PROOF, LOGIC, AND CONJECTURE: THE MATHEMATICIAN’S TOOLBOX Robert S. Wolf© 2008 by Robert S. Wolf. All rights reserved. Reproduced by permission of the author and copyright holder, Robert S. Wolf, exclusively for use during the 2009-2010 academic year by students of Daniel Goroff. No part of this book may be reproduced by any mechanical, photographic, or electronic process, or in the form of a phonographic recording, nor may it be stored in a retrieval system, transmitted, or otherwise copied for public or private use, without written permission from the author. Printed in the United States of America Robert S. Wolf, Ph.D. 279 Hermosa Way San Luis Obispo CA 93405, U.S.A.Contents Asterisks (*) denote optional sections. Preface Chapter 1 Al 12 Chapter 2 2 22 23 Chapter 3 Ba 32 33 34 Chapter 4 41 42 Note to the Student ‘Thanks Unit 1 Logic and Proofs Introduction Knowledge and Proof Proofs in Mathematics Propositional Logic ‘The Basics of Propositional Logic Conditionals and Biconditionals Propositional Consequence; Introduction to Proofs Predicate Logie ‘The Language and Grammar of Mathematics Quantifiers ‘Working with Quantifiers ‘The Equality Relation; Uniqueness Mathematical Proofs Different Types of Proofs ‘The Use of Propositional Logic in Proofs xii xiv 16 21 35 45 48 37 66 2 8 16 45 243 44 4s 46 Chapter 5 Sa 32 33 Chapter 6 61 62 *63 Chapter 7 1 12 73 14 15 16 “17 Chapter 8 81 ‘The Use of Quantifiers in Proofs ‘The Use of Equations in Proofs Mathematical Induction Hints for Finding Proofs Unit 2 Sets, Relations and Functions Sets ‘Naive Set Theory and Russell's Paradox Basic Set Operations More Advanced Set Operations Relations Ordered Pairs, Cartesian Products, and Relations Equivalence Relations Ordering Relations Functions Functions and Function Notation One-to-One and "Onto" Functions; Inverse ‘Functions and Compositions Proofs Involving Functions ‘Sequences and Inductive Definitions Cardinality Counting and Combinatorics ‘The Axiom of Choice and the Continuum Hypothesis Unit 3 Number Systems ‘The Integers and the Rationals ‘The Ring Z and the Field Q 92 107 114 126 133, 142 153 164 174 183 193 204 215 223 230 245 252 261 Contents 133 164 193 261Contents 82 783 "8.4 Chapter 9 94 92 93 "9.4 *95 Chapter 10 10.1 "102 Introduction to Number Theory More Examples of Rings and Fields Isomorphisms ‘The Real Number System ‘The Completeness Axiom Limits of Sequences and Sums of Series Limits of Functions and Continuity ‘Topology of the Real Line ‘The Construction of the Real Numbers The Complex Number System Complex Numbers Additional Algebraic Properties of C 269 282 291 297 304 314 328 339 352 363 Appendix 1A General-Purpose Axiom System for Mathematics Appendix 2 Elementary Results About Fields and Ordered Fields Appendix 3 Some of the More Useful Tautologies Solutions and Hints to Selected Exercises References List of Symbols and Notation Index 297 352 372 377 387 389 409 412 415Preface Almost all mathematicians will attest to the difficulty of making the transition from the lower division calculus sequence to upper division mathematics courses like abstract algebra and real analysis. One primary reason is that in a typical calculus course, where ‘most of the students are not mathematics majors, the emphasis is on applications rather than theory. As a result, students barely encounter deductive methods and proofs in these courses. Moving from problem solving to the proofs in higher mathemati is so difficult that many students, even some quite talented ones, quit mathematics. Until the 1970s, very few colleges or universities had a course designed to soften ‘his transition. There seemed to be a sink-or-swim attitude, a belief that the students who really should be math majors would be able to handle the transition and learn how to read and write proofs while they leamed the material in more advanced courses. This system may work well at some lite universities, but it has obvious drawbacks at colleges and universities that want to make higher mathematics accessible to more than a narrow audience, possibly even including students who are not mathematics majors. The “transition course” or “bridge course,” now fairly common, is designed to bridge the gap. Thelieve the jump from calculus to higher mathematics is as hard as it is because ‘two things occur simultaneously. First, the material and the concepts being taught become more and more difficult and abstract. Second, since students are expected to read and write proofs in upper-division courses, these courses are methodologically ‘much harder than calculus. Therefore, I believe that the most important role of the bridge course is methodological. Simply put, it should be more of a “how” course than a “what” course. This is perhaps what most sets this course apart from other ‘mathematics courses. About This Book The Approach In content, this book is similar to most of the other textbooks designed for this course; it differs in emphasis and method. Chapter 1 familiarizes the reader with the three main processes of mathematical activity: discovery, conjecture, and proof. While the main goal ofthe course is to lear to read and write proofs, this book views the understanding of the role of discovery and conjecture in mathematics as an important secondary goal and illustrates these processes with examples and exercises throughout. Chapter 1 also includes brief discussions of the way proofs are done in science and in law for the5 Preface purpose of contrasting these methods with the special meaning the word “proof” has in mathematics. Chapters 2 and 3 cover the basics of mathematical logic. These chapters emphasize the vital role that logic plays in proofs, and they include numerous proof previews that demonstrate the use of particular logical principles in proofs. These chapters also stress the need to pay attention to mathematical language and grammar. Many of the examples and exercises in these chapters involve analyzing the logical structure of complex English statements (with mathematical or nonmathematical content) and translating them into symbolic language (and vice versa). Unlike many texts that have just one short section on quantifiers, Chapter 3 provides a full explanation of how to understand and work with quantifiers; it includes many examples of altemations of quantifiers and negations of quantified statements. Without studying this material, students can get the impression that constructing truth tables is the main logic-based skill that is important for reading and writing proofs. Clearly, this impression can lead to frustration and failure down the road. Chapter 4, the last chapter of Unit 1, is thorough discussion of proofs in ‘mathematics. It carefully explains and illustrates all the standard methods of proof that have a basis in logic, plus mathematical induction. In addition, there are discussions of the meaning of style in proofs, including the importance of learning how to find a good balance between formality and informality; the connection between solving equations and doing proofs; and hints for finding proofs, including useful strategies such as examining examples and special cases before tackling the general case of a proof. ‘The remainder of the book is not directly about proofs, Rather, it covers the most basic subject matter of higher mathematics while providing practice at reading and ‘writing proofs. Unit 2 covers the essentials of sets, relations, and functions, including ‘many important special topics such as equivalence relations, sequences and inductive definitions, cardinality, and elementary combinatorics, Unit 3 discusses the standard number systems of mathematics—the integers, the rationals, the reals and the complex numbers. This unit also includes introductions to abstract algebra (primarily in terms of rings and fields rather than groups) and real ‘analysis, The material and the treatment inthis unit are intentionally more sophisticated than the earlier parts of the book. In fact, nearly half of the sections of this unit are designated “optional.” In a one-semester course, itis unlikely that most of this material can be covered; naturally, the intention isto give instructors the opportunity to pick and choose. On the other hand, instructors with the luxury of a one-year course will find that most or all of Unit 3 can be covered, as their students gain more and more confidence with abstract mathematics and proofs. Unique Features Iwould single out user-friendliness and flexibility as the main features that distinguish