MATHEMATICAL ANALYSISMATHEMATICAL ANALYSIS SECOND EDITION TOM M. APOSTOL California Institute of Technology ADDISON-WESLEY PUBLISHING COMPANY Reading, Massachusetts ‘Amsterdam + London + Manila + Singapore + Sydney + TokyoWORLD STUDENT SERIES EDITION FIFTH PRINTING 1981 ‘Acomplete and unabridged reprint of the original American textbook, ths Wold ‘Student Series edition may be sold only in those countries to which it is consigned by Addon Wedly or ts authorized trade datrbutor, I may not be re-exported from the country to which it has beon consigned, and jt may not be sold in the United States of America rts posessions Consulting Editor: Lyn Loomis Copyright © 1974 by Addison-Wesley Publishing Company, In. Philippines copy- tight 1974 by Addison-Wesley Publishing Company, Inc. ‘All sights reserved. No part of this publication may be reproduced, stored ina e- trieval system, or transmitted in any form or by any means, electronic, mechanical, ‘Photocopying, recording or otherwise, without the prior written permission of the ‘Original edition printed inthe United States of America. Published simul. {taneously in Canada. Library of Congress Catalog Card No. 72-1473. To my parentsPREFACE ‘A slance at the table of contents will reveal that this textbook treats topics in ‘analysis atthe “Advanced Calculus” level. The aim has been to provide a develop- ment of the subject which is hones, rigorous, up to date, and, at the same time, not too pedantic. The book provides a transition from elementary calculus to ‘advanced courses in real and complex function theory, and it introduces the reader to some of the abstract thinking that pervades modern analysis. “The second edition differs from the first in many respects. Point set topology is developed in the seting of general metric spaces as well asin Euclidean n-space, and two new chapters have been added on Lebesgue integration. The material on line integrals, vector analysis, an surface integrals bas been deleted. The order of some chapters has been rearranged, many sections have been completely rewritten, and several new exercises have been added. ‘The development of Lebesgue integration follows the Riess-Nagy approach which focuses directly on functions and their integrals and does not depend on measure theory. The treatment here is simplified, spread out, and somewhat rearranged for presentation at the undergraduate level. ‘The frst edition has been used in mathematics courses at a variety of levels, from first-year undergraduate to first-year graduate, both as text and as suppe= mentary tefereace. The second edition preserves this flexibility. For example, ‘Chapters | through 5,12, and 13 provide a course in differential calculus of func. tions of one or more variables. Chapters 6 through If, 14, and 15 provide a course integration theory. Many other combinations are possible; individual instractors ‘can choose topics to suit their needs by consulting the diagram on the next page, which displays the logical interdependence of the chapters. 1 would like to express my gratitude to the many people who have taken the trouble to write me about the first edition. Their comments and suggestions influenced the preparation of the second edition, Special thanks ate due Dr. Charalambos Aliprantis who carefully read the entire manuscript and made ‘numerous helpful suggestions. He also provided some of the new exercises. Finally, | would like to acknowledge my debt to the undergraduate students of Caltech whose enthusiasm for mathematics provided the original incentive for this work. Pasadena TMA. ‘September 1973"THE REAL AND COM. 2 AEA RSPR SoMe Basil, NoTIONS ‘OF Sor TaEORY CONTENTS ELEMENTS OF po Ser TOPOLOGY It (Chapter 1 The Real ad Complex Namie Sptes wrth ap 1.1 tntrootion 1 td 1.2 The field axioms 1 I 1.3. The onder axioms 2 H 14 Goometie representation of el numbers 3 penivirwves 15 Inerais 3 I 16 Ieper 4 f : 12 Tete erin rem fori 4 Foncroms fr soumseo 18 Rational