Undergraduate Texts in Mathematics Editors F. W. Gehring P.R. Halmos Advisory Board C, DePrima I. HersteinUndergraduate Texts in Mathematics Apostol: Introduction to Analytic Number Theory. 1976. xii, 338 pages. 24 illus. Armstrong: Basic Topology. 1983. xii, 260 pages. 132 illus. Bak/Newman: Complex Analysis. 1982. x, 224 pages. 69 illus. Banchoft/Wermer: Linear Algebra Through Geometry. 1983. x, 257 pages. 81 illus. Childs: A Concrete Introduction to Higher Algebra. 1979. xiv, 338 pages. 8 illus. Chung: Elementary Probability Theory with Stochastic Processes. 1975. xvi, 325 pages. 36 illus. Croom: Basic Concepts of Algebraic Topology I9"S. x, 177 pages. 46 illus. Curtis, Linear Algebra: Ar Introductory Approach (4th editiony I¥84. y. 337 pages. 37 illus. Dixmuer: General Topology 1984, \, 140 pages. 13 illus. Ebbinghaus/Fium/ Thomas Mathematical Logic. 1984, xii, 216 pages. 1 illus Fischer: Intermediate Real Analysis. 1983. xiv, 770 pages. 100 illus. Fleming: Functions of Several Variables. Second edition. 1977 1, 411 pages. 96 illus. Foulds: Optimization Techniques: An Introduction. 1981. xii, $02 pages. 72 illus. Franklin: Methods of Mathematical Economics. Linear and Nonlinear Programming, Fixed-Point Theorems, 1980. x, 297 pages. 38 illus. Halmos: Finite-Dimensional Vector Spaces. Second edition, 1974, viii, 200 pages. Halmos: Naive Set Theory. 1974, vii, 104 pages. looss/ Joseph: Elementary Stability and Bifurcation Theory. 1980. xv, 286 pages. 47 illus. Janich: Topology 1984. ix, 180 pages (approx.). 180 illus. Kemeny/Snell: Finite Markov Chains. 1976. ix, 224 pages. 11 illus. Lang: Undergraduate Analysis 1983. xiii, 545 pages. 52 illus. Lax/Burstein/Lax: Calculus with Applications and Computing, Volume 1 Corrected Second Printing. 1984. xi, 513 pages. 170 illus. LeCuyer: College Mathematics with A Programming Language. 1978. xii, 420 pages. 144 illus. Macki/Strauss: Introduction to Optimal Control Theory. 1981. xiti, 168 pages. 68 illus. Malitz: Introduction to Mathematical Logic: Set Theory - Computable Functions - Mode! Theory. 198 pages. 2 illus. continued after IndexLindsay Childs A Concrete Introduction to Higher Algebra Corrected Second Printing 5 Springer-Verlag New York Berlin Heidelberg TokyoLindsay Childs Department of Mathematics SUNY at Albany Albany, New York 12222 USA Editorial Board F, W. Gehring P. R. Halmos Department of Mathematics Department of Mathematics University of Michigan Indiana University ‘Ann Arbor, Michigan 48104 Bloomington, Indiana 47405 USA USA To Ashley and Nathan AMS Subject Classification: 12-01 Library of Congress Cataloging in Publication Data Childs, Lindsay N. A concrete introduction to higher algebra. (Undergraduate texts in mathematics) Bibliography: p. Includes index. 1. Algebra. 1. Title QAISS.C53 512.9 ———78-21870 With 9 Illustrations © 1979 by Springer-Verlag New York Inc. All rights reserved. No part of this book may be translated or reproduced in any form without written permission from Springer-Verlag, 175 Fifth Avenue, New York, New York 10010. U.S.A. Printed and bound by R. R. Donnelley & Sons, Harrisonburg, Virginia. Printed in the United States of America. 9876543 (Third printing, 1984) ISBN 0-387-90333-X Springer-Verlag New York Berlin Heidelberg Tokyo ISBN 3-540-90333-X Springer-Verlag Berlin Heidelberg New York TokyoPreface This book is written as an introduction to higher algebra for students with a background of a year of calculus. The book developed out of a set of notes for a sophomore—junior level course at the State University of New York at Albany entitled Classical Algebra. In the 1950s and before, it was customary for the first course in algebra to be a course in the theory of equations, consisting of a study of polynomials over the complex, real, and rational numbers, and, to a lesser extent, linear algebra from the point of view of systems of equations. Abstract algebra, that is, the study of groups, rings, and fields, usually followed such a course. In recent years the theory of equations course has disappeared. Without it, students entering abstract algebra courses tend to lack the experience in the algebraic theory of the basic classical examples of the integers and polynomials necessary for understanding, and more importantly, for ap- preciating the formalism. To meet this problem, several texts have recently appeared introducing algebra through number theory. This book combines the newer number-theoretic approach with the old theory of equations. In fact, the book contains enough of each of elemen- tary number theory and the theory of equations that a course in either could be taught from it (see below). But the algebraic similarities of the two subjects are such that both subjects can be developed in parallel, and ideas customarily associated with one can be transferred to the other. Thus the ideas of congruence and congruence classes, normally arising in elementary number theory, can also be used with polynomials. Doing so permits passage from the study of polynomials to the study of simple field extensions, and in particular, leads to an exposition of finite fields. There are, I feel, several advantages in beginning the study of higher algebra by studying number theory and polynomial theory.vw Preface First, the algebra is built on the student's entire mathematical experi- ence. The study of numbers and polynomial equations dominates the precollege mathematical training. By building on this background a course in algebra will be building on the strongest possible intuitive base. And given such a base, the potential for reaching results of significance is high. J hope that this potential is realized by this book’s theoretical development, numerous applications, and exercises. Second, the dominating algebraic idea in the development of the book is that of congruence classes. The concept of quotient structure is perhaps the most difficult of the concepts of abstract algebra. The experience of seeing it in a variety of concrete contexts, and seeing worthwhile consequences of its use, should greatly aid the student when subsequently it is seen in abstract presentations. This particular feature of our approach is one which was missing from traditional theory of equations courses, and also is missing in courses in linear algebra used as background for abstract algebra. Third, the subject matter of the book is intrinsically worth studying. Both number theory and the theory of equations have attracted the attention of the very greatest mathematicians. In particular, two of Gauss’s greatest achievements, the fundamental theorem of algebra and the law of