Consulting Editors




Department of Mathematics University of Washington







the ability to use mathematical technique and reasoning is more valuable than the ability to manipulate and calculate accurately. The second opinion is less widely held. this argument overlooks the obvious fact that in almost any situation. Mathematicians teaching in liberal arts colleges and universities are often under pressure from their colleagues in the humanities and social sciences to offer short courses which will painlessly explain mathematics to students with varying backgrounds who are seeking a broad. it is necessary to learn something of the logic. I t usually takes severa1 years to make this journey. because of the improving curriculum in high schools. the language. many students are completing the elementary mathematics included in algebra. For most people. I t is the authors' hope that this book will provide the means for this necessary contact. I t embraces a rigid method of reasoning. where the main emphasis is on mathematical formalism and manipulation. Fortunately. There is no such thing as "descriptive7' mathematics. Any enthusiasm for creative thinking which a student may carry into college will quickly be blunted by such a course. The extent to which such courses do not exist is a credit to the good sense of professional mathematicians. It is often claimed that the manipulative skills acquired in elementary algebra and calculus are what a student needs for the application of mathematics to science and engineering. In order to find answers to the questions "What is mathematics?" and "What do mathematicians do?". it is impossible to learn much about mathematics without doing mathematics. but only by active contact with the content of real mathematics. The first of these convictions seems to be accepted by most educated people. a concise form of expression. Too often. and trigonometry before entering college. geometry. and the philosophy of mathematics. Mathematics is a big and difficult subject. so that as college freshmen they can begin to appreciate the attractions of sophisticated mathematical ideas. liberal education. and a variety of new concepts and viewpoints which are quite different from those encountered in everyday life. such students are shunted into a college algebra or elementary calculus course. Many of these students have even been exposed to the new programs for school mathematics which introduce modern mathematical ideas and methods. v . and indeed to the practica1 problems of life.PREFACE This book is an offspring of two beliefs which the authors have held for many years: it is worthwhile for the average person to understand what rnathematics is al1 about. This cannot be done by listening to a few entertaining lectures. Although not altogether wrong. the road from marketplace arithmetic to the border of real mathematics is long and steep.

Such a mastery is hard to achieve. enough difficult material is included in most sections and chapters so that even the best students will be challenged. Appropriate places for the use of this book include: a freshman course to replace the standard precalculus college algebra for students who will progress to a rigorous treatment of calculus. These are designated by a "star. but his progress may not be rapid. a course to follow a traditional calculus course to develop maturity. The object of this book is to present in a form suitable for student consumption a small but important part of real mathematics. On the contrary. the book should be useful not only to students majoring in mathematics. I t is not meant for the college freshman with minimum preparation from high school. much of the material . That is. and they give the student almost no idea of what mathematics is really like. This is the reason t. a terminal course for liberal arts students with a good background in mathematics. and a refresher course for high school mathematics teachers. it is appropriate that any book on real mathematics should emphasize mathematical proofs. " In accord with the philosophy that students should be taught mathematics by exposing them to the mathematics of professional mathematicians. but it is within the reach of a large percentage of the college population. Since mathematics is a logical science. but also to adequately prepared students of any speciality. The student of more modest ability should keep this in mind in order to combat discouragement. An apt student with three years of high school mathematics should be able to study most parts of the book with profit.hat we have entitled our book "The Algebraic Foundations of Mathematics. The book treats those topics of algebra which are basic for advanced studies in mathematics and of fundamental importance for al1 working mathematicians. The book is written in such a way that the law of diminishing returns will not set in too quiekly. a senior leve1 course in abstract algebra that the average mathematics major in his class has a very distorted idea of the nature of mathematics. This book is not intended to be an easy one. I t is concerned with topics related to the principal number systems of mathematics. It is often painfully evident to an instructor in. Some sections digress from the main theme of the book. It should be emphasized that the starred sections are not the most difficult parts of the book. The student who masters the technique and acquires the habit of mathematical proof is well on his way toward understanding the nature of mathematics. starred sections can be omitted without loss of PREFACE Elementary college algebra and calculus courses usually cultivate manipulation a t the expense of logical reasoning. an elementary honors course for mathematics majors. say." For the most part. although it may be necessary to refer to them for definitions.

The appreciation of mathematical beauty is not like the enjoyment of literature. but only communicate with . 6-5) 7 (Omit 7-1. 4-6) (Omit starred sections) (Omit 6-1. 3 hours 1 (Omit starred sections) 2 (Omit starred sections) 4 5 Above all. 3 hours 1 Quarter. 7-3. and starred sections) 8 9 (Omit starred section) 10 (Omit 10-4) Development of the classical number systems Theory of equations 1 Semester. I t is much more difficult for a mathematician to explain his triumphs and masterpieces than for any other kind of artist or scientist.PREFACE vii in these sections is very elementary. music. 7-6. A star has been attached to just those sections which are not sufficiently important to be considered indispensable. this book represents an effort to show college students some of the real beauty of mathernatics. 4-3. and other art forms. 4-5. 3 hours 1 through 8 (Omit starred sections) 4 5 6 8 9 10 (Omit 4-1. but which are still too interesting to omit. 2 hours 1 Quarter. 6-4. Consequently. most mathematicains do not try to interpret their work to the general public. 6-5) (Omit starred section) (Omit 10-3. 2 hours 1 Quarter. The following table suggests how the book can be used for shorter courses. I t requires serious effort and hard study. 6-4. 5 hours 1 Semester. 5 hours Chapter 1 (Omit 1-3. 10-4) Elementary theory of numbers 1 Semester. 7-2. Course College algebra Time required 1 Semester. and starred sections) 2 (Omit starred sections) 4 (Omit 4-1) 5 (Omit starred sections) 6 (Omit 6-1. The complete book can be covered in a two semester or three quarter course meeting three hours per week. 3 hours 1 Quarter. 1-5.

who read most of the manuscript of this book. Addison-Wesley. Seattle. Curtis. and especially to our wives. Finally. For this reason. We are particularly indebted to Professors C. and anyone who will make the effort can enjoy the beauties of an intellectual domain which comes closer to aesthetic perfection than any other science. Washington January 1963 . Zuckerman. The swift and expert typing of Mary Pierce is sincerely appreciated. a mathematician is often considered to be a rather aloof person who lives partly in this world and partly in some other mysterious realm. However. we might never have finished this one. we are grateful to many friends for sincere encouragement during the last two years.viii PREFACE colleagues having similar interests. A. Acknowledgements. and H. R. This is in fact a fairly accurate conception. Our publisher. W. Writing a textbook is not a routine chore. and gave us many valuable suggestions. the door to the world of mathematics is never locked. S. Without the help of many people. Dean. has watched over our work from beginning to end with remarkable patience and benevolence. who have lived with us through these trying times.

. . . . . . . . . The cardinal number of a set . . . . . . . . . . . . . . Properties and esamples of measures Proof by induction . . . . . . . . Generalizations of the induction principlc . Properties of order . ix . . . Basic properties of the rational numbcrs .1-1 1-2 1-3 1-4 1-5 "1-6 *1-7 2-1 2-2 2-3 *2-4 2-5 *2-6 3-1 3-2 3-3 4-1 4-2 4-3 4 4 4-5 4-6 5-1 5-2 5-3 *5-4 *5-5 5-6 5-7 *5-8 6-1 6-2 6-3 Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The construction of sets from given sets . . . . . S h e division algorithm Greatest common divisor . . . . . . . More about primes Applications of the fundamental theorem of arithmetic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fields The characteristic of integral domains and fields . . . . . . . . . . . . . . . . . . . . The binomial theorem . . . . . . . . . . . . . . . . . . . Thc ordering of the integers . . . . . . . . . . . . . Linear congruences The theorems of Fermat and Euler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rings Generalized sums and products Integral domains . . . . . . S h e technique of induction . . . . . . . . . . . . . . . . . . . . . The fundamental theorem of arithmetic . . . . . . . . . . . . . Congruences . . . . . . . . . . . . . . . . . . . . . . . . Inductive properties of the natural numbcrs Inductive definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . General rules of operation Measures on sets . . . . . . . . . . . . . The algebra of sets Further algcbra of sets . . . . . . . . . . . . . . . . . . . . . . The definition of numbers Operations with the natural numbers The ordcring of the natural numbers Construction of the integers . . . . . .

. . . . . . . . . . . . . . . . 286 . . . . . . . . . . The algebra of matrices . . . . . . . . . . . . . . . . Polynomialsinseveralindeterminates . . . . . The unique factorization theorem for polynomials Derivatives . . . . . . The construction of & . . . . . . . . . . . . . . . . . . . . . . . . . . . Polynomials with rational coefficients . . . . . . . . . . . . . . . The fundamental theorem of algebra . . . Construction of the real numbers The completeness of the real numbers Properties of complete ordered fields Infinite sequences . . . . . . . . . . . The inverse of a square matrix . . The geometrical representation of complex numbers Polar representation . 213 . . . . . Polynomials . Applications of decimal representations . . . . . . . . . Greatest common divisor in F[x] . . . The solution of third. . 303 The construction of the complesnumbers . . . . . . . . . Sturm's theorem . . . . . . . . . . . . . . . . The division algorithm for polynomials . . . . . . . . . . Development of the real numbers . . . . . . . . . . . . . . . . . . . . The roots of a polynomial . . . . . . . . 298 . . . . . . . . . . . . . . . . 394 410 428 443 .and fourth-degree equations Graphs of real polynomials . . . . . . . . . . . . . . . . . . . . . . . . Comples conjugates and the absolute value in C . . . . Infinite series . . . Decimal representation . . Algebraic equations . . . . . . The coordinate line Dedekind cuts . . . . . . . . . . Systems of linear equations . . . 291 . . . . . . . . . . . . .X CONTENTS 6-4 6-5 7-1 7-2 7-3 7-4 7-5 7-6 "7-7 "7-8 *7-9 *7-10 8-1 8-2 8-3 8-4 9-1 9-2 9-3 9-4 9-5 9-6 9-7 9-8 "9-9 9-10 9-1 1 9-12 10-1 10-2 10-3 10-4 Equivalence relations . 218 . .

so that the student is exposed to a representative cross section of mathematics. We might select some important unifying concept of modern mathematics. The selection of a principal subject for this project poses difficulties similar to those which a blind man faces when he tries to discover the shape of an elephant by means of his "sense of feel. The reader should have successfully completed two years of high-school algebra and a year of geometry. Familiarity with ordinary numbers hides subtle difficulties which must be overcome before it is even possible to give an exact definition of them. However. An attempt will be made to answer the question "what are numbers?" in a way which meets the standards of logical precision demanded in modern mathematics. Our prerequisites for understanding this book are more modest. the development of these systems offers an opportunity to exhibit a wide variety of mathematical techniques and ideas. an older and perhaps familiar topic can be examined in depth. are objects of great usefulness and importance in mathematics. A typical description of such requirements in mathematical textbooks runs as follows: "This book has no particular prerequisites. This program has certain dangers. the real and complex number systems. I t is customary in technical books to te11 the reader what he will need to know in order to understand the text. the reader will need a certain amount of mathematical maturity. On the other hand. I t is this last more conservative program which will be followed. the end products of this work. Alternatively. part of the world of mathematics. " Usually such a statement means that the book is written for graduate students and seasoned mathematicians. will be very helpful. such as the notion of a group. For certain topics in the chapters on the complex numbers and the theory of equations." Only a few aspects of the subject are within reach. and it is necessary to exercise care to be sure the part examined is truly representative. a knowledge of the rudiments of trigonometry is assumed. The geometry. and explore the ramifications of this idea. Moreover. We will study the principal number systems of mathematics and some of the theories related to them. although not an absolute prerequisite. We do not expect that the reader will have much "mathe1 . especially for a student who does not see the point of this effort. but significant and representative. the purpose of this book is to exhibit a small. Checking the details in the construction of the various number systems is often tedious.INTRODUCTION As we explained in the preface.

1. . and use them in examples from the first chapter on. 4. 1. the real numbers:O. the integers are sometimes called whole numbers. or he can study these chapters in detail. but the resulting statements lack the clarity of these algebraic identities.-$. the integers: 0. can be stated without using the variables x.ernatives. +. 7.. the complez numbers: 0. relying on the knowledge of numbers which he already possesses. S T . On the other hand. -3. the rational numbers: 0. 4. -2. On the one hand. In this book the number systems will be considered at two levels. This innovation. i.. There are five principal number systems in mathematics: the natural numbers: 1. and z. The purpose of the remainder of this introduction is to provide some encouraging words on a variety of subjects. mathematics would not have progressed very far beyond what we now think of as its "beginnings." Indeed. due largely to the French inathematician Francois Vihta (1540-1603). the study of these number systems is the principal subject of arithmetic courses in elementary school and of algebra courses in high school. With a little encouragement almost any intelligent person can become a better mathematician than he would imagine possible. then it may possibly be the introduction of variables as a systematic notational device. Of course. The number systems. Basic laws of operation. we will assume at least a superficial knowledge of numbers. The latter road is longer and more tedious. With the possible exception of the complex numbers. 3. He can either skim the material in these chapters. etc. the names of these systems may not be familiar. -1..etc.. 1. 3 .and a. Variables. 1. 3.2 INTRODUCTION matical maturity. 2. *. 6. We suggest that this material be read quickly.$'a. y. For example. -1. For exarnple. occurred about 1590. 4. -*.etc. such as m. and 8 each present a critica1 study of one of these systems. The reader has two alt. Indeed." By using variables it is possible to express complicated properties of numbers in a very simple way. but it leads to a very solid foundation for advanced courses in mathematical analysis. Chapters 3. I a single event can be called the beginning of modern mathef matics. -3. -+. etc. Without variables.-4. 3. and the rational numbers are often referred to as fractions. one of the main purposes of this book is to put the reader in touch with mature mathematics. Some of the obstacles which a beginning student of mathematics faces seem more formidable than they reafly are. 2. each of these systems should be familiar. -3. then referred to later as it is needed. I t is hoped that our discussion will smooth the reader's way throughout the book. etc. 1 G. + + z+ y =y+z and x + (y+z) = (x-ky) i-2.