numbers 6 AgIAtioN nN mae 19. tational aumbers © 7 ‘TviaBia: Conve 10 Upper bunds, makin cia, et opr Hot (spreoum secs 4 4 1.11 The completeness atiom | : Lilo yee, Senes aw acura oie. 1512 Some properties ofthe spremn 9 INET PRobuene PERENTSL CAUCUS 1513 Properties ofthe integer dese from the competenes axiom | 310 1514 The Archimedean property ofthe ea-nomter system fa) 1.15 Rational numbers with Bite decimal represetaton rn Sa isc Nc 1116 Finite dca approximation o real nambers : in sributvls INTOORAL AND TMU 1117 Infiie decinal represeatation of real nanbes fon | 18_Abpoote values andthe ingle newality 2 4 4 119 The Cauchy Scar inegalty 3 ; 7 1.20 Ph and mis infty and te extended rel number sysiemR* | 16 SRINSONS MOYNTEOIAS 121 Complex sambers 4 122 Grane repent of coms mane 123. Theimagiary ont uy 124 Abwolute vale of « comple aimer loa Tie Bsa 125 Ingmar oorne compres ze 126 Complex exponentials fills | 1.27 Ruther proper of complex exponcatiais Er a 128 Thearpunentofacompkx number = = ss 129 Integral powers and rots ofeomplex numbers 2S Lat POUnIEH UNcaeals, 1.30 Complex logarithms . . . 2 131 Complex powers 2 ia i 132 Complex snes and cies : i caucus titponest aD swormipas uenesaue 133 Infinity and the extended complex lane C>- » Sis HaSIOUE CALS | TWeHtats eee FyCoapter 2 2 23 24 28 26 a 28 29 210 2 22 2 24 2us Elements of Polat Set Topology Itroduction Euclidean space Rt ‘Open balls and open sts in RO ‘The structure of open sets in Rt Goved sts Adherent poists. Accumulation points - (Closed se aed adherent points ‘The Boteano-Weienras theorem ‘The Cantor inteneon theorem ‘The Lindslof covering theorem Be Heo Bol ovine eae Compactnes in : Metrc spaces. Point set topology in metic spaces Cone set omic ace Boundary of set Erercnes ‘mts and Continity Introduction oe ‘Convergent sequences in a mettle Space Cauchy sequences. ‘Complete metric spaces : Limit ofa funetion : Limits of complex-valued functions 47 48 49 Onaper 6 64 62 63. 4 65 Contents Limits of vectorvatued fonctions Continuous fusctions Continuity of composite functions Centos complsaued nd vetoed intone : [Examples of continuous functions. Continuity and inverse ages of ope of cloned et ‘Functions continuous on compact sets “Topolgial mains (homeomorphisms) olvano's theorem : Connectedness ‘Components of a metic space |. ‘Arciie connectedness : ‘Uniform cootinwty - ‘Usiform cootioity and compact sets Fined-point theorem for contractions Discontinuities of real-valued functions ‘Monotonie functions... F Exercises » a5 Dectatives Introduetion . Deinition of derivative - : Derivatives and continuity | |. ‘Alacra of derivatives ‘The chain rule One-sided derivatives and infinite derivatives Functions with nonzero derivative Zeco derivatives and Tocal extrema Roles theorem ‘The Meas. Value Theorem for derivatives Tntermediate-value theorem for derivatives Taylor's formula with remainder Derivatives of vector-valued functions Partial derivatives 7 Differentiation of functions ofa complex variable ‘The Cauchy-Riemann equations Exercises cae ‘Functions of Bounded Variation and Rectifable Curves Introduction - Properties of monotonic functions ‘Functions of bounded variation ‘Total variation, ‘Additive property of total variation x SESNeeesReeseesyas m ira] 128 1 13066 61 oa 69 610 sit 62 (Chapter 7 1" 22 33 14 13 16 7 7a 19 2.10 ™m m2 m3 74 ras 16 17 78 19 10 72 im 72 124 725 126 1m Chapter 8 a a2 rey a4 as Caen Total variation on fa, x}. a function of x. ry Function of founded varaton exesed at the sites of increasing functions. - . 