quadratic reciprocity, are important results in this book. One of the important lines of research in modern algebraic geometry stems from A. Weil's 1949 paper on solutions of equations in finite fields, a topic which is beyond the level, but very much in the tradition, of the material in this book (see Ireland and Rosen (1982) for an exposition). But even at the level of this book the subjects have attracted the notice of combinatorial analysts and computer scientists in recent decades. A surprising amount of the material in this book dates from since 1940. As I discovered only late in the writing of this book, there is considerable overlap between it and the mathematics in Chapter 4, “Arithmetic,” of D. E. Knuth’s fundamental treatise, The Art of Computer Programming (Knuth, 1969). Thus the mathematics in this book is worth learning for its own sake, apart from any value it has in-preparing for more advanced mathematics. The explicit prerequisites of the book consist for the most part only of high school algebra (in the de facto sense, not in the sense of Abhyankar (1976)—in his sense this is predominantly a high school algebra text). In various places we assume some acquaintance with calculus; however, the subject of differentiating polynomials is developed from the beginning (Chapter II-6), and the one place where integration occurs explicitly (Section II-5C) may be omitted.' Two-variable calculus is mentioned only in the proof of the fundamental theorem of algebra (II-3), but either the facts needed there can be taken on faith or the proof can be omitted. The ' Chapter 6 of Part II is referred to as Chapter II-6 or simply as 11-6; if the reference occurs within Part II, simply as Chapter 6, II-5C refers to Section C of Chapter II-5.Preface vin use of infinite series is more substantial, however, particularly in connec- tion with decimal expansions of fractions. Several of the applications and a few of the theoretical sections use matrices and ideas from linear algebra. Chapter 1-9 is an overly concise review of the necessary ideas. Sections C-F of I-9 should be used only for reference. If linear algebra is lacking in the student's background, the chapters particularly to avoid are II-12 and the last half of III-9. The remaining uses of linear algebra mainly involve simple matrix manipula- tions as described in Section I-9A, and these are quickly learned. Exercises are scattered throughout the text as well as collected at the ends of sections. They range from routine examples to ingenious problems to extensions of theory. The most nontrivial ones are starred; comments on them are collected in the back of the book. Exercises which are mentioned subsequently either in the text or in exercises are marked with a dagger, and are indexed together with the subsequent references in the back of the book. There is more material in this book than would be appropriate for a one semester course. For a year course it could be supplemented with a not-too-geometric introduction to linear algebra (such as Zelinsky (1973)). For a one semester course there are a variety of routes through the book. The main development in Parts I and II is contained in I—: 1-3, 4A, 5A, B, D, 6-8, 11, 14A; Ti—: 1, 2, 3A, 4, 6, 8-10, IIA. To get to the classification of finite fields (III-14) most efficiently, follow the main development with HI—: 1, 4, 7, 8, 10, 11, (12), 13, 14. To get to algebraic numbers (III-18) most efficiently, follow the main development with III—: 1, 4, 7, 8, 10, 16A, B, 18-21. To concentrate on elementary number theory, see I—: all except 9C-F and 4B, 4C, 5C, 9B, 10, 13, 14B, 15 as desired; II—: 1, 2A; II[—: 1-5, 16A, B, 17; then II—: 3A, 8-10, 11A; TI—: 7, 8, 10, 18-21. To concentrate on theory of equations, see: I—; 1-3, 4A, 6-8, 11, 14A; II—: all, omitting 3C, 5, 7, 11B, 12 as desired; TiI—: 1, 4, 6-8, 10, 11, 13-16.viil Preface It is probably unwise to spend too much time in Chapter I-2. Also, part of the uniqueness of this book lies in the chapters on applications, so it is hoped that any route through the book will be chosen to allow time for visits to some of the scenic wayside areas. Finally, I wish to acknowledge with appreciation the contributions of people who in various ways influenced the book: Ed Davis, for developing and selling the idea for the course for which the book was written, and for a number of useful comments on an early version of the notes; Bill Hammond, for teaching from the notes and offering a number of improve- ments; Violet Larney, for teaching from the notes graciously even though her book (for a competing course) had just appeared; Morris Orzech, Paulo Ribenboim, Tony Geramita, Ted Turner, and Ivan Niven, for a variety of mathematical insights and ideas; and, especially, Malcolm Smiley, for reading through and teaching from the manuscript in its late stages and offering many substantial suggestions for improving the exposi- tion, and David Drasin, for reading through and making many helpful comments on the nearly completed manuscript. Also I wish to thank: Michele Palleschi for typing most of the manuscript even though she didn’t have to, and Mrs. Betty Turner and her staff for typing most of the manuscript even though they weren’t supposed to; the Universities of Illinois (Urbana) and Oregon for their hospitality during part of the time the book was written; and Springer-Verlag, particularly Walter Kaufmann-Buehler and Joe Gannon, for their professional treatment of the manuscript. Most of all, my greatest thanks go to my wife Rhonda, for putting up with my working on the manuscript at inconvenient hours at inconvenient locations. Fall, 1978 L. CHILDS My thanks to those who pointed out misprints and errors in the first printing, including Louis Brickman, Linda Deneen, William Hammond, Irving Kaplansky, Keith Kendig, Richard Patterson, Alan Sprague, Mel Thornton. Ted Turner, S. Wang, and especially Ernst S. Selmer. May, 1984 L. CHiLpsPart I INTEGERS Chapter 1 Numbers Chapter 2 Induction; the Binomial Theorem A. Induction B. Another Form of Induction C. Well-ordering D. The Binomial Theorem Chapter 3 Unique Factorization into Products of Primes A. Euclid’s Algorithm B. Greatest Common Divisors C. Unique Factorization D. Exponential Notation; Least Common Multiples Chapter 4 Primes A. Euclid B. Some Analytic Results . The Prime Number Theorem Contents W 1B 14 19 19 2 28 31 3 32 35x ‘Contents Chapter 5 Bases 37 A. Numbers in Base a 37 B. Operations in Base a 39 C. Multiple Precision Long Division 4 D. Decimal Expansions 4 Chapter 6 Congruences 47 A. Definition of Congruence 4 B. Basic Properties 48 C. Divisibility