In this book variables will be used to denote many kinds of objects. some mathematical statements involve a very large number of variables. but he must also learn numerous abbreviations and symbols for common words. . y. such as numbers. etc. In these cases. representing arbitrary lines in a plane. a2. x2. or possibly an integer. In the simplest cases. etc. b7. in al1 cases. ak. it is intended that these variables stand for real numbers. y3. and in some cases. However. zi. Except for the use of abbreviations. xl. or lines. 215. One of the difficulties in learning mathematics is the language barrier. or perhaps only for integers. However. letter symbols with subscripts are usually employed.. even infinitely many. ~ Variable symbols are often used to denote a subscript on a variable letter. However. for example. the variable subscript is usually assumed to stand for a natural number. a variable is a symbol which represents an unspecified member of some definite collection of objects. Mathematical language. the grammar of mathematics is the same as that of the language in which it is written. we find expressions such as x l . ~~. 7 x2~. To accommodate the need for many variables. . Thus. then 1 and m have exactly one point in common" are variables. the letters of the alphabet are used as variable symbols. the symbols 1and m in the statement "if 1and m are two different nonparallel lines.j. points. The reader who doubts this should try to express in words the relatively simple identity The variables which are encountered in high-school algebra courses usually range over systems of numbers. x7.52. The given collection is called the range of the variable. Not only must the student master many new concepts and the riames of these concepts. variable symbols are often useful in other contexts. ~ etc. The notations used for variables in mathematical literature often puzzle students.the statement that "the product of a number by the sum of two other ilumbers is equal to the sum of the product of the first number by the secoild with the product of the first number by the third" is more simply and clearly-expressed by the identity More complicated laws would be almost impossible to stat*ewithout using variables. For instance xi. that is. For example. Sometimes double subscripts are more convenient than single ones. rational numbers. and a particular object in the range is called a value of the variable.

except that variables occur in place of the subject or object. Sentences may contain variables. the statement "There is a real number x such that xlOO5 7 ~ -~ 5 25x7 500 = O" is an assertion which is either true or false. These are formulas. the sentential function x y = 1 is transformed into the sentences + I t makes no sense to ask whether or not a formula such as x y = 1 is true. and 0=0 + must be counted as sentences. or y that it is understood that x and y are variables which range over real numbers." where p and q are sentences or sentential functions. the statement x" has this property of universal validity. although it is not obvious which is the case. by suitable substitutions for x and y. either true or false. Por example. On the other hand. + + + < j7 - - * It is true. the formula x y=y x has the property that every substitution of numbers for x and y leads to a true sentence. and x > 2. for example. For example. of the form " p implies q. and expressions which have the form of sentences." Expressions such as these are called sentential functions. 2 1=2. A sentential function which is true for al1 values of the variables in it is said to be identically true or identically valid (the adjective "identically is sometimes omitted) . "x is an integer" and "2 divides n.. Such a sentential function is usually called an identity.* There are other expressions of importance in mathematics which cannot be called sentences. provided "either x < y. For instance. x2 + 2x + 1 = 0. .2.x + zj = 1. According to this definition. such formulas as 2 . They have the property that substituting numerical values (or whatever objects the variables represent) for the variables converts them into sentences.4 INTRODUCTION A sentence in mathematical writing is any expression which is a meaningful assertion. As many as one-half of al1 statements in a mathematical proof may be implications. 1 1<3-2. such as + . for others it is not. that is. Many beginning students of mathematics have trouble understanding the idea of logical implication. Implications. Sentential functions which are not formulas may also have the property of being true for al1 values of the variables occurring in them. ' . For some values of x and y it is true.

then the implicatioii "p implics q" is a sentence. theii "p implies q" is a sentential fuiictioii. then its truth is completely dctcrmiiicd by the truth or falsity of p aiid q. " 3 = 1 implies 1 < 1 " is truc. i t is nriessary t h a t z be nonncgative l 1 I for p i t is necessary t h a t p i Icor this reasoii. For thc readcr's coilveiiience. Specifically. "if x = 1. this implicatioii is truc cither if p is false or q is true. The idea which C the statement "p implies q" 11s~a11y O ~ V C Y Sis that thc validity of the . then x is nonnegative x is nonnegative if x is positive x is positive only if x is nonnegative only if x is nonnepative is x positive x positive is a sufficicnt condition for x t o bc nonnegative for x t o be norinegative.or cxample. if cither p. "3 = 3 implies 1 < 3" is triie. togcthcr with cxamples of thesc locutions. aiid "for . I. and to undcrstand what it means. i t is sufficient t h a t x be positive x nonnegative is a necessary condition for x to be positive for x t o be positive. but "3 = 3 implics 1 < 1 " is falsc. "3 = 2 implies 1 < 2 " is truc. it is important to be able to recogiiize a11 implicatioii. Such statements as thcsc occur rcpeatcdly in aiiy book or paper oii mathcmatics.-c to eyual 1.'E = 1 orily if x is an integcr ". If p and q are both sciltences. or both p arid q are sciitential functions. ".ISTRODUCTION 5 1 p irnplies q q is implicd by p 1 If p. Iii case " p implies q" is a seiitence. Thc variety of ways in which mathcmaticians say "p implies q" is ofteri bcwildcring to studeiits. TVC list in Table 1 some of the forms in which "p implies q" may be written. thcn x is an integer ".r = 1 is a siificierit conditioii for x to be an iiitcgcr ". it is neccssary that x be ari iiiteger" al1 have thc samc mcaiiiilg. then q q if P p only if q only if q is p p is a sufricient condition for q i t is suficient t h a t p q is a necessary condition 1 1 x positive implies t h a t x is nonnrgative x noniicgativc is implicd by x bcing positive if x is positive. The expressions "x = 1 iniplics x is an iiiteger". (. or q. I t may seem strange to coiisidcr a scn tcilce "p implies q" to bc true eveii though there is iio apparent conilectioil 1)etwceii p aiid q. I t is false only if p is true and q is false.

" whose truth was previously admitted only with reluctance. In fact. the implication "y 2 = x implies y < x" is a sentential function which everyone would agree is identically valid. the first statement is identically true. An implication does not ordinarily have the same meaning as its converse. "p implies q" and "q implies p. Most of these forms are derived from the terminology for implications. Our convention concerning the truth of an implication becomes more understandable when we consider how a sentence of the form "p implies q" may be obtained from a sentential function by substitution of numerical values for the variables. Severa1 examples are given in Table 2. it is true for al1 values of x and y. From any statements p and q it is possible to form two different implications. I the implication "p implies q" and its converse f "q implies p" are both true. we obtain the sentence "3 = 1 implies 1 < 1.INTRODUCTION TABLE2 Form p is equivalent to q p if and only if q p is a necessary and sufficient condition for q p implies q. then n > O. and conversely Example +1 y+1 +1 y+1 y is a necessary and sufficient condition for x + 1 y+1 x y implies x + 1 y + 1.I ) ~ = 1 > O and -1 < O." These assertions obviously have different meanings. then the formulas x = y 1=y 1 are equivalent. For example. I t is hard to see how the truth of such an implication as "3 = 1 implies 1 < 1" fits this conception. + + + + + + + . For example." Each of these implications is called the converse of the other. Converse and equivalence. if x and y are variables which range over numbers. whereas the second statement is not true for al1 n . ( . and conx x x = y is equivalent to x = = y if and only if x = = = = = versely sentence q is somehow a consequence of the truth of p. In practice. then the statements p and q are said to be equivalent. for example. For example. However. That is. since "x = y implies x 1=y 1" and x and "x 1= y 1 implies x = y" are identically valid. the notion of equivalence of p and q is most frequently applied when p and q are sentential functions. then n2 > 0 " is the implication "if n2 > O. There are various ways of saying that two statements p and q are equivalent. the converse of the statement "if n > O. by substituting 1 for x and 1 for y.

The fact that an implication is logically the same as its contrapositive is often very useful in mathematical proofs. " This is because "not p implies not q" is the contrapositive of "q implies p. then x is an integer " is the implication "if x is not an integer. DeJinitions." For example. and a similar relation exists between the fourth and second columns." The implication "not q implies not p" is called the contrapositive of "p implies q. that is. it is not correct to claim that if p implies q and not q implies not p. it is permissible to establish that "p implies q" and "not p implies not q. " However. if we wish to prove that p and q are equivalent." I t is easy to see that the contrapositive of "p implies q" is true under exactly the same circumstances that this implication is itself true. The entries in the fifth column of Table 3 are determined by the combinations of true and false in the first two columns. Kevertheless. for example." while "not p implies not q" is called the inverse of ('p implies q. rather than proving a statement of the form "p implies q. the problem of showing that 222 is . beginning students frequently find such arguments difficult to understand. while the entries of the last column are determined from the combinations which occur in the third and fourth columns. beware . the entries of the third column are just the opposite of those in the first column. These are "not q implies not p" and "not p implies not q." This is logically acceptable.INTRODUCTION TABLE3 P Q not p false false true true not q false true p implies q not q implies not p true false true true true true false false true false true false f alse true true false true true Contrapositive and inverse. together with the corresponding truth or falsity of " p implies q" and its contrapositive (Table 3). Of course." it is easier to prove the contrapositive "not q implies not p. then p is equivalent to q. Sometimes. Simple mathematical proofs often consist of nothing more than showing that the conditions of some definition are satisfied. Also." two other implications can be forrned using p and q. Consider. In addition to the implication "p implies q" and its converse "q implies p. " p implies q" and "q implies p " are valid. the contrapositive of the statement "if x = 1. The most convincing way to demonstrate this fact is to make a table listing al1 of the possible combinations of truth values of any two sentences p and q. then x is not equal to l.


together with the appropriate axioms and definitions. theii n 1. must work hard to follow a difficult proof. The direct proof starts from certain axioms or definitions. it is a matter of experience that understanding proofs is the most difficult aspect of mathematics. Applying the rule of detachment to (a) aiid (i) gives (iii) n 1 is a natural number. then n 2 m is impossible. then n n 1 is impossible. Substituting n 1 for m in (ii). a contradiction of some kind is obtained by means of a logical argument. The statements follow each other relent- + + > + + + + + + > + + > + . together with (b) and the rule of detachmeiit yields (vii) n 2 n 1 is impossible. even though it is often used unconsciously in everyday thinking. which. The rule of detachment can nolv be applied to (iii) and (iv) to conclude that (v) n 2 n 1. then n m. we assume that there is a largest natural number. The statements (v) and (vii) provide the contradiction which completes this typical indirect proof. Most people. (b) n < n (c) if n < m. Let this number be deiioted by n. Then. let us show by an indirect proof that there is no largest natural iiumber. The indirect proof begins by assuming "hat the statement to be proved is false. Our indirect proof begins with the assumption that the statement to be proved is false. using this assumption. This. The second method.INTRODUCTION From this we can infer such formulas as The logical structure of a mathematical proof may have one of two forms. For example. substituting n (vi) if n < n 1. that is. the so-called indirect proof. is perhaps less familiar. mathematiciaiis included. 1 for m iii (c) gives However. (ii) if m is a natural number. In spite of the elementary character of the logic used by mathematicians. (a) if n is a natural iiumber. This proof uses three general properties of numbers. From this contradiction it is inferred that the statement originally assumed to be false must actually be true. and proceeds by application of logical rules to the required conclusion. To say that n is the largest natural number means two things: (i) n is a natural number. then n n 1. for our purposes can be considered as axioms: 1 is a natural iiumber. we obtain (iv) if n 1 is a natural number.

The result of this labor is only the beginning.he argument has been checked. which may not be easy to find. . After the step-by-step correctness of t.10 INTRODUCTION lessly. real mathematics is not easy. Each step requires logical justification. it is necessary to go on and find the mathematical ideas behind the proof. Truly.

" The problem of finding a satisfactory mathematical definition is far more difficult than it might seem. Mathematics is not alone in using this idea.CHAPTER 1 SET THEORY 1-1 Sets. one speaks of the Smith family. I t should be emphasized that in thinking of a collection of objects as a set. and geometry borrow freely from elementary set theory and its terminology. meaning the collection of people consisting of John Smith. if we referred to Mrs. The central idea of set theory is that of dealing with a collection of objects as an individual thing. Indeed.1. A set is an entity consisting of a collection of objects. Thus. Smith's wardrobe. no account is takeil of the arrangement of the objects or any relations between them. analysis. al1 of mathematics can be founded on the theory of sets. we say that the sets are equal. When this is the case. piece of clothing belonging to Mrs. 11 . The notion of a set enters into al1 branches of modern mathematics. As is to be expected. The uncritical use of sets can lead to contradictions which are avoided only by imposing restrictions on the naive concept of a set. a deck of * This statement cannot be considered as a mathematical definition of the terni "set. and any intelligent person can learn enough about set theory for most useful applications of the subject. we would be treating as a single thing the collection of individual pieces of clothing belonging to Mrs. * Two sets are considered to be the same if they contain exactly the same objects. Finding a definition of "set" which is free from contradictions and which satisfies al1 mathematical needs has for 75 years been a central problem of the logical foundations of mathematics. Thus. DEFINITION 1-1. Here we have only supplied the synonym "collection" for the less familiar term "set. and their son William. Smith. The mathematical use of this device of lumping things together into a single entity differs from common usage only in the frequency and systematic manner of its application. an idea with such a wide range of application is quite simple. In the example given above. these difficult aspects of set theory can be ignored in almost al1 mathematical applications of the theory. Smith's wardrobe." I n mathematics. A set is usually determined by some property which the elements of the set have in common. Smith defines the set which we cal1 Mrs. the property of being a. his wife Mary. a definition is supposed to completely identify the object being defined. for example. for example. Also. and many occurrences of it are found in everyday experience. Algebra. The objects belonging to a set are called the elements of the set. Fortunately.




