You are on page 1of 66

FINITE ABELIAN GROUPS

by
AGAYTHA HOPE REED, B.A.
A THESIS
IN
MATHEMATICS
Submitted to the Graduate Faculty
of Texas Tech University in
Partial Fulfillment of
the Requirements for
the Degree of
MASTER OF SCIENCE

Approved

Accepted

August, 1992
ACKNOWLEDGMENTS

I am most grateful to Professor Harold R. Bennett for his patience,

encouragement, and critique during the preparation of this thesis. I would also like

to extend my appreciation to Professors J. Dalton Tarwater and Robert E. Byerly

for their assistance.

..
11
TABLE OF CONTENTS

ACKNOWLEDGMENTS 11

CHAPTER

I. INTRODUCTION 1

II. PRELIMINARY DEFINITIONS AND THEOREMS 4

III. BASIC RESULTS ON FINITE ABELIAN GROUPS 15

IV. THE FUNDAMENTAL THEOREM OF FINITE ABELIAN GROUPS 46

V. INVARIANTS OF FINITE ABELIAN GROUPS 56

REFERENCES 63

lll
CHAPTER I

INTRODUCTION

The roots of finite abelian group theory are grounded primarily in number

theory and in the theory of quadratic forms. The emergence of a theory of finite

abelian groups was first discernible in the late eighteenth and early nineteenth

centuries in the arguments of Euler, Lagrange, and Gauss. Indeed, many of their

results, though achieved while attacking very specific problems, eventually became

part of the foundation upon which the theory grew. Most notably, in his

Disquisitiones Arithmeticae of 1801, C. F. Gauss established several important

properties of finite abelian groups. Although Gauss did not use the terminology of

groups and did not unify his results, but instead treated them as separate cases,

his examination of the integers modulo m with addition, the integers relatively

prime to m modulo m with multiplication, the equivalence classes of binary

quadratic forms, and the nth roots of unity, exploited their abelian group

characteristics. Another milestone in the development of the theory was N. H.

Abel's discovery in the 1820's that algebraic equations are solvable by radicals if

the permutations of their roots commute. This result revealed and underscored

the impact of the property of commutativity; hence Abel's name has since been

attached to the property. E. Schering's study of arithmetic forms resulted, in 1868,

in a decomposition similar to that of a finite abelian group into cyclic summands.

In 1870, L. Kronecker authored a paper on algebraic number theory in which he

1
implicitly defined a finite abelian group and also discussed several associated

conclusions which closely resemble well-known results in modern finite abelian

group theory. Among the results derived from Kronecker's implicit definition is a

result which can be interpreted as a basis theorem for finite abelian groups. As

algebraic number theory progressed through the nineteenth century, numerous

examples of finite abelian groups surfaced. The theory continued to develop in this

piecemeal, largely implicit fashion, without a palpable effort to merge the various

results into a cohesive, independent theory, until 1878 when G. Frobenius and L.

Stickelberger first attempted to classify finite abelian groups and to exhibit finite

abelian group theory as an explicit, independent theory built upon its own

foundation. In more than a century that has intervened since finite abelian group

theory emerged in its own right, it has undergone the expected metamorphosis

and has extended its influence into the other branches of mathematics [5], [10].

Our purpose in this examination of finite abelian groups is to accumulate in

one work the basic theory of finite abelian groups, to illuminate the features of

finite abelian groups which make them of special interest within the more general,

more complex context of groups, and to present numerous examples of finite

abelian groups in this process. Furthermore, we will investigate the impact that

the properties of finiteness and commutativity have on a group's structure,

especially how these properties, in a sense, make finite abelian groups more

manageable than groups at large.

2
We begin our pursuit of these ends in Chapter II with a delineation of the

vocabulary and concepts which are prerequisite to our discussion of finite abelian

groups. The results in Chapter II are offered without proof as the proofs are

readily available and would constitute an unnecessary digression from our central

purpose. Chapter III consists of the basic results on finite abelian groups, with a

view to highlighting general group concepts which are simplified by the properties

of finiteness and commutativity. In addition, Chapter III contains a variety of

examples of finite abelian groups. The focus of Chapter IV is on a discussion and

proof of the pivotal and decisive Fundamental Theorem of Finite Abelian Groups,

a theorem which guarantees a complete determination of all finite abelian groups

up to isomorphism and a decomposition of any finite abelian group into a sum of

indecomposable components. Finally, in Chapter V we consider an invariant of

finite abelian groups which arises as a natural result of the Fundamental Theorem.

3
CHAPTER II

PRELIMINARY DEFINITIONS AND THEOREMS

Before launching into our discussion of finite abelian groups, it will be helpful

to recall several definitions and theorems from group theory, beginning with the

fundamentals. When appropriate, some results which are traditionally stated

separately have been merged for the sake of succinctness. Also, note that Z will

be used throughout this discussion to represent the integers.

Definition ILl A group is a nonempty set G along with a binary operation *


which satisfy the following conditions:

{i) Closure. If a, b E G, then a* b E G.

{ii) Associative law. If a, b, c E G, then (a* b)* c =a* (b *c).

{iii) Existence of an identity element. There exists e E G such that

a * e = e * a = a for all a E G.

{iv) Existence of inverses. If a E G, then there exists a- 1 E G such that

a* a- 1 = a- 1 *a = e.
Definition 1!.2 A group G which satisfies the commutative law, i.e., for all

a, bE G, a* b = b *a, is called an abelian group.

Notice that in identifying a group it is necessary to specify both a set and an

operation. Except when the operation is not apparent from the context, we will

represent a group using its set name. Furthermore, we will suppress the use of"*"

to indicate the group operation in favor of juxtaposition notation, unless the

context dictates the use of a particular operation symbol.

4
Definition 11.3 The number of elements contained in a group G is called the

order of G and is denoted by IG I· G is a finite group if IG I is finite.

Some properties of groups which are readily derived from the preceding

definition are enumerated in the following theorem.

Theorem 11.1 Let G be a group and let a, b, c E G.

{i} The identity element e of G is unique.

{ii) Every a E G has a unique inverse in G.

{iii) If ab = ac, then b = c {left cancellation).

{iv) If ba = ca, then b = c (right cancellation).

{v) The equations ax = b and ya = b have unique solutions for x andy in

.
G , z.e., x =a - 1 b an d y = ba - 1 , . ly.
respectwe

(vi) ( a- 1 t 1
= a.

{viii} aman = am+n for all integers m, n.

(ix) (am t = amn for all integers m, n.

Definition 11.4 If G is a finite group with a E G, then the least positive integer m

for which am = e is called the order of a. We denote the order of an element a as

o(a).

Theorem 11.2 Let a be an element of a group G such that o( a) = n and let t be

an integer. Then at = e if and only if n divides t.

Definition 11.5 A nonempty subset H of a group G is a subgroup of G if H is

itself a group with respect to the binary operation of G.

5
Obviously, every group G has two trivial subgroups; namely, the group G itself

and { e }. Moreover, the condition of being a subgroup is transitive: If His a

subgroup of G and K is a subgroup of H, then K is a subgroup of G.

The next theorem establishes the criteria which must be satisfied in order for

H to be a subgroup of G.

Theorem 11.3 A nonempty subset H of a group G with the operation * is a

subgroup of G if and only if

{i} H is closed; i.e., if a, bE H, then a* bE H; and

{ii) H contains the inverses of all of its elements; i.e., if a E H, then

a- 1 E H.

Since our scope is limited to finite abelian groups, it is useful to note that in

the case of finite groups, we need only verify closure for subgroups.

Theorem 11.4 A nonempty subset H of a finite group G is a subgroup of G if and

only if H is closed with respect to the binary operation on G.

Theorem 11.5 If {Hi}iEJ, where I is a nonempty indexing set, is a family of

subgroups of the group G, then niE/ Hi is a subgroup of G. In fact, niE/ Hi is a

subgroup of each Hi.

Definition 11.6 If H and K are subgroups of G, define

HK = {x E G: x = hk,h E H,k E K}.

Theorem 11.6 Let H and K be subgroups of G, then H K is a subgroup of G if

and only if HK = KH.

6
1 Zg~\·
Theorem II. 7 Let H and K be finite subgroups of G. Then IHKl = 1

Definition II. 7 Let S be a subset of a group G. The smallest subgroup of G

which contains S is called the subgroup generated by S and is denoted <S>.

Theorem II.8 If S is a nonempty subset of a group G, then <S> consists of all

possible products of powers of the elements of S.

Of special interest in our present investigation is the case when S consists of a

single element that generates the entire group G.

Corollary II.1 If g E G, then <{g}>= {gn : n E Z}.

For the remainder of our discussion we will eliminate the set notation and

replace <{g} > with the less clumsy <g>.


Definition II.8 <g> is called the cyclic subgroup of G generated by the element

g. If G =<g>, we say that G is a cyclic group.

Definition II. 9 Let G be a group with subgroup H and g E G. Then

Hg = {hg: hE H} is called a right coset of H in G. Accordingly,

gH = {gh : h E H} is a left coset. An element in a coset of H is called a coset

representative.

Clearly, if G is an abelian group, H g = gH for all subgroups H of G; hence, in

this context we may drop the qualifiers "right" and "left."

Theorem II.9 If H is a subgroup of G with a, bE G, then H a= Hb if and only

if ab- 1 E H.

Theorem II.10 If H is a subgroup of G, then the set of right (or left) cosets of H

in G form a partition of G.

7
Definition 11.10 The number of right cosets of a subgroup H in a group G is

called the index of H in G and is denoted [G : H]. If G is a finite group, then

[G: H] = mt·
The following well-known theorems and corollaries figure prominently in the

development to follow.

Theorem 11.11 (Lagrange's theorem) Let G be a finite group with a subgroup

H. Then IHI is a divisor of IGI.

Corollary 11.2 Let G be a finite group with g E G. Then o(g) is a divisor of IGI.