Tricks 49 D. More Properties of Congruence 51 E. Congruence Problems 2 F. Round Robin Tournaments 34 Chapter 7 Congruence Classes 36 Chapter 8 Rings and Fields 62 A. Axioms 2 B. Z,, 65 Chapter 9 Matrices and Vectors 68 A. Matrix Multiplication 68 B. The Ring of 1 Xn Matrices 70 . Linear Equations B D. Determinants and Inverses 6 E. Row Operations 6 F. Subspaces, Bases, Dimension 80 Chapter 10 Secret Codes, I 4 Chapter 11 Fermat's Theorem, I: Abelian Groups 90 A. Fermat's Theorem 0 B. Abelian Groups 91 C. Euler's Theorem 34 D. Finding High Powers mod m 95 E. The Order of an Element 96 F. About Finite Fields ” G. Nonabelian Groups 98 Chapter 12 Repeating Decimals, I 101Contents Chapter 13 Error Correcting Codes, I Chapter 14 The Chinese Remainder Theorem A. The Theorem B. A Generalization of Fermat's Theorem Chapter 15 Secret Codes, II Part II POLYNOMIALS Chapter 1 Polynomials Chapter 2 Unique Factorization A. Division Theorem B. Greatest Common Divisors C. Factorization into Irreducible Polynomials Chapter 3 The Fundamental Theorem of Algebra A. Inreducible Polynomials in C[x] B. Proof of the Fundamental Theorem Chapter 4 Irreducible Polynomials in R[x] Chapter 5 Partial Fractions A. Rational Functions B. Partial Fractions C. Integrating D. A Partitioning Formula Chapter 6 The Derivative of a Polynomial Chapter 7 Sturm’s Algorithm Chapter 8 Factoring in Q[x], I A. Gauss's Lemma B. Finding Roots C. Testing for Irreducibility xI 105 112 112 116 118, 123 125 129 129 132 134 136 136 138 142 144 144 145 148 151 157 160 166 166 168 169XII Chapter 9 Congruences Modulo a Polynomial Chapter 10 Fermat's Theorem, IT A. The Characteristic of a Field B. Applications of the Binomial Theorem Chapter 11 Factoring in Q[x], Il: Lagrange Interpolation A. The Chinese Remainder Theorem B. The Method of Lagrange Interpolation Chapter 12 Factoring in Z,[x] Chapter 13 Factoring in Q[x], III: Mod m A. Bounding the Coefficients of Factors of a Polynomial B. Factoring Modulo High Powers of Primes Part IIL FIELDS Chapter 1 Primitive Elements Chapter 2 Repeating Decimals, II Chapter 3 Testing for Primeness Chapter 4 Fourth Roots of One in Z, A. Primes B. Finite Fields of Complex Numbers Chapter 5 Telephone Cable Splicing Chapter 6 Factoring in Q[x], IV: Bad Examples Mod p Chapter 7 Congruence Classes Modulo a Polynomial: Simple Field Extensions Contents. 173 175 175 176 180, 180 181 185 193 194 198 205 207 212 218 222 223 226 229 231Contents Chapter 8 Polynomials and Roots A. Inventing Roots of Polynomials B. Finding Polynomials with Given Roots Chapter 9 Error Correcting Codes, II Chapter 10 Tsomorphisms, I A. Definitions B. Examples Involving Z C. Examples Involving F[x] D. Automorphisms Chapter II Finite Fields are Simple Chapter 12 Latin Squares Chapter 13 Irreducible Polynomials in Z,[x] A. Factoring x?" — x. B. Counting Irreducible Polynomials Chapter 14 Finite Fields Chapter 15 The Discriminant and Stickelberger’s Theorem A. The Discriminant B. Roots of Irreducible Polynomials in Z,[x} C. Stickelberger’s Theorem Chapter 16 Quadratic Residues A. Reduction to the Odd Prime Case B. The Legendre Symbol C. Proof of the Law of Quadratic Reciprocity Chapter 17 Duplicate Bridge Tournaments ‘A. Hadamard Matrices B. Duplicate Bridge Tournaments C. Bridge for 8 D. Bridge forp +1 Chapter 18 Algebraic Number Fields xu 237 237 238 242 255 255 257 259 261 264 267 273 273 275 280 282 282 287 288 291 291 293 296 8 309xiv ‘Chapter 19 Isomorphisms, IT Chapter 20 Sums of Two Squares Chapter 21 On Unique Factorization Exercises Used in Subsequent Chapters Comments on the Starred Problems References Index Contents 314 316 320 323 325 332 337I. INTEGERS This part of the book is about the natural numbers and integers. Among the highlights of this part, we show that every natural number factors uniquely into a product of primes, define congruence mod m, and invent new sets called congruence classes mod m, which for each m > 2 add and multiply to form a new algebraic system called Z,,. Various related results about numbers, and applications, fill out this part.Numbers This book is about the algebraic structure of sets of numbers and poly- nomials. Here are the most familiar sets of numbers and the way we shall denote them. N is the set of natural or counting numbers: 1,2,3,4,5 0... Z is the set of integers: 22,73, -2,-10,14234 0... Zis obtained from N by including 0 (zero) and the negatives of the natural numbers. Q is the set of rational numbers, that is, the set of all fractions a/b, where 6 #0 and a are integers. The set of integers Z is thought of as a subset of @ by identifying the integer a with a/1. Q is the smallest set containing Z in which every nonzero integer has an inverse, or reciprocal. (See Exercise E6.) R is the set of real numbers. A useful way to think of R is as the set of all infinite decimals. Real numbers can also be viewed geometrically, as lengths of directed line segments on a given straight line, or equivalently as coordinates on a given line (such as the x-axis used in calculus). Any rational number is a real number; the decimal expansion of a rational number a/b may be obtained by the familiar process of dividing b into a. One way to define the real numbers is as limits of sequences (Cauchy sequences) of rational numbers. (In fact, an infinite decimal .a,a)a, ... is a limit of finite decimals .a,a, . . . a,, and each such is a rational number.) However, it is possible to use real numbers without having a precise understanding of how they are defined. Calculus, which uses real numbers in an essential way, was invented and used for 200 years before a rigorous4 1 Numbers definition of real numbers was achieved. So we shall adopt the 18th- century point of view and assume that the real numbers are understood, at least intuitively if not rigorously, and we shall say no more about their definition in this book C is the set of complex numbers, that is, the set of numbers of the form a+t bi, a, b in R, where i=V—1. You are probably less familiar with complex numbers than with real or rational numbers. Historically, they only became well understood and accepted early in the 19th century, well over a century after the development of the calculus. We may think of C as the set of vectors (= directed line segments) in the plane, with a + bi corresponding to the vector from the origin to the point with coordinates (a, 6). If v =a + bi, a is called the real part of v, and b the imaginary part of v. A convenient way to represent elements of C is in terms of polar coordinates (see diagram). If r is the distance from the origin to the point (a, b) (that