— —






















— —
— —














































— €



























































— —
— — —




— —


— —

— —






















































































— —
— — —


































































































— —
— — — —
— —





































— —


























— — —
— —
— —







— — —































































— —














— —


— —









— —




















— —






— —


— —









































































— — —









































































— —




































— — —








— —


































































— —























































— —





















— —

— — — €






— —




















— — — — €










— —

















































































— —















— —



























— —














— —





— —
— —





— —





















— —







— —









































































































— —



























• • •
• • •
























































































































































































































































— — —































• • •




• • •
























— — —























— —




















— —
























— — —
























— — —





• —






— —
— —
— —




— — —







— — —
— — —

— —



































































• —






































































































— —



— — —
— —
— —


































































— —










— —






































































































— —




















— —











































































































































































— —






















































— —















































— —













































































































































































































































































































— —




























































































— —
















































— €

















































— —

— — •





































































— —



















— —




































































































































































































































— —


— — €























































































































































• •




























































— — —
— — —









— — —











































































































































































































































































— —








— — —





















































































— —











































— — —
— — — —









— —

— —

— —

— — — —
































— —






































































































































— — —


— —
— —
— —






















































— — — — — —



















































— — — —


— —


— —
— —




— —








— —


— — —


— — — — — — — —

— — — — — —






— —











— —
— —






























































— —












































































— —



















































































— —






















































— — — —
— — —
— — —


— — —








— — —










































































— —












— — — —

— — — —




— — — — — — —

































— —




— —












— —


— —
















— — —






€ €






— —
— —


— —























— —


























— — — — —










— —











— —



— —


— — • —


— —















— € —



















































































— —










































































































































— —















































































— — —








— —































































— —




— —
























— — —












— — —








— — —
— —








— —













































— —














































— € —





— —



















































— — —





















— —














































— €





























— —

















— —
















— —



















— —























































— —






















— —



























— —

















































— —

































— —
















































— €























— —
— — €
— — — —


— —










































— —
— —











— —






— —














— —






— —




— —
























































— —



— —

— — —










— — —
























































€ € — —




























































































































































— — —












— €
— —



— — €


— —



— —
— —









































— —

















— —













— —






























— — —





































































— —







— —

— —
— —










































































— —




























































































— — —





— —























































— —














































— —







— —













— — —
— —

— —

























— —

— —





— —
— — —
— —


— — — —

















— — —





















— —




































— €



































— —
















































































— —












































































































































































— —

— — —

— —







— — €








— —

— — — — —


































• •








































— — —













— — — — —


— — — — — —
— —
— — — —
— — — —

















— — —


— — — —





— —
















































— —

— —









— —


























— — —

— —























— —





— —



— —


















— —














































































— —





























— —


































































































































































— — —
















































































— —























































— —



















































— —










— — —



















— —














— —



















— —






— — — —
























— —




























































































— —































































































— —




























— —






— —

— —














— —
















— — —






























— —



















— —































— —































— —

— —





















































































































































— —


— —
— — — —
— —







— — —












































— —


— —


— —


























— —




















— —








• —









— —







































— —














— —








— — —








— —






— —






















— — — — —
















— —
























































— —






































— —
— —


— —







— —
— —





— — —





— —









— —




€ —



— —
— —




— — —


— —












— —

































€ — — — €








— — — —
— — — — —
— — —
— —



— — — — — —






— —





— —










— —

— —

































— —









— —



— —


































— — — — — — —




























— —



— •


— —
— —
— —






— €
















— —
— —





























— — —



























































— —

































• • •























— —

— — —





— —

— —

— €






































































































































— —






















— —































































— —
— — —






















— — — — —







— —




— —


















— —















— —


















— — —

— —


















— — —
— — —




























































































































— — —
— —








— —









— —




























































— —

— —




























































































































— €

— —












— — —


— —























— — —





— — — —












— —








































































— —












































































• •














































— — —
— —















— —





— — —
— —

— —



— —






























• •









































— —

































— —































































































— € —
































































— —





































































— —












— —















— —




















• •











— —





— — — —

— —











• • •



— —


















— — — —


— —









— — —







— —


— —







— —

— —
— —

— — —

— —





— —



















— —

























€ —




























• • •


€ €



















— — —













— —





























— —










— —

























































— — — — —
— —




— — —
















— —

— —
— —
— —











• •













































— —

— —



— — —


— —

— —



















— —





































• —





















— —




— —







































— —








— — — — —






































• • •


• • •


• • •









— —









— —

— — — —





— — — —

— —

— — — —












— — —






— — —













































— —
— —

















— —








































— —
































— —























— — —













— —
— —




































— — — — — —






— —

— —


— —
— —

— —




















































— —



















— — —

























— — — —



— —

— —
























— — —




— — — — — — —


— — — —




— — —




— — —

— —







— — — —
• — • — • • •

• • — —

• —





— —





















— •










— — — —

— —








— — — —
— — — —
— —



— —
— —
— —
— —

— — —
— — — —





— —
— —

— — — —
— —

— —











— —

— — — — —

— —
— —
— — — — — — — —



— —



— —


— —
— —






— —
— —
— —


— —
— —

— — —
— —

— — — —

— — —













— — —














— — — —










— —

— —
— • —


— —






— —
— —

— € —








— —

— —
— — —


— —


























— • • — — —



















— —








— —












— •







— —





























— —

























— —
— — • —
— • —




— —





— — — —










— —



— — —
— — —





— —
















































— —




— — —
— —
— € —




— — — — —







— —

— — — —


— —










— — — —






— —
— — — — —
— — — —


— —




— —








— —

— — — — —


— —


— —

— —
— — — — —
— — — — — — —


— — — —
— — —
— —
— — — —
— — — —
— — —

— € — —
— — — — — —


— €
— — —
— —


— —






— —














— —


€ — — — — — —

— —





— —
— — —










— — — —







— —









• •


— —
— — —


— €



€ —




— —



— — —





— —










— —



— —














— —




— —













— — — —







— —
— —

— —
— — — — —











— —




— — —



















— — —






























— — —
— —

— —




— — — —



















































— • —



— —
— —


— — — — —
— — — — — —





— — —

— —

— —

— — — — — — —

— — —





— —
— — — —

— — —

— — — — —


— — —


— — — —


— — —

— — — — — —


— — —



— — —











— — — —
— —
— — — — — —
— — — — — —
— —









— — — — —

— — —



— — — — —

— — —


— — —

— — —







— — — — —

— —



— —


— —


— —

— —



— —

— — —

— — — — — —




— — —








— — — —
— —























— —
— —
— — —



— —



















— — —


— — —



€ — —
























— — —






— — — —
— —
















— —











— —







— — — — — —



— — — —

— —



— — — — —

— — — —










— — — —

— — —

— —
— — — — —

— —
— —
— — —






— — —




















— — —







— —














— —




— — — —

— —
— —









— —


— — —

— —

— —
— —


— —























— —
— — — —





— —








— — —












— —

— —


— —





— —






— — —


— — — —


— — —
— — —
— — —

— —

— —





— — —

— —
— — —




— —


— —
— — —
— —














— — —







— —



— —



— —




— — —





— —



— —
— — —
— — — —





— — —

















— —
— — — — —
— —




— —





— €
















— —












— — —








































— —

— —



— —













— —
— —

— —






— —







— —













































— — —























































































































































































































































— —























































































































































































































































































































































































































































— —



























































































































































































— —











































































































































— —
















































BASIC ABSTRACT ALGEBRA
Second edition

Basic abstract algebra
Second edition

R B. BHATTACHARYA
Formerly. Urdverslty of Delhi

tK.JAIN
Ohio University
&

P. NAGPAUL

Si. Siephens College. Delhi

CAMBRIDGE
UNIVERSITY PRESS

Published by thc Press Syndicate of the University of Cambridge The Pitt Building. Trumpington Street, Cambridge CB2 1RP 40 West 20th Street, New York, NY 10011-4211. USA 10 Stamfurd Road, Oaklcigh, Melbourne 3166, Australia © Cambridge University Press 1986, 1994 Firsi edition publIshed 1986 Second edition publIshed 1994 Reprinted 1995

Ubrary of Congress Cataloging-in-Publication Data Is available.

A catalogue record for this book is available from the British Library. ISBN 0-521-46081-6 hardback ISBN 0-521-46629-6 paperback

Transferred to digital printing 1999

For PARVESH JAIN To whom we owe more than we can possibly express

Contents

Preface to the second edition Preface to the first edition Glossary of symbols

page xiii
XjV

xviii

Part I
Chapter I Sets and mappings
I. 2.
3. 4.
5.

Preliminaries
3 3

Sets

Relations Mappings Binary operations Cardinality of a set

9
14
21

25

Chapter 2

Integers, real numbers, and complex numbers
I. 2. 3.

30
30 35 36

Integers

Rational, real, and complex numbers
Fields

Chapter 3

Matrices and determinants I. Matrices 2. Operations on matrices Partitions of a matrix 3. 4. The determinant function Properties of the determinant function 5. Expansion of det A 6.

39 39
41

46 47 49
53

vi'

viii

Contents

Part 11

Groups
61 61

Chapter 4

Groups I. Semigroups and groups 2. Homomorphisms 3. Subgroups and cosets
4.

69 72

5. 6.

Cyclic groups Permutation groups Generators and relations

82 84

90
91 91

Chapter 5 Normal subgroups 1. Normal subgroups and quotient groups 2. Isomorphism theorems 3. Automorphisms 4. Conjugacy and G-sets Chapter 6 Normal series 1. Normal series Solvable groups 2. 3. Nilpotent groups Chapter 7

97 104 107
120 120 124
126

Permutation groups 1. Cyclic decomposition 2. Alternating group Simplicity of 3.
Structure theorems of groups 1. Direct products 2. Finitely generated abelian groups 3. Invariants of a finite abelian group 4. Sylow theorems 5. Groups of orders p2. pq
Part 111

129 129
132 135

Chapter 8

138 138
141 143

146 152

Rings and modules
159 159
161 163 168

Chapter 9

Rings
1.

2. 3.

4.
5.

Definition and examples Elementary properties of rings Types of rings Subrings and characteristic of a ring Additional examples of rings

176

Contents

ix

Chapter 10

Ideals and homomorphisms
1.

2. 3. 4. 5. 6.

Ideals Homomorphisms

Sum and direct sum of ideals Maximal and prime ideals Nilpotent and nil ideals Zorn's lemma

179 179 187 196

203 209 210

Chapter II

Unique factorization domains and euclidean domains 1. Unique factorization domains Principal ideal domains 2. 3. Euclidean domains 4. Polynomial rings over UFD
Rings of fractions 1. Rings of fractions 2. Rings with Ore condition
Integers 1, Peano's axioms Integers 2.

212 212 216 217 219

Chapter 12

224 224 228
233 233

Chapter 13

240 246 246 248
253 260 263 268 273

Chapter 14

Modules and vector spaces 1. Definition and examples 2. Submodules and direct sums 3. R-homomorphisms and quotient modules 4. Completely reducible modules 5. Free modules 6. Representation of linear mappings 7. Rank of a linear mapping

Part IV
Chapter 15

Field theory
281

Algebraic extensions of fields 1. Irreducible polynomials and Eisenstein criterion 2. Adjunction of roots 3. Algebraic extensions 4. Algebraically closed fields

281

285 289 295

2. Preliminaries 2. 3. HomR 2. 5. Fundamental theorem of Galois theory 3. Symmetric functions Ruler and compass constructions 5. Row module.x Contents Chapter 16 Normal and separable extensions 1. Smith normal form Chapter 21 Finitely generated modules over a PID Decomposition theorem 1. 4. primary modules. and Noether—Lasker theorem 367 367 368 382 388 392 392 393 394 Chapter 20 Smith normal form over a PID and rank 1. Generalized Jordan form over any field 402 402 404 408 409 418 . 2. 2. and rank 3. 4. Fundamental theorem of algebra Applications of Galois theory to classical problems Roots of unity and cyclotomic polynomials 1. 3. Cyclic extensions Polynomials solvable by radicals 3. Uniqueness of the decomposition 3. Additional topics 322 322 330 338 Chapter 18 340 340 344 348 355 358 Part V Chapter 19 Noetherian and artinian modules and rings 1. 2. Application to finitely generated abelian groups 4. Splitting fields Normal extensions Multiple roots Finite fields Separable extensions 300 300 304 307 310 316 Chapter 17 Galois theory Automorphism groups and fixed fields 1. 4. Noetherian and artinian modules Wedderburn—Artin theorem Uniform modules. Rational canonical form 5. column module.

Tensor products 2. 4. Tensor product of homomorphisms 5. Tensor product of algebras 426 426 428 431 433 436 438 476 477 Solutions to odd-numbered problems Selected bibliography Index . Module structure of tensor product 3.Contents xl Chapter 22 Tensor products Categories and functors 1.

.

The new problems include several that relate abstract concepts to concrete situations. Finally. R. we should mention the combinatorial applications of the Burnside theorem to real. More than 150 new problems and examples have been added. This chapter provides basic results on tensor products that are useful and important in present-day mathematics. Chapter 22 on tensor products with an introduction to categories and functors is a new addition to Part IV. Among others. due to Osofsky. P. In particular. life problems. B. Lopez-Permouth for his help during the time when the revised edition was being prepared. We are pleased to thank all of the professors and students in the many universities who used this textbook during the past seven years and contributed their useful feedback. we would like to acknowledge the staff of Cambridge University Press for their help in bringing out this second edition so efficiently. Jam S. Nagpaul xli' . the division algorithm and greatest common divisors in a given euclidean domain. Bhattacharya S.Preface to the second edition The following are the main features of the second edition. of the celebrated Noether—Lasker theorem. K. A proof for the constructibility of a regular n-gon has been included in Chapter 18. We have included a recent elegant and elementary proof. we present applications of G-sets. we would like to thank Sergio R. In particular.

3) is a prerequisite for the rest of the book. number systems. illustrated by numerous examples. The invariants of a finite abelian group and the structure of groups of orders p2. It is expected that students would already be familiar with most of the material in Part I before reaching their senior year. n > 4. The book is divided into live parts (see diagram). Chapter 12 deals with the ring of fractions of a commutative ring with respect to a multiplicative set. where p. In Chapter 2. and the nonsolvability of S. Chapters 9— II cover the basic concepts of rings. Part III (Chapters 9— 14) deals with rings and modules. maximal ideals. pq. including prime ideals. Part I (Chapters 1. or simply referred to as necessary. are proved in Chapter 7. it can be completed rapidly. Part II (Chapters 4—8) deals with groups.. It contains an informal introduc(ion to sets. q are primes. some important algebraic properties of integers have been proved. Normal series. Results proved in Chapter I include the Schröder-Bernstein theorem and the cardinality of the set of real numbers. It is self-contained and covers the topics usually taught at this level. UFD. including G-sets and their applications.Preface to the first edition This book is intended for seniors and beginning graduate students. Chapters 4 and 5 provide a foundation in the basic concepts in groups. Chapter 13 contains a systematic development of xiv . The simplicity of the alternating group A. starting from the well-ordering principle of natural numbers. PID. and the Sylow theorems. Therefore. and determinants. and so forth. matrices.. Chapter 3 deals with matrices and determinants. and the Jordan—Holder theorem are given in Chapter 6. Chapter 8 contains the theorem on the decomposition of a finitely generated abelian group as a direct sum of cyclic groups. are given as applications. skipped altogether. solvable groups.

'U L) U- 0 LU Q z z 'U LU 'U U- C LU I- I .

The book can be used for a one-year course on abstract algerba. and IV. Numerous examples have been worked out throughout the book to illustrate the concepts and to show the techniques of solving problems. including existence and uniqueness of algebraic closure. starting from Peano's axioms. ll. Therefore. Chapters 15 and 16 contain the usual material on algebraic extensions. it provides flexibility in selection of the topics to be taught. The material presented here is in fact somewhat more than most instructors would normally teach. Part V (Chapters 19—21) covers sOme additional topics not usually taught at the undergraduate level. and rank. Chapter 21 gives the structure of a finitely generated module over a ND and its applications to linear algebra. 10. Chapter 19 deals with modules with chain conditions leading to the Wedderburn—Artin theorem for semisimple artinian rings. .and 14(Sections I—3). Part IV (Chapters 15-18) is concerned with field theory.field theory — Chapters 15. If reference is made to a result occurring in a previous chapter. Chapter 17 gives the fundamental theorem of Galois theory and its application to the fundamental theorem of algebra. Chapter 18 gives applications of Galois theory to some classical problems in algebra and geometry. Topics discussed include completely reducible modules. Parts II and Ill arc almost independent and may be studied in any order. 16. The book has evolved from our experience in teaching algebra for many years at the undergraduate and graduate levels. We have also provided solutions to the odd-numbered problems at the end of the book.and 18 (Section 5). rings. A two-semester course in abstract algebra can cover all of the material in Parts II. The precise dependence of Part IV on the rest of the book can be found from the table of interdependence of chapters. and vector spaces. III.xvi Preface to the first edition integers. and normal and separable extensions. 5. A two-quarter course in abstract algebra may cover the following: groups — Chapters 4. We hope these will be used by students mostly for comparison with their own solutions. then only the chapter number is mentioned alongside. Chapter 14 is an introduction to modules and vector spaces. and 7 (Section 1) and 8. The material has been class tested through mimeographed notes distributed to the students. It can be studied after acquiring a basic knowledge of groups. rings — Chapters9. In all cases the additional information needed to identify a reference is provided. Part IV requires a knowledge of portions to Parts 11 and Ill. Chapter 20 deals with the rank of a matrix over a PID through Smith normal form. There are also many challenging problems for the talented student. lemmas. and examples is done afresh in each chapter by section. Numbering of theorems. free modules.

R. Nagpaul . 1965. MA. Jam S. and S.Preface to the first edition xvii We acknowledge our indebtedness to numerous authors whose books have influenced our writing. Addison-Wesley. It is our pleasant duty to express our gratitude to Professor Donald 0. Stephanie Goldsbcrry for the splendid job of typing. Cohn's Algebra. Lang's Algebra. we mention P. We also express our gratefulness to Ohio University for providing us the facilities to work together in the congenial environment of its beautiful campus. whose encouragement and unstinted support enabled us o complete our project. Bhattacharya S. Reading. Ohio University. 1977. Vols. 2. M. In particular. We also thank Mrs. I. Norris. B. Chairman. During the preparation of this book we received valuable help from several colleagues and graduate students. New York. 1974. P. Department of Mathematics. K. John Wiley. We express our gratitude to all of them.

)jEA intersection of (Xjj€A 3 (x E A IP(x)) 9(x) (X.Glossary of symbols V forall there exists is an element of is not an element of set of all x E A satisfying condition P(x) power set of x family indexed by set A union of (X.)IEA LEA U X1 lEA fl XXY 0 corç ç J or Cartesian product of X and Y empty set isasubsetof is a proper subset of contains properly contains implies if and only if if and only if image f:X—'Y f(x) if fisamapofXintoY Y y=f(x)wheref:X—'Y.x€X.yEY composition Euler's function xviii .

Glossary of symbols xix (a.. the degree of extension of L over K normalizer of S (in H) conjugate class of S(with respect to I!) product of (A's) ISA X1 1€ A det sgnc e.. the dimension of a vector space Lover K. the greatest common divisor of a and b.. the ring of n X n matrices over S alternating group of degree n dihedral group of degree n group of invertible m X m matrices over F center of G is a normal subgroup of quotient group (ring. module) of A modulo B in groups. as a ring... 0 elsewhere the set of positive integers (I . in fields.2.1 n N Z Q R C Z/(n) or lxi or card X IGI (SI A. according as the permutation is even or odd square matrix with I in (i.b) in number theory. in rings and modules.j) position. in vector spaces. the ideal or submodule generated by a and b alb a%b adividesb a does not divide b Kronecker delta determinant ± I. GL(m.n) set of all natural numbers set of all integers set of all rational numbers set of all real numbers set of all complex numbers the cardinal of the continuum (cardinality of the reals) integers modulo n cardinality of X order of groupG subgroup generated by S cyclic group of order n as a group. the index of a subgroup K in a group L. F) Z(G) <1 A/B [L:K] N(S) (NH(S)) C(S) (CM(S)) ISA fl direct sum of(Xt)IEA Im Ken f C image of homomorphism f kernel of homomorphismf is isomorphic into (embeddable) is isomorphic onto . the symmetric group of degree n.3.

Y) rationals between 0 and I of the form rn/ps. G(E/F) M®RN 0 end of the proof .0 A polynomial ring over R in one indeterminate x polynomial ring over R in n indeterminates. mn> 0.. R([xJj R<x> HomR(X. x1 formal power series ring ring of formal Laurent series RtxJ R[x1 .. under the binary operation "addition modulo 1" set of all R-homomorphisms of R-module I to Rmodule Y set of all homomorphisms of I to V endomorphisms of X Hom(X.Y) End(X) Aut(X) R5 F(a) F(SJ F(S) automorphisms of I localization of a ring R at S subfield generated by F and a subring generated by F and S subfield generated by F and S Galois field (finite field) with q elements algebraic closure of F fixed field of H Galois group of automorphisms of E over F cyclotomic polynomial of degree n tensor product of MR and RN F E41.xx Glossary to Symbols (S) (S)1 lEA opposite ring of R ideal (submodule) generated by S right ideal generated by S left ideal generated by S sum of right or left ideals (submodules) (X1).

PART I .

.

2. For example. we write B = (x E A IP(x)).1. we say x belongs toS.2.3. and we write x E S. Instead we shall assume an intuitive understanding of the terms "set" and "belongs to. IfS is a set and xis an element of the set S. repetition of an element has no effect. Given a set A and a statement P(x). It is not our purpose to present here an axiomatic account of set theory. There are standard symbols for several sets that we frequently deal 3 .3) denotes the set whose elements are 1. we say that a set is a collection of objects (or elements). Thus.2. we write x S. and 3. Also. The order in which the elements are written makes no difference. In symbols. we sometimes also write B = (xlP(x)).2. When the context is clear. XE A XE B.2. (1. We say that A and B are equal. there is a unique set B whose elements are precisely those elements x of A for which P(x) is true." Informally speaking. if they consist of the same elements.3).") A set is thus determined by its elements. A set with a finite number of elements can be exhibited by writing all of its elements between braces and inserting commas between elements.3) denote the same set. (1.2) is the same set as (1. Ifx does not belong to S.3) and (2. that is. Let A and B be sets. written A = B. Thus. An element ofa set Sis also called a member ofS.CHAPTER 1 Sets and mappings I Sets The concept of set is fundamental in mathematics. (1. (The symbol stands for "if and only if.

. A is called a subset of B if every element of A is an element of B. if A is a subset of B. where a. R is the set of all real numbers. The empty set is also called the null set or the void set. fractions a/b. or if its elements can be listed (counted. Thus. the empty set 0 is a subset of every set because the condition x E 0 xEA is satisfied vacuously. (In such a case we say that the condition is satisfied vacuously.2.n for any n whatsoever is called an infinite set.2. labeled) by natural numbers 1... Moreover.. IfS and Tare both empty sets. then S = since the condition x E S x E Tis satisfied because there is no element x in either Sor Tto which the condition may be applied.± 1... It follows immediately from the definition that A and Bare equal ifand only ifA C B and B C A. A set S is said to be empty ifS has no elements. but S '# A.±2 Q is the set of all rational numbers — that is. C is the set of all complex numbers x + iy.) Because any two empty sets are equal.b are integers and b # 0.. that is. (Some authors write A C B. and the process of listing stops at a certain number.n) is denoted by n. Some of these are given below. the set (I . and we n. A set whose elements cannot be listed by the natural numwrite IS1 bers I . say n. Others will be introduced subsequently as the occasion arises..y are real numbers and i2 = — 1. there is just one empty set. N denotes the set of all natural numbers 1. the statement xE S is not true for any x.2. which is denoted by 0.2. Let A and B be sets.. For any positive integer n. we also say that B contains (or includes) A. where x.3 Z is the set of all integers O. we write A C B. then S is a proper subset of A written as Sc A. Definition. A set S is calledflnite if it has no elements.. every set is a subset of itself.3. and we write B A (or B A). Let A andB be subsets ofa set U.4 Sets and mappings with. (The symbol stands for "implies. The union ofA and B.") If A is a subset of B. IfS CA. that is. . if aEA a E B. is AUB(xE UIxEA orxEB).) Further. The number n is called the cardinality of S.. Definition. written the set A U B.

.

Likewise. AU(AflB)=A=Afl(AUB). (ii) X—(AuB)=(X—A)n(X—B). (iii) Proof of(i): XnA. B. and C. X — (A U B) = (X — A) fl (X — B). and X be sets. B. B. Let A. Then (i) X—(X—A)=XnA. Moreover. LetS be a Set whose elements are themselves sets.l).(AnhJ)nC=An(BnC). 0 Inviewoftheequality(A UB)UC=A U(BUC)(Theorem l. The union of all sets in S is defined to be the set {xlxeXforsomeXinS) . Hence. A fl(BUC)=(AflB)u(A flC). B. Definition. The last part is proved similarly. it is clear that the set A U B U C consists of all those elements that belong to at least one of the sets A. is the set of those elements that belong to each of the sets A. (iv) (v) AU(BflC)=(AUB)fl(AuC). and C. B.6 Sets and mappings (iii) (AUB)uC=Au(BuC).2 Theorem (DeMorgan's rules).we can do away with the parentheses and simply write A U B U C to denote unambiguously the union of the sets A. 0 Proof of (ii) Xand xEA xE X — A and x B x E (X — A) fl (X — B). A fl B fl C. the intersection of A. and C. as Proof Left an exercise. 0 1. and C. This suggests the following definition for the union and intersection of an arbitrary number of sets. Hence.X—(X—A)=XflA.

('onsequently.Sets 7 and is denoted by the set The intersection of all sets in S is defined to be S) and is denoted by flXESX. The set of all subsets ofX is called the power set ofX and is denoted by X).. IfS contains only a finite number of sets. let X = (1. andX.. Let X be afinite set having n elements.1 X be a set. Recall that the empty set 0 and the set X itself are subsets ofX and are therefore elements For example. then {a) E 1. = (0. Hence. Hence. Then the subsets ofXareO. If a E X. where 0 r n.{t). then has just one element: 0. It should be noted that a and (a) are not the same. Then )t = subsets. (2). the total number of subsets ofXis On putting a = I in the binomial expansion (I we get + . It is shown in high school algebra that the number of ways in which r elements can be selected out of n elements is (n\ n! r!(n—r)!' which is therefore the number of subsets of X having r elements each.3 2 Theorem. orX1 U U their union is and their intersection is written fl7. has Proof Let us first consider those subsets of X that have r elements each.(2}. That is. IfXis the empty set 0..( 1 ).X).. say X1 written LJ7.2).

Proof Exercise. the following statistics were revealed: 60 were taking English. During quality control checking of a sample of 1000 TV sets.8 Sets and mappinp This proves that I has exactly 2" subsets. 75 sets had a defective sound system. 20 sets had a defective picture tube and a defective remote control. find the number of elements in A U B. 5. If a set A has m elements and a set B has n elements. 6 were taking both English and physics but not French. 1.4 Theorem. Let it be a set whose elements are nonempty andø it. and the The next theorem follows immediately from the definition. Let it be a set whose elements are only ([each element nonempty subsets ofX.(4. After the registration of 100 freshmen. subsets ofX. that is. 15 were taking physics and French. For example. (a) Show that 54 were enrolled in only one of the three subjects. it has only the partition it Problems Prove Theorem 1. 4. 15 sets had a defective sound system and a defec1. it was found that 100 sets had a defective picture tube. it C Definition. 0 0. and 10 were taking all three subjects. = 21X1 explains why the set of all sub- sets of Xis called the power set of X. 44 were taking physics. 30 sets had a defective picture tube and a defective sound system. lithe elements ofit are pairwise elements of it are called blocks of the partition it.1. 3. then it is called a partition of I. Then it is a partition ofl of I belongs to exactly one element of it. 30 were taking French. Hence. the equality has 2" elements. Let I be a set. 2.2). Assume that A fl B has k elements.(3). Prove Theorem I . Note that if X is empty. (b) Show that 35 were enrolled in at least two of them. the set it = ((l. 80 sets had a defective remote control system.5)) isa partition of the set X (1 disjoint and their union is I. = 21X1 0 we get Since n = Incidentally. Let X be a set.2(iii). . 24 were taking English and French.

805 sets had no defects. then trivially (a.b)) is called an ordered pair and is denoted by (a. AX B((x.b) = (c. We now show that (a.c) then AXB= The term "cartesian" is borrowed from coordinate geometry.(a. is called the cartesian product ofA and B.b) = (c. So a = c = d and a = b = c. Then ((a). then we must have (a.(a. 35 sets had a defective sound system only. This yields a = c.{c.d). . and is denoted by A X B. where x C A andy B. 50 sets had a defective remote control only. If a = c and b = d.( I . A relation from A to A is called a relation on A (or in 4).y) called its cartesian coordinates. ((1 .d). on the other hand. E R.b).d). If.Relations tive remote control system. Then the set ((a).b be elements of a se( S.b). b = d. Then R is called a relationfrom A to B.b) = (c.(2. Definition.y)Ix€A. Conversely. The cartesian product R X R is then the set of cartesian coordinates of all points in the plane.b) = (c). Use (a) (b) (c) (d) (e) Venn diagrams to show that 195 sets had at least one defect. In symbols. Let A and B he sets. and B (a.2) I .d). let (a. in that order. The set of all ordered pairs (x.a). Relations 2 Definition. Let A. By definition of equality of sets. Let a. then xis said to he in relation R toy. If (a) = (c).h.b). and b is the second component (or coordinate).c). (a) = (c. and 5 sets had all three defects.y). this implies (a) = (c) or (a) = (c.c)). 55 sets had a defective picture tube only. which implies a = c = h = d. then we must have (a.B be sets.( if A = (1.d) if and only if a = c and b = d.d). For example. and let R he a subset of A X B. where a point in the plane is represented by an ordered pair of real numbers (x.yE B). written x R y.a). Definition. a is called the first component (or coordinate).(2.b)) = ((c).(2.d)).b) = (c.

symmetric if x R y implies y R xfor all x. If R is reflexive. congruence of triangles and similarity of triangles are equivalence relations.y E X. antisymmetric. . Let us further agree that every line is parallel to itself. x is said to be congruent to y (modulo n). therefore it is a partial order on X.y E Xlet xIIy mean that xis parallel toy. set inclusion is reflexive. and transitive. symmetric.z E X." For any sets A. written x (i) (ii) (iii) y (mod n). if n divides x — y.thenAcC.y.y. (c) antisymmetric (Ix R y andy R x imply x = yfor all x.y E X. B"=D. For any x.C E Xwe see that (i) ACA. and transitive. then n divides y — x. then n divides x — z. Let n be a fixed positive integer. antisymmetric. 2. For x. (a) (b) If R is reflexive.10 Sets mid Strictly speaking. if n divides x — y. (c) The relation ("less than or equal to") on the set R of real numbers is reflexive. Let R be a relation in the set X.z in Z. then R and S are equal ifA =C. (ii) (iii) if A C B and B C A. antisymmetric. R is said to be reflexive tfx R xfor all x E X.B and a subset R of A X B.and.1 Examples (a) Let Xbe the set of all lines in a plane. a relation is determined by three sets. (d) The relation congruence modulo n on Z is defined as follows. and transitive. ifACBandBCC. Now. Definition. x = x (mod n).y E Z. (b) Let X be a set whose elements are themselves sets. it is a partial order on R. for any x. and S is a relation from C to D.B. and transitive. Consider the relation C determined by "set inclusion. A. although we call it simply the relation R. it is true that n divides x — x = 0. then R is called a partial order on X. If R is a relation from A to B. Hence. hence. Then II is an equivalence relation on X. (d) transitive if x R y andy R z imply x R zfor all x. if n divides x — y and also y — z. therefore. then A B. then R is called an equivalence relation on X. Similarly.

A lower bound and a greatest lower bound (g.Lb. It can be easily shown that a l. then xis said to be properlycontainedin y.6) ordered by divisibility. join x andy by a straight segment whenever y covers x.b. we give below the diagrams for the following three posets: (1.2.) (1. ordered by set inclusion.) are defined analogously.) of S. a poset.Relations II This proves that congruence modulo n is an equivalence relation on Z.b. written x < y. If x < y and there is no element a in X such that x < a <y. (The symbol here denotes an arbitrary partial order and does not necessarily have its usual meaning of less than or equal to" in real numbers.) of S is an element rneX such that(i)x <rn for all xeS and (ii) if x <rn' for all then m rn'.3. then xis said to be contained in y. Let S be a subset of X.) or simply the poset X.3) (1 (2. (g.2. then y is higber than x in the diagram.4. if it exists. A least upper bound (l.3) (I 3> 34 2t (a) 0 (b) (c) Let (I. (1. Represent each element in I by a small circle (or a point) in such a way that whenever x < y.4.5. briefly. and let x.6) ordered by the usual relation of "less than or equal to". and let be a partial order on X. ) be a partially ordered set. As an illustration. Let be a poset.I. Further. .) The set X together with the partial order is called a partially ordered set or. (c) (a) (1.u. We refer to it as the poset (X. then y is said to cover x.yE X. is unique. A finite poset X can be represented by a diagram in the following manner.u.b.2.5. An upper bound of S is an element beX such that x b for all XES. Ifx y and x # y.3. Let X be a set.

--n+2. let X = Z.. in this example. Suppose x y means x y (mod n).). For any set S. Thai is. there is a unique equivalence relation E such that X/E = it.e.. given a partition it of X. y€X. A partially ordered set L is called a lattice if every pair of elements in L has a least upper bound and a greatest lower bound. Conversely. That is. Then is an equivalence relation on Z [Example 2.—2n+2.. The following theorem shows that this property holds for every equivalence relation. n divides x — y. The diagrams (a) and (c) above represent lattices. The set of all equivalence classes of E in X is called the quotient set of X by E and is writien X/E. is a lattice. either x y or y x is called a chain or a totally ordered set. Let E be an equivalence relation on a set X.2 Theorem. that integers a. the power set Here A vB=AuB and A A ordered by set inclusion. A subset C of X is called an equivalence class of E (or an E-class) in X if C = E(a) for some a in X. A poset (X. respectively). E(a)=(xEXIxEa)..12 Sets and mappings Definition.n+2. 2. The equivalence classes Ii.. Then X/E is a partition of X. or Z/(n)J are called the We observe.. ) in which for all x. For example. Let E be an equivalence relation on a set X. is the set denoted by — 1)).2. that is. the elements Of residue classes of integers mod n. and let n be a fixed positive integer. . where 0= 2=(. and any two equivalence classes are either the same or disjoint. The least upper bound and the greatest lower bound of {x. The set of all elements in X that are in relation E to a is called the equivalence class of a under E and is denoted by E(a). X/E=(CC X]C= E(a)forsomea€X).. Definition. y} are written x v y and x A y.b belong to the same equivalence class if and only if they differ by a multiple of n.2n+ 2. Every chain is a lattice. Let x. and so on. respectively (called join and meet of x. and let a E I. for all A. y.y E Z. l(d)J. The set of equivalence classes or Z/(n).

Let A = [0.yare in the same block of it..B.4... E(a) C E(b). for every x E E(a).d) ill a+d=b+c is an equivalence relation. Conversely. a E E(a). 6.5.Relations 13 Proof For any a E X..xRx. Show that XX YCAXB. Therefore. E(b) C E(a). Foreveryx. 3.6. does it necessarily followthatXCAand YCB? 4. by Theorem 1. Show that the relation defined on N X N by (a.y are in the same block of it xE' y.—7.xRyandyRzimplyzRx.—5. E(a) = E(b). This proves that each a in X belongs to exactly one equivalence class. namely. Hence. B = (0. hence. By symmetry.. hence...5).8. Suppose E' is also an equivalence relation on X such that X/E' = it.) U (.—8.—3. x E a. 0 Problems 1. Therefore E is the unique equivalence relation on X such thatX/E=. Hence. b E a. therefore.l. Define the relation E as follows. we have X X V C A X B..—4..) U {. Suppose the following properties hold: (i) (ii) ForeveryxES.y. 2. let it be a given partition of I.z€S. Then a E b. hence.. Let Ube the set of all people..3.y E X. X/E = it. X/E is a partition of X.. Clearly.. Let Sbe a nonempty set.. a E b.7.. Hence. Hence.—-2..b)—(c. Y.A.— 1.0.). what isA X B?lfA X B = 0. . and the equivalence classes of E are the blocks of it. Draw the graph of A X B in the xyplane.—6. a E a. E is an equivalence relation on X.2. and let R be a relation on S. Then for all x. Is A X B = B X A? IfeitherA orB = 0.1). 7.4.. what can you say aboutA and B? Suppose XcA and YCB.. Suppose a E E(b). E = E'. is a roommate of is a friend of Describe the equivalence relation corresponding to the following partition of Z: (b) (c) (.... For any x. and..r. x E y ifand only ifx.3. x E E(b). E(a).. If for given sets X.) U (.. Which of the following relations on U are equivalence relations? (a) is an uncle of 5.y X. xEy x.

respectively. If V is the image of x underf we writef(x) = y and say that ftakes x to y. in symbols.4LB. which attempts 9. A mapping f X —. Is (a) reflexive. if (c) transitive? If so. prove that any equivalence relation satisfies (I) and (ii). 1ff is a mapping from A to B. they are not equal even iftheir domains are equal andf(x) g(x) for every x in the domain. iff(x) = y. If R = 0. G is (the relation)f itself interpreted as a subset ofA X B.14 Sets and mappIngs Prove that R 8. ifA = C. Determine the fallacy in the following argument. Again. then x is called a preimage of y. X such that f(x) = x for all x E X is called the identity mapping on X and is denoted by (or Ii). write down the quotient set Z/—. We emphasize that if the mappingsf and g have different codomains. The graph of a mappingf A —' B is defined to be the set G = ((x. by transitivity x R y and y R x imply x R x. is an equivalence relation. The sets A and B are called. Also.y) E A X BIf(x) = y}. As a matter of fact. 3 Mappings Delinti ion.] Define a relation in the set of integers Z as follows: a b a + b is an even integer. X'I-'y or simply x y. then R is symmetric and transitive but not reflexive. andf(x) = g(x) for all x E A. we write fA—'B or . B = D." [Hint: A relation R is a subset of XX X. Let A be a nonempty subset ofa set X. Two mappingsf A —'Band g: C—' D are said to be equal. A relaiionf from A to B is called a mapping (or a map or a function) from A to B iffor each element x in A there is exactly one element y in B (called the image ofx underJ) such that x is in relation fto y. the domain and codomain of the mappingf Let XE A. to show that a symmetric and transitive relation R on a set X implies that the relation is also reflexive: "By symmetry x R y implies y R x. Letf A B. Xsuch thatf(a) = a for each a E A is called an inclusion (or insertion) map of A intoX. Let A and B be sets. written f= g. Then the mappingf A —. . (b) symmetric.

Therefore. we have the result that For example. Then the mapping h: A —' C given by h(x) = g(f(x)) for all x E A is called the composite off followed by g and is denoted by g a [(or more commonly by gf). The set of all possible mappings from X to S is denoted by SX. let X = (0. The mappings land g are called factors of the composite h g of Note that the notation for the composite mapping is such that the order in which the mappings act is from right to left. This just means that assigned images in n X n X S there are exactly ntm distinct mappings from X to S. the empty mapping in which there is no element to which an image is to be assigned. facts first. Then there are 22 = from X to X: 4 distinct mappings /:X—"X f4:X—. The image ofany given element in Xcan be any one of then elements inS.Mappings 15 Let fi A —' B.' =f(x) for some A). That is. them elements in Xcan be X n = nm ways. S. lmf= (yE Be. Let X and S be sets. Thus.X withO'—'O. g: B C can be represented by . The composite g ofis defined whenever the domain of g contains the range off The composite of the diagram —° B. In go f. g of(x) = g(f(x)) for all x E A. On the contrary. Thus. but the definition of a mapping is satisfied vacuously. S. with0'—.1). The subset of B consisting of every element that is the image of some element in A is called the image (or range) of the mapping f and is denoted by lmf That is. This might seem strange. We observe that if X is empty. then there can be no mapping from X to S. l'—'O. A —' B and g: B —' C be mappings. there is just one mapping X namely.l. and fi X —. Note that this is true even if S is also empty. ifS is empty but X is not empty. l—'l. withO'—O. withO'—l. for finite sets X = and 5.

16 Sets and mappings One can go from A to C directly. A mapping that is infective and surjeclive is said to be bijective. C. g. y =f(x) for some x E A. Let x E A. Hence. x1 x2 #f(x2) (b) or.1 Theorem. Letf A B. surjective (or onto) '1 for every y E B. h(gf) and (hg)f have the same domain A and the same codomain D. h(gfXx) = h(gf(x)) = and (hg)f(x) = hg(f(x)) = h(g(f(x))). It is clear that the diagram commutes if and only if h = g of In the sequel we denote the composite by gf(instead of g of). 3.I) . h(gf) = (hg)f Definition. A bijective mapping .ifor all x1 . h( gf Xx) = (gf)f(x) for every x in A. B —. f(x1) =f(x2) x1 = x2. and I:: C D.x2 E A. we may writef A B. by definition of the composite. the image of any x E A is the same along either path — then we say that the diagram commutes. therefore. A —' B is a bijective mapping. Then. A mappingf A (a) 0 B is injective (or one-to-one or I . Then h(gf) = (hg)f Proof Clearly. equivalent/v. or along the path If the result is the same—that is.

be a mapping from the set of natural numbers to itself. Then f is injective but not surjective. f2(a). For example. bijection). we sayfis invertible. which proves that f is surjective and. in general.. where r> s. because f is injective. (iii) f is bijective if and only if every element in B has exactly one preimage. 1ff has an inverse. Now f°(a). bijective) mapping is also called an injection (surjection.. then its inverse. A mapping g: B A is called an inverse off if and gf—i4. f(a). are all in a finite set A.. 0 An infective mapping from a finite set to itself is Theorem. (ii) f is surf ective if and only if every element in B has a preimage. Leif A —' B. Leif? A — B. hence. The elements a = and. is unique. there must be repetitions. Proof Obvious. Otherwise.Mappings A 17 A is called a permutation of A. A bifective mapping is also called a one-to-one correspondence. Definition. An infective (curjective. defined byf(x) = x + 1. The following theorem characterizes injective. 3. Proof Letf A —. 3. Why? Recall that denotes the identity mapping on X. say. bijective. Ietf N N.3 does not hold for infinite sets. consider the effect of performing f repeatedly. We shall find b C A such that f(b) = a. . Let a C A. and bijective mappings. 1ff is invertible.3 bijective. Then f is infective if and only if no element in B has more than one preimage. Ifs = 0 then choose b = f''(a). if f'(a) =fs(a) where b fr—s—I(a) 0 We wish to remark that Theorem 3.2 Theorem. hence. surjective. Let us write f2 for if (with n factors) as ffl. A be injective. So assume that f'(a) — f3(a).

Then finduces the mapping g: T given by g(x) =f(x) for all x E S. Let ScA and TC Bsuch thatf(x)E TforallxES. Leif. A A —* B is infective if and only (if has a left inverse. gf(x) = gf(y) Conversely. Therefore. A (a) (b) — B.5 (i) (ii) Theorem. where a is the unique preimage of b B under f Obviously. The mapping g is called a restriction off (b) There is an important mapping determined by a subset of a set. Let Sc X and let A = (0. . Hence. is a left inverse as well as a right inverse off 3. the inverse off if it exists.B. In the next section we describe some of these induced mappings. letfi A —.4 Theorem.18 Sets and mappings 3. Proof Exercise. as we say — by another mapping or by some inherent property of the domain or codomain. A is a left inverse off (1 gf= a right inverse off zffg = i5. A mapping g: B —. Define g: B — A as follows: Let b E B and define g(b) = a. f is 1-1. if z B. 0 We shall frequently come across mappings that are determined in a natural way — induced. surf ective if and only iff/sas a right inverse. g is the inverse off 0 Definition. then fg(z) = z.! is a bijection. B be invertible with gas its inverse.1). Then Sdetermines the mapping f5: X A given by f5(x)=l The mapping =0 ifxES. Further. f is invertible if and only (if is bijective. which implies that g(z) E A is a preimage of z underf This provesfis onto. B be a bijection. is called the characteristic function of S (as a subset of X). Clearly. Thenf(x) =f(y) Proof Letf: A x = y.6 Examples of induced mappings (a)Letf A —. 3.

y)ESX T. It is usual to drop the subscript * and to denote the induced mapping by I also. The context usually makes clear whetherfdenotes the mapping A —' B or The mappingf: A B also induces the induced mapping . (x. B. let us define AXB SXT with for all (x. That is. The mappings p and q are called the projections from S X Tonto Sand T. place ofA. with x '—' E(x). with forall(x. there are two natural mappings with domain there S X T— namely. Thus. (x.g). Then it is clear that the characteristic function of S is the given mapping the unique subset of Xwithfas its characteristic function. Then E induces the suijective mapping p: X — X/E.g(v)) A X B. Note. (d) Given sets S and T. that for every singleton (a) C = (f(a)). The mappingp is called the canonical(or natural) mapping from X to the quotient set X/E.2) in the mapping yi: is bijective. given by S is clear that we could have used the set 2 = (1. let S=(xEXIf(x)= f Moreover.Mappings 19 Conversely. where E(x) is the equivalence class ofx under E. Sand g: B —' T.y) The mapping is called the cartesian product offand g and is denoted by fX g or (f. f(S) = (yE Bly =f(x) for some x E S). the set is also written as (c) Let E be an equivalence relation on the set X. and 2X• Beis a one-to-one correspondence between the set cause of this. Ax.y)'—'y and q: SX T— T with forall(x. (f) Letfi A —. A. S is This proves that It 1). Then f induces a mapping given by fe(S) = (yE Bly =f(x) for some xc S) 1mJf4(ø) = 0. given a mapping X —.y)ESXT. and for every Sc A.v) '—p (f(x). (e) Given mappingsf A —. respectively. in particular.