52 cards, considered as a set, remains the same whether it is in its original package or is shuffled and distributed into four bridge hands.

x2 = o. EXAMPLE 4. The set of numbers a on the real line (Fig. 1-1) which satisfy -1 5 a 5 l.

EXAMPLE 1. The set consisting of the numbers O and 1. EXAMPLE 2. The set of numbers which are roots of the equation x2 - x = 0. EXAMPLE 3. The set of numbers which are roots of the equation " x

EXAMPLE 5. The set of al1 numbers x/2, where x is a real number which satisfies -2 5 x 5 2. EXAMPLE 6. The set of al1 points a t a distance less than one from a point p in some plane. EXAMPLE 7. The set of al1 points inside a circle of radius one with center a t the point p in the plane of the preceding example. EXAMPLE 8. The set of al1 circles with center a t the point p in the plane of Example 6. EXAMPLE 9. The set consisting of ¿he single number O. EXAMPLE The set which contains no objects whatsoever. 10.

According to our definition of equality of sets, we see that the sets of Examples 1, 2, and 3 are the same. Although O occurs as a so-called double root of the equation x 3 - x 2 = O in Exaniple 3, only its presence or absence matters when speaking of the set of roots. The sets of Exanlples 4 and 5 are the same, as are those of Examples 6 arid 7. I we consider f a circle to be the same thing as the set of al1 of its poiiits, then the eleineiits of the set of Example 8 are themselves sets. Sets of sets will be studied more thoroughly in Section 1-5. The set described in Example 9 contains a single element. Such sets are quite common. I t is conventional to regard such a set as an eiltity which is different from the eleinent which is its only member. Even in ordinary conversation this distinctioii is often made. If Robert Brown is a bachelor with no known rclsttions, then we would say that the Brown family consists of one meniber, but we would not say that Mr. Brown consists of oiie nlember. The reader may feel that the set of Example 10 does not satisfy the description given in




Definition 1-1.1. However, it is customary in mathcmatics to interpret the term "collcction" i11 such a ivay that this ~iotion includes the collection of no objects. Actiially, thc sct containing no elements arises quite naturally in many situations. For instante, in consideriiig thc scts of real numbers which are roots of algebraic equatioris, it ~voilldbe awkjvard to makc a special rase for equations like x2 1 = O, ivhich has no real roots. 13ecause of its importailcc, thc set coiitaiilirig no elemcilts has a special namc, thc empty set, and it is rcprcscntcd by a special symbol, a. When it is neccssary to cal1 atteiltioii to the fuct that a set A is not the empty set, then ve will say that A is nonempty. Oiie reason that set theory is used in so many brailches of mathcmatics is thc versatility of its notation. As aiiyoiie ~ v h o has studied elcmentary algebra might expect, thc, letters of the alphabet are used to rcpresent sets. 111 this book, sets will be represeiitcd by capital letters, and the elements of sets ~ i l usually bc represeiitcd by small letters. l Thc statemeiit that an object a is an elemeiit of a set A is symbolized by


We rcad thc cxpression a E A as "a is in A," or sometimes, "a in A." To give a specific example, lct, A be t,he set of roots of the equation x2 - x = O (Example 2). Theii

IVe often wish to cxpress the fact t,hat an object is not contained in a certaiii set. If a is not an element of t,he set A, \ve writc

and rcad this cxpressioii "a is iiot in A . " Thus if A again is the set of Examplc 2, wc would havc 2 4 A, 3 4 A, 4 4 A, etc. I t \vas mentioned earlier that a set is ofteri dcfincd by some property possessed by its elements. There is a very uscful ilotatioiial device iri set theory which gives a standard method of symbolizing thc sct of al1 objects having a certain property. For instaiice, the sets of Kxamples 2 and 4 are respcct ivcly writtcn {2-~z2-x=0], and {al-l5a<l).

The symbolic form (*I*] is sometimcs called the set builder. In using it, we replace ttie first asterisk by a variable elemcnt symbol (x aild a in the examples), aiid the second astcrisk is rcplaced by a meaningful condition which the object represented by thc variable miist satisfy to be an element of the set (x2 - n: = O and - 1 5 a 5 1 in the examplcs). Thus, the


set builder occurs in the form (xlcondition on x) (or with some variable other than x), and this expression represents the set of al1 elements which satisfy the stated condition. Often, the totality of possible objects for which the variable stands is evident from the condition required of the variable. For example, if the real roots of algebraic equations are under discussion, then in (x/x2 - x = O), it is clear that x stands for a real number, and that the set consists of the real numbers which satisfy x2 - x = O In {al-1 5 a 5 11, it may not be clear . what kind of numbers are allowed as values of the variable. If it is necessary to be more explicit, we would write

where R is the set of al1 real numbers. Similarly, in Example 6, the set builder notation would be

where P is the set of al1 points in some plane and d(p, q) is the distance between points p and q in the plane. Here, the variable q can take as its value any point in the plane P, and the set in question consists of those points in P which satisfy d(p, q) < l. Other forms of the set builder notation for this example are

I t is often convenient to use general symbols or expressions in place of the variable element in the set builder notation. For example, (x21x E R ) , where R is the set of al1 real numbers, is the set of al1 squares of real numbers; (x/ylx E N, y E N), where N is the set of al1 natural numbers, is the set of al1 positive fractions. A variation of the set builder notation can be used to denote sets which contain only a few elements. This coilsists of listing al1 of the elements of the set between braces. For exalnple, the sets of Examples 1 and 8 would be written (0,l) and (01, respectively. I t is sometimes convenient to repeat the same element one or more times in the notation for the set. Thus, in Example 3, we might first write the set of roots of x3 - .r2 = x x (x - 1) = O as (O, 0, 1) , since O is a double root. Of course, by the definition of equality, {OJO,1) = 0 l . The notation (0, O, 1) conveys no more iiiformation about

b) as the sct containing variable quantities a and b. 3. As siich. &4. b) = (1. There is aiiother good reason for allowing repetitioii of one or more elements in the notation for a set. A A . denotes the set of al1 natural iiumbers. a. . it is often coiivenient to regard {a. a. Consider the example {a. b. Wc can think of this as the set whose members are the lettcrs a and b. equal to oiie f r o n ~ . 1)' = (1).). it would become a specific sct if particular values where substituted for a arid b. Similarly.if a = 1. A is a subset of R by writing A 2 B or B 2 A. . b = 1. If A is iiot a subset of B. If A 2 B. . 2 = (0. t h e i ~{a. lior examplc. the collection of sets (a. A is not thc same as the set R).1-11 SETS 15 xs . then each choice of valucs for a and b determines a set whose members are these selected natural riumbers. . b). ~ v c writc A g R or B 2 A . The set of al1 even integers. is a proper 1l 3 subsct of the sct Z of al1 intcgers. For example. . EXAMPLE The set of a11 poirits a t distance Iess than one from a point p 12. Scts containing many elements can often be represented by listing some of the elements between braces aild using a sequence of dots to indicate omitted elemciits. For instai~ce. 111 mathematical applications. then A is called a proper subset of B and iii this case ~ v c writc A C R or B > A . a and b may take on the samc value. 2. b] . Aily set A is a subset of itself. Some infinite sets caii also he represented in this way. EXAMPLE. DEFISITIOS 1-1. .x2 = O than {O. This difficiilty would be increased iii more complicated examples. I t is customary t o express the fact that. (a. it is clear that {1. in a plane P is a proper subset of the set of points of P a t distance less than or p. In this example. according to this definition. but A # R (that is. b) represeiits the same set a s {a. 2165) represents thc set of al1 natural numbers from 1 t o 2165. if we allow natural numbers to be substituted for a and b. 1) does. . &2. b) dctermined by substituting values for a and b would be considcrably more difficult t o describe. however. E'or example.2. A set A is called a subset of the set R (or A is included i n R) if every element of A is an element of B. If we did not allow repetition of the elements in designsting sets.

If A c B and B E A . then A = B. 1) C (O. . However. 2. although direct consequences of our definition. no element of can be found which is not in A. In this book. That is. Thus. The inclusion relation has three properties which. 2. Z designates the set of al1 integers: {. I n mathematical literature. 5 1). A = B. E A for every set A . Q. We will prove property (b) in detail. EXAMPLE (a10 < a 5 1) C {a10 5 a 14. This notation. is a subset of every set may seem straiige. a considerable amount of variation in notation can be found.16 SET THEORY EXAMPLE {O. The terminology and symbolism introduced in this section will be used in the remainder of this book. 3. . 1. For any sets A. The reader should carefully check to see that in each of these examples the condition of Definition 1-1. . the number systems of mathematics. EXAMPLE Q. The fact that Q. . The first of these has already been noted. . (a) A E A . -3. 13. by Definition 1-1. then every element of A is an element of B and every element of B is an element of A. (c) if A E B and B S C. . this condition is certainly satisfied. (b) if A c B and B 5 A. would be recognized by most modern mathematicians. 2 .3. are quite important. An example is the practice of denoting the empty set by the symbol a. 3. R designates the set of al1 real numbers. Certain sets occur so frequently in mathematical work that it is convenient to use particular symbols to designate them throughout any mathematical paper or book. 4). . b E N) . the letters N.2 is satisfied. leaving the proof of (c) t o the reader.1. and R will not be used to denote any set other than the corresponding ones listed above. -1. 1. Proof. . -2.0. Q designates the set of al1 rational numbers: (albja E 2. but it is by no means . B. it is certainly true according to our definitions: every element of is an element of A.) . considered as sets. THEOREM 1-1. . or in other words. has no elements. though not universal. Since Q. 15. 2. Throughout this book. A and B contain exactly the same elements. and C. will occur repeatedly.) . We therefore adopt the f ollowing conventions : N designates the set of al1 natural numbers : (1. then A E C.

write expressions for the follon-ing sets. y E Q). (c) The set of al1 integers which leave a remainder of one mhen divided by five.y)(x = y)) 3. and c are natural numbers less than or equal to 3. and if A C B and B G C. null set. 2. .2x2 . ensemble. F o r t h e reader's coiivenience. b. (d) The set of al1 rational numbers greater than five. Element of a set: member of a set. a ~ A : A 3 a . A ~ ~ . l. b.1-11 SETS 17 universal. 2 ) (e) { X ~ X y2 z2.3(c). (*]*) : (* : *). 7. b. 5. List the following collections of sets. b.x 2 = 0. CP: o. collect ion. Te11 in words what sets are represented by the following expressions. vaciious set. [* : *l. . and c are integers between -2 and 4. .n. ~ E A . Set : class. then A c C. (f) The set of solutions of the equation x3 . 6. [*)*]. where a. n E 2). Prove that if A B. -1 (4 {a. (a) The set of al1 even integers. A. (b) The sets {a2 a 1). (a) The sets {a. Describe the following sets by listing their elements. (b) The set of al1 integers which are divisible by five. zero set. then A C C. {n(n = m2. + + + + + + . (a) {x E NIx > 10) (b) {x E QIX . B C C. m E Z). 6 .2x 4. R. Prove Theorem 1-1. aggregate. poiiit of a set. y E R. E m p t y set : void set. State al1 inclusion relations which exist between the folloñing sets: N.3 E N) ( 4 ( 5 . 2. Y. a 4 A : ~ E ' A .2x = 0) 1 = O ) (c) {x1x2 . Q. z E R. c). Using the set builder form. y E R. x2 . {zjx = y2. 7. (c) The three element sets (a. c). b. where a is a natural number less than or equal to 5 .y2 = (x . (a) {xIx2 = 1) (b) {x1x2 . \ve list some common alternative termii~ology . (e) The set of al1 points in space which are inside a sphere with center a t the point p and radius r . where a. { X ~ X = y2 . the set of al1 even integers.

the number n of elements in the set may be one. Roughly speaking. There are many synonyms for the expression "cardinal number of A. it is natural to say that the cardinality of (P is zero. the set R of al1 real numbers. . two. . His researches have had a profound influence on al1 aspects of modern mathematics. and the sets of Examples 8 and 9 are finite. Of course. These descriptions of finite and infinite sets. the set of Examples 1. if its elements cannot be counted." Besides the term "cardinality of A. we will examine one of Cantor's most important ideas. until at some stage we reach an nth element and find that there are no more left. 0 . Returning to the examples considered in Section 1-1. In the remainder of this section. four. and 3." which we have already mentioned. three. a finite set is either the empty set.4)1=3. If A is a finite set. and of the cardinal number of a finite set are too vague to be called mathematical definitions. and such expressions as card A or N(A) can also be found. I t was remarked above that the empty set 4> is regarded as being finite: Since @ has no elements.1. or any natural number whatsoever. The first man to systematically study the cardinal number concept for arbitrary sets (both finite and infinite) was Georg Cantor (1845-1918). . 1 1-2 The cardinal number of a set. The symbolism A is perhaps even more common (but difficult to print and type).18 SET THEORY [CHAP. one finds such expressions as "power of A " and "potency of A. etc. . and so on. Note that the empty set @ is considered to be finite. . 2. (@l = 0. or a set in which we can designate a first element. We will use the notation iA 1 to designate the cardinal number of A . the set Q of al1 rational numbers. Thus. the set Z of al1 integers. and see in particular how it enables as to explain the concept of a finite set in more exact terms. although almost everyone with some experieiice learns to distinguish finite from infinite sets. that is. it is meaningful to speak of the number of elements of A. I t is not altogether easy to explain the difference between a finite and an infinite set. a third element. while the sets of Examples 4 through 7 are infinite. Examples of infinite sets are the set N of al1 natural numbers. we have not said anything about the cardinal numbers of sets which are not finite. A set is said to be infinite if it is not finite. in symbols. " The notation IA( for the cardinal number of the set A is not universal either. These examples show that some of the most important sets encountered in mathematics are infinite. The simplest and most important classification of sets is given by the distinction between finite and infinite sets. Moreover. As examples. a million. This number is called the cardinal number (or cardinality) of the set A. a second element. l a = l(O.