is, r= Va? + B° ) and @ is the angle (measured counterclock- wise) from the positive real axis to the vector v, then a=r cos 8, b= rsin 0, so v=a+bi=r cos @ + ir sin 8. The real number r is called the length of v and is denoted |v|; @ is called the argument of v and is sometimes denoted arg v. Multiplication of complex numbers when de- scribed in polar coordinates works rather neatly: (r cos 6 + ir sin @ )(s cos @ + is sin ¢) rs((cos # cos @ — sin @ sin $) + i(cos # sin @+sin 8 cos })) (1) rs(cos(# + ) + i sin(@ + )) That is, when multiplying two complex numbers, lengths multiply and arguments add. imaginary axis (a,b) real axis complex plane If you have had some acquaintance with infinite series, then you probably know the Taylor series for e* = exp(x), ea ltxt a > as well as for sin x and cos x, sinx = x- 24H SEX at Sat1 Numbers 5 and vox x cox = loot a at which converge for all real x. Then cos x + i sin x has a Taylor series xt it at ied Ita tata at which would be the same as the Taylor series for the complex function e* if we knew what e’* was. So we define e™ for x real by setting e = cos x +i sin x: that is, we define e* by replacing x by ix in the Taylor series for e. Then an arbitrary complex number v can be written in polar form as v = r(cos6 +i sind) = re; if w is another complex number, w = se’*, the multiplication of formula (1) above becomes vw = reMse® = psei+9), which is exactly what one would expect from the laws of exponents that you learned in high school. From an algebraic point of view Q, R, and € are different from Z. In Z the equation ax = | can be solved only when a = | or —1, but in Q, R, or € (see E5) every nonzero element has a reciprocal, or inverse. Thus, as we shall see, Q, R, and C are examples of fields of numbers, while Z is not. We conclude this introductory chapter with a remark on logical nota- tion. Frequently we shall make statements like “P is true iff Q is true,” where P, Q are assertions. “iff” means “if and only if.” The statement means: “If P is true then Q is true and if P is false then Q is false.” If an exercise contains a statement to be proved which includes an “iff,” there are really two statements which are to be proved. See Exercise E7 for an example. El. Find inverses of: 1+ i; 3—2i; 1+ 67. E2, Write in polar form: (1 + )/2; (—1- V— 3/2. E3. Solve in C these equations: x? = 1; x° = 1; x"=1. g Prove: (cos @ + i sin 8)" = cos nO +i sin nd. E5, What is the inverse of re'*(r + 0)? TE6. Let a, b, c, d, e, f be integers with b, d, f 0. For ordered pairs (a, 6) and (c, d) define the notation (a, 6) = (c, d) by (a,b) = (c,d) iff ad = be. Show: () (a, 6)= (a, 6); (i) if (a, ) =(c, d), then (c, d) =(a, b); Gi) if (a, 6) =(c, d) and (c, d) =(e, f), then (a, 6) =(e, f).Ev. 3 1 Numbers (These conditions are called reflexivity, symmetry, and transitivity respec- tively. If = is a relation such that (i), (ii), (iii) hold, then = is called an equivalence relation. The set of all pairs (c, d) which are = to (a, ) is called the equivalence class of (a, 6). For = as defined here, the equivalence class of (a, b) is denoted a/b. Thus, a/b = c/d means: the equivalence class of (a, 6) is the same as the equivalence class of (c,d); this in turn means that (a,6)=(c, 4), which in turn means that ad= be. The set @ of rational numbers is defined to be the set of equivalence classes with respect to = of ordered pairs (a, 6) of integers with b #0.) Is the following fact from calculus true? If so, how is it proved? Let f(x) be a funetion which is differentiable for all x, a no If (a) P(g) is true, and (b) If P(n) is true, then P(n + 1) is true for all n > m in Z then P(n) is true for all n > ng in Z. You have probably seen this principle used before, perhaps in evaluat- ing sums arising in connection with the definite integral. Here are some examples of arguments by induction. Example 1.14+3454+--- +(2n—-1)=7? foralln> 1. Proor. Denote the assertion by P(n). The statement P(1) is true, since 1 =, Assume that for some number k, P(A) is true, that is, ]+3+5+ + + (2k — 1) = k2, We look at the left side of the assertion P(k + 1): 1TH345+-5-+(2kK+1)=[143454+--++(2k-1] + Qk).8 2 Induction; the Binomial Theorem By assumption that P(k) is true we can replace the expression inside the square brackets by &?: =k 4+ (2k+1) = (k+l). Thus assuming P(K) is true, it follows that P(k + 1) is true. By induction P(n) is true for all n > 1. oO The rationale behind induction is that if statements (a) and (b) are true, then for any n > mp one can prove, in n — no Steps, that P(n) is true. For example, if P(n) is the equation (1) above and we wish to prove that P(5) is true, we can argue logically using statements (a) and (b), as follows: P(1) is true, by statement (a). Since P(1) is true, P(2) is true, by statement (b) with n Since P(2) is true, P(3) is true, by statement (b) with . Since P(3) is true, P(4) is true, by statement (b) with n = 3. Since P(4) is true, P(5) is true, by statement (b) with n = 4. This same kind of argument can be done, once we have shown (a) P(1) is true, and (b) For all n, if P(n) is true, then P(n + 1) is true, to show in n— | logical steps that P(m) is true for any n> 1. The principle of induction simply asserts that given (a) and (b), and any n= 1, P(n) can be shown true, and that therefore P(m) is true for any n. Here are some more examples Exampte 2. For all n > 1, 2" > 1 +n. Proor. The statement is clearly true when n = 1. Supposing that 2 > 1 + k for some k > 1, then it follows that 24+! = 24.2 >(1+k)2=2k+2>1 +(k +1). Thus the inequality is true for all n > 1 by induction. go ExaMpte 3. For all n > 0, 8 divides 32" — 1 (that is, 3%" — 1 = 8k for some integer k). Poor. The statement is true for m= 0 since 8 divides 3°— 1 = 0, Suppose 8 divides 3° — 1. We examine 24+) — |: uD) | a Pp ae. PPE R= 32(3% — 1) + (37-1). Since 8 divides 3-1 and 8=3?— 1, therefore 8 divides 3°(3** ~ 1) + (# = 1) =36+" — |. Thus the statement is true for all n > 0. o EXxaMpLe 4. For all n > — 2, 2n? ~3n? +n +31>0.A Induction 9 PROOF. Set f(n) = 2n? — 3n? +n + 31. Then f(—2) = 1 > 0. Suppose f(k) > 0 for some k > — 2. Then AK +1) = Uk+ 1 — 3K + 1P + (kK +1) +31 = 2k +3K7 43k +1) — KP + 2k +1) + (K+ 1) + 31 = (2k3—3k7 +k +31) + 6k? + 6k +2-6k-341 = flk) + 6k? > f(k) > 0. So fin) > Ofor alln > — 2. a EXAMPLE 5, dx"/dx = nx"~! for any integer n > 0. PRoor. 7 = 0 is clear. If dx*/dx = kx*—! then (x+Ax)i— xt lim = kxt lh; Ax30 Ax thus dxht! de El. Prove that for all n> 4, n! > 2”. E2. Prove that P+ 23+ --- +0 =[n(n + 1)/2f for all n > 1. E3. Prove that for any integers x, y, x —y divides x"— y" for all n> 1. EA. (a) Prove that Slt xtxt pee pal 4 T=x Ix for any n> 1 (b) Prove that 1+2+2?