One may visualize a set of pigeonholes or post office boxes. is written or A1 X A2 X .. namely. each labeled with exactly one element ofX. are called members or elements of the list. When we refer to the mappingf X — S as a family.. in general. so that used f Once again. The cartesian product of a finite family (that is. As usual. we think of the elements of X as labels for locating elements of S.. the cartesian product fl.. The difference lies in the point of view. always exists whetherfis invertible or not. But it should be carefully noted in this sense.. =f(i). that is." since. Note that each element of is a family where a1 E A1. A mappingf n elements of S. More generally.... In fact.n). f: given by f*(T)=(xEApf(x)E T) forevery TCB.. A mapping N —.. as before.. = A for each i E X. Let S be a nonempty set and n a positive integer. A list of length n is also called an n-iuple.f2. is in fact Ax.A2. In the particular case when A. let n denote S is called a list (or an ordered set) of the set (1 . a list) of sets A1 . The element f(i) is usually written J. In each box we put exactly one element of S (with repetitions allowed. and the list f is exhibited as The positive integer n is called the length of the list. there seems to be no difference between a family and a mapping. The cartesian is defined as product denoted by element in several IEX fi A1={f IEX UAjf(i)EAI for each iEX}.). the induced mappingf is commonly '(T) = (x E A f(x) E T). a mapping!: X —p S is called afamily of elements of S and represented as The set Xis called the index set of the and fis said to be indexed by X or X-indexed.. we see that the elements of the index set A' serve as addresses for the locations of the elements of S in the range of! Let be an arbitrary family of sets indexed by X. the induced mappingf *: —. Thus. S is called a sequence of elements ofS and is written (J or .20 Sets and mappings a mapping in the reverse direction.. is not the inverse offas defined earlier. One may wonder why we introduced the new term "family. X . andf . we may put the same being the clement in the box labeled i.EXA..

Let C[0. Again. 3. 5.l) be the set of all continuous functions from the closed interval [0. intersection. show that J'has a left inverse but no right inverse. what can be said about f and g? Prove Theorem 3. and so on. B and g: B — A be mappings such that gf— 14. Find the number of injective mappings from a set with in elements to a set with n elements if(a) m = n.Binary operations 21 Problems I. Also show that f has infinitely many left inverses. 9. B and g: B — C be mappings.2. 2. given and difference are all binary operations on the set . IfA and 13 are subsets of X. Addition (+) and multiplication in the set Z of integers or. How many are bijective? 7. Letf A —. and let S be the set of all differentiable func- tionsfE C[0.' instead of *(x. Then show thatfg = 6.A fl B. (c) gf is bijective if both land g are bijective. Iffis a mapping from N to N defined by 1(x) = 2x + I.lJ such thatf(0) = 0 andf' is continuous.1 ]to R.f is surjective if bothfand g are surjective.5. Prove Theorem 3. lffg is injective (or surjective). more generally. instead of lettersf. Let S be a set with n elements. then A U B. 4 Binary operations Definition. (b) in < n. A mapping *: is called a binary operation on the set S. (c) in> n. the image of(x.y). (b) g. 4. A binary operation on S thus assigns to each ordered pair of elements of S exactly one element of S.g. union.o.andA — BarealsosubsetsofX.11 to S.y) under a binary operation * is written x * . without having called them by that name. 8. Hence. lffis surjective org is injective. Binary operations are usually represented by symbols like .+. Moreover. in the set R of real numbers are the most familiar examples of binary operations. show that fand g are bijective. Show that Letfi A (a) gf is injective if both fand g are injective. Show that the mapping is a hijection from ('(0. Find the number of distinct mappings from S to S. We have come across a few others in this chapter.

i/zen * is said to be Examples (a) Addition (+) and multiplication () in Z (or. Let us call . U: S S. we get a unary operation. each is distributive over the other. Let * be a binary operation (not necessarily associative) on S. is distributive over + (but both corn the converse is not true). One would intuitively expect that the result can be generalized to any finite number of elements. right-distributive over o (yez)*x=(y*x)o(z. we first introduce some notation.y. the composition (e) of mappings is a binary operation on the set S = X to X. their composite gf is also a mapping X —' X. is called an n-ary operation on S.c C S. for any positive integer n. To prove it formally. then • is (c) left-distributive over o x * (ye z) = (x * y). Hence.b. the set then o is not commutative. (c) Composition (o) of mappings is an associative binary operation on X X has more than one element. A binary operation *: S X S (a) S on the set S is commutative (f x*y=y*x (b) forallx. Moreover. Moreover.z e S. Ifs is both left.22 Sets and mappings mappingsf X X and g: X —' X. (b) Union and intersection in the set are both commutative and associative. Hence. another binary operation on S. we may omit the parentheses and simply write a * b * c without ambiguity.and riglzi-distributive over o.y. (x * z) (d) for all x.z ES. Let * be an associative binary operation on the set S.y.1 for all x. Definition.x) distributive over o• 4. When S S n = 1.zES. * y) * z associative if x * (y * If o is for all x. a mappingf S (n factors). more generally. More generally.yES. This is indeed true. Then a * (b * c) = (a * b) * c for all a. in R) are and associative.

we have k respectively. Chapter 2) as follows: a6 = a1. For four elements.a1. one without ambiguity. let /(a.. For example.a5 in that order is (((a. Given a list a.a4 . every meaningful product of a. be any meaningful product of a.d.7.. a. = ai) for n> 1.Binary operations 23 S.. . (a * (b * c)) * d. Given a list a.b.. Let f. a. b(a. there are five meaningful products: ((a * b) * c) * d.. Proof We prove the theorem by induction on n. .a2 .. * a2) * a3) S a4) * a5.c in that order are (a * b) * c and a * (b * c). in several ways by inserting various parentheses and repeatedly applying the binary operaBut the parentheses must be suitably placed so as to get a meaningful product. a * (b * (c * d)). 4.. the standard product of a. In general. The standard product of elements a. .2 Theorem. Let n> 1 and suppose the result holds for products of m elements for every m < n.. 4. Then...b in that order. hypothesis. where I k n and are them. where I k n and 4. in that order. By the induction j—I fl ..(a. denote any meaningful product of a.. . clearly.(a1 are meanak) and ingful products of a.ak) and selves meaningful products. in that order is defined inductively (see Theorem 1.. Then b(a.ak) * cb2(ak+. hence....ak and ak+. in that order. in that order E 5. the expression a * b c is ambiguous and.c. some k. that is. Let * be an associative binary operation on the set S. = for . in that order. if * is not associative. not a meaningful product.(a.b. a * ((b * c) * d). (a * b) * (c * d). we can a * b the product of a. For n = I the result is trivially true.. The meaningful products of a.ak) * some k. For example.a3 . form a product of the elements a. is equal to the standard product fl7_ .

The validity of this definition follows from the Recursion Theorem (Theorem 1. Let * be an associative binary operation on S. Therefore.3 Theorem.a. by the induction a hypothesis. Let a C S. Chapter 2).. 4.n be positive integers. it is equal to fl7:a. once again 0 Theorem 4. \ In—k—I H I elements. Then (i) am * a" = Qm+n (atm)" = atm". As a result of the theorem.7.. (ii) . C S there is a unique product of a1 in that order that is * written a1 * (without any parentheses). jf* is an associative binary operation on S. Definition. and let rn. = 1k I-I ae) \ in—k i-I !fk=n— 1. Let a S and n C N. then given a list a1 . therefore. then ai) Ifk<n— I. Let * be an associative binary operation on a set S.2 is known as the generalized associative law. (A Now aa) ak+J) * an) ((n ai) I k (ri ak÷J)) * an...24 Sets and mappings Hence... We define inductively the power of an element a as follows: a'=a.

. Suppose e.. 4.B be sets... it can be easily proved that the sets m = (I . Let x. If (here exists a bijection A and B have the same cardinal number (or cardinality. Suppose x commutes with y and z. Determine if the binary operation oØfl R is associative (commutative) in each of the following cases: (a) (b) (c) (d) xoy=x2y xey=min(x. show that y= z. How many binary operations in Problem 4 are commutative? Suppose S is a nonempty set with associative binary operation °. Let A and B be sets. This enables us to assign to every set Sa cardinal numberlSl in accordance with the following definition. we say that A is equipoteni to B and write A B. 0 Problems 1.y) xoy=xY+yx xoy=l 2. B. Let A.fE S such that x e = x and Jo x = x for all x E S.z E S. 5 f Cardinality of a set Earlier in this chapter when we defined a finite set as having a finite number of elements.. We shall now make this notion precise and extend it to arbitrary sets. we relied on our intuitive notion of counting. Let S be a nonempty set with associative binary operation°. If o is commutative.. if x o y = e = z ° x. Let S be a nonempty set with associative binary operation °. Show that the binary operation o given by xo y = x a ° y is also associative. then A Definition..n) are equipotent if and only if m = n. 6. Prove that e = Furthermore.Cardinslity of a set 25 Proof Follows from the definition of a" and generalized associative law. 3. It is clear that equipotence is an equivalence relation (on any given collection of sets). If m and n are natural numbers.) IAI = IBI. We use this fact to define a finite set and its cardinality..m) and n = (l. Let a E S.y. iso commutative? Let Sbe a set with n elements.. B. If there exists a bijective mappingf A —. 5. Show that x commutes with y o z also. Find the number of binary operations on S.

. let = o o . the relation S coincides with the usual ordering of natural numbers. Theorem (Shröder—Bernstein).e. If mappingsf A is injective. We extend this ordering to arbitrary (finite or infinite) sets. If S is equipozeni to she set (I. Moreover. The following theorem shows that s is also antisymmetric and. Definition..) We shall prove that Z and Q are both countable sets. The cardinal number of N (and hence of every countable set) is denoted by (Some authors call a set countable if it is finite or equipotent to N.. is the first letter of the Hebrew alphabet. For any IAI. Let A and B be sets. For any mapping X —' X. Let S be a set. A set that is not finite is called an iitfinite set. The cardinal number ofan infinite set is called a transfinize cardinal number. Then there exists a bijection w: A—'B. If there exists an infective Pnapping . then gf A —. • 4. and let 4° denote the identity mapping. Care both injective. C is also injective. n 0) and 4* = g(B) = (g(b)fb E B). and let f 5. a partial order on the set of cardinal numbers. hence. then S is called a finite set ofcardinal number ii. We shall prove that. (n times)1. in a sense to be explained later. IZI = IQI = But the set R of real numbers is uncountable.1 A —' B andg: B —'A be infective mappings. IAI fBI. The letter pronounced aleph. therefore. Hence. IBI s ICI imply fAl 1(1. let C= B — Set B* Imf = c E C. .1 A —' B. Proof 1ff is not suijective. the cardinal number c of R is strictly greater than The usual ordering of natural numbers gives an ordering of the cardinal numbers of finite sets. hence. setA the identity mapA A —' B and g: B —. For cardinal numbers of finite sets. An empty set is a finite set of cardinal number 0. n E N.26 Sets and mappings Definition. n) for some n e N. have the usual meaning [i. it is clear that for cardinal numbers of arbitrary sets the relation is reflexive and transitive. then 141 is said to be less than or equal to IBI (written Al IBD. An infinite set S is said to be countable (or denumerable) if S is equipotent to the set N. Let A and B be sets.

B. Sincef(a) a A. hence. but g . hence. = 0. lxi < The following theorem shows that this result holds for any set. let j denote the being an element of (0. which yields a = a'.1). a = g(fg)" '(c) = then 1(a) = B. B.2 Theorem.) 5. if f(a) E B. Consider the mapping every a in A* has a unique preimage g'(a) in w: A B given by ifaEA. hence. b C because Cc B. then a. 0 has 2" We saw earlier that if X is a finite set with n elements. where b Let a. bijective. let b = b.: X — (0. If çv(a) Thus. Hence. Consider now the mapping g: X (0. the X for all x' C X. then a =g(b) EA*. lmf Hence. IXI have to show that there is no suijective (and hence no bijective) mapping Since we have already proved [Example 3.6(b)1 that from X to is equipotent to (0. For any set X. To prove cv is surjective. then a.Ifa EA. Hence.a' A. Leta E A. is a image of x under the mappingf mappingf. then IfbE B*. g '(a) = g '(a'). Conversely. g(b).1 )X.1 )X.3). Hence.Cardinality of a set 27 The restriction of g to B is obviously a bijection B* A. then a =g(b)forsome b E B*. Hence. a E A*. IA I < IBI means that IA BI and IAI IBI.l)X. it will suffice to prove that there is no surjective mapping from X to (0. This proves cv is injective.a' E A and = çv(a') E B*. Therefore. cv(a) = f(a) = b. n 0. Hence.1) given by g(x') = 0 = I = 1.u is bijective. f(a) =fg(b) = for some c E C and n > 0. (As usual. g C (0. b =f(a) for some a E A. = =f(a') a a'. Therefore.. IXI < given by j(x) = {x) for all x C X is Proof The mapping j: X we To prove that IXI obviously injective. We first prove that a wherec E C. then elements (Theorem 1. =f(a) if A* if and only iff(a) E We claim that . If b Ba.1) X• For any x in A'. is surjective and. hence. Thus.a' E A.1 )X• Consider any mappingf X —' (0. Therefore.

r is not surjective..1 )N R. 0 By Theorem 5.. Hence.. This proves that IXI Therefore XI < 0 This by For any set X let us denote the cardinal number for finite sets = notation is in agreement with the equality (Theorem 1.. but the S1 are not all distinct.1... hence. we can delete all repetitions in the sequence (S1 .. Proof Any mappingf N —' (0. IfS = (S1 . any real number x can be expressed in the binary scale in the form x=. where =0 or I for every n E N.3).). R is not countable.. Put differently..1) determines an infinite decimal = Thenf'-. It follows immedi- ately from the definition that a set S is countable if and only if S = .S2.28 Sets and mappings (O.S2. Therefore. R is uncountable. R is equipotent to (0.1 )N. x determines a mapping with Thus. then S is either countable or finite... we have an injective mapping givenby Therefore.x2x4x6.. On the other hand.x5x3x1. x1is a mapping (0.) and be left with either a sequence or a finite list. where the S1 are all distinct..3 Theorem.2. if there is a surjective mapping N 5. We shall now prove the important result that the set of real numbers is not countable.. S is either countable or finite. which is clearly injective.S2. by Theorem 5. =IRI=I(0. 5.l).). (S1 .l)NI= 2K0• < 2K0. INI = We next prove some properties of countable sets. The cardinal number of the set R of real nzimbers is = Hence. Hence. so there is no bijection from N to R.

) be a sequence of finite sets.5.a2. A U B is countable.. A subset of a countable set is countable or finite.. Because S is countable.).a2.... Z is countable.4 Theorem. and so on.b2.6 5. More generally.. Hence.2 Hence.b1..bi.. there is a bijectionf N g: N —' Tgiven by S that induces a surjective mapping g(n) =f(n) =a if f(n) E T.). Let A B= (b1. by the previous theorem.). Therefore.5 0 Theorem.b2.B=(b1. T is either countable or finite. Proof Z = N U N'.bq). Z and Q are countable sets. there is a surjective mapping N 5.. Hence.}... Proof LetA=(a1.. there is nothing to prove.bq.a2. Proof Let S be a countable set. p= 1. so S is countable or finite. The union of a sequence of unite sets is countable or finite..b2.. Thus. Z X N is countable... Let A and B be countable sets... A )< B is countable. Hence.— l.—2.. Suppose T is nonempty and let a E T. 0 Theorem. 5. A U B is not finite.ap. and let Sbe their union. 0 ... Then A U B and A X B are countable.. Then S={aj. 5. Proof Let {A. The set A X B can be partitioned into finite subsets I).. if f(n) T. where N' = {O. Q is 4(a. If T is empty.7 0 Theorem.b2. the union and Cartesian product of a finite number of countable sets are countable.)..Csrdinallty of $ set 29 5. Hence...b)=a/b countable. The mapping givenby is surjective. and let Tc S.. Because A C A U B..ThenAUB={a1...B. The more general result folows by induction on the number of sets. by Theorem 5.

A nonzero integer p is called a prime ifp ± I and the only factors of p are ± p and ± I. which is proved in Chapter 13. IfS is a nonempty subset of N. then S contains a smallest element (also called least element). An integer b is a factor or divisor of an integer a (or a is a multiple of b) if there exists an integer c such that a = bc. real numbers. ifa<0. This property along with our assumed familiarity with addition and multiplication of integers provides a basis for proving a number of other important properties of Z. We say that b divides a and write bla (b%a means b does not divide a).CHAPTER 2 Integers. We now state the well-ordering property of positive integers. The absolute value (or modulus) IaI of a is defined as follows: a ifa>O. The set of integers is denoted by Z. and the set of positive integers (also called natural numbers) by N. multiplication of integers. and complex numbers I Integers In this section we give some properties of integers that the reader will encounter throughout the book. rather we assume our familiarity with addition. 1a 0 ifa=0. 30 . We do not intend to give here an axiomatic development of this topic (we do this in Chapter 13). We first give some definitions and notation. and the usual properties of these operations. Let a be an integer.

x E S. But then by (ii). Assume that the theorem is true for all positive integers <n. Then for all positive integers x < m. Thus.2 Second principle of induction Let S be a subset of N such that (i. So let n> I. To begin.and n E S whenever m E Sfor all positive integers m <n. (ii) Then S=N. m — I S. Proof The theorem is true ifn = 1.wearedone. a contradiction.Intergers 31 We now derive some consequences of the well-ordering property of positive integers. T = 0. Proof Let T= (x E Nix S). by (ii). If n is prime. Otherwise by the well-ordering property T contains a smallest element m. we are done. Thas a smallest element m. Then by the well-ordering property of positive integers. so n1. n can be written as a . m # I. we are done. S = N. m E S. By (i). Now. If T= ø. Proof Let T= (xc 5). If T= 0. hence. 0 Next we derive 1. Every positive integer is either I or it can be written in one and only one way as a product of positive primes. then a + I E S. Then. by the induction hypothesis each n1 . Clearly. We shall prove that it is true also for n.1 First principle of induction Let S be a subset of N satisfying the following properties: (i) IES. (ii) Then S = N. and. Thus. which contra- dicts the minimality of m. Else.n2.) IES. n2<n. where I.3 Theorem (fundamental theorem of arithmetic). Solet T# 0. we derive 1. m # 1 because I E S. If a c S.n2 can be written uniquely as a product of positive primes. Therefore T = 0. 0 One of the important results that follow from the second principle of induction is 1. let n=n.

. say. Thus. let p1 > q1. and r the remainder Proof Let S = (a — bxlx E Z).. =p1. i = 2.. Thus.n — b = a — b(x ÷ 1) €S.b E Z. q = q'. q — q' = 0. of a modulo b. which gives. real numbers. in turn. by the induction hypothesis. then chooseq =xand r= For uniqueness of q and r.. urn <b. S has a positive element a — b(—IaI).. q. Let a. therefore. r=r'. If m = b.. then a — bx = b implies a = b(x + I). let a=bq+r=bq'+r'. which contradicts the minimality of rn. where q is called the quotient. S contains a smallest positive integer. Then there exist unique integers q and r such that a=bq+r.. 0 .n b. r'I < b. b> 0. Let where p1 and q1 are positive primes. choose q = x+ I and r= 0.4 Theorem (division algorithm). Or<b. this implies s — I = I — I. m = a — bx. Then q1 (or if some p1 = So let By the induction hypothesis either q1 = Pi for some i = If 2.s or q11(p1 — q. By the well-ordering property of positive integers. and complex numbers product of positive primes. for some 2 The proof of the theorem then follows by the second principle of induc- hon. and by renumbering the subscripts (if necessary) = q. We proceed to prove that the factorization so obtained is unique. 0 Next we prove 1. For definiteness.s.32 Integers.. a contradiction. q1 =p1 which reduces to the case considered in the beginning.. If m> b.. then . Thus. This proves the uniqueness of prime factorization ifp1 = p1 # q1. . then But. Then b(q—q')=(r'— Because Ir — r).). If p1 = q.

does there exist a unique mapping f: N —.c. such that up + va = 1. .5 (i.d.b). let d = am + bn be the smallest positive integer in S. of a and b. S.: N X S —. Then This implies that there exist u.b E Z. Hence.) of a and b (1) the following hold: dlaanddib. Let a.Integers 33 Definition. cia.d. We prove that d is the g. Then (p. and cib. Let p be prime and a. there exist integers m. by givingf(l) and prescribing a rule that determinesf(n + I) whenf(n) is known. 0 1. and clb. Then uph + vab = Sometimes a mapping f: N S is described inductively. For example. dis unique to within the factors ± I. of a and b is usually denoted by (a. Now the question is whether this is a valid procedure for describing a mapping.y Z). the answer seems to be yes. Let S be a set and a S. 1. 1.) Theorem. 0 Z orplb. Further. S satisfying the conditions stated earlier? Intuitively. Remark.7 Theorem (recursion theorem).b have a positive g. the standard product a1 was defined in this manner in the last chapter (Section 4).d. Hence.6 Corollary. let c E Z.b).c. Then (ii) a.c. r = 0 and Similarly. that is. there exists a unique mappingf N —' S such that f(l)=a and f(n+ foralln€N. sap d. d = (a. so since piab.. Then r=a—dq=a—(aml-bn)q=a(l—-mq)—hnq€S.c. Proof Suppose p%a. Let a and h be two nonzero integers. Then d is called a greatest common divisor (g. Another example is the definition of the factorial function: 1! 1.n such that d = am + bn. Proof Let S = (ax + bylx. then cld.v b. Then obviously S contains positive integers. Then d = am + bn implies ci d. Given a mapping4.b E Z such that plab. dib.a) = 1. (n + I)! = (n + 1)'n! for all n N. The positive g. (ii) If c Z. By the well-ordering property.d. We prove it formally in the following theorem. In other words. where 0 r < d. cia. By the division algorithm a = dq + r. which contradicts the minimality ofdin Sunless r 0.

be subsets of a set. I E P. Therefore. hence. it is easily verified that R — ((n + I . R M. hence. f(1) = q'(n. A. Then there is a (n + such Since n+ 1 there exists cES.. a contradiction.+ [n(n+ l)]2 Prove by induction that (I 3. n(4n2— I) + (2n — 1)2 = (a) 12 + 32 + (b) 2. Now (n R. hence R C R — ((n + I . the set N X S itself satisfies (*). Prove by induction the following laws: (a) . (*) Clearly.f(n)) for all n C N. a contradiction.a)EX. . hence.b)) satisfies (*). But then R — ((I .then(n+ N. thenf(n+ l)= cb(n.b) E R. Let fg: N —' S be two such mappings. + . iff(n)=g(n) forany nEN.34 Integers. P = N. Let R = (X) be the intersection of all sets in M.andif(n. So. To prove the existence off let M be the set of all subsets X of N X S satisfying the following conditions: (l. Therefore. such that (I .b)). real numbers.c)) satisfies (s). 0 a and f(n + I) = Problems Prove by induction the following summation formulas: 1. but n + I unique E S such that P. if n E P. E R.. Further. Then R satisfies (*). Therefore. M is nonempty. b # a. there exists bE S. by the induction principle. If I P. Hence. Then f(l)=a=g(l). by uniqueness of = Therefore the mappingf: N —p S given byf(n) = satisfies the required conditions. then that (n + I n + I E P. f(n) = g(n) for all n E N. Now let n E N and suppose n E P.f(n)) = = g(n + 1).... C R.x)EX. Let P be the set of positive integers n for which there exists a unique element in S such that E R. Again.c)).. then it is unique. R C R — ((I . . namely. Let A. for every n C N there is a unique element C S such that (n. and complex numbers Proof We first show that if a mapping satisfying the given conditions exists. hence.

the construction of real numbers is not given because it is normally done in books on analysis (see. a" a (modp).c E Z such that aib and aic.p E Z with p prime. I (mod n).Rational. Also show that any integral root must be a divisor of 0 8. Let a. (p — 1)! + 1 9. Prove Wilson's theorem: For any prime p. equivalently. note that for a prime p. 7. (mod p).f).b E Z and b . where (a. real. Furthermore. = max(e. show that m. and complex numbers 35 (b) 4. whenever aix and bix.m. 4(p) = p — 1. Prove Euler's theorem: For any integer a relatively prime to n. A terminating or a recumng decimal is a rational number that is. The least common multiple (l. (b) [a.c.b E Z. Thus. where 0. for example. If a. + nc. Constructing the rational numbers from integers is given in Chapter 12. b]. A real number may be taken to be a (terminating or nonterminating) decimal. (c) [a. denoted by [a. 6.b] is unique. where a. The system of complex numbers is defined to be the set C R X R .b) is the greatest common divisor of a and b.. real.n E Z. Euler's theorem generalizes Fermat's theorem. then dix. Prove that (a) [a.b] = ab/(a. Landau 1960). for 5.b). is defined to be a positive integer d such that (1) aidand bid.) of a and b. Show that any integer n> I is either a prime or has a prime factor (ii) rn. and complex numbers We assume that the reader is familiar with rational and real numbers and the usual operations of addition and multiplication in them. Prove Fermat's theorem: For any a. However.b. 2 Rational. a fraction of the form a/b." 0. Show that the polynomial where a. 10.b] = where g. factorizationsofa Suppose a = and b = and b as a product of primes. (4 is called Euler's function). has no rational root that is not integral. E Z. and let 4(n) denote the number of positive integers less than and prime to n. Let n be a positive integer.

3 Fields Introductory algebra is concerned with the operations of addition and multiplication of real or complex numbers as well as the related opposite operations of subtraction and division. moreover. (a. —b\ Thus. The complex number I = a — lb is called the conjugate of z. (a. b + d).0) = (ab. that multiplication is distributive over addition. and complex numbers (where R is the set of real numbers) with addition and multiplication given by (a. addition and multiplication in C have the same properties as the corresponding operations in R (or Q). where a.0) = (a + b.b) E C. respectively. Thus. if (a.0) + (b.0)' (b.—b) = (0. Then for any (a.0). (a.0) + (0.0) and (1.0). Let z = a + lb C.b) = (a.36 Integers. This allows us to identify every complex number (x. Now j2(0.O). Furthermore.0). (a.O).0) with the real numberx and thereby treat R as a subset ofC.1) .0). (0. where a and b are real and i2 = — I. It is easily verified that addition and multiplication in C are both associative and commutative and. abstract algebra is concerned with deriving properties of an abstract system (such as .d) = (a + c.b) = (a.d) = (ac — bd. every complex number can be written as a + ib.0) playing the roles of 0 and 1.b) ía (0.0) wherei=(0.0)=—1. For any (a.b) (c. ad + bc). respectively.b R.b) + (c. Let us henceforth write x for (x.l). real numbers. Consider complex numbers of the form (x.b)( 1.b) E C.0) = (a. The nonnegative real number (zi + 'Ia2 + b2 is called the modulus of z.1 )(b.b) + (—a. Then a and b are called the real =a+ib part and the imaginary part of z. Addition and multiplication of two such complex numbers give (a.l)=(—l.0).b) + (0. (a. This is the common notation for complex numbers. with (0. On the other hand.

. —(xO) E F. Z is an example of another algebraic structure—"commutative ring"—discussed later in the book. The system of integers Z is not a field under the usual operations because each nonzero integer is not invertible in Z. Definition. The multiplicative operation distributes over the additive operation. (iv) (v) (vi) F such that x + The additive operation is commutative: x + y = y + x. F has a zero element 0 sue/i that x + 0 = x = 0 + x. and complex numbers are concrete examples. and complex numbers under the usual operations of addition and multiplication. Every element x E F has a negative —x E (—x)=0=(—x)+x. ). (vii) Every element 0 * x E F has an inverse E F such that (viii) (19 xx__l = I = The multiplicative operation is commutative: xy = yx. The purpose of this section is to introduce the richest algebraic structure. x(y + z) = xy + xz. +. where F is a nonemply set and +. Not every field is infinite. real numbers. they can be easily deduced from them. For example. for which the systems of rational. to prove (a) consider xO = x(0 + 0) = xO + xO [axiom (ii)J [axiom (ix)] By axiom (iii). are binary operations on F. is called a field (1 the following axioms hold: q'i.) The additive operation is associative: (x + v) + z = x + (v + z). The multiplicative operation is associative: (xy)z = x(yz). xO + (—(xO)) = (xO + xO) + (—xO) = xO + (xO + (—xO)) [axiom (i)J. F has I 0 such that x I x = I x. or complex numbers) from its axioms in a formal. A system (F. Thus. Examples of fields are the systems of rational numbers. real. We close this section by giving an example of a finite field. rigorous fashion. so0=xO.Fields 37 the systems of integers. . real numbers. Note that although familiar statements such as (a) xO==0 (b) (— x)y = — (xy) = x(— y) (c) x(y—z)=xy—xz are not included in the axioms.

1 Example (integers modulo n). is a field with p elements. we proceed to show that axiom jvii) does hold. and let or Z/(n) = (O. We define addition and multiplication in as follows: x+y=x+y.38 Integers. but axiom (vii) need not hold. where n is any positive integer. But then Hence = Ou + ji7=0 + = Therefore.). E Since i= x + kn for all k E Z.. real numbers. That addition satisfies field axioms (i)—(iv) for is trivial. We shall return to the algebraic structure . (vi). It is also trivial that multiplication satisfies axioms (v). question arises: lfi= = in other words.5). x. Chapter 1).. and complex nwnhers 3. This proves that addition in Similarly. So there exist integers u and v such that pu + xv = I (Theorem 1. Let ii be a fixed positive integer. later in Chapter 9. and (ix). However. This implies that pfx.. Now 1= and 5= 57' n divides x — x' and n divides y — n divides (x — x') + (y — y') n divides (x + v) — (x' + y') by(1). is well defined.. the and = j7'..1. is it true that i + =1' + and are addition and multiplication in (1) well defined? We proceed to show that this is indeed the case. for n = p. for an example of a "finite ring. xy=xy. and the negative of is (i)." . Let 0 IE Z. The zero element is 0. p prime. n divides (xe' + x'y') — xy' — x'y = (xy — x'y') + x'(v' — y) + y'(x' — x) proving that multiplication is also well defined.(n — 1)) be the set of equivalence classes of integers mod n (Section 2. (viii).

.. A Pnapping A:mXn—F is called an m X n matrix over F.n.n) will be denoted by n..n) that is the size.. if m = p..ThenA(i.. which we shall study later.j) E m X ii.andlet(i.j)ti = l.n). They are a rich source of examples of algebraic structures. and a13 = b0foreach(i. As usual. for any positive integer n. written A = B. is called the (i. n = q.j) under the mapping A.. (Note that it is really the ordered pair (m. When the (i.. LetAbeanm X nmatrixoverF.CHAPTER 3 Matrices and determinants I Matrices Matrices originated in the study of linear equations but have now acquired an independent status and constitute one of the most important systems in algebra.the image of the ordered pair (i.....j = 1.. Let F be a field and m and n positive integers. The cartesian product m X n is therefore the set m X n = ((i. Definition.) 39 . A is said to be of size m X n.j) entry of A is written we say that A= An m X n matrix A over F and a p X q matrix B over F are said to be equal. IfA is an m x n matrix.j)E mX n.j). the set { I .j) entry (or element) of the matrix A and is denoted by A11 or the latter being more common.

the set I X n matrices (row vectors) and the set F"<' of all n X 1 matrices (col- umn vectors) over F are both written F" when the context makes the meaning unambiguous. for the sake of convenience. Again. written F.. An m X I matrix is called a column matrix (or a column vector). a I X n matrix is called a row matrix (or a row vector).. .amj is called thejth column (column)) of A.40 Matrices and determinants An m X n matrix is usually exhibited by writing the mn entries a..) If every entry mA that is not in its diagonal is zero. = 0 for all i j...)..... A diagonal matrix diag(d1 ... is called the ith row (or row i) of a..n.d.. The(i.. = a. The list (n-tuple) of entnes a1 .... for the sake of conveof all nience.. denoted by A.. A (strictly) lower triangular matrix is defined similarly. and the list of entries a.a22.. ami am2 a. is called the diagonal of A.j)entry in A Thus..... where d. ..... The list (m-tuple) A......).. The set The set of all m X n matrices over a field F is denoted by of all square matrices of order n over F is.1) is called an upper triangularmatrix if a....2 a1.. (A matrix that is not necessarily square is said to be rectangular. A square matrix A = (a. and denoted by In keeping with the topography of the rectangular array. Similarly.a.a.. then A is called a diagonal matrix and denoted by diag(d1.. a.... A1 is shown as a1..1...1 = 0 for all i > j.av .aa..d. and exhibited as (a..1 in a rectangular array as shown below: a11 a21 a12 a22 a1.. and strictly upper triangular if a.. i = l.. An n X n matrix A is called a square matrix of order n.1 a.Am and n is common to the ith row and thef th column.. a21. A has m rowsA...) is called a scalar matrix if d1 = d2 = d. ..

These three operations are induced. a11 ami — ami — amn Next we define multiplication of matrices. The negative of the matrix A is the m X n matrix B = such that (i. Let A = (afl) and B = (b0) be two m X n matrices over F. Let A = (a.p.Operations on matrices 2 41 Operations on matrices In order to set up an algebraic system of matrices.1) such that (i.. = . The sum of A and B. The negative of A is written — A. is the m X n matrix C = (c. the definition implies that a11 a11 +b11 + . we denote it simply by 0. * bmi bmn ami + bmi amn + b. and a third operation called scalar multiplication. written A + B. Note that matrices A and B can be added if and only if they have the same size. The m X n matrix in which every entry is zero is called the m X n zero matrix (or null matrix) and is denoted by When the size of the m X n zero matrix is clear from the context. When the matrices are exhibited as rectangular arrays. addition and multiplication.j)EmXn. Definition. Definition. .1) be an m x n matrix. by the binary operations of addition and multiplication in the field F. in a manner to be described hereafter. . we are going to intro- duce two binary operations.j)EmXn. Thai is.

Thus. BA have the same size if and only if in = n. The product AB is the m X p matrix C = (i. A and B are in X n and n X in matrices. A B and BA are not necessarily equal.. But even when A and B are both n X n matrices.. respectively.. Let A and B be matrices. This is easily proved by the following example. a. ... we have the rule: LetA = B= and C' be their product AB. X p where matrices over F.+ Then the (i.n... Let us define the dot product of two n-tuples of elements ofFas follows.j) entry in the product AB is the dot product of the n-tuplcs (a.j) entry in AB is the dot product of the !th row in A and the jth column in B..n X n and.j)EmXp. which displays the fact that the (i. The products AB and BA both exist if and only if. Note that matrices A and B can be multiplied (in that order) if and only if the number of columns in A is equal to the number of rows in B. AB. for some m.. The reader may also find it helpful to keep in mind the following.a.. Given a = (a1 . j • j = ••• . respectively. therefore. • • • b. a. Then = A. Let A = and B = be.42 Matrices and determinants Definition. . But then AB is an m X m matrix and BA is ann X n matrix. Let and g]. and —that is.) and b = (b1 we define ab=a1b. the dot product of the ith row in A and thejth column in B.

and B and Ccan he multiplied.j) entry of (A B)C. Then A(BC) and (AB)C are both in X q matrices.j) E m X q.) If A and B can be multiplied and B and C can he added. This proves A(BC) = (AB)C. The proofs for (ii) and (iii) are similar to the one for (I) and are left as exercises. B. This proves that multiplication of matrices is not commutative. Let (i. except in the trivial case of I X I matrices. then A(BC) = (AB)C.Operations on matrIces 43 Then AB BA. n X p. C = (c.kbk/) = the (i.1 (ij) * Theorem. 0 The Kronecker delta function is defined to be ô: N X such that jfj'j ifi#j a1ô11 = for z. B = (b11).j) entry of A(BC) is 0k (± bk/cl)) = ± (± a. Let A.j EN It is worth noting that given an n-tuple (a1 . Then the (i. Proof (i) Let A = (a11). then A(B+ C)=AB+AC. (ii. 2. and C be matrices. respectively. If A and B can be multiplied.1) be m X n. and B and C can he multiplied. If A and B can be added. p X q matrices. then (A + B)C=AC+ BC.

is clear from the context.j) entry in A by a. Let A be an m X n matrix.j) element is all (I. 2. Remark.a)A. scalar multiplication is also a poiniwise (or component-wise) operation.j) a X a. The n X n matrix in which..2 Theorem. If there is an n X n matrix B such that AB=l=BA.... is denoted by A = (a11) be an m X n matrix over F.44 MatrIces and determinants Definition.. and every other entry is zero. Then =A A be an n X n matrix.. is 1.0 00"' When the order of the identity matrix I. Therefore the multiplication ofa matrix by an element of Fis called scalar multiplication.]) E m X fl A. aA = diag(a. for each (i. The product aA (also written Aa) is the m X n matrix whose (i... The following theorem justifies the nomenclature of the identity matrix. then A is said to be invertible and B is called an inverse The inverse of a matrix A. the elements of Fare called scalars. Thus an identity matrix looks like 01 l0. the (i.a.j) is called the identity matrix of order n and denoted by entry is In other words.a. each entry in the diagonal of I. and likewise. we drop the subscript n and simply write I. Like addition of matrices.. where diag(a..j) entry in aA is obtained by multiplying the (i. if it exists. By that term we mean that the (i.. the . In the context of matrices over F.a) is an m X m scalar matrix. and let a E F..

and let (1) Then (ii) (iii) (iv) (v. .j = l.4 (i) Theorem.j) entry in A + B is obtained by adding the (i. If A and B can be added. The transpose of A is the (i. Let A and B be matrices over F and let a E F. Let A and B ben x n matrices over F. Proof Obvious. = where is the Kronecker delta. (a+/J)AaA+flA. On the contrary. (The order n of the matrix is usually known from the context and therefore not included in the notation. a(AB) = (cv.) can be written as A= i—I The following result about the product of two matrix units will be used frequently. 2. The following theorem gives properties of scalar multiplication. For any two matrix units eu. (amA = a(/)A). 0 then X n matrix in which the (i. of order n. 2. '(aA)—a('A).) a(A+B)=aA+aB. IA=A..3 Theorem.) It is clear that any n X n matrix A (at. multiplication of matrices is not a pointwise operation. where b0 = Definition.j)entry is I and all other Let entries are 0. Let A The transpose of A is generally denoted by 'A and is obtained by interchanging rows and columns of A.j) E n X m. (afl) be an m X n matrix. The n2 matrices i. then '(A + B) — 'A + 'B.Operatloas on matrIces 45 (i. n X m matrix B = (ba).. Then (ii) (iii) '('A)=A..A)B — A(aB). are called matrix units of order n..j) entries in A and B.n.

Similarly.1 1/121 A. A matrix is said to be partitioned if it is divided into submatnces (called blocks) by horizontal and vertical lines between the rows and the columns.46 Matrices and determinants (iv) ifA and B can be multiplied. we can consider B as the 2 X 1 matrix [B. b41 b22 b32 b42 be matrices that have been partitioned by the horizontal and vertical lines as shown. (iii) follow immediately from the definition. Proof (i). 0 Problems I. Sometimes the multiplication of two matrices A and B becomes simpler by partitioning them into suitable blocks.l. To prove (iv). Let Then and 'B = = and = 'A = Suppose AB (cu). If we delete any p rows and any qcolumns ofA. 2. The proof of (v) is left as an exercise. let A = (afl) and B = (by) bern X n and n X p matrices. For example. respectively. Then the (i.2 A22 whose elements A11 are the submatrices of A given by the partitioning. . 3 Partitions of a matrix Let A be an m X n matrix. Prove Theorem 2. We can now consider A as the 2 X 2 matrix [A.. ProveTheorem2. then 'A is invertible and = '(A'). b3.j) entry of '(AB) is Cfl = (ljkbkj = = = (i.4. ' b12 B= b2.j) entry of'BtA. the resulting (m — p) X (n — q) matrix is called a submatrix of A. then t(AB) = tB'A. let a11 A= a21 a3I a41 a12 a22 a32 a42 a13 a23 a33 a43 a14 a24 bI. (ii). (v) if A is invertible.

is absent.2. a certain number.n).4. one must partition the two matrices in such a way that the product of the blocks occurring in the process of multiplication is meaningful. n a(2) a(n) the fact that a maps I to a( I). For example.. Note that the order of the columns in this representation of a is imma12 4 3 terial. The product of two permutations of n is again a permutation of n. Ii represent 2 4 3 I 4'\ 3) and 3 1 1 2 the same permutation. Of course. ar(2) = 4. and so on. 4 For The determinant function set (l.. o(i) = i.. similarly. if 11 2 3 3 4\ 4 i)' (I I) = 2 3 I 3 4 I 2 are permutations of degree 4. if in such a representation of a. it is understood that a keeps i fixed.2. and so on. For example. this notation the first column expresses the second column. The method of block multiplication is sometimes quite handy when multiplying large matrices... say i. Furthermore. (1 In 2 \a(l) .The determinant functIon 47 whose elements are the submatrices of B given by the partitioning. that is. that a maps 2 to a(2). thus 43 In the same way 234 234 21 4 . . The set of permutations of n is denoted by convenience a permutation a E is usually represented as Let n be the Sn. will be considered as a permutation a of the set (1. then ar( because r maps I to 4 and a maps 4 to I. For example. Then it is easily checked that AB is the same as the product [As.5) such that a(3) = 3 and a(4) = 4.3.... LA2I +A12B21 1422JLB21J L1421B11 +14221321 Multiplying two matrices A and B in this manner is called block multiplication of A and B.

... not necessarily uniquely. . For example. is a transposition. if any. Observe that (a1 a2 . that is.a2 . a(a2) = a3. It can be shown that if a permutation is a product of r transpositions. A permutation a of a is called a cycle of length k if a maps a set of k elements. (a3 a4 . a1). if n the inverse of a.. (4 5 3 2 1) 1X3 1X5 1X4 I). then 12 0=Ifi ('I 2 ar. a. and also of s transpositions.a*.r E 2 n ft 2 2 j i j n n which maps I —' andj —' I and keeps other elements fixed.. For example. \1 2 Thus. We write such a cycle as (a1 a2 ... the number of factors in any representation of a as a product of transpositions .. The set Sn is called the symmetric group of degree a. For example. fixed. as (if). Thus. more usually. then r. ak).... (a. A cycle of length 2 is a transposition.. every permutation in can be expressed. and a(a) = a if a is not one of the a1 . is the permutation i\ which can also be written as Vi is a permutation of degree n.. It is easily seen that any permutation a of a can be expressed as a product of disjoint cycles.1.. if -l \l (2 3 2 4 3 4)' More generally.cr E Sn.. That is. a1 .. o(a1) = a2.. (a2 a3 .ak).48 MatrIces and determinants The inverse of a permutation of a is also a permutation... keeping the remaining elements.... cycles for which no two have any common element. ak). Note that the inverse of a j transposition (if) is (if) itself. A permutation that interchanges two elements and keeps others fixed is called a transposition. We write it as or. that is.a(ak) = a1. cyclically. _(i although (2 5 6\(l4x2653 462135 2 3 4 Further. a1 a2).s are both even or both odd (Theorem 2. as a product of transpositions. For example. not necessarily uniquely. say (a1. ifa... every cycle can be expressed as a product of transpositions. ak_I) are the same cycle... Chapter 7).

if it is a product of an odd number of permutations.3a22a3. a2 denote a fixed * The authors would like to thank Professor R.. Hence. 1 Definition.. Let A =(a11) be an n x n matrix and let R... If a permutation is a product of an even number of transpositions. Let a = then r' = . Also. .) denote its ith row. because the inverse of any transposition is itself. In the following sections we prove some results that enable us to evaluate the determinant of any square matrix in a less tedious manner. then it is called an even permutation. For example.= a12 (a. if a permutation a is even (odd).. det A = + a. then er 'is also even (odd). Let A = be an n x n matrix over a field F. Indeed. the first three are even. Let R = (a. For readers familiar with the concept of a ring (for the definition see Chapter 9). The identity permutation I of n is even because I (ijXj I). where i.2a21a33 — a.. most of the results proved for determinants over fields also hold for determinants over a commutative ring. it is of interest to note that the determinant of a matrix over a commutative ring is defined exactly in the same manner. it is called an odd permutalion. ..Properties of the determinant function 49 is always even or else always odd. consider a 3 X 3 matrix A = given by (a11).3a21a32 — —a. a.j E n.a23a32... 5 5. For any permutation a we define sgn a to be I or — according to whether a is even or odd. N.1 Properties of the determinant function* Theorem. Gupta for his suggestions on this section. S3 has 3! = 6 elements e=I(1 \l Ii 2 23/' 2 1 3\I ft 1 \23 ft 2 2 2 3\I 3\ a='fi \3 2 2 2 3 3\ 3)' + i)' Il 12' 3 3 2 Of these. and the others are odd. a. Then the sum is called the determinant of A and denoted by det A or IA. ii. ...

R1 R. (i) The left-hand side is (sgn 6ES... Then R1 R1 R1 R2 (i) R2 R. R R2 det R. T= The mapping aI— I Let t=(i. . Proof. then det A = det 'A = det A 0. Suppose the ith and jth rows are equal. Suppose i <j..50 Matrices and determinants I x n matrix. . det R. (ii) (iii) Clear. det (iii) (iv) If two rows of a matrix A are equal. 0: S —+ T such that 8(a) = at is easily seen to be onto mapping. proving (i). + alj))•••ana(n) = (sgn o)a Ia( j)• R1 . Let S={aeS1: a(i)<a(j)}. Thus (sgn a)(a a IA = .j). = det R1 + det . R1 R2 (ii) R2 det R= R.

+ det R. = det B+0+0+det A. (iv) det(1A) = = = t(n) =detA. . = det . by Theorem 5. + det .. If a matrix B is obtained from a matrix A by interchanging the ith and jth ron's. R3+R. 5.. R.2 Coroilary. . R1 R. R1 R1 R1 R1 R1 R. + det .1 (iii).. then det B = — det A. Proof. . = — (since ith row =jth row) (sgn =0.1 (i). Thus det B = — det A .. + 0 = det by Theorem 5.Properties of the determinant function 51 = (sgn cr)(a 1)••• .