2 t 2. we would use up al1 of N and have nothing left to associate with O. (. .1. then 3 must correspond . .2.1-21 THE CARDINAL KUMBER OF A SET 19 DEFINITION 1-2. . 0. The reader should study the following examples to be sure that he fully understands the meaning of the fundamental concept defined in Definition 1-2. b. DEFINITION 1-2. . The more general fact that there is no one-to-one correspondence between (1. . . not al1 of the numbers of N can be paired with themselves. -3. m) and (1. .c~-eenthe set Z = 3. 3. e]. . . and each element of B is matched with exactly one element of A . Usiiig Defiiiitioii 1-2. -2.1. -2. 2. 2). 2. EXAMPLE Let A = (1. and B = (a. . we find that more than one element of A must correspond to a single element of B. . sible one-to-one correspondences between -4 and B: 1 2 b 3 1 2 3 1 2 3 1 2 3 1 2 3 a 1 2 3 5 a 5 c 5 b c a 5 5 5 5 c a b 5 5 5 b a c 5 5 5 c b 5 5 5 c b a EXAMPLE It is impossible to obtain a one-to-one correspondence between 2. . 3 is paired with both 1 E A and 3 E A. I n the correspondence 1 t 1. . -1. . Let A be a set and let n be a natural number. which will be discussed in the next two chapters. . . -1. 3. 2. 1. Then the cardinal iliimber of A is n if there is a one-to-oiie correspondence between A and the set (1. the element 1 E B . .) of al1 natural numbers. . 1. 2. 3. is called a one-to-one correspondence between A and B. . The elements of Z and N can be paired off as follows: Note that in order to construct a one-to-one correspondence between Z and N. so that the correspondence is not one-toone.1. If 1 t 1 and 2 t 2. to 1 or 2 in B. . 2. . . A similar situation occurs in al1 possible correspondences between A and B. 3. . . No matter how we try to pair . . 31. off the elements of 4 and B. . EXAMPLE There is a one-to-one correspondence bet. 2. consisting of the first n natural . Otherwise. 3) and the set B = (1. n). we caii clarify the notioil of a fiiiite set. Then there are six pos1. A pairing between the elements of two sets A and B such that each element of A is matched with exactly one element of B. n) if m < n is also true. . 3. 2.) of al1 integers and the set N = (1. This can be proved using the properties of the natural numbers. the set A = (1.

so do the sets N and 2. it can be asserted that the number of students in the class is the same as the number of chairs in the room. the two sets {1. 2 t 2 and 1 * 2. a3. Two sets A and B are said to have the same cardinal number. . Any one. EXAMPLE Every set *4 is equivalent to itself.) (without repetition) exhibits the one-to-one correspondence between A and (1. every chair in the room is occupied and no students are standing. n). The usual practice of writing a finite set in the form {al. a. b. without referring to the exact number of elements in A and B. to-one correspondence of a set with itself is called a permutation of the set. . DEFINITION 1-3. a. For example. 2). The reason is obvious. By Example 3. 1 numbers.3. obviously a one-to-one correspondence of A with itself. az. and their cardinal numbers. Otherwise A is called in$nite. or the same cardinality. A set A is Jinite if A = @. . the existence of any one-to-one correspondence between A and B is enough to guarantee that A and B have the same cardinal number. In accordance with Definition 1-2. By Example 1. Then without counting the number of students and the number of chairs.2. let L4 = (1. This idea is illustrated in the following example. according to Example 2. Cantor observed that it is possible to say when two sets A and B have the same number of elements. 2 o 1. then there are other ways of defining a one-to-one correspondence of A with itself. . there is a one-to-one correspondence between the set of al1 students in the class and the set of al1 chairs in the room. a l a 2 a3 . which were given a t the beginning of this chapter. If A contains more than one element. 2. This definition is no more than a careful restatement of the informal descriptions of finite and infinite sets. c ) have the same cardinality. Suppose that in a certain mathematics class. . or there is a natural number n such that the cardinal number of A is n. 3) and (1. However. namely. . or to be equivalent if there exists a one-to-one correspondence between A and B. 3. . . there may be many one-to-one correspondences between A and B. As in Example 1.2.3) and {a. Then there are two one-to-one correspondences of A with itself : 1 * 1. . the sets {1.3.20 SET THEORY [CHAP.2) do not have the same cardinal number . since a ++ a for a E A is 4.

then there is a one-to-one correspondence between A and B : so that A and B are equivalent in the sense of Definition 1-2. .) and B = {bl. EXAMPLE The set iV of natural numbers has the same cardinality as the 5. in some sense.gram as in Fig. This can be seen with the aid of a dia. and 2.1-21 THE CARDINAL KUMBER OF A SET 21 If A = {al. n E N) of al1 positive rational numbers.) are finite sets which both have the cardinal number n. certain infinite sets are "bigger than" others. if A and B are finite sets. set F = (m/nlm E N. . which are equal to numbers which have skipping fractions like S. . An even more striking example of this phenomenon is the following one. By following the indicated path. One of Cantor's most remarkable discoveries was that infinite sets can have different magnitudes. a2. b2. that is. Example 3 has already illustrated the fact that infinite sets which seem to have different magnitudes may in fact have the same cardinality. The important fact to observe about Definition 1-2. 2. To appreciate this fact requires some work. b. . .3 is that it applies to infinite as well as finite sets. . then A and B are equivalent if 1 A 1 = 1 BI.3. . . I we number the fractions in the order that they are encountered. S. each fraction will eventually f be passed. That is. 1-2. a. .

1 been ~reviouslypassed. For example. nk-l. . Although Cantor's work on the theory of sets was highly successful in many ways. we will be able to present Cantor's proof that these sets do not have the same cardinal numbers. . The fact that the set A of real algebraic numbers has the same cardinality as N and the result that R and N do not have the same cardinal numbers together imply that R # A . ni. But this is not the case. but Cantor's results immediately imply that they do exist. one might guess that al1 infinite sets have the same cardinality. . Later. . . nk-1. real numbers r which are solutions of an equation of the form where no. 4 3 . 1%-e the desired one-to-one correspondence between get N and F: Cantor showed that many important collections of numbers have the same cardinality as the set N of al1 natural numbers. . Judging from these examples. . . since m / n is a root of the equation nx . Cantor proved that it is impossible to give a pairing between N and the set R of al1 real numbers.22 SET THEORY [CHAP. this is true for the set d of al1 real algebraic numbers. . that is. nl. nk integers. n k are any integers. A also includes numbers like d2. . One of these ranks among the three most famous unsolved problems in mathematics (the other two: the Fermat conjecture. . This interesting fact is by no means evident. it raised numerous new and difficult problems.m = O. The conjecture that no such set S exists . and the Riemann hypothesis. 4%) . I t is fairly hard to exhibit such a real number. there are real numbers which are not solutions of any equation with no. urhich is too technical t o explain in this book. a. That is. which we will describe in Chapter 5 . The set A includes al1 rational numbers.) Cantor posed the problem of urhether or not there is some set S of real numbers whose cardinality is different from both the cardinality of N and the cardinality of R.

ed {al.E N. That is. a3. the definition says the same thing as before. State which of the following sets are finite. it has been neither proved nor disproved.) is denumerable (or possibly finite. Thus. List al1 one-to-one correspondences between -4 and B. .411 (b) ( X ~E Z. 3.1-21 THE CARDINAL NUMBER OF A SET 23 is known as the continuum hypothesis. . x < 5 ) X (e) {xlx E N. The converse statement is also true. . if S is denumerable. . I t was first suggested in 1878. . a2. then B is equivalent to A. 2. . 2 . . 1. Y. where a. a2. An infinite set is called denumerable if it has the same cardinality as the set N of al1 natural numbers. n < 1000) (d) (nln E N. a3. then S can be written {al. n2 5 36) (e) {n31n. O < x < 1) 2. then it is possible to pair off the elements of S with the numbers 1 . l. 2. (1-2. Let A. a2. then A is equivalent to C. We conclude this section by listing for future reference the following important propert'ies of the equivalence of sets. (a) {(x. Thus. the elements can be labeled a l . a one-to-one correspondence between A and B is a one-to-one correspondence between B and A. and (c) if A is equivalent to B and B is equivalent to C. 2 E (0. That is. . If S is denumerable. . 3 . the set of al1 integers and the set of al1 positive rational numbers are examples of denumerable sets. . (b) if A is equivalent to B. c. As we have shown in this section. . b. n3 27) 3. B.1. 41. a3. . Then (a) A is equivalent to A . n3 5 27) (e) (n21n E Z. I t has already been noted in Example 4 that (a) is satisfied.3 = 0) (d) (x1x E Q. . Hence. a set which can be designat. 2)) Y E (3. is the symbol which stands for the element corresponding to the number n. d). if A and B are interchanged in Definition 1-2. since distinct symbols might represent the same element of S). . Let A = (1. 4) and B = (a.z>lxE (0. n2 5 86) (a) {nln E N. and to date. and C be arbitrary sets.4). Property (b) follows from the fact that the definition of a one-to-one correspondence is symmetric. What is the cardinal number of the following finite sets? (b) {nln E 2. The proof of (c) is left as an exercise for the reader (see Problem 8). < .). with the elements of S listed in the form of a sequence. x2 .

l ) . elements of A can be fornied: (1. EXAMPLE A man has tulo pairs of shoes. b). 6. 1 4. 5. (3. There are severa1 other methods of building sets from given sets.24 SET THEORY [CHAP. namely.1. 2).) of natural numbers. Let A be a denumerable set. + 1-3 The construction of sets from given sets. and let B be a finite set. Two ordered pairs are the same if and only if they have the same first element and the same second element. (3. b) is called an ordered pair of elements. b) = (e. 7. The second construction leads from a single set X to another set called the power set of X. Show that the set S = ((2. Show that the pairing x ++ y.1." DEFINITION 1-3. where y = 1/(1 x) is a one-to-one correspondence between A and B. List the pairs which correspond to al1 numbers up to 21. Then the following distinct ordered pairs of 1. b) = (e. 2)) (1. Using the method by which we proved that the positive rational numbers have the same cardinality as the set N of natural numbers. n. where a is the first elernent and b is the second element. b) = ({a. then the resulting object (a. l). The definition of the product of two sets is based on the concept of an ordered pair of elements. (1. b). what are the possible combinations of shoes * There is a simple way to define an ordered pair in the framework of set theory. and that sets B and C have the same cardinality. a) only if a = b. indicate how to prove that N has the same cardinal number as the set of al1 pairs (m. However. one brown pair and one black 2. An ordered pair is then a definite object. Suppose that a and b denote any objects whatsoever. 2. and let B be the set of al1 real numbers y satisfying O < y < l. pair. we will discuss two important methods of constructing sets from given sets. 3). Thus we arrive a t the following definition. we will use the informal description given in the text and regard this property of ordered pairs as a definition. a). By Definition 1-3. I the elements a and b are grouped together in a definite order f (a. but they will not be considered in this book. 3). . Prove that the set of al1 rational numbers Q has the same cardinality as the set of al1 natural numbers N. 2). b) = (b. and i t is possible to prove that (a. (2. y)lx E A. I n this section. for objects a and b let (a. The first process combines two sets X and Y to obtain a set called the product of X and Y. y E B) is denumerable. 3)) (2. this is true in general. d) if and only if a = c and b = d. 8. If he dresses in the dark. (a. Suppose that sets A and B have the same cardinality. Note that (a. (3. (2. Show that A and C have the same cardinality. d) if and only if a = c and b = d. l). EXAMPLE Let A = (1. 3). Let A be the set of al1 positive real numbers x.

