iE!
iO .
by h( n, m) = gf(n) (m) for, ~.ny (n, m) E Z+ X Z+ . If a E i~! Ai, then there exists i E I such that

a E Ai . Since t Z+ ~ I and gi :Z+ ~ Ai are surjections, there exist n, m e Z+ such that / (n) = i
and gi (m) =a, so that

h(n,m)=g f(x) (m)=gi (m)=a

Thereforeh is a surjection. If we compose h with a bijection of Z+ onto z+ x z: (use

27.3.3 to get such a bijection), then we get a surjection of Z+ onto U Ai ?Thus, again by 27.4.4,
iEI

U Ai is countable.
iEI
\
 C Aigebra.~
(3) It is enough if we'prove that product of any two countable sets is countable (for, you--.lan use
induction), let A and B be countable sets. If A or B is empty, then A x B is also empty. lj);refore
we can assume that A and B are both nonempty. By theorem 27.4.4; there exists surjections.

r.r: ~A and z: ~B· Now define h: Z+xZ+ ~AxB by

h(n, m )=(j(n) ,g(m))

for any n ,m « Z+ . You can easily verify that h is a surjection. Composing h with a bijection of Z+

onto z+ X Z+ , we get a surjection of Z+ onto A x B . Thus A x B is countable.

In the above, we have proved that any finite product of countable sets is countable. Regarding
infinite products, this is not the case. Infact a countably infinite product of finite sets (even two
elements) is unco~ntable. This is proved in the following. First let us agree to denote the set of all

mappings of a set A into a set X by XA.

27.4.7 Theorem : Let X={O,l}. Then XA::;:P(A) and hence there is no injection of XA into-Ll_.c,

for ~ny nonernpty set A ; in particular Xz+ is uncountable.

Proof: Let A be any non empty set. For any B ~ A , let us define

ifaEB
if a ~ B'

'l' B is called the membership function or characteristic function of B on A . You can easily check
·s ~
that B H '¥ &cis a bijection of the set JPl(A ) of all subsets of A onto th'e set X A . Thus X A ::;:JlD( A ).

A ) into A ' Therefore, there is no injection of

By theorem 27.3.6, there is no injection of JP>( XA into

A. In particular, there is no injection of XZ+ into Z+. Thus XZ+ is uncountable (by theorem
27.4.4).-----'

In the above theorem X Z+ is a countable product at copies-of the two element set X = {o.i] .
-' . . Z+
-For-anytwoelernent set X, X is uncountable:

2.7.4.8 c6rollary: Any intei'va~re than one -e1Em,~t, in the real number system is
uncountable. '-, .
 Centre for Distance Education 12 Acharya Nagarjuna University

Proof: Let X = {O, I} and S be the set of all real numbers in [0, 1] in whose decimal expansion!

are only D's and 1's occure. By theorem 27.4.8, S is uncountable and hence so is [O,Jl
I
(by

27.4.6(1)). For any real numbers a and h with a < h , [0,1] is bijective with [a, h] (for, consi\d~r the

bijection f:[O,q~[a,h]'given by f(x)=a+(h-a)x) and therefore [a,b] is uncountable and

hence so is every set containing [a, b] . Now, let I be an interval such that I contains atleast two

elements. Choose a, b e I such that a < h . Then, I being an interval, [a, h] c I and therefore I
is uncountable.
As a consequence of the above, it follows that the set lR of all real numbers and the set
lR - Q of irrational numbers are uncountable (since Q is countable).
Let us recall that a finite set cann't be bijective with a proper subset of itself (see corollary
27.4.2(2)) and therefore, if a set X is bijective with one of its proper subsets; then X is infinite.
Further, if X contains a countably infinite set, then X has to be infinite. Now, we shall prove that
either of these properties are necessary also for a set to be infinite.

27.4.9 Theorem: The following cr~ equivalent for any set X .

(1) X is infinite

(2) There exists an injection f: z: ~ X

(3) There exists a bijection of X with a proper subset of itself.

Proof: (1) => (2): suppse X is infinite. Then, for any finite subset A of X , X - A is a nonempty
set. Now, we shall define a function f:
.. no
z+ ~ X inductively as follows. Choose an arbitrary element
of X and call it f (1) . Let n> 1 and assume that the element, f (1)..,f (2), ...., f (n -1) are choosen

in X. Then. X -{f(1),f(2), .....:,f(n-l)} is a nonempty set and hence 'fe can choose an
element .

f(n) EX -{f(1),f(2), ,f(n~1)}.