Then det(X Y) = (det X)(det Y). .. by Theorem 5.1 (iii). (x r(1) xflT(fl))( 4TES. then det L : = (sgn a) det A. Then A = R2 can Re(l) be transformed to : by m interchanges of rows. Let X. det(XY) = (det X)(det Y)..3 are Corollary. . . Suppose a is a product of m transpositions.4 Theorem.52 Matrices and determinants 5. then det(Y) = Thus det (X Y) = = 0. det (XV) = (sgn a){( i1=1 it I where i2.3. n} and the rows of A respectively.. Proof. are not all distinct. .. =( tES. by Corollary 5.. . R1 Proof. Hence by Corollary 5. (sgn a)y14 i)o(l) Y). Hence R1 D det : = (— 1)'" det R2 = (sgn a) det A. (sgn . If a is a permutation of{1.2 5. V he n x n matrices. may be assumed to be distinct because if i1..

2. If A is a square matrix such that A" = 0 for some positive integer k. If A is an invertible matrix. then det A = 0. prove det A = det A 5. —1 5 2 a b c a2 b2 3. 3 123 4 I I I 31 (b) 4 0 1 I 0. then detA Evaluate the following determinants: (a) 234. Now detA= = 'I i_.of det A 53 Problems 1. They are given by those permutations Q that satisfy the condition c(i) j. 6. the sum of all such terms is that contain a given entry (— 1)ka CES. 4. Prove that = (a — b)(b — c)(c — a). IfA isa square matrix such that A = A2. 6 ExpansIon of det A det A Let A be an n X n matrix. in det A.. I'tk UES. = where o(i)-f is called the cofacior of a. Consider all the terms in the sum = > (— [)kaa a factor. Therefore. is an n x n upper (or lower) triangular Show that if A = matrix. prove detA =Oor 1. f1 .

.1 Theorem. Combining the last two results and (2) and (3). 1/ = 0. we get 6.W(fl)..3. By definition. Substituting (I) for A1. Similarly. The expression on the right side is the determinant of the (n — 1) X .6. det A.a. (— by Corollary 5.the cofactor ofa11. (3) (2) is called the expansion of det A according to the ith row. = 0 if p #j. Now consider the sum p i.. detA Similarly.n) and k = 0 or I according to whether a is even or odd. we get V f—I I.54 MatrIces and determinants That is. Lel A be an n X ii matrix. (3) is the expansion of det A according to the jth column. (2) detA = i1 a(s)—f = That is. Then (i) = det A.. (ii) the matrix A = A = for In the following we develop the method to evaluate the First we find A11. — — . A pJ 5j V f—I V cES. a where a is a permutation of (2. Similarly..

detB. a23 a33 + =a If a. a2. so that it becomes the first column. Let A be an n X n matrix.3 a23 +a33(_l)3431at1 . To find the value of a general cofactorA1. we move the ith row up past row i — I. so that after i — interchanges it becomes the first row. a22 a32 . we get det A = a. If we expand det A by the first row.1.3(— I)143 1a.3(a21a32 — a31a22).. So by Theorem 5. we get detA = a31(— a12 a22 a.. equal to (— I 6. We now have a matrix that is obtained from A by i — 1 interchanges of rows and) — I interchanges of columns. a3. a23) + a. We next move thejth column to the left past the firstj — I columns in turn.. of a..2(— 1)142 a2.1) position. The cofactor of a11 in this new determinant is the (n — I )th-order determinant obtained by omitting the new first row and first column. so that we can apply (I).1 the determinant of the new matrix is (— det A.1 a23 + a32(— 1)3+2 a.. — a3.1) position.2 Theorem. Thus.(— 1)14 I a22 a32 a23 a33 + a.. we expand det A by the third row. Thus.(a22a13 — a32a23) — a 12(a2. a. The foregoing discussion gives the following 6.Expansion of det 4 55 (n — I) submatrix of A. the ith row and jth column of A. which is obtained by omitting the first row and the first column ofA. we first perform some row and column interchanges on A to bring a11 to the (1.3 Example LetA = (a1)be a 3 X 3 matrix. The cofactor ofa13 in del A is times the determinant oft/ic submatth of A obtained by deleting the it/I row and the jth column.. a3. to bring a3 to the (1. row I. that is. where B is the submatnx of A obtained by omitting the ith row andjth column of A.

a23 a. a. Prove that I 1 a1 2 2 . a7' n—I 2 2 A A •. Prove that (b+c)2 (a) b2 a2 a2 (c + a)2 c2 b2 c2 (a+b)2 2a = 2abc(a + b + c)3.1 + a32(— 1)3+2 a2. c . we get det A = a12(— 1)1+2 a3.. an—I A This determinant is known as the Vandermonde determinant. a—b—c (b) 4.3 a23 Problems I. c—a—b a2+b2+c2 Prove that (a) a2+b2+c2 a3+b3+c3 a2+b2+c2 a3+b3+c3 a4+b4+c4 I a+b+c a b I 1 3 a+b+c = a2 b2 c2 2 = (a — b)2(b — c)2(c — a)2.1 + a22(— l)2f2 a33 a3..56 MatrIces and determinants If we expand A by the second column. Prove the following result: I I I a a3 b c b3 =(b—cXc—aXa—b)(a+b+c).. a13 a33 a. c3 2. 2b 2c b—c—a 2c 2a 2b =(a+b+c)3. 3.

of det 4 57 abc2 (b) c 2bc—a2 c2 b2 b a c b = c2 2ca—b2 a2 a2 a b2 2ab—c2 a2—bc b2—ca c2—ab = c2—ab a2—bc b2—ca b2—ca c2—ab a2—bc = (a3 + b3 + c3 — 3abc)2. .

.

PART II Groups .

.

which is defined as a nonempty setS with an associative binary operation. Any algebraic structure S with a binary operation + or is normally written (S. 61 . and so on.AnelementeinSiscalledalefiidentityifea=aforallaES. The simplest algebraic structure to recognize is a semigroup. modules. fields. groups.b E S. It is possible to have a semigroup with several left identities or several right identities. We usually write ab instead of a b. rings. it is also customary to use an expression such as "the algebraic structure S under addition or multiplication. Algebraic structures whose binary operations satisfy particularly important properties are semigroups.•). if a semigroup S has both a left identity eand a right then e = ef=f Therefore. e is the unique two-sided identity of the semigroup.CHAPTER 4 Groups 1 Semigroups and groups An algebraic structure or algebraic system is a nonempty set together with one or more binary operations on that set. or complex numbers under usual multiplication (or addition) (b) (c) The set of mappings from a nonempty set S into itself under composition of mappings The set of n X n matrices over complex numbers under multiplication (or addition) of matrices Let (S. ) be a semigroup and let a.A right identity is defined similarly.+) or (S. However. However. reals." Examples of semigroups are (a) The systems of integers.

(iii) For every a E G there exists a' G such that a'a = e.) is a group according to this definition. However. Definition. Let a E G. Definition. A nonempty set G with a binary operation . if e is a two-sided identity and a has a left inverse b and a right inverse c. Let a E S. Henceforth. then b = be = bac = ec = c.c E G. or two-sided) identity in S. Our objective in this chapter and the following four chapters is to give basic results and techniques in the theory of a somewhat richer algebraic structure called a group. Let (S. b is a right inverse if ab e. This proves that G is a group according to the definition that it is a semigroup with identity in which every element has an inverse. ) be a semigroup. (ii) There exists e E G such that ea = afor all a E G. let S be a semigroup and let e be a left (right. on G is called a group jf the following axioms hold: (i) a(bc) = (ab)c for all a. Hence.62 Groups Definition. Hence. Ifthere is an element e inS such that ex—x=xe for aJixES. Then a'a = e = a"a'. unless otherwise stated.). let a' be a left inverse ofa. Again. aa' = eaa' a"a'aa' = a"ea' a"a' = e. and a is said to be invertible. A group may be defined as a semigroup with identity in which every element has an inverse (that must be unique). Suppose (G. Now ae = aa'a = ea = a for every a E G. then b is called the inverse of a. the binary operation in a group . We now give an equivalent definition with weaker postulates.b. then e is called the identity of the semigroup (S. An element bin Sis called a left inverse of a (with respect to e) if ba = e. Let a E S. Therefore a' is also a right inverse of a. and let a" be a left inverse ofa'. If there exists an element b in S such that ab = e = ba. Let be a semigroup with identity e. An element in S may have several left inverses or several right inverses. e is also a right identity.

which is called the nih power of a. Therefore. called the negative of a. an element a is written the identity element is denoted by 0. we define.+). so that ea = a. Further. IfS has an identity e. in particular.b E G. that are written For any positive integer n we define.is associative.2 (Chapter 1) holds here. G is a group.1 each (n-foldsum).) simply by G. of the equations ax = b and ya = b has a solution. a = aa. ab = ba for all a. D . we define +a 1. eb = eax0 = ax0 = b. and we shall denote the group (G. Theorem 4.. In the case of an additive semigroup (S. = (a—'r. A semigroup G is a group and on/v effor a/la. in that order. so that ax0 = b. Let (S. Proof If G is a group.a (n-fold product). if a is an invertible element in S. Hence. and therefore there is a unique product of any given elements E S. Therefore. Now for any b in G. a has a left inverse. as before. we speak of sums and multiples instead of products and powers.. Since. Then. When a group A is abelian. the equation ax = b has a solution. b in G. then clearly the equations ax = band ya b have solutions x = a'b and y = respectively. Conversely. say y = e. and the inverse of an element a is written —a. for any positive integer n. the equation ya = e has a solution. The identity element is denoted by e or 1. Theorem. If the binary operation in a group (G.') be a semigroup. we also defineS a° = e. for every a in G. say x = x0. its binary operation is generally written additively. e is a left identity.•) is commutative — that is. Further. let G be a semigroup and suppose ax = band ya = b have solutions for all a. and the inverse of If the binary operation is written additively. where is an inverse of a. Hence. Let a E G.b E G — then the group is called commutative or abelian. the equation ya = a has a solution.Semigroups and groups 63 will be written multiplicatively. For any positive integer n. and a E S.

We shall prove that for all a. This proves that the equation ax = b is solvable. the equations ax = b and ya — b are solvable.. By the left cancellation law. 0 is the identity. Afinite semigroup G is a group if and only :fthe cancellation laws hold for all elements in G. Thus. let G be a finite semigroup and suppose that the cancellation laws hold. Con- sider the set H = (aa1 Since aa. by the last theorem. = b for some a. the cancellation laws hold trivially. Clearly. Indeed. H has n distinct elements. 0 1. (b) Integers modulo n under multiplication Consider the set Z/(n) as in (a). hence. Let Z/(n) denote the set lence classes. Z/(n) is an additive abelian group of order n. x + y = x' + y'. Denote by ithe equivalence class containing x. Conversely. and for all a.cE G. Define x x—y y (mod n) if for some q E Z. and therefore. Hence. E G for each I = H C G. nI(x + y) — (x' + y'). Let G = are distinct. for if I = and = then nix — x' and niy — y'.1. and is the inverse ofl Thus.. ba=ca=*b=c If G is a group. qn = i' Z/(n) = (O. This is well defined.3 Examples of groups (a) Integers modulo a under addition Let x.y c Z and let n be a positive integer. Hence..bE G.2 Theorem. Hence. we can prove that multi- . is an equivalence relation.) is said to be finite or infinite according to whether the set G is finite or infinite. The cardinality of the set G is called the order of the group G and denoted by IGI or o(G). that is. Clearly. = a1.64 Groups The following theorem gives a simpler characterization for a finite group. Similarly. b E II and therefore aa. it follows from the right cancellation law that the equation ya = b is solvable. G is a group. and therefore H = G. G is a group. We define multiplication in Z/(n) as = As proved for addition in example (a). A group (or a semigroup) (G.b in G. by Theorem 1.b. E G. aa. We define addition in Z/(n) as follows: i + = x + y..(n — I)). 1. where Let a. = aa1 a.1.

that is. Call a permutation a: X —' X a "symmetry" of S when it preserves diswhere d(a.. and let G be the set of bijective mappings on Xto X(i. and the other group postulates are easily verified by direct applications of results on mappings (see Chapter 1).b) denotes the distances.r(b)) = d(a. Thus. Hence. and 4ir/3 are three of the symmetries that move the vertices in the following manner: . if a is a symmetry then = d(o a symmetry. Then G is a group under the usual composition of mappings. Then (Z/(n).t13. 2. where is the Euler function. Clearly. Let us consider a special case when Xis the set of points on the perimeter of an equilateral triangle: 2 3 The counterclockwise rotations through 0.e. the identity permutation is a symmetry. (c) Permutations under usual composition Let X be a nonempty set.r are two symmetries.(orXb)) = d(c(r(a)). The unit element of G is the identity map of X. If a.Semigroups and groups 65 plication is well defined. or is also a symmetry.c(r(b))) = d(r(a). Further. This group is called the group ofpermutazions of X (or the symmetric group on X) and is denoted as is a group of order n!.b) = tance between the points a. the set of symmetries of S forms a group under composition of mappings. and the set (Z/(n))5 consisting of the invertible elements in Z/(n) forms a multiplicative group of order /. !fIXl = (d) Symmetries of a geometric figure Consider permutations of the set Xof all points of some geometric figures. when d(a.i(n). then d((ar)(a). permutations of X).b).b E X.) is a semigroup with identity.

(e) Linear groups Let GL(n. Also. Then det(AB) = (det AXdet B) — I. D3 = S3. I—.. Similar considerations apply to any regular polygon of n sides.F) is also a group under multiplication. Then GL(n. and 3—.l. These are commonly written as e=(l 2 a 3 2 a rotation through 4ir/3 is equivalent to. Since D3 is a subset of 53 and each has six elements. aearly.F).F) consisting of matrices of determinant I. SL(n. so Therefore. or is a resultand then again through 2it/3.F) be the set of n X n invertible matrices over a field F. the set of six symmetries is a complete list of symmetries of an equilateral triangle. and 2—.! 2—'3. We denote this group by D3. ant of. Let A. so ABE SL(n.F). called the general linear group (in dimension n). Since any symmetry of the equilateral triangle is determined by its effect on three vertices.B E SL(n. performing a rotation through This explains our symbol a2 for the rotation through Three other symmetries are the reflections in the altitudes through the three vertices. Consider the subset SL(n. . where the "product" is composition of mappings.2 2—'2 3—'l.3 l—'3 These may be rewritten as and respectively.66 Groups l—. called the dihedral group of degree 3. 3—*2 3—'l respectively.F). E SL(n. I.F) of GL(n. 3—.2 2—'2.F).F) is a group under multiplication. 1 A) = = implies = I. namely. E SL(n. This is discussed later in Section 5.

X and .. a a2 ab a2b ab a2b a a2 e ab a e In the following examples we provide solutions to a couple of problems illustrating computational techniques. the symmetries of an equilateral triangle [Example 1... ife1 is the identity ofG1..g E G.. Clearly. I n. Associativity of the binary operation is an immediate consequence of the associativity of the binary operations in each G... 2301 33012 We observe that the elements in each row (column) are distinct. then (e1 is the identity of G1 X . X (g1 be a family of groups. This is a consequence of the cancellation law.and is also written as I1?... ..j) entry is * aj. . I I n. where G (a1 For example.Semigroups and groups 67 (f) Direct product of groups Let G1 G1 X .*) of order n can be exhibited by its multiplication table...g...') is the inverse of(g. as we saw in the proof of Theorem 1.IG. is e e a a2 e a a a2 a2 a2 b b ab a2b a2b ab a a2 b e a2b b b e a ab a2b b ab a2b e a2b b a2 b ab.). Then the cartesian product is a group under the "pointwise" binary operation = whereg. The multiplication table for the group D3 (=S3).. the multiplication table for the group (Z/(4). which is an ii x n matrix whose (i.g.. A finite group (G.3(d)]...+) is +0123 00123 2 11230.2.

so ab Solution. Show that (A. and hence ab = ba. Let H be the subset of C consisting of the nih roots of unity. = a group such that (ab)2 = (ba)2 for all a. Let a. Let A be the set of mappings from R to R. so. that H is a group under multiplication. (a/r'b)2 that is. Solution. a2='ba2b' or a2b—ba2. 3. 'aXba 'b1) Then. 4. by hypothesis. b"a = ab". Prove 2. Then. c Problems 1. then ab = ba. e. from (2) and (3).b E G. a2b2 for all a. Set c = and consider c2 = ab(a 'b— 'aXba 'b') = ab(aa2b = ab(ab 'a2aXba 'b') = ab(ab )(ba 'b = ab(abb2a 1Xba 'b') (abXaba 'b2Xba = = = (abX aba = e.b E G. Similarly.b E G. Show that G is abelian. Let G be a group such that a2 = e for all a E G. Show that . in other words. We have (1) (ab)" = a"b". Suppose G also has the property that x2 — e. by appropriate cancellations. we get = But then b(b"a) = b(ab"). x E G implies x = e.68 G is a group such that (ab)" — a"b" for three consecutive integers. by hypothesis. Let G be a group such that (ab)2 G is abelian. (2) = and (3) From (1) and (2) we get (a"b"Xab) = = This implies.+) is a group under the usual addition of mappings. the square of every element in G commutes with every element in G. Then G is abelian.

ol 1—i 01 0 1]' and 1—i L 0 0 —l form a multiplicative group.tI be groups. (iv) For all a.andb'ab = a3. there is a unique a*EG such that = a.) 6. is abelian. Suppose G is a semigroup with the property that for each aeG. 9. = a*. 69 Prove that the matrices Ii ol Ii [0 four-group. 01101110 Li 0 0 0 —ii' o 0]' Further.Homomorphisms 5. a*aa* = a* and a** = a. (v) G is a group.y E G = . Prove (i) If e is an idempotent in G. aeG. Show that a group in which all the mth powers commute with each other and all the nth powers commute with each other. Homomorphisms 2 Definition. ['PT 0 ol lo b=1. Prove that there exists e E 6 such that e = e2. then x = (ii) If (iii) For all aEG.) Let G be a finite semigroup.0 ol ii' 7.b2 = a2. 1]' 1. . (This group is known as the Klein Prove that the following matrices form a nonabelian group of order 8 under multiplication of matrices: [101 Lo [—1 1. 8. A mapping 4: G-'H is called a homomorphism tffor all x.j1 P11 0 thena4 = e. Let G. x. bEG. 0 01 —'PT]' [—i oJ' [PT [—JET 0 1 [o —11 [ 1. show that if Ii e=1.(Thisgroupisknownasthe quaternion group.x = is a solution of xb = a.. then e* = e. m and n relatively prime.

An endomorphism of G that is both 1-I and onto is called an automorphism of G. For any x in G. then we say that 4) is an isomorphism (or monomorphism) of G into H. then 4) is called anepimorphism. Consequently. onto.2 Theorem. (c) Let G be the group (R'y) of positive real numbers under multiplication. also. 2. though the actual binary operations in G and H may be different. that is. ff4): G —. then H is called a homomorphic image of G. The mapping 4): R given by 4)(x) log x is an isomorphism. if 4) is bzjeciive. and let e' be the identity of H. H is an onto homomorphism. respec- .) 2.70 Groups Furthermore. Hence. is surjective. then G is said to be embeddable in H. I-I. = = a. and let H be the additive group R. Let G and H be groups with identities e and e'. = x= x = y. (d) Let G be a group. then 4) is called an isomorphism of G onto I!. G is said lobe homomorphic to H. (b) For any group G the identity mapping i: G —' G is an automorphism of G. called the inner auzomorphism of G determined by a. and we write GGH. is an automorphism of G. (Note that in the definition of homomorphism we have written both the groups multiplicatively. The mapping givenby 8(x)—e' forallxEG is trivially a homomorphism.A homomorphism ofGintouse!f is called an endomorphism of G. ff4) is just infective. and we write G H. = Hence.xya' = '0 is a homomorphism.1 Examples (a) Let G and H be groups. that is. 'a is injective. !f4): G —' H is a 1-I homomorphism. For a given a E G consider the mapping 1a:GG Since givenby )(aya 1) forallxEG. By the cancellation laws. ff4) is surjeclive.

suppose Ker 4) = (e}. Therefore. where e' is the identity in H. by the cancellation law. H is infective if and only Ker4)=(e). Conversely. Ker 4) = (e}. Let G be the multiplicative group of the nth roots of unity. Let G andif be groups. Then (1) 4)(e)—e'. (ii) = each x E G. Proof (i) 4)(e)4)(e) = 4)(ee) = 4)(e) = e'4)(e). Prove that G (Z/(n). and let 4): G H be a homomorphism. . We noted earlier that every group is isomorphic to itself. 0 It can be easily proved that iff: G H and g: H K are homomorphisms (isomorphisms) of groups. iff G H is an isomorphism of G onto H. Then 4)(x) = 4)(y) 4) is injective. Then 4)(x) = e' = Hence.+). The kernel of4) is defined to be the set Ker4)=(xE GI4)(x)=e'). A homomorphism 4): G —. = Thus. then the composite map K is G also a homomorphism (isomorphism). Further. 0 (ii) = 4)(xx') = gb(e) = e'. Because 4)(e) = e'. This proves that isomorphism of groups is an equivalence relation. Definition. Proof Suppose 4) is injective.Homomorphisms 71 lively. it follows that Ker 4) is not empty. a homomorphism 4): G H not only "preserves" the group operation but also the group identity and the unary operation x '—' x -'. and let x Ker 4). then the inverse mappingf': G is also an isomorphism. H be a homomorphism. Hence. Hence. x = e. çb(e) = e'. and let 4): G —. 23 Theorem. Problems I.

(a) (b) 5. II is a group relative to the binary operation in G. and/el H be a subset of G. = eb E II. suppose that H satisfies (ii).72 Groups Prove that a group G is abelian if and only if the mapping f G —' G. Show that there does not exist any nonzero homomorphism of the group D3 (or S3) (see Example 1. 0 There is a simpler criterion for a finite subgroup. is a homomorphism. written H < G. Subgroups and cosets 3 Definition. For any group G the singleton (e) and G itself are subgroups of G. A subgroup !!of G is said to be a proper subgroup if!! (e)andH*G.) be a group. Section 1) and f(a) = 0. abE II. Let G be a group. suppose H satisfies (i). Then e = bb' C H.3(d)) to the group Z3. Z2. ) be a group. 4. Let (G. and (ii) For all a. given by f(x) = x2. Show that a group G is abelian if and only if the mapping f: G —. Hence. Next. f: Z —. A a subgroup and only if ab C H for all a.2 Theorem. G. which proves that I! is a subgroup of G. Conversely. called trivial subgroups. where G is the quatern on group (see Problem 6. a' C H. a subgroup. (i) and (ii) are obviously true. Find the kernel of each of the following homomorphism: 2. subgroup of G if and only if either of the following holds: (1) For all a.b C H.1 Theorem.b C H.b C H. given by f(x) = f: G -. 1(b) = 1. e = Let a. F) S3. Let F be a field with two elements. H is called a subgroup of G. Therefore.b H. finite subset H of G is . Show that GL(2. Hence. ZN. 3. a subgroup. It is easily seen that the identity element of a subgroup of a group must be same as that of the group. 6. given byf(x) = is a homomorphism: 3. 3. Then for any a C H. ab = C H. Let (G. A nonemply subset 11 of G is a C H.

3 Ker 4) is a subgroup of G and Im 4) is a subgroup qf H. written H = [a). Hence. 0 For an example of a subgroup. then [a) = (kalk E Z) is the set of all multiples of a. Z(G) is a subgroup of G. Then k. 0 are both nonempty. 0 alr'x = = = . hence. ab' E Z(G). and a is called a generator of H. nZ = (nklk E Z). The subgroup nZ in the previous example is thus the cyclic subgroup of Z generated by n. Then for all k. = aexir' = ax/r' = xab'.Subgroups and cosets 73 Proof If H is a finite semigroup. Proof Since ex = x = xc for all x E G. nZ is a subgroup of the additive group Z. The center of a group G.e E Z(G). written Z(G).fl II. Let Hbe the set of general ed by a. then cancellation laws hold for all elements in G. H is a group and. let ) be any group. 3. is the set of those elements in G that commute wit/i every clement in G: that is. Since is closed under . Then for allxE G. Then = 4)(a)4)(b) = e'e' = e' (where e' is the identity of H). = E H for all all powers of a: H = (akIk E Z}. Therefore. let = Im 4). The next two theorems give further examples of subgroups. The center of a group G is a subgroup of G. /1 = 4)(v) for some x. If G is an additive group. Proof Ker 4) and Im 4) 4)(ab 1) a. which proves that Im 4) is a subgroup of Definition. that is. H is called the cyclic subgroup of G More generally. Let a. Hence. Next. Theorem.k' E Z. Hence. = E lm 4). II is a subgroup of G. A group G is said to be cyclic if G [a) for some a in G. where n is any given integer. by Theorem 1. Z(G) = (a E GIax = xafor alix E G}. Let a. Ker 4).h E Z(G). Let 4): G —. Then a = 4)(x). The converse is obvious.b E Ker 4). a subgroup of G.2. consider the set nZ of all multiples of n. Hence.4 Theorem.k' e Z.v E G. H be a homomorphism of groups. nk — nk' = n(k — k') E nZ. Then 3. Therefore. This proves that Ker 4) is a subgroup of G. and let a E G. Hence. they must hold for all elements in H.

if ab c K. Since e = ee C ilK. J1Ci K is not empty. = h3k4 for where h4 = h. Let H and K be subgroups of a group G. then a = abir' E K. = = where k3 = e K.B of a group G. for any subsets A. we define AB = (xy A. Then there exist elements a E I!— K and b K — H. but neither If C K nor K C H. Then h'k' for some h' H. k' K. b = ilK C K!!. then IlK is a subgroup for any subgroups H.b Hfl K. yE B). Let H and K be subgroups of G. Then e C 1!. e K. let be IlK. Hence. Now C KH = I/K. a similar argument shows that the intersection of any number of subgroups of G is again a subgroup of G. air' E 11K. Now a. = h. Hence. air' K. Hence.K of G. (For an additive group G. Hence. in the sense that if L is any subgroup con- . Then a = h. Conversely. air' H fl K. then b = a contradiction. On the other hand. KH C HK. abE H U K. some h3 H.. Hence. k E K. The subgroups of a group G can be partially ordered by set inclusion.74 Groups Let us next see how new subgroups may be obtained from given sub- groups.h3k4 = I:4k4. k4 E K. 0 ilK. Therefore. group Then HK is Proof Let HK =KY. This proves that I/K = K!!. Let H and K be subgroups a subgroup of G if and only if ilK = Ku.b E IlK.) 33 Theorem. Hence.k. Therefore. Hence.b E H U K.h3 C H. If ab H. = C K/I. set A binary operation on a set S induces a binary operation on the power Accordingly. hence. suppose that ilK is a subgroup. yE B). This proves that HK is a sub- group. But the union of subgroups H and K is a subgroup of G if and only if H C K or K C If. Next. we define A + B = (x + ylx C A. It follows from the theorem that if G is an abelian group. lfa. HKis not empty. More generally. . then air' E H. again a contradiction. Therefore. so that a = k/i ilK. H fl K is a subgroup of G. Hence. Then a' = Therefore. Let a for some h E II.k2 E K. Let a. Then H fl K is the largest subgroup of G contained in both H and K. b = /i2k2 for some E H and k. For suppose H U K is a subgroup. a K/f. HK.

Let S be a subset of G. Let S be a nonempty subset of a group G. Hence. ThereforeM'containsall Then foreachx S. Therefore M is the smallest subgroup containing S. then S is called a set of generators of G. then G is said to be a finitely generated group. which is denoted by H V K. for each i. Let G be a group. . as shown earlier. (Trivially. = (ALA is a subgroup of G. IfS is a finite set and G = [5]. Let M' be any subgroup of G containing S.3. and a E G. is not empty.6 Theorem. then [S) is the trivial subgroup (e). therefore. n.x EW. then a is said to be of infinite order. the set G itself is a set of generators of the group G. then HK is the smallest subgroup containing both H and K.thenhkE Mforallh E H. Sc M. If ilK = KH.) IfS is empty. the subgroup generated by S. It is called the subgroup generated by S and denoted by [SI. then m is ca/led the order ofa.Subgroups and cosets 75 tamed in both H and K. EM'. that. Definition. if no such positive integer exists. if there exists a least positive integer m such that am = e. and b = a subgroup of G. 0 IfS is a singleton (a). This proves that M is the smallest subgroup containing Sand. x. If M' is a subgroup and S C M'.k E K. Then Mis a subgroup Since G is in ?. M C M'. written o(a). such that XES or Xr1ES. To go back to the previous case of subgroups H and K. E S or Proof Clearly. For any two elements a = x in M. For if M is any subgroup containingilandK. Consider the family of all subgroups of G containing S. Sc A). . the subgroup generated by Sis the cyclic subgroup [a] = (a'Ii E Z). i= I finite products x1 Mc M'. Let M be the intersection of all subgroups A in of G and S C M. we see that the smallest subgroup containing H and K is the subgroup generated by ii U K. then M' E Hence. Then (lie such subgroup generated by S is the set M of al/finite products E S. That is. If G = [SI for some subset S of G.EvenifHK#KH. we can still find a smallest subgroup containing both H and K. then L C if fl K.

0 I <j m — 1. a' = This implies that H has exactly m elements Conversely. In the multiplicative group G = {xeCIx3 = 1}. For. then the elements a.7 (i) (ii) Theorem. where r(i) is the (iii) [a) is of order m and only t(o(a) = m.Theorderofi in G={xeClx4 = 1} 2 is 4. Find one. Hence.j. (iii) Let 0(a) = m. Hence. 0 (ii) Again by the division algorithm. Therefore by the well-ordering property of N. Further. the element ---hasorder3. a3. of 6 cannot all be distinct.. o(3)=4. i = mq + r. 0 r < m. then o(a)In. 3. (There exist infinite groups in which each element is of finite order. since 2 + 2 = 0. Therefore r = 0 and mm. r < m. Each element different from e in the Klein four-group (see Problem 5. For any i Z. a2. Let H = (a] be the cyclic subgroup generated by a. Then are distinct. a contradiction. if of G. (a) (b) In the additive group Z4.) 3. Hence a' = e for some I> 0. So there exist distinct positive integers i. a' = If a" remainder of i modulo m.. +) is infinite. Then a is of finite order. a' = a'. .8 0 Corollary. Section 1) is of order 2. then there exists a positive integer k such that x* = efor all x E G.j E Z. then a" — e. i <j. Then a"s are not distinct forall i E Z. Hence. efor some integer n 0. (i) If a" = e. suppose that His of finite order. By the division algorithm.j such that a' = (c) (f) = e. there is a least positive integer m = 0(a) such that atm = e. where r = r(i) is the remainder of i modulo m. for otherwise a' for some i. o(2) = 2. the order is 2 and that of a finite group 6 is finite. e = a" = (am )Ga' = a'. for any i E Z. If 0(a) = m. But then H has exactly m elements. If G is aJinite group. which gives The order of any nonzero element in the group (Z. then for all integers i. Let G be a group and a E G. (d) In S3. say m. Hence. as proved earlier.. Hence 0(a) is finite. n = mq + r.76 Groups We now give some examples to illustrate this concept. a' = forsome i.

For all a. G the mappingfi H —' aH. Let H be a subgroup of G.b in G belong to the same right coset if and only if ab' E H. and the set H\G of all right cosets ofH is a partition of G. Thus. — is an equivalence relation on G. Consider the mapping w: H\G given by aH—* Ha'. The set of all right cosets of Hin G is written H\G. Consider the relation on G.b. therefore w is bijective. where a b means a 'b E H. is obviously surjective. Hence. Definition. the set G/H of left cosets of H in G is a partition of G. we can prove that any two right cosets of H have the same cardinality. G/H and H\G have the same cardinality. A subset C of G is called a left coset of H in G if C = aHfor some a in G.Subgroups and cosets 77 Proof Since G is finite. b'c E a'c = (a'bXb'c) E H. Then x E G. 0 Definition. Choose k = IICEGn(a). = Hb'. By a similar argument. For any a E a'a = e E H. Note that H itself is a left coset of H. a'b. say n(a). the cyclic subgroup (a] is also finite. since eH = H. is clearly a bijection. Let H be a subgroup of G. Consider now the right cosets of H in G. The set of all left cosels of H in G is written G/H. Let us confine our attention for the present to left cosets aH of H in G. Two elements a. the distinct left cosets of H are pairwise disjoint and their union is G. we have the following definition. Consequentl). That is. It is easily verified that the equivalence classes of— are precisely the left cosets of H in G. Given a E G. Hence.' == Hb' = all = b!!. every left coset of H has the same cardinality as H. The cardinai number of the set of . Hence.c E G. By arguments similar to those given above. 0(a) is finite. w is injective. a'b E and (a'b)' E H. the set aH=(ahlhE H) is called the left coset of H determined by a. A right coset Ha is defined similarly. Hence. Therefore. given byf(h) = ah. The mapping is well defined because H=*a (b')' E Moreover.

. Then for every a E G. hence. the index of H is equal to the order of G.9 Theorem (Lagrange). Let G be afinite group oforder n. and therefore G is cyclic. We shall now derive some important results from Lagrange's theorem. Otherwise.. every finite group of prime order is cyclic and. abelian.78 Groups left (right) cosets of H in G is called the index of H in G and denoted by [G:H]. [Z: K] is finite. This proves the following theorem. By Lagrange's theorem the order of the cyclic subgroup [a] divides n. Let G be a finite group. [G: 1] = [G:HJ[H: I].10 Corollary. Indeed it is easy tc• check that K = nZ. n 0. and let If be any subgroup of G. Hence. hence. each left (right) coset of If in G is a singleton. If G has an element a of order 4. 3. Writing I for the trivial subgroup {e). and so Z/nZ = 7. In other words. Hence. Proof Let a C G. [a] = G.. For each nonzero subgroup K of the additive group Z. Then every left coset of H has m elements. Let IGI = IHI = m. 0 3. then [a] = G. and. consequently. which is of fundamental importance in the theory of finite groups. we get IGI = (G: 11. Then the order of any subgroup of G divides the order of G.3. o(a)In. = e. So o(a)In. If II is the trivial subgroup {e). Consider a group G of order 4. 3. or 5 are cyclic. Hence. every element # e is of order 2.11 Examples (a) (Groups of order <6) It follows from Corollary 3.10 that groups of orders 2. where k is the number of left cosets of H in G. If n is prime and a e.3(a)). G is cyclic. Let G be a finite group. we must have n = km. Since the distinct left cosets of If are pairwise disjoint and their union is G. the order of [a] must ben. (Example 1.

Solution. If o(a) = m. This yields (c) (Poincaré's theorem) The intersection of two subgroups of finite index is of finite index. = I. and (m. a contradiction.m) I. • = akbk = e which gives kfrnn (Theorem 3.) Thus. then where — 1 (mod rn).Fermat theorem) If a is an integer prime to the positive integer m. Let H. o(b) = n. ba = c.m) = 1. any right coset of H fl K is the intersection of a right coset of H and a right coset of K. Let o(ab) = k. b€G such that ab = ba. Consider the product ab. Hence. but the number of such intersections is finite. G is abelian. Hence. Also ak = (Problem 10). then o(ab) = mn. Then it is trivial that (H fl K)a = Ha fl Ka. o(ak) = o(b") = o(bk) . which implies b = a.7).K < G. is the Euler function. ab e. (b) (Euler. bc = a = cb. then ab = aa. Hence. and ca = b = ac. (d) Let G be a group and a. ab = c. is called the Klein four-group. ab b. Solution. ab '# a. Note that an element i E Z/(m) is invertible if and only if (x. Solution. This proves that the number of right cosets of H fl K is finite. Let a E G.b.a. or any group isomorphic to it. all groups of order <6 are abelian. Consider = (ab)k = e Thus.K of finite indices. By the same argument. Then (ab)k = e. n) = 1.Subgroups and cosets 79 that is. H.c) and a2 = b2 = c2 = e. If ab = e. G = (e. Thus. äE (Z/(m)). so tible elements in Z/(m) is I (mod m). and the order of the multiplicative group (Z/(m)) of the inverSince (a. (This group. Similarly.

(e) Let G be a finite group. Then ISTI= ISIITI ISn 71• in S X Tby (s.f is clearly surjective.t1) be the disjoint union of equivalence classes. Consider S X Tand define a relation if and only if s' — sa and 1' (s.t) — (s'. Let (s.80 Groups But am = e = e : o(ak)Im Therefore. Then — is an equivalence relation. it follows that k = mn.t) E S X T.z) denote the equivalence class of Solution. ISX T/-I=ISTI. so from (I) we get ISX TI=ISTIISfl fl. (Theorem 3. and S. ISXTJ=kISflTI. k = ISTI. . 1. Consequently. and let S X T/— be the set of equivalence classes.fis well forsomeaESfl Also. T < G. But then m(k. The result contained in the following problem is sometimes quite handy when studying structure of finite groups. Since klmn.l') for some a E Sn T. Similarly.. 1. We now define a mappingf S X TI— defined and injective. divides (m. Hence.7). Therefore..n) soak = e. It follows from the definition of equivalence relation that (. Hence. because (I) = s1t. E T= (s. Similarly.)= S Therefore.

Let Gbeagroupand H<G. Show that the vectors that issue from the origin 0 and have endpoints on a fixed line through 0 form a subgroup. where p is prime.b E G. Let G be a group and a. Also show that HH = H. 10. 5. 2. ShowthatforallxEG. If a. Suppose IG/HI = 2. 13. Show that 0(a) = ii. Find the subgroups of the groups given in Problems 5 and 6 of Section I.b are any two elements in a group G.x'Hxis a subgroup of G of the same cardinality as that of H. Show that there exists an element c E G such that 0(c) is the least common multiple of m and n. (a) If H and K are subgroups whose orders are relatively prime. 8. . Show that for any subgroup of a group the inverses of the ele6. 4.and cosets 81 which gives ISV TI = ISTIIS fl fl or ISTI = Problems 1. respectively. IfA is a left coset of some subgroup of G. Let H be a subgroup of a group G. LetGbeagroupanda. ments of a left coset form a right coset. 3. More generally. 9. Show that 0(a) = Ifa group G has only one element a of order n. Let Vbe the group of vectors in the plane. IS. Prove that aH Ha for all a E G. prove that H and K are subgroups of order p and n.b E Gsuchthatab —ba. 12. Show that a subset A of a group G cannot be a left coset of two distinct subgrotps of G. Let G be a group and H < G. o(a') for any element a in a group G. Show that if G is a group of even order then there are exactly an odd number of elements of order 2. with vector addition as the binary operation. Show that the elements of finite order in any abelian group form a subgroup. if A is a nonempty subset of G then show that AH = H if and only if A H. then prove that A is also a right coset of some subgroup of G. show that ab and ba have the same order. 14. Prove that Ha = H if and only if H. What are the cosets relative to this subgroup? 7. then either Hr' K = {e} or H <K. Leto(a) = mand o(b) = n. then a E Z(G) and n=2.

It is clear that w is a surjective homomorphism.7). cv is an isomorphism. Prove Wilson's theorem that. a' = is surjective. Let G be a finite group. Therefore.1 E Z/(n). tralizer of S in G. Prove that IGIISI+ITI. 20. for any prime p. and {(l 2 3). Let S be a nonempty subset of a group G. {(1 2)). for all s E S). Next. then o(g(a)) = o(a). a group G is said to be cyclic if G = [a] = (a'Ii E Z) for some a E G. and let C(S) = {x E GIxs = sx. 21. Cyclic groups As already defined. 0 . Theorem. for let 1. G given by qi(i) = a'. Every cyclic group is isomorphic to Z or to Z. Hence. Let m be a positive integer such that (m. 17. Then G = j and o(a)= n (Theorem 3. In fact. suppose G = [a] is a cyclic group of finite order n. Let m be a positive integer. Further. is Hence.1 Proof If G = [a] is an infinite cyclic group. (1 2)). 19. Letg: G —p G' be a homomorphism of groups. Prove that o(g(a))Io(a). What is C(G)? (C(S) is called the cenFind the centralizer of each of the following subsets of S3: {(l 4 2 3)). Let a E G.r) I.) 22.r). Further prove that if g is injective.n. cv = ci4i+j) — = a'a1. ci' is well defined and also injective. Prove that o(am) r. consider the mapping w: Z —. I a' * a1. Let a be an element of a group G such that 0(a) = r. these are the only cyclic groups up to isomorphism. Then = Clearly. and let S. Prove that o(am) = r/(. The most important examples of cyclic groups are the additive group Z of integers and the additive groups Z/(n) of integers modulo n. Consider the mapping cv: Z/(n) —' G given by yi(i) = a'. 18. Show that C(S) < G. wis injective. Moreover. an isomorphism. (p — I)! — (modp). 4.1n) for some n E N. Let a be an element of a group G such that 0(a) = r. T be nonempty subsets of G such that G ST.82 Groups 16. for otherwise a would be of finite order.

Proof Let G [a) be a cyclic group. Theorem. So let Jibe a proper subgroup of G.5 Example Let H = [a) and K [bJ be cyclic groups of orders m and n. and let Jibe a subgroup of G. such that (m.b) generates the group H X K. By the division algorithm Then a' a'" = a' H [bJ.b)"" (a"". Let G be afinite cyclic group of order n. Hence. nisd. Let G = [aJ.3. So let 1 <d < n and put n/d = m.b)" = (a41. mnld. Ifa1 E H. This proves H = [b]. Then b atm is of order d. Consequently. Let a' E H. respectively.Cyclic groups 83 The following is immediate from the above theorem. Hence. To prove the uniqueness. let Jibe any subgroup of G of order d. If H is a trivial subgroup. We prove that H [bJ. Then H X K is a cyclic group of order mn.b) — d.3 Theorem. Let —s.e) = (a. the result is obvious. mn — d. .n) I. Solution. 4.bd) implies a" = e b1. Then azd=cd—_e. By Theorem 4. but the following theorem shows that it does hold for cyclic groups. 0 The converse of Lagrange's theorem is not true in general. Also (e. 4. Hence. Prove that the multiplicative group of the nth roots of unity is a cyclic group of order n. Let o(a. [a'] = [a"].e) implies dimn. that is. Proof The result holds trivially if d = I or n. Now (a. mdIsd and so mis. But since each of the subgroups [a'] and [a"] is of order d. Since Ill X = mn. This yields [a'] c= [atm] = [b].2 flue) are isomorphic.4 Theorem. Therefore. 4. (b) is a cyclic subgroup of order d. and let d be a positive divisor of n. 4. H is generated by an element c = as. there is a least positive integer m such that b atm E H. Problems I. so mid and nid. it follows that (a. Every subgroup of a cyclic group is cyclic.bm") = (e. Then G has exactly one subgroup of order d. then a' E H. Any two cyclic groups of the same order (finite orinfi.

and the bottom row has 0(1) below i for each i — .. E n. The group of all permutations of X under composition of mappings is called the symmetric group on X and is denoted by A subgroup of is called a permutation group on X. Is Z2 x Z2 cyclic? Give an example of a nonabelian group such that each of its proper subgroups is cyclic. 4. Show that every finitely generated subgroup of(Q. +). S1 is denoted by symmetric group of degree n.. This is called a two-row notation for a permutation. (Z/(n). Prove that there are exactly two elements of an infinite cyclic group G that can generate it.n.. There is a simpler.84 Groups 2. 8.. 6.n. 0(n))' consisting of two rows of integers. and 12 of Z12. S Permutation groups Definition.. one-row notation for a special kind ofpermutation called a cycle. 7. 3. .. Let a E If there exists a list ofdis:inct integers x1 .x.. .. Show also (Q.1 Show that a group G has no proper subgroups if and only if it is a cyclic group of prime order. [Hint: If a is a generator of 5. the top row has integers I .. usually (but not necessarily) in their natural order.. and called the If lxi = n.. Let X be a nonempty set. where represents positive rational numbers. so ma 1 (mod n)... A permutation a E can be exhibited in the form (i \c(l) 2 c(2) . Prove that there are exactly çb(n) elements of a cyclic group of order n that_can generate it. 10.. 9.. then I = ma for some meZ. Show that Z x Z is not cyclic. It is easily seen that a bijection X V induces in a natural way an isomorphism S1. 1 Definition.. Find a homomorphism from S3 onto a nontrivial cyclic group. which yields (a. + ) is cyclic.4. Find all possible sets of generators of the subgroups of orders 3.n)= 1.

that is... then a is called a cycle of length rand denoted by (x1 .Permutation groups 85 such that o(x1) = i= 1. e=(l). We can find the product of any two permutations by the usual rule for . Two permutations a....2.)... ... Note that the one-row notation for a cycle does not indicate the degree n.3) given by II (I 2 2 3\ 3)' 11 2 3 2 2 3\ i)' Ii 2 I 3\ 2)' 3 3 2 3 3\ 2)' a1 (1 3\ i)' 3 fi 2 1 On examination we find that they are all cycles.. which has to be understood from the context. one step around a circle as shown in the diagram (for r = 5) and leaves every other integer in ii unmoved..) moves the integers x. A cycle oflength is called a transposition. we say a does not move x.. a2=(l I). z1=(2 3). let us consider the symmetric group S3.r E are said to be disjoint if they do not both move the same integer. a(x) =x 2 if x {x1 .x..Xr)..) = x1. t3=(l 2). any cycle of length I is the identity mapping e.) x x4 2 I Trivially. t2=(3 =(I 2 3).. For the sake of illustration. In other words. There are six permutations of the set (1.x. a(x) = x or r(x) = x. that is.x. a cycle (x1. 2). for all x E n.r— I a(x. (If a(x) = x.

4:G—'S0 = abx Hence. In A subgroup of a symmetric group other words.b.a2r=r2=w. a permutation group is a group ) such that the elements of G are permutations of some set X.r. because ax = ax' x.1 Theorem (Cayley).y E G.a. Every group is isomorphic to a permutation group. 5.r2=e. We shall later see further examples of these relations.x E G.or. Consider the mapping givenby forallaEG. (This is not true for symmetric groups of degree >3.) We can. For all a.86 Groups composition of mappings. Thus.a2. is called a permutation group. =fa(f.(x)) = 4 is a subgroup of SG. Hence. ro=a2r (*) and the associative property of multiplication. For any given a E G. 4 givenby L(x)=ax forallxEG. give a simpler description of the multiplication structure in S3.c3=e.a2r). More- . followed by r. how- ever. a acts first. S3 = (e. Relations of the type (*) that completely determine the multiplication table of the group are called its defining relations. and• is composition of mappings. the mapping fa:G'G.x '. 11 2 2 3\ (Recall that in the product ra. The following theorem indicates the importance of permutation groups. We easily verify that o2=a2. Proof Let G be a group.ar=r3. and the product of any two elements in S3 can be computed by using the relations o3=e=r2. where SG is the symmetric group on the set G.) Let us write a for a1 = (1 2 3)and r for = (2 3). x = x' and y = for all is a bijection.

