Undergraduate Texts in Mathematics Bruce P. Palka An Introduction to Complex Function TheorySpringer New York Beriin Heidelberg Barcelona Budapest Hong Kong London Milan Paris Tokyo Undergraduate Texts in Mathematics EW, Gehring PR. HalmosUndergraduate Texts in Mathematics Anglin: Mathematics: A Concise History and Philosophy. Readings in Mathematics. Anglin/Lambek: The Heritage of Thales. Readings in Mathematics. Apostol: Introduction to Analytic Number Theory. Second edition. Armstrong: Basic Topology. Armstrong: Groups and Symmetry. Bak/Newman: Complex Analysis. Banchoff/Wermer: Linear Algebra Through Geometry. Second edition. Berberian: A First Course in Real Analysis. Brémaud: An Introduction to Probabilistic Modeling. Rressand: Factorization and Primality Testing. Bressoud: Second Year Calculus. Readings in Mathematics. Brickman: Mathematical Introduction to Linear Programming and Game Theory. Cederberg: Geometries. Childs: A Concrete Introduction to Higher Algebra. Second edition. Chung: Elementary Probability Theory with Stochastic Processes. Third edition. Cox/Little/O’Shea: Ideals, Varieties, and Algorithms. Croom: Basic Concepts of Algebraic Topology. Curtis: Linear Algebra: Approach. Fourth edition. Devlin: The Joy of Sets: Fundamentals of Contemporary Set Theory. Second Course in Modern Ebbinghaus/Flun/Thomas: Mathematical Logic. Second edition. Edgar: Measure, Topology, and Fractal Geometry. Elaydi: Introduction to Difference Eq Fischer: Intermediate Real Analysis. Flanigan/Kazdan: Calculus Two: Linear and Nonlinear Functions. Second edition, Fleming: Functions of Several Variables Second edition. Foulds: Combinatorial Optimization for Undergraduates. Foulds: Optimization Techniques: An Introduction. Franklin: Methods of Mathematical Economics. Hairer/Wanner: Analysis by Its History. Readings in Mathematics. Halmos: Finite-Dimensional Vector Spaces. Second edition. Halmos: Naive Set Theory. Hiirmmertin/Hoffmann: Numerical Mathematics. Readings in Mathematics. Tooss/Joseph: Elementary Stability and Bifurcation Theory. Second edition. Isaac: The Pleasures of Probability Readings in Mathematics. James: Tapalogical and Uniform Spaces. Sinich: Linear Algebra. Janich: Topology Kemeny/Snell: Finite Markov Chains. Kinsey: Topology of Surtaces. Klambauer: Aspects of Calculus, Lang: A First Course in Calculus. Fifth edition Lang: Calculus of Several Variables. Third edition. Lang: Introduction to Linear Algebra Second edition. Lang: Linear Algebra. Third edition. Lang: Undergraduate Algebra. Second edition Applications and Computing. Volume 1. LeCuyer: College Mathematics with APL. LidVPilz: Applied Abstract Algebra (continued after index)Bruce P. Palka With 138 IllustrationsBruce P. Palka Department of Mathematics University of Texas at Austi Austin, TX 78712-1082 USA Editorial Boarc S. Axler Frederick W. Gehring Department of Mathematics Department of Mathematics Michigan State University University of Michigan Ezst Lansing, M! ! 2100 USA Paul R. Halmos Department of Mathematics Santa Clara University Santa Clara, CA 95053 USA Mathematics Subject Classifications (1991): 30-01 Library of Congress Cataloging-in-Publication Data Palka, Bruce P. ‘An introduction to complex function theory / Bruce P. Palka. Pp. em. Includes index. ISBN 0-387-97427-X 1. Functions of complex variables. 1. Title. QA331.7.P35 1990 515'.9—de20 90-47375, cr Printed on acid-free paper. © 1991 Springer-Verlag New York Inc. All rights reserved. This work may not be translated or copied in whole or in part without the written permission of the publisher (Springer-Verlag New York, Inc., 175 Fifth Avenue, New York, NY 10010 USA), except for brief excerpts in connection with reviews or scholarly analysis. Use in connection with any form of information storage and retrieval, electronic adaptation, computer soft- ware, or by similar or dissimilar methodology now known of hereafter developed is torbidden. The use of general descriptive names, trade names, trademarks etc., in this publication, even if the Camera-ready copy prepared by TeXniques, Austin, Texas, using LaTeX. Printed and bound by: R.R. Donnelley and Sons, Harrisonburg, Virginia. Printed in the United States of America 9 8765 4 3 2 (Corrected second printing. 1995) ISBN 0-387-97427-X Springer-Verlag New York Berlin Heidelberg ISBN 3-340-97427-X Springer-Verlag Bertin Heidelberg New York ‘not to be taken as a sign that such names, as understood by ks and Meichaudise Marks Avi, may wcourdingly bc used ficely by anyone.To my parents, Leonard Palka and Charlotte Fogarty PalkaPreface The book at hand has its origins in and reflects the structure of a course that I have given regularly over the years at the University of ‘Texas. The course in question is an undergraduate honors course in complex analysis. Its subscribers are for the most part math and physics majors, but a smat- tering of engineering students, those interested in a more substantial and more theoretically oriented introduction to the subject than our normal theory. (To be honest, I have included some material in the book that I do not ordinarily cover in the course, this with the admitted purpose of making the book a suitable text for a first course in complex analysis at the graduate level.) Thus, the tone of the course is quite rigorous, while its pace is rather deliberate. Faced with a clientele that is bright, but math- ematically less sophisticated than, say, a class of mathematics graduate students would be, I considered it imperative to give students access to a complete written record of the goings-on in my lectures, one containing full details of proofs that I might only sketch in class, the accent there being on the central idea involved in an argument rather than on the nitty-gritty technicalities of the proof. I also deemed it wise to provide the students with a generous supply of worked-out examples appropriate to the lecture material. Since none of the textbooks available when I started Leaching the course had exactly the emphasis 1 was looking for, | began to compile my own set of lecture notes. It is these notes that have evolved into the present book. In rough terms the course I have been describing comprises Chapters I, III, IV, V, VIL, and VIII of the book, together with the first three sections of Chapter IX. Chapter II, a resume of information from plane topology, is A complex variables course. Instead, the ideas from Chapter II get dispersedvit Preface throughout my ieciures, each topoiogicai notion being brought up as it becomes germane to the development of complex function theory. While this system works fine in the setting of a lecture, I find it disruptive to the ongoing narrative of a book. Therefore, just as other authors have done before me, I have chosen to assemble all the background material from elementary topology in a single place for ease of reference. Chapters VI and X, and with them the last two sections of Chapter IX, furnish “enrichment topics” to those who wish to proceed slightly beyond the essential core of basic complex analysis. The subject matter in Chapters VI and X would, I think, be regarded as standard in most beginning graduate courses. Located at the end of each chapter is a collection of exercises. Though some of these are intended to foster the development of the computational skills pertinent to complex analysis, most have a pronounced theorctical flavor to them, in keeping with the course for which they were designed. Many of the “classic” exercises in function theory turn up among these problems. Quite a few of the exercises, on the other hand, are original to this book (or they are, at least, to the best of my knowledge). It is high time that I expressed my gratitude to everyone who has had a hand in the creation of this book. These individuals include a num- ber « of graduate st students at Texas — Michael Pearson, Michael Westmore- fully read through carly versivus of ib of numerous errors, and, most importantly, identified places that from a student’s perspective were badly in need of change. I am grateful to col- leagues (in particular, to Barbara Flinn and Jean McKemie) who agreed to “field test” portions of the manuscript in their own classes. Their input has greatly improved the finished product. My special thanks go to Aimo Hinkkanen, with whom I’ve had many useful conversations during the final stages of preparation of the book and who has been an invaluable source of suggestions for problems. This book would have remained a pipedream were it not for the diligent efforts of Suzy Crumley, who typed it, and Buff Miner, who did the graphics and generally oversaw the production of the manuscript. Both patiently bore the brunt of my revisionist. tendencies Needless to say, they share none of the blame for the inevitable errors that have crept into the text and managed to escape detection under my proof- reading. ‘he editorial staff at Springer-Verlag (notably, Hob ‘lorop and his successor, Ulrike Schmickler-Hirzebruch) have been extremely helpful and understanding. Above all, I appreciate the fact that they did not pressure me with deadlines during my stint as graduate advisor, when my literary output slowed to a trickle. A “tusen tack” goes to the Mittag-Leffler In- stitute in Djursholm, Sweden, where some finishing touches were applied to the mannscript in the course of my stay y there during the academi valued proofreading). My teacher, Fred Gana. has been a source of both inspiration and encouragement for the undertaking. Finally, I would like toPreface ix ackuuwledge the support of my wife, Mary Auu, and my sons, Kevin aud Sean. Despite being innocent bystanders, they were often in perfect posi- tion to catch the flak of my frustration when things did not go as planned with this project. To them I say: the struggle is over and dad is a happy camper again.Contents Preface I The Complex Number System i The Aigebra and Geomeiry of Compiex Numbers 1.1 The Field of Complex Numbers 1.2 Conjugate, Modulus, and Argument . 