I~Jt)is process, we have a function f: Z+ -bX such that f (n) "* 1M for any m < n in

Z+ . This f is clearly an injection.

(2) => (3} ~:_Suppose f: Z+ ~ X is an injection. Put Y =X - {f (I)}. Then Yis a proper subset

of X We shall prove that X is bijective witl) Y . Defining g: X ~ Y by

 · If(n+l) if x= f(n)forsomenE Z+
g(x):::lx
if x;t: f (x ) for all n E 71.'+

You can easily check that g is a bijection of X onto y

(3) => (1) : This is a consequence of corollary 27.4.2(2).

The choosing of elements f (1), f (2), .... in the proof (1) => (2) above, is not an acceptable
induction formula. Earlier, in proving (3) => (1) of theorem 27.4.4, we have defined a function

·a: 7l* ~ A making use of the induction. But, there a (n) is defined uniquely in terms of the restriction

of a to {I, 2, ....) n -I} . But f (n) is not unique here. Therefore) the definition of f involves infinitely
many arbitrary choices. We shall have a brief discussion of this situation in the next section.

27.5 AXIOM OF CHOICE

Ti.1Inow we have come across certain definite allowable methods for specifying sets. We
can define aset I:lY listing its elements, or by taking a given X and specifying a subset y of X by
giving a property that the elements of yare to satisfy. We can giving a property that the elernets of
yare to satify. We can define a set by taking unions or intersections of members of a given
collection of sets or by taking the difference of two sets or by taking the set of all subsets of a given
set or by taking the Cartesian products of sets. Th~ function f defined in the proof of (1) => (2) of

27.4.9 must be a well-defined subset of Z+ x X, constructed using the above allowed methods of
forming sets. The methods given above are not adequate1~r this purpose. We need a new method
of assertaining the existance of such a function. Theref3re, we add the following new methods of
forming sets.
27.5.1 Axiom of Choice: Given a collection ci7 of pair-wise disjoint nonempty sets, there exists a
set C having exactly one element with each member of G9"; that is, for each S E ci7~_theset S nC
contains a simple element.

The axiom of choice asserts the existance of a set that can be thought of as having been
obtained by choosing one element from each member of the collection ci7. This seems to be a
trivial assertion and, infact, most mathematicians today accept it as part of the set theory on which
they base their mathematical arguements. But, in the past, a good deal of contraversy raged around
this particular assertion concerning set theory, for the simple reason that there are certain theorems
one can prove with the aid of the axiom of choice that some rnathernaticians were reluctant to
accept. For example, the well-ordering theorem which we shall discuss in the next section.

The following is an equivalent form of the axiom of choice and, in this form, it is morehandy]
~ Centre for Distance Education 14 t.charya Nagarjuna University B
and useful onseveral occations, in particular, in the proof of (1) => ( 2) of the theorem 27.4.9.

27.5.2 Theorem (existance of a choice function) :

\

Let B be a collection of nonernpty sets (not necessirily pairwise disjoint). Then there exists
a function

//
such that C (B) is an element of B , for each B E lffi.

Proof: For each BE B, let us define a set by

{B}xB = {(B,X)jXEB}.

Let rj)9= { {B} x B / B E B }. You can easily check that, for any B 7: C E B,

=({B} xB)n({C}xc)=¢
and hence rj)9is a collection of pairwise disjointnonempty sets. By the axiom' of choice
(27.5.1), there exists a s~t W such that W..~( {B} x B) has exactly one element, for each BE lffi,

that is, for each BE lffi,there exists uniqueelement xB EB such that

I.'

Now, the set C =.{(B, XB)j BE B} i~ a subset of for which (B, XB) E C.
eln.
Bx:dJ B
B.El]

such that, for each B E B, there is a unique x BE, U B

BElffi

Therefore, C: lffi~ U B is a function, and obviously C (B) = xB EB for any BE lffi

BElffi '

The function C in the above is called a choice function for the collection lffi.In the light of the
above theorem, let us reformulate the proof of (1) => (2) of theorem 27.4.9.
 ( Algebra E
We are given that X is an infinite set and we have to obtain an injection f: z: ~ X. Let
lffi be the class of all nonempty subsets of X. Then, by theorem 27.5.2, there exists a choice
function.

such that C (B) is an element of B , for each BE lffi . Now, we shall define f: z' ~ X inductively
as follows.

f(l)=C(X)

f(n)=C(X -{f(l), ,f(n-l)}) for n >1

Then clearly f is an injection.