The above isomorphism is called the lefi regular representation of G. Therefore.. This proves that is a subgroup of and. Let us now consider the general case of a regular polygon (see the if and only is asymmetry figure).y) for all x.acr '(y)) = d(x.rcr t(y)) = d(a '(x). that is.Permutation groups 87 over.y E X. denote the set of all symmetries of X. rcr' E is itself a group under composition of mappings.a(y)) = d(x.T E and x. DefinitIon.y).3(d). Similarly. Consider.y is given for all x.. which can be labeled as I . if and .. In other words.a '(y)) = d(ao '(x). the group of symmetries of a polygon of n sides is a subgroup of Hence.y) between points x. forall is an injective homomorphism.y E X. Let Xbe a set of points in space.. Let T1. It is clear that a permutation a if a takes any two adjacent vertices of to adjacent vertices. we have a right regular representation. A permutation tiofXis called a symmetry of X if d(o(x). G is isomorphic to a 0 SubgroUp of Hence. a symmetry is a permutation that preserves distance between every two points. It is clear from geometrical considerations that any symmetry ofXis determined uniquely by its effect on the vertices of the polygon. Thus. Therefore we need consider only the symmetries of the set of vertices.2. in particular. Groups of Symmetries We now describe an important class of permutation groups known as groups of symmetries.n. therefore. The group of symmetries of a regular polygon called the dihedral group of degree n and denoted by The particular case n = of n sides is 3 was considered in Example 1. the case when Xis the set of points constituting a polygon of n sides in a plane. It is called the group of symmetries of X. Then for all O. d(ra '(x).yEX. so that the distance d(x.

. Hence...n — I). the symmetries of Pn can be classified into two kinds: those that preserve the cyclic order I .. Hence. Clearly. The dihed ral group isa group oforder 2n generated ...n — I .a(n) are either in the cyclic order 1.n. Thus... 1. Therefore..ando"takesl to l. a'r = ... Writing a for venfythato'takesl toi+ 1(1= l.1..n).2 n or in the reverse cyclic order n. takes I to i. is the identity we easily permutation e. Now consider the product ra. Moreover. we conclude that ta= discussion in 5. i=I. is the symmetry taking I to i.. Let a be a symmetry that preserves the cyclic order.... Now a( 1) can have any one of the values 1 . the n symmetries that preserve the cyclic order of vertices are a' (i=O..n— 1). 2. Hence r2 — e.88 Groups 3 2 n only if a( I ).2..2. a(2).n— 1).2 We sum up the above Theorem. By the same argument there are exactly n symmetries that reverse the cyclic order.Sincec' (I> 1) preserves the cyclic order.. This proves that the dihedral has 2n elements a..n. Since ta reverses the cyclic order.a(2)... where r....n). and once a( I) is fixed. Let us call them a1 .. and w(l)= r(2)= n.a2 where a.. . = e.2. and 02 is the cycle (1 2. there are exactly n symmetries that preserve the cyclic order of the vertices....n—l...+i.. where a is the cycle (I Writing t for 11 2 n .. 2)' we clearly see that a'r reverses the cyclic order and takes ito i + I.o(n) are uniquely determined by the cyclic order... it follows that 01=C.1. and those that reverse the order... r2 takes Ito 1 and preserves the order... (i = group We now give a simpler description of the elements of that enables us to write the product of any two elements easily.÷1 (i = O....

3. we shall consider the symmetries of a rectangle in Example 5.j. DefinitIon. Problems 1. ab = c. i *j.t satisfying a" = e t2 and to. (Hint. . Let a and b be elements of order 2 in a group G.Permutation groups 89 by two elements a. and r is a reflection (or a turning over) in the diameter through the vertex 1. Let e # a E Suppose o(i) =j. = I or 2 according to whether n is odd or even. 5. and b and c are reflections in the axes of symmetry parallel to the sides. r(k) =j. Let k i. Suppose o(ab) 4. a is a rotation of the regular polygon through an angle 2n/n in its own plane. the group of symmetries of a rectangle is the Klein four-group. 2.3 Example The symmetries of a rectangle 4 3 I 2 are easily seen to be e=1'I 2 3 4'\ a=(i 2 4 3 2 3 4 1 4 Geometrically. Thus. 4. Find all subgroups of 53 and S4. ca = b. As an example of the group of symmetries of a nonregular polygon. bc = a. so at # to-. a is a rotation through an angle t.= a" 'r. Then (arXi) # (taXi). It is easily verified that a2 = b2 = c2 = e. The dihedral group D4 is called the octic group. prove that Show that I = (e). Show that the subgroup generated by a and b is D4. If n 3. Choose r E such that t(i) = r(j) = k.) 3. where n 2 t1(1 2 n Geometrically.

and also in the set of equations satisfied by elements in the generating set that would suffice to give us all products of elements in the group. but clearly this is an impossible act for groups of large orders (without computers!). A set of equations (r1 = l)JEA that suffice to construct the multiplication table of (i is called a set ofdefining relationsfor the group are products of ci eFnents of X). The group S3 is generated by {a.. The set X is called a set of generators.+) is generated by I such that n I = 0. is generated by (a. in gen= eral.b) such that = I = b2. ')JEA) is . Let G be a group generated by a subset X of G. Hence. we are naturally interested in a minimal subset Xof the group that generates it. A finite group. may also be described by listing its elements and by writing down the multiplication table. b 'ab = a2. Definition.90 Generators and relations 6 Generators and relations Sometimes it is convenient to define a group by a subset X of the group and a set of equations satisfied by elements of X. The system called a presentation of the group. the group (Z/(n).b) such that a3 = I = b2. For example. The dihedral group 0.

Let G be a group. If G is abelian. we write aB for (a)B and A1 for A(b). and if 4): G of groups.5(e)1. Definition.CHAPTER 5 Normal subgroups 1 Normal subgroups and quotient groups We recall that multiplication in a group G induces a product of any two subsets A and B of G. ifxNx' C Nfor every x E G. 1. Trivially. . Then the following (I) (ii) (iv) are equivalent. Let N be a subgroup of a group G. given by AB = (xylx E A.1 Theorem. If A or B is a singleton. The following theorem gives several characterizations of a normal subgroup. A subgroup N of G is called a normal subgroup of G. the induced multiplication of subsets of G is also associative. NIG. It is easily proved that the center of a H is a homomorphism group G is a normal subgroup of G. the subgroups (e) and G itself are normal subgroups of G. But the con- verse is not true: A group in which every subgroup is normal is not necessarily abelian [Example 1 . xyN for all x. written N G. then Ker 4) <1 G. then every subgroup of G is a normal subgroup. Since multiplication in G is associative.y G. x (xNXyN) 91 every x G. yE B).

Let N be a normal subgroup of the group G. N = Hence. Because multiplication is associative in G. Now NN C N eN= N. xNx (iii) (iv) = x(Ny)N = x(yN)N — xy(NN). We shall write all cosets of N as left cosets and denote the set of all cosets of N in G by GIN. Hence. Hence. C xNx'. Also. The homomorphism G GiN. Proof By Theorem 1.1. (xNXyN) = (xy)N. is a surjective homomorphism. Definition. we need not distinguish between left cosets and right cosets of N. C N. C N. Definition. and let S be a nonempty subset of G. Let x E G. 1. The group GIN is called the quotient group of G by N. For any non empty subset SofG. and Ker 4 = N. x' E G. Theorem. by definition.1 G. 1. Therefore. 0 (iv) (i) xNx' = xNx'eC xNx'N= Let Nbc a normal subgroup of G. because N is closed under multiplication. Hence. xN eN if and only if x E N.2 is a group under multiplication. given by '-' xN. This proves that GIN is a group. (xNXyN) (xy)N for all x. multiplication is also associative in GIN. Further. Therefore. N = eN C NN. eN. given by x xN. NN = N. The mapping 4 is obviously surjective. The above theorem has shown that any left coset of N in G is a right coset and vice versa. and for any x E G. N(S) Ls a subgroup of G. Hence. qS is a homomorphism. Further.y G.3 . Let G be a group. Let G bea group. Let N be a normal subgroup of G. Hence. The mapping x Theorem. The normalizer of S in G is the set N(S) (x E S). GIN is closed under multiplication. Then. The coset eN = N is the identity = = for multiplication in GIN. The normalizer of a singleton (a) is written N(a). for any subgroup H of G.y G.92 Normal subgroups (1) Proof (ii) Suppose N <. çb(xy) = (xy)N = 0 Hence. For all x. On the other hand. N<G. Then GIN G —' GIN. is called the natural (or canonical) homomorphism of G onto GIN.

then H is a normal subgroup of KY. e E N(S). and H C KH. Consequently. = = G'. N(S) is a subgroup of G. G' is a normal subgroup of G. K C N(H'). If H G. Therefore. II IlK. commutator in G. hence. by definition. Hence. then = x '(ySy' )x = x'Sx = )x = S. Let G be a group.5 of Chapter 4.b E G. H is a subset and. Hence.. If x. (ii) . G'. Then for all k E K. Proof Clearly. Further.. Hence N(S). 1. kH = Hence. Let H be a subgroup of G. 0 a Definition. The subgroup of G generated by tile set of all commutators in G is called the commutator subgroup of G (or the derived group of G) and denoted by G'.Normal subgroups and quotient groups 93 N(H) is the largest subgroup of G in which H is normal: if K is a subgroup of N(J-I). Then k E K.4 Theorem. (iii. for any gin G. Then (1) G"IG. GIG' is abelian. This proves that N(H) is the largest subgroup of G containing H as a normal subgroup. then G/H is abelian if and only if G' C 1!. H N(H). Then for any g G. say where x1 are commutators. Let Kbe a subgroup of N(H). a subgroup of N(H). Moreover. Let K be any subgroup of G such that II 'I K.) Proof (i) Let x= be any commutator in G. by Theorem 3. b E G. xHx' = H for all x E N(H). . XII is a subgroup of N(If). and let G' be the den ved group of G. = (gag Xgbg'Xgag E G'. For all a. yE N(S). Let G be a group. For any a. Then bab'a' is also a commutator. Therefore. = Hence. Then hllh -' = H for all h E H. Now any element yin G' is a product of a finite number of commutators.

. G = H U all and aH fl H = 0. Ha 0 H = 0. H. C Nfl M= {e).'ar' E AB. let a E A. The converse is proved (a)lfA <Gand BIG. axa' E A. m E M. Va E H. because AB<G. let a. But clearly aH = Ha. (iii) Suppose G/H is abelian... I = l.VnEN.Hk < G. GIG' is abelian. Hence. then nm=mn. (d) If N and Mare normal subgroups of G such that Nfl M = (e). Thus. similarly. Then for all a.. To prove its normality. Also.5 Examples = II.thenA flBlAandAB<G. Solution. by hypothesis. 0 1. (aG')(bG') = (bG'XaG'). 1m1 nm = (n and also NN=N. VxEAflB. proving that H is normal mG.k. then H is a normal subgroup of G. Then VaEA. Then ab(a1 b1 = B. To prove AB < G. Solution. gH Hg.94 Normal subgroups Hence. b1 for some b2 B. a H. Thus. (b) b. Thus. all = Ha.. Solution. Obviously A fl B <A. Vg G. A fl B. then n)m E MM = M.b E G.2. because B <I G. a1 E A. If n N. This proves that G' C H. )' ci G. If a H.. then HIH2. Trivially. = abb. (aba 'b ')JJ = (aIIXbHXaH) Hence. Solution. (c) If G is a group and H is a subgroup of index 2 in G. x E E B.VmEM. G = H U Ha. . Follows by induction on k. Therefore. Thus. then.

and thus K <3 T. Clearly. IG/N1) = 1. Chap- J(i o\ f—i o\l 0 —i)J' I\o is trivially normal All the subgroups of order 4 have index 2 and.Normal subgroups and quotient groups 95 Hence.2 in Chapter 4). IG/NI) = I.b E G. (f) Give an example of a group G having subgroups Kand Tsuch that K 1 T <3 G. aZ(G) = xmZ(G). Choose T = (e. Let K < G such that 1K] = Ni. The only subgroup of order 2.r. hence. Let a. are normal by Example (c). K = K fl N. Hence. (e) Give an example of a nonabelian group each of whose subgroups is normal. Solution. but K is not a normal subgroup of G. Section ter 4). that is.r). that is. from the defining relations of the octic group. K is not normal in G.c2r) and K = (e. and if G/Z(G) is cyclic. this implies IKI/IK n NI = I.c2. Solution. (h) If G is a group with center Z(G). Let G be the quaternion group (see Problem 6. because if we choose a E G and r E K. Solution. XE G. Now KN IKNI N IN1IKflNI So by Lagrange's theorem KI/IK fl divides IG/NI. 1. then a'w K. T and K are subgroups. Let G/Z(G) be generated by xZ(G). then G must be abelian. Since the 2. Solution. G is a desired group. so K=N. Then aZ(G). But since = INI and (INI. However. Also K as a subgroup of T has index 2. T <1 G by index of T is Example (c). Let G be the octic group 1)4 (see 5. Then KN/N < GIN. as an element in G/Z(G). nm = mn. . Then a normal subgroup N such that N is the unique subgroup of G of order IN1. Hence. (g) Let G be a finite group with (IN1. must be of the form xmZ(G) for some integer m.

Prove that 11<1 G if and only if the product of two left cosets of H in G is again a left coset. (1 32)) is a normal subgroup. 9. show that H G. since yE for some z E Z(G). Prove that in S3 the set N= (e. prove that NM is also a normal subgroup of G.96 Normal subgroups Thus. 12. Generalize this result to subgroups of other orders. Prove that H ci G and G/H is abelian. If N is a normal subgroup of a group G. 7. b = Z(G) and some integer n. z E Z(G).b E G. Find the elements of S3/N. Similarly. If N and Mare normal subgroups of G. 2. a = xmy for some yE Z(G). so that ab = (xmyXxnz) = xmyxnz = xmxnyz = xm+nyz. Va. Show that the order of each element of the quotient group Q/Z is finite. Find the elements of S4/N and show that they can be made into a group. If there exists a unique subgroup H of order 10 (or 20) in a group G. prove that NC Z(G). 4.) is Show that the cyclic subgroup of S3 generated by not normal in S3. If N is a normal subgroup of a group G such that Nfl G' = (e). (I 23). 13. = since And ba = (xnzXxmy) = Hence ab = ba. prove that NH is a subgroup of G. Suppose o(gN) is finite. (Hint: See Problem 3. R). Let GIN be the quotient group of G. Let H be a subgroup of G. 5. and His any subgroup of G. Prove that the center Z(G) of a group G isa normal subgroup. R) of invertible n x n matrices over R having determinant 1 forms a normal subgroup of the linear group GL (n. Show that o(gN) divides o(g). 8. . 10. 6. Also show that g'4cN if and only if o(gN) divides m. Section 3. Problems 1. Let H be a subgroup of a group G such that x2 E I! for every x E G. Find a subgroup N of order 12 in the symmetric group S4 on four elements. 3. Show that the set SL (ii. 11.

compute the first and second derived groups. 16. Let L(G) be the set of normal subgroups of a group G. 17. BIG. Then . then G is called a semidirect product (or split extension) of A and B. extension of two suitable proper subgroups. that is.cEL. Show that the quaternion group (see Problem 4. nontrivial subgroups. For G = S3. for all a. [A lattice L is called modular if for all a. Show that L(G) is a modular lattice. Show that there does not exist any group G such that IG/Z(G)I =37. (ab)"' = a"b"'. G and the order of each element of GIG"' is finite. with a c. If A. bEG. a v (b A c) =(a v b) A c.] 2 Isomorphism theorems In this section we prove some important theorems on group homomorphisms known as isornorphism theorems. = G' and = (G'). and let = Show that G"' 15. B are subgroups of a group G such that G=AB.1 Theorem (first isomorphism theorem). every homomorphic image of G is isomorphic to a quotient group of G. Section 1. Chapter 4) cannot be expressed as a split extension of two 19. Show also that the lattice of all subgroups need not be modular. We saw in Theorem 1. Let 4: G/Ker 4 Im G —. 2. 97 Let G be a group such that for some integer rn> I.2 that every quotient group G/Nofa group G isa homomorphic image of G.Isomorphism theorems 14. Show that the dihedral group can be expressed as a split 18. G' be a homomorphism of groups. The first isomorphism theorem (which is also known as the fundamental theorem of homomorphisms) proves the converse.b.

Any homomorphism 4): G —* G' ofgroups can befac4) =j. G 17 / G/KerØ Proof Clear.n. and N G. Since IL" is obviously suijective. if 4) is surjeclive. sv is a homomorphism." . Then H/Nfl N fIN/N.98 Nonnal subgroups Hence.3 Theorem (second isomorphism theorem). Let H and N be subgroups of G. Im 4) is the isomorphism obtained in the theore.2 loved as Corollary. 0 2. For any x.p is an isomorphism of groups. = cb(xy) = Hence. Further. we conclude that . Proof Consider the mapping v:G/K—1m4) xK = yK givenby where K = Ker 4). Because of this.y C y XE K = x) = e' 4)(x) = Hence. w is well defined and injective. G/Ker 4) — n'here G —' G/Ker 4) is the natural homomorphism. andj: Im 4) —' G' is the inclusion map. 0 2. in particular. The inclusion diagram shown below is helpful in visualizing the theorem. the theorem is also known as the "diamond isomorphism theorem. 1/zen G'.

is clearly surjective. Consider the mapping givenby forallhElf. Further. 4. Let H and K be normal subgroups of G and K C H. Hence. (G/K)/(!f/K) GIll. Now 4. 4." Proof Consider the mapping 4. Hence. H/Il fl N HN/N.IsomorpMsm G 99 FIN N HnN Proof Since N <1 G. is a homomorphism. 0 The next theorem shows that the direct product of two quotient groups is isomorphic to a quotient group. Hence. is the restriction of the natural homomorphism p: G —. is obviously surjeclive. = (xKlxH = H) = (xKIx E if) Hence. = H fl N. This theorem is also known as the "double quotient isomorphism theorem. GIN to H.(xyK) = xyH = (xH)(yH).: Gill given by xH. by the first isomorphism theorem. The mapping is well defined.yE G. HN = NH is a subgroup of G and N <1 HN.4 Theorem (third isomorphism theorem). 4. for xK = yK x'y E K EH xH = yH. Then (G/K)/(H/K) G/H.((xKXyK)) = 4. H/K. Ker 4. In fact. for all x. . Moreover. 4. by the first isomorphism theorem. 0 2. and Ker 4.

We may emphasize that c 'isa mapping not from Tto Sbut rather from to However. N2 (G. (x.b E H. G2./N1) X (G2/N2) given by (x. Thus. and if X C T. .: G — G' be a homomorphism of a group G onto a group G'. so 4. ifais bijection then defined as above coincides with the usual inverse. X G2)/(N. furthermore. = N. 1 G. H' <G' The 4r'(H') 1 G./N1) X (GilN2). the result follows by the first isomorphism theorem. Proof (i) Let a. X N2) = (G.b so = H'. X N2. < G. Theorem (correspondence theorem).napping H 4. çb(H) < G'. Let G1 and G2 be groups.N.x2) '—.(a). Then 4. Now (i)' Let a. Then the following are true: 2. It is easily verified that 4.100 Normal subgroups 23 Theorem. and the family of subgroups of G'.cb(b) E 4. Let 4.6 (i) (i)' H' <G' (ii) (iv) H<G=ç6(JI)<G'. Hence.: G. and N.(H). . H< G andH3 H= . then r'(X) denotes the set (s E Sic(s) E X). Then Proof Consider 4. X G2 —' (G.(H) is a 1-1 correspondence between the family of subgroups of G containing Ker 4. 4r'(H'). e H..x2N2).(H). normal subgroups of G correspond to normal subgroups of G'. 4r'(H') <G. . c '(X) is known as the inverse image of X under a. is a surjective homomorphism and Ker 4. 0 We recall that if c: S Tis a mapping from a setS into a set T.(a)4.(b) E H'. Thus. Then = since = E 4.

Let G be a group. 4r'(H'). 0 Definition. Given any subgroup H' ofG/N. Corollary. let = 4)(H2). then Theorem 2. Therefore. By Theorem 2. so by (iii). . 4)(4r '(H')) = II'. A group G is said to be simple if G has no proper normal subgroups. by (iii). To show that the mapping is injective. g E G. So the mapping H'—' 4)(H) is surjective. = The last part of(iv) follows from (ii). Therefore. Further. that is. Theng' =4)(g) for someg E G.6 there is a unique subgroup HofG containing Nsuch that H'==4)(H)=H/N. there is a unique subgroup H of G such that H' = H/N. Then by (i)' isa subgroup of G containing Ker 4). Hc 4)(x) E 4)(H) I.et xE Hence. 2.H2 are subgroups of G containing Ker 4). (iii) Trivially. 4r'(H') <1G. Definition. 0 Remark. Then 4)(h) E H'. Then = so. where H. Now g' 1(4(h))g' = (4(g)) (ii)' Let Ii L = since H <1 G. H. 4)(H) <G'.Isomorphism theorems 101 (ii)Let4)(h) E4)(H)andg' E G'.6 remains true if G' is replaced by Im 4). A normal subgroup N of G is called a maximal normal subgroup if (i) (ii) N*G. Then 4) 4)(x) = 4)(h) (for some h E H) xE H= (iv) Let H' < G'.7 Consider the natural homomorphism 4): G GIN given by x '—' xN. G has no normal subgroups except {e) and G. Let N be a normal subgroup of G. lf4): G —' G' is any homomorphism. and only if H/N<1G/N.. Now E since H' <IG'.

102

Normal subgroups

The following result is an immediate consequence of Corollary 2.7.

Corollary. Let N be a proper normal subgroup of G. Then N is a maximal normal subgroup of G if and only if GIN is simple.
2.8

Corollary. Let H and K be distinct maximal normal subgroups of G. Then H fl K is a maximal normal subgroup of H and also of K.
2.9

Proof By the diamond isomorphism theorem

H/H fl K HK/K.
Now K < HK

HK=

G. Hence, HK = Kor HK = G, since K is maximal. But Hc K,acontradiction, sinceHandKareboth maximaland
G.

distinct. Hence, HK =

Therefore,

H/H fl K

G/K.

Hence, by Corollary 2.8, H fl Kis a maximal normal subgroup ofHand,

similarly, of K. 0
2.10

Examples
for

(a) Let G be a group such that for some fixed integer n> I, = all a,b E G. Let = (a E = e) and = E G). Then

G,,IG,
Solution. Let a,b

and

and x

G. Then (ab')" =

=

e,

so

E Also, Hence, = = e implies E <1 G. Similarly, G" <1 G. Define a mappingf G —, G" byf(a) = a". Then, for all a,b G,f(ab) = (ab)" = a"b". Thus,fis a homomorphism.

Now Ker f= (ala"
theorem GIG,,
G".

e) =

Therefore, by the first isomorphism

(b) Let G be a finite group, and let Tbe an automorphism of G with the property T(x) = x if and only ifx = e. Then every g E G can be expressed as g = x E G.

Solution. We assert that

x = y.

=

=

theorems

103

by hypothesis. Therefore

G=

G).

(c) If in Example (b) we also have the condition that T2 = I, then G is abelian.
Solution. Let x E G. Then

x'(T(x)) =
=
Therefore, forailgE G, On the other hand,

=

=
ab.

g.Nowifa,bE G,then

=
Therefore, ab = ba.

=

= ba.

(d) A nonabelian group of order 6 is isomorphic to S3.

Solution. If each element is of order 2, then G is abelian. So there must be an element a of order 3. Let b E G be such that b {e,a,a2). Then it is easy to check that e,a ,a2,b,ab,a2b are all distinct elements and, thus, constitute the whole group G. Now b2 a or a2. For let b2 = a. Then b6 = e. This implies that the order of b = 2,3,6. But then the order of b = 2 implies a = e, a contradiction. And the order of b = 3, along with b2 = a, implies ab = e, a contradiction. Also the order of b = 6 implies that G is cyclic, a contradiction to the fact that G is not abelian. Hence, b2 * a. Similarly, b2 * a2. Also b2 = b, ab, or a2b would imply b = e, a, or a2, which is not the case. Therefore the only possibility for b2 E G is that b2 = e. Further, the subgroup [aJ = (e,a,a2) generated by a is of index 2 and, thus, normal. Therefore, bab' = e, a, or a2. But bab' = e gives a = e, which is not

so; and

bab' = a2. Hence, G is generated by a,b with the defining relations
a3=e=b2,
a'3 =
= b'2,
a2.

= a says that G is abelian, which is also not so. Thus,

On the other hand, S3, is also generated by a' and b', where

= a'2.
ab'—'a'b',

Then the mapping

a'—'a',
is an isomorphism of G onto S3.

104

Normal subgroups

Problems

If a is an isomorphism of a group G into a group H, prove that (o(a))" e' if and only if a" = e. Give an example to show this is not true if a is not an isomorphism. 2. If a group G is generated by a subset S, prove that a(S) also generates G, where a is an automorphism of G. 3. Show that a cyclic group of order 8 is homomorphic to a cyclic group of order 4. 4. Write down all the homomorphic images of (i) the Klein four-group, (ii) the octic group. 5. Show that each dihedral group is homomorphic to the group of order 2.
1.

6.

Let [a] be a cyclic group of order m and (bJ a cyclic group of order n. Show that there is a homomorphism a of[a] into [bi such that o(a) = bk if and only if mk is a multiple of n. Further, if mk qn,

show that a is an isomorphism if and only if (m,q) =
3

1.

Automorphisms

Recall that an aulomorphfsm of a group G is an isomorphism of G onto G. The set of all automorphisms of G is denoted by Aut(G). We have seen (Example 2.1(d), Chapter 4) that every g E G determines an automorphism I, of G (called an inner automorphism) given by xi—' gxg'. The set of all inner automorphisms of G is denoted by In(G).

Theorem. The set Aut(G) of all automorphisms of a group G isa group under composition of mappings, and In(G) Aut(G). Moreover,
3.1

G/Z(G)
x,y C G,

In(G).

Proof Clearly, Aut(G) is nonempty. Let a,r C Aut(G). Then for all
ar(xy) = o((rxXry)) = (ar(x)Xat(y)).
Hence, aT C Aut(G). Again,

= aa'(x)a(r'(y) = .xy.
Therefore, E Aut(G). This proves = Hence that Aut(G) is a subgroup of the symmetric group S0 and, hence, is itself a group.

Automorphisms

105

Consider the mapping

4)G—+Aut(G) givenby
For any a,b E G,

=
for all x
Im
4) is

a(bxb ')a' = 1a4(x)

G. Hence, 4) is a homomorphism, and, therefore, ln(G) = a subgroup of Aut(G). Further, is the identity automorphism if and only if = x for all x G. Hence, Ker 4) = Z(G), and by the fundamental theorem of homomorphisms

G/Z(G)

ln(G).

Finally, for any a E Aut(G),

(aJ4a 'Xx) = a(aa '(x)a

hence, a!4tr' =
G

= E In(G). Therefore, ln(G) <1 Aut(G). 0
1)

It follows from the theorem that if the center of a group G is trivial, then In(G). A group G is said to be complete if Z(G) = (e) and every automorphism of G is an inner automorphism; that is, G In(G) Aut(G). When considering the possible automorphisms a of a group G, it is useful to remember that, for any x E G, x and a(x) must be of the same order (see Problem 19, Section 3, Chapter 4).
3.2

Examples
S3.

(a) The symmetric group S3 has a trivial center {e}. Hence, In(S3) Now consider all automorphisms of S3. We have seen that
S3 = (e,a,a2,b,ab,a2b)

with the defining relations

a3—e=b2,

ba=a2b.

The elements a and a2 are of order 3, and b,ab, and a2b are all of order 2.

Hence, for any a Aut(S3), c(a) = a or a2, a(b) = b, ab, or a2b. Moreover, when o(a) and o-(b) are fixed, a(x) is known for every x ES3. Hence,
o is completely determined. Thus, there cannot be more than six automorphisms of S3. Hence, Aut(S3) In(S3) S3. Therefore, S3 is a complete group. (b) Let G be a finite abelian group of order n, and let m be a positive

106

Normal subgroups

integer prime to n. Then the mapping o: x '—b x'" is an automorphism

of G.

Solution. (m,n) = 1 I Yx E G,

3

integers u
= XWPS,

and v such

that mu + nv =

x=
since

=
n.

0(G) =

Now Vx E G, x = (Xu)m implies that a is onto. Further, Xm = e Xmu = e x = e, showing that a is I - I. That a is a homomorphism fol-

lows from the fact that G is abelian. Hence, a is an automorphism of G.

(c) A finite group G having more than two elements and with the condition that x2 # e for some x E G must have a nontrivial automorphism. (Also see Problem 1.)
is an automorphism, and, Solution. When G is abelian, then a: x 'clearly, a is not an identity automorphism. When G is not abelian, there exists a nontrivial inner automorphism.
a '—' atm

(d) Let G = [a] be a finite cyclic group of order n. Then the mapping a: is an automorphism of G if (,n,n) = I.

Solution. lf(m,n) = I, then it has been shown in Example (b) that a is an

automorphism. So let us assume now that a is an automorphism. Then the order of o(a) = am is the same as that of a, which is n. Further, if (m,n) d, then (am)hh/d = (anrid = e. Thus, the order of am divides n/d; that is, nln/d. Hence, d = I, and the solution is complete.
(e) If G is a finite cyclic group of order n, then the order of Aut(G), the

group of automorphisms of G, is 4(n), where 4

is

Euler's function.

Solution. Let G = [aJ and a E Aut(G). Since a(a') = (a(a))' for each integer i, a is completely known if we know o(a). Now let a(a) = am, m n. Then as in Example (d), we have (m,n) = I. Thus, each a determines a unique integer m less than and prime to n, and conversely, which completes the solution.

Problems
1.

2.

Find the group of automorphisms of (Z, +). Find the group of automorphism of (ZN, +).

Conjugacy and G-sets
3.

107

4. 5. 6.

7. 8.

Find Aut(K) where K is the Klein four-group. What about Aut(Z2 x Z2)? Show that the group of automorphism of D4 is of order 8. Show that Aut(Z2 x Z4) consists of eight elements sending (1,0) to (1,0) or (1,2) and (0,1) to (0,1), (0,3), (1, 1), or (1,3). Show that Aut(Z2 x Z3) Aut(Z2) x Aut(Z3). in the following problems 7 and 8, you may invoke Theorem 2.1, Chapter 8. Let A be a noncyclic finite abelian group. Prove that Aut(A) is not abelian. Let 6 be a finite group such that IAut(G)t = p. Prove IGI 3. (Hint: Note G is abelian)
Conjugacy and G-sets

4

Definition. Let G be a group andX a set. Then G is said to act on X there is a mapping written a x, such that for all 6 X X —' X, with

a,bE G, xEX,
(i)

a*(b*x)=(ab)ax,

(ii)
The mapping 4 is called the action of G on X, and Xis said zobe a G-set.
One of the most important examples of G-sets is the action of the group 6 on itself by conjugation defined by

a *x Example 4. l(b)j. The reader interested in conjugation and its applications only may go directly to Theorem 4.8 without breaking continuity. To be more precise, we have just defined an action on the left. Simi(see

X X G —, X with (x,a) '— x

larly, an action of G on X on the right is defined to be a mapping * a such that x * (ab) = (x * a) * b and

e=xfor all a,bEGandxEX.
We shall confine ourselves to groups acting on the left, and for the sake of convenience we sometimes write ax instead of a . x.
4.1

Examples of G-sets
(a)

Let G be the additive group U, and X be the set of complex z such that fri = 1. Then X is a G-set under the action y* where yeR and ceX. Here the action of y is the rotation through an angle 0= y radians,
numbers
anticlockwise.

108

Normal subgroups

(b) Let G = S5 and X = {x1,x2,x3,x4,x5} be a set of beads forming a circular ring. Then X is a G-set under the action
g*xj—xg(j), gES5.
(c) Let G = D4 and X be the vertices 1,2,3,4 of a square. X is a

G-set under the action

g*i=g(i), gED4, IE{l,2,3,4).
(d) Let G be a group. Define

a*x=ax, aEG,xEG.
Then, clearly, the set G is a G-set. This action of the group G on itself is

called translation.
(e)

Let G be a group. Define

aEG,xEG.
We show that G is a G-set. Let a,b E G. Then

(ab) * x =

=a(b*x)irl =a*(b*x).

=

Also, e * x = .x. This proves G is a G-set. This action of the group G on itself is called conjugation.

(f) Let G be a group acting on a set X. The action of G on X can be as follows. Define extended to the power set
a * S=(a
for
e* S

S)

a * (1' * 5) and a,b E G and Sc X. Therefore is a G-set. (g) Let G be a group and H < G. Then the set G/H of left cosets can be made into a G-set by defining a E G, xH E G/H. a * xH = ax/f,

all a E G and Sc X. Then, clearly, (ab) * S =
=
S for

(h) Let G be a group and H <1 G. Then the set G/H of left cosets is a G-set if we define
a * xH =

a E G, xH E Gill.
a * (b
*

To see this, let a,b E G and xli E Gil-I. Then

(ab) • xH= Also, e * xH = xli. Hence, G/H is a G-set.

xH).

If X is a G-set, then, as stated earlier, we generally write ax instead of a * x for the sake of simplicity.

Conjugacy and G-sets
4.2

109

Theorem. Let G be a group and let X be a set.

(i) If X is a G-set, then the action of G on X induces a homomorphism
(ii) Any homomorphism 4G —,
induces an action of G onto X.

Proof. (I) We define 4: G by (4(a))(x) = ax, aeG, xEX. Clearly çb(a)€S5, aeG. Let a, bEG. Then
(rb(ab)Xx)

(ab)x = a(bx) = a((cb(b)Xx)) = (4)(a)X(4)(b)Xx)) for all x E X. = (4)(a)4)(b)Xx)

Hence, 4)(ab) =

çb(a)çb(b).

(ii) Define a*x = (4i(a))(x); that is, ax = (#(a))(x). Then
(ab)x = (4)(ab)Xx) = (4)(a)4)(b)Xx)

4,(aX4i(bXx))

4)(aXbx) = a(bx).

Also, ex = (4)(e)Xx) = x. Hence, Xis a G-set.

0

Regarding G as a G-set as in Example 4.1(d), we obtain Cayley's

theorem (see Theorem 5.1, Chapter 4).

43

Cayley's theorem. Let G be a group. Then G is isomorphic into the symmetric group SG.

Proof By Theorem 4.2 there exists a homomorphism 4): G where (4)(a)Xx) = ax, a E G, x E G. Suppose 4)(a) the identity element in SG. Then for all x E G, (4)(a)Xx) = x. This implies ax = x for all x E X, and hence a = e, the identity in G. Therefore, 4) is injective. 0
Remark. An isomorphism of G into a group of permutations is called a faithful representation of G by a group ofpermutaiions.

Further, if H < G the set Gill of left cosets of H in G is a G-set [Example 4.1(g)]. The action of G on 6/H gives us another representation of 6 by a group of permutations, which is not necessarily faithful. This is contained in

Theorem. Let G be a group and H < G of finite index n. Then 4.4 such that Ker 4) = flX€GXHX there is a homomorphism 4): G

110

Normal subgroups

Proof Since 16/HI = n, SG,,, morphism 4): G such that

By Theorem 4.2 there exists a homo= axH. Now

Ker 4) = (a E Gf4)(a) = identity permutation)

= (a E GIaxH

xl!, Vx E G)

=(aEGIx'axEH, YxE G)

=(aE
XE 0

=flxllx-'.

YxE 6)

0

Remark. We obtain the Cayley representation of G by taking H = (e).

Corollary. Let G be a group with a 4.5 Then GIN is isomorphic into

subgroup H of index n.

Proof Follows from Theorem 4.4, since H <16 implies Ker 4) = 1!. 0
Corollary. Let G be a simple group with a subgroup 4.6 index n. Then G is isomorphic into

G offinite

Proof Follows from Theorem 4.4 by recalling that Ker 4) <1 G. 0
Definition. Let G be a group acting on a set X, and let x E I. Then the set

G(gx=x),
which can be easily shown to be a subgroup, is called the stabilizer (or isotropy) group of x in G.
For example, if G acts on itself by conjugation [Example 4.1(e)], then, for x E G,

=x)=N(x),
the normalizer of x in G. Thus, in this case the stabilizer of any element x in G is the normalizer of x in G. Another example is the 6-set 6/H, where H < G [Example 4.1(g)]. Here the stabilizer of a left coset xH is the subgroup

(gE

) = (gE GJx'gxE H)

={gE

Conjugacy and G-sets

Ill

Definition. Let G be a group acting on a set X, and let x E X. Then the set

Gx={axlaE G)
is called the orbit of x in G.

In case G acts on itself as in Example 4.1(d), the orbit ofxeG is Gx = {axIaEG} = G. Further, considering G as a G-set as in Example 4.1(e), where the

action of the group G is by conjugation, the orbit of x E G is Gx =
E G), called the conjugate class of x and denoted by C(x). We now prove a very important result.
4.7 Theorem. Let G be a group acting on a set X. Then the set of all orbits in X under G is a partition of X. For any x X there is a bijection Gx G/GX and, hence,

lGxl=
Therefore,

is a finite set,
XE C

111=
where C is a subset of X containing exactly one element from each orbit.

Proof For any x,y X, let x — y mean that x = ay for some a E G. Now ex for all x E X; and ifx = ay, then y = a'x. Ifx = ar andy = bz, then x = (ab)z. Hence, — is an equivalence relation on X, and the equiva-

x=

lence class of x Xis the orbit Gx. Hence, the set of all orbits is a partition of X. Therefore,

X=
where

Gx
XEC

(disjoint),

C is any subset of X containing exactly one element from each

orbit. Given x C X, consider the mapping
givenby

forallacG.
bG,r.

For any a,b E G,
Hence, is well defined and injective. Moreover, and, therefore, bijective. Hence, IGxI =
is obviously surjective

we give an independent proof of Theorem 4. then IXI= xEC xEC [G:G. Then the following are true: (i) The set of conjugate classes of G is a partition of G. which is called the conjugate class of a in G. 0 The partition of G given by the conjugate classes in Theorem 4. which is a subgroup of G. — is an equivalence relation on G.r1 = Hence. (iii) Follows from (I) and (ii). This proves G = UC(a). The equivalence class ofa in G is then the set e G}.8 (i) is . For the reader interested in conjugation only.112 Normal subgroups Since X is the disjoint union of orbits Gx.7 for the case when G acts on itself by conjugation. Also. To see this. Clearly. Then the mapping o: C(a) G/N(a) given by xN(a) is trivially onto. (ii) Let a G. (ii) IC(a)I = [G:N(a)J. let x. Then ay_tx=xa.8 Theorem. it follows that if X is finite.y C G. In Theorem 4. Proof (i) Define a relation — on G as follows: a—b if forsomex€G. a disjoint union of conjugate classes. a running over exactly one element from each conjugate class. = EfG:N(a)J. Let a E G. Let G be a group. (iii) If G is finite. it is I-I. 4.8 we give a version of Theorem 4.8.]. which is also the conjugate class C(a) of a. Then we define qa) = E G). a is a bijection. 0 The partition of X given by the orbits in X under G is called the orbit decomposition of X under G. Recall that the normalizer of a in G is the set =a). proving IC(a)I — IG/N(a)t.

then the set = is E S) called a conjugate of S. H is a subgroup of order then H 1 G.10 (1) Theorem. hence. XEC (I) . then H is properly contained in N(H). [G:N(x)]. Then T is said to be conjugate to S if there exists x E G such that T = Clearly. the class equation can be expressed as IGI'=IZ(G)I+ XE C where C contains exactly one element from each conjugate class with more than one element. Definition. Let G be a group. where p is prime and n>O. we Theorem. Therefore Gisthedisjoint union ofZ(G)andallconjugateclasseshaving more than one element. "being conjugate to" is an equivalence relation in the power set of the set G.9 where G acts on by conjugation.8). Then (ii) Z fl N is nont rivial for any nontrivial normal subgroup N of G. Then for any subset S of G. Proof (1) Consider the class equation of G.8 or by specializing Theorem 4. and let Sbe a subset of G. Let us first observe that xE Z(G) if and only if the conjugate class C(x) has just one element. We now give some important applications of the class equation of a finite group G(Theorem 4. Following the method of proof of Theorem 4. Let Sand T be two subsets of a group G. Let G be a finite group order G has a nontrivial center Z. Let G be a group. Hence. namely x itself.7 to the G-set can easily prove 4. If x E G.Conjugacy and G-sets 113 called the class decomposition of G. The equation in part (iii) is called the class equation (or class equation formula) of G. IC(S)I = [G:N(S)J [N(S) = (x E = S)]. (iii) If H is a proper subgroup of G. 4.

Since LIX '1 G/K. = (XE XIgx = x). since K is a maximal normal subgroup of G contained in H. and C is a subset of G having exactly one element from each conjugate class not contained in Z. Hence.11 Corollary. and therefore C(x) fl is either zero or equal to [G:N(x)J. Hence. (2) If x E N. 4. Because L/K is the center of G/K. Hence. / E L. therefore. a E Z. (iii) Let Kbe a maximal normal subgroup of G contained in H. then C(x) C N. and if x Z. (hKX/K) = (IK)(hK). for every x C. 0 We close this section by proving Burnside's theorem. This proves that Z is not trivial. which has applications in combinatorics. p divides Because p divides INI. p divides now from the class equation (I) that p divides ZI. L C N(H). (ii) Because G Z U C(x)) (disjoint). and K ç L.12 Theorem (Burnside). By Theorem 4. p divides It follows IG:N(x)] for every x Z.10. it follows from the equation (2) that p divides IZ fl This proves that Z fl N is not trivial. IN(x)I < ph. N(a) must be of order p2. Hence. Let G be afinite group acting on afinize set . Every group of order p2 (p prime) is abe/ian. let X. Hence. For a fixed g E G. Therefore. say L/K. then N(H) must be Hence.114 Normal subgroups where Z = Z(G). it follows by Theorem 2. Then the quotient group G/K is of order p' for some r> 0. This implies that H # N(H). Proof Suppose G is a nonabelian group of order p2. Then Z is a proper subset of N(a). 0 4. This implies N(a) = G. 1 'hlehK C H. a Z. Now L is not contained in I!. By Lagrange's theorem IN(x)I divides pfl for every x E G. by the first part of the theorem.6 that L G. Hence. If H is a subgroup N(H) = G and. Therefore. a contradiction. G/K has a nontrivial center. H . Let h E H. C(x) fl N is either empty or equal to C(x). Let a E G. it follows that His properly contained in N(H). Because H C N(1f). N = G fl Hence.c3 G. Hence. N (Zfl u (u C(x) fl N) XE C (disjoint). G has a nontrivial center Z that must be of order p.

But. and o(b) = m.x) in S. We claim that a2 = e. Let A. Define A to be conjugate to B with respect to H if B = hE H. If a2 e. so for a2' = a. Let mln.7. Prove that IGI = Solution. a contradiction. Because = e.B E 9'(G). IGXI is the number of ordered pairs (g. 0 4. then a2 E C(a). because G has only two conjugate classes. Therefore. for any fixed x E X. which proves the theorem. then b E C(a). xeX aeCxEGa where C is a subset of X containing exactly one element from each orbit. these must be (e) and C(a).x) E G X XIgx =x). g2 = e. Then .Conjegacy and G-sets 115 X. a2 = e. Hence. gEG XEX By Theorem 4. Hence. the power set of G. Then there exists an element b such that b E f a]. IGI = 2. Let e a E G such that o(a) n. It then follows that m = n.x) inS is exactly lXgI. Therefore. if e b E G. Thus o(b) = 0(a) = n. a2 = x for i = n. n is prime. the cyclic group generated by a. (b) Let H be a subgroup of a finite group G. Hence. IGxI= [G:Gj = xEX Hence. For any fixedg E Gthe number of ordered pairs (g. Then the number k of orbits in X under G is Proof LetS = ((g. Because there are only two conjugate classes. Now IGxI = IGal for every x E Ga. IGXI = k. This implies G is abelian. Hence.13 Examples 2. Therefore. This yields all g G. Similarly. we obtain 1 (mod n). (a) Let G be a group containing an element of finite order n> I and exactly two conjugacy classes.