2 Exponentials and Logarithms of Complex Numbers .. . . 2.1 Raising e to Complex Powers . 2.2. Logarithms of Complex Numbers. . . 2.3 Raising Compiex Numbers to Compiex Powers... . 3 Functions of a Complex Variable . 3.1 Complex Functions 3.2 Combining Functions 3.3 Functions as Mappings 4 Exercises for Chapter I Il The Rudiments of Plane Topology 1 Basic Notation and Terminology . 11 Disks ........... 1.2 Interior Points, Open Scts 1.3 Closed Sets 1.4 Boundary, Closure, Interior . 1.5 Sequences 1.6 Convergence of Complex Sequences 1.7 Accumulation Points of Complex Sequences... .. . 2 Continuity and Limits of Functions 2.1 Continuity ........ 2.2 Limits of Functions 3 Conuecied Seisxii It Iv Contents 3.1 Disconuecied Seis . 3.2 Connected Sets. . . 3.3 [3 Domains = =. : 50 3.4 Components of Open Sets . 50 4 Compact Sets 52 4.1 Bounded Sets and Sequences oe tae 52 4.2 Cauchy Sequences . 53 4.3 Compact Sets beens soe 54 4.4 Uniform Continuity ...............0. .. 57 5 Exercises for Chapter IT... . 6.2... eee eee eee 58 Analytic Functions 62 1 Gompiex Derivatives .........- 62 1.1 Differentiability 62 1.2. Differentiation Rules 64 1.3 Analytic Functions 67 2 The Cauchy-Riemann Equations : 68 21 The Cauchy-Riemann System of Equations ...... 68 2.2 Consequences of the Cauchy-Riemann Relations ... 73 3 Exponential and Trigonometric Functions . . . tee 3.1 Entire Functions .........-.... see 3.2 Trigonometric Functions ........ we 77 3.3. The Principal Arcsine and Arctangent Functions... 81 4. Branches of Inverse Functions : : 85 4.1 Branches of Inverse Functions 85 4.2 Branches of the p**-root Function 87 4.3 Branches of the Logarithm Function . . _ o1 4.4 Branches of the A-power Function ........... 92 5 Differentiability in the Real Sense ...........-.-. 96 5.1 Real Differentiability . . . : 96 5.2 The Functions f; and fz 98 & 101 Complex Integration 1 Paths in the Complex Plane .............---- V1 Paths Parameirizing Line Seginenis Reverse Paths, Path Sums ......Contents xiii 1.5 Change of Parameier 2 Integrals Along Paths . . . . 2.1 Complex Line Integrals 2:2. Properties of Contour Integrals . 122 2.3 Primitives........ 125 2.4 Some Notation 129 2 Rectifiable Paths... 2... 131 3.1 Rectifiable Paths... 2.0.2... eee eee 131 3.2. Integrals Along Rectifiable Paths............ 133 4° Exercises for Chapter IV... 0.0... 136 V Cauchy’s Theorem and its Consequences 140 1 The Local Cauchy Theorem 00... = 140 1.1 Cauchy’s Theorem For Rectangles 140 1.2 Integrals and Primitives... ... . 144 1.3 The Local Cauchy Theorem............... 148 2 Winding Numbers and the Local Cavey Integral Formula 153 2.1 Winding Numbers . . 153 160 161 3. Consequences of the Local Cauchy Integral Formula . 164 3.1 Analyticity of Derivatives 164 3.2 Derivative Estimates........ 167 3.3. The Maximum Principle 170 4 More About Logarithm and Power Functions... .... « 115 4.1 Branches of Logarithms of Functions woe 175 4.2. Logarithms of Rational Functions 178 4.3 Branches of Powers of Functions . 182 5 The Global Cauchy Theorems ... . . 185 5.1 Iterated Line Integrals. ... . . 185 52 Cries. bee 186 52 Canchy’s Theorem and Integral Forms la 2... 18h 6 Simply Connected Domains. ..... . 194 6.1 Simply Connected Domains 194 6.2. Simple Connectivity, Primitives, and Logarithns 195 7 Homotopy and Winding Numbers . . 197 7.1 Honwiopic Paths i97xiv Contents 72 ColatbePals.. =... 203 8 Exercises for Chapter V......---- 2220002005 204 VI Harmonic Functions 214 1 Harmonic Functio 215 Ld fe Co 215 2 The Mean Value Property . . 219 2.1 The Mean Value Property . 2.2 Functions Harmonic in Annuli 3 The Dirichlet Problem for a Disk 3.1 A Heat Flow Problem . 3.2 Poisson Integrals... . 4 Exercises for Chapter VI . . VII Sequences and Series of Analytic Functions 243 1 Sequences of Functions . . . 1.1 Uniform Convergence . 1.2. Normal Convergence 2 Infinite Series .............0.. 2.1 Complex Series 2.2 Series of Functions... 0.0 ee eee eee 3 Sequences and Series of Analytic Functions 3.1 General Resulls ...........- 3.2 Limit Superior of a Sequence 3.3. Taylor Series 3.4 Laurent Series 4 Normal Families ............... 4.1 Normal Subfamilics of C(U) 4.2 Equicontinuity ............. 4.3. The Arzela-Ascoli and Montel Theorems .... . . . 282 5 Exercises for Chapter VII... 6.0... eee eee 286 VIII Isolated Singularities of Analytic Functions 300 1 Zeros of Analytic Functions... 0... 2. eee 300 1.1 The Factor Theorem for Analytic Functions... . . . 300 1.2 Multiplicity... ..........- . 303 1.3. Discrete Sets, Discrete Mappings . . 306 2 leolatedSingulaitics = ttt re)Contents xv Definition and Classifivatiun of Isuiaied Singularities Removabie Singularities 23 Poles .......... 2.4 Meromorphie Functions 2.5 Essential Singularities . 2.6 Isolated Singularities at Infinity Theo © 3.i The Residue 3.2 Evaluating Integrals with the Residue Theorem . 3.3. Consequences of the Residue Theorem... .. . 4 Function Theory on the Extended Plane .........- 41 The Extended Complex Plane 4.2 The Extended Plane and Stereographic Projection .. 350 4.3 Functions in the Extended Setting . . 4.4 Topology in the Extended Plane . . . 4.5 Meromorphic Functions and the Extended Plane 356 5 Exercises for Chapter VIII ...............008 362 IX Conformal Mapping 374 1 Conformal Mappings 375 1.1 Curvilinear Angles . . 375 1.2 Diffeomorphisms . . . 377 1.3 Conformal Mappings 379 1.4 Some Standard Conformal Mappings... . . 383 1.5 SelfMappings of the Plane and Unit Disk... . . 388 1.6 Conformal Mappings in the Extended Plane . 389 2 Mobius Transformati 391 2.1 Elementary Mébius Transformations . 391 2.2 Mobius Transformations and Matrices . 392 2.0 Baked VOUS oes ce se ae . 394 24 Cross-ratios - 396 2.5 . 398 2.6 . 399 2.1 Classification of Mobius ‘Iranstormations - 402 2.8 Invariant Circles ......... - 408 3 Riemann’s Mapping Theorem . 416 3.1 Preparations ........... 416 3.2 The Mapping Theorem 419 4 The Carathéodory-Osgood Theorem... ... . 492 4.1 Topological Preliminaries . : 423, 4.2 Double Integrals . 426xvi Contents 45 CoufuuslModulus =... 4.4 wxtending Conformal Mappings of the Unit Disk . 4.5 Jordan Domains 4.6 Oriented Boundaries 5 Conformal Mappings onto Polygons BA Polygons... - 5.2 The Reflection Princ 5.3 The Schwarz-Christofiel Formuia............ 6 ExacmefrChaplam = =... X Constructing Analytic Functions 1 The Theorem of Mittag-Leffler Lee ii. Series of Meromorphic Functions . . . 1.2 Constructing Meromorphic Functions Lee 1.3 The Weierstrass p-function ............... 2 The Theorem of Weierstrass ..........------- 2.1 Infinite Products 2.2 Infinite Products of Functions 2.3 Infinite Products and Analytic Functions ....... 495 24 The Gamma Function. ................5 504 3 Analytic Continuation... . 2... 6... ee ee eee 507 3.1 Extending Functions by Means of Taylor Series . . 507 3.2 Analytic Continuation... .....- tee 510 3.3 Analytic Continuation Along Paths . 512 3.4 Analytic Continuation and Homotopy 517 3.5 Algebraic Function Elements 520 3.6 Global Analytic Functions 527 4 Exercises for Chapter X... 2.2.2 2-2. eee eee 535 Appendix A Background on Fields 1 Fields... 2... eee eee 11. The Field Axioms 12 Subflds = 1.3 Isomorphic Fields 2 Order in Fields... 2... 2.1 Ordered Fields eee 2.2 Complete Ordered Fields ..... 2.3. Implications for Real SequencesContents Appeudia DB Winding Nuubers Revisited Index 1 Technical Facts About Winding Numbers . 1.1 The Geometric Interpretation... . . 1.2 Winding Numbers and Jordan Curves . xviiChapter I The Compiex Number System Most of us met complex numbers for the first time in high school algebra, where the imaginary unit i = /—TI and expressions of the type z + iy arose quite naturally in the study of quadratic equations. Back then we didn’t balk for an instant at the prospect of extracting square roots of negative numbers. We simply learned to deal with complex numbers on a formal level by mastering the few innovations that caused the algebra of the complex number system to differ from that of the real numbers — and we managed just fine. It is in much the same spirit that we wish to begin the study of complex analysis in this chapter. After a nod in the direction of more rigorous mathematics in which we outline a proper definition of the complex number system, we review for the reader some of the standard elementary facts of complex arithmetic. By then embcllishing these facts with detail of a perhaps less familiar character, we lay the groundwork for a careful treatment of the theory of analytic functions of a single complex variable. 