Corollary 11.3 Let G be a finite group with g E G. Then giGI = e.

Theorem 11.12 (Cauchy's theorem) Let G be a finite group and let p be a

prime number. If p is a divisor of IGI, then G contains an element of order p.

Definition 11.11 A subgroup N of a group G is called a normal subgroup of G if

g- 1 ng E N for every g E G and n E N. To denote that N is a normal subgroup of

G we write N <l G.

Theorem 11.13 Let N be a subgroup of G. The following are equivalent:

(i) N <l G.

(ii} g- 1 Ng =N for all g E G.

{iii} Ng = gN for all g E G.

(iv} g- 1 ng EN for all g E G and n E N.

(v) (N g)(N g') = N gg' for all g, g' E G.

8
Although (iii) by no means implies that ng = gn for all g E G and n E N, if G

is an abelian group this certainly is the case. Hence, for any subgroup N of an

abelian group G, N g = gN, so that every subgroup of an abelian group is a

normal subgroup.

The concept of mappings is central to the study of algebra. In this study we

will be most interested in mappings between groups which preserve the group

structure. In that regard, we now turn our consideration toward group

homomorphisms.

Definition 11.12 Let G and H be groups with operations * and*, respectively. A

mapping f : G ~ H is called a homomorphism from G into H if

f(z * y) = f(z) * f(y) for all z, y E G. {In general, we will suppress "*"and''*"

and simply write "f( zy) = f( z )f(y) ", where the operation on the left is

understood to be that of G and on the right to be that of H.)

Definition 11.13 A homomorphism f: G ~ H is called a monomorphism iff

is a one-to-one mapping. Iff maps G onto H, then it is called an epimorphism. f

is called an isomorphism if it is both one-to-one and onto. If there exists an

isomorphism between two groups G and H, we say that G is isomorphic to H and

write G ~ H. Also, a homomorphism which maps a group G to itself is called an

endomorphism.

As we progress in our discussion, the significance of isomorphisms of groups

will become abundantly clear.

9
Definition !1.14 Let f: G -----? H be a homomorphism. The kernel off, denoted

kerf, consists of all elements x E G such that f( x) = eH, where eH is the identity

element of H. Alternatively, kerf= f- 1 (eH)· Furthermore, the image off,

denoted Imf, consists of all elements f(x) E H such that x E G.

Theorem !1.14 Let f: G-----? H be a homomorphism.

{i) f is a monomorphism if and only if kerf= ec.

{ii) f is an epimorphism if and only if Imf = H.

{iii) f is an isomorphism if and only if kerf = {ec} and Imf = H.


Theorem !1.15 Let f : G -----? H be a homomorphism. Let ec and eH represent

the identity elements of G and H, respectively. Then

{i) f(ec) = eH.

{ii) f(x- 1 ) = f(xt 1 for all x E G.

{iii) f(xk) = f(x)k for all x E G, k E Z.

(iv) f sends subgroups of G to subgroups of H. In particular,

f( G) = Imf is a subgroup of H.

{v) f- 1 sends subgroups of H to subgroups of G.

{vi) Iff is one-to-one, G ~ f( G).

{vii) f(x1x2 · · · Xn) = f(xt)f(x2) · · · f(xn) for all Xi E G, 1 < i < n.

(viii) If g : H -----? K is a homomorphism, then go f : G -----? K is a

homomorphism.

(ix) Iff : G -----? H is an isomorphism, then f- 1 : H -----? G is an

isomorphism.

10
{x) kerf <1 G.
Theorem 11.16 Let G be a group with subgroup N such that N <1 G. The

collection of distinct cosets of N in G forms a group called the quotient group of G

by N and is represented by GIN.

Suppose that N <1 G and a, bE G. Then Na, Nb E GIN. The operation on

GIN is defined, then, in the fashion suggested by Theorem 11.13( v),

(Na)(Nb) =Nab; i.e., right cosets combine to form right cosets. To clarify,

suppose that G is the group Z 9 = {0, 1, 2, 3, 4, 5, 6, 7, 8} with addition modulo 9 as

the operation and N = {0,3,6}. Then GIN= {N,N + 1,N + 2}. So, for

example, (N + 1) + (N + 2) = N + 3 = N, since 3 EN and

(N + 2) + (N + 2) = N +4= N + 1 since 4 EN+ 1.

Theorem 11.17 Let G be a group with N <1 G. Define a mapping "' : G -----+ GIN

by "!( x) = N x for all x E G. Then "' is a homomorphism of G onto GIN. "' zs

often called the natural homomorphism.

Since much of what is to follow involves isomorphisms, it will be beneficial to

note several group characteristics which are preserved under isomorphisms.

Theorem 11.18 Let G and H be groups such that G ~H. Then

{i) IGI = IHI.


(ii) If G is an abelian group, H is an abelian group.

(iii) If G is a cyclic group, H is a cyclic group.

(iv) If G contains an element of order n, H contains an element of order

n.

11
(v) If every element of G is its own inverse, every element of H is its

own ?.nverse.

(vi) If every element of G has finite order, every element of H has finite

order.

(vii) If 1/; : G ~ H is an isomorphism, then for all x E G, o(x) =


o('f/;(x)).

(viii) If K is a group such that H ::::: K, then G ::::: K.

The next three theorems, the Isomorphism Theorems, clarify the

interrelationships among quotient groups, normal subgroups, and homomorphisms.

While our interest herein is restricted to groups, it is worthwhile to note that

analogs of these theorems hold for every type of algebraic system [14].

Theorem II.19 (First Isomorphism theorem) Let G and H be groups with

f :G ~ H a homomorphism with kernel K. Then K <l G and GI K ::::: Imf.

Corollary II.4 If Imf = H, i.e., f is an epimorphism, then G I K ::::: H.

If we let ¢> represent the isomorphism of G I K to Im f, we see that any

homomorphism f: G ~ H can be decomposed into the composition of three

homomorphisms: (1) the natural homomorphism, 1}, (2) ¢, and (3) an inclusion:

G~ GIK ~Im f ~H.

Theorem II.20 (Second Isomorphism theorem) Let G be a group with

subgroups M, N such that N C M C G and M, N <l G. Then MIN <l GIN and

GIM ~ (GIN)I(MIN).
12
Theorem 11.21 (Third Isomorphism theorem) Let H, N be subgroups of G

with N <J G. Then H N is a subgroup of G, (H n N) <J H, and Hj(H n N) ~ H NjN.

Definition 11.15 An isomorphism f from G onto itself is called an

automorphism.

Of course, automorphisms inherit all the properties of isomorphisms. In

addition, we have the following result which will resurface in the examples of

Chapter III.

Theorem 11.22 If G is a group, then the collection of all automorphisms of G,

denoted Aut( G), is a group with composition of mappings as its operation.

We conclude our compendium of results with a thumbnail sketch of Sylow

theory. This theory is most significant for its contribution to the classification of

finite groups.

Definition 11.16 Let G be a group and pn the highest power of a prime p that

divides jGj. If Sis a subgroup ofG of order pn, we say that Sis a Sylow

p-subgroup of G.

Definition 11.17 Let S and T be subgroups of G. S and T are said to be

conjugate provided that there is an element g E G such that gS g- 1 = T.

Theorem 11.23 (Sylow's theorem) Let G be a finite group, p a prime number,

and k, m, n nonnegative integers.

(i) If pm divides jGj, but pm+I does not, then G has a subgroup of order

pm. Moreover, G contains subgroups of order pn, 0 < n < m, each of

which is contained in at least one Sylow p-subgroup.

13
{ii) Any two Sylow p-subgroups are conjugate.

{iii) For a given prime p, the number of Sylow p-subgroups in G is of the

form 1 + kp. Furthermore, 1 + kp is a factor of IGI.

14
CHAPTER III

BASIC RESULTS ON FINITE ABELIAN GROUPS

A finite abelian group is a nonempty finite set with a binary operation which

satisfies, in addition to the group axioms, the commutative law. Given an

arbitrary group, there is no reason to expect, if a and b are elements of the group,

that the symbols ab and ba should represent the same element. Consider, for

example, the group of 2 x 2 nonsingular matrices with entries from the integers

modulo p with the operation of matrix multiplication. Since matrix multiplication

is notoriously noncommutative, we would not, in general, expect the product AB

of two arbitrary matrices A and B to equal the product BA. For example,

I::], I:~1 I::]; I:~1 I::], I:~: ],


], while ],
we shall see, however, when the group at hand is endowed with the property of
As

commutativity, many of the notions of group theory are simplified. This

simplification admits a thorough investigation of finite abelian groups and makes

the complete classification of finite abelian groups possible.

In this chapter we will examine examples of finite abelian groups and several

fundamental results. Additionally, we will consider a few theorems from the

broader theory which are made more definitive in the finite abelian case. Finally,

we will observe several instances in which the restrictions "finite" and "abelian"

significantly simplify and, in some instances, trivialize, group concepts. In this

15
process we will lay the groundwork for our discussion of the Fundamental

Theorem of Finite Abelian Groups in Chapter IV.

Before forging ahead a brief comment about notation is in order. It is common

practice to use additive notation rather than multiplicative notation when working

in the context of finite abelian groups. When it seems natural in the context we

shall adopt this convention, otherwise we will use juxtaposition to represent the

combining of elements. Following is a table of analogous notations. For a group G

let a, bE G and let H and K be subgroups of G.


Multiplicative notation Additive notation

1, e 0, e

ab a+b
a-1 -a

ab- 1 a-b
an na

HK H+K

Ha H+a

HxK HGJK
As a starting point for our study, let us consider the most basic of finite

abelian groups, the cyclic groups. Recall from Chapter II that if a single element 9

generates the entire group G, we say that G is a cyclic group. So, if IGI = n and

G =<9>, then G consists entirely of "powers" of 9; i.e.,

G = {9 , 9 2,93 , ••• , 9n- 1 , 9n = e}. In additive terminology, G contains "multiples"

16
of g, G = {g, 2g, 3g, ... , (n- 1)g, ng = 0}. Clearly, then, if G =<g>, then

IGI = o(g).