132 ‘Continuous functions of bounded variauon | |) ) ) 132 ‘Curves and paths : do 13 Recifiabe paths and arcleagti || : be ‘Additive and continuity properties of ar length us Equivalence of paths. Chang: of parameter. | |. | 136 Exercises. Peco es 3 4 at ‘The Ricmano-Stieijes Integral Introdection . coo05 Notation “The definition of the Rion Sis ‘sega Linear properties Integration by parts ‘Change of variable in a Riemann-Stetjes integral [Reduction o a Riemasn integral . Step functions as integrators Reduction of a Riemann kes ine inte wm Euler's summation forma Monotonically increasing integrator. | Uppecandlowet integrals» Addiuive aol linesty proper. of upper and lower integra Riemano’s condition See Comparison theorems Integrators of bounded variation Sulficient conditions for existence of Riemans-Siieles integrals ‘Necessary conditions for existence of Riemann-Stielj integrals. Mean Valve Theorem f+ Riemann ina ‘The integral asa funetion ofthe interval. ‘Second fundamental theorem of integral calculus Change of variable ina Riemann integral. ‘Second Mean-Value Theorem for Riemann integrals Ricmann-Stilfjes integrals depending oa a parameter Differentiation under the integral sign Interchanging the order of integration 5 Lebesgue’ criterion for existence of Riemann integrals Compecnived Kievan Stes neers Exercises . Innit Series and Intinte Products troduction 5 183 Convergent and divergent sequences of complex numbers - 183 Limit superior and 184 Monotonic sequences of re] numbers 135 Init series . : 185 a6 a7 a8 39 10 Bil a2 a3 aid Bis ais 817 a8 B19 820 21 an an. 82 #25, 826 827 9.10 oun 912 913 94 sas 9116 97 918 919 924 Catone Tsering and removing parentheses ‘Alternating eeries “Absolute and conditional convergence Real and imaginary parss of a complex series “Tess for convergence of series with positive ters ‘The peomeric series ae ‘The integral test ‘The big oh and litle oh notation — ‘The ratio test and the root text Dirchle’s text and Abel's test. Pia sa of the ome ois Eon tit ie Rearrangements of series. igman's foram on conden convergent series Subseries ‘Double sequences : Double series Rearrangement theorem for double series Asufficient condition for equality of iterated series. ‘Muhipication of series. Cesiro summability : Infinite products. « [Euler's product for the Riemann zeta function Exerciser ‘Sequences of Functions Pointwise convergence of sequences of functions Examples of sequences of real-valued functions Definition of uniform convergence Uniform convergence and continuity |< “The Cauchy condition for uniform convergence Uniform convergence of infinite series of functions» A spacefiling curve ‘Uniform convergence and Rierann-Stieie interation Nonutifomlycomerget queen ht canbe intepratad term by ‘Uniform convergence and difeentiation ‘Salficient conditions for uniform convergence ofa sxics Reciprocal of a power series Real power series ‘The Taylor's series generated by a function ‘emstein’s theorem ‘The binomial series 28 219 at ma as 20 21 24 BB 2» cm 2om 923 Copter 10 101 102 103 104 105 106 107 103 109 1010 ion 1012 1013 1014 ous 1016 1017 1018 1019 1020 1021 102 1023 1024 3025 wn ‘Abel's iit theorem ‘Tanber's theorem ‘The Lebesgue Integral Introduction. - ot ee ‘The integral ofa step function Monotonic sequences of step functions Upper functions and their integrals. ‘Rieanann-integrable functions as examples of upper functions |. ‘The class of Lebeogue-integrable functions on a general interval ‘Basic properties of the Lebesgue integral... - Lebesgue integration and sets of measure zero. || ‘The Levi monotone convergence theorems. a “The Lebesgue dominated convergence theorem Applications of Lebesgue's dominated convergeace theorem ete negra on nbounded inal Ets of eg on Yvounded intervals... ae Improper Riemann integrals Meagurable functions. Continuity of functions defined by Lebesgue integrals Differentiation under the integral sign Tnterchanging the order of integration ‘Measurable sets on the real ine. ‘The Lebesque integral over arbitrary subsets of Lebesgue integrals of complex-valued functions . Inner products and nonns . ‘The set 4() of squareintegrable functions - ‘The set (0) as a semimettic space ‘A omerenge hese fo Sis of acon 13) ‘The Riest Fischer theorem. Brercias ao Fourier Series and Fourier Integrals Introduction... ‘Orthogonal systems of fnctions = “The theorem