+---+2""!=2"—1 for any n= 1. () Given any integer n>1 and any integers a, m,7,---yfq—1 With a>2 and 0 1 42th tat a(at DQat yn +30)10 2 Induction; the Binomial Theorem E6, Consider the ancient puzzle called the Tower of Hanoi. The puzzle consists of n disks of decreasing diameter placed on a pole, There are two other poles. The problem is to move the entire pile to another pole by moving one disk at a time to any other pole, except that no disk may be placed on top of a smaller disk. Find a formula for the least number of moves needed to move n disks from one pole to another, and prove the formula by induction. E7. What is wrong with the following theorem and proof? Theorem. All babies have the same color eyes. PRoor. n= | is obvious. Suppose that in any set of n babies all have the same color eyes. Consider a set of n + | babies. OOSGGQS We may assume by induction that in the set L of the n babies to the left all have the same color eyes, and similarly that in the set R of the n babies to the right all have the same color eyes. But then evidently all the n +1 babies have the same color eyes, for the leftmost and rightmost babies have the same color eyes as all the babies in between. By induction, for any n, in every set of n babies all have the same color eyes. Since the set of all babies is one such set, the theorem is proved. [JB Another Form of Induction IL B. Another Form of Induction Induction (2). Let P(n) be a statement which makes sense for any integer n> ny. If (a’) Pi) is true, and (b’) For all n> ng, if P(m) is true for all m, ny ng. Here is an example of the use of induction (2), an example which is not so easily done using induction (1): ExampLe 6. Any natural number n > 2 factors into a product of prime numbers. Recall that a natural number n is prime if n does not factor into the product of two natural numbers each smaller than n. We adopt the convention that a prime is a product of (one!) primes. Proor. Let P(n) be the statement “n is a product of primes.” Then P(2) is true since 2 is prime (hence is a product of primes by our convention). Suppose P(m) is true for all m 2. a Note. The statement of Example (6) is part of the fundamental theorem of arithmetic, which we will prove in full in Chapter 3. Theorem. /f induction (1) is true then induction (2) is true. Proor. Let P(n) be an assertion and suppose we know of it: (a’) P(np) is true; (b’) For all n> 7p, if P(m) is true for all m, ng no, if Q(n — 1) is true then Q(n) is true. Assertion (a) is the same as (a’), so is true. For (b), suppose Q(n — 1) is true, Then P(m) is true for all-m, ny | ‘E10. The Fibonacci sequence is an example of an inductively defined sequence of numbers. It is defined as {a9, a), a3,... } where ay =0, a= 1, and a,= a,_\ + 4,2 for n > 2. Thus it begins 0112358 13 21 3455 89 (a) Prove by induction that a, and a,4, have no common factor >0 except 1. The golden mean is the ratio ba such that b:a=(a+5): 6. If we consider a fraction b/a where b : a is the golden mean, then b_atb a b $0C Well-ordering 13 The golden mean b : a was considered to be the most perfect ratio for the lengths of the sides of rectangular designs, such as portraits, The golden mean relates to the Fibonacci sequence by the following two facts: (b) a,,= (e+ sallo-eal" (prove this by induction); (©) lim Suet = Los (prove this using (b)) It follows from (c) that the golden mean can be approximated by ratios of consecutive numbers in the Fibonacci sequence. C. Well-ordering Here is another version of induction. Well-ordering principle. Let ny be any fixed integer. Then any nonempty set of integers > no has a least element. We prove the well-ordering principle using induction (1). To do so we prove that if there is a set of integers > ny with no least element, then it must be empty. (This approach uses the standard logical strategy for proving statements of the form “if A then B”—we prove that if B is false then A must be false. The reason that the strategy works is that the only situation under which the statement “if A then B” is false occurs when A is true and B is false. If we assume B is false and are able to show thereby that A is false, then the situation “4 true and B false” cannot hold and so “if A then B” is true.) ProoF. Let S be a set of integers > my with no least element. Let P(m) be the statement: “No number mq we will show that S is empty, which will prove the principle. Evidently P(19) is true, for if not, np is in S, and since all numbers in S are > no, therefore S would have a least element. Suppose P(k) is true for some k > no. If P(k +1) is false, then some number 0 and 6 there exist integers q > 0 and r with 0 0, b — ax > 0}. Since 6 = 6 — a-Ois in S, Sis a nonempty set of integers > 0 (ng = 0). So S has a least element, r. We show r satisfies the conditions of the theorem. Since r is in S, r > 0, and r= b— aq for some q > 0. We must show that r a, then r—a=6 — a(q + 1) > 0 would be in S, But r—a 0, so it must be that r 0. The binomial theorem. For n any integer > 1, cern (§)e+(Ievt “(pees where n\___al () = aGen for O 0 assume that for all r, O | and the theorem is true for n— 1, that is, (xty)" b= ("Se +("7 ero + Lae + ("7 aerhyr ee +("2! md, We compute (x + y)" as follows: (x ty)" = (ety) (ety) = x(x ty) + (ety)D The Binomial Theorem 7 substituting for (x + y)"~! we get meter a (™ Narva (7 Lget ("5") ("Ip yee (Sto apie (hosp + (Sop Collecting coefficients, this becomes ase (oat) eee n-1)\L(\. n—1)\_/n)_ By the lemma, and the fact that ( ° Je(a)-8 (TaN -()=1 this becomes Yon nonyr ” (9) te (Maton g ee + (My, which proves the binomial theorem by induction. oO E13. Prove that the sum of the elements of the nth row of Pascal’s triangle is 2” for each n. How many subsets of a set with n elements are there? E14. Prove that (C2) #0) =F) for any s, N >s. E15, Prove that V (MP a (TP a ep = (20 (0) + (1) t+ GY = (7) E16. Let ,5, denote the sum of the first rth powers: Wee DRS VE bo tat mr You know ,63 by E2, and ,s by ES. (a) Find ,5). (b) Find jsp. *(o) Prove that for each r > 1, (nt It! = (ntl) = ("¢!) . r+ + (7t ys. *(d) Derive from (c) that ,.5, is a polynomial function of n whose degree is r+. (©) View ,5, as the sum of the areas of a collection of rectangles lying inside the region between the x axis and the curve y = x" between x= land x= + I. Does (d) seem reasonable when ,.5, is viewed this way? (You may wish to recall the integral test if you have studied infinite series.)18 EIT. E18. 