Theorem 111.1 All cyclic groups are abelian.

PROOF: Let G =<g> be a cyclic group of order n. Let r, s be integers less than

n. We want to prove that

If we expand the left-hand side and apply the associative law, we have

r factors s factors

g·g···g·g·g···g
'---v-' '---v-'
s factors r factors

Example 111.1 For any positive integer n, the roots of the polynomial :z:n =1
form a cyclic group with the operation of ordinary multiplication of complex

numbers. Consider, in particular, the 4th roots of unity, {1, -1, i, -i}. Below is a

multiplication table, or Cayley table, for this group.

0 1 z -1 -z

1 1 z -1 -z

z z -1 -z 1

-1 -1 -z 1 z

-z -t 1 t -1

17
Inspection of this table will verify closure, the existence of inverses, the role of

1 as the identity element, and commutativity. Associativity follows from the fact

that multiplication of complex numbers is associative. Note that i 1 = i, i2 = -1,


3
i = -i, i4 = 1, and that ( -i) 1 = -i, ( -i) 2 = -1, ( -i) 3 = i, ( -i) 4 = 1, so that

this group is indeed cyclic and is generated by either of the elements i or -i.

The very name "cyclic group" is suggestive of geometric associations. Let 'ljJ

2
represent a rotation in the plane about a fixed point through an angle : and let e

represent the identity rotation. Then n successive applications of 1/; sweeps out a

complete circle; i.e., the collection of rotations {1/;,1/; 2, ... ,1/;n-I,'l/;n = e} is a cyclic

group generated by 1/;. Furthermore, if we superimpose this result upon the unit

circle in the complex plane, it is clear that {1/;, 1/; 2, ... , 1/;n-l, 1/;n = e} are the same

rotations which determine the nth roots of unity discussed in Example III.l.

Example 111.2 The group of roots of x 8 = 1 are exhibited in the Cayley table

below.

18
0 1 -1 't -1, v'Z -v'Z ivz -iv'Z
1 1 -1 't -1, v'Z -v'Z iv'Z -iv'Z
-1 -1 1 -1, 't -v'Z v'Z -iv'Z iv'Z
't 't -1, -1 1 iv'Z -iv'Z -v'Z v'Z
-1, -1, 't 1 -1 -iv'Z ivz v'Z -v'Z
v'Z v'Z -v'Z iv'Z -iv'Z 't -1, -1 1

-v'Z -v'Z v'Z -iv'Z ivz -1, 't 1 -1

iv'Z iv'Z -iv'Z -v'Z v'Z -1 1 -1, 't

-iv'Z -iv'Z iv'Z v'Z -v'Z 1 -1 't -1,

The entire group is generated by any of the following elements: v'Z, -vz, ivz,
-ivz. By Lagrange's theorem, the order of any subgroups of this group must

divide 8; therefore, any nontrivial subgroups of this group must have order 2 or 4.

Upon examination of the table it is clear that the cyclic groups {1, -1, i, -i} and

{1, -1} are just such subgroups.

Theorem 111.2 Let G =<g> be a cyclic group of order n.

(i) All subgroups of G are cyclic.

{ii) If k is an integer such that 0 < k < n, then <gk> is a subgroup of

order (k~n), where ( k, n) is the greatest common divisor of k and n.

{iii) To every divisor d of n there corresponds exactly one subgroup of

order d. This subgroup may be generated by g!fi.

PROOF:
19
{i) Let H be a subgroup of G. If H =<e>, then His obviously cyclic.

Suppose H =J.<e>. Since any element in His also in G, it can be

expressed as a "power" of g. Let m be the least integer such that

0 < m <nand gm E H. Now suppose for some integer t, that gt E H.

The division algorithm guarantees that there are integers q and r such

that

t= mq + r, 0 < r < m.

Now

so that

However, gt, gm E H; therefore, gr E H. Because 0 :S 1' < m and since

we chose m as the least positive integer such that gm E H, we are forced

to conclude that 1' = 0. Thus, gt = gmq and so gt E<gm>. We have

demonstrated that any element in H can be expressed as a power of gm;

hence, H =<gm> and His cyclic.

{ii) Let m = (k,n) and lets= o(gk). First, observe that (gk)~ = gn~ = e;

hence, by Theorem 11.2, s divides ; . On the other hand, since gks = e,

Theorem II.2 implies that n divides ks and so ~


m
divides !c_s.
m
Since

20
( ~,!) = 1, we have that .!!..
m
divides s. Therefore, since .!!..
m
divides s and s

divides .!!.. ,
m
we conclude that s = .!!.. = _n_
m (k,n)
and that I <gk> I = _n_.
(k,n)

(iii) Suppose d > 0 and d divides n. By (ii), o(g~) = (E.n ) = d since


d'n

(J, n) = J. Therefore, I <g ~ > I = d and G has at least one subgroup of

order d. Now assume that <gw> is another subgroup of order d. Then

gwd = e and by Theorem 11.2 n divides wd; as a consequence, J divides w


and so <gw> is a subgroup of <g~>. However, I <g~> I = I <gw> I = d;

hence, <g~>=<gw> so that the subgroup of order dis unique.

The following corollary to Lagrange's theorem considers the implications of

that theorem when the order of the group is a prime number.

Corollary 111.1 Let G be a group and let IGl = p, where p is a prime number.

Then G is cyclic and its only subgroups are { e} and G.

PROOF: The only positive divisors of p are 1 and p, so by Lagrange's theorem

the only subgroups of G are { e} and G. Now if g E G, then <g> is a subgroup of

G. If g-=/= e, then <g>-=/= { e }, hence <g>= G and G is cyclic.

Definition 111.1 A group G is said to be a simple group if it has no nontrivial

normal subgroups.

In fact, the only simple abelian groups are precisely those prime order groups

considered in Corollary III.l.

21
Theorem 111.3 An abelian group G is simple if and only if it is finite and of

prime order.

PROOF: Suppose G is a simple abelian group. By definition, its only normal

subgroups are { e} and G. Since all subgroups of an abelian group are normal,

these are its only subgroups. If g is a non-identity element of G, then <g>= G, so

that G is cyclic. By Theorem III.2( iii), IGI and 1 must be the only divisors of IGI,
hence, IGI is a prime number.

On the other hand, if G is a finite group of prime order, then by Corollary III.l

it has only trivial subgroups and must be simple.

Example 111.3 Consider the group G determined by the following Cayley table:

* e a b c d

e e a b c d

a a b c d e

b b c d e a

c c d e a b

d d e a b c

G is a simple abelian group of order 5 generated by any non-identity element.

One notable feature of the Cayley tables in the preceding examples is the

symmetry of the entries about the main diagonal. A cursory examination of the

22
tables reveals that this symmetry is directly attributable to commutativity. In

fact, if a group table lacks this symmetry, the group is not abelian.

Example III.4 The integers modulo n, denoted Zn, with the operation of

addition modulo n, is a cyclic group of order n. Herein we will represent Zn by the

standard residue class representatives, Zn = {0, 1, 2, ... , n- 1}. For arbitrary

positive n, Zn can be generated by any nonzero element which is relatively prime

to n. In particular, Zn can be generated by the element 1.

Theorem III.4 If G is a group and IGI = p, where p is a prime number, then

PROOF: By Corollary III.1, G is a cyclic group, so for some g =j:. e,

G =<g>= { e,g,g 2 ' ... ,gp-1} .

Define B : G ---+ Zp by B(gk) = [k]P, where [k]P represents the congruence class

of k modulo p. B is well-defined and one-to-one since for integers k1 , k2 , we have

gk 1 = gk2 if and only if gk 1 -k2 = e if and only if p divides ( k1 - kz) if and only if

[ki]P = [k 2 ]p· B is obviously onto. Finally, B is a homomorphism, for if gm, gn E G,

then

Therefore, B : G ---+ Zp is an isomorphism.

After a few more examples, we will generalize this result for all cyclic groups of

order n.

23
Example III.5 If p is a prime number, the operation of multiplication modulo p

on the set of nonzero integers modulo p makes this set a cyclic group of order

p - 1. We will denote this group by Zp•. For instance, let p = 7; then

Z7· = {1, 2, 3, 4, 5, 6} and <3>=<5>= Z 7•. The subgroups of Z 7• are {1}, {1, 6},

{1, 2, 4}, and z7 ..


Example III.6 Although matrix multiplication is typically noncommutative, the

following collection of 2 X 2 matrices comprise a cyclic group under matrix

multiplication:

!I::I,I~1 I,I~ ~1 I,I:~ II·


:
1 1

The identity element of this group is 1 011 and the group is generated by either
10

I I I
0 1
-1 0
I· or 0 -1
1 0

Example III. 7 The following subgroup of the permutation groups Sn, n > 4, is a

cyclic group with respect to composition of permutations:

{(1),(1 2 3 4),(1 3)(2 4),(1 4 3 2)}.

24
0 (1) (1 2 3 4) (1 3)(2 4) (1 4 3 2)

(1) (1) (1 2 3 4) (1 3)(2 4) (1 4 3 2)

(1 2 3 4) (1 2 3 4) (1 3)(2 4) (1 4 3 2) (1)

(1 3)(2 4) (1 3)(2 4) (1 4 3 2) (1) (1 2 3 4)

(1 4 3 2) (1 4 3 2) (1) (1 2 3 4) (1 3)(2 4)

Note that

(1 2 3 4) 1 - (1 2 3 4),

(1 2 3 4) 2 - (1 3)(2 4),

(1 2 3 4) 3 - (1 4 3 2),

( 1 2 3 4 )4 - ( 1)'

and

(1 4 3 2) 1 - (1 4 3 2),

(1 4 3 2) 2 - (1 3)(2 4),

(1 4 3 2) 3 - (1 2 3 4),

(1 4 3 2t - (1),

so that (1 2 3 4) and (1 4 3 2) are generators for this group.