on best approximation ‘The Fou ses offen in on toner ont Properties ofthe Fourier coefficients “The Ricsz~Fischer theorem Tas cong ad epreenation poten or wignometic in ‘The Riemann-Lebesque lemma ‘The Dirichlet integrals. ‘a ner repent ft he patil of «Po ese ‘Riemann’ localization theorem SSE SSBRSRBREER BUSsBRsBEeR ESS SeSeseseses a2 a3 na mas 16 nu nas m9 120 m2 nz (Chapter 12 2 2 23 n4 25 126 128 129 210 Rn Ri 3 nia Chapter 13, Ba 42 na ba Bs G6 a7 Chapter 14 Mat 42 Content Sufficient conditions for convergence of Fourier sxis at a particular point Csiro suramability of Fourier series. Consequences of Fes theorem ‘The Weierstrass approximation theorem Other forms of Fourier series ‘The Fourier integral theorem < “The exponential form of the Fourier integral theorem Integral transforms Comvolutions The convolution theorem for Fourier transforms ‘The Poision summation formula Exercises Malivariable Diterntia Caleutus Introduction “The dicectional derivative Directional derivatives and continuity “The total derivative The total deri ‘An application to complex "The matrix of linear fonction ‘The Jacobian matrix “The chain rule Mate form of the chain rule ‘The Mean: Value Theorem for diferentiable functions A sufficient condition for diferentibility {A sulficent condition for equality of mixed partial derivatives “Taylors formula for functions from RY 10 RE Exercises plicit Fonetons and Extremum Problems Introduction Functions with nonzero Jacobian determinant - “The inverse function theorera ‘The implicit fonction theorem Extrema of real-valued functions of one variable Extrema of real-valued function of several variables _Extremum problems with side conditions Exercises Multiple Riemann Ttegrals Introduction ‘The measure of « bounded interval ia RY 36 38 32 33 35 a6 340 ae 388 3843 14a us 46 147 43 49 1410 Chapter 15 15a 152 133 154 153 156 130 138 159 1510 isa 1512 1513 Chapter 16 161 162 163 164 63 166 167 168 169 16.10 161 1612 1613 16.18 165 Coane ‘The Riemann integral of a bounded function defined on # compact interval in Se fear oa Tabeng cris fresno maine “Riemann integral 5 Evaluation ofa rmitipl integral by iterated integration Jordan-meagurale sets ia R" “Multiple integration over Jordan measurable sots Jordan content exprened as « Riemann integral ‘Additive property ofthe Riemann integral Mean-Value Theorem for multiple integrals Maltiple Lebesrwe Integrats Tntroduetion os fe ‘Step functions and ther Integrals - : : ‘Upper functions and Lebespue-intgrable functions : ‘Measurable functions and measurable sets ia R™ | ‘i bi reton tere forte double of se fncton ‘Some properties of sts of meature 220. Fubin's reduction theorem for double integrals “The Tonelli-Hotson test for integrability Coordinate wansformations ‘The transformation formula for multiple inuegrais Proof of the transformation formula for linear coordinate transforma: tions. Proof of tie transformation formula for the characteristic incon ofa compact cube. ample of roo of he tomato emule Gg Exercises : Cauchy's Theorem and the Reside Cabelas Analytic functions. fo Paths and curves in the complex plane Contour integrals. eine alon cca ath fn fhe adi CCanchy’s integral theorem for a circle a Homotopic curves - : Invariance of contour integrals under homotopy ‘General form of Cauchy's integral theorem . Cauchy's integral formal. ‘The winding number of a circuit with respect toa point ‘Pee abound of he et foi wit wind number ro . ‘Analytic functions defined by contour itegrals . : Power-series expansions for analytic functions - (Cauchy's inequalities. Liouvil's theorem Isolation of the zeros of an analytic function 16.16 1617 1618 1619 1620 1621 1622 1623, 1624 1625 1626 1627 Contents “The identity theorem for analytic functions The maximum and minim modulus of an analytic fonction ‘The open mapping theorem Laurent expansions for functions analytic in an annulus Isolated singularities. The residue ofa function at an isolated singular