2 Induction; the Binomial Theorem Multinomial theorem, Yi you multiply out (x; + +++ +," and collect coefficients you get a sum in which each term has the form (ayn (2am) where aj +a) + +++ +a, =2, and each a, > 0. Show that (4, . (where 0!= 1) (a) when m= 3, (b) for any m > 2. (©) Find the coefficient of x? in (x? + 3x +2), xh for some coefficient Let a be an integer > 2. Prove that for any > 0 there exist unique integers Jos Pipes, With Or, 0 and b > 0 there exist unique integers q> 0 and r,0 0 there exist g, r with 0 0 we use induction (2). Suppose that for all c, 0a, let c=b—a. Then 0 O may be written as b= a-q +r with 0 r and subtract b = aq’ +r’ from b= aq +r to get a(q — 9’) +(r—r')=0, or aq — q') =r’ — r. Since r’—r>0 and a> 0,q¢-—q' 20. Also, r —ra) using Euclid’s algorithm, (i) Let {q,} be the Fibonacci sequence. What is N(d,4, 4,)? Gi) Suppose that a is an integer a, N(b, a) < N(@qx» &n)- (Hint: Use induction.) B. Greatest Common Divisors We observed that the last nonzero remainder in Euclid’s algorithm applied to a and b is the greatest common divisor of a and b. Thus finding the greatest common divisor is an effective computational process. Euclid’s algorithm also has the following useful consequence sometimes called Bezout’s Identity. [f the greatest common divisor of a and b is d, then d= ax + by for some integers x and y.B Greatest Common Divisors 23 Before we show how this is done, hide the rest of this page and try the numbers 365 and 1876, It is easy to see they are relatively prime (365 = 5+ 73 and neither 5 nor 73 divides 1876). Try to write 1 = 365x + 1876y for some integers x and y. It is not obvious how to do this! Here is how. Do the Euclidean algorithm: 1876 = 365-5+ 51, 365 = 51-748, 51 =8-6+3, 8=3-24+2, 3=2-1+1; so | is the greatest common divisor. Now solve for the remainders (the remainders are in boldface to distinguish them from the quotients) 1=3-2-1, 2=8-3-2, 3=51-8-6, 8 = 365- 51-7, 51 = 1876 — 365-5, and successively substitute the remainders into the equation 1=3-2-1, starting with 2: 1=3-2 =3-(8-3-2)=3-3-8 3($1-8-6) -8 =3-51—8-19 3-51 — 19(365 — 51-7) = 136-51 — 19-365 136(1876 — 5-365) — 19 - 365 = 136 - 1876 — 699 - 365. So x = — 699, y = 136. Another, easier, way to write d= (a, b) as d= ax + by is to keep track of how to write each successive remainder in Euclid’s algorithm in terms of aand 4, using the following layout illustrated with 1876 and 365. The first column consists of numbers obtained as remainders in Euclid’s algorithm, together with a and 5; the second column describes the coefficient x of 365; and the third column describes the coefficient y of 1876, when writing the number in the first column in terms of 365 and 1876.24 3 Unique Factorization into Products of Primes r= x-365 + y+ 1876 r x y 1876 0 1 365 1 0 1876 = 365-5 +51 1876 0 1 365-5 3 0 : subtract a =) 1 365= S1-7 +8 365 1 0 51-7 —35 7 : subtract 8 36 -7 Sl= 8-6 +3 51 -5 1 8-6 216 -42 : subtract 221 43 8= 3-2 +2 8 36 -7 3-2 —442 86 : subtract 2 478 93 3= 2-1+ 1 3 -221 4B 2 478 -93 : subtract 1 — 699 136 Therefore, 1 = — 699 - 365 + 136 - 1876. Here is a more formal proof of the last assertion of Euclid’s algorithm and of Bezout’s identity using induction (1). Theorem. If ry is the last nonzero remainder in Euclid’s algorithm for a and b, then ry is the greatest common divisor of a and b and ry = ax + by for some x,y PRoor. If ry is the last nonzero remainder in Euclid’s algorithm for a and 6, then the number of steps in the algorithm is N + 1. We shall prove the theorem by induction on N. If N = 0, then a divides 6, and the theorem is trivial, If N = 1, then Euclid’s algorithm for a and 6 has the form: b=aq+n, a=rq,+0. Then it is easy to see that r, is the greatest common divisor of a and 5; also r= 6-1 + a-(—q,), so Bezout’s identity holds. Assume the theorem is true for N = 7 — 1, so that the theorem is true for any two numbers whose Euclid’s algorithm takes n steps. SupposeEuclid’s algorithm takes +1 steps for a and 6. If b= aq, +7, then the rest of the algorithm for a and b is Euclid’s algorithm for r; and a. So Euclid’s algorithm takes n steps for r, and a. If r, is the last nonzero remainder for Euclid’s algorithm applied to a and b, it is the last nonzero remainder for Euclid’s algorithm applied to r, and a. By induction, r,, is the greatest common divisor of a and r), and r, = au + rv. Now b = aq, + ry, so it is very easy to see that r,, the greatest common divisor of a and r,, is also the greatest common divisor of 6 and a. Moreover, substituting for r, in the expression r, = au + r,v yields r, = au + v(b — aq,) = a(u — 0q,) + bv. So Bezout’s identity holds. The theorem is true by induction (1). Oo Corollary. If e divides a and e divides b, then e divides (a, b). 9. Why is the above corollary true? E10. Write the g.c.d. of a and b in terms of a and b as in Bezout’s identity, where a and b are: (i) 267 and 112; (ii) 222 and 1870; ii) 500 and 11312; (iv) 11312 and 11213; (v) 135 and 216. Ell. Solve (i) 283x + 1722 =31, (ii) 365x + 72y = 18, Gi) 11x + 2345y E12. Prove: If a divides and a divides c then a divides bx + cy, for any x, y in Z. 6. E13. What is the g.c.d. of a, b, c? Call it (a, 6, c). Prove that ((a, b), ¢) =(a, (b, ¢)) = (a,b, 0) El4, Let d be the gcd. of a),a3,...,a,. Prove: d=a,x, + ++» +a,x, for some integers x4,...4%,- EIS. Prove: (a, b) = (a, b + ax) for any integer x. E16. Prove: m(a, 6) = (ma, mb). E17. Prove: If (a, r) = d and (b, r)= 1, then (ab, r) = d. E18, (13750, 12222) =? E19. (336366, 488448) =? E20. Prove: If d=(a, 6) and ar + bs = d, then (r,s) = E21. You are given two “hour”-glasses—-a 6-minute and an I -minute hourglass— and you wish to measure 13 minytes. How do you do it? E22. Suppose axy + byy = 1. (a) Show that if x = x9 + nb, y = yo — na for some n, then ax + by = I. (b) Show, conversely, if ax, + by; = 1, then there is some integer n such that x, = xo + nb and y, = yo — na.26 3 Unique Factorization into Products of Primes C. Unique Factorization A natural number p > | is prime if the only divisor of p greater than 1 is p itself. Note. | is not prime, by convention. Primes are the building blocks of natural numbers, for Theorem. Any natural number > is prime or factors into a product of primes. You have no doubt factored many numbers into products of primes. For example, 368 = 2-2-2-2-23, 369 = 3-3-4], 370 = 2-5-37, 371 = 53°7, 372 =2-2-3-31, 373 = 373 (a prime number), 374 = 2-11-17. The proof was given in Chapter 2 as an example of induction. We are going to prove the fundamental theorem of arithmetic, namely, that factorization of a natural number into a product of primes is unique. What does “unique” mean? Suppose a is a natural number. If a=p,- +p, and a=q,-°~q,, are factorizations