1 The Algebra and Geometry of Complex Numbers 1.1 The Field of Complex Numbers ‘We assume on the part of the reader a knowledge of calculus, including a basic familiarity with the structure of the real number system IR and with analytic geometry in two-dimensional real euclidean space IR. An element z of IR? is nothing but an ordered pair z = (z,y) of real numbers x and 2 = (2,y) and w — (u,») of IR? to he declare == (2,9) and w — ( declan2 I. The Complex Number System iequiied that 2 — u aud y — v. The addition of such pairs is carried vut in the most straightforward fashion, ztwa(ttu, yt). So-called “imaginary numbers” involving the quantity YT were in rea- sonably wide-scale use long before the time of the Irish mathematician William Rowan Hamilton (1805-1865). That use amounted in large part to not unlike That most of us experience today in our initial contact with complex numbers. It was Hamilton, however, who in an 1833 paper finally “demystified” these numbers and placed them on a firm logical footing. He did so by observing that there is an advantageous way to define a product zw of the elements z and w; namely, ‘The term “advantageous” alludes to the fact that, with the operations of addition and multiplication in IR? so defined, the ordinary rules governing the arithmetic of real numbers — e.g., the associative laws for addition and multiplication, the commutative laws for these operations, the distributive law — see their validity extended to IR. In standard mathematical par- lance, when endowed with these two algebraic operatious IR? acquires the structure of a field. (See Appendix A for the exact definition of this concept and for other material pertaining to fields.) It has become traditional to employ the designation C for IR? in its manifestation as a field and to refer to the clements of IR?, in this context, as complex numbers. The role of zero in the field © is played by (0,0) and the element (1,0) serves as its multi- plicative identity. The additive inverse of the complex number z = (z,y) is the element! (-z,—y); the multiplicative inverse z~1 (or 1/7) of z is given by 27? = (z[z? + y?J-! , —y[z? + y*]-1) — provided, of course, that z is different from zero. Just how does the mysterious number i fit into this scheme? In order to answer this question, we first remark that the set of complex numbers of the form z = (z,0) constitutes a subfield of € that is structurally indistin- guishable from the field of real numbers. For this reason it is customary to blur any distinction between the real number z and the complex number (z,0) and actually to employ z as a convenient abbreviation for (z,0). Un- der this convention IR itself is regarded as a subfield of ©. If we introduce the symbol i as a special notation for the complex number (0,1), we observe that # = (0,1)(0,1) = (-1,0) = -— in accordance with the convention just established. In the field © the real number —1 does, indeed, have t as a square root! Furthermore, an arbitrary complex number z = (z,y) can be rewritten as follows: z= (z,y) = (x0) + (0,y) = (2,0) + (0,1)(¥,0) = 2+ ty.1. The Algebra and Geometry of Complex Numbers a z=x+iy Rez Figure 1. Figure 2.4 I. The Complex Number System ‘The upshot of the preceding couuneuts is that we are justified in think ing of a complex number z as a quantity of the type z = z + ty, where z and y are real numbers and where i? = ~1. In fact, the representation of complex numbers in the form z = 2 + iy — as opposed to the ordered pair notation z = (z,y) — is the universally preferred one. (On aesthetic grounds we may occasionally opt to write z+ yi. Somehow 1 + i5 looks peculiar, while 1+ 5 doesn’t.) For reasons chiefly of historical interest x is called the real part of z, and y — not. iy, as one might expect — is termed its imaginary part. Use of the notations Rez and Imz to signify the real and imaginary parts of z is quile common. The formulas that define the sum and product of complex numbers z = z + iy and w = u+ iv now take the forms zetw=ztutilytr) , zu— yu +i(zv t yu). For exampie, we compuie (144i) +(-2- i) =-143i, 8-4, ioaey( 3 a) 14, 238i ~ (2455) (35 + 3B) = : One big difference between IR and € must be stressed from the outset. The complex number field is an “unorderable field” — we once again refer the reader to Appendix A for a precise definition of the term — so that expressions like z < w are not generally meaningful in ©. When inequalities appear in this book it will always be tacitly assumed that the quantities under comparison are real numbers. While few mathematical illustrations can claim to be worth the thou- sand words promised by the adage, there is often genuine benefit to be gained by studying a mathematical concept from a geometric viewpoint ‘The usual way of representing complex numbers graphically is to depict z= z+ iy, depending on the context, either as the point with coordinates (z,y) in the cartesian plane or as the vector from the origin to that point (Figure 1) In this setting the cartesian plane is spoken of as the campler plane, the z-axis as the real azis, and the y-axis as the imaginary azis. ‘The graphical interpretation of complex addition is indicated in Figure 2. ‘The geometric meaning of multiplication is less transparent and will be described shortly, following some additional preparation. Mathematicians tend to use the expressions “complex numbers” and “complex plane” inter- chaugeably, despite the geometric ove crowd in our usage of these terms, drawing iittie or no distinction between them.1. The Algebra and Geometry of Complex Numbers ulus, and Argument The conjugate of the complex number z = z + iy is the complex number z= 2—iy. For instance, 7 = ~i and 3 — iV2 = 34 iV2. Graphically z and z correspond to points that are mirror images of one another with respect to the real axis (Figure 3). Nt Figure 3. One verifies with little effort the following elementary identities: Faz; (1.1) yEu=7+0 , w= , zw Rez=(z+2)/2 , Imz=(z-2)/2i. A real number z in @ is characterized by the property that ¥ = z; an clement z of © is purely imaginary, meaning that Rez = 0, precisely when z=-z. The modulus |2| of z = 2+ iy is defined by |z| = \/z? +9”. (The terms magnitude and absolute nalue are commonly used synonyms for “modu- lus.”) As an example, | — 3 + 4i| = 5. Clearly |z| gives the length of the vector corresponding to z in the complex plane (Figure 4). More gener- ally, |z — w| is the ordinary distance between the points of the plane that represent z and w. Some basic properties of the modulus are summarized in (1.2). Only the cation. For these we refer the reader to Examples i.i and of the other statements is left as an exercise. The checking6 I. The Complex Number System Izl Figure 4. (|2wl = lzllw! . lz/w! = |2l/lwl : lel=Pi . 22 =[eP 5 (1.2) [Rezl< lz], [Imez| <2]; |z+ wl <|el+luls |z+w] > |lz|— wl]. If z #0, we remark that 1.3 hs as) rem In particular, z- = 7 if |z| = 1. Identity (1.3) shows how z and 27? compare graphically: 27! points in the direction of 7 and has modulus 1/lz| (Figure 5). Consider next a complex number z other than zero. It is always possible (14) z z|(cos0 + isind) , where @ is a real number: if z = 2 + iy, we merely choose any @ that satisfies cos@ = z/|z| and sin@ = y/|z|. For instance, 1+ has a polar representation z a itis v3 (os4 + isin Z) Each real number @ for which (1.4) holds is termed an argument (or an amplitude) of z. Geometrically, @ simply provides a measurement in radians1, The Algebra and Geometry of Complex Numbers 7 Figure 5. of the angle from the positive real axis to the vector depicting z in the complex plane. The usual sign conventions for angles in polar coordinates are to be observed. In Figure 6 we indicate with 61,0, and 43 three arguments of z; 6; and 62 are positive arguments, whereas 63 is a negative argument. The notation argz will be used in this text to designate the set of all arguments of z. Assuming that one such argument @ is known, this set is readily described: argz consists of all real numbers @ having the Figure 68 I. The Complex Number System form & — 09 + Sha, where & is au integei. To give au Husiration of this, arg(—1 {(2k + ijw: & = 0,+i,---}. (NB. It is customary in compiex analysis to write argz = 6 — rather than @ € argz — to express the fact that @ is an argument of z. This mild abuse of notation is so well established that any attempt to avoid it would ultimately do the reader a disservice.) We select one special member of arg z, the unique argument 6 of z in the interval (-7, 7], for “preferential treatment” and employ it, cirenmstances permitting, in situations where a definite argument is called for. This argument is referred to as the principal argument of z. It is denoted by Argz. For example, Argi= 5 Arg(1— -+ , Arg(-m)=n. In general,the principal argument of z = z+ iy (# 0) is given by Arcsin(y/lz/) if2>0, Argz = § r—Arcsin(y/|z]) if 0, x—Arcsin(y/|z|) if ||z|— wll. The trick is to write z = (z-+w)—w and to apply the triangle inequality, lz] =[(2+w)- wl < [z+ wl] +]-vl=|z+ul4+ |v}. This leads to the inequality e+ ul 2 bel- Repetition of the preceding computation with the roles of z and w inter- changed gives {z+ w| > |w| - Now |lz|—|w|| = |z| —|w] if [2] > [w| and |[z|— wl] = feo — Jz | if [rol > |]. In alll cases, therefore, we can assert that |= + w EXAMPLE 1.3. Describe geometrically the set 5 of complex numbers z that obey the condition |z — 1] = 2lz + 1].12 I. The Complex Number System Write z= «+ iy. Recalling (1.9), we reimatk that re So |z—1[= 241] |z-1P? =4|z 417 # |z? —2Rez4+1=4|e|? +8Rez+4 © 3\z/? +10Rez+3=0 4 3(0? + y*) +102 +3=0 2? 