point | The Cauchy residue theorem. (Counting zr0s and poles ina region | Evaluation of real-valued integrals by means of residues Evaluation of Gauss’ sum by residue calculus ‘Application ofthe reside theorem tothe inversion formula for Laplace transforms Des : -- Conformal mappings Index of Special Symbots Index© CHAPTER THE REAL AND COMPLEX NUMBER SYSTEMS AL INTRODUCTION ‘Mathematical analysis studies concepts related in some way to real number, 90 ‘we begin our study of analysis with a discussion ofthe real-number system. Several methods are used to introduce real numbers. One method starts with the positive integers 1,2, 3,... a8 undefined concepts and uses them to build a larger system, the postive rational numbers (quotients of positive integers), their negatives, and zero. ‘The rational numbers, in tur, ee then used to construct the ‘irrational numbers, real numbers like V2 and nwhich are not rational. The rational and irrational aumbers together conatitte the real-number system. ‘Although these matters are an important part of the foundations of math- matics, they will not be described in detail here. AS a matter of fact, in mow phases ofanalyis its only the properties of real numbers that concern us, rather than the methods used to construct ther. Therefore, we Shall take the real numbers themselves as undefined objects satisfying certain axioms from which further properties willbe derived. Since the reader is probably familia with most of the ‘properties of real numbers discussed in the next few pages the presentation will ‘be rather brief. Its purpose is to review the important features and persuade the reader that, if t were necessary to do 60, all the properties could be traced back to the axioms. More detailed treatments can be found in the references atthe end ofthis chapter. For convenience we use some elementary set notation and terminology. Let ‘Sdenote a set (a collection of objets). The notation x € S smeans thatthe object x isin the set 5, and we write x S to indicate that x is notin S. A set Sis sui to bea subser of 7, end we write $ & T, if every object in Sis «also in 7. A set i called nonempty if t contains atleast one object. ‘We assume there exists a nonempty set R of objets, called real numbers, which satisfy the ten axioms listed below. The axioms fall ia a natural way into three groups which we refer to as the feld axioms, the order axioms, and the completeness axiom: (also called the least-upper bound axiom oc the axiom of contin). 1.2 THE FIELD AXIOMS ‘Along with the set R of real numbers we assume the existence of two operations, called addition and matiplcation, such that for every pair of real numbers x and y2 Real and Complex Number Systems Aut the sum x + y and the product xy are real numbers uniquely determined by x and y satisfying the following axioms. (In the axioms that appear below, x, y, 2 represent arbitrary real numbers unless something is said to the contrary.) Axiom}. x+y = yx = Oe (commutative laws. (84+ aG2) = (ye (ossoriaive laws). Axiom 3. xy + 2) = ay 4x2 (distributive tax, Axiom 4. Given any two real mumbers x and y, there exits real number 2 such that x + z= y. This zis denoted by y —'x; the number x — x is denoted by 0, (It ‘can be proved that O is independent of x.) We write ~ x for 0 ~ x and call —x the negative of x Axiom 5. There exists at least one real number x #0. If x and y are two real Imumbere with x # 0, then there exists @ real mumber 2 such that xz = y. This 2 ts denoted by yx; the number x{x is denoted by | and can he shown to be independent of x. We write x" for Vix if x O-and call x°* the reciprocal of x. From these axioms all the usual laws of arithmetic can be derived; for example, =x) = x) = x ay a yet (Op) ete. (For ‘a more detailed explanation, see Reference |.1,) Axiom 2. x + (9 + 2) 1.3 THE ORDER AXIOMS We also assume the existence ofa relation < which establishes an oxderin the real numbers and which satisfies the following