3)). 3)) (2? 1)) (2. = ((n. left black shoe) and Y = {right brown shoe. 1). in symbols X x Y = {(x. 2). 2). m E N ) . 3). Thus. I n U X ( V X TV) i t is just the other way around: the elements are ordered pairs in which the first element is a number and the second element is an ordered pair of numbers. 2). ((2. l). and the second element is a number. ((1) 3). Note that the elements of ( U X V ) X TV are different from al1 of the elements of U X ( V X TV). the order in which the sets are taken is significant. 2. (left brown shoe. The reader must be careful to make a distinction between ((2. Then the product of the sets X and Y is t. that is. 0. 3)). (3. I n fact. (3. the elements of ( U X V ) X T are ordered pairs whose V first element is an ordered pair of numbers. The product of X and Y is denoted by X X Y. ((2.2. 3). ((3. y). for example. ((3) 3). right black shoe). V = i(1. and 3 are exactly the 4. I t follows that U X V f V X U . l). l). (3. V = (1. and V X U = ((1. (3. (1. Then the set of al1 possible combinations which the man might wear is the set of al1 ordered pairs with the first element taken from the set X and the second element taken from the set Y. (left black shoe. the set of al1 pairs (left brown shoe. right black shoe). right black shoe). 3)) ((2. EXAMPLE The set of al1 ordered pairs of natural numbers is the set 3. 2). ((1) 3). . 3). m)(n E N. (1. right brown shoe). (left black shoe. ((3. ((2) 1).1-31 THE CONSTRUCTION OF SETS FROM GIVEN SETS 25 which he can put on? Let X = {left brown shoe. and P = (2. 2). 31. 2. DEFINITION 1-3. S Thus. 3). Thus. 1). (3. 3). 3). 3). 5. 2). We also have ux (U x x = {(O. l). 3)). 3). y)lx E X ) y E Y). V Then EXAMPLE Let U = (1. 3)). in forming the product of two sets. 3)) 2)) ((1) l). Let X and Y be sets. (1. l). elements of the products A X A. EXAMPLE The ordered pairs listed in Examples 1. and N X N. respectively. where x E X and y E Y. right brown shoe). 3) and (2. (1. l ) . X X Y.he set of al1 ordered pairs (x. 3). 2)) ((3. 3).

It is easy to prove that these results hold in general. every element of Y X X is an ordered pair (z. . x. .(xl. . 2. Indeed. = (1. it is true that U X V is equivalent to V X U and ( U X V) x W is equivalent to U X (V X W). . Xn. - The definition of the product of two sets can be generalized to a finite collection of sets X1. if n = 3.). and TV be the sets defined in Example 5. is the set of al1 ordered strings of elements . y) between X X Y and Y x X. X X. so that (x. We will prove (a) and leave the proof of (b) as an exercise for the reader. with x E X y E Y.26 SET THEORY [CHAP. Then I t is possible to generalize Theorem 1-3. For example. and Z be sets. Y. (y. 3).3. . . V = (1. X B . The product of these sets. x) E Y X X. that is 7. DEFIXITION 1-3. We turn now to a second method of obtaining a new set from a given set X.4.3 to products of finite collections of sets (see Problems 6. . . then U EXAMPLE Let U. 31. V. and (b) (X X Y) X Z is equivalent to X X (Y X 2 ) . The set of al1 subsets of X is called the power set of X. THEOREM 1-3. . where xi E Xi for i = 1. Then (a) X X Y is equivalent to Y X X. . Even though U X V # V X U and (U x V) X W # U X (V x W) in Example 5. as we see by counting the elements in each of these sets.. x) is the desired one-to-one correspondence The pairing (x. Proof. and is denoted by P(X). n. Every element of X X Y is an ordered pair (x. and TV = (2. then (y. y) E f X X Y. then X X @ = cP X X = cP. 1 EXAMPLE If X is any set. and 8). x2. empty set contains no element. deiloted by X1 x Xz X . since the 6. we must show that there is a one-to-one correspondence between X X Y and Y X X. y). Let X be any set. 7. . . x). w). Let X. According to Definition 1-2. . with z E Y and w E X. y) can be matched with (y. 3). there cannot be any ordered pair whose first or second element belongs to the empty set. 2.3. I (x.

Suppose that X is a finite set. it is possible to add elements one by one. and if A # B. a. Let X1 be a set which is obtained by adjoining to X a new element a which is not in X. EXAMPLE If X 8. (b).. associate a symbol xk.1-31 THE CONSTRUCTION OF SETS FROM GIVEN SETS 27 Thus. = (a. then P(X) = (a). . I f 11 = n. the elements of P(X) are precisely the subsets of X. . then {x) E P(X).5. The empty set corresponds to l. If X is an infinite set. then P(X) (a. For instance. That is. .. P(X) is infinite if X is infinite. a. . EXAMPLE If X '9. then A # Bl. the set A itself and the set Al obtained by adjoining a to A.. x. = (a). THEOREM 1-3. .. until a set X containing n elements is obtained. With each ak in X. the result is a sum of distinct products f of x'a (except the first term. x There is another way to prove this theorem which is worth examining. Therefore.. (a}). the elements of X1 are al1 of the elements of X..x. . . Note that al1 of the sets so constructed are different. (a). a.). = = = a . there are just twice as many subsets of X1 as there are subsets of X. b). Then every subset of X1 either does not contain a and is therefore a subset A of X. a. . A # Al.. then P(X) (a.. or else it contains a and is therefore obtained from a subset A of X by adjoining the element a to A. EXAMPLE If X 10. then it has infinitely many distinct subsets. There is exactly one such product for every subset {a. 1 P(X1)1 = 21P(X) 1. That is. . if x E X. Starting with the empty set @ (for which IP(X)I = 1). a. That is. Let X consist of the distinct elements al. every subset A of X gives rise to two distinct subsets of X1. if . doubling the cardinality of the resulting power set each time an element is added. since it gives additional information about the number of subsets of a finite set. namely x. Thus. Al # B. so that P(X) contains a t least as many elements as X. . . In particular. then IP(X)I = 2". E P ( X ) a n d X E P(X). and Al Z B1.) of {al... (a. . which is 1). a.. .. . and consider the formal product I this expression is multiplied out. b)) . Our reasoning shows that the power set of X will contain 2" elements. In fact. . That is. together with the new element a.

until we reach N. al1 products corresponding to sets containing the same number j of elements become t'. a2. the formulas for the binomial coefficients are correct in the cases j = O and j = n. plus the number of subsets containing one element of X. 2! = 1 . Then the product becomes (1 t)(l t) . .28 SET THEORY Kow replace each x k by the symbol t. al1 of the terms t j can be collected into a single expression of the form Nj. Thus. We get + The coefficient of t j is * - n(n . X = {ai. as)..4 = 24. this sum is just the total number of subset. while in its expansion. .s of X. and so on.2 . Clearly. By using the binomial theorem of algebra (see Section 2-2) to expand (1 t). Therefore + + + + + + + + + + + + + We can specialize even more by letting t have the value l . plus the number of subsets containing two elements of X. .tj. 2 . 3 . (1 t).. In the example.. . Then the identity (1-1) becomes The sum 011 the right-hand side of this identity represents the number of subsets containing no elements of X (the empty set).. l! = 1. it is possible to squeeze more information from identity (1-1). since every subset contains some number of elements of X between zero and n. As in this example. we have arrived a t the same conclusion as before : there are exactly 2" subsets of a set X with n elements. For example. we obtain (1 t)3 = 1 t t 4-t t2 t2 t2 t3 = 1 3t 3t2 t3.l ) . 3 = 6. the number of subsets containing n elements of X. I t i s a l s o customary to define O! to be 1.. With this convention. 4! = 1 . 3! = 1 . . is precisely the number of subsets of X which have cardinality j. where Nj. (n j! j + 1) - n! j!(n - j)! * An exclamation mark (!) following a natural number n denotes the number obtained by multiplying together al1 the numbers from 1 to n. 2 = 2..

. No. U X W XV.1-31 THE CONSTRUCTION OF SETS FROM GIVEX SETS 29 Comparing the identities (1-1) and (1-3) we see that for al1 numbers t. . = 1. then U X V is equivalent to V. a l = bl.j) ! different subsets containing exactly j elements. where U. V X U. For example. V. a.. and Y then IXnJ = J P ( Y ) I . 2). 2. Xn. Nz. 8. Ni. 3. 9.. . V X W X u . there are n!/j!(n . then a. X X . 4.. . 2 . . in a set containing ten elements. 6. This will justify (1-4).5 yields the interesting fact that in a set X containing n distinct elements. That is. w). Prove Theorem 1-3. B X A. For any set X. define n factors 7-7 xn = X X X X . N. Show that if X = (1. w x U X V . U X V. and W are arbitrary sets.1)/2. 1.3 for a finite collection of sets X1) X2) . Thus. = n. = b. 7. = 1. . Prove that the following sets are equivalent: U X V X W. where A = (x.y. 2)) and C = ( a ) . V X U X W . Let U = (1. B = (1. = (1. where n is a natural number. V . our somewhat longer proof of Theorem 1-3. . and in general Later we will be able to prove that if for al1 values of t. .3(b). . and A X (B X C). State the generalization of Theorem 1-3.. . Prove that if U and V are denumerable sets. W X V X U. Prove that U X N is equivalent to N. 2. = n(n . List the elements of the sets A X B. ( A X B) X C. where N is the set of al1 natural numbers. Prove that if U is a finite nonempty set and V is a denumerable set. . there are 10!/4!6! = 210 subsets of cardinality 4. ... This leads us to expect that the coefficients of the same powers of t on each side of the equation must be equal. 5 . . Prove that U X V X W is equivalent to ( U X V) X TV. 2). = bo. then the following sets are equivalent: U. n). ..

14. b ) c important laws of operation. such as a and a (b c) = a b a c. 11. where X = {1. that is. or in B.] 1-4 The algebra of sets. [Hint: Suppose that a o A is such a correspondence. then X and P(X) do not have the same cardinal number. As we pointed out in the Introductioii. Let B = (aja E X . 4). Show that if b ++ B. + + + + DEFIXITION 1-4. a - A a n d a G? A ) . then so do P ( A ) and P(B). that is. (a) Let t = -1 in equation (1-1) and interpret the meaning of the resulting identity. The ordinary number systems satisfy severa1 b = b a . Thus. . then both b E B and b 4 B. Our objective in this section is to study the principal operating rules for sets.1. (b) What is the number of subsets of even cardinality of a set containing n elements? 13.30 SET THEORY [CHAP. so that the statement "x E A or x E B" includes the case where z is in both A and B. The first two basic operations of set theory are analogous to addition and multiplication of numbers. They are binary operations. Let X be a set with 7 objects. Prove this fact. List the elements of P(X). Moderii algebra is largely eoncerned with systems which satisfy various laws of operation. the word "or" in mathematics is interpreted in the inclusive sense. c) = ( a . or in both A and B. Let A and B be sets. so it is natural that the algebra of sets should be a part of this subject. (a) How many subsets of X of cardinality a t most three are there? (b) How many subsets of X of cardinality a t least three are there? 12. 3. it is impossible to give a one-to-one correspondence between the elements of X and the elements of P(X). Theil A UB A nB = = (xlx E A o r x E B ) . those elements which are in both A and B. The intersectioii of 4 and B contaii~s . (XIXE Aandx E B ) . the union of A and B contains those elements which are in A. Show that if the sets A and B have the same cardinality. There are also natural operations of combining sets which satisfy rules analogous to these identities. The set A U B is called the union (or join or set s u m ) of A and B. The set A n B is called the intersection (or meet or set product) of A and B. Cantor proved that if X is an infinite set. they are performed on a pair of sets to obtain a new set. 1 10. a ( b . 2.

= = ( 1 . EXAMPLE (a10 4. EXAMPLE If *4 is the set of al1 points of a line through the point p and if 6. 3. The set A" is called the complement of A in X (or simply the complement of A if it is understood that A is being considered as a subset of the universal set X). 5. it is performed on a single set to obtain a new set. 3. 1-3. (a10 = @. EXAMPLE 3. The elements of the sets are represented by the points inside a closed curve in the plane. 3. B is the set of al1 points of a second (different) line through p. 7 ) 2. Let A be a subset of the set X. 2. 4 . 6 ) (2. 5. called the universal set. al1 of the sets under consideration will be subsets of some particular set X. This set. In Fig. 3. 3. 6 . 4. 7 ) U (2. EXAMPLE ( 1 . the total shaded area is A U B and the doubly shaded area is A n B. = 1) U {O. but can be any objects whatsoever. The third basic operation of set theory is analogous to forming the negative of a number. For the purposes of developing the algebra of sets. It should be emphasized that these diagrams are only symbolic. 5. 3. we will fix a universal set X once and for all. may be different for different problems.1-41 THE ALGEBRA OF SETS 31 These sets can be illustrated by means of simple pictures called Venn diagrams. 4 . that is. 4 . but it will usually be fixed throughout any discussion. 7 ) n <a< <a< 1) ( 2 . 7 ) . I t is a unary operation. DEFINITION 1-4.2. n (1. 1 ) < a 2 1). . Then Ac = (X~X E X . 6 ) (3). and that the elements of the sets which they represent are not necessarily points in the plane. Al1 of the sets under consideration are assumed to be subsets of X. AnB EXAMPLE (1. x g A). In most mathematical applications of set theory. then A í l B = ( p ) . 5. = n (al$ < a < 2) {al$ <a< 1). 6 ) 1. EXAMPLE (a10 5.