this book from the other available bridge course books. User-friendliness could also be called readability. One hears continually that reading is a lost art, that students (as well as the general population) don’t read any more. I believe people will read books they find enjoyable to read. Every effort has been made to make this book engaging, witty,and thought-provoking. The tone is conversational without being imprecise. New concepts are explained thoroughly from scratch, and complex ideas are often explained in more than one way, with plenty of helpful remarks and pointers, There are abundant examples and exercises, not only mathematical ones but also ones from the real world that show the roles logic and deductive reasoning play in everyday life. ‘The flexibility ofthis text is a response to the different approaches to teaching the bridge course. In this course, probably the most important decision the instructor must make is how much emphasis to put on logic and axiomatics. Mathematicians would generally agree that proofs proceed from axioms and that the methods we use in proofs are based on principles of logic. Mathematicians would also generally agree that Jeaming to prove things in mathematics involves much more than learning to follow a set of rules. Constructing proofs isa skill that depends to a great extent on commonsense reasoning, and the formal rules involved must become so ingrained that one is barely aware of them. Different instructors have very different solutions to this dichotomy. Some believe itis necessary to give their students a thorough introduction to logic and to teach the major methods of proof explicitly before this knowledge can be intemalized. Others believe the exact opposite—that much coverage of these topics is a waste of time and perhaps even counterproductive to the real purpose of the course. These instructors prefer to start showing their students proofs right away and to discuss logic and rules primarily when questions arise. They believe that reading and writing proofs is a natural skill that, ike speech and walking, is best acquired by practice rather than by formal instruction, readily admit to being closer tothe first point of view. Twenty years of teaching and thinking about this course has convinced me that, while some students are capable of leaming how to read and write proofs by osmosis, many other good students are not quite able to do this, Also, if students never see the structure and rules that gover proofs, they might get the impression that writing proofs is a mystical or magical activity or thatthe corectness of proofs is based solely on the authority of the instructor. Therefore, this text carefully covers the essentials of mathematical logic, the role of logic in proofs, and the axiomatic method. Furthermore, this book is the only one that includes, as an appendix, a mathematically complete axiom system that is meant to be an important reference for students. ‘At the same time, this text is also an appropriate choice for instructors who prefer not to spend much time discussing logic and its relationship to proofs. Many of the sections in Unit 1 can be skimmed if desired, enabling instructors to spend most ofthe course teaching (and proving things) about sets, relations, functions, and number systems, The axiom system in Appendix 1 does not need to be covered. “Appendix 2 deserves special mention, It contains many basic results about the real numbers proved from scratch, using the ordered field axioms. Ifthe unit on logic and proofs is covered thoroughly, itis natural to study this appendix in conjunction with the chapter on proofs, It is also possible instead to delay this appendix until the unit on number systems. But an interesting altemative for instructors who prefer to introduce proofs early is to start the course with Appendix 2! The rationale is that all students, understand the basic algebraic properties of real numbers, which means that they are familiar with the ordered field axioms even if they do not know them by that name.xii Preface Furthermore, many of these proofs, especially those that do not involve inequalities, require very little logic. So Appendix 2 provides an ideal context for introducing students to proofs gently, without needing to explain any abstract concepts or complicated use of logic. ‘The exercises in this text enhance its flexibility. For one thing, they vary greatly in difficulty. In almost every section, there are some very easy problems and some rather difficult ones (marked with asterisks). There are also many types of exercises. Some problems are straightforward computations. Quite a few problems are intended to encourage the discovery process by asking the student to investigate a situation and then ‘make a conjecture (with or without proof). Since the goal of the bridge course is to teach students to read proofs as well as write them, almost every section (starting with Chapter 4) has exercises that ask the student to critique purported proofs. Of course, in a text of this type, most of the exercises ask the student to prove something or perhaps complete a proof started in the text. The Solutions and Hints to Selected Exercises at the end of the book include a few complete proofs, but they more often provide hints or outlines {to help students get started with their proofs. Additional complete proofs and teaching suggestions are provided in the Instructor's Manual. Every chapter ends with Suggestions for Further Reading that point out several possibilities in the reference list at the end of the book. These suggestions are intended both for students who might be helped by seeing more than one approach to basic ‘material and for students who are interested in pursuing a topic in more depth. Itis my sincere hope that students and instructors will find this text an enjoyable and valuable introduction to higher mathematics and its methodology. I am always interested in any type of honest feedback, including corrections and criticisms. I can be contacted by email at remeli@enipniendis. cobertswolFady choo. com. Note to the Student If you are using this book, then I presume that you are a student who has completed most or all of the undergraduate calculus sequence and that your experience in ‘mathematics $0 far has been satisfying enough that you are now planning to study some “higher” mathematics. This text and the course for which it is written are designed to provide you with a smooth introduction to higher mathematics. The existence of such books and courses should be viewed as a genuine attempt to make abstract mathematics ‘more accessible than ever before, A thorough discussion of this point is found in the preface. (If you have not read the preface, please do so. It outlines the objectives of this course and the approach this textbook takes.) Based on many years of teaching this course, I have one primary piece of advice for you: approach your study of higher mathematics with a positive and active attitude! ‘You have almost certainly heard that post-caleulus mathematics is difficult. I would not contradict that opinion. Higher mathematics is not simple. Much of it is abstract and complex and challenging to most students. If you are looking for an easy subject to study, there are better choices. But mathematics is fascinating (in fact, most ‘mathematicians consider it “beautiful"), and leaming it can be extremely rewarding. IfPreface xi you have been reasonably successful in mathematics so far, itis likely that you are capable of learning and appreciating much of post-calculus mathematics. However, your chances of succeeding in higher mathematics are very slim if you wait for it to happen to you. Unfortunately, many students enter a course like this one with an attitude that can only be described as passive, even fearful. They listen passively to lectures and take notes unguestioningly, they wait until assignments are given out before attempting problems from the text, and they wait until just before quizzes and exams to actually read the text. Even if you are somewhat apprehensive about studying abstract mathematics, you will benefit greatly if you can go into it assertively. + Ifyou find something in the text or in a lecture confusing, you may or may not choose to ask your instructor about it right away. But you will probably benefit most if ‘you tackle the point yourself—by thinking about it, reading the text and your notes to ‘ty to understand the rationale for it, and by thinking about examples