axioms: Axiom 6, Exactly one of the relations x = y,x < yx > y holds Non x > y means the same as y Oand y > 0, then xy > 0. Asiom 9. If > y and y > 2, then 8 > 2, Note. A real number x i called postive if x > 0, and negative if x <0. We denote by R* the set ofall positive eal nmbers, and by R= the et of all negative real numbers. From these axioms we can derive the usual rules for operating with inequalities. For example, if we have x < 3, then x7 < 32 ifs postive, whereas xz > if is negative Alto, fx > y and 7 > w where both y and w are positive, then x2 > yw. (For a complete diseussion ofthese rules see Reference 1!) Nove. The symbolism x < y is used as an abbreviation for the statement “rey or xy" mu eters 3 Thus we have 2 < 3 since 2 < 3; and 2 <2 since 2= 2. The symbol > is similaly used. A real number xis called nomegarve ix 2 0. A pai of imul- taneous inequalities such as x < y, y 0. o imple consequence of the foregoing axioms, Then a ah, [a ta) = brie 2a) isix 1 and ifthe only postive divisors of mare 1 and n. I'm > 1 and n {snot prime, then is called composite. The integer lisnether prime or composite. “This section derives some elementary results om factorization of inlegers, culminating inthe unique fctrization theorem, alo calle the fundamental theorem of arithmetic. ‘The fundamental cbeorem sates that (1) every integer m > 1 can be represented 45a product of prime factors, and () this factorization can be done in only one way, epact from the order ofthe factors. Ii easy to prove pact (1). Theorem 1S. Every integer n> |i either a prime or a product of primes. Proof. We vse induction on n. The theorem hokis trivially for n = 2. Assume itis true for every integer & with 1 < K 0, 6 2 0 und use induction on n= a +6. If n= Othen a= b = 0, and we can take d = O with x = y = 0. Assume, then, that the theorem has been proved for 0, 1, 2,...,n — 1. By symmetry, we can. asume a>}. Ifb = Otaked = a,x— 1, y—0. If} > 1 we can apply the induction hypothesis to a — 6 and b, since their sum is a=" —- 5 \ can be repre sented a5 a product of prime factors in only one way, apart from the order of the factors, Proof. We use induction on n. ‘The theorem is true for m = 2. Assume, then, that itis true forall integers greater than 1 and less than n. IC mis prime there is nothing more to prove. Therefore assume that 1 is composite and that m has two factorization into prime factors, say m= PPS Pe 992 2 ‘We wish 1 show that s = 1 and that each p equals some g. Since py divides the product 442 °°" du it divides at least one factor. Relabel the q's if necessary so that 7,l¢;. Then p; = 9, since both p, and 4, are primes. In (2) we cancel p, ‘om both sides to obtain m Since is composite, 1 < njp, I. We assume that Vn is rational and obtain a contradiction. Let vin = a/b, where a and 6 are integers having no factor in common. Then 2b? = a? and, since the left side of thisequation isa multiple of n, so to0 isa. However, ia isa multiple ofn, a ivelf must be a iultiple of, since m has no square factors >. (This is easily seen by examining the factorization of into its prime factors.) This means that a = en, where cis some integer. Then the equation nb? = a? becomes ni? = cn, or BF = nc The seme argument shows that & must also be a multiple of n. Thos a and b are both multiples ofr, which contradicts the fact that they have no factor in common This completes the proof im has to square ctor > 1 rm has a aquare factor, we can write n = mk, where K > Vand i hae no square factor >. Then Yn = mv/k; and it Vn were cational, the number vk would also be rational, contradicting that which was just proved A different type of argument is needed to prove that the number ei irrational (We assume familiarity with the exponential e* from elementary celeulus and its representation as an infinite series.) Theorem 1.11, If e* number eis irratlonal Proof. We shal prove that e~! is rational. The series fore“! i an akernating series with terms