Let G = [aJ be the cyclic group of order p. where the beads can have any of n different colors. First. Then the orbit of and the stabilizer is H fl N(A). Let X be the set of all possible necklaces.7 by taking X to be the power set of G and H to be the group that acts on Xby conjugation. a is bijective. Then E H fl N(A) h1(Hn N(A)) = h2(Hn N(A)) : hT'h2A = Ahj1h2 = h1A/zj' = Hence. (c) Find the number of different necklaces with p beads.116 Normal (i) (ii) Conjugacy defined in is an equivalence relation. Solution. For the proof of (ii). then (CH(A)I = [H: fin N(A)1. (called the conjuIf CH(A) is the equivalence class of A E gate class of A with respect to H). for let h1 . consider the mapping a: where a a is also 1-1. A E X is the conjugate class The direct argument is as follows.h2 E H. p prime. Clearly IXI = Let be a necklace. Follows from Theorem 4. Solution. The example which follows gives a beautiful illustration of the Burnside Theorem. Then X is a G-set under the action . (i) is obvious.

for all geG. Hence IXgI = n. Let H be a proper subgroup of a finite group G. Xg consists of all those necklaces which are unchanged by any permutation and these are precisely those which are of one color. 2. Find the number of conjugates of the element (1 3) in D4. I timcs) I)]. Determine the conjugate classes of the symmetric group of degree 3 and verify that the number of elements in each conjugate class is a divisor of the order of the group. We now compute Xg for each gEG. Since G is cyclic of prime order. Let G be a group. x€G. VaeG}.) So the number of orbits is the same as the number of different necklaces. The Burnside Theorem then gives the number of different necklaces as (p -. Xg = = = =x. Therefore. (The beads are just being permuted cyclically.Conjugacy . Show that Z(G) = UC(x).nd G-sets 117 = where subscripts are modulo p. Then Let g # e. g generates = x} = x. [Hint: IC(H)I = [G:N(H)] and 1x' HxI =IHLI . = n". Problems 1. 3. So G. 4. g e. First Xe = {XEXIe*X = x} = X. Clearly the action of any fixed element a given necklace yields the same necklace. Show that G contains at least one element that is not in any conjugate off!.

and let the index of H in G be finite. (iii) Set of faces. Prove that each of the following sets is a G-set: 8. Show that every element of order dividing INI is contained in N. If every two maximal subgroups of G are conjugate in G. 2 / 1—3 2 2 1 /\ \ /\ 2 2 3 . Let H < G. (i) Set of vertices. 6. Show G acts transitively on X if and only if there is only one orbit. 1—3. prove that each element in G has at most m conjugates. Prove that in any group the subset of all elements that have only a finite number of conjugates is a subgroup. Let G be a group in which each proper subgroup is contained in a maximal subgroup of finite index in G. then G acts on X transitively.2.IG/NI) = 1. 9. (ii) Set of long diagonals.] Let G be a group and X be a G-set.118 Normal subgroups 5. Let G be the group of symmetries of a cube. prove that G is cyclic. Let G be a finite group with a normal subgroup N such that (INI. If the commutator group G' ofa group G is of order m. (iv) Set of edges. Also show if H < G and X = G/H. Use the Burnside Theorem to 14.3 consisting of 2. Prove that H contains a normal subgroup N that is of finite index. Let G = S3 and X be the undirected graph with vertices 1. I—3 . the set of left cosets of H. 13.3 . 1 2 3 . Prove that if a finite group G has a normal subgroup N with INI = 3 such that N Z(G). 7. x2EX there exists geG such that g*x1 = x2. 1—3 3• Show G acts on X canonically. show that the number of orbits is four. 2 . 11. 10. We say G acts transitively on X if for all x1. then G has a subgroup K such that [G:KJ = 12. [Hint: G acts by conjugation on N. Let X = be the set consisting of diagonals of a square .

a= and Regard and X x X as G-sets in a natural way (as ordinary matrix multi- plication).S3 with = {I.3. Find the orbit \OJJ x x. Regard X as a D4-set under the action a diaa a + 2) = {g(a). Regarding each of the sets X and X x X as a K-set in a canonical way.g*x2). and let X = (1.x2)=(g*x1.3.2. (1 3). . and 4.3.Coaijugacy and G-sets 119 15. (1 3). Find the number of necklaces with six beads and two colors. Consider the Klein fourgroup K as the group of symmetries of a rectangle with vertices 1. 17. 18. Show that there exists a homomorphism G—. where gonal with vertices a. and let G = 1/ab\ beZ1. a + 2. + kt(" Show that there are neckties having n strips (of equal width) of k distinct colors.2. and 4.4. 19.3.2)eX x X.g(a + 2)).4). Consider additive and multiplicative groups (Z7.2. find the orbits of IEX and (1.5. Show that X x X is also a G-set under the action g*(x1. = {O. 16. (2 4).2.6). Suppose X is a G-set. with vertices 1. 1. Show that there are six different necklaces with four beads and two colors and eight different ones with five beads and two colors. (2 4)). +) and For convenience write Z.

G1. . Proof If IGI 1 or if G is simple. {e} = G0C G1 = G is trivially a normal series of G.. then (e) C G is the only composition series of G. Every finite group has a composition series..Gr) without repetition whosefactors are all simple groups.1 Lemma. Thefactors G1/G1_1 are called composition factors of G. G. . By the induction hypothesis H has a composition 120 . ../G.. then (G) isa composition series of G without factors). Suppose G is not simple. and the result holds for all groups of order < Let H be a maximal normal subgroup of G. C is =G a normal series of G.G1 The factors of a normal series are the quotient groups G. For any group G. A sequence (G0. Definition._I... then the result is trivial (ifiGi I. 1 i r.. A composition series of a group G is a normal series (G0. We often refer to a normal series (G0. 1. If G is a simple group.CHAPTER 6 Normal series I Normal series Gr) of subgroups of a group G is called a normal series (or subnormal series) of G if Definition.) by saying that {e} = G0 C G1C .

i') C G is a composition series for the quaternion group G.l. written S S'. G has a composition series Examples (a) (e) C (e..t) C (e. (d) (0) C (0.. By Corollary 2.. ifthefaciors of one series are isomorphic to the factors of the other after some permutation..1 = H5_1._1 #H5_1. S2bytheinductionhypothesis._1 () H3_1.r.2 C H..thenS1 let K = G. Therefore.i7) = Z/(18) isa composition for Z/( 18).12.. Kis a maximal normal ..3 Theorem (Jordan — Holder)." in the following sense.15) C (0.9) C (0.=G.G1 Gr) and S' = G are said to be equivalent. (c) (e) C (e. We remark that a composition series is not necessarily unique.e'.i3) C 1)4 is a composition series for the octic group 1)4 (see Section 5.r.a2. Definition. Chapter 4).9. another composition series for the octic group 1)4 is (e) C (e. 1.. For example.(l 2 3). I = 1 . say S1: S2: (e)=G0cG1c•"cG. 0 1..a2r) C However. (1 3 2)) C S3 is a composition series for S3. Any two composition series ofafinite group are equivalent.c2. Evidently. Chapter 5.i.. If G.2. Two normal series S = (G0.lfG.c. — is an equivalence relation..3. Proof Let G be a finite group and suppose the theorem holds for every group of order < IGI.a2) C (e.9. for some aE S.. it is shown in the next section that any two composition series of the same group are "equivalent.6. Consider any two composition series of G.e') C (e. that is. where e' = —e and 01 0 (b) (e) C (e.Normal serIes 121 series {e) C H1 C (e)CH1C"CHCG.

(e)=G0CG1 C G. I = . = G.4 Examples (a) An abelian group G has a composition series if and only if G is finite. (b) If a cyclic group has exactly one composition series. . . 53 — S4./G1_11= Pi Pr. then it is a p-group.thatis.and also of Now K has a composition series. Further. Thus. (e) K0 C K1 c namely. must be cyclic groups of prime order p (say). the composition factors — are determined uniquely up to isomorphism and ordering. Therefore. This proves that S1 and S2 are equivalent.. 113_1/K Now and G/1131 G.._1/K. By Theorem 4. by the induction hypothesis. i = I permutation of the prime factors of IGI determines a unique composition Pi series of G. Hence. Then the quotient groups G. every composition series of G such that 1G1/G1_1 I = p. 1 IGI = fllG..—G is a unique r.1CG..JGj=rpr. S1 S2 —S4. by the diamond isomorphism theorem. Hence.r. G has a unique composition series if and only if .=p. Hence. Therefore. being abelian and simple.122 Nonnal series subgroup of G. ... 0 S3 and Theorem 1. G1 = [aP'*' Pj. 1.wherep. p1. say C K.1. Solution. Moreover the composition factors of G are determined by the prime factors of JGJ. 1=1 r— I.4 in Chapter 4. Hence. = G.. is finite. namely. .3 shows that the factors of any composition series ofa finite group G — that is. SupposeG = [a)isacyclicgroupoforderp1 "p. G has a unique subgroup of order p1 ../G. Solution. Pr are primes (not necessarily distinct). 53: S4: {e)=K0CK1C"CKCG. an abelian group G has a composition series ifand only if (. This gives two more composition series of G.. Let (e) G be an abelian group with a composition series = G0 C G1 C C G.

any composition factor is of order pk for some k> 0.2. But since a group of prime power order has a nontrivial center.p.. . 5. 3. Write down a composition series for the Klein four-group. Write down all the composition series for the octic group. Let us define a composition series of any group G to be a descending chain (possibly infinite) G0=GJG1JG2J..Normal serfes 123 (c) The Jordan — HOlder theorem implies the fundamental theorem of arithmetic. Solution. Then G has a composition series such that all its composition factors are of order p...J{e) such that Show that Z has is simple. it follows that each composition factor is a simple abelian group and. Problems Write down all the composition series for the quatern ion group. Show that G has a composiI. and also a comq.. 2. By the Jordan — HOlder theorem the two series must be equivalent.. show that the number of distinct composition series of G is r!... p prime. Solution. and n= n = p1 p. therefore. Verify that they are equivalent. Find all composition series for Z/(30). (after reordering if necessary). distinct primes. p. r = s and p1 = q. 7. 4. as shown in Example (b). If G is a cyclic group such that IGI = . 6. Then. and no finite composition series but does possess . Let G be a cyclic group of order n. H which have isomorphic composition factors.. Let G be a finite group and N <1 G. say q3. 8. Give an example of two nonisomorphic finite groups G. Clearly. respecposition series whose factors are cyclic groups of orders q1 tively. i= 1. tion series in which N appears as a term. respectively. 9. and each composition factor is a simple group.3. Suppose n has two factonzations into positive primes. (d) Let G be a group of order pfl.. Thus. of order p.. G has a composition series whose factors are cyclic groups of orders p.• p.. Find the composition factors of the additive group of integers modulo 8.

0 C N. written as follows: = G'. Then N(k) = and (G/N)(') = (e) for some positive integers k. then every subgroup of G and every homomorphic image of G are solvable. Thus. C H G H be a surjective homomorphism. we define the nih derived group of G. fl(n) = Hence. Hence. if N is a normal subgroup of G such that NandG/N are solvable. A group G is said to be solvable integer k. Because GIN is a homomorphic image of G (under the natural homomorphism for every positive integer n. then G (e) = G(k) C . group H of G. This proves that every homo. Hence. Jf(k) 4((e)) = (e). (e) 2. then G' = (e). morphic image of G is solvable. G is solvable. then G is solvable. by induction. Further.2 Theorem. so that for every positive integer n. Definition. conversely. Conversely.124 Normal series infinite composition series.l. 2. every abelian group is solvable. trivially. For any positive integer n. Theorem. Therefore. a finite group is solvable if and only if its composition factors are cyclic groups of prime orders. Write two such series and show that the Jordan—Holder Theorem does not generalize to infinite groups. = (n > 1).b . Let G be a group. Therefore. let N '1 G such that Nand G/N are solvable. A group G is solvable jfand only if G has a norma/series with abe/ian factors. = (e)for some positive It is obvious that if G is abelian. = for any positive and.1 (e). 2 Solvable groups Recall that the subgroup G' generated by the set of all commutators aba'b' in a group G is called the derived group of G. Then for any subProof Let G be a solvable group. = G —' GIN). Proof If {e). H' = integer n. For any a. JIG is solvable.

. i = . Therefore. for 23). Now C . being a finite abelian group.CER.. has a composition series whose factors are cyclic of prime order.. is solvable.a'"') C is solvable... Problems 1.. for some prime p. between the terms get a composition series of G in which every factor is a cyclic group of prime order.. for is a normal series with abelian factors. Generalize this to n x n upper triangular matrices.. Then M is cyclic of order p. G" = (e). a. B are groups. G in which each factor (e} H.C"CH. Inserting the corresponding subgroups of C H. and G is solvable...—G is abelian./H1_I. Show that the group of all upper triangular matrices of the form lab i o 001 2.. .Solvable groups 125 is a normal series of G in which each factor a normal series G (e)=H0CH. but solvable. We may first show that M is an abelian p-group. Show that a simple group is solvable if and only if it is cyclic. G' C I . C H. c . and. by induction.r..b. Then G has a normal series is abelian. Hence. B are solvable. is not (b) The dihedral group (e) C (e.1CH. We shall see in the next chapter that S4 is solvable.. 3. The converse follows from the first part of the theorem.in the normal series C H.a. we and H. n 5.(1 32))CS3 is a normal series such that SilN is abelian.4 of Chapter C !4_.. 0 such that 1 5.. Show that if A. H C 2.. then A x B is solvable if and only if both A. Then by Theorem 1.3 Examples (a) S3 is solvable. I Now suppose G is a finite solvable group.r. (c) Let G be a finite solvable group and M be a minimal normal subgroup of G.

126

Normal series
4.

Let A, B be normal solvable subgroups of a group G. Show that AB is also solvable.

Additional problems on solvable groups are given in the problem set at the end of Chapter 8.
3

Nilpotent groups

We define inductively the nih center of a group Gas follows. For n I, Z1(G) Z(G). Consider the center of the quotient group G/Z1(G). Beis a normal subgroup of G/Z(G), by Corollary 2.7 of cause Chapter 5 there is a unique normal subgroup Z2(G) of G such that Z2(G )/ Z1(G) Z(G/Z1(G)). Thus, inductively we obtain a normal subgroup = for every positive Z,,(G) of G such that integer n> 1. is called the nih center of G. Setting 4(G) = (e), we 1(G)) for all positive integers n. It follows have 1(G) = immediately from the definition that (XE
Hence,
C

E

for all yE G).

The ascending series

of subgroups of a group G is called the upper central series of G.
Definition. A group G is said to be nilpotent if Z,,,(G) = GJor some iii. The smallest m such that Zm(G) = G is ca/led the class of nilpotency of G. If G is an abelian group, then Z1(G) = Z(G) = G. Thus, trivially, every abelian group is nilpotent. For a nontrivial example we prove
3.1

Theorem. A group of order p" (p prime) is nilpotent.

Proof By Theorem 4.10 of Chapter 5, G has a nontrivial center Z,(G). Now G/ZI(G), being of order Pr <p1. has a nontrivial center. Hence,
I Z1(G)I <1Z2(G)I. By repeating the argument we get IZm(G)t = p1 for some n. Hence, Zm(G) = G. 0

m

3.2
series

Theorem. A group G is ni/potent
(e)

and only if G has a normal

= G0 C G1C •"C
C Z(G/G1...1)for

G

such that

al/i =

I

m.

Nilpotent groups

127

Proof Let G be a nilpotent group of class m, so that Zm(G) =

(1.

Then

(e)=Z0(G)CZ1(G)C"C Zm(G)=G
is a normal seties that satisfies the required condition, since Z.(G)/
(e} =
Z1_ 1(G) = G0 C G1 C

1(G)). Conversely, suppose that G has a normal series C Gm = G such that G/G11 C Z(G/GI_L) for all i.

Then G1 C Z(G). Now GilGI C Z(G/G1). Hence, for every x E G2, yE G, C G1 C Z1(G). Therefore,G2 C Z2(G). By repeating the ari = l,...,m. Hence, G = Gm C Zm(G), and G is gument we get G1 C therefore nilpotent. 0
3.3

Corollary. Every ni/potent group is solvable.

Proof Let G be a nilpotent group. By Theorem 3.2, G has a normal series

with abelian factors. Hence, by Theorem 2.2, G is solvable. 0
The converse of this corollary is not true. For example, the symmetric group S3 is not nilpotent because its center is trivial. But it is easily verified

that is trivial. Hence, S3 is solvable. Thus, we see that the class of nilpotent groups is contained strictly between the classes ofabelian groups and solvable groups.
3.4

Theorem. Let G be a ni/potent group. Then even' subgroup of G

and every homomorphic image of G are ni/potent.
Proof Let G be a nilpotent group of class m, so that Zm(G) = G. Let Jibe a

subgroup of G. Then, clearly, H fl Z(G) C Z(H). For all x C Z2(G), E Z1(G). Hence, for all x C H fl Z2(G), and y C it, andy C G, xp. xx 1y' C H fl Z1(G). Therefore, H fl Z2(G) C Z2(H). By repeating the argument we see that H fl Z.(G) C Z,(H), i I m. Hence, H = H fl G H fl Zm(G) C Zm(Ji). Thus. ii is nilpotent. Let G —. H be a surjective homomorphism. Then

=
Hence, y C G. Hence
C Z(H). Let XE Z2(G). Then

for all x,i' E G.

C Z(G) for all

C 4(Z(G)) C Z(H).

Because d is surjective, it follows that

Z2(H). Therefore,

C Z2(H). By repeated use of the argument we see that /(Z1(G))c Z,(H), i= 1 in. Zm(H). Therefore, H is nilpotent. 0

128

Nonnal series

3.5
H1

X"

Theorem. Let H1
X

.

be a family of ni/potent groups. Then

is also nilpotent.

Proof We show that ifH and Kare nilpotent groups, then II X Kis also nilpotent. The proof of the theorem then follows by induction. it can be easily shown that Zm(H X K) Zm(H) X Zm(K) for any m. Hence, Zm(H>( K) = H X K for some m. Therefore, H X K is nilpotent. 0
Problems
1.

2. 3.

Show that is not nilpotent for n 3. nilpotent? solvable? Is Give an example of a group G such that G has a normal subgroup H with both H and G/H nilpotent but G nonnilpotent.

4.

Let a group G be a product of normal subgroups H and K of
orders pA and qP, respectively, where p and q are distinct primes.

Show that G is nilpotent.
5.

Let G be a finite nilpotent group, and let p be prime. Show that ifp divides then p also divides JZ(G)J.

CHAPTER 7
Permutation groups

1

Cyclic decomposition

Let us recall that a cycle a = (a1

a,) in

is a permutation such that

inn. Two cycles (a1 ..a,),(b1 b3) in are disjoint permutations if and only if the sets {a1,. . . , a,) and {b1,. . , bj are disjoint. Note that a cycle
.

of length r can be written in r ways, namely, as (a1 a,) and

i=2,...,r.

A cycle of length r is also called an

r-cycle.
1.1

Theorem. Any permutation a E

is a product of pairwise dis-

joint cycles. This cyclic factorization is unique except for the order in which the cycles are written and the inclusion or omission of cycles of length 1.

Proof We prove the theorem by induction on n. If n = the theorem is obvious. Let i1 E (I ,2,...,n). Then there exists a smallest positive integer r such that a'(i1) = i1. Let i2 = c(11), i3 = a(i2) = a2(i1),...,i, = a(i,_1) Next, let
1

I = (1 ,2,...,n) — (i1
IfX = 0, then oisacycleand wearedone. SoletX
c5c1, where c1 = (i1
cycles.
j2 . . . i,),

0,andletos = oil.

Then is a permutation of a set consisting of n — r elements. By inductiOn 0* = C2C3 ... Cm, where are pairwise disjoint cycles. But since a

it follows that a is a product of disjoint
.

To prove uniqueness, suppose a has two decompositions into disjoint

cycles of length> I, say a =
129

fi,.

130

PermutatIon groups

Let x E n. If x does not appear in any then a(x) = x. Hence, x does not appear in any f3,. Ifx is in some y1, then a(x) x. Hence, x must be in some fi,. But then a'(x) occurs in both y, and for every r E Z. Hence, y, and are identical cycles. This proves that the two decompositions written earlier are identical, except for the ordering of factors. 0

Corollary. Every permutation can be expressed as a product of 1.2 transpositions.
Proof We need only show that every cycle is a product of transpositions. It is easily verified that

(aia2...am)=(aiam)(ajam_i)...(aja2).
factors need not be pairwise disjoints. For example,
(1

D

Note that decomposition into transpositions is not unique, and the
2
3

4

S)=(4 =(5

=(5 2X4 2X3
3X2

5X3

5X2

5X1 2X1

lX3

5)(4

5) 5) 5X2

3X3

5)

Let a E S,. Let us write the decomposition of a into disjoint cycles, including cycles of length 1, and write the factors in order of increasing the cycles ,...,ykare of lengths length. Supposea= y1 -" nkandnl + n2 + + nk = n. Thus,a respectively. Then n1 n2 determines a partition (n ,. ..,flk) of n, which is called the cycle structure of a. Conversely, given a partition (n1 ,...,flk) of n, there always exists (not

uniquely) a permutation a E Sn such that the cycle structure of a is for the product of any disjoint cycles y, ,...,Yk of length n1 ,...,nk,
respectively, satisfies that condition.
1.3

Theorem.

if a,c E
a

then r =

the symbols in a. Hence, any Iwo conjugate permutations in have the same cycle structure. C'onversely, any two permutations in with the same cycle structure are conjugate.

= ao(i) a(j). Therefore, if Proof If a(i) =j, then za(i) is a cycle in the decomposition of o, then is a (a1
cycle in the decomposition of r. Hence, the cycle decomposition of t is obtained by substituting a(x) forx everywhere in the decomposition of a. Thus, a and have the same cycle structure.

Cyclic decomposition

131

Conversely, suppose that a and r have the same cycle structure
(p,q,...,r). Then a and t
have cycle decompositions

Define a E by = I = l,...,n. Then it is easily verified that = t(b,), I = l,...,n. Hence, a and r are conjugate.

0

1.4 Corollary. There is a one-to-one correspondence between the set of conjugate classes of S, and the set of partitions of n.

Proof Obvious. 0
Examples

1.5

(a) Express the following permutations as a product of disjoint cycles:
k2 5 (ii) (4 2
(i)

/1

2
1

3
I

4
6

5

6
3

4

(in)

Ii

2

3
1

5X3 4 4 5
2
5

2

6X5

6

7

I)

4

Solution. (1) We start with I and obtain the cycle (I 2 5 4 6 3). (ii) The given permutation can be rewritten as

°\4 6

_(i

2

3 2

4
1

5

6
7

3

7 5

Starting with 1, we obtain the cycle (I 4). Then start with any other
element not in (1 4), say 2. Starting with 2, we get a cycle (2 6 7 5 3). These

two cycles exhaust the list (1,2,3,4,5,6,7). Thus, a = (26 7 5 3)0 4). (iii) Proceeding as before, we obtain that the given permutation is (I 3X2 4X5) by the convention of omitting cycles of length 1, equal to (1 3X24).

(b)Let a=(l 5 3)0 2), r =(l 6 79). Compute
Solution.

By Theorem 1.3, ata' = (a(l) a(6) a(7) a(9)) = (26 7 9). The interested reader may actually verify this by first computing a'.

132

Permutation groups

°'k2 5

fi

2
5

3
1

4 4
1

234 5)k3
OW
._,

5\ 3)' 4 3\

fi
6 7

2
1

3

4

5

Thus, by composing the mappings we obtain

fi

2

3
3

4

5

7

8

9\

6

4

5

9

8

2)=(2 6

7

9).

Problems
I. 2. 3.

4.

If a and fi are disjoint cycles, then a/i = fla. If a is a cycle of length r, then a' is the identity permutation. Show that the order ofany c E is the least common riiultiple of the lengths of its disjoint cycles. Any two cycles in Sn are conjugate if and only if they are of the same length.
Alternating group

2

We showed in Section I that every permutation in be expressed as a product of transpositions, and the number of transpositions in any given product is not necessarily unique. However, we have the following important result showing that the number of factors is always even or always odd.
2.1

Theorem. If a permutation c E Sn isa product of r transpositions

and also a product of s transpositions, ihen rand s are either both even or
both odd.

Proof Let C' Let P = 111<1(x1

where

and

are

transpositions.

Xj) be a polynomial in the variablesx1 ,...,x,,. Define
— X,1(f)),

—P. Let is a transposition, then where E We show that = (k 1), k < 1. Now one of the factors in the polynomial P1SXk — Xg, and

the corresponding factor becomes x, — xk. Any factor ofF of the form x, — x1, where neither i non is equal to kor I, is unaltered under the mapping All other factors can be paired to form products of the form — XkXX, — x,), with the sign determined by the
in

relative magnitudes of i, k, and I. But since the effect of

is just to

Alternating group A,

133

interchange Xk and x,, any such product of pairs is unaltered. Therefore, the above argument gives that the only effect of P.

This proves our assertion. Finally,

Also

Thus (— 1)' =

(—

1)', proving that r and s are both even or both odd. 0
is

Definition. A permutation in is called an even (odd) permutation a product of an even (odd) number of transpositions.

The sign of a permutation a, written sgn(a) or e(a), is defined to be + I or — I, according to whether a is even or odd. More generally, we define the sign of any mapping from n to n as follows. Definition. Let
: n —' n. Then

1+ 1

j

—I

0

if 4) is an even permutation, if 4) is an odd permutation, if 4) is not a permutation.
be mappings from n to n. Then

2.2

Lemma. Let

=
Hence for any a E S,,,

= e(a).

Proof If 4) and ware both permutations, the result foLlows from Theorem 2.1. If 4) or w is not a permutation, then is not a permutation. Hence,

= 0=
Hence,

If a is a permutation,

= €(a). 0

=

1.

Let A, denote the set of all even permutations in S,,.

2.3 Theorem. A,, is a normal subgroup of S,,. Jfn> I, A,, is of index 2 in S,,. and, hence. (A,,I =

Proof If n

1, S1 (e) = A1. Hence, A1 is trivially a normal subgroup. Letn> 1. Let G=(1,— 1) be the multiplicative group of integers 1,—f.

134

PermutatIon

group
givenby

Consider the mapping

4):S,,—'G

a'—'€(a).

By Lemma 2.2,4) is a homomorphism. Because n> 1, the transposition = (1 2) is inS, and 4)(r) = — 1. Hence, 4) is surjective. Therefore, by the fundamental theorem of homomorphisms, S,,/Ker 4) G. But Ker 4) is precisely the set A,, of all even permutations. Hence, A,, is a normal subgroup of S,,, and [S,:A,,J = IGI = 2. Consequently, IA,,I = IIS,,I =
3n!.

0

Definition. The subgroup A,, of all even permutations in S, is called the alternating group of degree n.
2.4

Examples

(a) In S3 the identity and the cycles (I 2 3), (I 3 2) are the only even permutations. Hence,

((l),(l 2 3),(l 3 2)). (b) A,, is nonabelian if n> 3. For, if a, b, c, d are distinct, then
A3 =

(a b cXa b d)

(a cXb d), whereas (a b dXa b c) = (a dXb c). Hence,

(c) The only possible even permutations inS4 are 3-cycles and products of two transpositions, in addition to the identity. Hence, A4 consists of the following elements:

e=(l), a=(I 2X3 a=(I 2 3), fl=(l
a2=(1
3

4),
3

2),

fl2=(I

4), y(l 4
4
3),

b='(l

3X2

y2(l

4), c=(l 4X2 3), 2), 3), 2 4), o2=(2 3 4).

ö=(2 4

It is easily verified that a, b, and c are conjugate to one another. For

example, aaa' = c.

Similarly, a, fi, y, and cS are conjugate elements, and so are a2, fl2, y2, and ö2. Thus, the conjugate classes in A4 are (e), {a,b,c), (a,fl,y,ö), and (a2,fi2,y2,ô2). Note that two elements of A,, may be conjugate in S,, but not in A,,. For example, a and a2, being of the same length, are conjugate in 54 but not in
A4.

(d) A4 has only one proper normal subgroup, which is the Klein fourgroup. Consequently, A4 has no subgroup of order 6. (This example shows

that the converse of Lagrange's theorem does not hold.)
c2

With the same notation as before, we easily verify that a2 = b2 = = e, ab = c, bc — a, ca = b. Hence, H {e,a,b,c) is a subgroup of A4.

Moreover, since a, b, and c constitute a conjugate class, H is normal. It is

Simplicity of

135

clear from the relations satisfied by a, h, and c that H is the Klein fourgroup.

Let K be any normal subgroup of A4. If XE K, then x2 E K, and because K is normal, every element conjugate to x or to x2 is also in K. then K must Hence, if K contains any of the elements contain all of them. But then KI> 8. Hence, by Lagrange's theorem,K = 44. This proves that H = (e,a,b,c) is the only normal subgroup of A4. Finally, any subgroup of order 6 of 44, being of index 2, would be a normal subgroup. Hence, A4 has no subgroup of order 6.

3
3.1

Simplicity of A,,
Lemma. The alternating group

is generated by the set of all

3-cycles in
1,2, for then A,, = (e) and S,, does not have any 3-cycle. So let n> 2. Now every 3-cycle is an even permutation and, hence, an element of A,,. Conversely, every element in A,, is a product of an even number of transpositions. Consider the product of any two transpositions a and r. If a and r are disjoint, say a = (a b), r = (c d),

Proof The result holds trivially for n =

thenar

(a bcXb c d).Otherwise,a = (a b),r = (b c)givesar =(a b c).

Thus, every even permutation is a product of 3-cycles. Hence, A,, is generated by the set of all 3-cycles in Sn. 0
3.2

Lemma. The derived group of S,,

is A,,.

Proof Forn= 1,2, fl=(1 23). Then
2X1
2

2 andleta=(l 2)and
3X2
1X3 2

l)=(I

2

3).

must contain every conjugate of Hence, (1 2 3) E S',,. Because Sn, On the other (1 2 3). Hence, every 3-cycle; therefore, An C hand, every commutator in an even permutation. Hence, C A,,. This proves that = A,,. °

Theorem. The alternating group A,, is simple if n > 4. Conse3.3 quently. S,, is not solvable if n > 4
Proof Suppose H is a nontrivial normal subgroup Of A,,. We first prove
that H must contain a 3-cycle. Let a
e

be a permutation in H that

.4 Example A. Hence.. Then r(a) = a. But r E H. is a normal subConsider now the derived group of A. . which proves that A. and t(x) = x whenever a(x) = x. and for every x other Let /3 = than e.. Because any two cycles of the same length are a transpoa conjugate in Sn. a cannot be a cycle of even length.. for all k 1.. Consequently. = An for every = A. Thus. Being an even permutation. Thus. A.... has no proper normal subgroup. r moves fewer elements than a.? — sition /1 in 5... With a (c d e) as before.b. Consider first case (I)..'. Let r be any 3-cycle inS.) Then are disjoint. H = A. is group of A. r E H. n > 4. it must move at least two more elements...(e d c) Now let r — = (c d eXa bXc d) =(a bxd (e d c) Then /3(a) a. Therefore. This = = A. a contradiction. Because a cannot be a 4-cycle. Consider now case (2).. Then d . is not solvable.c. either nonabelian if n> 3. Hence. But A.. Let a = (c d e). a and r are conjugate in A.. (afiXaXafiY' = Hence. 0 proves that S.. and An is simple. H contains every 3-cycle in 5. a must be a 3-cycle or have a decomposition of the form moves a=(a or b C (1) a=(a bXc d)".. =(c d =(a b e)(a b C •. Hence. flE Hand moves fewer integers than a. /3(b) b. 3. is not trivial. we conclude that a must be a 3-cycle. (2) where a. is the only nontrivial normal subgroup of Sn.). (This is possible because aca' = r.. a contradiction. say d and e. is simple if n> 4. Thus.. Therefore.136 PermutatIon groups the least number of integers in n. Therefore. /1(x) = x if o(x) = x. n >4.. Hence. Because is trivial or = A. such that a and /1 and n> 4....). Hence..d are distinct...

Hence. Consider a regular tetrahedron with vertices numbered I through 4. then IH = IHI 2 or )H1 2 ci E = because H * (e). the only nontrivial normal subgroup of S. so H or If H n (e).. b)fr d e).. If H — (e.. H A. Consider a cube with vertices numbered I through 8. n — 1) and Let H < S. Problems I.. Let H be the subgroup of the group G of rotations and reflections keeping the vertex I fixed.. Prove that S.) — (e). Show that there are 420 elements of S7 having disjoint cyclic decomposition of the type (a 6. Let K and H be. . then Z(S. Prove that every finite group is isomorphic to a subgroup of the alternating group A.. S3 and 161 = 48. This implies C H. Show that any group of order 4n + 2 has a subgroup of index 2. 7.. 5. is simple..SimplidtyofA. Show H 8. a contradiction. is generated by the cycles (I 2 (ii — I n). 4. But then — 2 implies 2. 9. Let (e) # H <1 Suppose H r) # (e). respectively. Show H S3 and IGI=24. isA.. Show that exactly half the permutations in H are even. = because A. Let S5 act on itself by conjugation. for some n> 1. 137 Solution. Show that there are 1120 elements of S8 having disjoint cyclic decomposition of the type (a b c)(d e f)(g h).. 3. Let H be the subgroup of the group 6 of rotations and reflections keeping one vertex fixed. the orbit and stabilizer of (1 2) S5. Then II n gives H fl A. 2.. Find also the number of elements in the orbit of (a b)(c d e). .tl). Show IKI = 10 and 1111=12.

x. under the given mapping.<IH. (ii) (iii) x.H. x. then the structure of the group can generally be determined. = H..CHAPTER 8 Structure theorems of groups I Direct products If a group is isomorphic to the direct product of a family of its subgroups (known as summands) whose structures are known to us. Then and (xç have the same image under the given isomorphism. Also e " e e. let x=x. Hence. = e. then each x. Proof (i) X fin.. •" •" H. When is a group isomorphic to the direct product of a given finite family of its subgroups? 1. to x1 Theorem. By uniqueness of representation.. and if x1 '' (iv) x E ii can be uniquely expressed as e. X (ii) Let H = H. Let H.. I 138 . Trivially.' H. these summands are like the "building blocks" of a structure. <1 Further.andll.4. X '' X H under the canonical mapping that sends (x.. be afamily ofsubgroups ofa group G. Then the following are equivalent. = e.=e.1 let H = (i) ii. x.).fl(H1 "'H. Thus.. Theorem 1. Also H. and XX1 (iii) H. <1 H.1 answers the question. . fin. 1!._. !1 <I H..x.

whereas the internal direct product exists if H. a(x.. . that the external direct product X H. But then.. and called the direct sum of H. Thus. for let x.)(a2) . and only if the canonical map H. 3 a2 E G such that a2 [a.][a2] is direct. So if a. X . I n... . we say that!! is an internal direct product of!!. we ultimately get G = (a. X H. yE To prove (iv). then ab = Let e '1' a. = e =x7'. is a homomorphism.. = e.. = e... (iv) (i) First observe that for all x E H. x C2 x x C. X H. The mapping is injective. then = and G C. (ah)' = Solution. so x..b E G. i #j. Under any one of the equivalent statements in Theorem I.]. X phism.]. (a.1. given by x.H. tion. we shall generally omit the word "internal" or "external" whenever the subgroups H......H..](a2] [a. = ba.. The mapping is clearly surjective. E G. I. it follows immediately that the mapping a: H. In the second situation. the (internal) direct product of subgroups H. Then X If. always exists. proving (i). let xs=xI where xfl. the cyclic - group generated by a.][a2]. ® ® I-f. .. i xy = yx.].. If G = (a. the direct product of cyclic groups each of order 2. however. X H. Then either G = [a.. Then that x1' = e..... Now x2 = e x = x' for all xE G. This proves (iv)... where C1 are cyclic and lCd = 2. ii follows that xy = for all x E H. . The emphasis on the word internal is more psychological than mathematical because the "external II.. = x. (b) A group G of order 4 is either cyclic of G product of two cyclic groups each of order 2.... we are through. of G is also written as H. C2 x C'2.... E since 1-11. in our situadirect product" H.H..' impliesx...Direct products 139 (iv) The hypothesis implies (iii) fl H3 = {e). This implies by hypothesis.. Continuing thus.2 (a) IGI Examples If each element e of a finite group G is of order 2. x1 = E H. Note. satisfy any one of the conditions in Theorem 1.]. .. Then the product (a. otherwise G [a. If the group is additive. For let Then e = xx. is isomorphic to I!..]..or G [a. yE Hi. a direct .. is an isomorX H... x. H.... = x2 x. . 0 x1eH1.H. 1. e = so by (iii) each x1 = e.11) '1 H. I Then since all xis commute.

+) cannot be written as the direct sum of two nontrivial subgroups. Solution. where H is a subgroup of order m.6. then G is cyclic. Since H and K are hk=kh for all hEH and keK. By Lagrange's theorem normal and HK Hence G is generated by hk. then H x K.2. Let N <1 G H X K. Solution. (e) X K nontrivially.11(e) in Chapter 4. (d) If G is a cyclic group of order mn. Proceed as in (d). then G is a direct product of cyclic groups of orders 0. and K is a subgroup of order n. 1 i S k. where (m.4. there exist a unique subgroup!! of order rn and a unique subgroup Kof order n. then 0(411k) = o(h)o(k) = pq. If H = [h] and K = [k]. G is a group of orderpq.8). Hence.n) = G Solution. and K = 2. Because G is cyclic. Prove that either Nis abelian or N intersects one of the subgroups H X (e). 4. and if G has a normal subgroup H of order p and a normal subgroup K of order q. Show that the group (Z/(4). Show that the group Z/( 10) is a direct sum of H = (0. Problems 1. Also then If x K. wherep and qare distinct primes. . By Lagrange's theorem H fl K = (e}. distinct primes. Show that the group (Z/(8). then by Lagrange's theorem G must contain an element a of order 2 and also another element b of order 2. = {e}. G is a finite cyclic group of order n p. IHKI = IHDKI = mn. Obviously. 1. If G is not cyclic.140 Structure theorems of groups Solution. 3. Then by Example 3. +) cannot be written as the direct sum of two subgroups of order 2.

.. the equation x1a1+ implies •. or. Let A be a finitely generated abelian group. for all integers x1 . the theorem is trivially true.ak) with the property that.. hence. This decomposition. C1 are of Proof Let k be the smallest number such that A is generated by a set of k elements.. +xkak=xtal+ '. and assume that the theorem holds for every group generated by a set of k — I elements.. Precisely.1 Theorem (fundamental theorem of finitely generated abelian groups). for some) k. .. (Z. We prove the theorem by induction on k. A is a cyclic group. are all infinite. Then A can be decomposed as a direct sum of a finite nwnber of cyclic groups Ci. 2. In this section we show that any finitely generated abelian group can be decomposed as a finite direct sum of cyclic groups. First consider the case when A has a generating set (a1 .Xk.Finitely generated abelian groups 2 141 Finitely generated abelian groups An important question in the study of an algebraic structure is to know if it can be decomposed into a direct sum of certain "well-behaved" direct summands. +xkak=O XkO This implies that every a E A has a unique representation of the form a = x1a1 + For if a=x1a1+ ". . . We may remark that not every finitely generated group is finite (consider..+)). when applied to finite abelian groups.. then + xkak. Let k> 1. X1 E Z. such that either nile. for example. If k = I. enables us to find the number of nonisomorphic abelian groups of a given order (Section 3). .... .

m1b1 = 0. Therefore.. . which proves our claim that A = C. (b1 . We claim that So suppose x1b1 CA1 for some x1... x. for some generating set (1) By the division algorithm we can write x1=q1m1+r1. Therefore. which implies x1 =0 by the minimal property olin1. C1 is a cyclic subgroup of order Let A1 be the subgroup generated by (a2. so that. Therefore.n1b1 + r2a2 + where0r.. Hence. namely.. Now A1 is generated by a set of k — I elements... Hence. A = where C1 = [a6] is the cyclic subgroup generated by a..142 Structure theorems of groups I = 1.. Now suppose that A has no generating set of k elements with the property stated in the last paragraph.<. Hence... C1 ® elements. Moreover. Then. i = ® l.ak) is a generating set of A. + rkak = 0. Then (I) becomes + where b1 = a1 + q2a2 + Now b1 # 0.. x2.. A is the direct sum ofa finite number ofinfinite cyclic subgroups. a contradiction.Xk (not all zero) such that x1a1 + + xkak = 0.a&} of A... and let Xdenote the set of all k-tuples (x1 . there exist integers x1 . we can assume that x > 0 for some i. Consider now all possible generating sets ofA with k Hence. in this case.a2. is an infinite cyclic group. @ A1. (a2 ak). by the minimal property of in1.. = (0). k. = 0 implies x1 0..xkEZ. Since =0 = 0. A1 cannot be generated by a set with less than k — elements.. Let C1 = [b1].n.k. a1 = b1 — q2a2 — — Hence.. by Theorem LI. for otherwise a1 —q2a2 — which implies that A is generated by a set of k — I elements.. given any generating set (a .. fl A. we may take rn1 to be the first component. Without loss of generality..b1—x2a2— — =0. C. Moreover... Because m1 is the least positive integer such that m1b1 = + 0a2 + + Oak =0.x&) of integers such that x1>Oforsomci. it follows from (2) that r2 = = rk = 0. (2) for each i = 2. a 1 . Then x1b1=x2a2+ C.ak).. Let m1 be the least positive integer that occurs as a component in any k-tuple in X.. Moreover.... for some generating set (a1 ofA.. since x1a. A =C1 < in. for otherwise A would be generated by a set with less than k elements..k.

since X = ISH TI for any finite sets S and T it follows that IAI = ''• = P"lI Moreover...mk are determined uniquely.. we conclude that m11m2. 3 Invarlants of a finite abelian group Throughout this section A denotes a finite abelian group written additively.. any cyclic group of order m is isomorphic to Zm.mk. i = 2. Hence... Then there exists a (all> I) such that unique list of integers m1 ...Ck are cyclic subgroups that are either all infinite or. A = C... if A is a finite abelian group. Hence.. ® ® respectively... C2.. by the induction hypothesis.. IAI = andA = C1 ® .. are finite cyclic groups of orders m2.b1 By + m2b2 + 0b3 + + = 0. where C2. are infinite for i >j. then C1 Ck are all finite..Invariants of a finite abelian group 143 contradiction. 0 Evidently. This completes the proof of the theorem.. consequently..... and suppose that C2 is of order m2. such that subgroups of orders Further. for some j k. and m.. are cyclic subgroups ofA of order mk.. Let A be a finite abe/ian group..m. ® where C... suppose . respectively. respectively.. repeating the argument given for equation (I)... .bk) is a generating set of A.1 Theorem... To prove the uniqueness of the list rn. 3....k. are cyclic where C.. and with m21m31 Let C1 = [b1].. Then {b1 . .C. We prove in the next section that for a given A the positive integers nz3 .

..!— I.1..144 Structure theorems of groups where are all cyclic subgroups of A.k— I..... we canshowthatmk_..e3).. mk. is obviously the preimage of (e1 ...2 Proof Let A E F. Proceeding in this manner....Imk. a is clearly injective (Theorem 3..k.. then A is said to be oftype (m1 ...lmk_I for Ci) ® •. 0 0 Z.k.. Now D.... This completes the proof.e.. Therefore. Definition. To show that a is surjective.=n.A of A assumed above.... mk=IAI=nI n..... Therefore. and the integers m1 .. 3. it follows that mk... Byasymmetricargument.IC.. 0 . Then A determines a unique type where se.) of positive integers such that k = 1k.. 0 ._... mk Now consider mk_ = (mk_ 1ala E A).. From the two decompositions ..2 i= 1.•(m&.. Then the group Z.l. A partition of a positive integer k is an r-tuple (k1 .. Hence.... mijm2I..1). . Hence.. IPnk_IDJI= I forj= 1. But every element in A is of order Hence..... and each Lemma. n. = n.ek). we get ?nk_IA = (Pnk_ICI) ® ._Ifrnk_I... n. p prime. and the set P(e) of partitions of e. Hence.mk). mk_dn._I.r=O.k— 1.mk are called the invariants of A..)DJ) Because ?n. let (e. The fundamental theorem of finite abelian groups enables us to determine the number of isomorphism classes of abelian groups of order n. If A where I <mlImlt.. There is a 1-I correspondence between the family F of nonisomorphic abe/ian groups of order p'. and e1+e2+ Define a mapping a: F—' P(e) by o(A) = (e1 . is trivial for i= l. 0 (mk_IDI_I) 0 i= 1... with fC1J=m. JDjJ=nj..) E P(e).. ® = (Ink... By the same argument mk n1.. has an element of order n... Let A be afinite abe/ian group..

By Theorem 3. ® By Lemma 3. By Lagrange's theorem . • 0.)® ® S(p. Let A C Abe... We have IA = .. say.. of orders for all j So 1 .® Put S(p1) fA. e..> 0.. Then A p. by Theorem 1.. .4 0 Theorem. A.).)I IS(pk)I = ps'. .IP(f1)I. respectively..3 Lemma. that is.. where A. 3. Let A be a finite abelian group of order distinct primes. Then by comparing orders on both sides. The last statement can be proved easily by observing that each of the subgroups S(p. = S(p.. contains unique subgroups A. 0.1. A = Proof.=A.. However. A. e.) and H. A = S(p1) ® . we give a proof based on the fundamental theorem of finitely generated abelian groups.. are cyclic groups of orders. if where then H. ® . 3..J.k.. By Lemma 3. Let n = p1 distinct primes.Invariants of a finite abelian group 145 The following lemma is an immediate consequence of Sylow's theorems discussed in the next section. Proof Let be the family of nonisomorphic abelian groups of order n.1. is the subgroup (x C A 0(x) is a power of p.). where This decomposition of A is unique. Because A1 is cyclic. it follows that IS(p. ® A = S(p. Therefore. Then the number of nonisomorphic abelian groups of order n is ..3.2 .

The Sylow theorems. There is no such general result for finite nonabelian groups.146 Structure theorems of groups the number of nonisomorphic abelian groups S(p1) is IP(fJI. 4. IP(2)l = 2. Z2®Z4®Z. Z2®Z2®Z2®4®Z5. A group G is called a p-group if the order of every element in G is some power ofp. Let G be a finite group. 51.5 Example Find the nonisomorphic abelian groups of order 360 Solution. m > 0.9®Z5. Hence. zs®z. the existence or nonexistence ofa simple group ofa given order can sometimes be determined by application of the Sylow theorems. a subgroup H ofany group G is called a p-subgroup of 0 if the order of every element in H is some power of p. 32. Let p be prime. among other machinery. and let p be a prime. 4® Z3 ® Z3 Z5.1 . Let PtmI KIGI. In particular. n = 360 32. Likewise. Hence. Then any subgroup of G of order ptm is called a Sylow p-subgroup of G. IP(3)l = 3.®Z3. and let p be a prime. IP( 1)1 =1. Lemma (Cauchy's theorem for abelian groups). 4 ® Z4 ® 4 ® Z3 ® Z5. Let A be afinite abe/ian group. there are six nonisomorphic abelian groups of order 360. Definition. which occupy an important place in the theory of finite groups. 4 Sylow theorems In the last section we obtained the decomposition of a finite abelian group A that provides a complete description of the structure of A. Jfp divides IAI. k I—I ° 3. Z2®Z2®Z2®Z3®Z3®Z5. 5'. and these are = . Definition. then A has an element of order p. are a powerful tool in studying the structure of finite nonabelian groups.