that might clarify it, + Specific, concrete examples are one of the major keys to understanding abstract ‘mathematical concepts. The many examples in this book will help you. But you wi benefit even more if you try to construct your own examples. When something seems difficult to understand, ask yourself, “Can I come up with an easier version of this or a simple instance or situation that might illustrate this concept?” + Similarly, you will benefit if you do the homework that is assigned in this course thoughtfully and thoroughly. But you will benefit even more if you view the assigned problems not just as a task to get through quickly but as investigations or stepping-stones to discovery. What is the purpose of this problem? What points does it illustrate? Why is it worded the way itis? Are the restrictions in it necessary, or could itsill be solved with some restrictions loosened? What further questions does it raise? Asking such questions makes a sucessful mathematics student. These are a few suggestions that could help you take control of your study of higher mathematics rather than the other way around. Am I simply suggesting that you spend lots of time studying? No, not reaily. In the short term, itis true that approaching mathematics actively takes more time than being passive. But in the long term, an active, inquisitive attitude will actually save you time, because you will develop tools and habits that enable you to study efficiently and get to the core of concepts and problems quickly. One last piece of advice: in spite of your positive atitude, you should expect some failures. In this course, besides learning some abstract concepts, you wall be leaming a very special way of gaining knowledge. Unless you worked with proofs in high school or in your calculus courses, you probably have very little experience reading or writing them. Almost no one leams these skills quickly and painlessly. Just as in leaning to walk, everyone has to fall down many times and struggle through many halting little steps before mastering proofs. Eventually, a skill hat was a major challenge can become so muuch second nature that it's impossible to remember that it was ever difficult. With work, perseverance and a positive attitude, the ideas of higher mathematics and the language of proofs can become comfortable and familiar to you.Unit 1 Logic and ProofsChapter 1 Introduction 1.1 Knowledge and Proof ‘The purpose of many professions and subjects is to gain knowledge about some aspect of reality. Mathematics and science would seem to fit this description. (You might try to think of subjects you have studied that are not inthis category. For example, do you think that the main goal of learning to paint or to play tennis is to gain knowledge?) At some point, if you want to become proficient in such a subject, you have to understand, how knowledge can be acquired in it. In other words, you have to understand what you ‘mean when you say you “know” something, in a technical subject like mathematies or even in ordinary life. ‘What do you mean when you say you “know” something? Do you just mean that you believe itor think that it’s true? No; clearly, to know something is stronger than just to believe or have an opinion, Somehow, there's more certainty involved when you say you know something, and usually you can also provide some kind of reasons and/or justification for how you know something, How do you acquire enough grounds and/or certainty to say you know something? Here is a random sample of facts I would say I know: like chocolate chip cookies. Paper burns more easily than steel, ‘The world’s highest mountain is in Nepal Mars has two moons. ‘The Bastille was overrun on July 14, 1789, If you examine this list, you'll see that there seem to be two obvious sources of this. knowledge. One source is firsthand experience; consider the first two statements. The other source is things read in books or heard from other people, such as the last three statements, But how reliable are these sources of knowledge? No one has ever been to ‘Mars. From what I've read, everyone who has ever observed Mars carefully through a ‘good telescope has concluded that it has two moons, and s0 I confidently believe it, But do I really know it? Would I stake my life on it? Would Tbe completely devastated and4 Chapter Introduction disillusioned if someone announced that a third moon had been discovered or that the storming of the Bastille actually occurred early in the moming of July 15? Regarding the statement about buming, all my experience (and pethaps even some understanding of physies and chemistry) indicates that this statement is true, But do I really know it in any general or universal sense? Do I know that paper burns more easily than steel in subzero temperatures? At altitudes over two miles? Or even on February 29? ‘A branch of philosophy called epistemology studies questions like these. It can be defined as the study of knowledge and how itis acquired. Ina sense, this book is about the epistemology of mathematics, but it concentrates on mathematical methods rather than on philosophical issues. The purpose of this chapter is simply to start you thinking about what you mean when you say that you know something, especially in mathematics. Mathematics is a subject that is supposed to be very exact and certain. Over thousands of years, mathematicians have learned to be extremely careful about what they accept as an established fact. There are several reasons for this, The most obvious is that ‘much of mathematics is very abstract and even the most talented mathematician’ intuition, can be led astray. As a result, mathematics has evolved into a discipline where nothing is considered to be known unless it has been “proved.” In other words, any serious work in ‘mathematics must involve reading and writing mathematical proofs, sine they are the only accepted way of definitively establishing new knowledge in the field. ‘Before we begin our study of proofs in mathematics, let's take a look at what the word “proof” means in some other subjects besides mathematics. There are many other subjects in which people talk about proving things. These include all the natural sciences such as physics, chemistry, biology, and astronomy; disciplines based on the application of science such as medicine and engineering; social sciences like anthropology and sociology; and various other fields such as philosophy and law. In every subject we can expect to find slightly different criteria for what constitutes a proof. However, it tums out that all of the sciences have a pretty similar standard of what a proof is. So we begin by discussing briefly what proofs are supposed to be in science, since they are quite different from proofs in mathematies. Then we also take a ook at what a proofs in law, since it provides a sharp contrast to both mathematical proof and scientific proof. Proofs in Science ‘We all have some idea of what scientists do to prove things. When a scientist wants to prove a certain hypothesis (an assertion or theory whose truth has not yet been proved), she will usually design some sort of experiment to test the hypothesis. The experiment ‘might consist primarily of observing certain phenomena as they occur naturally, or it ‘might involve a very contrived laboratory setting. In either case, the experiment is used to obtain data—factual results observed in the experiment. (In recent years, the word “data” has been borrowed and popularized by the computer industry, which uses the word to refer to any numerical or symbolic information. This is somewhat different from the scientific meaning.) Then comes a process, usually very difficult and sometimesLA Knowledge and Proof - hotly disputed, of trying to determine whether the data support the hypothesis under investigation. ‘This description of what a scientist does is 50 oversimplified that it leaves many ‘more questions unanswered than it answers. How do scientists arrive at hypotheses to test in the first place? How do they design an experiment to test a hypothesis? Does it ‘make sense to conduct an experiment without having a particular hypothesis that you're ‘trying to prove? How well do the data from an experiment have to fit @ hypothesis in order to prove the hypothesis? Do scientists have to have a logical explanation, as well as supporting experiments, for why their hypotheses are true? And how do scientists handle apparently contradictory experimental results, in which one experiment seems to