4 ee ty=-1 o7+%,.2, p16 sto Tyo o( +) 4 28 et5) tay. (NR. The symbol < stands for “if and only if” or “is eqnivalent to.” The ‘ and & Sis now easily seen to be a ‘circle of radius 4/3 centered at the point —5/3 on the real axis. EXAMPLE 1.4. Find a “complex” equation for the hyperbola with “real” equation 2? — y? =1. Make the substitutions 2 = (z + Z)/2, y = (z— 2)/2i. Then (z+2)? | (@-zP? 247 4 a It follows that the given hyperbola is described by the complex equation 24s? 2477 =2. EXAMPLE 1.5. Determine all solutions of the equation z4 + 16 = 0. It is required to find ali 4**-roots of —i6. Noting that W/16 that Arg(—16) = 7, we invoke Theorem 1.1 to identify these roo! 2[cos(w/4) + isin(x/4)] , 2[cos(3x/4) + ésin(3x/4)] , 2[cos(5x/4) + isin(5x/4)} , 2cos(7m/4) + isin(7x/4)] . to VERDE fb enh 83 and JE 3 to V2 + iv2, —V2 + iv’2, —V2 — iv’2, aud V2 — iv’2. EXAMPLE 1.6. Solve the quadratic equation z? — 32 +3+i=0.2, Exponentials and Logarithms of Complex Numbers 13 The quadraiic formula works just as efficienily in sulving quadr equations with complex coefficients as it does in solving the corresponding equations with real coefficients. (See Exercise 4.13.) As a consequence, the desired roots are given by _3tf0-40 Gt) _34V-3-F SE ‘Yo compute Vv set @ = Arg(—3— 4i). Then —x < 8 < —n/2, cos8 = —3/5, and sind = —4/5. In view of Theorem 1.1 and the definition of the principal square root, V-3—4i = V5 [cos(8/2) + isin(6/2)] Appealing to the half-angle formulas from trigonometry — note that cos(6/2) > 6 and sin(/2) < 6 — we compuie cos(0/2) = piteet [=2 - + and sin(9/9) = —y/ 125098 f14 0/5) __ 2 ep NF 2 a whence Y=3—4i = 1 — 2i. The solutions of the given equation are, there- fore, 2—i and 1+ 2 Exponentials and Logarithms of Complex Numbers 2.1 Raising e to Complex Powers It is our goal in this section to define the quantity ¢* for complex z and to derive its elementary properties. Recognizing that from a strictly rigorous outlook the introduction of this notion here might be judged premature, we feel that this point is far outweighed by the enormous convenience of If we substitute iy for t in this series and compute formally, fighting off any qualms about the precise meaning of convergence, we arrive at 2 vu appiy-F- ev =1ltiy— a4 I. The Complex Number System , ‘}tilu- J X The two series in the last line should again evoke memories from calcu- lus — they are Taylor expansions of cos y and sin y, respectively. In other words, the suspicion arises that e” = cosy + isin y represents the proper interpretation of e'¥. Next, since e**! = ee! for real numbers s and t, in ‘an attempt to accign meaning to the quantity ettiy difficult to avoid Sere the suggestivn that wu fact, motivated by the preceding con siderations, we now actually take the step of defining e* for z = z+ fy via the formula (1.10) ef = e*(cosy +isiny) . The notation exp(z is a frequently employed substitute for e*, used espe- cially in situations where z is replaced by a ture complicated eapiessivn. For instance, we might write exp|(z + 1)/(z* + 4)] in preference to the cor- responding expression involving ¢. Here are a few sample computations of fees Mar, ear, MMe | e cos] +isin1 . ‘he definition (1.10) makes it obvious that e* # U holds for every com- plex number z and that Rez (1.11) Furthermore, de Moivre’s formula implies directly that , arg(e?) =Imz. (1.12) (ery) =e" for every integer n. In particular, (e*)~? =e Let z = + iy and w = w+ iv. Bearing in mind formula (1.6), we calculate ete” = e*(cosy + isiny) e(cosv + isin v) = oH [ooe(y 40) + isin(y + 0)] = ett! and so verify a fundamental law of exponents: (1.18) erev =ert™ For which values of z is it the case that e? = 17 Certainly z = 0 has this property, hut many other camplex numbers share it. For e* = 1 to hald, ' only purely imaginary numbers z = iy fall under consideration. But ¢ cosy + isiny = 1 is true if and only if cosy = 1 and siny = 0, conditions2. Exponentials and Logarithms of Complex Numbers 15 satisfied precisely when y = Zé for some integer &. The conclusion: ¢* — 1 if and only it z = 2kmi for some integer &. More generally, since by (1-12) and (1.13) the equation e* = e” is equivalent to e”-* = 1, we remark that (1.14) e” & w= z+ 2kmi for some integer k . We have earlier represented a non-zero complex number z in the polar form z= r(cosé + ising) , where r= |z| and @ is any argument of z. This polar description of z can now be expressed more economically as (1.18) z=re®. In most situations (1.15) will henceforth be our preferred mode of polar representation. Notice that if we set r = 0 in (1.15) we obtain what amounts to a polar representation of z = 0 — 8 can be chosen arbitrarily — even though we assign no argument to 0. 2.2 Logarithms of Complex Numbers What does it mean to state that s = int, the (natural or base e) logarithm of the positive real number t? The answer confided to us by our high school math teachers is, of course, that e* = t. We would like to mimic this idea and say what it means for a complex number w to be the logarithm of a non- zero complex number z. The analogous requirement would be that e” = z and, indeed, any w enjoying this property is pronounced a logarithm of z. The difference between the real and complex sitnations is that, whereas e* = 1 is satisfied for a unique real value of s, there are infinitely many complex values of w for which e” = z. For example, e” = 1 is satisfied by w = 2kmi for k = 0,41,--- . Each such w is to qualify as a complex logarithm of 1! As happened with arguments, one of the logarithms of a given non-zero complex number z will be singled out for distinction, that being w = In|z| +i Arg z. Owing to (1.13) and (1.15), the calculation iare© izie =e confirms that w definitely is a logarithm of z. It is termed the principal logarithm of z and is denoted by Log z. (Our use of the notation log z, on the other hand, parallels the usage we’ve established for arg z; ie., log indicates the set of all logarithms of z. To write log z = w merely signifies that w is one of the possible logarithms of z.) In order to reinforce it, we Testate this notational convention: throughout this text Logz designates the principal logarithm of z, the one defined by Log z = In|z| + #Argz .16 I. The Complex Number System ‘This definition insures thai Loge = ln for a positive real 2. (Frum now on we shall write Log z instead of In z when z > U.) As examples, we record Log(-1) =i , Log(ei)=1+(x/2)i_ , Log(1—-i) = Log V2-(1/4)i. Because an arbitrary logarithm w of z satisfies e” = z = e8*, itis a consequence of (1.14) that w acquires the form for some integer k. Furthermore, any complex number of this type is a logarithm of z. In other words, the complete set logz of logarithms of z is made up of all complex numbers Log |z| + i#, where @ ranges over the set of arguments of z. For z = —1, to give one example, we have log(=1) = {(2k + 1)xi: k =0,41,--}. If wy is a logarithun of 21 aud wo # logarithun of zp, then ewtes = oie’? = a2, which demonstrates that wy + w2 is a logarithm of 212. This fact leads to the statement log(z122) = log 21 + log 22 , interpreted as follows: the set of logarithms of 212 consists of all sums of logarithms of z; with logarithms of za. It is not, in general, to be expected that Log(21z2) = Logz + Log zz. (For instance, Log(—i) = (—x/2)i # (3n/2)i = Log(—1) + Log#.) Certainly Log 21 + Log zz is a logarithm of 2129, just not necessarily the principal one. 2.3. Raising Complex Numbers to Complex Powers One standard way to define t*, where t is a positive real number and a is an arbitrary real number, is by means of the formula t9 = e*1°8*. If we at- tempt to carry this formula over to the complex setting with a mind toward raising a non-zero complex number z to a complex power A, we immediately encounter an apparent obstacle: z has many different logarithms. Which id there to be many different ways — but not always infinitely many, as it turns out — to raise z to the power 4. To be precise, for any logarithm w of z the complex number ¢” is called the \-power of z associated with w. ‘The choice w = Log = gives rise to the principal \-power of =. The familiar notation z* will see its use restricted to this special A-power of z. To re- peat, lest there be any contusion later on: z* will be employed in this book exclusively to symbolize the principal \-power of z, the one given by Palos:3. Functions of a Complex Variable Here are two simpie exampies: G77 = oe? To8t — @*t = cos(x?) + isin(x”) , gent Since an arbitrary logarithm of z has the structure Log z + 2k7i for a a suitable intoger b, it follows eecily thet the general l-power of z inh the form er, y Where © is au mbeger. (By way of and = 2zi, then z* = e7 e~4**”, 50 the collection of 1- powers of z in this instance is seen to consist ‘of the numbers e~@#+D** for k = 0,41,--- .) There are two cases meriting special commentary. Should A itself be an integer, say A = n, then quite clearly e7**¥ = 1 whenever k is an integer, in which event the sundry A-powers of z all reduce to 2”, the “nenal” n*-power of z. Secondly, if \ = 1/n for a positive integer n, PMA a exp (#8) exp (48) 7 2 = 2 (8222) a(n) = Vielexp [i (eet2ie)) and conclude, as might be anticipated, that the set of \-powers of z co- incides with the set of n'*-roots of z listed in Theorem 1.1. In particular, 2m = yz, Although it is true that 2+# = 242#, other “laws of exponents” have only limited validity for Principal pores To single out one such law, (zw)* is not generally the same as z\w*; e.g., (—i)'/? = (V2 ~ i¥2)/2, which is different from (—1)/?i4/? = (-. -Vi+ are Similarly, (z*)* can differ from z*#. As a rule it is advisable to exercise due caution in carrying out algeb: ation, if 2 =i 3 Functions of a Complex