At this point, let us pause to consider the following collection of cyclic groups

of order 4: Z 4 , Z 5 ., {1, -1, i, -i}, {(1), (1 2 3 4), (1 3)(2 4), (1 4 3 2)}, and

25
~ [ : : ] ,[ ~ 1
: ] ,[
1
~ ~
1
] ,[ : ~ 1
] ) . Although their elements

appear different and their operations are diverse, aside from these superficial

characteristics, all of these groups are essentially the same. If we establish a

correspondence between the identity elements, the generating elements, and the

elements of order 2 of each group, it is clear that all these groups are isomorphic

to one another. We previously established a similar result for a group G of prime

order. The following theorem generalizes this result to all cyclic groups of order n.

Theorem 111.5 If G is a finite cyclic group and IGI = n, then G ~ Zn.

PROOF: Let G =<g>. Define a mapping f: Z -----+ G by f(n) = ng for n E Z.

Form, n E Z we have f(m + n) = (m + n)g = mg + ng = f(m) + f(n), so that f

is a homomorphism. In fact, since G is cyclic, f is clearly an epimorphism. If

o(g) = n, then ng = 0 and <n>= kerf; i.e., all integer multiples of n are in the

kernel. By the First Isomorphism theorem, Z /kerf= Z /<n>~ Im f = G for

some n > 0. But Z /<n> consists of the cosets of <n> in Z, which we can

represent by {0, 1,2, ... ,n -1} = Zn. Therefore, we have G ~ Zn·

The following corollary is an obvious consequence of this result and the

transitivity of isomorphisms.

Corollary 111.2 Two cyclic groups are isomorphic if and only if they have the

same order.

26
From this point onward, then, we will use Zn to represent arbitrary cyclic

groups of order n.

As we will exhibit below, finite abelian groups are not necessarily cyclic.

However, as we shall see in Chapter IV, an examination of the structure of any

finite abelian group can ultimately be reduced to an examination of cyclic groups.

The following theorem which concerns both cyclic and noncyclic groups will be

useful in what is to follow.

Theorem III.6 If G is a finite group of composite order, then G has proper

subgroups.

PROOF: If G is a cyclic group, then Theorem III.2( iii) guarantees that G has

unique subgroups corresponding to each divisor of IGI.


If G is noncyclic, then no single element of G generates the entire group. So if

g =/:. e is an element of G, then <g> is a proper subgroup of G.

Example IlLS For an integer n > 1, let Un represent the collection of all positive

integers less than n and relatively prime to n. Under the operation of

multiplication modulo n, Un is a group. Un, called the group of units of Zn, is

simply the reduced residue system modulo n of number theory fame. IUnl = </>(n),

where </> is the Euler </>-function. If n = p, where p is a prime number, then Up is a

cyclic group; in fact, Up is just ZP. of Example III.5. However, for some values of

n, Un is a noncyclic group, e.g., n = 8, 12, 15, 16, 20, 21, .... As a case in point, let

us consider the group Us = {1, 3, 5, 7}.

27
·s 1 3 5 7

1 1 3 5 7

3 3 1 7 5

5 5 7 1 3

7 7 5 3 1

Note that all of the non-identity elements of Us are of order 2, so that no

single element generates the entire group; hence, Us is noncyclic.

In Chapter II we asserted that Aut( G), the collection of all automorphisms of

a group G, is itself a group with respect to composition of mappings. At this

point, we are ready to prove our assertion.

Theorem III. 7 If G is a group, then Aut( G) is a group with composition of

mappings as its operation.

PROOF: Aut( G) is certainly nonempty, since the identity mapping on G is an

automorphism. If a:, f3 E Aut( G), then a: o f3 E Aut( G) since the composition of

isomorphisms is an isomorphism. Moreover, if a: E Aut(G), then a:- 1 E Aut(G)

because a: is an isomorphism of G and, therefore, must have an inverse which is

also an isomorphism of G. In short, in light of the fact that composition of

mappings is an associative operation, we have demonstrated that Aut( G) is a

group.

28
In general, Aut( G) is nonabelian. However, in the case that the underlying

group G is a cyclic group, Aut( G) is abelian, although not necessarily cyclic.

Example 111.9 Aut(Zn), the group of automorphisms of Zn, is a finite abelian

group of order </>(n). In fact, Aut(Zn) ~ Un. Like Un, Aut(Zn) is cyclic if n is a

prime number.

Example III.lO The set of functions {z, -z, ~' -~} is a noncyclic abelian group

of order 4 where functions are composed by substitution of one function in

another. The relevant group properties may be verified in the Cayley table below.

0 z -z 1 _l
z z

z z -z 1 _l
z z

1 1
-z -z z z z

l 1 1
z -z
z z z

_l 1 1
-z z
z z z

Note that z serves as the identity element for this group and that all three

non-identity elements are of order 2.

is a noncyclic abelian group.

29
Example 111.12 The four plane symmetries of a chessboard, S, consist of the

identity rotation e, the rotation r through 1r about the center, and the reflections q1

and q2 in the two diagonals. Under composition of motions these symmetries form

an abelian group.

30
By relabeling the elements and generalizing the operations of the groups in

Examples III.8, III.lO, III.ll, and III.12, we can derive the following generic

Cayley table for these noncyclic abelian groups of order 4.

e a b c

e e a b c

a a e c b

b b c e a

c c b a e

We can conclude that these noncyclic abelian groups of order 4 are isomorphic

to one another and also to Aut( Z 8 ). Furthermore, comparing this table to the

table below which is a general table for a cyclic group of order 4, we can see that

groups of the same order need not have the same underlying structure.

e a b c

e e a b c

a a b c e

b b c e a

c c e a b

Example 111.13 Let S be any finite set and let P(S) represent the set of all

subsets of S. P(S) is called the power set of S. For A, BE P(S), define an

operation * by A* B = (AU B)\(A n B). With this operation P(S) is a finite

31
abelian group. The identity element of P( S) is 0 and each element of P( S) acts as

its own inverse since every non-identity element has order 2. If lSI = n, then

IP(S)I = L:~=o (: ) = 2".

Let S = {a, b, c}, then P( S) contains 23 = 8 elements and exhibits the following

Cayley table.

* 0 {a} {b} {c} {a, b} {a,c} {b,c} {a, b, c}

0 0 {a} {b} {c} {a, b} {a, c} {b,c} {a,b,c}

{a} {a} 0 {a,b} {a,c} {b} {c} {a,b,c} {b, c}

{b} {b} {a,b} 0 {b,c} {a} {a, b, c} {c} {a, c}

{c} {c} {a, c} {b, c} 0 {a,b,c} {a} {b} {a, b}

{a, b} {a, b} {b} {a} {a,b,c} 0 {b, c} {a, c} {c}

{a,c} {a, c} {c} {a,b,c} {a} {b,c} 0 {a,b} {b}

{b,c} {b,c} {a,b,c} {c} {b} {a,c} {a,b} 0 {a}

{a,b,c} {a, b, c} {b,c} {a,c} {a,b} {c} {b} {a} 0

So for n > 1, P(S) is a noncyclic abelian group. As we shall see in Chapter

IV, however, every noncyclic abelian group can be readily decomposed into cyclic

constituents.

Example 111.14 Let {0, 1}n represent all n-tuples of 0 's and 1 's. Define an

operation of component-wise addition on these n-tuples by ai + bi{mod 2}, where i

32
represents the ith component. Then { 0, 1}n is a finite abelian group of order 2n.

The identity element of {0, 1Y is (0, 0, ... , 0), i.e., then-tuple consisting of n 0 's.

Since 0 + 0 =0 and 1 + 1 =0 modulo 2 every element is its own inverse· hence


' ' '
every non-identity element has order 2. In fact, it should be obvious that

{0, 1}n ~ P(S), where lSI= n.


Definition 111.2 Let G and H be groups and define the operation on G x H to be

the Cartesian product:

(a,b)(c,d) = (ac,bd).

This product makes G x H a group with identity element ( ec, eH) and with

(g, ht 1 = (9- 1 , h- 1 ). G X H is called the external direct product of G and H. This

definition generalizes to any finite number of groups.

If G and H are abelian groups, it follows that G x H is an abelian group since

(a,b)(c,d) = (ac,bd) = (ca,db) = (c,d)(a,b).


Definition 111.3 If Gb G2, ... , Gn are normal subgroups of a group G and if

every element of G can be expressed uniquely in the form g = 9192 · · · 9n 1 where

9i E Gi, we say that G is the internal direct product of the Gi. If G = G 1 G2 · · · Gn,

each of the Gi 's is called a direct factor of G. In additive notation we have that if

every element of G can be expressed uniquely in the form g = 9I + 92 + · · · + 9n 1

where 9i E Gi, then G is the direct sum of the Gi 's. Furthermore, if

G = G1 E9 G 2 E9 · · • E9 Gn, then each of the Gi 's is called a direct summand of G.


Notice that the external direct product essentially allows us to construct

"new" finite abelian groups by forming the Cartesian product of two or more finite

33
abelian groups. On the other hand, the internal direct product suggests a

decomposition of a given group G into smaller components. Hence, we refer to the

former as "external," since we are imposing a method for constructing larger

groups from smaller ones, and to the latter as "internal," since we are dissecting

the larger group and investigating the smaller constituents from within.

Theorem III.S Let G be a group with subgroups H and K such that G = H K. If

H nK = { e} and if every element in H commutes with every element in K, then

G ~ H x K; i.e., if H n K = {e}, the external direct product and the internal

direct product are essentially the same.