On the other hand. 3). 3. A n (B n C ) u A = A . .1 Thus. Then (1. f and consequeiitly x E (A U B) U C. either x E A or x E B U C. by 1-4. Then (1. A n A " = iP. A. then x E A U B. (ala 2 o> Then {ala < O)" = THEOREM 1-4. u @ = A . I x E B.AnB=BnA. if x E A U (B U C). Proof. then either x E B or x E C. Let A.1. C(A). then we conclude imf mediately that x E (A U B) U C. Suppose that x E A U (B U C). U A " = X. Suppose that x E A. 3) " = = (2. A n A = A. Then thefollowing identities are satisfied: (a) (b) (c) (d) A A A A A (e) A (f) A (g) A uB=BuA.An@=@. Some which the reader may encounter are A'. In Fig. 5). = = (1. Then according to Definition 1-4. A n X = A. (2). in every case. intersection. 2. (A = n B) n C . and c(A). EXAMPLE Let X 7. u(BnC)= (AuB)n(AuC). and complement. n (B u C ) = (A n B ) u (A n C ) . I x E C. u (B u C) = (A u B) u C. Then by Definition 1-4.3. Again. Thus. if x E B U C. x E (A U B) U C. The remaining identities are left for the reader to check. 4. a n d C be subsets of X. uX=X. 3)" (1. EXAMPLE Let X be the set of al1 real numbers. Al1 of these identities are simple consequences of the definition of union. 4. 1-4.32 SET THEORY [CHAP.1. the shaded area represents AC.1 again. 2. x E A U B. A" consists of those elements of X which are ilot elements of A. EXAMPLE Let X 8. We will illustrate this assertion by giving the detailed proofs of (b) and (d). There are many different notations in mathematical literature for the complement of a set A. 5). B. 9.

A U (B U C) = (A U B) U C. A similar argument shows that (A U B) U C 2 A u (B u C). We have shown that This inclusion relatioii. it is possible to derive from these numerous other laws of operation. . If x E A n B. This shows that A n (B n C) = (A n B) n C. WehaveshownthatA n (B u C ) E ( A n B ) u (A n C). x E A n B and x E C. The heavily outliiied region in Fig. Consequently. Therefore x E (&4 n B) n C. x E A. Similarly.3 are the basic rules of operation in the algebra of sets.rate the other identities of Theorem 1-4. this means that A U (B U C) E (A U B) U C. The reader should illust. if x E A n C. yields Let us illust'rate by a Venn diagram the identity (d) which we have just proved. combined with the one obtained above.1.2. . The identities (a) through (g) in Theorem 1-4. Similarly. Then either x E A n B or x E A n C. in any case. then x E A n (B U C). Therefore X E A n ( B u C ) . either x E A n B or x E A n C. x E (A n B) U (A n C ) . suppose that x E A n (B U C). If x E A n (B n C). By Definitioii 1-1. 1-5 represents either side of the identity. x E A n (B U C). suppose that x E (A n B) U (A n C). and x E C. This proves the first half of (b). Hence. That is. To prove the first equality of (d). then by Definition 1-4.1-41 THE ALGEBRA O F SETS 33 theii x E (A U B) U C. Then x E A and x E B U C. Consequently. Thus. x E A and x E B n C. Hence. Hence. On the other hand. By algebraic manipulations alone. x E B. (A n B) n C G A n (B n C). Hence.3 by Venn diagrams. then x E A and ~ E B so that X E A and ~ E B u C . either x E A and x E B or x E A and x E C. A n (B n C) G (A n B) n C.

A = . ( A (c) ( A C ) C A . and C be subsets of a universal set X. and hence x E A n B. if A = A n B. I A c B.4 are again simple applications of the definitions. 10. f (c) I A S B. Therefore A 2 B. THEOREM 1-4. then Ac 2 Bc. (a) A c A u B . n B)" = u B". then x E A implies z E B. B c A u B . (d) A E B if and only if A n B = A .4. XC = @.34 SET THEORY [CHAP. f A n B. Let A and B be subsets of the set X. I x E A n B. Indeed. A = A n B. 1 EXAMPLE Let A. then in particular. t h a t is. if A 2 C and B 2 C. B. and C be sets. . Therefore. However. Let A. Conversely. let us prove the first part of (d).3. A n B . then A n B 2 C. For exahple. in particular. x E A. = ( C n A) U ( C n B) = (b) A U (B n A) = (A U B) n (A U A) = ( A U B) n A = A n (B U A) ( . as y e did for the proof of Theorem 1-4. then B = C. (a) (A U B) n C = C n ( A U B) (A n C) U ( B n C). (b) If A 2 C and B C. = (d) ac = X. (b) (A U B)" = Ac n Bc. B A n B . THEOREM 1-4. so that f Thus A A n B c A . B = B U ( A ~ B ) B U ( A ~ C ) ( B U A ) ~B U C ) = ( A U B ) ~ = = ( ( B u C ) = (14 u C ) n ( B u C ) = ( A n B ) U C = ( A n C ) U C = c U n c ) = C. f (a) I A C B. then every element of A is in A n B and. A U ( B n A) = = A sncl A n ( B U A) = A- (e) If A n B = A n C and A U B = A U C. Identities such as those of Example 10 can of course always be obtained directly from the definitions of the set operations. The proofs of the various statements in Theorem 1-4. B. and similarly. A 2 B if and only if AuB=A.5. identities which involve severa1 sets can usually be derived more easily by algebraic manipulations. then A U B E C.~ ~ ( B u A ) ) u @ (= A ~ ( B u A ) ) u ( A ~ A ~A = ( ( B u A ) u A ' ) = ) n A n (B U (A U AC)) = A n ( B U X) = A n X = A. then A U C B U C and A n C 2 B n C. in B.

We illustrate the identity (A n B)" = A" U BC by a Venn diagram. . 8. x E AC and x E Bc. (e). show that B = Ac if and only if A U B = X and A n B = a. using the results of Theorem 1-4. Justify each step of the computations in Example 10. In Fig. A n B. That is. 9. 1.4 U B. the region outside of the doubly shaded region represents each of the sets (A n B)" and A" U Bc. so they are equal. Let us examine (b).5 should be clear. (f). which in turn amounts to x é A and x é B. (a) A U ( A c n B) = A U B. Thus. then Ac B.3 where they are needed. (c). B = {nln/2 E N) + + 2. (e). 4.3 and Example 10. Prove Theorem 1-4.3(a). show that if A U B = X. Also. A n ( A C uB) = A n B (b) A U (B n (A U C)) = A U (B n C) (c) ( n n B) U (B n c>) (C n A) ( U = ( ( A U B) n ( B u C ) ) n ( C U A) 5.3(d). Prove Theorem 1-4. 3. then B E A". Show that if A E C. B = {nln < 10) . Illustrate Theorem 1-4. and (d) of Theorem 1-4. (g) and Theorem 1-4. (e). Prove the following identities by algebraic manipulations. and (d).1). 6. Thus (A U B)" and A" n Bc contain exactly the same elements. To say that x E (A U B)" is the same as saying x é A U B.4(c) by a Venn diagram.1-41 T H E ALGEBRA O F SETS 35 The statements (a). B = {nln2 = 2n 3) (e) A = {nl(n 1)/2 E N).4(d). 1-6. and Ac in the following cases. 7. (b) A = {n1n2 > 2n .5(b). which means x E Ac n Bc. using the results of Theorem 1-4. and (g). Use the result of Exercise 8 to give a new proof of Theorem 1-4. show that if A n B = <P. Using Theorem 1-4. then A U (B n C) = (A U B) í7 C. The proof that (A n B)" = A" U Bc is similar. If the universal set is the collection N of al1 natural numbers. (a) 4 = (nln is even) . (e).5(a). determine .

Then u(S) = {xlx E A for some A E S). B). 2. {l. S need not be a finite collectioiz of sets (see Example 3 below). Define . (a) (. 1 10. 5:-. 2).1. General rules of operation.B = A n Be. 3. A @ @ = A (b) (A O B) O C = A O ( B O C) (c) A n ( B o C) = ( A n B) O ( A n C ) 1-5 Further algebra of sets.(A n B) U (A n ( B U Cc) ~ (b) ( A U (B n c ) ) ~= n c)) U (A n D) 11.( B U C ) (b) A . then A . In Fig. 6)). (b) (A/A)/(B/B) = (a) A/A = 14c (c) (A/B)/(A4/B) = A ri B UB The binary operation (*/*) is called the Schefler stroke operation. E As in the case of two sets.2: A for al1 A E S).4(b) into rules involving only the Scheff er stroke operation. It is possible to extend many of the identities in the previous section to theorems concerning operations on any number of sets. = { A .. then the diflerence between A and B is defined to be A . Rlake Venn diagrams to illustrate the following identities. Then U(S) = (1. Let S be a set whose elemeiits are sets. (a) A n (B U (C U D)) = (. .B = ( a E Ala B).(B .3(d) and Theorem 1-4. D). u(S) contains those elemeizts which are in any one or more of the sets in S. u(S) is called the u n i o n of the sets of S and n(S) is called the intersection of the sets in S. 13. B. 3.4 . (a) A O A = @ . then U(S) = A U B and n(S) = *4 í l B. C. Prove the following. 5.( A . Translate the identities of Theorem 1-4. For these definitions. DEFINITION 1-5.B) = A n B 12. If A and B are any sets. S = {A. Prove that the following are true. inside the heavy outline. 6) 1. and n(S) contains those elemeizts which are in every set in S. EXAMPLE Let S = ((1. n(S) = (~1. Define A O B = (A n Be) U ( A c n B).36 SET THEORY [CHAP. 5. and n(S) = @. if A and B are subsets of some universal set X. 14.4/B = Ac n Bc.A ) = A (c) A .B) . Thus. (2. u(S) is the total shaded area and n(S) is the most heavily shaded area.C = A . Show that the following are true. I n particular. EXAMPLE If S 2. 1-7.

it is customary to write uieICi for u(S) and n i ~ ~ fori n(S). and n(S) = X. 3. I the sets of the collection ( A l. I S is a collection of sets. For example. . the intersection or union of the empty set of sets is often eiicountered. GENERAL RULES O F OPERATION 37 EXAMPLE Let C be a circle in soine plane P. . (See Fig. An) are f united in any way. .. the set I is called aiz index set. Shen 4. U(S) = @. .3(b) are called the associative laws for the operations of set union and set intersection.2. .). C The identities of Theorem 1-4. ) let I = N = (1.1-51 FURTHER ALGEBRA O F SETS. f When this notation is used. . . I S = (CiJi E I ) . Moreover. . two a t a time.1.2.) Then U(S) is the set of al1 points in P. Example 4 may perhaps be surprising. 2. EXAMPLE Let S be the empty set of subsets of the universal set X. . i) can be denoted by Ci7and we have S = {Cili E N). and let S be the collection of sets Then the set {1. the resulting set is equal to .then we write S = (CJi E 1 . using each of the sets a t least once. 1-8. A s . but it is nevertheless correct according to the definition. which satisfy the following specifications: the elements of A are al1 points of P lying on the side containing C of some tangent line to C. These are special cases of a general associativity principle. and if its member sets can be labeled by the f elements of another set 1. The reader should carefully check these examples to be sure that they satisfy the condition of Definition 1-5. THEOREM 1-5. Let S consist of al1 sets A 3. while n(S) consists of al1 points inside C. It would be a nuisance if these operations were undefined in this case. .

two at a time. A3 U (Al U A2). to show that (Al U (A2 U A3)) U A4 = u ( { A ~ .2. is just what is obtained from unions of the type considered by omitting parentheses. The possible ways to unite the sets (A 1.2. 1 If the sets of this collection are intersected in any way. A3). and A4 in various ways. A3. A3 U (A2 U Al). we may adopt the notation since the expression A l U A2 U . using each set once are : Al U (A2 U A3). A2. A2 U (A3 U Al). A2 U (Al U A3). (A3 U Al) U A2. using each set a t least once.2 is out of reach until we have discussed inductive proofs. THEOREM 1-5. A2. for example. A4)). (Al U A3) U A2. Then (a) u(S U T) = u(S) U u(T). (b) n ( S U T) = n(S) n n(T). two a t a time using each set once. in 120 ways: (Al U (A2 U A3)) U A4. the meaning becomes clear if we look a t examples. As a consequence of Theorem 1-5. ((Al U Az) U A3) U A47 A1 u ((A2 u A31 u A. the resulting set is equal to The phrases "united in any way. A l U (A2 U (A3 U Aa)). Al U (A3 U A2). A3. I t is possible to give a proof now of a theorem which is closely related to Theorem 1-5. A mathematically correct proof of Theorem 1-5. and a proof of the theorem will be given there. (A2 U Al) U A3. (A1 U A2) U (A3 U A. this method is not a "proof" and will not satisfy the mathematician who demands a general method which will cover al1 possible cases at once. two at a time" and "intersected in any way. (Al U A2) U A3. The theorem says that the arrangement of parentheses is of no consequence anyway.38 SET THEORY [CHAP. A2. However. (A2 U A3) U Al.2. This is perhaps the best way for the reader to convince himself that the theorem is true. two a t a time" are somewhat vague. U A. Of course.). .). I t is a simple chore to prove any individual instance of Theorem 1-5. together with the cornbinations obtained by interchanging Al. (A3 U A2) U Al- Four sets can be united.3. These will be considered in Chapter 2. Let S and T be two sets whose elements are sets.