of a into products of primes, we shall say that the factoriza- tions are the same if the set of p,’s is the same as the set of qs (including repetitions), That is, m= and each prime occurs exactly as many times among the p;’s as it occurs among the q,’s. Thus we consider the factoriza- tions 2-2+3-31 and 3-2-31+2 as being the same, because each prime occurs an equal number of times in each; whereas 2: 3-3-3] and 2-3-2- 31 are different. Factorization of a is unique if any two factorizations of a are the same. Theorem. Any natural number > 2 factors uniquely into a product of primes. ProorF BY INDUCTION (2). Suppose that the result is true for all numbers less than a. Suppose a=p,- ++ p, and a=q,--- q,, ate two factoriza- tions of a. We want to show that the two factorizations are the same. Suppose it is true that p,; =, for some j. Then a/p,=p,: ++ p, = 1°" + G-14)41 7 + Gq: Since a/p, Im)C Unique Factorization 27 But since p; = q;, the set of primes {p,, 7, ...,P,} is then the same as the set of primes (4)... G1» Go Ga +++ Gud» SO the two factorizations of a are the same, and the result is true for the number a. That would prove the theorem by induction (2). We prove that if p,---p,=4,-~*q then p,=4, for some i by means of the following important Lemma 1. If p is a prime and p divides ab, then p divides a or p divides b. Proor. Let d=(p, a). If d > 1 then d=p since p is a prime. In this case, pla. If d= 1, by Bezout’s identity we can write l=ax+py for some integers x and y. Then 6 = bax + bpy. Now p divides ab and also divides p. So p divides bax + bpy = b, completing the proof of the lemma, Oo From the lemma it follows by induction (Exercise E32) that if a prime divides a product of m numbers it must divide one of the factors. To complete the proof of uniqueness of factorization, suppose we have PiP2*** P= 4%" * * Gm. Then p, divides q, - - - Yn. Since p, is prime, p, must divide one of the q’s, say q,. Since q, is prime, q; is divisible only by itself and 1. Since p, #1, p, = 4- Thus the induction argument described above for proving uniqueness of factorization can always be used, and the proof of uniqueness of factoriza- tion is complete. o The argument of Lemma | can be used to prove Lemma 2. Suppose a, b, ¢ are natural numbers > | such that a divides be. If the greatest common divisor of a and b is 1, then a divides c. E23, Prove Lemma 2. E24, Give six counterexamples to the assertion: If a divides bc and a does not divide b, then a divides c. The problem of factoring a natural number 7 into a product of primes is much harder in practice than the problem of finding the greatest common divisor of two numbers, One needs to find divisors of n, and that is a matter of trial and error. For example, to factor 3372, we first see that 2 is a divisor: 3372 =2- 1686. Then we look for a divisor of 1686: 1686 = 2: 843, so 3372 = 2-2. 843. Then we see that 3 divides 843: 843 = 3 - 281, so 3372 =2-2-+3-281. Finally, we check that 281 is not divisible by 2, 3, 5, 7, 11 or 13, and 17 > V281 , so 281 must be prime by exercise E25, and we have the factorization of 3372. There are obvious tricks for testing a number n to see whether it is divisible by 2, 3, or 5; we shall see later (Chapter 6) tests for 7, 11 and 13,28 3 Unique Factorization into Products of Primes In general, however, unless n happens to be prime it is a slow process looking for divisors. So much so that it is claimed that if one had a 126-digit number NV which happened to be the product of two unknown 63-digit prime numbers, and one wished to factor N, even using the best available (in 1977) methods and computers it would take about 40 x 10'° years, By comparison, finding the greatest common divisor of two 126-digit numbers would take seconds. For a discussion of techniques for factoring composite numbers, see Knuth (1969, pp. 338-360). Some of the classical tricks are in Ore (1948). E25, Prove that if n is not prime, m has a prime divisor < Va . E26. One can write down all primes <7 for any given n by the simple device, called the sieve of Eratosthenes, of writing down all the numbers > ph where p,,...,p, includes all primes which divide either m or n and some of the exponents may be zero.D Exponential Notation; Least Common Multiples 29 E27. Complete with a statement involving a's and 7s: () m divides nif... ; ii) the greatest common divisor of m and nis... ; it) the least common multiple of m and 7 is » (Gee E34); (iv) m/n is an integer if (v) m/n is an integer d ble by pif... . Here is an application of the fact that any natural number factors uniquely into a product of primes. Theorem. V2 is irrational. Proor. If not, ¥2 =6/a, a, 6 natural numbers. So 2a? = 6. Write a= pi! ++ pf, b= pf'+-+ pf. Then in the equation 2a?=6?, 2 occurs to an even power on the right side, and to an odd power on the left side, which is impossible. o E28, Prove: If (a, 5) = | and ab is a square, then a is a square. E29, Prove: If d=(a, 6), then {ax + by|x, y in Z) = {zd|z in Z). E30. If (a, 6) = 8, what are the possible values of (a®, b*)? E31. If (a, b) =p’, p prime, what is (a2, b?)? E32. Prove by induction that if a prime number p divides a,-a,+ + + a,, then p must divide one of the factors aj. E33, Show V1000 is irrational. Definition. The /east common multiple of two natural numbers a and 6 is the smallest natural number which is a common multiple of @ and 6, More precisely, m is the least common multiple of a and 6 if (1) alm and b|m, and (2) if alr and dlr, then m b, You want to bring back c quarts, where c is some integer, 0 <¢ < a+ 6, For which c can it be done? How? Prove that limp. ..(a, 6/(a”, b)) = 1. Show that 7700 is irrational. Let 20. Prove that there exist unique integers q and r with 0 2 and 6 with (a, b) = 1, there are infinitely many primes of the form an + b. We give a brief introduction to some of the ideas which go into the proof of that theorem in Section B.32 4 Primes FA. Here is another proof of the infinitude of primes, due to G. Polya. *(a) Show that 2”) + 1 and 22+ | are relatively prime if mn. (b) Use (a) to show that there are at least m primes less than 2°” + 1 for each n, hence there are infinitely many primes. ES. Prove that for any n there exist n consecutive natural numbers none of which are prime. (Hint: Start with (n + 1)!