PROOF: Define¢: H x K ----t G by ¢((h, k)) = hk for all hE H, k E K. Then

for h' E H, k' E K

¢((h, k)(h', k')) ¢((hh', kk'))

hh'kk'

hkh'k'

¢((h, k))¢((h', k')).

Therefore,</> is a homomorphism. Now if </>((h,k)) = ¢((h',k')), then hk = h'k'

and (h't 1h = k'k- 1. Since (h't 1h E Hand k'k- 1 E K and H n K = {e}, we have

(h')-1h = e and k'k- 1 = e, thus h' =hand k' = k, so that¢ is one-to-one. Since

G = H K, we know that every element of G can be expressed as some product hk

for hE H, k E K, so that¢ is onto. Therefore, G ~ H x K.

34
0

Example 111.15 Consider the direct sum Z2 E9 Z 2 = {(0, 0), (1, 0), (0, 1), (1, 1)}.
In the table below notice that the non-identity elements are all of order 2. In fact,

z2 E9 z2 is a noncyclic abelian group of order 4 which is isomorphic to the groups

of Examples III.8, III.10, III.11, and III.12.

E9 (0,0) (1,0) (0,1) (1,1)


(0,0) (0,0) (1,0) (0, 1) (1, 1)
(1,0) (1,0) (0,0) (1, 1) (0,1)
(0, 1) (0, 1) (1, 1) (0,0) (1,0)
(1,1) (1, 1) (0, 1) (1,0) (0,0)

Example 111.16 Z 2 E9 Z 3 = {(0, 0), (0, 1), (0, 2), (1, 0), (1, 1), (1, 2)}.

E9 (0,0) (0,1) (0,2) (1,0) (1,1) (1,2)


(0,0) (0,0) (0, 1) (0,2) (1,0) (1,1) (1,2)
(0,1) (0,1) (0,2) (0,0) (1, 1) (1,2) (1,0)
(0,2) (0,2) (0,0) (0,1) (1,2) (1,0) (1, 1)
(1,0) (1,0) (1,1) (1,2) (0,0) (0,1) (0,2)
(1,1) (1,1) (1,2) (1,0) (0,1) (0,2) (0,0)
(1,2) (1,2) (1,0) (1,1) (0,2) (0,0) (0, 1)

Note that

o((O, 0)) = 1
35
o((O, 1)) o((0,2))=3

o((1, 0)) 2

o((1, 1)) o((1, 2)) = 6.

So Z2 EB Z3 is a cyclic group of order 6, and by Theorem III.5, Z 2 EB Z3 ~ Z6.

Theorem 111.9 Zm EB Zn is a cyclic group if and only if (m, n) = 1.

PROOF: First, suppose that (m,n) = 1 and that o((1,1)) = k, i.e.,

(k(mod m),k(mod n)) = (0,0). This implies that m divides k and n divides k,

and since m and n are relatively prime, it implies that mn divides k. By the

definition of the order of an element, k must be the least positive integer such that

k · (1, 1) = (0, 0), so we have that mn = k. But IZm EB Znl = mn = k, so (1, 1)

must generate the entire group; hence, Zm EB Zn is cyclic.

Now suppose (m,n) = d > 1. Let m' = ';; and n' = J and let (x,y) E Zm EB Zn.
Then

m' dn' ( :z:, y) (m'dn'x(mod m),m'dn'y(mod n))

(mn':z:(mod m),m'ny(mod n))

(0, 0),

so o( ( :z:, y)) < m' dn'. Consequently, Zm EB Zn does not contain an element of order

mn and thus is not cyclic.

36
As we suggested at the outset of this chapter, many group theoretic notions

become significantly simplified when we are working in the context of abelian

groups. In the remainder of this chapter we will explore some of these results and

concepts which become more manageable in this context.

Definition 111.4 The center of a group G, denoted Z( G), is the set of all

elements of G which commute with every element in G; i.e.,

Z(G) = {z E G: gz = zg for all g E G}.

Definition 111.5 For a E G, N(a) = {z E G: az = za} is called the centralizer,

or normalizer, of a in G. N(a) consists, then, of all elements of G which commute

with a.

For arbitrary groups, the center and the normalizers are subgroups. In fact,

the center of a group may be thought of, in a sense, as the "abelian" part of the

group. Moreover, if G is an abelian group, then Z(G) = N(a) = G for every

a E G, since every element of G commutes with every other element of G.

Definition 111.6 Let G be a group with a, b E G. We say that b is a conjugate of

a if there exists c E G such that b = cac- 1 . The set of all elements of G which are

conjugate to a, denoted C(a), is called the conjugacy class of a.

Conjugacy is an equivalence relation on groups; hence, the conjugacy classes of

a group form a partition of the group. This implies that to count the order of a

finite group we need only sum up the number of elements in each of its conjugacy

classes. Denote the number of elements in C(a) by Ca. Then IGI = L: Ca as the

37
sum ranges over one element in each conjugacy class. The number of elements in

the conjugacy class of a is equal to the index of N( a) in G; i.e.,

Ca = [G: N(a)] = 1 J~lw Therefore, IGI = 2:: 1 J~l)l as the sum ranges over one

element in each conjugacy class. Now if a E Z(G), then gag- 1 =a for all g E G

and, thus, C(a) ={a}; i.e., each element of the center is self-conjugate and thus

has a conjugacy class consisting of itself alone. If G is abelian, every element is

self-conjugate. Also, if a E Z(G), then N(a) = G and [G: N(a)] = 1. In view of

this we have

IGI
IGI
2: IN(a)l
IGI
IZ(G)I + 2: IN( a )I
N(a)"f:.G

This equation is called the class equation of G. When G is an abelian group every

element is in the center of G, so the class equation collapses to

IGI = IZ(G)I.

Definition III. 7 Let G be a group with a, b E G. We call the element aba-lb- 1

the commutator of a and b. The subgroup generated by all elements of the form

aba- 1 b- 1 in G is called the commutator subgroup of G. We shall denote the

commutator subgroup as G'.

Let us emphasize that G' is the subgroup generated by all commutators in G.

The collection of all commutators in G may or may not comprise a subgroup on

its own.

38
If G is an abelian group, then aba- 1 b- 1 = e for every a, b E G; thus, every

commutator of an abelian group is the identity element and G' consists solely of

the identity.

The concepts of center, normalizer, conjugate, and commutator certainly have

a reduced impact in the abelian realm. It is just this sort of reduction which is

afforded in the abelian context, however, which sometimes provides crisper results,

or, at the least, simpler proofs.

In the general case, a standard proof of Cauchy's theorem entails a

manipulation of the class equation. However, for the abelian version of Cauchy's

theorem, dependence on the class equation is obviated.

Theorem III.lO (Cauchy's theorem for finite abelian groups) Let G be a

finite abelian group and p a prime number such that p divides IGI. Then there

exists a E G, a =/:. e, such that aP = e.

PROOF: We will use induction on IGI. First, note that the result holds vacuously

in the case G = { e}. Next, let us assume that the theorem holds for all abelian

groups of order smaller than IGI.


If G is a simple group, then by Theorem III.3 and Corollary III.l, it must be a

cyclic group of prime order. As a consequence, IGI = p and G contains p- 1

elements a =/:. e such that aP = alGI = e.


Suppose instead that G is not simple and has a proper subgroup H. Because

H is an abelian group and IHI < IGI, then if p divides IHI, by the induction

39
hypothesis there exists h E H, h ¥= e such that hP = e. Since h E G as well, we

have found an element of G which satisfies our requirements. Therefore, we may

assume that p does not divide JHJ. H <1 G since G is an abelian group; hence, we

can form Gj H, likewise an abelian group. Since p does not divide JHJ, then p

1
must divide JG/HJ, because JG/HJ = ~~1 < JGJ. Since GJH is an abelian group of

order less than JGJ and since p divides JG/ HJ, the induction hypothesis guarantees

the existence of k E Gj H, k i= ec;H, such that kP = ec;H· Now the elements of

Gj H have the form k = Hb, where bE G, so that kP = (Hb)P = HbP. Since

ec;H =He, then kP = ec;H, but k i= ec;H, implies that HbP = H, Hb i= H;

therefore, bP E H, b ~ H. By Corollary II.3, (bP)IHI = e; i.e., (biHI)P = e. Let

c = biHI, so that cP = (biHI)P = e. We need only establish that c i= e. Suppose, to

the contrary, that c =e. Then biHI = e and (Hb)IHI =H. But (Hb)P =Hand p

does not divide JHJ, so Hb = H and bE H, contrary to our previous observation.

It follows, then, that c i= e, cP = e, and our induction is complete.

In Chapter II, we stated Sylow's theorem, which, in part, asserts that if G is a

finite group and p a prime number such that m is the maximal power of p that

divides JGJ, then G has a subgroup of order pm. If G is a finite abelian group, this

Sylow p-subgroup is unique.

40
Corollary 111.3 (to Sylow's theorem) Let G be a finite abelian group and p a

prime number such that pm divides IGI, but pm+l does not for some positive

integer m. Then G has a unique subgroupS of order pm.

PROOF: Sylow's theorem guarantees the existence of S, so we want to show here

that S is unique. Suppose that T is another subgroup of G of order pm, S =f. T.

Since S and T are abelian, ST = T S. Now by Theorem II. 7, IST I = ~~g~ 1


1 = f;rf~ .
Since S =f. T, IS n Tl < pm, so ISTI = pn for some n > m. Because G is an abelian

group, ST is a subgroup of G (Theorem II.6) and ISTI divides IGI. This, of

course, implies that pn divides IGI. However, by assumption, pm is the greatest

power of p which divides IGI, hence, we have a contradiction. We must conclude

that S =T and that S is the unique Sylow p-subgroup of G.

Recall that Lagrange's theorem asserts that if G is any finite group, then IGI is
divisible by the order of any subgroup of G. In general, the converse of Lagrange's

theorem is false, the fact that a given positive integer n divides IGI does not

guarantee the existence of a subgroup of G with order n. Once again, however,

including commutativity in our assumptions provides a nice result, namely the

following partial converse to Lagrange's theorem in the finite abelian case.