If n = l. Let G be afinite group. a E A. G/C contains a subgroup H/C of order pm—I. Ifp I lEbli. then p divides the order ofA/[b]. and therefore there exists an element oforderp. which is divisible by ptm I. then there exists a cyclic subgroup of A oforderp. Proof By Sylow's theorem there exists a subgroup of order p.the result is obvious. The cyclic group C generated by a is clearly normal in G. Let k = o(a). Suppose now [bJ. and this yields the required subgroup of G. If ptm divides 1GI. pim. Then since o(a) = p. the order of A. and the quotient group G/C has order nip. By the induction hypothesis N(a) has a subgroup of order ptm. Suppose that the result is true for all groups of orders m. some power of the element a is of order p. Now pjn. that o(ã) = p. but pf IZ(G)I. p1k. hence. Therefore. Proof We prove the theorem by induction on n =IGI. Hence. the cyclic subgroup generated by there exists b ('# e) E A such that A b. 0 . such = e implies ãk = Therefore. proving the theorem in this case. If the order of the center of G is divisible by p. then it follows from Cauchy's theorem for abelian groups that the center of G contains an element a of order p. a Z(G). If the order of a finite group G is divisible by a prime number p. then G has an element of order p.2 be a prime.Sylow theorems 147 Proof We prove the theorem by induction on n. where pin. and let p 4. 0 A special case of the theorem is 43 Corollary (Cauchy's theorem). p%[G: N(a)] for some a E G. by the induction hypothesis. So let us assume that the result holds for all groups of orders less than n. This means p divides the order of[aJ. then G has a subgroup of order ptm. This implies that IN(a)i and IN(a)i <IGI. Hence. 0 Theorem (first Sylow theorem). by the induction hypothesis there exists liE A/[b]. Therefore. The class equation of G is n =1GI=1Z(G)I+ where summation runs over one element from each conjugate class having more than one element. Now consider the case when the order of the center is not divisible by p. we are done. there are at least p — I elements of order p (namely. and m <n. But ifp KiIb]i. all but the identity element of the cyclic subgroup of order p). If IAI = p or even if A is a cyclic group.

therefore. a contradiction. let G. and their number divides 0(G) and satisfies n. and/el p be a prime. any L e C(K). G is of order a power ofp. So let us assume that each element in G has order a power ofp and prove that order of G is a power ofp. = index of!! fl N(L) in H = p'. If 11= L. Regard the set C(K) as an 11-set by conjugation. and C(K) = U CH(L) LEC(K) (disjoint). e 0. (I) = 1.5 Theorem (second and third Sylow theorems). 0 4. Then all Sylow p-subgroups ofG are conjugate. Because K is a Sylow p-subgroup. Proof If G is of prime power order.. Let C(K) denote the family of subgroups of G conjugate to K. p= I But then HL/L established. then. Proof Let K be a Sylow p-subgroup in G. A finite group G is a p-group if its order is a power ofp. H/H fl L HL/L is a p-group IlL = I since L is a Sylow p-subgroup.4 Corollary. the orbit of L.148 Structure theorems of groups 4. Then for ER). Then by Cauchy's theorem there exists an element of order q. Obviously. Let G be a finite group. So our assertion that p' = if H L is . I (mod p). (2) where the union runs over one element L from each orbit [= conjugate class CH(L)J.. then each element has. C(K) = C G). Next let H be any Sylow p-subgroup. Then = IG/N(K)I..p)= 1. trivially. We first show that pt = I if I! = L. If possible.13 (b) in Chapter 5. (IC(K)I. trivially. Hence. order a power of p. p' = 1. Conversely.(L) C C(K). (3) because H is Sylow p-subgroup. let q # p be a prime number dividing the order of G. By Example 4. that is.(L) = C'. We want to show that 11 is conjugate to K..

and G is therefore a cyclic group generated by a. by what we established in the last paragraph.. a H. then A = pr'.) of order group of order d. Thus.. If n = p1' . Because all Sylow p-subgroups of G are conjugate to K. it follows that all the proper subgroups of G are contained in H. Hence.. 4. n. (a) If a group of order p" contains exactly one subgroup each of orders Solution.3). that H must be equal to some conjugate L of K. so we have from (4) that the number of distinct conjugates of K is (modp). p2. Let G be a group of order pfl. and let H be a subgroup of order By Sylow's fIrst theorem H contains subgroups of orders p..) is a subThen B = S'(p1) ® . pr'. e>O proving the third Sylow theorem. it follows that their number Is equal to IC(K)I = IGI/IN(K)I. 4.7 Examples then it is cyclic. Because these subgroups are also subgroups of G.. otherwise a will generate a subgroup K of order <p'. This would mean a E H. and this in turn implies.)..Sylow theorems 149 By (2) and (3) we have IC(K)I = (4) at least one term pe must But then (I) implies that in the summation be equal to I. . then the order of a must be pfl. Now if a E G. 1 S(p1) ® ® S(p. the order of a is p".6 0 Corollary. because a E K C H. Solution. then A contains a subgroup of order d.. (b) If d is a divisor of the order n of a finite abelian group A..I IGI.. that is It is also clear that there is one and only one term in the sum equal to I. subgroup S'(p. A Sylow p-subgroup ofa finite group G is unique only (fit is normal. and = i r (Lemma 3. proving the second Sylow theorem. By Sylow's first theorem each S(p1) contains a Let d = S'(p.

(d) Let G be a group of order 108. and therefore G cannot be a simple group. there is a unique Sylow 7-subgroup that must be a normal subgroup. Thus Kerq is a nontrivial = 1541 = 24. the number of Sylow 7-subgroups n7 = I + 7k 7. fin K 'I Ji and i/fl K 1 K because subgroups of order are normal in a group of order p" (or. Further. and 36. (1 + 7k)12'. N(HflK)Jii IHKI=lHflK. Chapter 5.11 + 3kand(l + 33. we must divide are done.11(e). since the index of H in G is 4. 33• The number of Sylow 3-subgroups is a3 = 1. (c) Prove that there are no simple groups of orders 63. and 1 + 7k divides 32. This yields a7 I or 8. and N(HnK)JK. Therefore. H fl K = H fl N(K) <1 H). The number of Sylow 7-subgroups is n7 = + 7k. Because H fl K is a normal subgroup in H as well as in K. Example (b) proves the converse of Lagrange's theorem for abehan groups. we have a unique Sylow 3-subgroup that is then a normal subgroup Solution. = l G. = 108 = 22. Now consider N(H n K). For groups of order 56. Show that there exists a normal subgroup of order 27 or 9.56.Thus. 27)/IH fl 27/4 < hUn KL so IH fl JCI= 9. If k = I.S4. then there are 8 6 = 48 elements of order 7. Note implies G s S4. Because IKI 27 =81. But then I + 7k divides 32• This yields k 0. But then the remaining eight elements must form a unique Sylow 2-subgroup. So 1 again we have a normal subgroup.= . 27 Then HK C N(H fl K). Thisgives (Example 3. . lfn7 = 1.4. Solution. Here is an alternative method. If n7 = 8. alternatively. which is not possible since jGI = 36 and Kerq. The nonsimplicity of groups of order 36 may be proved similarly. 7.(I + k =0. which is sometimes easier. Let H be a Sylow 3-subgroup of G. Then from the result II1KI=IHIIKI/IH fl 108. By Theorem 4. IGI I of order 27. we have four Sylow 3-subgroups. Also normal subgroup of G and so G is not simple. there exists a homomorphism p:G—. Let H and K be two distinct Sylow 3-subgroups.150 Structure theorems of groups Note.jHKp= (27 . Thus. Chapter4).

Solution. and. By the third Sylow theorem the number is equal 'o I + kp for some positive integer k.andthen we get by Corollary 4. (p. and.q) = 1. Let IGI = p"q.Sylow theorems 151 and N(H n K) < G. clearly. Further. where p and q are prime numbers such thatp> qand q%(p— 1). G is generated by the element xi'. (e) If 11 is a normal subgroup ofa finite group G. . then the order of the element xy is equal to IHIIKI = pq. all Sylow p-subgroups of G are contained in 11. say 11. then H contains every Sylow p-subgroup of G. IN(H fl HflK<1G. Let K1 be another Sylow p-subgroup of G. K is also a Sylow p-subgroup of G. G is cyclic. thus. and thus q=A1(1 +kp) for some positive integer A1. the number flg of Sylow q-subgroups is equal to I + k'q for some integer k'. Hence. that must be normal. H (as a group by itself) contains a Sylow p-subgroup K. (f) Let G be a group of order pq. IHI = p"q1. uk V/i E H and Vk E K. since 11<1G. p-subgroups pq=A(l +kp). and. Hence. This implies pIA. Hence. for some positive integer A. But sincep>q. Then by Sylow's second theorem K1 = x E G. By Sylow's first theorem.q1) = I. Then IKI = ptm. Because IKI = K1 = C C H. say K. K) = G.6 that there is only one Sylow p-subgroup. It is easy to see that K fl If = {e}. and if the index of 11 in G is prime to p. hence. Because the index of H — that is IGI/IHI — is prime top. This implies that if x and y generate the cyclic groups H and K. (p. and by the same argument pp1(l +k'q) for some positive integer Butsinceqj'(p— +k'q)ispossibleonly ifk'=O. Solution.q=A1(l +kp)isnotpossibleunlessk=0. respectively.6. ptm. K is normal by Corollary 4. Then Gis cyclic. Hence. and this implies that there is a unique Sylow q-subgroup.

Then G contains no element of order 8. Now b2 = a or o(b) is 8.4). and that a direct product of cyclic groups is abelian. there are two. Throughout this section p and q are distinct primes. We denote by [xJ the cyclic group generated by x. pq We recall that there are cyclic groups of every possible order. (8). we must have b2 = eor a2. b'ab = a3. That they are nonisomorphic is easily checked. Let b E G such that b [a]. a or a3.y.y] the group generated by x. and G is abelian. (ii) The group G generated by a and b with defining relations a4e. o(a). Solution. more generally. Then G = [a] U [ajb. We also recall the following properties of a finite p-group P: Z(P) is nontrivial. 5. Incidently. If each element of G is of order 2. and only two. there are three. Hence.1 Groups of order p2 We have already shown (Corollary 4. Because [a] is of index 2 in G. otherwise G is abelian. we have Thus. 5 Groups of orders p2. Thus G contains an element a of order 4. Hence. Solution. that two cyclic groups of the same order are isomorphic. The first is the octic group. [a] <1G.11. and.2).152 Structure theorems of groups (g) Any group of order 15 is cyclic. In the following. by [x. is normal. b2=e. (2. groups of order p2: . (h) There are only two nonabelian groups of order 8. Follows from (f). then every subgroup (1) H. then N(H) (ii) (iii) If 111= p".2. Chapter 5) that a group of order p2 is abelian. Consider a nonabelian group G of order 8. Thus. and the second is the quaternion group. two groups isomorphic to each other are identified. and b2 C [a]. and only three. we have two nonabelian groups of order 8: (i) The group G generated by a and b with defining relations a4=e. If H is a proper subgroup of P. abelian groups of order 8 that are of types (2. then G is abelian. Because a3 = C [a]. Hence. b2=a2.

(I) (2) The integer r in (I) is a solution of the congruence equation (modq) for = b' = b'3 = b2' by a'b2a — b2' induction = b" = b'2 a2ba2 = b'2 by induction (modq). = for some integer r. and thus there is exactly one Sylow q-subgroup. Since [a) and [b) are normal in G. which divides pq.r2.whence G = [a. a contradiction. for otherwise ba = ab. pq 153 (i) (ii) 5. be this unique Sylow q-subgroup.. thus. In this case q I (mod p). Groups of order pq. Case (ii). Conversely. G — lab). and (a] fl [b) — {e). Because[bJ 'I G. if r is a solution of the congruence equation (2).. and. because replacing a by a1 as a generator of [a] replaces r by r1. Also the number of Sylow p-subgroups is 1 + pp. . a'ba—b'. q > p Let G be a group of order pq. and all yield the same group. Then r 1 (mod q). which. Thus. then the defining relations (1) determine a group consisting of the pq distinct elementsa'bJ. Suppose the number of Sylow p-subgroups in G is I. being of order q. which in turn implies that the number of Sylow p-subgroups is 1.2 abe/ian group of type (p2). where a' — 1.Oi<p— l(modq)arer.b) contains a subgroup (a] oforderp and a subgroup (b) of order pq.. where I. Then the subgroup Ia. Because o(ab) ispq. is cyclic. This gives 1 + Aq = I. Let [bI. we have G [a] X [b). G is generated by a and b with defining relations a'— I —be. and G is abelian. We conclude that there are at most two groups of order pq: (i) The cyclic group of order pq.. Let [a) be one of the Sylow p-subgroups of G. where a' 1.b].p). This gives I + Case (1).Groups of orders p2. G is cyclic. or q. which divides pq. and abe/ian group of type (p. The number of Sylow p-subgroups is q. and Let 1 this be Ia). Then the number of Sylow q-subgroups of G is I + Aq.

Suppose a group G H X K is a product of groups II and K. Show that the set of positive integers less than and prime to 21 form a group with respect to multiplication modulo 21. 14. Using this 13. I (mod q). hr a and b with defining re/aI (mod q). Show that a group of order 1986 is not simple. 15. Show I divides (np)!. . p and q distinct primes. 9. Prove that any group G of order 12 that is not isomorphic to 44 contains an element of order 6. prove groups of order 48 cannot be simple.J Structure theorems of groups The nonabelian group generated lions I a'ha=b'. Prove that the group of all rational numbers under addition cannot be written as a direct sum of two nontrivial subgroups. 12. I. Show that G X H is cyclic if and only if(rn. and let H be a cyclic group of order n. Show that an abelian group oforderpk and of type contains — I elements of order p. then the group must be cyclic. 16. 3. prove that G is the direct product of its Sylow subgroups. 7. 8. prove that its Sylow 7-subgroup is normal.154 (1i. Prove that this group is isomorphic toll X K. Prove that a group of order 200 must contain a normal Sylow subgroup. 11. Show that a finite abelian group that is not cyclic contains a subgroup of type (p. or otherwise. where p is prime. Let G be a simple group and let n.n. must contain a normal Sylow subgroup and be solvable. If the order of a group is 42. provided p dividec q Problems I. 6.n) = I. Show that a group of order p2q. If n is a divisor of the order of an abelian group. Prove that G/ll X {e) K. 10. Deduce that G has a normal Sylow 3-subgroup. where p is prime. respectively. Ifeach Sylow subgroup ofa finite group G is normal. be the number of Sylow p-subgroups. 4. where H and K are cyclic groups of orders 6 and 2. Let G be a cyclic group of order . Prove that if the order of an abelian group is not divisible by a square. 5. 2. prove that the number of solutions of x" = e in G is a multiple of n. Prove that any group oforder 2p must have a normal subgroup of order p.p).

If M is a normal p-subgroup. Let H < G such that H H fl P. 22. show that 'PxrK is a Sylow p-subgroup of K for some x€H. 18. subgroups of G.j Prove that A5 is the first nonabelian simple group. Then use Problem 21 to obtain that if H is a 24.Groupsofordersp2. group and G is solvable. and let P be a Sylow p-subgroup of G. What is the order of Z(G)? Suppose G is a finite group of order PIPZ where are distinct primes. show that M c P. K <G. 21. Let G be a finite group. Prove that a finite group G is nilpotent if and only if it is the direct product of its Sylow subgroups. Show that N Ii P is a Sylow p-subgroup of N. Suppose P is a Sylow p-subgroup of a group G. Sylow p-subgroup. Let G be a finite group such that G = HK. 23. Let N = N(P). 155 Let G be a group. Also. Let G be a finite group with a Sylow p-subgroup P. that Z(G) = G'. Let P be a Sylow p-subgroup of a finite group G. Show that any subgroup of G that contains N (N included) is equal to its own normalizer. [Hint: First show that if!! < G. . where p is prime. Suppose A G <1 G and G be a nonabelian group of order p3. For a prime p. where H < G. Show 20. 25. then H <1 G. Show that G contains a normal Sylow sub.pq 17. show that NP/N is a Sylow p-subgroup of GIN. Let N <1 G. then N(H) # H. Let A be an abelian subgroup ofG. Show also that G con- tains a normal p-subgroup H that contains all normal p27. if P is a Sylow p-subgroup of 6. Prove that IG/Z(G)I # 77. 26.

.

PART Ill Rings and modules .

.

such that (R. Q: the ring of all rational numbers. .b E R.) a (b+c)=a b+a c. We denote the zero element of a ring by 0. a+(—a)0 forallaER. (The two distributive laws are respective/v called the left distributive law and the right distributive law. b. that is. Further.) is a multiplicative semigroup. E R.) We usually write ab instead of a b. (a+b) c=a c+b c. Thus.CHAPTER 9 Rings I Definition and examples Definition.+) shall. (R. Further. be denoted by — a. the additive inverse ofan element a of the additive abehian group (R. as usual. in a ring R. 159 . respectively.+) is an additive abe/ian group.) Multiplication is distributive (on both sides) over addition.c. if a.b.i. (iii. for all a. (U (ii. The identity of the additive abehan group is called zero element of the ring Rand is unique.nultiplication ('aLco called producf. By a ring we mean a nonemptv set R wit/i two binary operations + and called addition and . we denote a + (—b) by a — The most familiar examples of rings are Z: the ring of all integers.

) To illustrate. Other interesting examples are CEO.(n — I)). that is. The addition in each of the rings is the usual addition of mappings. that is.bEA The ringA so obtained is called a trivial ring. Generally. ab = 00000 iO!23 20202 30321 R such that A commutative ring is a ring R in which multiplication is commutative. 00123 11230 22301 33012 that is. We define a trivial ring as one in which the product (or multiplication) of any two elements is 0. The element e is called unity or the identity element of R. A ring with unity is a ring R in which the multiplicative semigroup has an identity element. the ring of all functions from a nonempty set X to the real numbers R. the unity or identity element is denoted by I.. = (0.b R. Another example of a ring is an additive abelian group A in which multiplication is defined by ab=0 foralla. . Addition and multiplication in 4 are defined as follows: (See Example 3.1 in Chapter 4. that is. We define multiplication by (JgXx) =f(x)g(x). we give the addition and multiplication tables of 4. a ring with finite number of (or Z/(n)): the ring of integers modulo n.. there exists e ae = a = ea for all a E R. C: the ring of all complex numbers.The elements of elements — is are the residue classes mod n.l. (f+ g)(x) =f(x) + g(x). If a ring is not commutative it is called noncommut alive. 1): the ring of all continuous functions from the interval 10.. ba for all a.1) to R. An example of a finite ring — that is.160 Rings R: the ring of all real numbers..

b. +(—(aO))= aO. aO So. —(ab) = (—a)b. Let R be a ring. (iii) a(b—c)=ab—ac. . We define their product a. (ii) o aO = a(b + (—b)) = ab + a(—b). Thus.(a—b)c=ac—bc.c C R. Proof (i) aO a(O +0) = aO + aO.Elementary properties of rhigs 2 161 Elementary properties of rings Theorem. (iii) a(b—c)= a(b +(— c))= ab+ a(—c)= ab— ac. o = Oa. (a—b)c=ac--bc. 0 of elements of a ring Let a1 be a sequence inductively: R. O=aO. 2. Similarly. Then for all a. n>!.1 (i) aO=O=Oa. Similarly. (ii) a(—b) = —(ab) = (—a)b. — (ab) = a(— b). = a1. Thus. Similarly.

(See Theorem 4. Also we define Oa to be the zero element of the ring.2 Theorem (n arn÷j) = njy a. we write in times and in times am=a.a. in a ring R.a2. by using induction on m and n..) 2.2 of Chapter I. a negative integer. then a—rn denotes the element (ar. The uniquely determined product of the sequence a1 is denoted a1a2 In the same way we can dewe can easily fine uniquely the sum of the sequence a1 + Also.. .. then ma stands for -m times (—a) + (— a) + + a). for al/a1 .... called the generalized associative law for products. If a E R and in is a positive integer. 2.b2+ •. essentially asserts that we can insert parentheses in any manner in our product without changing its value. This result.162 RIngs We then have the following result. If R. a2 + prove the generalized distributive law given in 2. which can be directly proved by induction on m and n... (a) For all positive integers in and n and for a/la in a ring R.am+. If a has an inverse I E R. in a ring R. then we define a° = 1. the following hold: (I) (ii) (a'? = a'"".4 Theorem.3 Theorem +amXbt+b2+ +a2b1+ for al/a1 am and b1 ±ambi±a..a If in is ma—a+a+"+a..

and the complex numbers C are examples of fields. It is left as an exercise for the reader. Thus. 3 Types of rings Definition.b in a ring R thefollowing hold. m(na) — (mn)a. The rational numbers Q. If in addition. and let exists. A ring R whose nonzero elements form a group under multiplication is called a division ring. (iii) ma + na (iv) (v) (m + n)a. implies x =0 or y =0. Ifx #0. x.—& a complex a a nonzero matrix. Every field (therefore every division ring) is an integral domain. then its determinant ô = + bb # 0.y E F. and. (maXnb) = (mnXab) — (naXmb). The ring of integers Z is an integral domain. hence. R is commutative. Let H be the set of2 X 2 matrices of the form I a bl a and a I. so x'(xy) —0. ã —b A ring R is called an integral domain zfxy = 0. then . (Let F be a field. x E R.Types of rings 163 (b) For all integers m and n andfor all a. The theorem may be proved by using induction on m and n. An example of a division ring that is not a field is the ring of real quaternions. y E R. then R is called a field. the real numbers R.) xy = 0. y = 0.

where aMER.e.J) entry is I and whose other entries are zero. One can similarly define left zero divisor. An element x in a ring R is called a right zero divisor if there exists a nonzero element yE R such that yx = 0.) I)efinition. Matrices with entries in a ring R (i. it follows that ej)ek. The i. all of whose entries below the main .j n.164 Rings However. Then with the usual addition and multiplication of matrices is called the ring ofnX n matrices over R. then is not commutative. It also follows that if n> I. are called matrix units. and = eth. Let R be a ring. = 0 if3 k. (For example.1 Some important examples of rings (a) The ring of matrices Rudiments of matrices over any field have been given in Chapter 3. R has unity I. contains nonzero zero divisors. An element is called a zero divisor if it is both a right and a left zero divisor. I This representation of matrices in terms of matrix units is often a conve- nient tool in certain problems involving algebraic manipulations.. that is. the ring of integers Z is neither. that is. Let Sbe the set of n X n matrices. if A then A can be E uniquely expressed as a linear combination of ç. From the definition of multiplication of matrices. We denote by the matrices in whose (i. We may note that a ring R is an integral domain ifand only if it has no right (or left) zero divisors except 0. Also. If n> I and R is not a trivial ring.'s over R. matrices over R) and their addition and multiplication can be defined in exactly the same manner. For if b=[g then ab ba. 3. and let be the set of all n X n matrices over R. that is. 1 eUek. an integral domain need not be a division ring or a field. = is the Kronecker delta.

Let b0 + b1x + a1ER._1. and let .= j+k—I aJbk. Alternatively. a1 E B. where f(i)= a.Types of rings 165 diagonal are zero. a E R. A polynomial with coefficients in R is a formal expression of the form a0 + a1x + a2x2 + are + amxm. one may look upon a0 + a1x + as + amxm. that is.). and + b1 R.ai. that is. a function f:N —'B.am.O.x + + called equal if m = n and a = b.x". f(i)—'O.. g(x)=b0+b1x+b2x2+"' +b.... for all i.. I I m + I. (b) The ring of polynomials Let R be a ring. Similarly. a sequence (ao. Two polynomials a0 + a1x + + amx'" and b0 + b. let S consist of matrices a11 a12 as.i>m+ I. +b1)x+(a2+b2)x2+ where c.O.. we have the ring of lower triangular matrices. Polynomials may be added and multiplied to form a ring in the following manner. be another polynomial. Let f(x) = Then + a1x + a2x2 + + amxm... o a ring with the usual addition and multiplication of matrices and is called the ring of upper triangular matrices..

namely. (f+ gXi) =f(i) + g(i).)... The sumf+gof the functionsfandgis defined to be the or. In what follows we assume I E R. RIxj contains a carbon copy of R. Then x2 (0. We call x an indeterminate and refer to the polynomials in R[xJ as polynomials in one indeterminate over R (sometimes we also call them polynomials in a variable x). (a0 + a1x + where Cl— J+k—i f(j)g(k).0...0.) = (a.. we say that R is a subring of R[xJ..1.0.).. R be the corresponding function. denoted by Further.. and so on.0. so that g(i) = b11. ab —p (ah..0. the mapping a —' (a.0...).).) = (a.. sequences with finitely many nonzero entries) forms a ring called the poiyring over R....0.0.0.a1 +(0 which yields the familiar expression of a polynomial in x...0.. Consider with finitely many nonzero entries.0. ..0. Thus..) + (b. + amxm) + (b0 + b1x + + b.e..0.0.. iEN...0.1.) E R[x].. I n+ I and g(i)'=O.)(b. if b —' (b.. that is.).. . We find a sequence (a0.166 Rings i g: N —..0. x3 = (0.0.0.) of the ring R into the ring R(xJ preserves binary operations.x") =(a0+b0)+(a1+b1)x+(a2+b2)x2+ The product of two functionsfand g is defined to be the function h.0. I> n +1.0...0. equivalently..). a + b —p (a and the mapping is also I-I.. (a0 + a1x + usual sum of functions.0...1.. then + b. equivalently...0... and we denote by x the sequence (0.0.x"Xb0 + b1x + = C0 + It is now straightforward to verify that the set of polynomials (i. where h(i) or. By identifying the element a E R with the sequence (a. + + + + a.

y C (—fXx + y) = — [f(x + y)] = — [f(x) +f(y)J —f(x) —f(y) = (—fXx) + (—f)(y). 1ff C End(M). Let End(M) [or Hom(M. We show that the set End(M) together with the above binary operations of addition and multiplication forms a ring. We define addition and multiplication in End(M) as follows: (i) (ii) (f+ gXx) =f(x) + g(x). Obviously. = [1(x) + g(x)] + [f(y) + g(y)] = (f+ g)(x) + (f+ guy) This shows (f+ g) C End(M). Then —f C End(M) because. Also it is easily shown that the two mappings 0: x and I: x '—' 0 (zero endomorphism of M) (identity endomorphism of U) x of M into M are also endomorphisms of U.f+ 0 =fandf+ (—f) 0 for alifE End(M). the set of endomorphisms of the group M into itself.y C M. (fg)(x) =f(g(x)). as usual.Types of rings 167 (c) Ring of endomorphisms of an abelian group Let M be an additive abelian group.g E End(M) and XE M. M)] denote. Let x. for allf. define —fto be the mapping given by (—fXx) = —f(x) for all x C M. for any x. then f+ g C End(M) and fg C End(M). Then (f+ gXx + y) =f(x + v) + g(x + y) [by (i)J = [f(x) +f(y)] + [g(x) + g(y)] (because fand g are homomorphisms) (because M is abelian) [by (i)]. . Further (fgXx + y) =f(g(x + y)) [by (ii)) =f[g(x) + g(y)] (because g is a homomorphism) =f(g(x)) +f(g(y)) (becausefis a homomorphism) = (fgXx) + [by (ii)). First we show that if fg E End(M). which shows thatfg E End(M).

Let (R. f+g=g+f (fg)h (f+ g)li =fh + gh. Similarly. (f+ g) + h f+ (g + h). (d) Boolean ring Let be the set of all subsets ofa nonempty set A [called the power set ofA and denoted by If a. It can be easily shown that x2 = x for all x E R implies 2x =0. it can be easily shown that for all jg. We note that addition and multiplication of elements of S are to coincide with addition and multiplication of these elements considered as elements of the larger ring R. Some of the important properties of the ring R are (i) (ii) &=a. ab = an b. The zero element of this ring is the empty set.168 Rings Furthermore. and let S be a nonelnpty subset of R.and 2a = 0. and f(gh).) be a ring. we can define a subdivision ring of a division ring and a subfield of a field. along the same lines.h E End(M). Hence (End(M).) is itself a ring. . It is interesting to point out that any Boolean ring is known to be a subring ofthe ring of all subsets of a set.+.b E we define a+b=(aUb)—(aflb). f(g+ h)—fg+fh. for all a E A ring R is called a Boolean ring ifx2 = x for all x E R.+. +. Then ) is a commutative ring with identity (which is the whole set A). 4 Subrings and characteristic of a ring Definition. Then S is called a subring ([(S.) is a ring.

The condition a — b e S for all a.+). ab E S. The x Definition.+. ) is a subring of(R. x E R. Therefore. n1 E Z.b E S implies (S. Let R be a ring. If there exists a positive integer n such that na = 0 for each element a of a ring R. Let (a—b)x——ax—bx=xa—xb=x(a—b). Because the intersection of a family of subrings is again a subring. The subring generated by the empty set is clearly (0). converse follows from the definition of a subring.ndcharacteristkofaring 169 Every ring R has two trivial subrings. Then the set 7(R) = (a E RIxa R) is called the center of the ring R.Sulwings. (ab)x — a(bx) = a(xb) = (ax)b = (xa)b = x(ab). Thus. Then the smallest subring of R containing S is called the subring generated by S.2 Theorem. it follows that the subring generated by a subset SofR is the intersection of all subrings of R containing S. Because 0 E S. Hence. The center of a ring is a subring. Because the two distributive laws hold in R. (0) or simply 0 and R. Write Z(R) = S. they also hold in S. Let R be a ring (with or without unity).1 Theorem. D ). Proof Let R be a ring. Hence.) is a semigroup. 0 Definition. The condition abE S for all a. A noneinpty subset Sofa ring R is a subring if and only iffor all a. Further. by Theorem 4. then S a. and the subring generated by a single element a in R consists of all elements of the form + nkak.b E S we have a — b E Sand ab E S. The following result is frequently useful: 4. A subring of R may have unity different from the unity of R [see Problem 10(b)). Proof(Jfpart). Then 0. Definition.+) is an additive subgroup of the group (R. a — 1' E S. and k a positive integer.b €5. S is a subring. the smallest such positive integer is called the . = ax for all 4.b E S implies (S.1. n1a + n2a2 + We now define the characteristic of a ring. Let S be a subset of a ring R.

n2a =0) for all a E R. a. The ring Z/(n) of integers modulo n has characteristic n. 0.x) ax ofF X A into A such that (1) (ii) (iii) (iv) (v) a(x+y)=ax+ay./JEF. The element (8 nilpotent because Io l\2 fo 0 0 If R is an integral domain. a(xy) = (ax)y = x(ay). a E F. But since F is a field. Then (n1n2)e = 0 yields (n1e)(n2e) = 0. a.y E A. x. (a/3)x=a(fJx). the characteristic ofF is max (n1 . then = 0. n2 <n.n2) <n. in a 2 X 2 matrix ring is also 0 is always nilpotent. 0 and I (if R has unity) are idempotent elements. R is said to have characteristic zero. This proves n is prime. n1 <n. Then we say that A is an algebra over F if there exists a mapping (a. An element a in a ring R is called n such that = 0.YEA. The characteristic of R is denoted char R.. either n1e = 0 or n2e = 0. .3 Proof Let n 0 be the characteristic of F If e is the identity of F. IEF. If possible. Thus. (a+/J)x=ax+fix. let n = n1is2.e. we have elements. Let A be an arbitrary ring.e. For example. Theorem.xEA./JEF.170 Rings characteristic of R. if no such positive integer exists. Definition. Continuing like this we obtain a = 0.xEA. aEF. so Definition. The element ax is called the scalar multiplication of a and x.X.xEA. But then either n1ea = 0 (i. because for = 0. =0 = e2.. the ring of integers has characteristic zero. An element e in a ring R is called idempotent if e = Clearly. a contradiction. n1a = 0) or n2ea =0 (i. then R does not possess any nonzero nilpotent a E R. 4. 0 Definition. lx=x. Then the characteristic ofF is either 0 or a prune number p. and let Fbe afield. Let F he afield.

. Also. i = 1... then ax = x.. and is of R1 for all i = denoted by® R is infinite. .. we define them here.. will not have unity even though each R. the field of real numbers R can be regarded as an algebra over the field of rational numbers Q. if R is a ring containing a subnng F C Z(R) such that F is a field.R. i = 1. then R can be regarded as an algebra over F.. 4.. These will be discussed in somewhat greater detail in a later chapter. may have unity.. It is easy to verify that S is a subring of R. If there exists a unique a E R such that xa = x. Note that if the family of rings R. R will have unity if every R..a2..b1.a2b2. is finite.) of R forms a subring of R.)—(a1+b.Subrings and cbaracterls*Ic of a ring Ill Any field Fcan be regarded as an algebra over itself by defining scalar multiplication as multiplication of elements in F.a2. Consider the set R of ((a1 ...) and (a1..).b2..).2.2 R7' of elements of the form (O.. it is convenient to call a the ith component of this element and to refer to addition and multiplication in R as componentwise addition and multiplication.)Ia.. The zero element of R is the element whose ith component is the zero Also. then direct product and direct sum coincide.) = It (a. Because of their many-sided importance and also for our present need to construct some interesting examples. E R) sequences (a.) (b.. Let R.. The rings R. .O....R1. Thus.. Consider a subset S of R such if(a1) E S..)+(b..a.2 and often designated by I1.. i = 1.b2. We now introduce the concepts of direct product and direct sum of a family of rings...2 be a family of rings.) E R. if (a.... This ring R is called the direct product of the rings R.). then® I.We define binary operations on R in the following natural way: (a1.4 Examples (a) Let R be a ring and let x E R... is then easy to verify that R is a ring...a...a2.. then all but a finite number of the components of(a1) are zero. has unity.. Moreover.. which we write as (a. We construct from this family a new ring as follows.. This ring S is called the direct sum of the family R. the polynomial ring Q[x) is an algebra over Q..O. may be called the component rings of this direct product.....O. The set 1..a2. More generally..a2.a2+b2.

(ba — e)b — bab — b — 0. 1). Solution. Then x(y + r)x — x and. for some C R. Similarly. e(ex — x) —0 yields ex — x. r E R. Then e * 0 and a(e — e2) — 0.I mapping of a finite set into itself is an onto mapping (Theorem 3. (d) Let R be a ring with more than one element. forall CE R. (e) Let R be a ring such that x3 = x for all x tive. if R has a unique right unity e. xr=O implies r='O. eisa nonzero idempo- tent. Let A — (a' E RIaa' = preimage). First we show R is an integral domain. ac 0 for every c R. Fix a0 E A and consider the mapping a' '—' a'a — + a0 ofA to A. R must be an integral domain. This proves our assertion. Thus. yx — 1 =0. Thus. we have (xe — x)e — Xe2 — xe —0. a *0. Now for all x R. Now. So a+ax—x—a and. If an element a of R has more than one right inverse. ac — a. If possible. Now xyx=x gives x(yx— I)—O. we get a —0.172 Rings In particular. Next let x e E R be a solution of ax — a. Let xr =0. y + r y. Because. x is invertible in R. Hence. let a b . Then R is commuta- .3. that is. (c) Let R be a ring with I. hence. thus. andab =0. It follows then that a does not have a two-sided inverse. then R is a division ring. A must be an infinite set because every 1. bx — c has a solution. Thus. if 0 * a R. a contradiction. This proves that each nonzero element a E R has an inverse. Thus. It can be easily checked that this mapping is I-I but not onto (a0 does not have a 1 Solution. Similarly. Solution. hence. y (b) Let x be a nonzero element ofa ring R with I. then xy — 1 — yx. Hence. so xe — x —0. by hypothesis. But this implies e — e2 0. By hypothesis A has more than one element. then e is the unity of R. We assert that e is an identity element in B. If ax — b has a solution for all nonzero a E R and for all b E R. then there exists b E R such that ab — e. R. Also. Finally. xy — I =0. ax = x. ba — e. If there exists a unique R such that xyx x. Solution. x(a+ax—x)—x. Chapter 1). Then abx 0 forallx E R. then it has infinitely many right inverses. which completes the solution. Clearly.eO.

) is a commutative ring with unity. respectively. and (x — y)3 = x — y implies xy2 — x2y (I) xyx — yx2 + yxy + y2x = 0. so (3xX3y) = (3y)(3x). Subtract (3) from (4) to get (3) (4) 2xy= Because 2yx. Hence. S has no idempotents #0. Let S = CEO. (x2 — x)3 = x2 — x implies 3x2 = 3x after simplifications. where we definef+ g andfg.1]. Consider S = (3xIx E R).y E R. 3xy = 3yx. Let a E [0. 2yxy2 + 2y2xy + 2yx = 0. Multiply the last equation by y on right and by y on left: 2xi' + 2yxy2 + 2y2xy = 0. gfE Tfor allfE Tand g ES. foryES. Thus. (2) By adding (I) and (2) we get 2xy2+2yxy+ 2y2x=0. as usual. R is commutative. Problems 1.1. Thus. y2 = y for all y E S implies S is commutative. Let 0 and I be the constant functions 0 and 1. we have xy = yx for all x. x3 = x for all x E R implies (x + x)3 = (x + x) for all x E R.Subrings and characteristic of a ring 173 Solution. Also. Show that (a) (b) (c) (d) (S. by (f+ g) (x) = f(x) + g(x) and (fg) (x) = f(x)g(x). S has nonzero zero divisors. Now (x + y)3 = x + y implies xy2 + x2y + xyx + yx2 + yxy + y2x = 0.11. + . 1] be the set of real-valued continuous functions defined on the closed interval [0. It can be easily checked that S subring of is a Rand. . Then the set T = (fE Slf(a) =0) is a subring such thatfg. 9xy = 9yx. which implies 3xy = 3yx. that is. 6x = 0 for all XE R. y2 = (3x)2 = 9x2 = 6x2 + 3x2 = 3X2 = 3x y.

Prove that the following statements for a ring R are equivalent: (a) R has no nonzero nilpotent elements. (b) Let S = (0. nilpotent elements. Let R be an integral domain and a. show that a = b. 4(n) is Euler's function (this is called the Euler— 2 Ferinat theorem). 3.n) = I. .b are halves of odd integers). Let Sbe the set of 2 X that (a) S is a ring. Show that nilpotent elements form a subring. (x. and (m.2. then a= 0. (gA matrices over Z of the form Show (b) = where * denotes some integer. 5. 8. lfa ER such that a2 = 0. (i) = (a + b a. Show that the characteristic of an integral domain is either 0 or a prime number. (b) Let e be an idem potent in a ring R.b E R. Show that the set cRc (eaela E R) is a subring of R with unity e. The set A is called the ring of Gaussian integers. 10. show that R contains a subring that is in I-I correspondence with Z (this subring is called the prime subring of R). show that a + b is also nilpotent. If atm = btm. Give an example to show that this may fail if the ring is not commutative.n) = l.then where 6. and invertible elements of the following rings: (i) Z/(4) (ii) Z/(20) (b) Show that the set U(R) ofunits of a ring R with unity forms a multiplicative group (cf.n) = I. Show that an integral domain contains no idempotents except 0 and I (if I exists).4. Furthermore.174 Rings 2. (a) Show that the following are subrings of C. Problem 18(c)). 4. (a) Ifa ring R with I has characteristic 0. show that under any of the conditions (a) or (b) all idempotents are central.b E Z a. Also find the idempotents and nilpotent elements of S.6. Show that S is a subring of Z10 with unity different from the unity of Z10. (b) 9. If a and bare nilpotent elements ofa commutative ring. 7. (c) Prove that an element E Z/(n) is invertible if and only if I (mod ii).8). Show also that if(x. (a) Determine the idempotents.

' = ya. axa = a. Let a. then is a group.e. Let R be a ring. If there exists an invertible element u R such that aua = a.Subrings and characteristic of a ring 175 11. we mean an element a E R such that the equation a+x—ax=0 (a+x—xa=0) has a solution x = b R. An element a C R is called quasi-regular if it is both right and left quasi-regular in R. Show that a finite integral domain is a division ring. Suppose R has unity.bE R. Show that a is right quasi-regular. (a) (b) (c) 19.' — ay is invertible. Let a. 15. Suppose. there exists bin R such that ab = a division ring. Let (U(R). where u = I + . Let R be a ring with unity such that for each a x R such that a2x = a..h R such that ah is nilpotent. Let e be an idem potent in a ring R. hence. Show that (S.b E R such that ha = h and a is right quasi-regular. Prove that R is In Problems 16—23. y2a = y. Show that b = 0. (b) axa — a is nilpotent and. 12.o) is a semigroup with identity. Show that if a R is nilpotent. 13.' R such that a2y = a. There exists. Show also that if ab is right quasi-regular. then I — a is invertible. show that e 0. Show that (R. 14.o). if R has a unity. Prove the following: (a) R has no nonzero nilpotent elements. where n is some positive integer. then ba is also right quasi-regular. a. 21. show that R is commutative. 18. (c) (d) (e) ax = xa. (i. for each nonzero element a in R. and a. ) denote the group of units 20. Show that ha is nilpotent. Define a circle composition ° in R by a b = a+b—ab. ax and xa are idempotents in the center of R. e. IfS is the set of all quasi-regular elements in R. show that ab = I for R there exists somebERimpliesba= 1. If x4 = x for all x in a ring R. then a is quasi-regular. Let R be a ring with left identity e. (f) aua = a. Let a be an element in a ring R with unity. Further. invertible elements) of R. 17. Let a R and let a" be right quasi-regular. . If e is right quasi-regular. by a right (left) quasi-regular element in a ring R. 16. and yE R is chosen as in (e).

Show that R is a division ring if and only if all but one element are right quasiregular.2. all invertible elements and idempotent elements are von Neumann regular)... Show the following: (a) 0 # a E R is invertible if and only if a is regular.n. i = 1. Also show that a = E R is nilpotent if and only ifeach a1. as usual. Show that R cannot be an integral domain. division ring.u2 . show that U(R2)x . (f) The set of invertible elements in R along with 0 forms a 23.. (e) Either 2 = 0 or 2 is invertible in R. (c) R has no nonzero nilpotent elements. ) becomes a ring.) = (a. that is. U(R. the group of units of the rings R and R1. Let R be a ring with more than one element.. Let R be a ring with I such that each regular element different from I is right quasi-regular.) = (a0. = . .u1). u....a.. + (a. i = 1.. the elements of R where c. (d) R has no idempotents different from 0 and I.. respectively.. Then (RIEx)).) = (c. As in a polynomial ring. n> I.2 n.Xb. 24. Show that the invertible elements of R are the elements (u1 .) denote...)1a1 E R).176 Rings 22. Let R be the direct sum of a family of rings R1 25. as in R[x]. In other words. an invertible element of R. (b) 0. 5 Additional examples of rings (a) Ring offormal power series Let R be a ring. called the ring offormalpower series over R. Call an element a in a ring R (von Neumann) regular if there exists an element b R such that aba a (for example. and let ((a. + . x where U(R). is (a1 nilpotent..). Define addition and multiplication in (b. Let R be the direct sum of a family of rings R. I # a E R is invertible if and only if a is right quasiregular..a2. aJbk.

R<x> is called the ring of formal Laurent series over R.a2..-if(a) of G into R such that f(a) = 0 for all but a finite number of elements of G.... the elements of R<x> are infinite sequences. (c) Group algebra Let G be a group. multiplication.Additional exampks of rings 177 themselves may be considered to be elements ofR[fxJJ. the elements of are generally written as (b) Ring offormal Laurent series Let R be a ring.. then R<x> becomes a ring containing as a subring. (d) Quaternions over any field Readers not familiar with the notion of a basis of a vector space may skip this.k) be a basis of Q over K. 0.. Show that if R is commutative.. and let R<x) = ((.i. are different from zero. Straightforward computations yield that R(x) isa field if R is a field.a. (fgXa) = bc—a f(b)g(c). If addition and multiplication of elements of R(x) are defined precisely as in REx] or in REIxI). Let Q be a vector space of dimension 4 over a field K. l(a)=0 =0..k. then R is contained in the center of R[G]. Define addition..0. R a ring with unity.)Ia. ER). with i a negative integer.. 0(a) l(e)=l. Show that R[G] is a ring called the group algebra of G over R. Let (I ..0.aO. We shall make Q into an algebra over K by suitably defining multiplication of these basis elements in Q as follows: . and 1 in RIG] by (f+ g)(a) =f(a) + g(a).j. In other words. As an exercise...a_k+.. and RIG] the set of mappings ai. write the multiplication table and list some properties for the group algebra FIGI. ifa*e. then RIIx]J is also an integral domain. Further. It is an interesting exercise to show that if R is an integral domain. where F — Z/(2) and G is a group of order 3..aI. with the understanding that at most a finite number of the components a.

As an exercise show that if K = C.) We extend multiplication to the whole set Q in the obvious manner. fo l\ is Io we see that Q is the same as the ring of real quaternions H. . (PT o \. This yields that Q is an algebra over K. then Q not a division ring. discussed earlier in this chapter. called the algebra of quaternions over K. If K= R.178 RIngs . then by taking Ii o\. I 1 1 ijk j i k I i —I k —j j —i —l (Recall that the product of any element in the left column by any element in the top row (in this order) is to be found at the intersection of the respective row and column.

(0) and R are ideals.CHAPTER 10 Ideals and homomorphisms The concept of an "ideal" in a ring is analogous to the concept of a "normal subgroup" in a group.bESimpliesa—bES." Theorems proved in this chapter on the direct sum of ideals in a ring and on homomorphisms between rings are parallel to the corresponding theorems for groups proved in Chapters 5 and 8. a. 179 . in a commutative ring every right ideal or left ideal is two-sided.bESimpliesa—bES. In every ring R. A nonempty subset Sofa ring R is called a right (left) ideal if (I) (ii) arES(raES)forallaESandrER. I Ideals Definition. a right ideal or a left ideal is a subring of R. Clearly. (ii) aESandrERimplyarESandraES. Trivially. Definition. called trivial ideals. The role of an ideal in a "homomorphism between rings" is similar to the role of a normal subgroup in a "homomorphism between groups. so an ideal is sometimes called a two-sided ideal. A nonemply subset S of a ring R is called an ideal of R if (i) a. Rings modulo an ideal are constructed in the same canonical fashion as groups modulo a normal subgroup. and every ideal is both right and left.