Variable 3.1 Complex Having reviewed the algebra of complex numbers and established some convenient notation and terminology, we conclude this chapter by casting18 I. The Complex Number System a preliminary glance ai the uliimaie objecis of siudy in comples analysis, complex-valued functions of a complex variable. (In truth, it is only a spe- cial class of these functions that comes under scrutiny in the subject, but this fact need not concern us for the time being.) To begin, we recall some pertinent background material concerning functions in general. If A and B are sets, a function f from A to B is arule of correspondence that assigns to each element z of A an element f(z) of B, the value of f at z. ‘The notation f: A — Ris used to indicate that f is snch a function. We refer to the set A as the domain-set of f and to B as its target-set. For a subset S of A the image of S under f, denoted f(S), is the set of values that f attains on S;i.e., f(S) = {f(z): 2 € S}. The particular set f(A) is called the range of f. For the most part the functions of interest. in this text. are those of the kind f:.A — ©, where A is a subset of C. In fact, we go so far as to lay down the following guidelines: in the absence of any statement to the contrary, f:A—C will always signify a function whose domain-set A is a subset of C; unless otherwise stipulated, any function referred to in the text is pre- sumed to be of this type. (The chief situation covered by “otherwise stip- ulated” occurs in Chapter VIII, where functions involving the “extended complex plane” are discussed.) Frequently such a function will be defined of the function in question the set of all complex numbers for which the defining formula “makes sense.” For example, f(z) = (2? -— 1)/(z?7 + 1) represents a function f with domain-set A = {z:z # ti} = C ~ {+i}, since the formula defining f only ceases to be meaningful at complex num- bers where its denominator vanishes. (N.B. The notation S ~ T is used for {z:z € S, z ¢ T}.) By the same convention, f(z) = Logz determines a function whose domain-set is A= C ~ {0} The class of rational functions of z is going to be a prominent source of examples for this book. We are thinking here of functions of the type a9 +ayz+ z)= A) bot biz+ where the coefficients ao, a1,..-,@, and bo,b,-..,b, are complex num- bers. Included in this class are the polynomial functions of 2, the rational functions having the special form (1:16) F(z) = a9 +1244 ane”. (Assuming that a, # 0 in (1.16), we say that f is a polynomial function of degree n.) Sometimes we shall also want to draw examples from a broader class of rational functions, the rational functions of z and y. This class consists of polunomial functions of z and y — the value of such a function at z= x + iy is given by an expression of the sort f(z) = don + ai02 + amy +--+ + amnz™y" ,3. Functions of a Complex Variable again with complex coefficients ag9, a10,-++;4mn — aud their quotic Every rational function of z is clearly a rational function of z and y, but not the other way around. By our convention polynomials are defined every- where in C, whereas to obtain the domain-set of a general rational function we must remove from € all points at which its denominator takes the value zero. It is occasionally instructive to look at a function f: A — C in terms of its real and imaginary parts; ie., to represent f in the form f = u+ iv, where u(z) = Re(f(z)] and o(z) = Im[f(z)]. The functions u and v are, naturally, real-valued functions on A. It is even possible to carry this a step further, by identifying z = z + iy with the ordered pair (z,y) of real numbers and by writing u(z) = u(z,y) and v(z) = v(z,y). In other words, we can regard u and v as real-valued functions of the two real variables z and y. Applied to the function f(z) = 2? this procedure yields u(z) = u(z,y) In case f(z) = e* the corresponding functions are u(z)=u(z,y)=e* cosy , v(2)=v(z,y) =e*siny. The elementary algebra of complex-valued functions of a complex variable does not differ radically from that of real-valued functions of a real variable. Presented with a pair of functions f: A + © and g: B — @, we can multiply f by a complex number c to obtain the function cf, add f and g to form £ +4, multiply these functions to produce fg, and take their quotient f/9. Both f +g and fg have domain-set AM B, and the domain-set of f/g is {2 € ANB: g(z) #0}, provided these sets are non-empty. Two other useful functions associated with a given function f = u+iv are its modulus |f| = (u? + v?)4/? and its conjugate f = u — iv. A more interesting way of combining functions f:A — C and g: B + to manufacture a new function arises when the range of f is contained in B. In this situation it is possible to form a function go f:A — C, the composition of g with f, by setting g o f(z) = g[f(z)]. In the case of the functions f(z) = 2? and g(z) = e*, for instance, we have go f(z) = e*” and fog(z) =e. An idea of major importance associated with the composition of func- tions is that of an “inverse function.” Suppose that a function f:.A — C is one-to-one (or, to use a synonym favored by complex analysts, univalent), which means that f(z,) = f(z2) only when +, — 2;. Under ¢ n de > Aas follo AY ofw) iwi de! > Aas follows: for w in f(A), g(w) is the unique element of A such that f[y(w)] = w. Thus, to find z = g(w) we must solve the equation f(z) = w for z in terms of w. The function g is called the20 I. The Complex Number System inverse of f aud ix usually denuied by f—'. (Cace must be exercised leat the inverse f—? of f be confused with the reciprocal of f, that is, with 1/f. The context will ordinarily rule out any possible ambiguity along these lines.) Setting B = f(A), one readily checks that f-!(B) = A, that f-lof(z) =z for all z in A, and that f o f-!(w) = w for all w in B. Consider, as an illustration, the exponential function f(z) = e*. This function fails to be univalent on its full domain-set C, and, consequently, it does not have an in- verse function. If, however, we modify the exponential function by suitably restricting its domain-set, we do arrive at functions that admit inverses. (This sounds a theme that we shall take up in earnest in Chapter III un- der the heading “branches of the logarithm.”) For example, it follows from (1.14) that the function f: A + ©, where A = {z:—x < Imz < 7} and f(z) = c%, is univalent, and it is easy to see that B = f(A) = C ~ {0}. ‘We conclude that f possesses an inverse function f~': B — A. In order to compute z = f~#(w) for w in & we must find the solution z of e* = w belonging to A, a task quickly accomplished: z = Log w, the principal log- arithm of w. Asa result, f~!(w) = Log w gives us a formula for the inverse of f. Of course, once the inverse of a function f has been determined, there is no real need to use different letters for the variables in f and f~!, as long as the domain-set of each function is kept in mind. We are thus free to write f-1(z) = Log i itvatis i i . {2} = Leg he pre: 2 of el 3.3. Functions as Mappings We have yet to mention a view of complex functions that occupies a central place in the conception of this book, a geometric view. From its perspective attention focuses on the role of a complex-valued function f of a complex variable as a mapping (or transformation) of the complex plane. In practice one often considers two separate copies of the complex plane — typically labeled the z-plane and the w-plane — and studies the manner in which geometric entities (lines, circles, etc.) in the z-plane are transformed into the w-plane under the correspondence w = f(z). Such input can often contribute appreciably to one’s understanding of a function. The following examples are meant to dramatize the geometric aspect of complex func- tions. EXAMPLE 3.1. Discuss the geometric effect on the complex plane of the transformation f(z) = az + 6, where a # 0 and 6 are complex numbers. Writing a = ale” with 6 = Arga, we realize f as the composition f =kohog of three extremely simple transformations: , A(z)=lalz , kz)=z40. (2) = Based on our geometric understanding of the algebraic operations in C, the action on the complex plane of each of the individual components of f is3. Functions of a Complex Variable a we2iz+1+i l+i — z-plane w-plane e | w=22' — zi-plane w-plane Figure 9. very easy tu describe. The geomeiric effect of f is just the cumulative effect, Produced when the transformations g,h, and k operate in sequence, as fol- lows. First, g rotates the plane about the origin through the angle 8, with Positive @ giving rise to a rotation in the counter-clockwise direction and negative 8 to a clockwise rotation; next, the mapping h dilates (= stretches or shrinks) the plane so that each ray issuing from the origin is mapped to itself and all distances get multiplied by the factor |a|; finally. k causes each point of the plane to be translated a distance |b] in the direction that the vector representing b points. (Figure 9 traces the successive images of @ particular line segment under the elementary transformations that com-22 I. The Complex Number System pose iv give f(z) = 2iz + 1+ i.) Trausfurmatious of the type treated in this example are called sense-preserving similarity transformations. Such transformations map any geometric figure in the complex plane — a trian- gle, for instance — onto a figure similar, in the strict geometric meaning of the word, to the given one. EXAMPLE 3.2. Let n be a positive integer. Discuss