Theorem 111.11 Let G be a finite abelian group. If n divides IGI, then G

contains a subgroup of order n.

PROOF: If IGI = 1, then the theorem holds trivially.

41
Case 1. If IGI is a power of a single prime p, we may write IGI = pm, where m

is a positive integer. A divisor of IGI, then, has the general form n = pmi, where
1 < mi < m. By Sylow's theorem, G has at least one subgroup corresponding to
each such n.

Case 2. If IGI involves powers of more than one distinct prime factor, we may

write IGI 1
= p'{' p'; 2 · · ·p7:k, where the Pi's are distinct primes and the mi's are

positive integers. We need to consider four subcases.

(a) n =Pi, 1 < i < k. By Cauchy's theorem, if Pi divides IGI, then G must

contain an element of order Pii therefore, by Sylow's theorem, G must contain a

subgroup of order Pi.

(b) n = pr;'i, 1 < i < k. If pr;'i is the maximal power of Pi dividing IGI, then G

contains subgroups of order p"(i, namely the Sylow Pi-subgroups.

(c) n = p"('i, 1 < i,j < k and mi <mi. The same argument applies here as in

Case 1. If p"(i is the highest power of Pi dividing IGI, then G has subgroups of

orders p"('i, 1 <j < k.

(d) n = p'('ipr:•, where Pi~ Pr, 1 < i,r :S k, mj < mi and ms < mr,

1 < j, s < k. Since G is an abelian group, the product of two of its subgroups is

again a subgroup of G by Theorem II.6. By (c), G has subgroups of order p'('i and

pr:•; hence, G has a subgroup of order n = p'('i pr:•. This result generalizes for all

suitable n which contain the products of powers of two or more distinct primes.

42
Definition 111.8 Let G be a group and p a prime number. If the order of every

element of G is a power of p, then G is called a p-group. Of course, IGI is likewise

a power of p.

The next theorem concerns the center of an arbitrary p-group. We will be most

interested in a corollary to this theorem concerning p-groups of order p 2 •

Theorem 111.12 Let G be a group with IG I = pn, where p is a prime number.

Then Z(G) =/: {e}.

PROOF: We want to show that Z( G) contains an element other than the identity.

If a E G, then the normalizer of a, N(a), is a subgroup of G; hence, IN(a)l divides

pn. So IN(a)l = pna for some na < n. Also, a E Z(G) if and only if na = n. The

class equation of G is

Since p divides pn and p divides ::a for each na < n, then p must divide IZ(G)I.

Since e E Z(G), then IZ(G)I > 1. Finally, because IZ(G)I must be a positive

integer divisible by p, we must conclude that IZ(G)I > 1; thus, Z(G) must contain

an element other than e.

Corollary 111.4 If IGI = p 2 , with p a prime number, then G is an abelian group.

PROOF: We want to show that Z( G) = G. By the preceding theorem, we know

that G has a nontrivial center and, since Z( G) is a subgroup of G, we have

IZ( G) I = p or p 2 • If IZ( G) I = p 2 , then Z( G) = G and our conclusion holds.

43
Suppose, on the other hand, that IZ(G)I = p. Let a E G, such that a~ Z(G).

Then N(a) is a subgroup of G, Z(G) ~ N(a) since all the elements in the center

commute with a, and a E N(a). This implies that IN(a)l > IZ(G)I = p. However,

by Lagrange's theorem, IN(a)l divides p 2 , implying that IN(a)l = p 2 • But this

forces a E Z(G), a contradiction. So IZ(G)I =pis not possible; thus, Z(G) =G.

Theorem 111.13 If IGI = p 2 , then either G is a cyclic group or G is isomorphic

to Zp EB ZP.

PROOF: If IGI = p 2 , then the non-identity elements of G must be of order p or p 2 •

If G contains an element of order p 2 , then G is obviously cyclic. If G does not

contain an element of order p 2 , then all elements of G except e must have order p.

By Theorem III.12, Z( G) is nontrivial, so choose a E Z( G) and bE G, b ~<a>.

There are p 2 distinct elements of the form (ma) + (nb ), 1 < m, n ~ p since

<a> n <b>= {e}, so <a> EB <b>= G. Also, since a E Z(G), every element of

<a> commutes with every element of <b>. By Theorems III.8 and III.5, then

G ::::;<a> EB <b> and since I <a> I= I <b> I= p, G:::::; Zp EB ZP.

In particular, this theorem says that all groups of order p 2 are isomorphic to

either ZP2 or Zp EB ZP. We have demonstrated several groups of order 4 in the

examples above. Notice that Examples III.l, III.5 (for n = 5), III.6, and III.7 are

all isomorphic to Z 22 = Z 4 , while examples III.8 (for n = 8), III.9 (for n = 8),

44
III.lO, III.ll, III.12, III.13 {for lSI = 2), III.l4 {for n = 2) are isomorphic to

Z2 EB Z2.

As we have demonstrated, the properties of finiteness and commutativity

seemingly simplify the groups which possess them. In Chapter IV we will prove

the decisive Fundamental Theorem of Finite Abelian Groups. This theorem,

notable for both its power and its simplicity, exploits these properties to provide a

complete decomposition of a finite abelian group into its simplest essential

components. It enables us to thoroughly examine the structure of finite abelian

groups and, hence, results in their complete classification, a result not realized for

groups in general.

45
CHAPTER IV

THE FUNDAMENTAL THEOREM OF FINITE

ABELIAN GROUPS

Frobenius and Stickelberger's proof in 1878 [4] of their version of the

Fundamental Theorem of Finite Abelian Groups was a milestone in the study of

group theory. The theorem thoroughly demystified the structure of finite abelian

groups and allowed for their complete classification. Moreover, an important

ancillary result of this theorem is an exact count of the number of nonisomorphic

abelian groups of any given order. Our focus in this chapter, then, will be on

offering a proof that any finite abelian group can be realized as the isomorphic

copy of a direct sum of cyclic groups.

As we demonstrated in Chapter III, the direct sum of abelian groups is

abelian; hence, the direct sum of cyclic groups must be abelian as well, though not

necessarily cyclic as Theorem III.9 concurs.

Recall that a p-group is a group in which every element has order a power of a

prime p. For instance, Z 2 E9 Z 2 and Z 8 are 2-groups and Zzs is a 5-group; however,

there is no prime p for which z12 is a p-group.

Definition IV .1 Let G be a finite abelian group and let p be a prime number.

Define Gp = {g E G: o(g) = pn for some integer n > 0}. We call Gp the

p-component of G.

46
Theorem IV.l (Primary decomposition theorem) Every finite abelian

group G is a direct sum of p-groups.

PROOF: Assume IGI > 1, since the theorem holds trivially in case IGI = 1. Let

P = {p: P divides jGj,p a prime}. Now for pEP, the p-component of G is

nonempty since 0 E Gp for all primes p E P. Furthermore, because G is an abelian

group, Gp must be a subgroup of G. We want to show that G = ffipEP Gp. To do

so we must be able to express every g E G uniquely in the form g = l:pEP gp,


where 9p E Gp. Since o(g) divides IGI we can express o(g) as o(g) = npEP np,
where each nP is a power of p. Next, let mP = ~for each pEP.
np
Since the

greatest common divisor of the mp 's is 1, the Euclidean algorithm insures the

existence of integers ap such that 1 = I:pEP apmp. Multiplying both sides by g we

Suppose that g can also be expressed as g = l:pEP hp with hp E Gp. Then

of p; hence, o(l:pEP(9p- hp)) is the least common multiple of the orders of all the

p E P. Therefore, gp - hp = 0; thus, gp = hp for all p E P. Hence, the expression

of each g E G as the sum of one element from each p-component of G is unique.

47
In other words, Theorem IV.1 says that the p-groups which do the trick are

precisely the p-components. So, alternatively, we can express Theorem IV.l as

follows: A finite abelian group G is the direct sum of its p-components for the

various primes p which divide IGI.


This last result significantly narrows our search for a method to express all

finite abelian groups in terms of cyclic groups. In essence, Theorem IV.1 allows us

to restrict our efforts to showing that all finite abelian p-groups can be expressed

as a direct sum of cyclic groups.

Example IV.l Consider Z 12 = {0,1, ... ,11}.

IOI 1,

Ill 151 = 171 = 1111 = 12,

121 1101 = 6,

131 191 = 4 = 22,

141 181 = 3,

161 2.

Then

G2 {0, 3, 6, 9},

G3 {0, 4, 8}.

48
So, by the primary decomposition theorem, Z 12 = G 2 EB G3 , which is clear since

G2 EB G3 = {0, 9 + 4 = 1, 6 + 8 = 2, 3, 4, 9 + 8 = 5, 6, 3 + 4 = 7, 8, 9, 6 + 4 =
10,3 + 8 = 11} = Z 12 •

Before we can proceed to show that every finite abelian p-group is a direct sum

of cyclic groups we need the following definition and pair of theorems.

Definition IV .2 A mapping f is said to be idempotent iff o f = f.


Theorem IV .2 Iff is any idempotent endomorphism of a group G with lmf <l G,

then G = ker fEB Im f.

PROOF: Suppose f :G ~ G is an idempotent endomorphism such that

lmf <l G. We want to show that each g E G can be represented uniquely as a sum

z + y, with z E kerf, y E lm f. Now g = (g- f(g)) + f(g), where f(g) is

obviously in Im f. Since f(g- f(g)) = f(g + f( -g))= f(g) + f(f( -g)) and

f of= f, then f(g + f( -g))= f(g) + f( -g)= 0 implies that g- f(g) E kerf.

To see that this sum is unique, suppose g = z +y with z E kerf and y E lm f, so

that f(z) = 0 and for some a E G, f(a) = y. Then

f(g) = f(z + y) = f(z) + f(y) = 0 + f(y) = f(f(a)) = f(a) = y, and, thus,


z = g- f(g).