So in every case arE I. rE Z. Then S is a subring of R. and B1 is a left ideal in R. I = (c) Let R be a ring and a R. a right ideal of R. I E I. . To see this.1 Examples of ideals (a) In the ring of integers Z every subring is an ideal.1] and! = (f C RJf(c) = 0). that is. If then I is an ideal of S. then aR is an ideal of R.11 to the field of real numbers. ifr=0. the2 X 2 matrix ringoverafieldF. we have r = rI C I. (f) Let R be the ring of all functions from the closed interval [0. Let CE [0. I is an ideal ofZ. (d)LetRbethen X nmatrixringoverafieldF. (g)LetR F2. Then the subset is an ideal in R. ifa right or a left ideal Iofa ring R contains a unit ofR. so for all r R. Hence. (b) The right as well as left ideals ofa division ring are trivial ideals only. Hence. because a unit u C I implies u:r' E I. except possibly the ith. but 1 is neither a right ideal nor a left ideal I= of R. We note that a need not belong to aR. Then I is an ideal of R. If R is commutative. Then A is a right ideal. in this case aR is the smallest right ideal containing a. let I be a subring ofZ and a El. Indeed. A sufficient condition for a to be in aR is that I E R. then! is the whole ring. Then aR = (axjx C R) is R. if r<0. LetS= [lflbelhe set of upper triangular matrices over F. (or B1) be the set of matrices in R having all rows (columns). Foranyl nletA.180 Ideals and homomorphisms 1. We later show that R has no nontrivial ideals. (e) Let R be the ring of 2 X 2 upper triangular matrices over a field F. Then r times =0 —r times ifr>0. zero. and Ra=(xaIxER) is a left ideal of R.

If a> I.2 does not hold. Then the matrix El. has nontrivial right..Ideab 181 If Next we prove an important result about ideals in the matrix ring A is an ideal in the ring R.. The following theorem states that the converse is true for rings with unity..... then it can be easily shown that the ring A. x= which shows a.. as well as left.3 E A. If D is a division ring. . On the other hand. Thus. ofall a X a matrices with entries from A is an ideal in R.2 ring RN is of the form where A is some ideal in R. there exists a matrix = Then = C 1... Then 1A. we get an important result contained in 1. if we replace the word "ideal" by "right ideal" (or "left ideal") in the theorem..j = 1.1 E A and. 0 Remark. Next. then R = D. are the only ideals in the ring!).b. a55r E A.. So A is an ideal in R.3e55. let E A... This completes the proof. so a11 — b11 CA. Set A = El). proving that (0) and D. A = D.j = 1. Similarly.1 (d)1... then every ideal fin the matrix 1.. El such that Because au C A. Because (0) and Dare also the only right (or left) ideals of D. Theorem. aiieii) e35 = a. Hence..1 E = in I..2 a. ideals Isee Example 1. XE A.. and fi = an ideal.. where A is some nonzero ideal in D. x E I. in general. Then we claim A is an ideal in R. C I for each i.. D. has no nontrivial ideals. as desired. 0 As an immediate consequence. a —/1 El. If a ring R has unity. Our purpose is to show that we can multiply x suitably on the right and left to produce an element in I whose entry at the (1.. Thus. hence.2 n.. denote the matrix units in R. Proof Let I be a nonzero ideal in D. it follows that Theorem 1. I C A.3 Corollary. and. ra51 E A..1) position This will mean that all a. that is. Clearly.. Here we will show that each term a13e11 of x is in I. Proof Let (a11 E i. hence. so! = D. For let a11 . let rE R and a51 E A with E I. Let x = C land let rand s be some fixed integers between I and n.. Because A. Then there exist matrices a = save. We now proceed to show that! = A. But the only ideals in any division ring Dare (0) and D.

. Define multiplication by ab = 0 for all a. .+) has no proper subgroups. Then for all i A. a — bE A. rE R. for some a R. the following example shows that Theorem 1. by (S)1 and (S). where p is prime. and Ic R2. lfXis an additive subgroup of(R.EAA. respectively. (0) and Rare the only ideals of R.+).b ER.182 Ideals and homomorphisms Further.4 Then Theorem. Definition. are right (left) ideals.+) is a subgroup of(R. I= Definition.. Theorem 1. then Xis also an ideal. we define the left ideal and the ideal generated by a subset S. 0 Next. in general.. Let (R.+) be an additive group of order p. A.am} is a finite set.. IfS = (a1 . But since (R.. Now. Let (A. But clearly I 0.EA be a family of right (left) ideals in a ring R. Hence. for matrix rings over rings without unity. A right ideal I of a ring R is called principal if! = (a). we have xr = 0 E X and rx 0 E X. proving that flIEAA. Then R isa ring without unity..+). Letd = (AlA is a right ideal ofR containing S). Then 0 because R E Let! = Then! is the smallest right ideal of R containing Sand is denoted by (S). because A. It then follows that a nonempty subset X of R is an ideal of R if (X. Example. R. and a principal ideal. is also a rig/u (left) ideal.am).2 is not necessarily true if R is a ring without unity. In a similar manner we define a finitely generated left ideal. is a right (left) ideal.. and ar (ra) A.... Thus a — b and ar (ra) E fl. This implies I (0) or R2.since (0) or Rare the only ideals in R. because for all x E X and r E R. let Sbe a subset ofa ring R. a principal left ideal. Simi- larly. denoted..). 1.b fl. Suppose I = A2 for some ideal A in R. I i m. then (S)r is also written (a1 . The smallest right ideal of R containing a subset S is called a right ideal generated by S..2 does not hold. a finitely generated ideal..EAA. Proof Let a. it is straightforward to verify that i={(g is an ideal in R2.. A right ideal I of a ring R is called finitely generated (a1 am)p for some a.

(a)1 = (rairE R). we write m = qn + r. Thus.h C R define a b(mod I) ifa — b C I. Z and F[x] are principal ideal rings. In this case the symbols RaR. We can make RI! into a ring by defining addition and multiplication as follows: 5b = ab. n E Z). Then for any element m C I.tesurn r1as1Ir1. and let RI! be the equivalence class containing a.183 We leave it as an exercise to verify that (a)— Unite turn (a). m = qn. Similarly. We now proceed to define the quotient ring of a ring R. we define principal righi (left) ideal rings. (a)1 = {ra + flair E R. We note that all ideals in the ring of integers Z and the polynomial ring F(x) over a field Fare principal. where 0 r < n.s1 C . Let! be an ideal in a ring R. Then is an equivalence relation in R. Definition. A ring in which each ideal is principal is called a principal ideal ring (PIR). and let n be the smallest positive integer in I (note that I does contain positive integers). Then r = m — qn C I. it is also denoted by a + I. For a.bn. by using the division algorithm for F[x] (Theorem 4. so I = (n). It can be easily checked that the class of elements of the form a + x. x C I. Similarly. and thus. Hence. Let RI! denote the set of equivalence classes. If I E R. Let I be a nonzero ideal in Z. . and Ra are also used for (a). these can be simplified to (a) = I. (a)1. Chapter Il). A commutative integral domain with unity that is a principal ideal ring is called a principal ideal domain (ND). n E Z). Because n is the smallest positive integer in I. respectively.1 (a)1 = (arir E R). r must be zero. = (ar + nair E R. aR.1. and (a)1. we can show that every ideal in F[xJ is principal.

•) is a ring.+) is an additive abeian group: 1+ (5+ E) — (b-I-c) — (a + (b + c))—(a+ b) + c (1) (2) (3) (4) a+b'=a+b=b+a=S+ä Second. (a) Let(n) — (b) Let R be a ring with unity. Definition.+. This proves that multiplication is well defined. which is an ideal. Let I (x) be the ideal in R(xJ consisting of the multiples of x. then I is the unity of R/!. then the quotient ring the usual ring of integers modulo n. Then the quotient ring R[xJI! = (ñIa R). the distnbutive laws hold: (6) (7) (i + E)i = + ëã. then RI! is the zero ring. The ring RI! is also sometimes denoted by R if there is no ambiguity If! R. Further. (R/I. We show (R/!. If ! = (0). the same as the ring R by identifying a + (0). ab = cd. b — d E I. If n #0.184 Ideals and homomorphisms We can easily show that these binary operations are well defined.5 about!.Thena—cE!andb—dEL Write ab — cd= a(b — d) + (a — c)d. Examples E Z) be an ideal in Z. If n —0. . if I R. There- fore. then Z/(n) is ZI(n) is the same as Z. abstractly speaking. ) is called the quotient ring of R modulo!. a E R.letñ—Eandb—d. Hence. similarly. we proceed to_show that multiplication is well defined. we have ab — cd E 1. 1. First. Then the ring (R/L+. then RI! is also commutative. then RI! is. Let ! be an idea! in a ring R. (RI! ñ(b E) ) is a semigroup: a(bc) = a(bc) — (ab)c — (ab)E—' (ab)ë (5) Third. Because a — c. and let R[xJ be the polynomial ring over R. with a. and if R is commutative. For example.

Thus. Let x E 1. R[x]/(x2 + 1) is the field of complex numbers. Thisyieldsi3 = and Then = ±iif n is odd. then a+bx+cx2+ showing that R[xJ/I = E R}. Then —1. Therefore.Ideals 185 Solution. + E R/A C R/A — . and i is identified with 1. where is identified with a. E R/A Hence. Then Solution.fiER.andsoon.i" = ±lifniseven forsomea. (d) Let R and let A be an ideal in R. Then x 0. Let (0 Then x\E(0 Q 0 (o x \o o Therefore. Lndeed. (c) Consider the quotient ring R(xJ/(x2 + 1). if ER[x]/!. R[xJ/(x2 + 1)= R) where = —T.soingeneral. Let a+bx+cx2+" ER[xJ/(x2+l). = i.

and A = Conversely. Let g)EA}. indeed. hence. X= (0). (e) Find the nontrivial (i) right ideals and (ii) ideals of the ring R = Solution. it is easy to verify that is indeed a right ideal of R. Case 1. X = n0Z for some n0 E Z. Let g)EAforsomeaEQ}. where is identified with n. (1) Let A be a nonzero right ideal of R. X is an additive subgroup of Z. Clearly. The proof also shows that A is a principal right ideal of R. Therefore. that is. Case 2. . in this case any right ideal of R is of the form where K is an additive subgroup of Q. AC('flbZ Q Therefore. 1(n0z \0 Q 0' as claimed. We show \0 Q 0 Let a E Q be such that Let zE Z and qE Q. Then (n0z q\(n0 \0 0/ \0 0)\0 Hence.186 Ideals and homomorphisms R/A is. Clearly. X (0). n0 0. q/n0 0 EA (n0ZQCA \O 0 Trivially. K is an additive subgroup of Q. jfK is an additive subgroup of Q. the ring of integers.

] 2 v c) for all HomomorphIsms Let R be a ring. Find all ideals in Z and also in Z/( 10). (2) of R are also left ideals. Generalize Problem I to noncommutative rings with unity having no nontrivial right ideals. and ideals of a ring R = Show that every nonzero ideal in the ring of formal power series F([x)) in an indeterminate x over a field Fis of the form (xi for some nonnegative integer m. Find right ideals. (b A c) = (a v I. Find all ideals in a polynomial ring Fixi over a field F. Let R/! be the quotient ring ofR modulo 1. Section 1. 6. Let R be a commutative ring with unity.) A (a 8. (Consider A={(T where n0 and a are fixed elements in Z and Q. 7. 4.Homolnorplllsm3 187 (ii) It is straightforward to verify that the nontrivial right ideals A an additive subgroup of Q. Show that L(R) is a modular lattice (see Problem 17. Prove the converse of Problem 1. left ideals. we have an example of a ring that is not commutative in which each right ideal is an ideal. Prove that R is a field. Then there exists a natural mapping that sends a R into c R/!. b. eeL.) Problems I. Suppose R has no nontrivial ideals. and let! be an ideal in R. Chapter 5). This mapping preserves binary operations . respectively. Let L(R) be the set of all right (left) ideals in a ring R. It is interesting to note that each left ideal of R is not an ideal. 3. 5. 2. Give an example to show that the lattice L(R) need not be distributive. [A lattice is called distributive if a v a. Therefore.

Ihenf(O) is tile zero of S. Definition. As stated above R S implies S R. Thus.188 Ideals and homomorphisms in the sense that = abER A mapping with these two properties is called a of R into RI!. . Leif be a mapping from a ring R into a ring S such that (i) (ii) f(a+b)=f(a)+f(b). 1ff is 1-1. thenfis called an isomorphism (or a mono. Also. and R may be identified with a subring of S.b€R. then there exists a homomorphism g from S into R that is also I-I and onto. these rings can be regarded as the same. (iv) The set (a E Rlf(a) =0) is an ideal in R called the kernel off and is denoted by Kerforf '(0). The mapping is called a natural or canonical homomorphism.1 ofa ring R into a ring (I) If 0 is the zero of R. then f(— a) = —f(a). we also say that Scontains a copy of R.nbedding of the ring R into the ring S (or R is embeddable ins). S means that R is embeddable in S. Then f is called a homomorphism of R into S.b E R. the identity mapping gives R R for any ring R. In this case we say that the two rings R and S are isomorphic. a. (iii) The set (f(a)la E R) is a subring of S called the homomorphic image of R by the mappingf and is denoted by Im forf(R). The symbol R C. then gfis also an isomorphism of R onto T.norphism)ofR into S. and. Then we have the following: 2.I homomorphism of R onto S. R Sand S = Timply R = T. Leif R —' S be a S. In this casef is also called an e. abstractly speaking. Hence. Let us now list a few elementary but fundamental properties of homomorphisms. (ii) If a E R. It is easy to verify that if!: R —' Sand g: Tare isomorphisms of R onto Sand S onto T. respectively. If a homomorphismffrom a ring R into a ring Sis both 1-1 and onto. a. Theorem. f(ab) =f(a)f(b). We write R S whenever there is a I . we have shown that Isomorphism is an equivalence relation in the class of rings.

f(R) is commutative. Then f(a) —f(b) —f(a) Also. =f(a). For if f(a) = f(b). We can define a mapping in an obvious way (called the canonical or natural homomorphism) by the . Let N be an ideal in a ring R. if the kernel of a homomorphismf: R S is (0). Similarly. 189 Proof (i) Let a R. (iv) Let a. then Kerf= (0). (v) = Of(r) = 0. Thenf(a) =f(a + 0) =f(a) +f(O). (vi) If R is commutative. ra R. thenf is follows that f(a —b) 0.b c R.Homomorphisms (v) If I E R. (vi) f(l) is the unity off(R). f(a)f(b) =f(ab) =f(ba) =f(b)f(a).f(R) is a subring of S.b Ef-'(O)= Kerf rE R. (ii)f(0) =f(a + (—a)) =f(a) +f(—a). thus. a — bE Kerf= (0). Thus. f(l)f(a) f(a). then f(a)f(l) =f(al) is an ideal in R. Form the quotient ring R/N of R modulo N. Hence.f(—a) = —f(a). We record it this fact separately in Theorem. a — b Also f(ar) Thus. and thereforefis 1-1. and thusf(0) is the zero of S. thenf(1) is the unity of the subringf(R).f(b) Ef(R). then Ef—'(O). If a. 0 It is clear that iff is 1-I. Then Kerf— (0) if and only iff is I-I. +f(—b) =f(a — b) Ef(R). (iii) Let f(a). f(0) will also be written as 0.2 S. necessarily 1. Similarly. ihenf(R) is commutative. Therefore. therefore. ar If a c Hence. Then f(a—b)=f(a)—f(b)——0—O——0. Letf R —' S be a homomorphism ofa ring R into a ring 2.I. For convenience. This shows a = b. f(a)f(b) =f(ab) Ef(R). Conversely.

gis an onto mapping. But then cib. Then 2. so a—bEN. This showsgis 1-1. Hence. Given a homomorphism of rings f R S. g(àb) =g(a1 )g(b). Thisgives = g(a + b) =f(a + b) =f(a) +f(b) = g(a) + g(b). This is stated more precisely in the following Theorem (fundamental theorem of homomorphisms). Similarly. R/N Imf 0 As remarked earlier. Letf(a)=f(b). g is a homomorphism. there exists a unique infective homomorphism g: R/Kerf— S such that the diagram 2. R/N. we have b) 1mJ fora+N=b+Nimpliesa—.bEN.4 . Thenf(a—b)=0. The fundamental theorem of homomorphisms between rings is also stated as follows. in a certain sense. Theorem (fundamental theorem of homomorphisms). Next. we know all the homomorphic images of R.190 Ideals and homomorphisms rule Then is a homomorphism of R onto R/N because . there exists XE R such that ij(x) = £ Thus. We showgis 1-1. every homomorphic image of a ring R is of this type — a quotient ring of R modulo some ideal Also. 'i is a homomorphism of R onto R/N. Then g is a mapping of R/N into 1(a) —f(b) = 0.3 R/N= Imf Proof Let us define g(a + N) =f(a). we regard isomorphic rings (abstractly) as the same algebraic systems. Clearly. Let fbe a homomorphism of a ring R into a ring S with kernel N. It is an important fact that. if of R. Writing a for a + N. if we know all the ideals N in R. Accordingly. orf(a—b)=O. so g is well defined. therefore R/N is a homomorphic image of R.i(ab) = ab + b) = a+ b =à+ h= = + (b). In this sense this theorem states that the only homomorphic images of a ring Rare the quotient rings of R.

that is. we show that any ideal Xin Sis of the form f(A) for some ideal A in R.3. f'(A)= (rE RLI(r) E A).Homomorphisms RN/S R/Kerf I commutes. left ideals) inS.Thisprovesg—h. Becausefis onto. Also.. The context will make it clear in which sense the symbolf' is used. let f— h. where q is the canonical homomorphism. g(a+N)—h(a+N). where r is the unique element of R such thatf(r) — s. X Df(f'(X)). S be a homomorphism ofa ringR onto a ringS. We denote byf '(A) the set of all elements ofR whose images under the mappingf are in A. Then gq(a) — hii(a) for all a E R. We remark is a mapping of subsets of S into subsets of R. since f(a) X and X is an ideal. and set A Then A is an ideal in R. 1= gq. f= gi. In fact.fis a 1-1 mapping of R onto S.so Next.I correspondence from the set of all ideals fright ideals. Similarly. then the is used to denote the mapping of S onto R defined byf'(s) r. So let x X. however. ra E f '(X). For if a. thenf(a — b) —f(a) —f(b) E X. as proved in Theorem 2. there exists an a E R such thatf(a) = x. . A f '(X) is an ideal in R. so x =f(a) €f(f'(X)) as desired. Let j? R —. 0 Clearly. To show that the mapping Fis onto.b E A and r E R. left ideals) in R that contain N onto the set ofall ideals fright ideals. that is. letf be a mapping ofa ring R into a ringS. To observe that g is unique. we show X Trivially. If. a — b E f '(X) and ar E f '(I). Theorem (correspondence theorem). and let A be a subset of S. and let N = Kerf Then the mapping F: A s-øf(A) defines a 1. Ii preserves ordering in the sensethat A B jfff(A) 2. g defined as g(a + N) =f(a) is injective.5 Proof Let Xbe an arbitrary ideal in S. Then a €f'(X). and f(ar) '=f(a)f(r) E X. where h:R/Kerf—' Sis a homomorphism. Thus. For any ideal A in R it is easy to check thatf(A) is an ideal in S. Hence.

an antihomomorphism preserves addition but reverses multipli- cation. R/K be the natural homomorphism of R onto R/K. Thenf(A) cf(B). But sincef(A) x—a we cannot have A — B. we get. XE A. and no ideal of R (other than R itself) contains Mpropenly. Here it is tn vial that A Cf-'(f(A)). 0 An important consequence of Theorem 2. and ideal(x2 + l)in Rix].y E f(x+y)—f(x)+f(y) anti-isomorphism. f(A) f(B). Let R andS be rings. defined next. as before. Conversely. and f(xy)—f(y) Thus. Finally. f(A)—f(B)impliesA —B. The 1-1 correspondence between the set of right or left ideals can be established exactly in the same way. So let x Ef'(f(A)). JfK is an ideal in a ring R. S is an antihomomorphism for all x. A concept opposite to that of homomorphism is that ofantihomomorphism. Then any ideal (right or left ideal) in R/Kis of the form where A is an ideal (right or left ideal) in R containing Ker — K. let A and B be ideals in R such that A B. This implies f(x) —1(a) for some a E A.6 Let R —. shows that if an idealMin aringRis such that M # R. which we state as a theorem in view of its importance throughout our study. If f(A) = f(B). E Kerfc A. An antibomomorphism that is both 1-1 and onto is called an . Theorem. The examples are ideals (p) in Z. Such ideals (called "maximal") exist. that is. — B. whereA is an ideal fright or left ideal) mR containing K 2. Hence. Therefore. where p is prime. 0 Theorem 2. A mappingf R —. in abundance and play a very important role in the theory of rings and fields. in particular. in general. let ThenA —f-'(f(A))çf'(f(B))— B. Suppose thatA and Bare ideals in R containing N such that f(A) —f(B).6. We claim that f'(f(A)) — A.192 Ideal3 aDd We now show that the mappingFis 1-1. then each ideal (right or left ideal) in R/K is ofzheformA/K. Then f(x) Ef(A). Thus. Definition. then RIM has no nontrivial ideals. Then Kis also an ideal in A (regarded as a ring by itself). This proves the theorem.5 is the following immediate result. We return to this discussion in Section 4. and A) is (q(a) = a + precisely A/K. A = B. Hence. Similarly. A ç B.

Definition.y E R. But the only nonzero idempotent hence. Then the opposite ring ofR. Now - Kerf—(x+BIx+A =0)=(x+BIxEA)—A/B. Solution. Define a mappingf: RIB — R/A by settingf(x + B) x + A. which completes the solution.+. Solution.. for all x.+. by (R/B)/(A/B) R/A.yER. Thereforef( 1) 1. x1+A=x2+A. that the system (R. f(l)If(l)=f(1 Sof(i) f(Z) c Z. Let (R.. n times (if n> —11 0) times =f(— 1—I—. fis well defined. Hence. Solution. 2.— 1)=f(— l)+ =0 +f(— 1)(ifn <0) (ifn=O).HomomorphIsms 193 Let R — (R. Define binary operation • in R as follows: xoy=yx It is easy to forallx.7 (a) Examples The only homomorphisms from the ring of integers Z to Z are the identity and zero mappings. Let f be a nonzero homomorphism. be the fundamental theorem of homomorphisms (c) Any ring R can embedded in a ring S with unity.) is a ring. Then is 1 — a nonzero idempotent in in Z is 1. (b) Let A and B be ideals in R such that B ç A.o). In each casef(n) = n. where x • y — x.. ) be a ring. that . written is defined to be the ring (R. Let Si.+. •) be a ring. Thus. forx1 +B=x2+B implies x1 —x2EBçA. Then R/A (R/B)/(A/B).e the cartesian product ofR and the set of integers Z.+.

5.1) is the unity of S. and bJ. Z[xJ/(x2 + 1) Z[i].j) entry of'(BA) is the (j.mn). (d) Let R be a ring. Show that there exists a ring homomorphism!: Z/(m) 1. R — and = Then by the definition of the transpose of a matrix. and f(B) — 'B = Then — aj. Also show that R/N has no nonzero nilpotent elements. Define a mapping!: byf(A) 'A.RjxS—0).n) = (ab + na + mb.fis clearly 1-i and onto. and (0. (a.n E Z. Prove that + E Z). S is made into a ring by defining (a. the transpose of A. Solution. Show that the set N of all nilpotent elements in a commutative ring R forms an ideal.b E Rand m. . Now f(A • B) — f(BA) = '(BA). where the multiplication of matricesf(A) andf(B) is in the ring Assume A (a. r(S)=(xERISx'O) and 1(S)—(x€. as sets (and indeed as additive groups). Give an example to show that N need not be an ideal if R is not commutative. Recall that. The (i. (a Z/(n) 3. '(BA) 'A'B.0) is a 1-1 homomorphism.)°° Problems Find the ideals of the ring Z/(n). where Z[i] — 2. which is given by — bfk — b'kj = (i. where a.. Let sending 4. Hence.i) entry of BA.194 Ideals and homomorphisms is S — R X Z.). — for all I i.J°" (R°'%.m) + (b.f is an embedding of R into S. It can be easily checked that Sis a ring with binary operations defined above.J n. We now showf(A o B) — f(A)f(B).mXb. '(A + B) = 'A + 'B.m + n).j) entry of 'A'B E Hence. Hence. Also the mapping f: R — S defined by f(a) — (a. ito I if and only if njm.n) = (a + b. Let S be a nonempty subset of R. sof(A + B) —f(A) +f(B). (R. This proves thatf(A B) —f(A)f(B). Then (R.

andCareright(left)idealsinaringRsuchthatA cBu C. 14. show that RIM is a division ring. Also show that Let A be a right ideal of R. 16. for each r R. respectively 6. 8.) where A is an ideal in a ring R. respectively]. [r(S) and i(S) are called right and left annihilators of S. Show that = ((x.) . LetR1 andR2beringsand. 17. ideals in R. Show that a ring R with unity is a division ring if and only ifR has no nontrivial right ideals. as defined in Chapter 12. Define 1(A) = {xeRIxa€A for all aEA). Find the ideal M in each case. Also show that this is so if. Show that any nonzero homomorphism of a field F into a ring R Fbe a nonzero homomorphism of a field Finto itself. In Problem 5 show that r(S) and I(S) are ideals in R ifSis an ideal mR.0)lxE R1). R2 as rings. Show that fg is a homomorphism (isomor18. 13.B. Letf: F IfA. Show that the following rings R satisfy the property of Problem 18: that is. is I-I. either r or 1 — r is invertible. phism). Show that the intersection of two ideals is an ideal. ring R into itself. Show that R is anti-isomorphic to R°". 11. 9. p prime.y)IyE R1. 7. Let R be a ring with unity. (a) R = Z/(p2). the set of all noninvertible elements forms a unique maximal ideal M.thedirectproductofR1 and R2. are = ((0. but that the union need not be ideal.R = R1 X R2. If the set of all noninvertible elements forms an ideal M. Letfg: R R be antihomomorphisms (anti-isomorphisms) ofa 15. (b) (Such rings are called local. In any ring R show that the set A = (x RIr(x) fl B # (0) for all nonzero right ideals B of R) is an ideal in R. 19. Show that f need not be onto. Show that Let R be a ring. Prove that 1(A) is the largest subring of R in which A is contained as an ideal. 10. show thatAçBorA cC.Homomoqlitiws 195 Then show that r(S) and i(S) are right and left ideals. (1(A) is called the idealizer of A in R. 12.

..A..A2.. the intersection of all the right ideals in R containing each A. One can similarly define the sum and direct sum of an infinite family of right (left) ideals in a ring R. If A. 3.Ia.. Also.A2 3. Similarly.n.a.n) isa right ideal in R.. and. then +a.A2. we write it as A=A. C S..ja...2 n) is the of right ideals Proof It is clear that (a. Then .).. IfthesumA = LI.... if Tis any right ideal in R containing each A. 0 Notation. From the definition of the sum it is clear that the order of A. i= 1. if a. S is the intersection of all the right ideals in R containing each A. + A2 + + A.. + A2 + A + A.A2 A. is contained in S.EA. Thus. in a ring R is denoted by A.. is immaterial.. then obviouslyT S. ... Then the smallest right ideal of R containing each A.EA... is called the sum of A. A.. Theorem. I I n. Although no extra effort is needed to talk about this.A2 A. We write A= of right (or left) ideals in a ring R is called a direct sum (leach element a E A is uniquely expressible in the for. then a1=a1 in S..196 Ideals and homomorphisms 3 Sum and direct sum of Ideals In this section we discuss the concept of the sum of ideals (right ideals. where we discuss sum and direct sum of a more general family...nilyofright ideals in a ringR. i= 1.2 Theorem.... LeIA... is a direct sum.n 17_.. 1 i n (that is.LetA1 the following are equivalent: right (or left) ideals inaringR. left ideals) in a ring R.'s in A. The sum of right (or left) ideals A1 .. + a2 + + a. are right ideals in a ring R. we prefer to postpone it to Chapter 14. i 2. C A.. hence. each A. a.A.®A2® Note.. beafa. Definition. Further.. E A..1 S(a1 +a2+ A1.

Then O=(a1—b1)+(a2—b2)+"+ (a. Similarly....... l.n. Then by (ii) we get x =0..a.=0. Let R? .. and let R = R1 X 3. Then R — rings...._1.R2 . Then (ii) (iii) Let x E fl x— a1+ Thus. (ii) ff0— E7_1a...0. Let 1...0)(a1... Hence.. fl Proof (I) : (ii) Follows from definition of direct sum. a2 b2._1+ a.n... R a R.a..a2... x =0. be afamily of rings.. is a direct sum of ideals R?. Proof Clearly..). fR = ®17.2....R.J.....0.2. i 1. and R? R. a. hence. Let j*.... R xERrfl Then x=(0... as R.3 E R2 X" X R. = b. be their direct product. a1 — b. Let where a=a1+a2+'"+a.. = 0 and. i= (0).2..'s considered as rings R A1 X A2 X on their own right. This gives a....+1. A. 0 Theorem. then X A.n.. On the other hand. A is a direct sum.Sum and direct sum of Ideals (I) A— is a direct sum.. and a=b1+b2+"'+b. adirect sum of ideals ofR.. (iii) then a.. .a.. — b. This gives a1 —b1 EA1 Hence.0.. the direct product oftheA...IA. Therefore.a. .a. i='l...).

This remark then immediately yields that f(xy) =f(x)f(y). thenf(x + y) =f(x) +f(y). E A.. then x can be uniquely by isdirect. then I — I. (a)... In the tatter case the notation A1 ®A2 0. fl 4 (0). let Jbe a right ideal of R such that 0 # IC I. is also frequently used. C I. Let I be a ajjejj E I. By assumption. if! is a right ideal of R with the property that each nonzero element generates!..EA. Definition.inR.2. It is also clearthatf is both 1-1 Because and onto. 0 A.f is an isomorphism...IA. Then nonzero right ideal of R contained in e..1R. A right (left) ideal fin a ring R is coiled minimal jf(i) I # (0).n. For the second part we note that ifxE R.1A. If R is a division ring. i — 1.. To see this. Let 0 . It is clear that if! is a minimal right ideal.. if x. Also. Suppose 0 # a E J.is also call ed the (internal) direct swn of idealsA. so I — I.198 Ideals and bomoinorpldsns expressed asa1+a2++ a. The context will make clear the sense in which the term "direct sum" is used. A nontrivial example of a minimal right ideal is I a1. then R itself is a minimal right ideal as well as a minimal left ideal..a. Indeed. (external) direct sum of the family of rings A. Hence.. 0 Thedirect sumR — ® 17.f is well defined.y ®17.b. To show we need to note that if a1 + since a.. then I is generated by any nonzero element of!. 0 e11R— : 0 : aGED 0 where R D. then! is minimal. and (ii) if I is a nonzero right (left) ideal of R contained in I... is the n X n matrix ring over a division ring D.

b B..x + B). • (i) A+B B A AflB B A (ii) A+B A+B A+B AflB R B A B A More generally.b + B) — (x + A. Then the mappingf: A —.4 Examples (a) For any two ideals A and B in a ring R. Also. We show that g is onto.j n. Cf C e11R. (A + B)/B. This gives = (jaijeij)(ar. a€A. bE B. 1 so exists. (ii) Note that A and B are also ideals in A + B considered as a ring by itself. The mapping g:A+B—(A+B)/AX(A+B)/B. Then g(b) = (b + A. showing that e11R is min- imal. Therefore. each element . Solution.O + B) E (A + B)/A X (A + B)/B. Hence.Ie. J= e. given by g(x) = (x + A. by the fundamental theorem of homomorphisms. is clearly a homomorphism.O + B). (i) It is clear that B is an ideal in A + B and A fl B is an ideal in A. a A. Hence. Thus. is clearly a homomorphism whose kernel is A fl B. x E A + B. thenf(a) = a + B x + B. ifA1 such that A. It is easy to check that Ker f= A fl B. +A1= R for all 1 R R A1 is a family of ideals in a ring R with unity then i. Let (x + A. 3.f is onto.i)EJ. R A2 flA. given byf(a) = a + B. A/(A fl B) (A + B)/B. if x+ Be(A +B)/B and x= a + b. Thus. Write x = a + b.Sumanddlrectsumofldeals 199 for some!. i R A.1R.

O + B) has a preimage in A + B by the mapping g. g is an onto mapping.y + B) have preimages. we assume the result is true for n by (ii) —1 ideals. J#' R=A. Z/(n) where n — X Z/(pp) X X Z/(pr).200 Ideals and homomorphisms of the type (x + A. Hence. Then R R n—i R A" R A R A flAt n flA. To complete the proof. are distinct pnmes.XAI so +fl4)c A. Z. i .+flAj. Then R = R2 = (A. p. Then by the fundamental theorem of homomorphisms we get A+B A+B x A+B AflB By (i) A B A+B A proving (ii). + 4. Becausep. WeshowA. (p7'). (b) For any positive integer n. k. Similarly. B AnB and A+B B A AflB' To prove the last part we first prove by induction that "Ii' Assume that this is true for n — I ideals. LetA. elements of the type (0+ A. +A. I *j.. we can find integers a and b such that Solution. 1 and p1 are relatively prime.

A is a direct sum of right ideals i n. implies e. e. each term is zero. 2 i n. I i n.) e. — 0. 1 (c) Let (e."j. + A1— i. and let (As). + "+ e.a.EA. Because 1 E R. I k. I I n.. be a family of idempotents in a ring R such that orthogonal family of idempotents if Solution. x = xapr + xbp. be a family of right Then there exists an orthogonal family of idempotents (e. Suppose R = A1 ®® Solution. I e1R. XE (n). Next.).'e(p7') + (p'). + the left by e.R..R. by I I n.. let x E This shows • Then a. Similarly.. we get 0= e. Then 1 e. = This yields But then by Theorem 3.Thenx= l. A.).'=O. Thus. Hence. Let 0 i n. Theorem 3. letX E A.. LetxE (n). 1 #j. such that A. = e.2.. that is. is called an = 0. + e. e1e.k.a1. Then forallxEZ.EZ. Then e. (d) Let R be a ring with unity.Sum and direct mm of ideals api'+bpy= I. + •+ + .2.. On the We further show (n) Recallthatn "pr. Because the p. 1 ideals of R. I i n. e.j I. ea.'s are all distinct. the solution is then complete.). E A. Multiplying both sides on + e1 a.. This proves that (n) — By example (a). Hence. E e.2. E Z.R CA. = flAt. we can write e. (A family of idemopotents (e. and thus XE This implies x E (0) for all I other hand. we get n = divides x.

1R (Re. SoA1 =A +M1 isadirect sum.n. thenthereexists an element x E R such thatx — x. + proving that A.. 3. M1 (after renumbering if necessary) such that M1 But then M1 flA —(0). = (0) or A fl M. i 1. Then every right (left) ideal A in R is of the formeR (Re) where e is an idempotent.. A = eR... say that contains all M. and R2 each with unity.. Denote by f the projections of R onto R.x.2.. Ifeis any idempotent in R show that R—eR®(I —e)R isa direct sum of right ideals. LetR=D. are minimal right ideals. Showthatifx1 ER.. BecauseA fl M. given byf1(a1 . Solution. respectively. be ideals in a ring R with unity such that A1 + = R foralli #j.202 Ideals sad bomomorphlsssa = e1 x + . 6.A. lfA1 R. By hypothesis R = M1 where M. (e) Let R be a ring with 1 such that R is a finite direct sum of minimal right (left) ideals. By continuing this process we must come to a right ideal..a2) — andf2(a11a2) = a2. x = e. x e1R. = e. where A1 and A2 are ideals (right ideals) in R1 and R2.A2. Let (efl).) 2. ç M. as before. say.R. Show that e. (This is called the Chinese remainder theorem.. Prove the following: (a) Eachf is an onto homomorphism. are minimal right ideals in R. (mod A. Let .ByExampIe (d). say Al2.) is a right ideal (left ideal) in R. R. then there exists some M. I = l.betheringofnXnmatricesoveradivisionringD.Thus.. Because A R.) 5. then (1 — e)R denotes Let R OR2 be the direct sum of the rings R. ShowthatA +A=AforanyidealAinaringR. 0 M2. (b) If A is an ideal (right ideal) in R. such that A1 (1 M2 —(0) and A ç A2 = A1 0 M2 — A 0 M. (In case 1 theset(a—eaIaER). and therefore coincides ®® withR. Show directly that Z/(12) Z/(3) X Z/(4). Assume thatA # (O)andA R. either A fl M.andM.. .. (c) The ideals (right ideals) in R are precisely those ideals (right ideals) that are of the form ®A2. gives x which gives. Hence. there exists.. 4. Problems 1. I i. then A=f1(A)®f2(A). j (a) be the set of matrix units.) for all 1. = M. n.thereexistsarightidealBsuchthatR=A ®B..2..

then A + B Z. A.1 Theorem. Similarly. for all ideals X A.6 the ideals of R/A are of the form B/A. B = R.e2 are positive integers.R is a direct sum of right ideals e. This implies that (A + (x))/A is a nonzero ideal in R/A and.Maximal aad prime IdeaM (b) (C) Let A be a nonzero right ideal of R contained in e. Show that RXS R S AXBAXB. LetRandSberings. Clearly. where p1 'P2 are distinct primes and e1 . R = ®17. +(x)—R.4(e)]. Then A — e11R.R. show this for left ideals. is comaximal.. hence. (ii) The quotient ring R/A has no nontrivial ideals. ideal. for left ideals. Re. the following are (i) A is maximal. Then A + (x) is an ideal of R properly containing A... B in any ring R are called comaximal A +B=R. Definition. If A is right (left) ideal of R. (iii) ForanyelementxER.. Two idea/s A.R. If A — (pt') Definition. An ideal A in a ring R is called maximal if if (I) A # R. A + (x) — R. Therefore. . Such ideals are called comaximal ideals in Z.1e. If B/A is not zero. 4. therefore. (ii) (iii) Let x E Rand x A. then B A and. For any ring R and any ideal A equivalent: R. and (ii) eitherB=A orB=R. Proof (i) where B is an ideal in R containing A. then A — eR (Re) for some idempotent e E R [see Example 3. an ideal A is maximal if and only if the pair X. respectively. the only ideals of R/A are R/A and the zero (ii) By Theorem 2. Also (d) 7.andletAandBbeidealsinRandS.1. 4 MaxImal and prime ideals respecand B = (pr) are ideals in Z generated by tively.

To prove the converse. for any ideal M. Proof Suppose R is a nonzero commutative ring with unity. But ifn is composite. so Mis a maximal ideal. let a E Z such that pf a. then the set M={[g ob]It. An immediate but important consequence of Theorem 4. RIM is also a ring. Let 0# 5E R RIM. where n E Z.y. aR R. Then aR isa nonzero ideal in R.b E F.sayn = n1n2. n is prime.1. Then Z/(n) is a field. x A. a commutative simple ring with unity must be a field (Problem 1. This implies ii + which proves that Z/(p) is a field. Then there exist integers — 1.11i12 — OinZl(n)andO # ñ1. Recall that each ideal in Z is of the form (n). So ifK Mis an ideal in R. where p is some prime. Hence. Section 1). K R. IfMis a maxmal ideal. the nonzero ideal K/Mm RIM must be RIM. so — 1. Conversely. where n is a nonnegative integer. x. then. Suppose (n) is a maximal ideal. (n) = (— n). in a nonzero commutative ring with unity.EF} . This shows that R is a RIM be a field. (b) If R is the ring of 2 X 2 matrices over a field F of the form [3 a. Now 0 Definition. let R — 4. RIM is also a commutative ring with unity. n> I. by Theorem 4. E Z such that ax + py — 1. 0 # p12. If B A.3 Examples of maxima) ideals (a) An ideal M in the ring of integers Z is a maximal ideal if and only if M = (p).This implies that there exists b e R such that ãb =1.1 is 4. Indeed. Hence.2 Theorem. choose an element x E B. An example of a noncommutative simple ring is the n X n matrix ring over a field F.204 Ideals and (iii) (1) Let B A. which is a contradiction because ZI(n) is afield. an idea/Mis maximal if and only if RIM is afield.then. Because R is simple. A ring R is called a simple ring 41 the only ideals ofR are the zero ideal and R itself A field is clearly a simple ring. clearly. 0 field. Then. So we can always assume that any ideal in Z is of the form (n). Then R has no nontrivial ideals.wheren1> 1.n2> 1. Now.

Definition. then their product AB is an ideal mR. Let Then S is a subring of R that is isomorphic to F under the mapping '—' a..AB+ACA(B+C). and C be right (left) ideals in a ring R. cE C. it follows that M is a maximal ideal in R. 0 (ii) Clearly AR. Hence..4 Theorem. if aEA. Clearly. Thus. bE B. Also. then theirproduct is defined 1fA1 inductively. Then the set I. Because associativity [Theorem 4. is called the product ofA and B (in this order) and written AB. Let A. where 4o is ra bl Ia 0 o an onto homomorphism whose kernel is M. It may AI. Because S is a field.. 4..4(i)J holds. finite sum ) which is easily seen lobe a right (left) ideal in R. . the product of any parentheses). (ii) A(B+C)AB+ACand(B+C)A"BA+CA. Note that if A and B are ideals in R. Let A and B be right (left) ideals in a ring R. a(b+c)=ab+acEAB+AC. by the fundamental theorem of homomorphisms. Further.Maximal and prime Ideals is 205 a maximal ideal in R. B. M is an ideal in R. Then (i) (AB)C — A(BC). RIM S. the mappingf: A —' 5.AA may be written asA1A2 ' be verified that finite sum (. ACCA(B+C). then are right (left) ideals in a ringR. (i) Follows from the associativity of multiplication in R.

(aXb) ç P. Suppose A M. a E B. But then by our hypothesis b P. Theorem. Suppose A P. IfR isa commutative ring. then an ideal Pin R is prime (land only (lab E P. each element of (aXb) is in P. then we show P is prime. Thus. Hence. in general. This suggests the following Definition. Now (na + arXmb + bs) — nmab ÷ nabs + mabr + abrs.206 Ideals and homomorphisms A prime integer p has the fundamental property that ifplab. This implies a P or b E P. Then A + M R. Clearly. and let A and B be ideals in R such that AB ç M. But since P is prime. then a E (p) or b E (p). then A ç P or B ç P. (a) ç P or (b) ç P. this property may be restated as follows: If ab E (p). Because ab P and P is an ideal in R. For each prime integerp. In fact. the ideal (p) in Z is a prime ideal. Choose an element a A such that a P. where n. 0 . Proof Let M be a maximal ideal. Then AB ç P implies aB c F. then each maximal ideal is 4. in any integral domain (0) is a prime ideal. Hence. We know (aXb) consists of finite sums of products of elements of the type na + ar and mb + bs. we get (na + arXmb + bs) or finite sums of such like products are in P. b E R implies a E P or b F. b E R implies a E P or b E P. then orpib.BçM. Conversely. Equivalently.s R. Then b= ab+ mbEAB+MçM. Hence. a commutative ring R is an integral domain if and only if(0) is a prime ideal. But the converse. If R is a ring with unity. mEAL Let bE B. if (aXb) ç (p).5 prime. Let A and B be ideals in R such that ABc P. Write 1 = a + m. if ab P. This proves M is prime. then (a) ç (p) or (b) ç (p). The ideal (0) in the ring of integers is prime but not maximal. a E B. 0 For commutative rings we have 4. B C P.m E Z and r. An ideal P in a ring R is called a prime ideal (fit has the following property: IfA andB are ideals in R such that AB ç I'. where a and bare integers. Proof Let ab Pand Pbe a prime ideal. is not true. a E A.6 Theorem. In terms of the ideals in Z. so ab E P for all b E B.

then (aXa) = (a2) because R is commutative. RIP = (0) implies R = P. Suppose P is maximal in the family F of all ideals in R that are disjoint . Then each prime ideal P # R is maximal.)..Maximal and prime Ideals 207 4.R be a commutative principal ideal domain with identity. However.we get I = ax.. Hence. aR ç hR implies a = bx. Let . Thus. ñb= 0) implies a E For bE P (i. Hence. This implies a E (a2). Consider RIP. because R is a domain. So RIP = 1 }. Next. Thus. Hence. ab —0 implies a =0 or b = 0. a is invertible. RIP = (0) or {O.1}. Solution. Thus. So RIP is also a Boolean ring in addition to being an integral domain. (a) = or (b) =0. there exists an elementxE Rsuchthata — a2xora(l — ax) O. Then some b a bx = bay implies I = by. But since (0) is a prime ideal. that is. then (aXb) (0) (0). M = R. a contradiction.7 Theorem. a2. Hence.. R is an integral domain. i= 0 or b — 0). Then R is a field. But we know that an integral domain has no idempotents except 0 and possibly unity. (b) Let R be a Boolean ring.e. that is. Then x= ayfor some y E R. Then any nonzero ideal P R is prime and only jf it is maximaL Proof Let P = aR be a nonzero prime ideal in R. which is not true. and b aR. So RIP is an integral domain. and. Now for all x E R. 0 4. But (a2) is prime. The converse follows from Theorem 4. P is a maximal ideal. x E R. we get (a) = (a2). which obviously forms a field. Because (a2) is trivially contained in (a). (c) Let a be a nonnilpotent element in a ring R. a3. (x + PXx + F) — x2 + P — x + P. 1fF is not maximal. if a E R. Therefore P is maximal. R is a field. hence. and let S = (a. Now. Solution. (a) ç (a2).. then there exists an ideal M such that P M and M R.Ifa 0. Because P—aR is a prime ideal and aR. Because abE P(equivalently.8 Examples (a) Let R be a commutative ring with unity in which each ideal is prime. If ab =0. M = bR for R. we havexE aR.5.

let A P and B P. is a maximal ideal of CEO.. is a field isomorphic to R.suchthatM—M.f(z) 0 for any z E [0. Let Jbe an ideal of CEO.1] — UXE(olJJX. Solution. If possible.. I] be the ring of all real-valued continuous functions on the closed unit interval.andconversely. But a *0 implies that a is invertible. that is. Then h — g — a is such that h(r) 0. Then g(r) '0' 0. J = R. It is easy to check that M. exists an open interval. I ]/M.) XE Fsuch Solution.—(fERjf(r)—O). there exist positive integers Then iandjsuchthata'EA +Fand a'aJE(A+PXB+P)-'AB+AP+PB+PPcP. Let g(r) — a. we may proceed as follows. Then Pis a prime ideal. we show that M.1]. Suppose thatf(z) Then impliesthatthere '0' 0. Let AB ç P. it means that there does not exist any ideal P. . By the Heine—Borel theorem in analysis of this family of open there exists a finite subfamily.1) is continuous. is an ideal. Otherwise for each x E [0. Let M be a maximal ideal of CEO. Now k n). Thus. Precisely. so a — g — h E J.!). By maximality of P. Consequently. Letg E J. we note that CEO. which is not true. I).1]. We claim that there exists rE [0.1] such that. [0.Or l.1] IL. which proves the converse. say. Conversely. IfMis a maximal ideal ofR. say I. To see that it is maximal. such Clearly. Because there exists E M such that #0 for all yE I.1). there 0. (Note that the statement does not say that P is maximal in A.f(r) 0. Thus. (d) Let A — C[O. 0 and (B + P) fl S (A + P) fl S 0. f(z) —0 —0 a contradiction. —0. Thus. where A and B are ideals in A. P implies either A ç P or B ç F. which in turn yields that f is invertible and a fortiori M — CEO.. Considerf— [0. which proves that P is a prime ideal. 1) properly containing M. because ABçPand because Pisan ideal mR. for allfE M. h E Mr. then there exists a realnumberr. Alternatively. g M. I) for any r E [0..208 Ideals and from S. Then A + P P and B + P P. But 0 for some z in [0.