prove a hypothesis and another seems just as clearly to disprove it? “These ae just a few ofthe difficult questions we could ask about proofs in science. But without straining ourselves to such an extent, we can certainly draw some obvious conclusions. First of all, there is general agreement among scientists that the most important test of a hypothesis is whether it fits real-world events. Therefore, the most ‘common and trusted way to prove something in science is to gather enough supporting data to convince people that this agreement exists. This method of establishing general Jaws by experimentation and observation is known as the seientifie method or the ‘empirical method. It normally involves inductive reasoning, which usually refers to the mental process of “jumping” from the specific to the general, that is, using a number of observations in particular situations to conclude some sort of universal law. Does pure thought, not connected with observing real-world events, have a role in science? It definitely does. Can you prove something in science by logic or deduction ot calculations made on paper without experimental evidence? Well, these methods are definitely important in science, and some of the most important discoveries in science have been brilliantly predicted on paper long before they could be observed. In fields Jike astronomy, nuclear physics, and microbiology, it’s getting so difficult to observe things in a direct, uncomplicated way that the use of theoretical arguments to prove hypotheses is becoming more and more acceptable. An interesting contemporary example in astronomy concems the existence of black holes in space. These were predicted by very convincing reasoning decades ago, but no one has observed one. Most astronomers are quite sure that black holes exist, but they would probably hesitate to say that their existence has been proven, no matter how ironclad the arguments seem. With few exceptions, scientific theories derived mentally are not considered proved until they are verified empirically. We will see that this type of attitude is very different from what goes on in mathematics. Proofs in Law Everyone also has some idea of what it means to prove something in law. First ofall, note that a proof in a court of law is a much less objective and permanent thing than a proof in mathematics or science. A proof in mathematics or science must stand the test of time: if it does not stand up under continual scrutiny and eriticism by experts in the field, can be rejected at any time in the future. In contrast, to prove something in a jury trial ina court of law, all you have to do (barring appeals and certain other6 Chapter 1 Introduetion complications) is convince one particular set of twelve people, just fora little while, The jurors aren’t experts in any sense. In fact, they aren’t even allowed to know very much in advance about what's going on; and you even have some say in who they are. Furthermore, it doesn’t even matter if they change their minds later on! ‘Now let's consider what kinds of methods are allowable in law proofs. Can a lawyer use the scientific method to convince the jury? In a loose sense, the answer to this is definitely yes. That is, he can certainly present evidence to the jury, and evidence ‘usually consists of facts and observations of actual events, A lawyer may also conduct simple experiments, try to convince the jury to make an inductive conclusion, and use various other methods that are similar to What a scientist does. Of course, lawyers are rately as rigorous as scientists in their argumentation. But at least we can say that most ‘proof methods that are scientifically acceptable would also be allowed in a court of law. ‘What other methods of proof are available to lawyers? Well, they can certainly use logic and deductive reasoning to sway the jury. As we will se, these are the main tools of the mathematician. Lawyers can also appeal to precedent (previous legal decisions) or to the law itself, although such appeals are generally made to the judge, not the jury This is analogous to the practice in science or mathematics of using a previously established result to prove something new. ‘Are there any methods of persuasion available to a lawyer that are totally different fom scientific and mathematical methods? Again, the answer is yes. A lawyer can use a variety of psychological and emotional tricks that would be completely improper in science or mathematics. The only time that a lawyer can use these psychological tools ‘freely is during opening and closing statements ("Ladies and gentlemen of the jury, lok at my client's face. How could this sweet old lady have committed these seventeen srisly..”), However, many psychological ploys can also be used with witnesses, as long as they dre used subtly. These include leading questions, attempts to confuse or badger wimesses, clever tricks with words, gestures, facial expressions and tones of voice used to create « mood or impression, and so on. Without going into greater detail, we can see that the guidelines for proofs in law are very broad and freewheeling, for they include almost everything thatthe scientist and the mathematician can use plus a good deal more. Exercises 1.1 (2) List six statements that you would say that you know, and explain how you know cach one. Pick statements with as much variety as possible. (2) (®) Briefly discuss the differences (in your own mind) among believing that something is true, thinking that something is true, and knowing that something is true. (6) Which combinations of these conditions do you think are possible? For example, is it possible to know something is true without believing itis? (3) Briefly discuss under what circumstances you think it's appropriate to use the inductive method of drawing a general conclusion from a number of specific instances, For example, if someone is chewing gum the first three times you meet him, would you bbe tempted to say he “always chews gum”?12 Proofein Mathematics 7 (4) Mention a few ways in which a lawyer can try to convince a jury to believe something that is not true, Give some specific examples, either made up or from actual cases you have heard about. 1.2 Proofs in Mathematies ‘The preceding discussions of proofs in science and proofs in law were included primarily to provide a contrast to the main subject of this book. In this section we begin to look at the very special meaning that the word “proof” has in mathematics. How do we prove something in mathematics? That is, how do we establish the correctness of a mathematical statement? This question was first answered by various Greek scholars well over two thousand years ago. Interestingly, their basic idea of what a mathematical proof should be has been accepted, with relatively minor modifications, right up until this day. This isin sharp contrast to the situation in science, where even in the last three hundred years there have been tremendous changes, advances, and controversy about what constitutes @ proof. In part, this is because the range of methods allowed in mathematical proofs is quite a bit more specific and narrow than in other fields. Basically, almost every mathematician who has ever addressed this issue has ‘agreed that the main mechanism for proving mathematical statements must be logic and deductive reasoning. That is, the reasoning that leads from previously accepted statements to new results in mathematics must be airtight, so that there is no doubt about the conclusion. Inductive reasoning, which is the mainstay of the sciences but by its very nature is not totally certain, is simply never allowed in mathematical proofs. ‘There are examples that dramatically illustrate this point. In number theory (the branch of mathematics that studies whole numbers) there are some very famous conjectures. (Like 2 hypothesis, a conjecture is a stateinent thet has not been proved, although there is usually evidence for believing i. The word “conjecture” is generally preferred by mathematicians.) One of these is Goldbach’s conjecture, which claims that every even number greater than 2 can be written as the sum of two prime numbers. Ina few minutes, you can easily verify this for mumbers up to 100 or so. In fact, it has, been verified by computer up into the trillions. Yet no finite number of examples can possibly constitute a mathematical proof of this statement, and in fact it is considered ‘unproved! Now imagine such a situation in science, where a proposed law turns out to