Theorem IV .3 Let G be a finite p-group with g E G of maximal order. Then

<g> is a direct summand of G.

PROOF: If G =<g>, we are done; therefore, assume G =J<g>. Let a E G, but

a fi_<g>. Since G is a p-group, then for some positive integer n, pna E<g>.

49
Therefore, there exists bE G, b ff.<g>, such that pb E<g> (for instance,

b = Pn-Ia). Since g is of maximal order, o(g) > o(pb), and for some integer m,

pb = pmg. So 0 = pb- pmg = p(b- mg) and b- mg ~<g>, since b ~<g>.


However, since o(b- mg) = p, it is clear that <b- mg> n <g>= {0}. Thus, there

exists a nonzero subgroup, call it S, of G such that Sn <g>= {0}.

Next, consider the quotient group, Gj S. Note that in G/ S, o(S +g)= o(g);

hence, S + g must be an element of maximal order in G/ S. By induction on IGI,

since JG/ Sl < IGI, we conclude that <S + g> is a direct summand of Gj S. Then

the following composition of mappings

where 1J is the natural homomorphism, '1/J is a projection of G / S onto its direct

summand,¢ maps m(S +g) to mg, and vis an inclusion mapping, is an

idempotent endomorphism of G with image <g>. Hence, by Theorem IV .2, <g>

is a direct summand of G.

Theorem IV .4 Every finite abelian p-group G is a direct sum of cyclic groups.

PROOF: Let g E G be of maximal order. By Theorem IV.3 there is a subgroup S

of G such that G = S+ <g>. Since lSI < JGJ, then by induction on IGI we have

that Sis a direct sum of cyclic groups; hence, S+ <g>= G is a direct sum of

cyclic groups.

50
Example IV .2 Recall from Example III.8 that for some values of n, Un is a

noncyclic finite abelian group. In particular, U15 = {1, 2, 4, 7, 8, 11, 13, 14} is a

noncyclic 2-group. Now, 2, 7, 8, and 13 are all elements of maximal order in U1s·

So, for example, <2> is a direct factor of U15 • In fact, since <2>= {1, 2, 4, 8} and

<11>= {1, 11}, we can see that we can express U15 as the direct product of cyclic

groups by u15 =<2> X <11>::::: z_. EB z2.


In the preceding theorems we have progressed from finite abelian groups to

p-groups to cyclic groups, reducing the complexity of the structure at each step.

We are now prepared to state and prove the Fundamental Theorem, which not

only asserts that we can decompose every finite abelian group, but also that this

decomposition is, in a sense, unique.

Theorem IV .5 (Fundamental Theorem of Finite Abelian Groups) Every

finite abelian group is a direct sum of cyclic groups. Moreover, if G is a finite

p-group and if

G G1 EB G2 EB · · · EB Gm

H1 EB H2 EB · · · EB Hn

are direct sum decompositions of G into nonzero cyclic p-groups, then m = n and

Gi ::::: Hi for all i, after suitable rearrangement of subscripts.

PROOF: The first statement, often by itself referred to as the Fundamental

Theorem of Finite Abelian Groups, follows immediately by combining Theorems

IV.l and IV.4.

51
We prove the second assertion by induction on m. If m = 1, then G = Gt is

itself a cyclic p-group; therefore, there exists 9 E G such that o(9) = IGI.

Expressing 9 in terms of the second decomposition, we have 9 = h 1 + h2 + · · · + hn,


where hi E Hi. Let o(9) = pt, i.e., tis the least positive integer such that Pt9 = 0.

Then 0 = Pt9 = pt(ht + h 2 + · · · + hn) = pth 1 + pth 2 + · · · + pthn; thus, for all i,

o(hi) <pt. Suppose, next, that o(hi) < pt for all i, say o(hi) = pr, r < t. Then

pr 9 = pr h1 + pr h2 + · · · + pr hn = 0, a contradiction since t is the least integer such

that pr 9 = 0. Therefore, we must conclude that for some i, pt =o(9) =o(hi)i thus,

IGI = IHil and G =Hi. As a consequence, n = 1 and we have G 1 ~ H 1.

Assume now that m > 1. Renumber the Gi so that IG 1 I > IGil for all i 2: 2.

Next choose 91 E G such that <9>= G, and write 91 = h 1 + h 2 + · · · + hn. Now, as

shown above, o(9t) is the maximum of the o(hi), say o(9t) =o(hj) for some j.

Since o(9t) = IG 1I and o(hj) < IHjl, we have that IG1I::; IHil· In a similar

fashion, suppose <hi'>= Hj. Then hj' = 91' + 92' + · · · + 9m' with o(hj') =o(9k' ),

where o(9k') is the maximum of the o(9i')· Since o(hj') = IHil and o(9k')::; IGkl,

we have IHj I < IG k I· Thus, combining these results we see that IGtl = IHj I·

Without loss of generality, let j = 1. Now G1 n (H2 E9 H3 E9 · · · E9 Hn) = {0}, since

if t91 = t(h 1 + h2 = · · · + hn) = :z: 2 + Z3 + · · · + Zn, where Zi E Hi, then th1 = 0

and, in turn, t91 = 0. Now, since IG 1 E9 (H2 E9 · · · E9 Hn)l = IH1 E9 H2 E9 · · · E9 Hnl,

we have that G = G1 E9 H2 E9 · · · E9 Hn. Therefore,

G/G1 ~ G2 E9 · · · E9 Gm ~ H2 E9 · · · E9 Hn.
52
Let 0
: H2 EB · · · EB Hn ~ G2 EB · · · EB Gm be an isomorphism. Then

G2 EB · · · EB Gm = o(H2) EB · · · EB o(Hn) and by the induction hypothesis,

Gi ~ o(Hi) for all i > 2 after suitable resubscripting. Clearly, H, ~ a(Hi) for all

i > 2; therefore, Gi : : : : Hi for all i. So any two decompositions of a finite p-group G

have the same number of summands of a given order.

Example IV .3 Consider G = Z 2 EB Z 2 and its cyclic subgroups:

H1 {(0, 0), (0, 1)},

H2 {(0, 0), (1, 0)},

H3 {(0, 0), (1, 1)}.

G can be written in the following ways as a direct sum of these cyclic subgroups:

G H1 EB H2

H1 EB H3

H2 EB H3.

While, as the example suggests, there may be several ways to express a given

finite abelian group as the direct sum of cyclic groups, the Fundamental Theorem

demonstrates that a correspondence exists in these direct sum decompositions

between summands of the same order which make the decomposition unique.

53
Example IV.4 Examples III.lO, III.ll, and III.12 are all isomorphic to Z 2 EB Z2.

For example, in III.12, the plane symmetries of a chessboard, S, can be expressed

as the direct product of cyclic groups in the following ways:

S { e, r} x {e, 9I}

{ e, r} x { e, 92}

{ e, 91} X { e, 92 }.

Example IV.5 In Example III.13, for lSI= 3, we see that P(S) can be realized

as a direct product of cyclic groups by <{a}> x <{ b} > x < { c} >; hence, for

lSI= 3, P(S);::::;: Z2 EB Z2 EB Z2. In general, then, for lSI= n, P(S) is isomorphic

to the direct summand of n copies of Z 2 •

Thus, the Fundamental Theorem fully discloses the structure of finite abelian

groups. Moreover, recalling Theorem III.5 which concluded that every finite cyclic

group of order n is isomorphic to Zn and applying this to the case when n = pk,
we see that it allows us to almost effortlessly exhibit one group from each

isomorphism class of finite abelian groups. In fact, to determine the number of

distinct isomorphism classes of abelian groups of a given order n, we need only

determine all possible ways of factoring n as a power of primes.

54
Example IV .6 There are six distinct prime power factorizations of

n = 108 == 22 • 33 ; thus, six isomorphism classes of abelian groups of order 108

exist. Following is a list of one representative from each class:

Example IV. 7 There are six isomorphism classes of abelian groups of order

n == 594, 4 73 == 11 2 • 17 3 • To derive a set of class representatives, simply replace the

primes 2 and 3 in Example IV .6 with 11 and 17, respectively.

Generalizing examples IV. 6 and IV. 7 it is clear that there are six isomorphism

classes of abelian groups of order n = p 2 q3 for any distinct primes p and q. In

Chapter V we will see that a finite abelian group is determined uniquely, not by

its order, but by the exponents associated with its order.

55
CHAPTER V

INVARIANTS OF FINITE ABELIAN GROUPS

In this chapter we will investigate an invariant of finite abelian groups which

arises as a direct result of the Fundamental Theorem. For our present purpose we

will define an invariant as a simple notion associated with a group G and

independent of any particular decomposition of G, which, when known, completely

determines G, or some characteristic of G, up to isomorphism. Along the way we

will arrive at a precise method to count nonisomorphic abelian groups of any finite

order.

If our interest in finding an invariant were limited to finite cyclic groups, our

search would be effortless. Theorem III.5 tells us that IGI completely determines a

finite cyclic group G up to isomorphism. Unfortunately, this result does not

directly generalize to all finite abelian groups as Examples IV .6 and IV. 7 illustrate.

However, the Fundamental Theorem leads us to the generalization we seek.

If G is a finite p-group, then G = G1 EB · · · Gm, where each Gi is cyclic. For

convenience, let us assume throughout this discussion that the Gi are written in

descending order of cardinality. Since each of the Gi 's is a cyclic group, then by

Theorem III.5 each is completely determined by IGJ Therefore,

{jG 1 j, IG2j, ... , IGml} serves as an invariant of G. The second assertion of the

Fundamental Theorem insures that these invariants do not depend on any

particular decomposition of G into a direct sum of cyclic groups; i.e., if G ~ H,

56
then G and H have the same invariants. As a consequence, if we know the

invariants { i1, i2, ... , im} of a p-group G, where the ij are powers of p, say ij = pki,

we can construct G simply by forming the direct sum Zi 1 EB · · · EB Zim, where Zij

represents the cyclic group of order ij.