] 5 Nilpotent and nil ideals Let A2 be right (or .2) is not of zero. It has a nonzero ideal A = (8 such that A2 — (0). but A" = (0) for some positive integer n> I. Show that the ideal (x3 — x — I) in the polynomial ring Z/(3)[xJ over the field Z/(3) is a maximal ideal and therefore. As another example.) A ring R may have nonzero nilpotent elements. [Hint: Use Example (c) and assume that a maximal member in the family F exists. Let Pbe a prime ideal in R.Nilpotent and nil ideals Problems Let Fbea field. Show that RIP is an integral domain. consider the ring R = 2 X 2 (upper) triangular matrices. every zero ideal is a nilpotent ideal. Show that (0) is a prime ideal in RIP. Let (P. but A2 = (0). Such right (or left) ideals are called ni/potent. . left) ideals in a ring R. For example. Forall left idealsA.. then their product is written as It is possible that a right (or left) ideal A is not zero. I. but it may not possess a nonzero nilpotent ideal.) Let Pbe a prime ideal in a ring R such that the quotient ring RIP has no nonzero nilpotent elements.BinR. (Indeed this set may not even be an ideal.aRbçPimpliesaEPorbEP. 7.EA be the family of all prime ideals in R and let A = fl. Show that the following are equivalent for any ring R..ABç PimpliesA ç PorB ç P. 2. 5. prime. A right (or left) ideal A in a ringR is called nilpotent (0) for some positive integer n. BinR. (A ring S is called a prime ring if(0) is a prime ideal in S. If A1 = A (say). The existence of a maximal member in F follows by Zorn's lemma. (a) (c) (d) (b) Foralla. However. as the following example shows. 6. Prove that the ideal (x3 + x + 1) in the polynomial ring Z/(2)(xJ over Z/(2) is a prime ideal. Definition.EAPI.bER. ForallrigbtidealsA. the set of nilpotent elements in a ring R does not necessarily form a nilpotent ideal. 3.). Prove that the ideal (x4 + 4) is not a prime ideal in the polynomial ring Q[xJ over the field of rational numbers. Clearly. and every element in a nilpotent ideal is a nilpotent element. given in Section 6.ABc PimpliesA ç PorBç P. PisaprimeidealinR. 4. Prove that (0) isa prime ideal in Fm. in the ring R = Z/(4) the ideal A = (0. Show that A is a nil ideal.

. a contradiction.. has no nontrivial ideals (Corollary 1. Hence. So I is not Show that if A and B are nilpotent ideals... Therefore..).p + (p2).. 3. But Xc nilpotent. Hence. such as eu.I—(0).0 + is nilpotent.p + (pk+I).A2. the converse is not true."j.j n. However.. RI is a nilpotent ideal in R. i .O + (p3). show that the sum of nil ideals 1.0 + (pk). 6 Zorn's lemma A partially oixlered set (also called a poset) is a system consisting of a nonempty set Sand a relation.a— IaERI=(O).. Let R = ® I — 1. Let I be the set of all nilpotent elements. So I is a nil ideal.. RI (0) or RI R. then the element 5. every nilpotent right (or left) ideal is nil. over a field F has no nonzero nil right or left ideals. Then consider the ideal 5. usually denoted by such that the follow- . Show that the n X n matrix ring F. Then R contains nonzero nilpotent elements... Definition.2 = (0 + (p). be the direct sum of the rings Z/(p').210 Ideals and bomomoq*Isnis Example. for if P =0 for some positive integer k> 1. k—ltimes ktimes Hence. is a nil ideal.. where k is some positive integer.1 1 ktimes -. But RI cannot be equal to R.. Let I be a nilpotent right ideal in R with 0. But then foranyaEl. i.. Let R = F. p is prime.. Clearly. so it belongs to I.3). 0 + (p2).. But we know that the ring R = F. as the following example shows. A1 .A.... Example. since R has unity 0. 2. be the ring of n X n matrices over a field F. RI = (0).2. Problems 0. But I is not nilpotent. Then R has nonzero nilpotent elements. Then I is an ideal in R because R is commutative. A right (or left) ideal A in a ring R is called a nil ideal if each element of A is nilpotent. If R is a commutative ring. such as (0 + (p). their sum A + B is nilpotent.

Hence. then there exist elements A and B of S such that a E A and b E B. ar U. N = R. For if a. as desired.b. a (reflexive).Zorn's lemma ing conditions are satisfied: For all a. a contradiction to the maximality of M in S.) isa subset of S such that. so I EXforsomeXEC. Zorn's lemma. c ) is a partially ordered set under inclusion. Then by Zorn's lemma S contains a maximal member M. then clearly NE S. Let S be the set of all ideals in R such that if XeS then X R and X I. implies N= Mor N= R. Similarly.Also. a — b Bor a—b€A. and ifM ç N. which is not true.impliesm=a. So let M N. b and b c a either a b orb a. a a a b and b a - a b (antisymmetric). Hence. then (S. accordingly. racBorraEAforallrER.1 Theorem. If U = R. then there exists a maximal ideal M of R such that Proof. then I EU. an element m S is a maximal element of a poset (S. we prove the following result. (S. U E S. where Nis an ideal in R.c (i) (ii) (iii) S. and thus Mis a maximal ideal in R.a—bE U and ra U. we haveA ç BorB ç A. U is an ideal in R. The following axiom from set theory about posets is widely used in mathematics. 6. But since Cisachain.b U. We claim that Mis a maximal ideal in R.a€S. Hence. 0 Problem Use Zorn's lemma to show that every proper right ideal A in a ring R with unity is contained in a maximal right ideal M. As a simple example of the use of Zorn's lemma. If N # R. for every a. An element u E Sis an upper bound of C if a u for every a E C. and. U# Rand. C.) has a maximal element. (M is called a maximal right ideal ifM R.Jnanycase. thus.b ma. and I is an ideal in R such that I R. We assert that the union Uofan arbitrary chain Cm Sis an element of S.s) if A chain C in a poset (S. If R is a nonzero ring with unity I. c (transitive).) . for some right ideal N. If every chain C in a poset has an upper bound in S.

a and b are associates. an element u R is a unit if and only if u is a divisor of 1. Of course. and. Definition. and we write bla or a 0 (mod b). Then b = byx. where v — This means that such that a — bu. this gives yx — I. hence. Because R is an integral domain with 1. then the converse is also true. Definition. Such a ring is also called a domain. 212 . Clearly. a — tic. Two elements a. In fact. and (ii) then plaorplb.R av. Therefore x and y are units. If a and bare nonzero elements in R. and (1 i) every divisorofa called an irreducible element is improper: that is. then integral domain. if R is a commutative if a and bare associates. implies either b or c is a unit. then b and bla. For let b = ax and a by. a. then a and bare associates. that is. we say that b divides a (orb is a divisor of a) and that a is divisible by b (or a is a muizipleof b) if there exists in R an element c such that a = tic.b in R are called associates if there exists a unit u E .CHAPTER 11 Unique factorization domains and euclidean domains 1 Unique factorization domains Throughout this chapter R is a commutative integral domain with unity. We call an element bin Ran improper divisorof an element a R jfb is either a unit or an associate of a. A nonzero element p of an integral domain R with unity is calleda prime element tf(i) it isnota unit. ifaib and bla. A nonzero element a of an integral domain R with unity is it is not a unit.bER.c E R. b.

cR = R.so pR + aR — R. pjb because plab.3. Definition. Because R is a PID. so pR + aR = pR.b E R. Define a function N: R .1). which contradicts the assumption that p Ja. Proof Let R be a PJD.2 Examples of UFDs (a) The ring of integers Z is a UFD.y E R such that px + ay = 1. (b) The polynomial ring F[xJ over a field F is a UFD. which implies p = cd. Then there exists x. Then p E cR. as will be shown in Theorem 1. c is a unit. Then pR = cR. there exists c E R such that pR+aR=cR. c or d must be a unit. But an irreducible element need not be prime (see Problem 5). The factonzation into irreducible elements is unique. therefore. (ii) Every irreducible element is prime. Hence. An irreducible element in a commutative principal is always prime. Suppose d is a unit. 0 We now define an important class of integral domains that contains the class of PIDs. A commutative integral domain R with unity is called a unique factorization domain (or a UFD) if it satisfies the following condilions.Unique factorization domains 213 It is quite easy to verify that a prime element is an irreducible element.1 ideal domain Theorem. E Z) is not (d) The commutative domain R = (a + Z such that a UFD. However.. we have the following 1. and let p E R be an irreducible element. But since p is irreducible. Suppose plab. a. (C) Each commutative principal ideal ring with unity that is also an integral domain [called a principal ideal domain (PID)J is a UFD (see Theorem 2. as shown later. Assume p Ja. This implies pbx + aby b. d E R. 1. (i) Every nonunit of R is a finite product of irreducible factors. This gives a E pR. for let r — a + b S E R. Hence.

the theorem states that if a q1q2 Pt P2 q. Now consider the factorizations 9 3 3 (2 + (2— claim that 3. and after cancellation of the common factor p3. the proof is complete. so a2+5b2—l. one of which involves exactly s + I factors. which is impossible for integers a and b. and on renumbering the q1.. then N(r)N(s) — N(rs) = N(I)= I.. by the induction hypothesis. So 2+ is irreducible. Thus. IfR is a UFD. hence.2. 2— and 3 are irreducible. We then prove it is true for any element that can be factored into s + I irreducible factors. (iii) N(rs) = N(r)N(s). Proof More explicitly. divides the product of the p's. Hence. by axiom (ii) for a UFD. 1. the only units are ± I. This gives 9 — N(r)N(s). then m = n. Let 2 + and 2— = rs. This gives a—± I.. Because and arc associates.m. we assume it is true for any element of R that can be factored into s irreducible factors. i = l. In particular. Let a—flp1—flp (I) be two factorizations of a into irreducible factors. N(r) is called the norm of. the only associates of an element are the element itself and its negative. we have that and q1 are associates. R is not a UFD in view of the next theorem. where u is a unit. Hence.andinthelattercasesisaunit. = Because the theorem is obvious for factorizations of irreducible etc. 2 + are all irreducible.3 Theorem. Then up3. and has the properties that (i) N(r) 0. then thefactorizazion ofany element in R as afinite product of irreducible factors is unique to within order and unit factors. Because we already know that p3 and differ by a unit factor.Intheformercaserisaunit. Thus. (1) gives (2) Hence. b=O.. N(r)— 1. It follows from property (iii) that if rs I. Similarly. Therefore. Then N(2 + N(r)N(s). and. p1 must divide one of the elements Let p3 divide is irreducible. (ii) N(r) = 0 iffr — 0.. N(r) = 1.214 Unique factorlzatlon domains and euclidean domains N(r) = ri= a2 + 5b2. We have that p. the two factorizations in (2) can differ only in the order of the factors and by unit factors. either N(r) — I or N(r) 9 and N(s)= I. ments.. or 9. where p1 and q1 are irreducible. 0 . 3. But if N(r) = 3 and r = a + then a2 + 5b2 = 3.

and (ii) (ffor c R.m. therefore. 1. Then did' and d'Id.b) = c to mean that (a. and d = Then . ceR.. Now suppose dand d' are the two greatest common divisors of a and b. But then cia and cjb i = 1. b€R. Proof Let us write we mean where p1 are irreducible.. We write (a..b) Problems In the following problems R is a commutative integral domain with unity in which for each pair a. Then there exists a greatest common divisor of a and b that is uniquely determined to within an arbitrary unit factor.. min(e1.. 0 The uniquely determined (within an arbitrary unit factor) greatest common divisor of a and b is denoted by (a..3. and A1 A.cb) are associates.d) defined as follows. dia and dib. Show that c(a. clearly.b).. In a UFD two elements a and b are called relatively prime if I.b) is a set in which every two elements are associates. I = 1. g.b E R. b. and cib. then abjc. by Theorem 1. Definition.. I. then cid. Definition.Unique (actorizatlon domains has In the next theorem we show that in a UFD any pair of elements a and b a greatest common divisor (g. they are associates because R is a where A. Let R be UFD. (a.c. = min(e11f). i = 1. Let a. Show that if(a. e1 ..b) consists of all unit multiples of c.2.j are nonnegative integers. and Set g. This proves cfd. cia. are nonnegative integers. Then cia implies a unit. (a. 1 . as desired. 2. and a...2. Let c R be such that cia and cib. Note that (a. commutative integral domain.c..m... imply. An element d in an integral domain R is called a greatest common divisor of elements a and b in R (i) dia and dib.2.b) and (ca.b) = and if aic and blc.f1) = g.m. and.d. that A1 e. Hence. b) exists.4 Theorem.

then blc.c.d. Also.c) 1.d. A is an ideal in A. 5..R. we are done. 1. A = aR for some a R. 10. 2.beZ. aIR. is irreducible but not prime in Prove that 2 + 1 6. Show that the units of ±1 ± 2 are ± 1. Because R is a principal ideal ring. So let b not be a product of irreducible elements. Therefore. 8.b€A. of 4 and 2 + Show that in the ring does not exist. but a UFD is no: necessarily a PID. a A. if it exists. = +b Ja. Next..216 Unique factorlzatlon doniahn and eudidean doinalue 3. 7.. a Then a— bE a. we are done. and a.RCA.c. Show that 3 is irreducible but not prime in the ring Show that in the ring I— = (1 + the g.b are both even or both odd}. akR = ak+ 1R = . let a1 R C a2RCa3RC beachain of ideals in R. b E for some i. Show that if 1 and (a.b) = and blac. So let a be. Show that if(a. of 10 + lii and 8 + i in Z [I]. Because either a. Now a A implies a E akR for some k. where neitherb nor c is a unit. . proving our assertion.R. LetA—Ua.R and a. Find the g. Every PID is a UFD. 2 Principal ideal domains Recall that a commutative integral domain R with unity is a principal ideal domain (PID) if each ideal in R is of the form (a) — aR. in a1R.R C b lie in one of the two ideals a. Hence. If both b and c are products of irreducible elements. and write b — xy. A aR C akR C A gives that A aR = akR..1 Theorem. arE a. hence. Then a E a.RCA.j. r E R. Proof First we show that if R is a principal ideal ring. we show that each element a E R is a finite product of irreducible elements. If a is irreducible. Show that there exist irreducible elements which are not prime in each of the rings and Let 11. then (a.bc) = I. then R cannot have any infinite properly ascending chain of ideals. say. 9. 4. Further.

proved in Section 4).y] cannot be of the form (f(x. But F[x. then pla orpib. b # 0.bE E= E—{0) and bl a. Finally.y)) x = cf(x.y)) for any polynomialf(x.1 (a) The ring of integers Z is a euclidean domain if we set /i(n) = ml.b E R. A commutative integral domain E with unity is called a euclidE —' Z satisfying the following (ii) Ifa. where p is an irreducible element in R.. then For each pair of elements a. But since R cannot possess any infinite properly ascending chain of ideals. y = df(x. and a.b E.yJ. for any nonzero polynomial ff(x) F[4 we set degree — offand 4(O) — —1. This gives x/c — y/d. To complete the proof that R is a UFD. n Z.yJ is a UFD (Theorem 4. 0 3 Euclidean domains An important class of unique factorization domains is given by euclidean domains or rings admitting a division algorithm.y) for some c. which is absurd because x and y are independent variables over F. the ordinary division algorithm yields integers q (= quotient) and r (= remainder) satisfying axiom (ii). Thus. there exist elements q and r in E such that a — bq + r. because (x) + (y) = (f(x. with çb(r) <çb(b). (b) The polynomial ring F Exl over a field Fin a variable xis a euclidean domain if. (c) The ring of Gaussian integers R = (m + n Z) is a eu- . Then for any two integers a and b (# 0). This process leads to a properly ascending chain of ideals (a) C (b) C (x) C .3. we give an example to show that a UFD need not be a PID.Eucildean doinahn 217 where x.1.d (#0) in F. we conclude that a must be a finite product of irreducible elements.yJ over a field Fin variables x and y. The ideal I = (x) + (y) in F[x. that is. This has been shown in Theorem 1. we need to show that if plab.y] is not a PID. These rings are defined as follows: can domain if there exists a function axioms: (I) DefinitIon. say.y). F[x. is not a product of irreducible elements. Consider the polynomial ring R = FIx. Examples of euclidean domains 3.. Axiom (i) is clear.y) E F(x. that will continue indefinitely if a is not a finite product of irreducible elements. dv — cy 0.

i( 1) as a lower bound. Hence.rER But since r"=a—qdeA and4i(r)4i(d) by the choice of d. Choose integers a0 and fi0 such that Ia — Then hO — fbi = Because b(a0 b(a— a0)+ Z. 3.) It requires quite tedious computations to show that is a PID but not a euclidean domain. and thus bla implies 4(a). Because for all a CA. Wilson. We claim A = (d). Math.2. 0 3. Mag. Consider. Every euclidean domain is a PID.3 Theorem.218 Unique factorizatlon domains and eudlidean donialns we have proves (i). = ÷ nPT) = m2 + n2.Notethatforx. .letO * be Rand write are rational numbers. then Proof Let A be a nonzero ideal in a euclidean domain R. The interested reader is referred to 3. If a C A. r must be zero. Every euclidean domain is a UFD. A principal ideal ring that is not a Euclidean ring. 46. which a +fi'/T. where q. Then the set a a non- empty set of integers with 4. There exists an interesting class of principal ideal domains that are not euclidean domains. R is a euclidean domain.2 Theorem. (a and b have the same parity if both are odd or both are even. for example. Thus.r çb(r) = R. = (a + b E Z and a. C.whereaandfi a01 Toprove(ii). q.b have the same parity). ha. Further — a0)2 + (P — Po)9 We can then wnte a= Hence. So by the principle of well-ordering of integers there exists an element d E A such that b(d) is the smallest in this set. a C (d). a— b(a —a0)+ bq + r. 34—8 (1973). Proof Follows from Theorems 2. write a = qd + r. 0 Remark.1 and 3.yE R.

4. Let R be a commutative integral domain with unity that is not a field. then = çti<b). Show that if P is a primary ideal then = {XERIfEP for some n > 11. Prove that in any PID every ideal is a unique product of prime ideals 8. is called the radical of P. (a) (b) (c) 2. 9.y]. Give an example to show that this may not be true if R is not a PID. Show that a€R is a unit if and only if q(a)= (p(l). a and b are associates. xyEP. Prove the following: If b 0.[2J and dean domain and (ii) UFD.Findqand r in Z[iJ such that a = bq + rand q.y) is in F[x. a prime ideal. where q. Explain why in the UFD even though each of (2— = (11 — (5 + + the factors is irreducible. Let R be a euclidean domain. then If a and b are associates. is a (i) eucliShow that each of the rings Z[. If 3. Conclude that the only units of Z[iJ are ± I.] 4 Polynomial rings over UFD Recall that the ring of polynomials R[xJ over a ring R in x (called a variable or an indeterminate) is the set of formal expressions a0 + . Leta=3+2iandb=2—3ibetwoelementsinZ[i].yeR.(r) < q(b). Let R be a euclidean domain. Prove that if D is a domain that is not a field. An ideal P in a commutative ring R is called primary if for all x. 7. Prove that every nonzero prime ideal in a euclidean domain is maximal.) Let R be a PID. +1.(x + iy) = xl + yl. for some prime element p. then D[xJ is not a euclidean domain. e> 0. Show each primary ideal P is of the form (pe).PoJynomlat rings over UFD 219 Problems 1. Show that is not a euclidean domain. [Consider the ideal(x2. 6. 5. 10. show that the polynomial ring R(x] in a variable x is not a PID. implies fEP for some positive integer n.

So let m> n and k = m — n + 1. which is given in the following where a(. y. and we prove it for polynomials of degree m.). We prove the theorem by induction on m.0.. respectively. Let k = max(m — n + 1. and a.0. then n is called the degree off. Also. a01 y' if i = 0.0. or. lff= (a1) is a nonzero polynomial and n is the greatest integer such that a. equivalently. R can be embedded in R[x]. We shall regard R as a subring of REX] by identifying a with (a. thenfis called monic. a typical element of REx.xJ. where all but a finite number of a's are zero. Let R be a ring. If we start a1 x with S and construct the polynomial ring S[y] over S in the indeterminate. has degree at most m — 1. The degree of a zero polynomial is generally defined to be — The natural mapping a —' (a.y] = REy.0.y over R.1 Theorem.0)... Let R = F[x] bea polynomial ring overa commutative integral domain F. We assume it is true for all polynomials of degree < m. is called the leading coej/icieni off Also. and if iand fare both zero. 1—0 j—0 ifj =0. and letS = REx] be a polynomial ring over R.) is an isomorphism of R into R[xJ. Then there exist unique polynomials q(x) and r(x) in F[x] such that acf(x) = q(x)g(x) + r(x).. for example.a11a2.220 Unique factorization domains and euclidean domains + + a1 E R. we take q(x) — 0 and r(x) =f(x). if I E Rand a. We write this ring as R[x..x'yJ denotes 4... where r(x) = 0 or r(x) has degree less than the degree of g(x). It follows from the definition that RIx.. Hence.. the set of sequences (a0.. a E R.). and let a be the leading coefficient of g(x). in other words.y] is of the form a11ER.. then S(yJ is called a polynomial ring in two indeterminates or variables x. we can define a polynomial ring in a finite number of variables. Now the polynomial af(x) — the leading coefficient off By the induction hypothesis there where b is . = 1..y]. a polynomialfis monic if its leading coefficient is I. Let f(x) and g(x) 0 be polynomials in F{x] of degrees m andn. Inductively. Proof If m <n. # 0. The polynomial rings over a commutative integral domain have a division algorithm.

we need only prove thatf1g.(x) is Proof Writing c = c(f) and d = c(g). 4. and g1. we have f(x) = dg.x' and g1(x) — b. Note that 1(x) is primitive if and only if c(f) is a unit. the product of two primitive polynomials is primitive. whercf1 and are primitive. Because a polynomial ring F[xJ over a field F. Becausefg — cJ (x) and g(x) = I I Because R is a UFD. which is trivially a UFD.2 Lemma (Gauss). c is determined to within a unit factor. therefore.d.. follows from the primitivity off.d. that are not divisible by p (the existence of a.Polynomial rings over UFD exist polynomials q. If f. is necessarily a g. and g1). wheref. The factor c is called the content off(x) and is denoted by c(f). In particular. by choosing c to be equal to the g. of the coefficients off(x). g(x) E R[x]. respectively.c.(x) — a. where c E R and f. p % b. Thenf(x) E R[x] is called primitive if the g. Supposef.1(b) and Theorem 3.(x). akf(x) = + q. as desired. of its coefficients is a unit.c. f(x) We observe that it is possible to write any nonzero polynomial f R[xj in the form f= cf1.d.g. Iff(x).. 0 Now our aim is to show that polynomial rings over a UFD are also UFD. and b.g1 is not primitive and letpbe an irreducible element of R that divides all the coefficients of f.(x). of the coefficients off(x).3]. Because p divides all terms of the above sum . (x) and r1 (x) such that 221 (af(x) Then — bxmMg(x)) — q. then c(fg) c(f)c(g). We first need some preliminaries. is also a UFD [Example 3.c.(x)g(x) + r. . In what follows we assume that R is a UFD. our next theorem may be regarded as a generalization of this fact. Uniqueness follows immediately. is primitive. (x). Any element c satisfying the conditionf= cf. (x) is primitive. is primitive. (x))g(x) + r. Definition. Let R be a UFD. Now the coefficient of x" inf1 (x)g.

Then p(x)If(x)g(x) ah(x) =f(x)g(x). first show that each nonunit element of R[x] factors into irre- ducible ones..f(x). so 4(x) a E R. which yields c(t) = Therefore. a contradic- tion. and. this yields alc(f) or alc(g). Now . yields dividesf(x). We prove the result by induction on the degree of a polynomial.222 Unique factorizatlon domains and eudildean domains that precede and follow a. The elements of S are of the form A(x)p(x) + S be of the smallest B(x)f(x). Let R bea uniquefaciorizazion domain. a contradiction unless r(x) —0. therefore. This implies r(x) S. c(f)c(g). divides implies akf(x). If the degree of p(x) is zero. Then the poly- nomial ring R[x] Proof We over R is also a unique factorizazion domain. Otherwise. hence. 0 We now prove 43 Theorem. By Theorem 4. it does not divide the sum itself. and let a be its leading coefficient. then p(x)If(x) orp(x)Ig(x). Thus.. it suffices to prove the theorem for polynomials f(x) E R[x] that are primitive. B(x) E R[x). If the degree off(x) I. Without any loss of generality. n1 <n and n2 <n. g(x) E R(xJ and p(x) is an irreducible element in REx] such that p(x)If(x)g(x).f1 (x)g1 (x) is primitive. alf(x) or alg(x). qS(x) = R. a'91(x). = This. Assume that the degree ofp(x) is positive. then p(x) = a E R. so akf(x) = = (x)h(x). Hence. the result follows by induction.1 there exists a nonnegative integer k and polynomials h(x) and r(x) such that akf(x) 4i(x)h(x) + r(x). Hence. Because is a unit in p(x) is irreducible and does not dividef(x). where either r(x) —0 or the degree of r(x) is less than that of 4(x). it follows that R(x].!. Consider the ideal S generated by f(x) and p(x) in R[x]. Let 0* degree. and let p(x) not divide f(x). But h(4 E R[x]. Thus. writef(x) — f1 (x)f2(x). where f1(x) andf2(x) are nonconstant polynomials over R of degrees n1 and n2. dividesp(x). Hence. Similarly. we arc done. and the result is true for all polynomials of degree <n. lff(x) is irreducible over R. where 41(x) is primitive. E R. then the result clearly holds because R is a UFD. Now assume that the degree off(x) — n> I. where A(x). This implies ac(h) since a is irreducible. respectively. along with and the fact that R(x] is an integral domain. We now show that if p(x).

2. since. Thus. By equating the contents of the polynomials on both sides and using Lemma 4. Hence. p(x)If(x)g(x). Show that the n X n matrix ring over a polynomial ring over the n X n REx] is isomorphic to the polynomial ring matrix ring ..Polynomial rings over UFD 223 and we can write a = A(x)p(x) + B(x)f(x) for some A(x). I l.. is also a UFD. 2. 0 over R in a finite number of Remark. we obtain ac(g) = c(p)c(1) = c(i). 1(x) E R[x]. o b R such that bf(x) = 0. because p(x) is irreducible. Show that an element f(x) E REx] is a zero divisor if and only if there exists an element I.y] in two variables over a field F is a UFD but not a PID. Show that a nonzero polynomialf(x) Fix] is a unit if and only if f(x) E F. 4. This proves that t'(x) as desired. ag(x) = p(x)i(x).. 3. by hypothesis. we get ag(x) = A(x)p(x)g(x) + B(x)f(x)g(x).n.2. g(x) = p(x)t'(x). It follows immediately from Theorem 4. Let R be a commutative ring with unity. Let F[x] be polynomial ring over a field F. Problems Show that the polynomial ring F[x.. R[xJ. B(x) in REx]. Thus. and therefore aIz(x). Then by multiplying by g(x) on both sides of the above equation. This yields p(x)Iag(x). then the polynomial ring variables x.3 that if R is a UFD..

Indeed. if R is an integral domain. 1 RIngs of fractions DefInitIon. relation — on R X S by (r. aearly. that is. we have S. A nonempty subset Sofa ring R is called a multiplicative set if for all s1 .CHAPTER 12 Rings of fractions Let R be a commutative ring containing regular elements. we obtain a ring R5 such that there is a canonical homomorphism from R to R5. In this chapter we show that any commutative ring R with regular elements can be embedded in a ring Q with unity such that every regular element ofR is invertible in Q.s) s"(rs' — r's) = 0.s2 E S. Define a (r'. In particular. any integral domain can be embedded in a field. then S is called a regular multiplicative set. then S = R — (0) is a regular multiplicative set. elements a E R such that a 0 and a is not a zero divisor.s') if there exists s" S such that . In particular. If in addition. the set of all regular elements is a regular multiplicative set. by defining the general notion of ring of fractions with respect to a multiplicative subset S. The conditions under which a noncommutative integral domain can be embedded in a division ring are also discussed. each element of S is regular. 224 Let R be a commutative ring and S a multiplicative set.

Then f(a + b) (a + b)s/s = as/s + bs/s —f(a) +f(b). and let R5 denote the set of all equivalence classes. a as/s — 0/s t(as2 — Os) 0 for some z S forsomexES. where a R. and letS be a multiplicative subset of R. . 1. RI!. a/s1 b/s2 = ab/s1 s2. is altogether a different notion. We define a/s1 + b/s2 = (as2 + bs1)/s.1 225 Lemma. we assume throughout this section that 0 S.2 Theorem. Further. that is. Proof It is routine to verify that addition and multiplication as defined preceding the theorem are well defined and that is a ring. a R implies a 0.ands/sistheunityforanysES.) IfS consists of all the regular elements of R. (Warning: The quotient ring of a ring R modulo an ideal I.b R. Let R be a commutative ring.3 Theorem. (R5.+.) is a ring with unity. Proof Let a. — is an equivalence relation. Then there exists a natural homomorphism R defined by f(a) as/s. then R5 is the zero ring. Next.Riap of fractions 1. if R5 is a ring of fractions where 0 E S. 1. Finally.s).fis a homomorphism. and s S is some fixed element. Hence. x E S. R5 is also called the quotient ringof R with respect to S. It may be noted that the zero element of R5 is 0/s for any 5ES. The reader is encouraged to verify these facts for the sake of getting familiar with the notion. Proof The proof follows by routine verification. 0 Let a/s denote the equivalence class determined by (a. as defined in Chapter 10. then R5 is usually known as the total quotient ring of R. reminiscent of addition and multiplication of "fractions" in arithmetic. Therefore. f(ab) = abs/s = as/s bs/s —f(a)f(b). f is a monomorphism ¶xa 0. 0 The ring R5 is called the ring offractions of R with respect to S or the localization of R at S.s2. Our purpose is to make into a ring by defining addition and multiplication in a natural manner.

Then the ring R5 has thefollowing properties. then Q is called a quotient ring of R. we have 1. Any commutative integral domain R can be embedded in afield R5 (called the field of fractions of A). Let S be the set of all regular elements in R. Proof That S is a multiplicative set follows from the definition ofa prime ideal. 0 As a consequence of Theorem 1. where b is some elemen' inS. (i) .R is embeddable in R5.4 Theorem. A ring R with unity is called a local ring if it has a unique maximal right ideal.3. Proof (i) follows from Theorem 1. IfR is commutative. Let a. Then S = R — P is a multiplicative set. where a E R.4.r'.226 Rings of fracdons Therefore. a2/s2 E . Kerf= (0) iii ax =0 implies a 0. Let R bea commutative ring with regular elements. where a E R and b is a regular element of R.3. 1. 0 Definition. Then Sis a regular multiplicative set. Finally.. Proof Let S = A — (0). We first show that is an ideal in R3. and/el Q be a ring with unity containing R as a subring such that each regular element of R is invertible in Q and each element of Q is ofiheform ab (a.'b). 1. and R5 is a local ring with unique maximal ideal (a/sea E F. because ab E S.6 Theorem. Regarding R as a subring of R5. s P)./s. Further. Then Q is called a right (left) quotient ring of R. then identifying a with ab/b E R5. Let R be a commutative ring andPa prime ideal. Let R be any ring (not necessarily commutative). we have (ii) each regular element of R is invertible in R5. 0 Definition. let a/s E R5.5 Theorem. R5 is a field containing A as a subring. it is clear that blab E R5 is the inverse of ab/b. to prove (iii). sE S. if a is a regular element in R. Then a/s (ab/bXb/bs) a. and (iii) each element of R5 is of the form ar'. By Theorem 1.

s/r C R5.Rings of 227 and r/s C R5. hence. so a is prime in R. Let!: R R5 be the canonical homomorphism given byf(a) = as/s for some fixed s E S. First we show that if a E R is irreducible in R. Then r P. Then a/i = (b/s1Xc/s2) = (b. that if a/s E R5 is irreducible then it is prime. and let S be a multiplicative subset of R containing the unity of R. We prove a/i is prime in Let a/i divide (b/s1)(c/s2). then a/I is irreducible in R5. 0 1. But then a divides b. Let R be a UFD. This proves one of the conditions for R5 to be a UFD.. or c/s2 as desired. let a/i — (b/s. This shows that all the elements of R5 that lie outside are invertible./s1Xr/s) a. — a2/s2 = (a.it follows that a/si!.s Ps. Now let a/seR5./s. lfg: R R' is any homomorphism of a ring R into a ring R' such that every element g(s). To show that P5 is (a. We proceed to prove the second condition. prime. it follows that aib or aic.s2 C Further. s C 5. Then a/s = (1/s)(a1/l)(a2/I). For definiteness let ab.c. or 53. Then (a/lXd/s3) — (b/s.s2 — a2s. Thus.r/s./s. Then R5 is also a UFD./l) is a product of irreducible elements in R5. so albcs3. a contradic1 tion. 1. is an ideal.)/s. Ps is the unique maximal ideal in R5. where b/s. so ab. let r/s P5. . so r/s is invertible. Then as.Xc/s2) for some d/s3 ER5. because c/s2 is not a unit. Solution.Xc/s2).(a.Xc/s2). 0 Problems In Problems I —3. s2 = bcs3. aib or aic. then there exists a homomorphism h: R5 following diagram is commutative: g R' such that the . IfaIs3. For. This implies ads. maximal. so a/s is a unit in R5. Hence. S is a multiplicative subset of a commutative ring R with unity and 1 C S. and c/s2 are nonunits in R5. Now a/s irreducible implies a is irreducible in R. Then a.s2 = bc. = b for some b. as a product of irreducibleelements in R. Write a = a1a2 a. This proves that R5 is a local ring. a contradiction. C R. Hence. This implies a/s divides b/s. is a unit. Because a is irreducible and.7 Example. Thus.

f(0) =0 andf(—x) = —1(x). A mapping with these two properties is known as an R-linear mapping.y) are isomorphic fields? RIngs with Ore condition 2 We now consider the question of embedding any integral domain in a division ring. Show that the ring Z(p) 4.R is an R-linear mapping. Then is closed under intersection.y R such that ax = by. that R isa right Ore domain. Show that the same holds for R8. p%b} 5.ShowthatR5isalsoaPlD. Choose 0 # a E land 0 b J. Since ax = by I n J. is a local ring. What is the unique maximal ideal? Let R(x) and R(x. Throughout this section '1 will denote the set of nonzero right ideals of aringR. then E '€ for any . Then there exist 0 # x.y E I. 3.b Z. An integral domain R is called a right (left) Ore domain tffor any pair of nonzero elements a. Thus. The converse is similar. first.1 ('ii) Lemma. Let R be an integral domain. 2. We note that iff is an R-linear mapping.228 Rings of frsctioin This is also expressed by saying that g factors throughf 2. Let R be a commutative ring with I having no infinite properly ascending chain of ideals in R. = (a/bla.b R there exist nonzero elements x. Definition. and let I be a nonzero right ideal of R. LetRbeaPlD. and (ii)f(xr) =f(x)r for all x. We also remark that it follows from the definition that any commutative integral domain is a right Ore domain as well as a left Ore domain. Denote by HomR(I. Let R be a right Ore domain. 1ff I—' . Is it true that R(x) and R(x. Suppose. Let p be prime. then f is an additive group homomorphism. Let land Jbe nonzero right ideals in R. it follows that I fl J # (0). r E R. y) be fields of rational functions over R.y E R such that ax by (xa = yb). We note that an integral domain R is a right Ore domain if and only if every pair of nonzero right ideals has a nonzero intersection.R) the set of mappings f: I —+ R such that (1) f(x + y) =f(x) +f(y).

Hence.'(X) is a right By using the hypothesis that R is a right Ore domain.+) is an additive abelian group. iff(I) = 0. Then because R is a right Ore domain. we show that (0). This implies ar €f '(X).b €f'(X) and r E R. Also. (ii) Let a. This yieldsf+g=f'+g' on AnBECe. 0 B E Ce. the addition is commutative and assodative. So IC f '(X). We now show — is transitive. then because R is a right Ore domain. In J X. (Q. — is reflexive and symmetric. J E Ce. Our purpose now is to make Q = HI— into a ring by suitably defining addition and multiplication. where f+g: DomfflDomg-'R. Proof Clearly. This proves that— is an equiva— Denote by [f] the equivalence class determined byfe H. 0 i' a This proves 0 Let R be a right Ore domain and H = tion — on H by f— g iff= g on some A E Define a rela- 2. Thus. and by Q the set HI— of equivalence classes. Letf—gandg— h. and g=g' on some BECe. [f + gJ — [f' + g' J.+) and (—fJ is the additive inverse of [f]. So f(ar) E X. Also. Thenf=f' on some A Ce .f(I) fl X # (0). Then f(a). thenf(I) C X. Therefore. Proof (i) Let I. To show that addition is well defined. sof(a) —f(b) E bEf'(X). But then!— hon A lence relation. [0] is the identity element in (Q.f(b) I. note that Domffl Dom g C Ce. For. This proves that!.Rings with Ore conditIon 229 (Recall that f'(X)= (0). — is an equivalence relation. This implies that there exists 0 # a E I such that f(a) E I. Thus a — ideal. Hencef—h.f(a)r€X.Thenf=gonsomeA C €andg= honsomeBE Ce. First. In what follows let Dom f denote the domain of a mapping f We define addition in H as follows: [fJ+[gJ= [f+gJ. let [fJ If'] and [gJ = [g'J. sof+ g €11. On the other hand. if f(I) 0.2 Lemma. because Xis a right ideal.

+ [g])[hJ — [fJ[h] + IgJ[hl ([fJ [f]([g] + [Ii]) hold.R)/— is a division ring. so [11 has a left inverse.) Proof Let R be a right Ore domain. Q — U.wherea* c . But then gf— identity mapping Ion Domf. if!: R R is the identity mapping.Weclaimthatfg =f'g' onB fl g'(A). (ii) Every element of Q is of the form ab'.R)/— is a division ring. x E Dom f The mapping g is well defined because Ker f— (0).B E re.isa 1-1 homoofRintoQ. This proves [fg] = (f'g']. sofg E H. Then x E B implies g(x) — g'(x) and g(x) E A implies f(g(x)) =f'(g(x)). This gives Ken— (0). then (I] is the unity of Q. (Recall that the ring Q with properties (i) and (ii) is called a right quotient ring of R. let If] — [f'] and [g] = Ig']. where fg: g'(Domf) -.3.1. Q is a ring with unity. Thenf=f' on.4 (i) R is a subring of Q.say. To show that multiplication is well defined. Define g: Imf—' R by g(f(x)) x. Then R is a right Ore domain jf and only there exists a division ring Q such that 2. Thus. [11= f 0]. Hence. R. Nowf= the zero mapping on Kerf Therefore.b. If Kerf# (0). Let R be an integral domain. 2. a contradiction. Hom(I.E. By Lemma 2.3 Theorem. andg= g'on. We assert that the mapping a '—' 1a] ax. This gives (g)(fJ (I]. say. for some a. right Ore domain. therefore [gf) (I). (fgXx) = (fg'Xx). Then by Theorem 2. Also. we define multiplication in H as follows: (f][g] = (fgj. We now show that Q is a division ring. then KerfE €. 0 Theorem. R.230 RIngs of fractions Next. Let R be a Hom(I. Then Q Proof Let [f] be a nonzero element in Q.A E €. g '(Domf) It is straightforward to check that distributive laws is a semigroup and that the [fJ[g] + [f][h]. This proves that Q is a division ring.

considering q and a (= as) as mappings. a3) of elements in A. We may remark that under this identification if q E Q and a E R. then q(a) is identified with qa.J. Further.13) of is positive (1. R is a right principal ideal domain). Then a5 = 0 on some I E This implies =0. let a and b be nonzero elements in R. For.3). By construction of Q there exists I that q(I) C R. by definition of an R-hnear of mappings q and a. This proves (i). a E R (i. For. where the summation runs over every permutation '13. Show that R is a right Ore domain and a left Ore domain. Let 0 '. qb — a.e. lfR is an integral domain with unity in which each right ideal is of the form aR. Then Q. (ab)"(x) = (ab)x = a(bx) Therefore. that is. This implies ax — by. Conversely. such To prove (ii). show that R is a right Ore domain. This implies q — is a' Problems I.i. Therefore. (qaXb) — — q(ab). Let R be an integral domain satisfying the standard identity of degree 3. . under the stated identification if — q Q.) is an even permutation or an odd permutation. ± — 0. mapping q. Also. so (a + b)' — a5 + This implies [(a + b)s] — (a* + b*J — (as) + (b.y R.2.Rings with Ore condItion 231 morphism. let [as] = [0]. so for some nonzero elements x.a2 . this implies 0. Then there exists a E R such that q(b) — a. This proves that the mapping is 1-1. a(b*x) = as(bsx) — (a*b*Xx). then q(a) — qa. 2. the composition a — q(ab). R can be embedded in a division ring Q. We may then identify R with its copy in Qand regard R as a subring of Q. b— Hence. a E A. that is. But because R is an integral domain. a! = 0. To show that the mapso ((ab)') ping is I-I. let q Q. Hence. R is a right Ore domain. because every nonzero element in Q invertible. and the sign of the corresponding term or negative according to whether (1 . for every ordered set (a . (a+b)a(x)=(a+b)x=ax+bx_a*(x)+bls(x)_(as+b*Xx)..' b E!. 0 that is.

232 RIngs of fractions Show that an integral domain satisfying a standard identity of any degree n is both a right and a left Ore domain. ring of quotients of R = ((3 E F) is the 2 X 2 matrix ring over F. . Show that (3 is regular if and only if a 0.) 3. as well as a left. and let R — ((3 E F). Conclude that R is its own right and left ring of quotients. Let F be a field. 4. (We may remark that a tight. and every regular element in R is invertible in R.

I Peano's axioms The traditional method of describing the set N of natural numbers axiomatically is by means of the following axioms of Peano: (i) (ii) lEN. 2' 3. We write 1' = 2. called the successor map.CHAPTER 13 Integers In this chapter our purpose is to establish all the familiar properties of the natural numbers (= positive integers) and to obtain the ring of integers by starting from the five axioms of Peano. For each a E N there exists a unique a' E N called the successor of a. (In other words there exists a map a a' of N into itself.(b)ifa S. (In other words.b N.) (iii) (iv) (v) I foranyaEN. S. Either of these sets of axioms determines a unique ring (up to isomorphism) called the ring of integers. The fifth axiom is called the axiom of induction or the ii rst principle of induction and is the basis of the proofs of many theorems in mathematics. We also demonstrate that the five axioms are equivalent to the axioms ofan ordered integral domain whose positive elements are well-ordered. For all a. We may restate the first principle of induction (also called the principle of mathematical induction) as follows: 233 .) LetSbeasubsetofNsuchthat(a)l Then S= N. and so on. 3' — 4. a' b' S a = b.thena' ES. the successor 1-1.

n' E £ Axiom (v) then gives S — N. a' a.. there exisisauniqueb E Nsuch that a — b'. LetS-'(nENIn'*n).Thenbydefini- tionofS. in N satisfying 1." The next theorem shows the existence and uniqueness of a binary operation in N (called plus) satisfying certain conditions. 1. (b) S(n) is true implies S(n') is also true. S — N. a # 1. n. so (n')' # n'.234 Integers (v') Suppose that for each natural number n E N we have associated a statement S(n) such that (a) S( I) is true.1 Proof Corollary. 0 1. For any a E N. Then S(n) is true for all n E N. for by axiom (iv). Further. The following are some of the immediate consequences of Peano's axioms. Thus n' E S. Thenn-' m' for somemEN. b = c. Then n' # n. by axiom (v). Byaxioms(i)and(iil). (n')' — n' implies n' — Consequently. if a — b' and a — C'. 1 ES. for all a. There exists one and only one binary operation "+.so n'=(m')'.b€ N. Proof We first show that there exists a binary operation "+" in N satisfying (1) and (ii). LetnES. then by axiom (iv). Theorem. I ES.2 Corollary. . LetnES. Hence. 0 Addition and multiplication of natural numbers can be defined in a "recursive manner.3 (1) (ii) a+b'(a+b)' a+l=a'. Proof LetS=(n Nm —1 orn — m'forsomeme N). Foranya EN. Consider the set S— (a Nia + b can be defined for all b in N satisfying conditions (i) and (ii)).

We now define a' + b= (a + b)'. Next.Peano's axIoms 235 To show that I E S. For any fixed a E N let Then I ESbecause a+1=a' a®1 [by(i)] [by(i)]. Hence.4 laws: (1) (since it E 5) [by(ii)1. a + b is defined for all b E N. So a' E S. The binary operation + in N satisfies the following Foralla. (a + b) + c — a + (b + c) (associative law of addition). (I) and (ii) are satisfied for the case Let a E S. let n E S. Hence. Hence. . This proves that "+&