the behavior of the mapping 1 = 2" on circles centered at. the origin and on rays emanating from the origin. Observe that w = 0 if and only if z = 0. For z # 0 we express z in the polar form z = |zle4"8?. Then w = z” has a polar representation w = |2[Pe" ATE*, which permits us to write || = |z|” and arg w =n Argz (But not necessarily Arg w = n Arg!) It follows that for r > 0 the circle K = {z: |z| = r} is transformed to the circle K* = {w: |w| = 1}. As a matter of fact, in view of Theorem 1.1 there are exactly n points on K that are carried under w = 2” to any given point of K*. In effect, this transformation takes K and “wraps” it n times around K*. For —1 <@< 4 the ray R = {re : r > 0} is transformed to the ray R* = {se'"*: s > 0}, this in a one-to-one fashion. Two rays forming an angle » at the origin lane have as their images a pair of rays meeting at the origin of reas : oe : ne & Figure 10. EXAMPLE 3.3. Describe the image of an arbitrary line L in the complex plane under the function f(z) = e We first deal with the case of a vertical line L = {z: Rez = zo}. A3. Functions of a Complex Variable 23 typical point z of Z has the form z = zo + iy; its image w = f{<) i by w = e%ottY = e7%e'Y, Accordingly, |w| = e*°, implying that f(L) is contained i in the circle K of radius e*° centered at the origin of the w- plane. Due to the relationship arg w = y, w is seen to visit every point of K as z varies over any half-open interval of length 2x on L. The image of L is, as a consequence, the complete circle A, each point of which has infinitely many pre-images on L. 1 ical I. line in the camplex plane is descrihed by another linear equation, mz +. A point z of L has the representation z=2+i(mz-+b), with the result that w =e? = ee(™*+), In particular, w has modulus e*. We can also specify an argument 6(tw) of w by taking 0(w) = maz +b, When m = 0 — meaning that I is a horizontal line — we see that w traverses the open ray R = {re!b: r > 0} as = ranges over L. If m > 0, both |w| and 6(w) increase steadily — |w| covering the interval (0,00) and 0(w) the interval (—oo,00) — as z increases from —oo to oo. ‘The image produced is a curve that winds infinitely often around the origin and is known as a logarithmic spiral. The spiral in question here is the one with polar equation r = e@-»)/™, When m < 0 the image of L is another logarithmic spiral. (See Figure 11.) z-plane w-plane Figure ii. From the preceding discussion we distill Figure 12, which is recorded for future reference. The mapping depicted is again w = e?, but with z ed to satisfy —r < Imz < ™ In diagrams of this kind the use its inverse. In this figure, for instance, the origin of the a plane corresponds to the point 1 in the w-plane.24 I. The Complex Number System ni z-plane w-plane Figure 12. EXAMPLE 3.4. Show that the function f(z) = (1 — z)/(1-+ 2) maps the disk D = {z: |z| <1} onto the half-plane H = {w: Rew > 0}. Set w = f(z) and rewrite w in the following manner: for z = 2 + iy, From this we read off the relationship 1-2? Rew= Ty ae’ which makes it evident that w belongs to H for z in D and for no other z. We infer that f(D) is contained in H. Does every point of HT belong to f(D)? To answer this, fix w in H and solve the equation w = (1—z)/(1+2) for z in terms of w. The solution is z = (1 — w)/(1+ w). By construction w= f(z), so the above remark shows that 2 lies in D. Therefore, f(D) = H. (Sce Figure 13.) The mapping in Example 3.4 harbors several other noteworthy prop erties of which we will later have occasion to take advantage. The first is that f 1s a univalent tunction whose domain-set A = © ~ {—1} coincides with its range and whose inverse is f itself. This is so because the equation 1—2)/(1+2) can be solved for z in terms of w precisely when w # —1, in which event the unique solution z is given by z = (1—w)/(1+w) = f(w). Considerations similar to those in Example 3.4 reveal that the image un- der f of the unit circle K = {z: |z| = 1} — or, more accurately, of the “punctured” circle K ~ {—1} — is the imaginary axis in the w-plane i the <-pl t ¢ the unit circle, less the point w 1, in the w-plane. (There is a standard way to “fill in” the “punctures” occurring here, that being to adjoin an f= fo}, thei4, Exercises for Chapter I 25 w=(1-2)/(142) Figure 13. ideal “point at infinity” to the complex plane and then to regard f as a function on the so-called extended complex plane © = € U {oo} satisfy- ing f(-1) = 00 and f(oo) = —1. So regarded, f maps the full circle K onto the extended imaginary axis, meaning the imaginary axis together with the point oo, and vice versa. These matters will be discussed morc full ) Under f the sei {z: |z{ > i} is seni io the left half-plane {w: Ke w < U} with the point —1 removed. ‘The punctured teal axis IR ~ {—1} is mapped by f to itself, the intervals (—co,—1) and (-1,00) each going to itself. The upper half-plane {z: Imz > 0} and lower half-plane {z: Imz < 0} are interchanged by f; i.e., they are carried to {w: Imw < 0} and to {w: Imw > 0}, respectively. The function f(z) = (1—z)/(1+z) is just one member of a very impor- tant class of complex functions, the class of Mébius transformations (also called linear fractional transformations). This class consists of all functions of the type f(z) = (az +)/(cz +d), where a,6,c, and d are complex num- bers for which ad — be # 0. Such functions will be subjected to a systematic examination in Chapter IX. Notice that every similarity transformation f(z) = az +b qualifies as a Mébius transformation, one for which c = 0 andd=1. 4 Exercises for Chapter I 4.1 Exercises for Section I.1 4.1. If z= 142i and w = 3+ -4i, express the following in the form 2 + iy: Gi) Bz + ten ; (ii) 22? — xm ; (iii) Abo] + (1 — i)2? 5 Gi 2); (v) (1 = és) /(1-+ dz) ; (vi) (2 + 5274)7} | (vii) Im(zu) + 264 a j26 I. The Complex Number System (viii) SeosfArg(2”)] + SisinfArg w] . 4.2. Verify that 7+ w = 74, FW = 270, and |zw| = |z||w| for all complex numbers z and w. Assuming that w # 0, show also that (z/w) = Z/@ and \2/w| = |z//\ul- 4.3. Confirm that the identity 1+2z+---+2" = (1—2"*1)/(1—z) holds for every non-negative integer n and every complex number z, save for z = 1. 4.4. Establish the socalled “paralle! law” fo 4.4, Establish the so-called “parallelogram law” foi and w: |z + wl? + |z — wl? = 2|z|? + 2|wl? . 4.5, Given that z # 0 and w # 0, demonstrate that |z + w| = |z| + |w| is true if and only if w = tz for some t > 0. 4.6. Provide geometric descriptions for the following subsets of @: (i) {2: fz — 1 = |e — if} 5 (i) (2: fz 1] = 2Jz A} ; (iid) {2: [2-1] = 2}; (iv) {z: (L4i)z+ (1 —-a)2 = 1); (v) fz: 27 +iz - 17-3 =0}; (vi) {22 fz — a] + fe + i] = 4} ; (wil) (2: fe iP + |e +P = 4}; (viii) {2 : 22? — z? = 1} ; (ix) {z: 2? — 2? = i} . In each instance, sketch a rough graph of the set. 4.7. Depict each of the following sets in the complex plane graphically: (i) {z: z= Oor Arg z = 1/4} ; (ii) {z: |Argz — Argil < 1/6} ; (iii) {z: | Arg(z — i)| < 1/6} ; (iv) {z: | Arg(iz + 1)| = 1/3} 5 (v) {2: Arg 2+ Arg(z — 1) = x} ; (vi) {2: Arg(z — 1) - Argz = 2/2}. 4.8. Verify that 2Arg(1 +z) = Argz when |z| = 1, but z # —1. (Hint. What is the set K = {1 42: |2 }?) 4.9. Establish de Moivre’s formula (1.7). (Hint. Deal with n = 0 directly, use induction for n > 1, and reduce the case of negative n back to the case n>1,) 4.10. Compute: (i) the square roots of —1+iV3; (ii) the cube roots of —8; (iii) the fourth roots of ¢ ; (iv) the square roots of 5 + 124 ; (v) the fourth roots of 7— 244 . The answers may involve radicals, but should not contain any trigonometric expressions. 4.11. Show by example that Zw = Z\/w need not be true for arbitrary complex numbers z and w. Confirm that this formula is valid, however, if either z or w is a non-negative real number. (N.B. Similar comments apply to the principal n‘*-root.) 4.12. Knowing that Ree > 0, demonstrate that |e + Ve?—1| > 1, with equality holding only when ¢ is a real number in the interval (0, 1]. (Hint. First check that Re(@ve He > 0 for every ¢ under consideration.) #. the solutions af a quad anda tO anda #0,4, Exercises for Chapter I aT (4 | ae oa) = (-b + VO? = 4ac j/{2a). given by ihe usual quadratic formuia; ie., by 4.14. Find all the solutions of the equations: (i) z? -4iz~4—-2i=0; (ii) 227 — (245i) 2-243 5 (iii) 24 + (442%) 2? -1=0. 4,15. When cis real and in the interval (~1, 1], the roots z of z2—2cz4+1 = 0 have |z| = 1; when c is any other complex number, this equation has one root z; with the property that |z;| > 1 and a second root z2 having |z2| < 1. Corroborate these statements and, in the latter case, identify z; and z2 explicitly. (Hint: Recall Exercise 4.12. The determinations of 2; and z2 will vary, depending on whether Ree > 0, Ree = 0, or Ree < 0.) 4.16. Determine all solutions of 2"! + 2? 4...424+1=0. Here nis an integer greater than onc. 4.17. It is plainly true of any complex number z that either V2=270r Identify the set of 3 the Identify the sot of 3 the proper choice. 4.18. For which complex numbers z is it the case that \/z/z = z/|z|? 4.19. Show that the locus of points z in the complex plane satisfying an equation of the type Az? + Bz + BZ4C = 0, in which both A and C ate real numbers and |B|? ~ AC > 0, is a circle when A # 0 and a line otherwise. Conversely, show that any circle or line in the compiex piane admits an equation of the kind just described. 