which decrease steadily in absolute value. In such an alternating series the error made by stopping atthe mth term has the algebraic sign ofthe Fst neglected term and is less in absotate value than the fist neglected term. Hence, ifs, = Boao (~ DIVE, we have the inequality FERRE OBER btn boo, them the O a for at least one x in S. ‘Theorem 1.18 (Additive property). Given nonempty subsets A and B of R, let C denote the set Cater yixed, yea). Uf each of A and B has a supremum, then C has a supremum and sup C = sup 4 + sup 8. Proof. Let a= sup A, b = sup B. If eC then z= x+y, where xe 4, yeBwor=x+ysa tb. Hewea + bisan upper bound for C, oC hasa supremum, say ¢ = sup C, and ¢ 0. By Theorem 1,14 there is an x in A and a y in B such that a-e 00, by Theorem L.1,a 4b < « “The proof of the next theorem is laft ae an enorioe forthe reader. ‘Theorem 1.16 (Comparison property). Given nonemply subsets $ and T of R such that $ <1 for every six Sand tin T. If T has a suprohum then S has @ supremum and sup S < sup T. 1.13. PROPERTIES OF THE INTEGERS DEDUCED FROM THE ‘COMPLETENESS AXIOM Theorem 117. The set Z* of positive integers t,2, 3... is unbounded above. Proof, If Z* were bounded above then Z* would have a supremum, say = sup Z*. By Theorem 1.14 we would have a — 1 < n for some nin Z*, Then n+ 1 > aforthism. Sincen + 1€ Z" this contradicts the fact that a = sup Z* Theorem 1.18, For every real x there is a positive integer n such that n > &. Proof. If this were not true, some x would be an upper bound for Z*, contra isting Theorem 1.17 1.14 THE ARCHIMEDEAN PROPERTY OF THE REAL NUMBER SYSTEM ‘The next theorem describes the Archimedean property of the real number system. Geometrically, it tells us that any line segment, no matter how long, can be m120 Fite Desi Approsimatons a covered by a finite number of line segments of a given positive length, no matter how small. Theorem 1.19. If x > 0 and if y is an arbitrary real mumber, there ls a positive integer m such that nx > y. Proof. Apply Theorem 1.18 wit x replaced by y/x. 415 RATIONAL NUMBERS WITH FINITE DECIMAL REPRESENTATION Avec suber of tbe font aot A aM, 10 * 10 w where ay isa nonnegative integer and a, is usually written more briefly a follows: re integers satistyingO < a, < 9, 103° ‘This is said to be a finite decimal representation of. For exaznple, 1s 12 Fe 2 270 sp IMO ET FG Real numbers lke thee are necessarily rational and in fact, they all have the form 1 = afiO, where ais an integer. However, not ali rational numbers can be ex: pressed with init decimal representations. For example, i} could be to expressed, then we would have $ = aj10* oF 3a = 10 for some integer c. But this is im possible sine 3 does not divide any power of 10. 5 = 725, oF 146 FINITE DECIMAL APPROXIMATIONS TO REAL. NUMBERS. This section uses the completeness axiom to show that real mumbers can be approximated to any desired degree of accuracy by rational numbers with finite decimal representations, Theorem 1.20, Assume x > 0. Then for every integer n > 1 there it a finite decimal r, = 0.240," 6, such that 1 nSrsnte Proof. Let S be the set of all nonnegative integers 4. Therefore we can write # = 0.125000 - If we interchange the inequality signs < and < in (6), we obtain a slightly different definition of decimal expansions. The finite decimals 7, satisly r, 0, we have |x| = x < a, whereas if x < 0, we have Ix] = —x 0, putx = —B/A to obtain B* — AC < 0, which is the desired inequ I-A = 0, the proof is teivial where 4 ee (arby < fa fbi, he m= fy seb (yoo) ar two meinen etor, wb Sake is their dot product, and fal = (ava)? isthe length of a 129 PLUS AND MINUS INFINITY AND THE EXTENDED REAL NUMBER SYSTEM R* [Next we extend the real number system by adjoining two “ideal points” denoted by the symbols +00 and —co (“plus iniity” and “minus infinity”). Defition 1.26. By the extended real number system R* we shall mean the st of eal numbers R together wih wo symbols +20 and ~ 20 which satisfy the allowing properties: 2) IER, then we have s+ Ga)= +0, x (-@) =~, x— (42) = 0, 5 = (20) = +0, 31420) m (00) = 0. Del 126 Complex Nembere s bb) Uf x > 0, then we have x00) = $0, af—c) = = ©) If x < 0, then we have 2(4+00) =, x(= 0) = te 4) 40) + (hoo) = (4. 