The above proof, based on the axiom of choice or the existence of a choice function, is a
logically correct one. 'One must make specific use of a choice function in such proofs. But in
practice most mathematicians do no such thing. They go on with no qualms giving proofs, as we
did for 27.4.9, proofs that involve an infinite number of arbitrary choices. They know that they are
really using the axiom of choice; and they know that if it were necessary, they could put their proofs
into a logically more satisfactory form by introducing a choice function specifically. But usually they
do not bother and neither were we in the main text of-this book.

27.6 WELL - ORDERING THEOREM :

<,
In this section, we present a theorem, which is equivalent to the axiom of choice and is
proved by Zermelo a century ago and which startled the mathematical world. There was a
considerable debate as to the correctness of the proof. When the proof was analyzed closely, the
only point at which it was found that there might be some question was a construction involvinq an
infinite number of arbitrary choices, that is, a construction involving the axiom of choice. But present
day mathematicians accept the axiom of choice as a reasonable assumption about set theory and
they accept this theorem of Zermelo. The proof of this theorem is considerably involved and we
present only the statement of the theorem. Let us begin with the followino,

27.6. V. Definition: Let X be a nonempty set. An ordering on X is defined to be a binary relation

R on X (that is, R c XxX) such that R is reflexive, transitive, antisymmetric and

RUKI'=XxX.

It is a habit to denote an ordering by the symbol :s; (and read it as 'less than or equal to') and
 Centre for Distance Education 16 ~charya Nagarjuna University

as usual VJe write a ~ b to mean (a, b) E~ . With this notation, a binary relation ~ on a set X is
called an Or~ing can X if it satisfies the following conditions for any a, b, C EX.

(1) ~a
\
\
(2) a ~ b -and b ~ c ='> a ~ c
\
(3) a z.b and b~a=,>a=b

(4) as; b or b sa
We frequently write a < b to mean a ~ b and a i b . Also we write a ~ b to mean b ~ a ; that is
?:=~-l . It is obvious that ~ is also an ordering if ~ is an ordering.

27.6.2 Definition: A pair (X,~) is called an ordered set if X is a nonempty set and:::; is an
ordering on X .

111ebest and familiar example of an ordered set is the set lR of the real numbers together
with the usual ordering defined by a s; b<;:::.> b - a is nonnegitive. If (X,:s;) is an ordered set and y

is a nonempty subset of X, then (y, ~y ) is also an ordered set where ~y is the restriction of <

to y; that is, :S;y = ~ n(y x Y). Therefore, simply, we can state that any nonempty subset of an
ordered can be treated as an ordered set.

27.6.3 Definition: Let (X,:s;) be an ordered set and A a non empty subset of X . An element
ao of A is called the least element of A if aO ~ a for all a EA .
. DnG .. . .
If a least element of A exists, then clearly it is unique. For example, in (JR.,:s;) , the interval
(0,1) has no least element while the interval [0,1] has least element, namely O. Similarly, the set of
positive real numbers has no least element while the set of non-negitive real numbers has least
element.' ;

27.6.4 Definition: An ordering s on a set X is called a well-ordeiing:f every nonempty subset of

X has a Ieast element. 11set together with a welt-ordering on it is called a well - ordered set.

With respect to the usual ordering, the set z'

of positive integers is a well-ordered set
(see 27.4.5). Also, on any finite set, every ordering is a well-ordering (prove this). The usual ordering
on the set.)R of real numbers· is not a well-ordering. In the following, there is a method of constructing
now well-ordered sets from the given ones.

27.6.5 Exercise: Let (X,:s;) and (Y,:s;) be well-ordered sets. Let us define an ordering :s; on the
 ( Algebra ~

Cartesian product X x Y as follows.

(XI'YI)~(X2'Y2)~ either xl<x2 or(xl=x2 and YI ~Y2)' Prove that ~ is a well-ordering

on Xxy.
The above ord~ring is called the dictionary ordering or lexicographic ordering. In general, if
Xl, X 2, , X n are well-ordered sets and X = Xl xX 2x x X n then we can define a well-
i

ordering on X as follows; For any X=(Xl>X2,·· .. ·,xn) and Y=(Yl,Y2, .... 'Yn) in X.

x ~ Y~ either X= y orthere exists r , 1~r ~ n, such that xi = Yi Ifor all i < rand x; <Yr.