be true in millions of test cases, without a single failure. Itis extremely unlikely that scientists would consider the law unproved, with such overwhelming evidence for it. (By the way, number theory is full of interesting conjectures that have remained ‘unproved for centuries. We encounter more of these in Section 8.2.) ‘Thus the scientist's most valuable proof method is not considered trustworthy in ‘mathematics. And, as we saw in the previous section, the mathematician’s most valuable proof method—deduction—is of only limited use in science. For these reasons, most specialists in the foundations of mathematics do not think that mathematics should be classified as a science. There are some respected scholars who do call it an exact, science, but then they are careful to distinguish it from the empirical sciences.: Chapter 1 Introduction Discovery and Conjecture in Mathematics Can we say that the scientific method—observation, experimentation, and the formation cof conclusions from data—has no place in mathematics? No, that would be going too fat, Even if empirical methods may not be used to prove a mathematical statement, they are used al the time to enable mathematicians to figure out whether a statement is likely to be provable in the first place. This process of discovery in mathematics often has a very different flavor from the process of proof. Higher mathematics can be very intimidating, and one of the reasons is that many proofs in mathematics seem extremely sophisticated, abstract, and nonintutive. Often, this is because most of the real work is hhidden from the reader. That five-line, slick proof might well be the result of months or even years of trial and error, guesswork, and dead ends, achieved finally through patience and a little bit of luck. After that it might have been refined many times to get it down from ten pages of grubby steps to five elegant lines. This point is worth remembering when your self-confidence begins to fail. Thomas Edison's famous remark— “\Genius is 1 percent inspiration and 99 percent perspiration" —is more true of mathematics than most people realize. Although the main goal of this book is to help you leam to read and write ‘mathematical proofs, a secondary goal is to acquaint you with how mathematicians investigate problems and formulate conjectures. Examples and exercises relating to discovery and conjecture appear throughout the text. The last seven exercises in this chapter are of this sort. ‘The process of discovering mathematical truths is sometimes very different from the process of proving them. In many cases, the discovery method is completely useless as a proof method, and vice versa. On the other hand, in many cases these two processes are intimately related. An investigation into whether a certain statement is true often Jeads to an understanding of why itis or isn’t true, That understanding in turn should normally form the basis for proving that the statement is or isn't rue. ‘There is another important use of empirical methods in mathematics. It was stated previously that deduction is the only way to prove new things from old in mathematics. ‘But this raises a big question: Where do you start? How do you prove the “first thing”? Classical Greek scholars such as Eudoxus, Euclid, and Archimedes provided the answer to this question. Since you can’t prove things deductively out of thin air, the study of every branch of mathematics must begin by accepting some statements without proof. ‘The idea was to single out afew simple, “obviously true” statements applicable to any given area of mathematics and to state clearly that these statements are assumed without proof. In the great works of Euclid and his contemporaries, some of these assumed statements were called axioms and others were called postulates. (Axioms were more universal, whereas postulates pertained more to the particular subject.) Today both types are usually called axioms, and this approach is called the axiomatic method. ‘When a new branch of mathematics is developed, itis important to work out the exact list of axioms that will be used for that subject. Once that is done, there should not be any controversy about what constitutes a proof in that system: a proof must be @ sequence of irrefutable, logical steps that proceed from axioms and previously proved statements,12 Proofein Mathemattes q Euclid was one of the most important mathematicians of ancient Greece, and yet very little is known of his life. Not even the years of hhis birth and death or his birthplace are known, As @ young man, he probably studied geometry at Plato's academy in Athens. Itis known that he spent much of his life in Alexandria and reached his creative prime there around 300 nc. He is most famous for his Elements, a monumental work consisting of thirteen books, most of which deal with geometry. ‘The Elements are the oldest surviving work in which mathematical subjects were developed from scratch in a thorough, rigorous, and axiomatic way. However, the great majority of the results in Euclid’s Elements were first proved by someone other than Euclid, Euclid is remembered less for his original contributions to geometry than for the impressive organization and rigor of his work. The Elements was viewed as the model of mathematical rigor for over two thousand years and is still used as a geometry textbook in some places. ‘Although it became clear in the last century that many of Euclid’s definitions and proofs are flawed by modem standards, this does not diminish the importance of his achievement. How are the axioms for any branch of mathematics determined? Here is where ‘empirical methods come in. Since the axioms are not expected to be proved deductively, the only way to verify that they are true is by intuition and common sense, experience and lots of examples—just the sorts of things a scientist is supposed to use. For example, in the study of the ordinary algebra of the real numbers, two of the usual axioms are the commutative laws: xtyey+x and xy=yx, forall numbers x andy ‘These are good choices for axioms, for they are extremely simple statements that virtually everyone over the age of eight would agree are clearly true, so clearly true that it would seem pointless even to try to prove them. The choice of axioms in mathematics is not always such a smooth and ‘uncontroversial affair. There have been cases in which the developers of a subject spit {nto two camps over whether a particular statement should be accepted as an axiom, and in which the disagreement went on for many years. There is usually no single correct ‘answer to such an issue, ‘The theory of the axiomatic method has been liberalized somewhat in the last two centuries. The classical Greek idea was that the axioms and postulates must be true. Moder mathematics realizes that the idea of truth is often dependent on one’s interpretation and that any axiom system that atleast fits some consistent interpretation,10 Chapter 1 Introduction or model, should be an allowable area of study. The most famous example of this liberalization pertains to the parallel postulate of Euclid’s geometry, which implies the existence of straight lines in a plane that don’t meet. This seems to be obviously true; but early in the nineteenth century, it was noted that this postulate is false on the surface of a sphere (with straight lines interpreted as great circles, since arcs of great circles are the shortest paths between points on the surface of a sphere). Any two great circles on a sphere must cross (see Figure 1.1). So if one wants fo study the important subject of spherical geometry, this postulate must be rejected and replaced with one that is false in the plane. The subject of non-Euclidean geometry may have seemed like a strange curiosity when it was first introduced, bt it took on added significance in the twentieth century when Albert Einstein's general theory of relativity showed that our physical universe is actually non-Euclidean. ‘As another example, consider the equations 1+ 1= 1 and 1 + 1=0. At first glance, these are just wrong equations, and it would seem ridiculous to call them axioms. But they are wrong only in our ordinary number systems. They are true (separately, not simultaneously) in some less familiar systems of algebra in which addition has a different ‘meaning. In fact, the first equation is an axiom of boolean algebra, and the second is an, axiom in the theory of fields of characteristic 2, Both of these subjects are related to the binary arithmetic