0)(+00) = (=m) + (-@) = (+00)(— ©) If-xe R, then we have ~20 OF #0. i) byl = He I Web = bhi #0 #9) 16%, 09 = bel Proof. Statements (i)and (iv) areimmediate, To prove (i), we writex = x, + ix, Y= +My $0 that xy = xy, — x72 + Ae + %¥,). Statement (i) follows from the relation lay? = xii + xed + aby] + dt = GT + xDOT + 9D = bale. ‘Equation (ii) can be derived from (i) by writing it fa the form tx} = [xf xsl. Geometrically, |x} represents the length of the segment joining the origin to the point x. More generally, [x — pis the distance between the points x and y. Using this geometric interpretation, the following theorem sates that one side of 4 triangle leas than the sum ofthe other two side, Det 140 Complex Expoomatials » Theorem 1.39. If x and y are complex numbers, then we have be + ys b+ Lyi riangle inequality). ‘The proof is eft as an exercise for the reader. 125 IMPOSSIBILITY OF ORDERING THE COMPLEX NUMBERS ‘As yet we have not defined a relation of the form x < y if x and y are arbitrary ‘complex numbers, forthe reason tha it is impossible to give a definition of < for ‘complex aumbers which will have all the properties in Axioms 6 through 8. To illustrate, suppose we were able to define an order relation < satisfying Axioms 6,7, and 8. Then, since i ¥ 0, we must have either # > 0 or i < 0, by Axiom 6, Let us assume / > 0. Then taking, x = y= fin Axiom 8, we get i? > 0, oF =I > 0. Adding 1 to both sides (Axiom 7), we get 0 > 1. On the other hand, applying Axiom 8 to —1 > 0 we find 1 >'0. Thus we have both 0 > I and 1'> 0, which, by Axiom 6, is impossible. Hence the assumption / > 0 leads us toacontradiction. [Why was the inequality —1 > Onot already a contradiction?) ‘A similar argument shows that we cannot have < 0. Hence the complex numbers cannot be ordered in such a way that Axioms 6, 7, and 8 will be satisfied. 1.26 COMPLEX. EXPONENTIALS “The exponential e* (x rel) was mentioned earlier, We now wish to define &* when 2 is a complex number in such a way that the principal properties of the real exponential function will be preserved. The main properties of e* for x real are the law of exponents, ee" = e*1"%, and the equation e® = 1. We shall give a 0 Theorem 144. ef = if, and only if, zis an integral multiple of 2. Proof. Wz = 2nin, where m isan integer, then © = 00 an) + isin en) = Conversely, suppose that & = 1, This means that e* cos y = 1 and e* sin Since e* # 0, we must have siny = 0, y= kx, where k is am integer. But cos (kn) = (—1) Hence e* = (=I) since e* cos (kx) = 1. Since e* > 0, ‘k must be even. Therefore e* = 1 and hence x = 0. This proves the theorem. Theorem 148. €*1 = & if, and only if, 2, — 23 = Pnin (where n isan integer). Proof. & if, and only ie = 1 1.28 THE ARGUMENT OF A COMPLEX. NUMBER We the point z = (x, ») = x + iy is represented by polar coordinates r and 8, we ‘can write x= ros and y = sind, so that z= cos + irsin® = re™ naan, calm leti, ANeNtae, e De 108 negra! Power nd Rats 2m ‘The two numbers ¢ and 0 uniquely determine z. Conversely the positive number + is uniquely determined by =; in fact,r = |z|. However, z determines the angle 0 ‘only up to multiples of 2x. There are infinitely many values of @ which satisfy the equations x = |2| cos 8, y = [z| sin @ but, of course, any two of them differ by some multiple of 2x. Each such 9 is called an argument of z but one of these values is singled out and is called the principal argument of z. Definition 146. Let z= x + iy be a nonzero complex number. The unique real mumber 0 which satisfies the conditions x= bleos8, y= flsin®, -n 0. ‘Theorem 1.50, which states thatthe usual laws of exponents hold, can be proved ‘by mathematical induction. ‘The proof i left as an exercise