If (X, ~) is a well-ordered set and y is any set which is bijective with X , then we can
define a well-ordering on y; for, consider a bijection t y ~ X and define, for any Yl> Y2 E Y,

you can easily verify that this is a well-ordering on y. Now, the usual ordering on Z+ is a well-

ordering and, since' Z+ x Z+ is bijective with Z+, we have a well-ordering on Z+ x Z+ which is

also a well-ordering. But these two wetl-ordsnnqs on' z+ xZ+ are different. If /:Z+ ~Z+ xZ+ is

a bijection then the elements in Z+ x Z+ can be expressed as

/(1)</ (2)</(3)<.: .....</ (n )< .....

where there is no other element in between /

, .
(n) .elo
and / (n + 1) .
r

If we look at the dictionary ordering Z+ x Z+ , then the elements of Z+ x Z+ can be expressed

as

(1,1)«1,2) < « 2,1)«2,2) < ~ (~,1)< (3,2) «3, 3) < « 4,1)< .

'In the former well-ordering , each element.except the least element f (1) , has an immediate

predecessor, whi'a in tile later well-ordering, all elements of the form' ('n;!) have noimmedi~te
. .';.'.

predecessor. Thus there are two distinct well-orderings 'on Z+; one Is.the usal order, while the

other is obtained by a biiection of Z+ x Z+ onto' Z+ and by the dictionary ordering on Z+ x Z+ .

Infact, there are infinitely many distinctwell-orderings on Z+ , since (Z+ r is bijective with Z+ anc
 -- ----- -- --- --- - . --------~-.-- -

( Algebra :s
A ={a E X/there is n with an =Zandcz, = l for alIi :;t: n}.
A consider of a" elements a where exactly one coordinate of a is equal to 2 and all the
others are equal to 1. This set A has no least element (If a E A is such that an = 2 , then the
element b e A for which b., + 1 = 2 is similar than a).

rhus the dictionary ordering on the above set X is not a well-ordering. Then, is there some
other ordering on X which is a well-ordering? No one has ever constructed a specific well-
ordering on an uncountable set. So is the case with the real number system IR . The usual ordering
on IR is not a well-ordering. However, the following celebrated theorem of Zermelo says that such
a well-ordering exists on ]]{ .

27.6.8 Theorem (Zermelo's well-ordering theorem) : On any given non-empty set, there exists
a well-ordering.
The lack of any constructive procedure for well-orderings on an arbitrary uncountable set
led to a considerable debate as to the correctness of the proof of Zermelo's theorem which asserts
the existence of a well-ordering. When the proof was closely scrutnised, it wa,s found that the only
unacceptable point is the use of infinitely many arbitrary choices, that is, a construction involving
the axiom of choice. For this reason, some mathematicians rejected the axiom of choice, which
looks like an innocent and simple axiom, and for many years a legitimate question about a new
theorem was: Does its proof involved the axiom of choice or not? A theorem was considered to be.
on some what shaky ground if one had to use the axiom of choice in its proof. The present-day
mathematicians, by and large, do not have such qualms. They accept the axiom of choice as a
reasonable assumption about set theory and included it in the list of axioms of set theory and they
accept the well-ordering theorem along with it.

Further it is proved that the axiom of choice is implied by the well-ordering theorem. In
otherwords, the well-ordering theorem and the axiom of choice are equivalent to each other.

27.7 ZORN'S LEMMA :

Zorn's lemma is yet another equivalent form-of the axiom of choice and of the well-ordering
theorem discussed in the previous section. Let us first have the following concept which is weaker
than that of a well-ordering.

27.7.1 Definition: Let X be a nonernpty set. A reflexive, transitive and antisymmetric binary
relation on X is called a partial order on X That is, a subset e of X x Xis called a partial order
_on X if the following are satisfied for any a, b, C EX.

(1) a s: a
(2) a :::;;
band b :::;;
C -=> a :::;;
C
 Centre for Distance Education 18 Acharya Nagarjuna University

we have the dictionary ordering on ( z+ r

27.6.6 Definition: Two ordered sets (X,::::;:) and (Y,~) are said to be order isomorphic ifthere is

a bijection j: X ~ Y such that, for any a, hEX,

a::::;:h¢::;>j(a)::::;:j(b)

Such a mapping f is called an order isomorphisrr r.

Two orderings ::::;:

and ::::;:'on a set X are said to be equivalent if the ordered sets (X,::::;:)
and (X, ::::;:')
are order isomorphic; that is, there is a bijection j of X onto itself such that

a s; b ¢::;> j (a)::::;:' j (b) for any a, hEX.

27.6.7 : Prove that any two finite ordered sets with equal number of elements are isomorphic.