TO say that x E u ~ E means~that x E Bi for some i E I by Definition 1-5. {2. We have A n B. Further. Part (b) of Theorem 1-5.1-51 FURTHER ALGEBRA O F SETS. EXAMPLE I n Theorem 1-5. These laws can also be generalized. 4)).4).4). if f x E u ( S ) .This argument S ~ O W S x that the elements of A n (uiEIBi) exactly the same as the elements are of uiEI(A n Bi). U(S) = (1. . But if A E S U T . Then {{l. I A E S. and U ~ E I ( ABi) = {2nln E N) = íl A. then A x E A n Bi for some i E I. Then A ri ( u ~ E I B ~ U E E I( A n Bi). 2n 5 i).1. if x E u ~ € I ( n Bi). i) for i E 1. then certainly A E S U T.3 amounts to a careful examination of Definitions 1-4. ncs u T ) = (21. .3. . U(S)U U(T) = {1. = uT= U(S u T ) = S The proof of Theorem 1-5.4(b). The identities (d) of Theorem 1-4. On the other hand. then either A E S or A E T. 2. Thus u(S) u u ( T ) u(S U T ) .3 is proved similarly. For this particular i. Define Bi 6. SO that A U ( n i ~ r B i= A U (1) = (1. I the opposite inclusions are combined.4). 3. . let I = N = ( Theorem 1-4.3117 {1. . ). In either case. '44. (2.).4. Similarly. x E u(S U T ) . 7 n ( S ) = {2).and let A = (2nIn E N) = (2.3). Proof. 2 3. x E A n Bi. . 3. . On the other hand. U(T). 31. 6.2. 2. .1. 2. x E ui€1(A n Bi). Therefore. {2. = ) A U (ni~~Bi) = ( A U Bi). and T = ((2. and A n (Ui~rBi) A n N = = A.2. Then = UiEIBi= N. x E u(S) U . . 31. 3. 6) i ) . then x E A for sorne A E S.4. ~ B E A and x E UiEIBi. I A E S. = (2nln E N. f then x E ~ ( 8 ) I A E T .n(T) = (21. it follows that u(S U T ) = f U(S) U u ( T ) . By Definition 1-5.4). then x E u ( T ) .). Suppose x E u(S U T ) . Thus.4)). The other statement is proved in a similar way. it follows that x E A and x E uiEIBi. and let A be any set. 4. n ( S ) ri n ( T ) = (2). n i ~ r B = (11. Let (Bili E 1)be a set of sets. Suppose that x E A n ( u i ~ ~ B i Then by Definition 1-4. Then x E A for some f A E S U T. {1. (2.1 and 1-5. This shows that u(S U T ) c u ( S ) U u ( T ) .2.4). .1. GENERAL RULES O F OPERATION 39 I~XAMPLE Let S 5. THEOREM 1-5. ((1. ü ( T ) c u(S u T ) . Therefore u ( S ) G u(S U T ) . Consequently.3 are the distributive laws for the set operations of union and intersection. U(T) = (2. x E A n (Ui€IBi).

. THEOREM 1-5. multiplication. . 4.ic of sets is that only a finite number of different sets can be constructed from a finite number of sets using the operations of union. . A little calculation will show that the only possible sets which can be constructed from two sets A and B in X are a. i + 1. Let A l . i. as well as A and A" again. An examination of the Venn diagram in Fig. Then every set which can be formed from these sets by union. A n Bc. each set in our list is the union of one. (A n B) U (Ac n Bc). or al1 of these fundamental sets. A 2 . we obtain the sets A". (A n Bc) U (Ac n B). i + 2. 6.) if i is odd. 2. The next step produces no new sets. A" n B. . . One of the surprising facts about the arithmet. and to build from it infinitely many other numbers by addition. For example. A.) if i is even (1.SET THEORY Finally. i + 4. or complementation is either or has a representation as a union of .AuBc. . . . B. .). At the second step. . This is an example of a general theorem which is usually called the disjunctiue normal form t heorem. i. . We see that except for a. A U A" = X. subtraction. A U A = A. 2. and complementation. . A n A = A. . two. we get A n A" = +. three. .AnB. Thus the first step of the construction yields one new set A". . In the ordinary arithmetic of numbers.5. A uB = (A n B) u ( A n Bc) u (Ac n B). 1-9 indicates why these sets are important. starting with a set A (contained in the universal set X).AcnB. A". X. and division. it is possible to start with a single nonzero number. be subsets of the set X.A~Bc. .An13c. . intersection.AcnBc. say 2. For example. (1. intersection. n i ~ l ( A Bi) U = Therefore. Bc. .AcuB. i + 3. . . and A" n Bc are particularly interesting. A. the four sets A n B.AuB. . nor does any step thereafter. In this list.

2 for the following particular combinations of sets: (Al U A2) U ( A 3 U A4). Suppose that S = {A).. (a) (uiEIAi)" = niéIA( (b) ( n i ~ r A i ) " u i € ~ A : = 5.3(b). The foundation of statistics is the theory of probability. or al1 of these Mijk. For instance. n A.. n A. A l U (A2 U A s ) ) U A4.. n A 2 n A3. and A 3 can be obtained as a union of one. B. two.5. five. i are O or 1 and A5i is A j if ij = O and Aj" if ij = 1. like the proof of Theorem 1-5. can be carried out only by mathematical induction. MOOO A l = MOOl = Moll = M. A2.2. every possible nonempty combination of A l . Al U (A2 U (A3 U ~ 4 ) ) . Show that the following are true. Let (Ai[i E 1) be a set of subsets of X.1-61 MEASURES ON SETS certain of the sets where il. A2 n A. if n = 3.. and complementation? I f C A C B C C C X. M010 = A l n A. how many different sets can be constructed? *1-6 Measures on sets. M100 = A ? n A2 n A. 4. ( 3.iz. By the theorem. four.. = M. One important application of set theory is its use in mathematical statistics. . intersection.... What is U(S)? What is n(S) ? 2. What is the largest number of different sets which can be constructed from three subsets A. A.. Since this result will not be needed in later parts of this book. n A. . . and in its mathematical form. seven. The proof of Theorem 1-5. 1. probability is the study of certain kinds of measures on sets.. C of a universal set X. For example. . Check Theorem 1-5. where A is a set. a formal proof will not be given.. In this section and the next the concept of measure of a set will be introduced. n A. n A. six. and some of its simplest properties will be examined. three. using the operations of union. = Al Al A". Mi10 = A" A. Prove Theorem 1-5. n A 2 n A. . . n A.

." then it is reasonable to supf pose that for any roll of the dice. 3.SET THEORY Severa1 ways of "measuring" sets are already known to the reader. the probability of making the point 2 is + . We will now consider a measure for finite sets which links our discussion to the application of set theory to probability. The probability of making a certain point is the ratio of the number of different ways that the point can be made. the 36 different pairs (m. . the measure of a line segment (which may be considered as a set of poiiits) is usually taken to be the length of the segment. Another useful measure for these line segments might be the annual cost of upkeep of that section of the rail line which is represented by them. . that is. the number of elements in A. . Here the length of each line segment is of little interest. Both A and B will come to rest with a number of dots from 1 to 6 on the "up" face. We now assign a measure to the subsets of (2. I S is such a subset. n). . For example. A good measure of a finite set A is 1 Al. a railroad map usually indicates the route between major cities by a sequence of line segments connecting intermediate points as shown in Fig. For example. Now it is customary to define the "point" which is made on any roll of the dice to be the total number of spots on the two "up" faces. The result of the roll can therefore be represented by an ordered pair (m. Suppose that a pair of dice. n) are equally likely to occur. where m gives the number of dots on the "up" face of A and n gives the number of dots on the "up" face of B. the set of possible points is (2. then the point made on the roll is m n. then the cost of upkeep of the part of the rail line represented by I l U I 2 would be the cost for I l plus the cost for 1 2 . m and n can be any natural numbers from 1 to 6. the number of possible results of a roll. 12). The important measure of these segments is the actual rail line distance between the cities corresponding to the points which the segments connect. I the dice are "honest. 3. . For example. Therefore. Thus. But there are situations where different measures of line segments and finite sets are more useful. labeled A and B. to 36. 12). n) of natural numbers. the possible points which can be made on a roll of the dice are the numbers from 2 to 12. if the outcome of the roll is represented by (m. Note that this measure has a natural extension to those subsets of the map which are unions of two or more segments. if I I represents the part of the rail line between Milwaukee and Chicago and if I 2 represents the part between Detroit and Buffalo. 1-10. . Thus. are rolled. For example. . assign as the measure of S the f probability that on a roll of the dice the point made will be a member of S.

the measure of the set (7. Thus. Let us now look for some common properties of the measures described in the above examples and try to arrive a t a suitable mathematical notion of measure. 11) is & = 4j. Such sets are called half-open intervals. take S = (7. then A n Bc E S. was actually defined for unions of segments of the map... by the roll (1. A nonempty collection S of subsets of X is called a ring of subsets of X (or just a ring of sets) if it satisfies the following two conditions.) (4. there is a rule for assigning a certaiii number to various subsets of a giveii set. In both cases. Thus our meassince DEFINITION 1-6. sets of the type I = {xja < x 5 b . unless these collections satisfy certain "closure " conditions. The second of these measures. b. measures need not be defined 011 al1 subsets of a given set. each set of S has the form I1U 1 2 U U In. In each case. b. . Then the collection of al1 finite subsets 2. f EXAMPLE If X is any set. It is clear now that this "probability measure" can be determined for each of the Z1 = 2048 different subsets of possible points.). but only on some collection of subsets.where Ii = {xlai < x 2 bi). 1 = {xla. EXAMPLE Let S be the set of al1 subsets of R which are finite unions of 3. (b) I A E S and B E S. < . a. There are tmo ways of making 11: ( 5 . 2 ) ) (6. We have seen that there are six ways of making 7. . for some real numbers al. .1. Let X be a set. (2. 1 2 = (xla2 < x 5 b2). That is. a2. another example. In general. The outcome of the roll will be in (7) only if the point made is 7. . of subsets of X. The point As made on a roll will be in this set if it is a 7 or 11. . then A U B E S. the measures are defined only for very special subsets of the whole map. EXAMPLE Let X be an infinite set. Suppose now that the subset S is the set (7).1-61 MEASURES ON SETS 43 ure would assign to the subset (2) the number h. Since 7 can be made in six possible ways: (1. and the measure of (7) is i. the cost of upkeep. . (a) If A E S and B E S. (3. 11). 4. 1). 6). . two different measures were suggested for line segments making up the map. b~ R). . bi. then the collection of al1 subsets of X is a ring 1. b2. Then S is a ring of sets. of X is a ring of subsets of X. . the measures on them will not be very useful. 3). X is not in this ring. a E R.5). One property is immediately evident. 2 can be made in only one way. 6. . (5. In the example of a railroad map. Moreover. < x . However. . 6) and (6. . the probability of making 7 is = 3. however.l). 5).

There is one more important property that our examples have in common. the measure of (7) is +. A n A" = is in S. A n (A n BC)"= A n (Ac u B) = (A n Ac) u (A n B) = u (A n B) = A n B. with A n Bc taking the place of B. A. . 1 The expression "ring of sets" is standard mathematical terminology. . S ~ u A ~ u . Note that the term "pairwise disjoint" refers to the collection of sets as a whole and not to the individual sets in the collectioii.1 (b). We obtain A n (A n Bc) E S.and the measure of (7) U = $. we find that Al n A. .1. if Al. This additivity property is shared by the probability measure example. A using repeatedly 1-6. . Here. . . .2(b). Suppose that A E S and B E S. Then E S. 12).44 SET THEORY [CHAP. One of the requirements in Definition 1-6.3. I t is derived from abstract algebra. E S . then the measure of A U B is the sum of the measures of A and B. A n BCE S. . Now use Definition 1-6. Thus. Let S be a ring of subsets of X. .5 and 1-4. then A n B E S . then using Definition 1-6. This is still clearly true if Il and I2are replaced by unions of segments. Al U A2 U A3 = ( A ~ u A ~ ) U A .l(b) again.3. A2.E . Thus.2. provided that these unions have no segment in common. . A . A n B E S.~ . . there is some subset A of X which belongs to S.n A. For example.ES. belong to S. then the measure of í 1U I2is the measure of Il plus the measure of 12. E S . I A and B are subsets such that no number of (2. + . A collection of sets is called pairwise disjoint if each pair of different sets in the collection is disjoint. (a) If A E S and B E S.1 (b). Then by Definition 1-6. E S and Proof. . A n B = a. 11) is 8 the mathematical notion of measure. .. This simple property is the essence of (11) = (7. U A. 12) is in both A and B (A n B = a).the measure of (1 1) is &. The "closure" conditions to which we alluded above are the properties (a) and (b) in Definition 1-6. . THEOREM 1-6. by Theorems 1-4. we noted that if Il and I2are distinct segments. E S . Finally.l(a) repeatedly gives Al U A2 E S.1 is that S be nonempty .. A2. . There are other important closure conditions which are satisfied by rings of sets.u A . (b) f (c) I Al. 3. which we have just proved. that is. by . I n the upkeep cost measure on the segments of the railroad map. Consequently. then Al U A2 U Al n A 2 n . the measure was defined for al1 subf sets of the set of points (2. nA . . Two sets A and B are said to be disjoint if they have no elements iii common. . . However. n . by Definition 1-6. . Similarly.

the cardinal number of A . + + + It is left to the reader to show that the condition of Definition 1-6.3 is satisfied. Al = (x(1< x 21. For this discussion the following example is important. provided we agree that each line segment includes its lefthand endpoint. .2 3 ) ) = m2 m3.x2)) = mi ma.3. . 12) are disjoint. A2 = {xl2 < x 5 31. < DEFINITION 1-6. For a nonempty subset *4 of X . m((x2)) = m2. 11) and (2. m(+) = O. x2 = 2 . define m ( A ) to be the sum of al1 those mi for which xi E A. A3.~ 3 ) = mi -i. so that a measure is defined on the collection P ( X ) of al1 subsets of X . EXAMPLE Let A l . . etc. . m. x2. X n = n. A2. 2 3 ) ) = mi 4. m. m ( { x i ) ) = mi. then We will be concerned principally with measures defined on the set of al1 subsets of a finite set. Let X be a set. Then m ( A ) is just the number of even numbers in A minus the number of odd numbers in A. m3 = -1. ) m({x2.1-61 MEASURES ON SETS 45 EXAMPLE The sets (7. EXAMPLE The collection of line segments in the railroad map example are pairwise disjoint. . For instance. = m. .m3. . . A3 = (213 < x 5 41. 5. then m ( A ) = Al. . x2. . . m({xa)>= ma. be the sets of real numbers x defined by 6. then m ( A ) = 1 if xi E A and m ( A ) = O if xi 4 A. . m((x1. . (1) If m1 = m2 = .. x. be a sequence of n real numbers. . Let m1 = -1. but not its right-hand endpoint. . If A = let m ( A ) = O. 1 I t is not surprising that there are so many interesting special cases of Example 7. Thus we can say that m measures whether or not xi is in A. A measure on the collection S is a rule which assigns to each set A in the collection S some real number m ( A ) . Particular cases are worth noting. . A2. (3) Let xi = 1.. Let mi. = 1. is pairwise disjoint. .m2 m. = m. = (-1)". A3. = O. (2) If mi = 1. . . subject to the conditioii that if A and B are disjoint sets in S . m2 . m ( { x i . . +. Then the collection A l . and let S be a ring of subsets of X. m ( { x i . m2 = 1. if n = 3. 7. EXAMPLE Let X be a set containing n distinct elements xl. 4. m2 = . since actually every measure on the collection P(X) of al1 .