4.20. Show that the locus S of points z in the complex plane subject to the condition |z + 1||z — 1| = 1 is the curve described in polar coordinates (r,8) by the equation r? = 2cos(24). (The curve in question is known as a lemniscate.) Sketch a rough graph of this curve in order to obtain an approximate picture of the set S. (N.B. In describing a point of IR? in terms of polar coordinates (r,@) we always assume that r > 0. Naturally, r = 0 holds only for the origin. Exercise 4.8 may be of help in establishing that every point of S lies on the given lemniscate.) 4.21. Let ¢ be a complex number satisfying |c| < 1. Demonstrate that lz +e] < |1 + 22| if and only if |z| <1, with equality holding if and only if lel= 2 : Yaw a (dsr) (mr) ~ 2 |2,@; — 250,)? . Igk ra?) (Sm)28 I. The Complex Number System Section 1.2 Section 1.2 4.23. Certify that e7 = z, Logz = Log? and 24 = z is both real and negative? and that, except for non-positive real values of 2%. What. happens to the latter formulas when z 4.24. If n is a positive integer, establish the truth of the identities 1+ cos 6 + cos(20) + +--+ cos(n#) = 1 2 and cot(0/2) _ cos{(2n + 14/2] sind + sin(20) + ---+sin(nd) = > Te tor all real 6 other than integral muitipies of Zz. (Hint. Make use of Exercise 4.3 with z =e.) 4.25. Express in the form z + iy: (i) Log(—e) ; (ii) Log(1 — iV3); (iit) (<1)! ; (iv) Bowe; (w) AV 5 (vi) # (wit) (VB DEF 4.26, Determine the collection of all A-powers of z when (i) z = —e and wi; (ii) z= ie*/? and \ =i; (iii) z =i and \ = Logi; (iv) z= 1 and 1-i. 4.27. As pointed out in the text, it is not always so that Log(z122--- Zn) = Log 214 Log 22 +--- + Log zn. The best one can usually say is that these quantities differ by an integral multiple of 27i, a state of affairs often indi- cated by writing Log(z1z2-+- zn) = Log 21 + Log z2-+- +--+ Log zp (mod 21). Show that actual equality Log(z: 22 -- +n) = Log 21 + Log 22 + +--+ Log zn occurs when and only when —x < Arg zi + Arg za +++: + Arg in < ™- 4.28. Verify that Log(1 - z?) = Log(1 — z) + Log(1 + 2) when |z| < 1. What can be said about Log{(1— z)/(1 + z)] for such z? 4.29, Establish that Log(z*) = A Log z (mod 2ri), but give an example of z and 2 for which Log(z*) # ALogz. Assuming that } is both real and positive, determine the set of all z such that Log(z*) = \ Log z. (Hint. For the jast part check first that A Arg z is an argument of z* when A is reai.) 4.30. Confirm that the law of exponents 2\z4 = z\+# is valid for all non-zero complex numbers z and all complex exponents \ and y. Give an example of complex numbers z, A, and yt for which (z*)# # z*#. 4.31. Prove that the only circumstances under which |z*] = |z[* are (a) when ) is a real number or (b) when z is a positive real number and the quantity (Im A) Log |z! is an integral multiple of 2x.4, Exercises for Chapter I 29 x Section 1.3 4.32. Determine the domain-sets of the following functions: (iz? + 22+ 5) Mz" + 322 — 4) ; (ii) g(z) = 22/(z? +B) ; (22 = 27417" 5 iv) HC2) = (€4 + /Ket ~1) 5 (¥) fl2) = Log let — e° (vi) m(z) = (e+ y)/(E~ 2? - ia*y’) . ; xpress the real a: 4.33. Express the real a; functions of the two real variabies z and y: (i 22 — 245? $1; (iii) A(z) = ze? + 274e77 ; (iv) A(z zLogz for Rez > 0. 4.34. If f(z) = co +c1z+-+-+¢n2", derive the ensuing polar coordinate representations for u = Re f and v =Imf: u(veit) = ag 4 Sot [ax com( kA) — hy sin(kA)) (re) 2a im and v(re*) = bo + D> r* [ax sin(kO) + by cos(kO)] , ti where cy = ay + iby for k= 0,1,...,0 4.35. Verify that the composition go f of Mobius transformations g and f is again a Mobius transformation. (N.B. It will follow from Exercises 4.36 and 4.48 that the inverse of a Mébius transformation is likewise a mapping of this type, so the class of Mébius transformations forms what is known a “group” under the operation of composition.) 4.36. Suppose that f(z) = (az +5)/(cz +4) is a Mébius transformation with the property that ¢ # 0. Through the process of actually computing J~1, which turns out to be another Mébius transformation, certify that f is a univalent function whose domain-set is © ~ {-d/c} and whose range is © ~ {a/c}. 4.37. Let A = {z: Rez > O}, and let f:A — C be the function given by f(z) = Log(z? + 1). Show that f is a univalent function. Find its range B = f(A) and its inverse function f~!. (Hint. Perhaps the most efficient way of identifying B is by geometrically tracking the image of A under successive application of the elementary functions g(z) = z?, h(z)=z+1, and k(z) = Log z that compose to produce f.) 4.38, Define f: A — ©, where A = {z: |z| > 1}, via the formula f(z) = (z + 274)/2. Verify that f is a univalent tunction with range € ~ [-1, 1] and compute f~!. (Hint. Exercise 4.15 is relevant here. The rule of corre- spondence giving f~!(w) involves two different. expressions, the applicable30 I. The Complex Number System one vaiying with the lovativa uf w.) 4.39. Let D={z: |z| < 1}. Under the — that |c| < 1, show that f(z) = (z+¢)/(1+ 22) satisfies f(D) = D. (Hint. Recall Exercise 4.21.) 4.40. Determine f(S), where f(z) = e!/* and S = {z:0< |z| 1 has at most n roots. (Hint. Make use of the previous exercise.) 4.44. A point z is termed a fized point of a function f provided f(z) = Locate all fixed points of the following functions: (i) f(z) = (1+ i)z+1 (ii) o(z) = (1 + )Z 41 ; (ili) A(z) = F +3 ; (iv) b(z) = ze? ; (v) &(z) (82 —4)/(2 — 1) 5 (vi) m(z) = Qz + D/(2 + 1); (wi) hae 142828; & ) q(z) = 2442-2, (ix) r(z) = (2? + 22)/(z? +1). 4.45, i ial f(z) = dg + az +--+ + an2" of degree n > 2 could have? What about a proper rational function f(z) = (ao + az +--+ +@n2")/(bo + biz +--+ +bmz™), where an # 0, bm #0, and m > 1? (The word “proper” here signifies that nem) 4.46. Suppose that f(z) = (az-+6)/(cz +d), where ad—be = 1 andc £0. Show that f has only one fixed point when a+d = +2, but two such points in all other cases. 4.47. Describe the geometric effect on the complex plane of the transfor- mation f(z) = aF +b, where a # 0 and b are complex numbers. (Such a mapping is called a sense-reversing similarity transformation.) 4.48. If f and g are sensc-preserving similarity transformations, show that the composition go f and the inverse function f~! are transformations of that same type. What happens to go f when either f or g is changed to a sense-reversing similarity? When both f and g are so changed? What is the character of the inverse of a sense-reversing similarity transformation? 4.49. A similarity transformation f(z) = az +b is devoid of fixed points in the complex plane precisely when a = 1 and 6 # 0; otherwise, f has exactly one fixed point or fixes every point, Support this claim. 4.50. Let the f(z) — ar th haw point. Show that f can be rewritten in the form f(z) = zo + a(z— 29). De- duce from this representation that the geometric effect of f on the complex as a fixed4, Exercises for Chapter I 31 plane ainounts to a rotation about the point z throught au augle 9 — Aiga, followed by a dilation with respect to zo in which every ray issuing from zo is mapped to itself and all distances get multiplied by the factor |a|. 4,51. Prove that a sense-reversing similarity transformation f(z) = a+b has a unique fixed point if |a| # 1, has a line as its set of fixed points if ‘= 1 and ab — —b, and has no fixed points in the complex plane if Jal = 1 4.52. Consider the transformation w = (1 —z)/(1+2z) from Example 3.4. Check that under this mapping the following sets transform as indicated. (i) {z:Imz > 0} > {w:Imw < 0} ; (ii) {2: Rez = -1,z # -1} > lw #1) 3 (iil) {2: [e+ 2] = 1,2 4-1} 9 {wi u=-2}; 1} = {w: w+ (1/2) = 1/2,w # -1) . 4.53. Demonstrate that under the correspondence w = z? any horizontal line in the z-plane, apart from the real axis, is transformed to a parabola in the w-plane. To what does the real axis get mapped? 4.54. Extend the previous exercise as follows: under the correspondence w = =? every line in the z-plane, except for lines passing through the origin, is carried to a parabola in the w-plane. What are the images of lines containing the origin? (Hint. Use Exercise 4.53 and the identity z? = e72i0(e!7)?) 4.55. By a cardioid is meant a plane curve that is similar, in the technical geometric meaning of the term, to the curve whose polar equation is r = 1+ cos 4. Show that the transformation w = z? maps an arbitrary circle K passing through the origin of the z-plane to a cardioid in the w-planc. (Hint. If K has center zo, then this circle admits the description K = {zo + zoe”: —1 < 0 < 1}. Begin by looking at the case zp = 1.) 4.56. If L is a line in the z-plane, determine its image under the mapping w = \/z when (i) L is horizontal, (ii) L is vertical, and (iii) L has equa- tion y = 2V3. (N.B. The character of the image in (i) and (ii) will vary, depending on whether L intersects the negative real axis or not. Keep in mind that Re Vz > 0.) 4.57. Let K be a circle in the z-plane. Show that the image of K under the mapping w = 27? is a circie in ihe w-piane if K does not pass through the origin, whereas its image — or, more correctly, the image of K ~ {0} — is a line when the origin is a point of K. If L is a line in the z-plane, prove that w = 2~! maps L to a circle punctured at the origin when the origin does not lie on L and transforms L — or, rather, L ~ {0} — toa line with the origin deleted when L contains the origin. (Hint. Put Exercise 4.19 to work.) 4.58, Assuming that r # i, check that ihe circle K = {z: |zj = rj is transformed under the correspondence w = (z + z~!)