that is used in designing computer circuits. So it can be very fruitful to have strange-looking statements be axioms in a specialized branch of mathematics. One ‘wentieth-century school of thought, called formalism, holds that mathematicians should not worry at all about whether their axioms are “true” or whether the things they study have any relationship at all to the “real world.” However, most modem mathematicians ‘would not go quite so far in their loosening of the ancient Greek viewpoint. Figure 1.1 On a sphere, “straight lines” (great circles) are never parallel‘Wolf 01-01 Page lo12 Proofs in Mathemattes a Organization of the Text ‘The main goal of this book is to teach you about mathematical proofs—how to read, understand, and write them, The rest of Unit 1 includes two chapters on logic, which are intended to provide enough of an understanding of logic to form a foundation for the ‘material on proofs that follows them. The last chapter of this unit is devoted to ‘mathematical proofs. It is perhaps the most important chapter of the book, ‘Since it has been pointed out that logic and deduction are the only mechanisms for proving new things in mathematics, you might expect this whole book to be about logic. But if you look at the table of contents, you will see that only the first unit is directly devoted to logic and proofs. This is because certain other subject matter is so basic and important in mathematics that you can’t understand any branch of mathematics (let alone do proofs in it) unless you understand this cote material. This materia is covered in the book’s two other units, Unit 2 is about sets, relations, and functions. These are all relatively new concepts, in the development of mathematics. The idea of a function is only two or three centuries ‘old, and yet in that time it has become an essential part of just about every branch of ‘mathematics, a concept almost as basic to modem mathematics as the concept of a ‘number. The idea of sets (including relations) and their use in mathematics is only about a hundred years old, and yet this concept has also become indispensable in most parts of contemporary mathematics. Chapter 7, on functions, includes several other important topics such as sequences, cardinality, and counting principles. Unit 3 is about number systems, The use of numbers and counting is almost certainly the oldest form of mathematics and the one that we all leam first as children. So it should come as no surprise to you that number systems like the integers and the real numbers play an important role in every branch of mathematis, from geometry and calculus to the most advanced and abstract subjects. This unit discusses the most ‘important properties of the natural numbers, the integers, the rational numbers, the real numbers, and the complex numbers. At the same time, it introduces some of the major concepts of abstract algebra, real analysis and topology. So that's what you will lear about in this book: logic and proofs; sets, relations, and functions; and number systems. [like to think of these three topics as the building ‘blocks or essential tools of mathematical proofs. The viewpoint of this book is that if (and only if!) you leam to understand and use these basic tools will you be well on your ‘way to success in the realm of higher mathematics. Exercises 1.2 ‘Throughout this text, particularly challenging exercises are marked with asterisks. For the first three problems, you will probably find it helpful to have @ list ofall prime numbers up to 200 or so. The most efficient way to get such a list is by @ technique called the sieve of Eratosthenes: first list all integers (whole numbers) from 2 up to wherever you want to stop, say 200. Now, 2 is the smallest number in the list, so circle it and cross out all larger multiples of 2. Then 3 is the smallest remaining ‘number in the list, so circle it and cross out all larger multiples of 3. Then circle 5 and2 Chapter 1 Inteodution cross out all larger multiples of S. Continue in this manner. When you're done, the circled numbers are all the prime numbers up to 200. (If your table goes up to 200, the largest number whose multiples you need to cross out is 13. Can you see why? See Exercise 8.) (1) (@) Consider the expression n° -n +41. Substitute at Ieast a half dozen small ‘nonnegative integers forthe variable 7 inthis expression, and in each case test whether the value of the expression turns out to be a prime number, Does it seem plausible that this expression yields a prime number for every nonnegative integer n? (©) Now find a positive integer value of m for which this expression is not a prime number. Hint: You probably won't find the right n by trial and error. Instead, try to think the problem through. Q) Verify Goldbach’s conjecture for all the even numbers from 4 to 20 and from 100 to 110. (3) An interesting variant of Goldbach’s conjecture, known as de Polignac’s conjecture, is the claim that every positive even number can be written as the difference of two prime numbers. As with Goldbach’s conjecture, it is not known whether this statement is true ot false, (a) Verify de Polignac’s conjecture for each positive even number up to 12. *(b) In the unlikely event that one or both of these conjectures is actually false, 4e Polignac’s conjecture would probably be much more difficult to disprove than Goldbach’s conjecture, Can you explain why? *(4) Try to prove each of the following statements, Since we have not begun our study of axiomatic mathematics, the word “prove” is being used here in an informal sense. That is, you should try to come up with what you think are convineing arguments or explanations for why these statements are tue. Pethaps you can succeed with pictures and/or words. Or, you might need to resort to more sophisticated methods, such as algebra or even calculus. (Don’t worry if you feel as if you're groping in the dark in this, problem, When we get to Chapter 4, we get much more exact and technical about what constitutes a proof.) (@) A negative number times a negative number always equals a positive number. (You may assume that the product of two positive numbers is always positive, as well as basic algebraic rules for manipulating minus signs.) (b) If you add a positive number to its reciprocal, the sum must be at least 2 (©) The area of a rectangle equals its length times its width, (You may assume that the area of a one-by-one square is one, but this problem is still not easy.) (@ A straight line and a circle meet in at most two points. “The remaining exercises have to do with the process of discovery in mathematics; as we have discussed, this often precedes proof but is no less important.12 Proofs in Mathematics B (5) (@) Complete the last three equations: 1 = 143 - 14345 - 1434547 143454749= 7 (b) On the basis of the equations in part (a), make a conjecture about the sum of the first n odd numbers, where n can be any positive integer. (©) Test your conjecture for at east four other values ofn, including two values that are greater than 10. (6) Consider the following equations: e 2 PHP = 9 = (142% PHI%S = 36 = (142437 (2) On the basis of these equations, make a conjecture, (b) Test your conjecture for at least two other cases. (7) (@) Carefully draw three triangles. Make their shapes quite different from each other. (b) Incach triangle, carefully draw all three medians. (A median is a line from a vertex of a triangle to the midpoint of the opposite side. Use a ruler to find these ‘idpoints, unless you prefer to use an exact geometric construction!) (©) Onthe basis of your figures, make a conjecture about the medians of any triangle. *(d)_ After making some careful measurements with a ruler, make a ecnjecture about how any median of a triangle is cut by the other medians. (8) @ Ifyou haven't already done so, construct the sieve of Eratosthenes for ‘numbers up to 200, as described before Exercise 1 (b) By trial and erro, fill in each of the following blanks with the smallest ‘number that makes the statement correct: () Every nonprime number less than 100 has a prime factor less than _ (ii) Every nonprime number less than 150 has a prime factor less than (Gi) Every nonprime mumber less than 200 has a prime factor less than“4 Chapter 1 Introduction (©)_ Using your results from part (b), additional investigation if you need it, and some logical analysis ofthe situation, fill inthe following blank with the expression that you think yields the smallest number that makes your conjecture correct: Every nonprime number 1 has a prime factor equal to or less than (9) ‘The numbers 3, 4, and 5 can be the sides of a right-angled triangle, since they satisfy Pythagoras’s theorem (the familiar a” + 6? = c*), Positive integers with this property are called Pythagorean triples. The triple 3, 4 5 also has the property that the largest numberof the triple (the hypotenuse) is only one more than the middle number. (@) Find two more Pythagorean triples with this property. (©) Could the smallest member of a Pythagorean triple with this property be an even number? Why or why not? *(¢)_ Try to find a general formula or rule that can be used to lis ll Pythagorean, triples ofthis type (@) Can two of the numbers ina Pythagorean triple be equal? Why or why not? (You may use the fact that 2 is not equal to any fraction.) (10) Starting with any positive integer, itis possible to generate a sequence of numbers by these rules: If the current number is even, the next number is half the current number. Ifthe current number is odd, the next number is ! more than 3 times the current number. For example, one such sequence begins 26, 13, 40, 20, 10, 5, 16, .. (@) Choose three or four starting numbers, and for each of them generate the sequence just described. Keep going until the sequence stabilizes in a clear-cut way. (A. good range for most of your starting numbers would be between 20 and 50.) (©) On the basis of your results in part (a), make a conjecture about what happens to these sequences, for any starting number. It tums out that @ general law does hold here; that is, al such sequences end in exactly the same pattem. However, itis quite difficult to prove this theorem, or even understand intuitively why it should be ‘rue. (11) The ancient game of Nim is very simple to play (in terms of both equipment and rules) but is quite entertaining and challenging. It is also a good setting for leaning about the mathematical theory of games. Here are the rules: ‘Nim is a competitive game between two players. To start the game, the players create two or more piles of match sticks, not necessarily equal in number. One classic starting configuration uses piles of three, four, and five, but the players can agree to any starting configuration (see Figure 1.2) ‘After the setup, the players take tums. When itis his or her turn, a player must remove at least one match stick from one pile. For instance, a player may remove an entire pile at one tur; but a player may not remove parts of more than one pile at one tum. The player who removes the last match stick wins the game. ‘Once the starting configuration is determined, Nim becomes a “finite two-person ‘win-lose game of perfect information.” The most important mathematical result about such a game is that one player (cither the one who plays first or the one who plays second) has a strategy that always wins for that player.1.2. Proofein Mathematics 1s Figure 1.2 One typical starting configuration for Nim (@) Play several games of Nim (by yourself or with someone else) using only two piles of sticks but of various sizes. On the basis of your experience, devise a rule for determining which player has the winning strategy for which games of this type, and what that strategy is. You will be asked to prove your conjecture in Section 8.2. (©) _Analtemate version of Nim states thatthe one who removes the ast match stick loses. Repeat part (a) with this alternate rule. *(€) Repeat part (2), now starting with three piles of sticks but with one of the piles having only one stick. “(@)_ Repeat part (c) using the alternate rule of part (b).. ‘Suggestions for Further Reading: Literally thousands of fine books have been written about the subjects touched on in this chapter, including inductive and deductive reasoning, the processes of discovery and proof in science and mathematics, and the history of the axiomatic method. A few of these appear in the References at the end of this text: Davis and Hersh (1980 and 1986), Eves (1995), Kline (1959 and 1980), Lakatos (1976), Polya (1954), and Stabler (1953). For a witty and informative discussion of Goldbach’s conjecture and related problems of number theory, see Hofstadter (1989).Chapter 2 Propositional Logic 21 The Basics of Propositional Logic ‘What is logic? Dictionaries define it to be the study of pure reasoning or the study of valid principles of making inferences and drawing conclusions. As Chapter 1 emphasized, logic plays an extremely important role in mathematics, more so than in the sciences or perhaps in any other subject or profession. The field of mathematical logic is divided into the branches of propositional logic and predicate logic. ‘This chapter is about propositional logic. This is a very old subject, first developed systematically by the Greek philosopher Aristotle. It has various other names, including the propositional calculus, sentential logic, and the sentential calculus. Basically, propositional logic studies the meaning of various simple words like “and,” “or,” and “not” and how these words are used in reasoning. Although it is possible to carry out this study without any special terminology or symbols, it’s convenient to introduce some, Definition: A proposition is any declarative sentence (including mathematical sentences such as equations) that is true or false. Example 1: (@) “Snow is white” is a typical example of a proposition. Most people would agree that it’s a trie one, but inthe real world few things are absolute: city dwellers will tell you that snow can be grey, black, or yellow. (b) “3 + 2=5” isa simple mathematical proposition. Under the most common interpretation of the symbols in it, it is of course true. (©) “3+2=7" is also proposition, even though its false inthe standard number system, Nothing says a proposition can't be false. Also, this equation could be true (and the previous one false) in a nonstandard number system. (@) “Is anybody home?” is not a proposition; questions are not declarative sentences, (@) “Shut the door!” and “Wow!” are also not propositions, because commands and exclamations are not declarative sentences. (®) “Ludwig van Beethoven sneezed at least 400 times in the year 1800” is a sentence whose truth is presumably hopeless to verify or refute, Nonetheless, such, sentences are generally considered to be propositions, 162 The Basis of Propositions Logie W Aristotle (384-322 8.¢), like his teacher Plato, was a philosopher who was very interested in mathematics but did not work in ‘mathematics o any extent. Aristotle was apparently the first person to develop formal logic in a systematic way. His treatment of ‘propositional logic does not differ greatly from the modem approach to the subject, and the study of logic based on truth conditions is still called Aristotelian logic. Besides writing extensively on other humanistic subjects such as ethics and political science, Aristotle also produced the first important ‘works on physics, astronomy, and biology. Some of his claims were rather crude by modem standards and others were simply wrong. For example, Aristotle asserted that heavy objects fall faster than light ones, a belief that was not refuted until the sixteenth century, by Galileo. Still, his scientific work was the starting point of much of modem science, Very few people in the history of humanity have contributed to as many fields as Aristotle. (g)_ "x > 5” is a mathematical inequality whose truth clearly depends on more information, namely what value is given to the variable x. Ina sense, the truth or falsity of this example is much easier to determine than that of example f. Even so, we follow standard practice and call such sentences predicates rather then propositions. (h) “Diane has beautiful eyes” is a sentence whose truth depends not only on getting more information (which Diane is being referred to?) but also on a value Judgment about beauty. Most logicians would say that a sentence whose truth involves 1 value judgment cannot be a proposition. We use the word statement as a more all-encompassing term that includes propositions as well as sentences like the last two examples. Section 3.2 clarifies this, terminology further. (®) °23 is a purple number” has more serious flaws than examples (g) and (1). ‘Neither more information nor a value judgment determines its truth or falsehood, Most people would say this sentence is meaningless and therefore not a statement. (j) “This sentence is false” is a simple example of a paradox. Ifit’s true, then it rust be false, and vice versa. So