) = di U di = di. A2. Since m is a measure. m(A2 U A3) = m(&) Thus.3. This will become clear after we observe t.) = (A1 n A. AS.) u (A1 n A. + I n = 2.3). A formal proof of this theorem will not be given here. that A2 n A3 = Qj. The reader should begin to be aware that mathematics leans heavily on this important method of proof which will be discussed in the next chapter. which is justified by the assumption that {Al. We used here the fact that Al n A2 = @ and A l n A3 = @. Consider the case n = 3. we can apply Definition 1-6. then m(Al U A2 U U A.4. namely m(A1 U A2) = m(Al) m(A2) if A 1 n A = di. The assertion is Since the collection {A1. By repeated application of the argument used in the case n = 3. this theorem is the same as the additivity condition for a f measure required in Definition 1-6. by Definition 1-6. we are ready to examine the assertion that every measure defined on P ( X ) for a finite set X is of the type given in . we know in particular.hat the additive property of measures has a simple generalization. Let m be a measure defined on a ring S of subsets of a set X.4 is true for any n. . A3) is pairwise disjoint. A2. we have + m(&).) is a collection of sets in S and this collection is pairwise disj oint. A3) is pairwise disjoint.3.3 again to the left side of the last equality to obtain the desired result: By the distributive law for the set operations (Theorem 1-4. Now if Al and A2 U A3 are disjoint. .) = m(&) + m(&) + + m(An). If (A 1. A. so that A l and A2 U A3 are indeed disjoint. we obtain Al n (A2 u A.4. Accepting Theorem 1-6.46 SET THEORY [CHAP. 1 subsets of a finite set S is of this form. it is possible to see that Theorem 1-6. . THEOREM 1-6. because such a proof is based on the principle of mathematical induction. .

12). 3) wliich are rings of sets. . suppose that X = (21. 2. m2. Indeed. x2. E A. For example. with xi E A. . Example 7 is the most general possibility for a measure on the set of al1 subsets of a finite set. Show that the number of sets in a ring of subsets of a finite set is always a power of 2. x. x2. $3. m4 = m((x4)) are certain real numbers. x3) = ( ~ 1 )U {XZ)U {x3). x3. . m. . 11. . is a measure. m ( A ) is the sum of al1 m. 41. . Show that I ++ U . m2 = m({x2)>. Hence. then {xi) n (xj) = <P.4 in the case n = 4. Give the details of the proof that if m is a measure defined on a finite set (21. However. the collection of al1 distinct one element sets {xi). [Hint: Let S be such a ring. 11. (10. 5. x2. find the measure of the following sets: (2. . Moreover. except possibly for A = <P. x3. Find al1 collections S of subsets of (1.] 4.3. . Thus. 8. 6 . Prove Theorem 1-6. m3 = ( b ) ) . we have shown that there are numbers mi corresponding to the distinct elements xi E X such that for any nonempty subset A of X.). 10. x2. by Theorem 1-6. that is. we can write A = Uzi~~{xi). 1.3. Let Al. 2. . 3. for which x. . 3. For simplicity. Suppose that m is a measure defined on P(X). . n) and S. {xl. x4 are distinct. . . where x l . I t is evident that if A is any nonempty subset of X. A. in the iiext section we will show that m(@) = O for every measure. A2. then there are real numbers mi. if i # j. . (There are 15 such collections. . if A = {xl. For example. the measure m(A) is precisely the sum of those mi for which xi is in A. 6. defined in Example 7. 4. 12)) (2. Then m1 = m((xl)). m satisfies the condition of Definition 1-6. 2 2 .1-61 MEASURES ON SETS 47 Example 7. such that for a nonempty subset A. I n the dice rolling example. starting with a measure m on P(X) for which nothing is assumed except that it satisfies the conditions of Definition 1-6. that is.). . x4). then This argument shows that any measure m on P(X) is a measure of the type described in Example 7. . E T A ~defines a one-to-one correspondence between the subsets I of (1. be al1 those nonempty sets in S which do not contain a smaller nonempty set of S. is pairwise disjoint. But this is just the measure of Example 7.) 3. 2. Show that m.4. m(A) is the sum of al1 mi = m((xi)) for which xi E A.

1. "1-7 Properties and examples of measures. Ac n B.2. 2 tails and a head = 5 points. ( a ) Since @ n @ = @.m(A). m(Ac) = m(A U A") = m(X). . by subtraction. (a) m(@) = 0. 10. For a subset A of the set {O. depending on the outcome of the toss.3(e). 15. as was done for the dice rolling example in the text. Define a probability measure m on the collection of subsets of the possible points (20. m(Ac) = m(X) . Let A. f (b) I A E S and A" E S. 1-11 shows that A uB (A n Bc) u (Ac n B ) A = (A n Bc) u (A n B). . What is the probability that a t least two heads will appear in the outcome of a toss? 8. 10 points. (b) By Theorem 1-4. queens 2 points. define m(A) to be the probability that on a given deal the number of points scored is an element of A. 51. Let m be a measure on a ring S of subsets of a set X. 10. = u ( A n B). . I n a given deal. two cards are dealt from a standard deck of 52 cards. In this section we derive some useful properties of measures. and m((5)). and m((l1). B be in S. Subtracting the nurnber m(@) from both sides of this equality gives O = m(@). + + THEOREM 1-7. jacks 1 point. kings 3 points. lo)). 3 tails = 15 points. Proof. B = (Ac n B ) n (A n B). A n B) is pairwise . and that the collection of subsets (A n Bc. 1. and al1 other cards O points.811. I n a certain card game. Aces count 4 points. as follows: 3 heads 2 heads and a tail = = 20 points. An examination of the Venn diagram in Fig. THEOREM 1-7. Thus m(@) = m(@U &) = m(@) m(@). Let m be a measure on a ring S of subsets of a set X. 2.48 SET THEORY [CHAP. 7. so that m(A) Again. Thus. Find m((20. three pennies are tossed a t the same time and points are scored.m(A). 5)). A and A" are disjoint. m((20. A n A" = and A U A" = X. n2({%9. . Find m((O)>. m((5. I n a certain game. then m(Ac) = m(X) . 15. the possible points range from O points (neither card is an ace or a face card) to 8 points (a pair of aces). 6. 1 7. the empty set is disjoint from itself (and it is the only set having this property). Theii Proof. 8) of possible points.

m ( A U B ) = m ( A n Bc) m ( A ) = m ( A n Bc) m(B) = m(Ac n B ) + + ( A V B) + m ( A n B ) . the subset of right-handed brunettes. then 1Cl/n gives the fraction of the class which is in C and 1001Cl/n gives the percentage of the class which is in C. Recalling that cardinality is a measure on the set of al1 subsets of a finite set.2 gives Now if there are n students in the class. brunettes.4. + m ( A n B). the rest are 1. 40% of the students are blonds. are right-handed brunettes. that is. since 47% of the class is in A U B.PROPERTIES AND EXAMPLES OF MEASURES disjoint .m(B) = -m(A n B). and 5% are both blond and left-handed. Moreover ( A U B)" is the subset of students who are neither blond nor left handed. Consequently. Subtracting the second and third equalities from the first one gives m ( A U B) - m ( A ) . 12% are left-handed. which when rearranged is the desired identity. by Theorem 1-6. Thus. . and C is any subset of students. Theorem 1-7. Then A 1 B is the subset of left-handed blonds and A U B is the subset of students 7 who are either blond or left handed. Let A be the subset of blonds in the class and B be the subset of left-handed students. + m(A n B). EXAMPLE In a certain class. We conclude this chapter by giving some practica1 examples of measures on finite sets. Find the percentage of atudents who are right-handed brunettes. 53% of the class is in (A U B)" that is. we have Therefore.

2. For a subset A of X = (xl. A3 = (x1. and x3. Because of our rough knowledge of the weights of $1.m((A n B) U ( A n C)) = - m(A) m(B) [ m ( An B) + + m(C) + m(B C) + m(C n A ) ]+ m(A n B n C). x3). 1 EXAMPLE A certain type of spring balance is constructed so that i t measures 2. let m(A) be the total weight of the steaks in the set A. how can we use the spring balance to determine their weights exactly? Let the steaks be denoted by xl. Clearly m is a measure on the subsets of X. I t is possible to extend the result of Theorem 1-7. by Theorem 1-7. we are certain that m(*41). B. and m(A3) can be accurately determined by the spring balance. A3 n A1 = (52). Then using Theorem 1-7. suppose that A. A2 n A3 = {xl). For example. x3. x2. If we have three steaks each of which is known to weigh between and 1 pound. A2 = (xi. m(Aa). = m(A) + m(B) + m(C) - m(B ri C) .50 SET THEORY [CHAP. and A l fl A2 = (x3).2 repeatedly. $31. 22. Now Al U A2 = A2 U A3 = A3 U Al = X.2 to an identity which involves more than two sets. + Thus. f l . since their weights are between 1 and 2 pounds. and C are subsets of X and that m is a measure defined on a ring of subsets of X. Adding these equalities gives Hence. x2. m(X) = + ( m ( ~ l ) m(A2) + + m(&)). Let Al = (x2. 52). x3). only weights between one and two pounds. Therefore.

2 have antigens A and B. ~h positive 1 Blood contains no antigens Rh A A and Rh B B and Rh A and B al1 antigens Suppose that in a group of ten people: 4 have antigen A. These antigens are denoted by A. Rh negative O. TB) TRh denote the sets of people having the respective antigens A. B. (A n Bc) u ( A n B). Rh positive B. Rh positive AB. Rh negative is The number of people with type O. it is possible to determine the number of people with each of the eight possible blood types.1-71 PROPERTIES AND EXAMPLES OF MEASURES 51 EXAMPLE The classification of blood type is made on the presence or 3. Rh negative 13. B. Rh negative AB. absence of three distinct antigens in the blood. . 1. (A" n B) U ( A n B). Determine the number of people in the group having type O. and 2 have al1 antigens. Rh positive is therefore 3 . By similar considerations. 5 have antigen B. Rh positive A. Rh negative A. and Rh. Let TA. and Rh. The number of people with type O (Rh positive or negative) is The number of people with type O. Rh positive blood.1 = 2. 3 have antigens B and Rh. 6 have antigen Rh. 3 have antigens A and Rh. The possible blood types are eight in number: Type O. Use the identities of Section 1-3 to show that A UB A B = = = (A n Bc) U ( A c n B) U ( A n B).

show how the spring balance can be used to determine their weights exactly. 4. 3 are written in random order. that is. m(C). or 3 occurs third? 7. 1 2. m(B). What is the probability that a t least one of the numbers will occupy its proper place.) Determine the approximate ratio of smokers with lung cancer t o smokers without lung cancer. A. (These are fictitious estimates. I n a certain sample of the population. or 2 occurs second.000 people. Determine m(A U B U C U D) in terms of m(A). 5. Suppose that a certain spring balance rneasures only weights between 1+ and 3 pounds. 6. AB. Rh positive. m(A n B n C). and AB. m(A n B ) . It is estimated that 80% of those with lung cancer smoke and that 65% of those without lung cancer smoke. Three numbers 1. find the number of people with blood types A. m(B í l C n D). m(C n D). and m(A n B n C r i D). 3. m(B n C). m(B n o ) . Assume that each possible ordering is equally likely. R h positive. If four steaks are known to weigh between 4 and 1 pound. nz(A n C n D). m(A n C). m(A n B n D). R h negative. Rh negative. Show that the empty set @ is the only set disjoint from itself. m(D).52 SET THEORY [CHAP. . 2. I n Example 3. 1 occurs first. m ( A n o ) . i t is found that lung cancer occurs in 15 cases per 100.

Sign up to vote on this title
UsefulNot useful