/2 to an ellipse in32 I. The Complex Number System the w= plane. Vii is ihe nage of K when r = i? (fini. Observe that K = {re’* 0s <0 < 2n}. Start by expressing u = Kew and v = imw for w on the image of K as functions of the parameter 9.) 4.59. Let w = (az + 6)/(cz +d), where ad — be = 1 and c # 0. Show that the circle K = {z: |z + (d/c)| = 1/|cl} in the z-plane is mapped by this transformation to the circle K* = {w: |w—(a/c)| = 1/|e|} in the w-plane 4.60. Given that 21,22, and zg are distinct points in the complex plane, verify that these points are the vertices of an equilateral triangle if and only if, (1.17) d+dtBanntantorn. (Hint. Exploit the fact that both the property of being the set of vertices of an equilateral triangle and property (1.17) are invariant under similarity d if and =. . # 21,22, and zy where a # 0, then wy, wa, and ws exhibit ihe same property — io reduce the problem to a simple case.) 4.61. Let z1,22,23, and 24 be distinct points of ©. If these points list in consecutive order the vertices of a square, demonstrate that (21 — za)(za — 24) (1.18) — {21 #2}{%3 2a} Provide an example to show that the validity of (1.18) does not, in general, imply that z1, 22, zs, and zq are the consecutive vertices of a square. Under the assumption that these points are the consecutive vertices of a parallel- ogram, however, and also satisfy (1.18), establish that the parallelogram in question is necessarily a square. (Iint. Appeal to similarity invariants.)hapter I a The Rudiments of Plane Topology Introduction ‘The present chapter is devoted to a survey of the basic definitions and theorems from plane point-set topology that are prerequisite to a careful oncepts will discussivn of complies function theory. (Additional topologi be introduced later in the text, as they become pertinent to developments.) We warn the reader that a number of the statements found in this chapter are just that, statements made without any attempt to justify them. Our objective here is simply to assemble in one place for handy reference the relevant background material from topology, not to provide an in-depth account of the subject. Those desirous of a more detailed treatment are encouraged to consult one of the many available textbooks on elementary topology, e.g., Topology - A First Course by J.R. Munkres (Prentice-Hall, Englewood Cliffs, N.J., 1975). Having said this, we hasten to add that the bulk of the results offered here without proof are really nothing more than exercises in sorting through the definitions of the concepts involved. They can be treated as such by the ambitious reader. i Basic Notation and Terminoiogy 1.1 Disks We start by fixing notation for a number of standard sets that will be in constant use throughout the book. If zo is a point in the complex plane and if 0 < r < on, we write A(eo,r) = {21 |e 201 <7,34 Il. The Rudiments of Plane Topology Aw Btzo,r) = ~aisrh, A*(z0,7) = {2:0 < |z— 20] 0 such that the open disk A(z,r) is contained in A. A set U in the complex plane with the property that every point of U is an interior point of U is known as an open set. (N.B. The letters U, V, and W, when employed in this book to represent sets, will invariably designate open sets.) Some elementary examples of open subsets of € are the empty set ¢, the entire complex plane C, the open disk A(zo, 7) the punctured disk A*(zo, r), the set © ~ A(zo,r) = {2: |x — 29] > r}, and th mez > 0}. b 1 f mez > O}. T ) ( ne { open sets are described in: Theorem 1.1, The union of an arbitrary collection of open subsets of © is an open set. The intersection of a finite collection of open subsets of © is an open set. 1.3. Closed Sets A subset A of C is pronounced a closed set provided its complement © ~ A is open. Examples of such sets include ¢, C, the closed disk A(zo,r), the circle K(zo,r), the closed upper half-plane {z: Imz > 0}, and the real line IR. The sets ¢ and C are the only subsets of the plane that are simultane- shy most subsets of © are neither open nor closed. There is a proposition dual to Theorem 1.1 dealing with closed sets. It can be inferred from the ear- lier theorem by application of de Morgan’s rules for taking complements of unions and complements of intcrsections. The: 1.2. Th of T is a ciosed sei. The union of a jiniie coticctiun uf closed subseis of T is a closed set.1, Basic Notation and Terminology 35 1.4 Boundary, Closure, inicrior A point z in C is termed a boundary point of the plane set A if for every > 0 the open disk A(z,r) has non-empty intersection with both A and © ~ A. The set consisting of all such boundary points is known as the boundary of A. We use the notation OA to signify this set. As examples, observe that OA(zo,r) = dA(z0,r) = OK (20,7) = K(z0,r), @A*(z0,7) = K(zo,r) U {zo}, and A{z:Imz > 0} = RR. An open set in the complex plane is characterized by the property that it contains none of its boundary points, whereas a closed set is distinguished by the feature of containing all of its boundary points. ‘The set A = AU AA associated with a subset A of © is referred to as the closure of A. (We rely on the context to make evident whether the symbol — above a letter stands for “conjugate” or for “closure.”) By way of illustration, the closure of the open disk A(zo,r) is the closed disk A(zo,r). Clearly a point z belongs to A if and only if AN A(z,r) is non-empty for every r > 0. We record the following observation. Theorem 1.3. The boundary 9A and closure A of a subset A of © are closed sets. Notice that a piane set A is ciosed preciseiy when A = A. To add one last bit of basic terminology, the set of all interior points of a set A is called the interior of A. The notation A® is commonly utilized to designate this set, which is easily seen to be open. 1.5 Sequences A sequence o in a set A — here we are not necessarily restricting A to be a subset of © — is usually defined to be a function o:IN — A, where IN is the set of positive integers. Writing a, = (n) we can view o as a rule for listing in a specified order — namely, ai, a2, a3, --- — the elements of its range. Since ¢ may fail to be one-to-one, we must allow for duplications to crop up in this list. In other words, we are free to regard a sequence o in A as an ordered, but possibly repetitious list a1, a2,a3,-++ of elements of A, indexed by the positive integers. The n* entry a, in the list is then called the n® term of the sequence. In this book the symbol (an) will be employed as shorthand notation for the sequence a1,a2,a3,---. (Sometimes we shall run into sequences indexed by sets of integers other than IN. Sequences 49,1, a9,+-+ will, for instance, occur quite regularly. Such minor deviations from the dieenasion at hand should cause no severe problems of a index set, but the context will ordinarily make even this unnecessary.) There are several options available for presenting sequences in the con-36 Il. The Rudiments of Plane Topology ttern to its terns, a sequence might be described by writing out enough of the terms so that the pattern becomes recognizable; a second option is simply to give the rule of correspondence that expresses aq as a function of n; a third way to define a sequence is to do so recursively by identifying a, and by detailing the procedure for obtaining Gy once @1,42,...,4n-1 are known. We can exemplify all three methods using the sequence a; = 1,az = —I,a3 = 1,a4 = —1,a5 = 1,ag = —1,---. The pattern here is evident. Also, the general term a,, is readily expressed: an = (—1)"+1, Lastly, the same sequence is presented recursively as follows: a, = 1 and ap = —ap-1 for n > 2. If (aq) is a sequence in a set A, then by a subsequence of (an) is meant a sequence, call it (bn), related to the given one in the following manner: there is a sequence nj < nz < ng --- of positive integers such that b: = any, by = Gny4b3 = Gpy,-- . The notation (aq,) will serve to indicate such a subsequence of (an). We stress again that the notion of a sequence is not confined to se- quences of complex numbers. At various times in this book we shall have occasion to consider sequences of sets Ay, Az, Ag, ++, sequences of complex functions f,, fo, fa,---, and, yes, even sequences of sequences! We now focus our attention on a sequence (zn) of complex numbers. It may happen that, as n increases, the term z, develops an overwhelming attraction for and finds itself inescapably drawn toward some particular complex number 2. When this behavior is exhibited, we characterize (zn) as a “convergent” sequence with zo as its “limit.” To make this precise, we define a sequence (zn) in © to be convergent if there exists a complex number zo of which the following is true: corresponding to each ¢ > 0 there is an index N = N(c) such that zq lies in the disk A(zo,€) for every n > N. (See Figure 1.) This being so, there is exactly one such number 29 and it is called the limit of (zn). We write zo = litMaeo Zn OF 2n — Zo to express the fact that (z,) is a convergent sequence with limit zo. In symbol-free language one speaks of zq “tending to” or “converging to” or “approaching” zo. We point out that zp — 20 is equivalent to jz, —zo| > U. This observation can be quite helpful in confirming that the suspected limit of a complex sequence actually is the limit, for the reason that (|z, — zo|), being a real sequence, is subject to analysis by techniques —- the “squeeze tule,” for one — without counterparts in C. The inequalities |Re zn — Rezo] < fen — 20], [Im zp — Imzo] < len ~ 20], and len — 20] < |Re2n — Rezo] + |Imzq —Im29| , when combined with the squeeze rule, lead to a useful remark.