+2.) E6. Prove that for any n there is a prime p with n dx m=0 Using this with x = 1/p,, we getB Some Analytic Results 33 Multiplying out the right side, Noe Fe E isimmoP moge0 yao Pi Now since p,,..., py are assumed to be all the primes, and any natural number is a product of primes, each natural number n has the form n= pj'- - » px for some e, - - - éy, uniquely. So the right-hand sum of (1) is the sum of all reciprocals 1/n such that no prime occurs in the prime factorization of n to an exponent greater than E. In particular, if n < E, 1/n is included in the right-hand side (why?). Thus w i 1 1 ——!] > — forany E > 0. ies) = n ¥ But lim, ,,,5£. (1/7) is infinite, and so choosing £ sufficiently large, SE_,(1/n) is larger than any fixed finite number, such as (a) We get a contradiction, proving the theorem. oO E7. Show that im Mtn) ~ Novo Euler's proof provides a starting point for a proof of Dirichlet’s theorem that given any a, b with (a, b)=1 there are infinitely many primes in the arithmetic progression an + b. The proof of Dirichlet’s theorem begins with the interesting observation, which we shall justify momentarily, that the infinite series > 1 p prime diverges. One wants to show that if P, is the set of primes of the form an + b, then pinP, diverges. This of course would mean that P, is an infinite set. To do that one shows that in fact the ratio i Pin Py pcn P 1 p prime p 00, where (a) is the number of natural numbers (-re(: + forO00. A much more precise result, called the prime number theorem, says: ax) wm, (flog x) This means that given any e > 0 there is some number N such that for any x>N, en |<« or |r) - lore 1 there 1s a prime p with n 1.a) A. Numbers in Base a Our way of writing numbers is biased in favor of the number 10, When we take a number 6, like b= MCMLXXVI, and write it as 6 = 1976, we mean that b=1X1P+9x 1P +7 x 10+ 6; we call this way of writing 6 the representation of b in base 10, or radix 10. If we write a number 6 as a sum of powers of a (a > 2), bea tye te brat ry with each of ro, ..., 7, between 0 and a — 1, this is the representation of b in base (or radix) a. There is no particular reason except convention or physiology why we have a bias towards 10. For example, we could write 6 = MCMLXXVI in base 2: b=1xWH 1x Pix Be ix Wei x P+ ix BA+1 KB, or b= 11110111000. If we used base 2 with all numbers, our favorite electronic computer, which understands “on” (1) and “off” (0) very well, would be very happy and efficient. Numbers themselves have no bias, as the following theorem shows. Theorem. Fix @ natural number a > 2. We may represent any integer b > 0 in base a: that is, b can be written uniquely as barattn a" ts trat+ rat ry with 0 @41 set q’ =the largest integer such that q’d, q—the guess q is always > the correct quotient q; (ii) if d, > b/2, then q > q’ — 2—the guess q' is equal either to q, to q + 1, or toq+2. Proor. Assume b is even (for b odd see Ell). For (i), since gde,,,, then qd, is defined to be the largest multiple of d, which is b/2. By choice of q it suffices to show that (q —2)d 6/2 and q! 6/2, the standard guess q’ will never be off by more than 2: either q’, q’ —1, or q’ —2 will be the correct quotient. Notice that it does not matter how large the base b is. To insure that the leading digit of the divisor is > 5/2, one “normal- izes.” Take the original divisor d and dividend e and multiply d and e by 2° where 2° is the largest power of 2 with 2d 8, where (a) p= (3, 11, Die g = (1 Hs 1S, 4 Ores (b) P= (1, 6, Dros @ = 9, 0. 13, 1, 10)r6- E13. How big must the leading digit of the divisor be in order that the guess 4 never differ from q by more than 3? E14, (a) In base 10 if we do not normalize, how far off can q’ be from the correct quotient q? (b) Same question in base b for any 6. D. Decimal Expansions We can do expansions in base @ also, In base 10 they are called decimal expansions, To begin with an example, to expand 1/7 into a decimal, divide 7 into 1 by the division algorithm, multiply the remainder (which is 1) by 10, divide that by 7 (with quotient 1), multiply the remainder (which is 3) by 10, divide that by 7 (with quotient 4), etc. It is what you do in long division: 14285... 7)1.00000 Thus 1/7 = 0.14285... .D Decimal Expansions 45 The base-free procedure for taking 6/c and finding its expansion in base a is the the same idea: Find b= c-q + rp where q is an integer and ry 0 congruent to a (mod m) where a and m are (i) 3312 (mod 4), (ii) 177 (mod 8), (ii) 31 (mod 37), (iv) 111 (mod 109). B. Basic Properties The congruence symbol looks like an equality, and this is no accident. In fact, one can view congruence geometrically as a kind of equality. Take the real number line and lay on it a circle of circumference n. Then wind the real number line around the circle. The picture shows n = 6. 3,9, 15 0, 6, 12, 18 Integers which lie over the same point on the circle are congruent mod n; integers lying over different points are not. Most of the things you normally can do with an equality you can do with congruence mod n. In particular, congruence satisfies these funda-C Divisibility Tricks 49 mental properties: (i if a=b (mod n) then ka (ii) if a=b (mod n) and b=c (mod n), then a=c (mod n); ii) if a=b (mod n) and a’ =b’ (mod n), then (iiia) a + a’ =b + B’ (mod n), and (iim) aa’ = 6’ (mod n). b (mod n); Alll of these follow easily from the condition that a=b (mod n) ifa= 6+ ng for some integer g. We prove (ii) and (iiim): PRooF OF (ii). if a= (mod n) and b=c (mod n), then a=b+sn,b=c + tn for some integers s and ¢, Substituting, we get a=(c + tn) +sn=c+ (multiple of n), so a=c (mod n). o PRooF oF (iim). if a=b (mod n) and a’ =5’ (mod n), then a= b + sn, a’ = b’ + In for some integers s and t. Then aa’ = bb’ + nsb’ + thn + stn? = bb’ + (multiple of 1). So aa’ = bb’ (mod n). oO E3. Prove (i) and (iiia). Property (iim) shows particularly well the virtue of the congruence notation. If we translate a =b (mod n) and a’ =b’ (mod n) into divisibil- ity notation, they read: n divides a — b and n divides a’ — b’. To conclude that n divides aa’ — bb’ is less natural than to conclude from a=b (mod n) and a’ =b’ (mod n) that aa’ = bb’ (mod n). The one thing which can be done with ordinary equality which cannot be done in general with congruences ‘mod n is cancellation: If ab =ac (mod n), it does not necessarily follow that b=c (mod n). For example, 2-1=2-3 (mod 4), but 1 #3 (mod 4). Similarly 6 = 12 (mod 6), but | # 2 (mod 12). We shall postpone for now the rule which replaces the usual rule of cancellation. Meanwhile, you try to invent a useful rule of your own. Do examples. FA, Prove that 6-4" =6 (mod 9) for all n > 0. ES. Prove that if n > 4 and n is not prime, then (n — 1)!=0 (mod n). C. Divisibility Tricks There is a very old trick to check addition and multiplication: sum the digits and do the operation on the sums of the digits. For example, multiplication: 3325 343424+5=13, 182 1+8+2=I1, 605150 6+0454+14+54+0=17,