Definition V .1 Let n be an integer. The sequence of positive integers

n1' n2' · · · , nr, where n 1 > n 2 > · · · > nr > 0, is called a partition of n provided

that n = L~=l ni. We shall denote the number of partitions of a given integer n by

P(n).

Example V.1 P(1) = 1 since 1 = 1 is the only partition of 1. 'P(2) = 2 since


2 = 1 + 1 = 2. P(4) = 5 since 4 = 1 + 1 + 1 + 1 = 2 + 1 + 1 = 2 + 2 = 3 + 1 = 4.
Theorem V .1 Let p be a prime number and n a positive integer. The number of

nonisomorphic abelian groups of order pn is equal to P( n).

PROOF: Let G be an abelian group of order pn with invariants {i 1, i 2, ... , im}

such that ij = pni and n = L:j= 1 nj. Then G is the direct sum of cyclic groups

G = G 1 EB ·· ·Gm, where !Gj! = pni; i.e., Gj :=::::: Zpni· However, there are P(n)

distinct partitions of n and, thus, P( n) isomorphic abelian groups of order pn.

Example V.2 The number of nonisomorphic abelian groups of order 75 zs

P(5) = 7. Since

5=1+1+1+1+1=2+1+1+1=3+1+1=4+1=2+2+1=3+2=5,

we can represent the isomorphic classes of abelian groups of order 75 as follows:

z1 EB Z1 EB z1 EB Z1 EB z1,

57
Z72 EB Z7 EB Z7 EB Z7,

Z7a EB Z7 EB Z7,

z74 EB z7,
Z72 EB Z72 EB Z7,

Note that the result in Theorem V .1 is entirely independent of the choice of p.

So, for a given n, there are precisely the same number of nonisomorphic abelian

groups of orders 2n, 3n, ... , pn, ... , if p is a prime number.

Using the primary decomposition theorem, we can generalize Theorem V.l to

all finite abelian groups and thus have a means to count the number of

nonisomorphic abelian groups of any given order.

Corollary V .1 Let n be a positive integer such that n = p~ 1


• • • p~m, where the Pi 's

are distinct prime numbers. The number of nonisomorphic abelian groups of order

n is equal to I1~ 1 P(ni)·

PROOF: Let Gp; be a Pi-component of a group G with IGp; I =Pi;. By Theorem

V.l there are P(ni) such nonisomorphic Pi-components possible. Since

G= 61i:: 1 Gp; and n = IGI = f1i:: 1 jGp;l, then there are Tii:: 1 P(ni) nonisomorphic

abelian groups of order n.

58
Example V .3 There are 105 nonisomorphic abelian groups of order

n = 24,149,664 = 11 3 • 7. 34 • 25 since P(3). P(1) · P( 4) · P(5) = 3 · 1 · 5 · 7 = 105.


So far we have examined an invariant for cyclic groups and p-groups only. Our

final step is to broaden these results to find the invariant which applies to all finite

abelian groups. In this regard, we will expand the result of the second assertion of

the Fundamental Theorem to encompass all finite abelian groups.

First, recall that Theorem III.9 says that the direct sum of two cyclic groups is

a cyclic group if and only if their orders are relatively prime. In general, if

A1, ... , An are cyclic groups such that (lA~ I, IAil) = 1 whenever i # j, and

<ai>= Ai, then Ea~ 1 Ai is itself a cyclic group generated by the element 2:7= 1 ai.
Theorem V .2 Let G be a finite abelian group. If

are direct sum decompositions of G into nonzero cyclic groups with IGil divisible by

IGi+II and IHil divisible by IHi+II for all suitable i, then m =n and Gi :::::::Hi for

all i.

PROOF: Let K be a p-component of G. Then K is the direct sum of the

p-components of each Gi; i.e., K = E9~ 1 ( G )w By Theorem III.2( i), since Gi is


1

cyclic, its subgroup (Gi)p is likewise cyclic. In similar fashion, K = E9~= 1 (H,)w By

our assumptions, we have I(Gi)pl > I(Gi+I)pl and I(Hi)pl > I(Hi+I)PI; therefore, we

have that I(Gi)pl = I(Hi)pl, whence (Gi)p::::::: (Hi)p for all i. Since this result holds

59
for all P dividing IGI and Gi = EBpeP( Gi)p, Hi = EBpeP(Hi)p, we have that G, ~ Hi
for all i.

As a result, we see that the family {IG 11, ... , IGml} implicit in Theorem V.2

does not depend upon a particular decomposition of G. Moreover, given such a

family, say {nb ... ,nm} where ni+I divides ni fori= 1, ... ,m- 1, we can

construct G = 2n 1 EB · · · EB 2nm. We conclude that the family associated with G

satisfying the aforementioned conditions is indeed an invariant of G.

Definition V .2 Let G be a finite abelian group written as the direct sum of cyclic

groups, say G = G1 EB · · · EB Gm and that IGil = ni, such that ni+l divides n for 1

i = 1, ... , m - 1. The family of positive integers {n 1, ... , nm} are called the

invariant factors of G. The invariant factors of the p-components of G are called

the elementary divisors of G.

Example V.4 Consider the finite abelian group

G = [25a EB Z 5 2 EB 2 5 ] EB [Z3 4EB 2 3 2] EB [22a EB 2 2 EB 22], where the brackets distinguish

the various p-components of G. Regroup the summands by the following method:

1. For all primes p dividing IGI, form the direct sum of their maximal

summands.

2. To step 1 add the direct sum of the maximal summands of the remaining

summands.

3. Repeat step 2 until all summands have been exhausted.

60
Our regrouping yields G = [Z5 a EB Z 3 4 EB Z 2 a] EB [Z5 2 EB Z 3 :z EB Z2] EB [Zs EB Z2].

Each of the bracketed direct sums is a direct sum of cyclic groups whose orders are

relatively prime, hence each bracketed sum is itself cyclic. Thus, we can write G as

the direct sum of cyclics G = Z 5 a_ 3 4.


2
a EB Z 52. 3 :z. 2 EB Z 5 .2 • Note that the order of each

summand is divisible by the order of the successive summands. Hence, the

invariant factors for G are {5 3 • 34 • 23 ,5 2 • 32 • 2, 5. 2}. The elementary divisors of


3
G are {5 ,5 2 ,5},{3\3 2 },{23 ,2,2}.

Notice that the invariant factors of a p-group are simply its elementary

divisors, i.e., the orders of its constituent cyclic p-groups. Furthermore, the

invariant factor of a cyclic group G is merely \G\. So this general invariant we have

discussed encompasses the particular invariants mentioned earlier as special cases.

A couple of other invariants of finite abelian groups merit mention as we

conclude this discussion. First, if \G\ = p~ 1 • • • p~m, then we define

T( G) = ~~ 1 P?;, with T( G) = 0 if G = { e}, to be the trace of G. As it turns out,

T( G) determines the degree k of the minimal symmetric group Sk in which G can

be imbedded [9]. Another invariant, the Ulm invariant, counts the number of

cyclic summands of order pn+I of an abelian p-group. In the finite case, the Ulm

invariant, denoted U( n, G), is given by

61
where d stands for the dimension of the parenthetical group when it is considered

as a vector space over Zp. A more general form of the Ulm invariant holds for

infinite abelian p-groups as well [14].

In our discussion we have demonstrated how the conditions of finiteness and

commutativity simplify broader group concepts and ultimately allow for the

complete decomposition of all finite abelian groups. Furthermore, we have seen

that, given an order n, we can readily determine how many nonisomorphic classes

of finite abelian groups of that order exist. Moreover, we can easily exhibit a

representative for each isomorphism class. In sum, the structural characteristics of

finite abelian groups make them highly accessible to investigation.

62
REFERENCES

[1] Armstrong, M. A. Groups and Symmetry. Undergraduate Texts in


Mathematics. New York: Springer-Verlag, 1988.
[2] Caskey, J. H. A Study of Abelian Groups. M.S. thesis, Texas Tech University,
May, 1955.
[3] Durbin, J. R. Modern Algebra: An Introduction, 2nd ed. New York: John
Wiley & Sons, 1985.
[4] Frobenius, G. and L. Stickelberger. "Ueber Gruppen von vertauschbaren
Elementen," J. Reine angew. Math., 86: 217-262, 1878.
[5] Fuchs, L. "Historical Survey of Abelian Groups," American Mathematical
Heritage: Algebra and Applied Mathematics, no. 13, 61-71, 1981.
[6] Gorenstein, D. Finite Groups. Harper's Series in Modern Mathematics. New
York: Harper & Row, Publishers, 1968.
[7] Hall, G. G. Applied Group Theory. London: Longmans, 1967.
[8] Herstein, I. N. Topics in Algebra, 2nd ed. New York: John Wiley & Sons,
1975.
[9] Hoffmann, M. "An Invariant of Finite Abelian Groups," American
Mathematical Monthly, Vol. 94, No. 7, August-September, 664-666, 1987.
[10] Kleiner, I. "The Evolution of Group Theory: A Brief Survey," Mathematics
Magazine, 59: 195-215, 1986.
[11] Ledermann, W. Introduction to the Theory of Finite Groups. Edinburgh and
London: Oliver and Boyd; New York: Interscience Publishers, Inc., 1949.
[12] MacDonald, I. D. The Theory of Groups. Oxford: The Clarendon Press, 1968.
[13] Robinson, D. A Course in the Theory of Groups. New York: Springer-Verlag,
1982.
[14] Rotman, J. J. The Theory of Groups, an Introduction. Boston: Allyn and
Bacon, Inc., 1965.
[15] Walker, E. A. Introduction to Abstract Algebra. New York: Random House,
1987.

63

You might also like