In other words, any two well-orderings on a finite set are equivalent. But the discussion prior

to 'definition 27.6.6 says that, on the countably infinite set Z+ , there are infinitely many distinct (not
equivalent) well-orderings. It is natural to ask whether one can find a well-ordering on an uncountable
set.

z+
Let us consider the uncountable set (z+) which is the countably infinite product.

X=z+ X Z+ x '.: =(z+YZ+

we can think of X as the set of all mappings of Z+ into itself or as the set of all 'sequences'
of positive integers. We can generalise the dictionary order to X , in a natural way, as given below
:Forany

in X, we shall define
... '
a ::::;:
b¢::;>·either a =b or there exists. n such that aj = hi for i < n and an < bn. .

You \can easily verify that this ::::;:

is an ordering on X . But this is not a well-ordering; for,
consider the set
 ~~c~e~nt~re~fo~rED~is~ta~n~ce~E~du~c~at~io~n~~~~~~~20ts~~~~~~~~A~ch~a~ry~a~N~a~g~ar~jU~n~a~U~n~iv~er~s~

(3) a s; band b S; a ~ a = b

Any ordering on X is a partial order bnX,. For any a, b.e Z+ , if we define as; b to mean

a divides b (that is, ac = b for some c E Z+), then ~ is a partial order on Z+ that is not an ordering;

for 2 i 3 and 3 i 2 under this partial order. This p~rtial order on Z+ is called the division order.

27.7.2 Definition: A nonempty set X together with a partial ordering on X is called a partially
ordered set. Let A c;:;;: X and x EX· Then x is called an upper bound (lower bound) of A if a ~ x
(x ~ a) for all a EA. xis called a maximal (minimal) element of X if there is no Y E X such-that
x ~ Y (Y < x) and x 7:- y. A subset A of X is'called a chain in X if,for any a, b E A ,either a ~ b
or b S; a.
The following result, known as Zorn's lemma, can be dedved from the- axiom of choice and
vice-versa. We are not going to gi!ve the proofs here for obvious reasons.

27.7.3 Zorn's Lemma: Let (X,~) be a partially orderedsetsuch that every chain in X has an
upper bound in X . The X has a maximal element.

If ~ is a partial order on X , then the inverse relation S;-l = {(a, b) E X X X / b S; a} is also a

partial order on X and is called the dual order for ~. For convenience, we write ~ for ~-l. The
, following is another form of Zorn's lemma.

27.7.4 Theorem: Let (X,~) be a partially ordered set such that every chain in X has a lower
bound in X . Then X has a minimal element.

On any set d1'of sets we can define a partial order by A ~ B <=> A c B . This partial order is
called the division order. Frequently Zorn's lemma is applied in the form given below. A collection d1'
of sets is said to be closed under unions of chains if, for any chain ~ in cf7, cf7 contains the union
of members of ~.

27.7.5 Theorem: Let X be a set and cf7 a set of subsets of X. Suppose cf7 is closed under
unions of. chains. Then d1' has a maximal member.

We shall exhibit two applications of the above theorem. The first one is infact another
equivalent statement for the Zorn's lemma.

27.7.6 Theorem (Hausdorff Maximal Principle) : Let (X,~) be a partially ordered and A be a

chain in X . Then A is contained ina maximal chain. That is, there exists a chain M such that
A c;:;;: M and M is not properly contained in any chain in X .

Proof: Let cf7 be the collection of all chains in X which contain A. Then cf7 is closed under
 ( Algebra ~

unions of chains (for, the union of a chain- of chains is..' Ciigaina chain) and hence ril' has a maximal
. . .

member which is a maximal chain cqntaining A.

27.7.7 Theorem: Let R be a finitely ~en,~rated,rjng. Then every proper ideal of R is contained in
a maximal ideal of R .

Proof: Let R be generated by ~ finite set S. Then an ideal I of R is proper if and only if S q;J .
". "" .
",

You can easily prove that the union of a chain of proper ideals is again a proper ideal (use the
','

finiteness of S). The remaining part of the proof follows again from 27.7.5.

The following is a summary,in a nutshell, of our discussion made in the present section as
well as the earlier two sections.
27.7.8 Theorem: Thefolfowlnqprcpositions.are equivalent to each other

(1) The axiom of choice

(2) The well-ordering theorem

(3) The Zorn's lemma

(4) The Hausdorff's maximalprinciple ,

Lesson Writer
Prof: U.M. Swamy
Andhra University,
Visakhapatnam.