there is no way it could sensibly be called true or false, and therefore itis not a statement, Notation: We use the letters P, Q, R, ... as propositional variables. That is, we let these letters stand for or represent statements, in much the same way that a ‘mathematical variable like x represents a number.18 Chapter? Propositional Logic Notation: Five symbols, called connectives, are used to stand for the following words: . \ for “and” . V. for “or” . ~ for “not” . + for “implies” or “if... then” . + for “if and only if” ‘The words themselves, as well as the symbols, may be called connectives. Using the connectives, we can build new statements from simpler ones. Specifically, if P and Qare any two statements, then PAQ, PVQ, ~ P+Q and PQ are also statements. Definitions: A statement that is nor built up from simpler ones by connectives and/or quantifiers is called atomic or simple. (Quantifiers are introduced in Chapter 3.) ‘A statement that is built up from simpler ones is called compound. Example 2: “I am not cold,” “Roses are red and violets are blue,” and “If a funetion is continuous, then it’s integrable” are compound statements because they contain connectives. On the other hand, the statements in Example 1 are all atomic. Remarks: That's pretty much all there is to the grammar of propositional logic. However, there are a few other details and subtleties that ought to be mentioned, (1) Notice that each connective is represented by both a symbol and a word (ot phrase). The symbols are handy abbreviations that are useful when studying logic or Jeaming about proofs. Otherwise, the usual practice in mathematics is fo use the words rather than the symbols, Similarly, propositional variables are seldom used except when studying logic. (2) Why do we use these particular five connectives? Is there something special about them or the number five? Not at all. It would be possible to have dozens of ‘connectives. Or we could have fewer than five connectives—even just one—and still keep the full “power” of propositional logic. (This type of reduction is discussed in the exercises for Section 2.3.) But it’s petty standard to use these five, because five seems like a good compromise numerically and because all these connectives correspond to familiar thought processes or words.21 The Basis of Propesitonal Logie 19 (3) When connectives are used to build up symbolic statements, parentheses are often needed to show the order of operations, just as in algebra. For example, it’s confusing to write PQ VR, since this could mean either P A (QV R) or ([Link])VR. However, just as in algebra, we give the connectives a priority ordering that resolves such ambiguities when parentheses are omitted, The priority of the connectives, from highest to lowest, is~,\,V,,—*, + . (This order is standard, except that some books give / and V equal priority.) How is a statement interpreted when the same connective is repeated and there are no parentheses? In the case of (or V, this is never a problem. The statement (P/AQ)AR has the same meaning as P A (QR), so it's perfectly unambiguous and acceptable to write PA QAR; and the same holds for V. (Note that this is completely analogous to the fact that we don’t need to put parentheses in algebraic expressions of the form a+ +c and abc.) On the other hand, repeating —* or +* can create ambiguity. In practice, when a mathematician writes a statement with the logical form P ~» Q—* R, the intended meaning is probably (P —* Q) A (Q—* R), rather than (P ~* Q) ++ R or P+ (QR) A similar convention holds for + . This is analogous to the meaning attached to extended equations and inequalities ofthe forms x= y=z, x~P means (P+Q) + [-Q4EPI] ‘Terminology: Each of the connectives has a more formal name than the word it stands for, and there are situations in which this formal terminology is useful. Specifically, the connective /\ (“and”) is also called conjunction. A statement of the form PA Q is called the conjunction of P and Q, and the separate statements P and Qare called the conjunets of this compound statement. ‘Similarly, the connective V (“or”) is called disjunction, and a statement P VQ is called the disjunction of the two disjunets P and Q. ‘The connectives ~, +, and <> are called negation, conditional (or implication), and biconditional (or equivalence), respectively. Now it's time to talk about what these connectives mean and what can be done with them. In propositional logic, we are primarily interested in determining when statements are true and when they are false, The main tool for doing this is the following. Definition: ‘The truth functions of the connectives are defined as follows: . PAQ is true provided P and Q are both true. . P-VQ is true provided at least one of the statements P and Q is true,20 Chapter? Propositional Loge + =P is tre provided P is false. + P-¥Q is true provided P is false, or Qis true (or both). + PQ is true provided P and Q are both true or both false. Note that these truth functions really are functions except that, instead of using ‘numbers for inputs and outputs, they use “truth values,” namely “true” and “false,” (If you are not very familiar with functions, don’t be concemed; we study them from Scratch and in depth in Chapter 7.) We usually abbreviate these truth values as T and F. ‘Since the domain of each truth function i a finite set of combinations of Ts and Fs, ‘we can show the complete definition of each truth function in a truth table, similar to the addition and multiplication tables you used in elementary school. The truth tables forthe five basic connectives are shown in Table 2.1, ‘Table 2.1 Truth tables of the connectives It is important to understand how the truth functions of the connectives relate to theirnormal English meanings. Inthe cases of ~ and A, the relationship is very clear, but itis less so with the others. For example, the truth function for V might not correspond to the most common English meaning of the word “or.” Consider the statement, “Tonight I'll go to the volleyball game or I'l see that movie.” Most likely, this means Iwill do one of these activities but of both. This use of the word “or,” which excludes the possibilty of both disjuncts being true, is called the exclusive or. The truth funtion ‘we have defined for V makes it the inclusive or, corresponding to “and/or.” In English, the word “or” can be used inclusively or exclusively; this can lead to ambiguity. For instance, suppose someone said, “I’m going to take some aspirin or call the doctor.” Does this statement leave open the possibility that the person takes aspirin and calls the doctor? It may or may not. In mathematics, the word “or” is generally used inclusively, If you want to express an exclusive or in a mathematical statement, you must use extra21 ‘The Basies of Propositional Logic 21 ‘words, such as “Either P or Q is true, but not both” or “Exactly one of the conditions P and Q is true” (see Exercise 8) “There are enough subtleties involving the connectives ~* and ++ that the entire next section is devoted to them. Using the five basic truth functions repeatedly, it’s simple to work out the truth function or truth table of any symbolic statement. (If you have studied composition of functions, perhaps you can see that the truth function of any statement must be a composition of te five basic truth functions.) Some examples are shown in Table 2.2. Note how systematically these truth tables are constructed. If there are n propositional variables, there must be 2" lines in the truth table, since this is the number of different ordered n-tuples that can be chosen from a two-element set (Exercise 11). So a truth table with more than four or five variables would get quite cumbersome. Notice that these tables use a simple pattem to achieve all possible combinations of the propositional variables. Also, note that before we can evaluate the output truth values of the entire statement, we have to figure out the truth values of each of its substatements. ‘We can now define some useful concepts. Definitions: A tautology, ora law of propositional logic, isa statement whose truth function has all Ts as outputs. A contradiction is a statement whose truth function has all Fs as outputs (in other words, t's a statement whose negation isa tautology). “Two statements are called propositionally equivalent ifa tautology results when the connective ++ is put between them. (Exercise 7 provides an alternate definition of this concept) Example 4: One simple tautology is the symbolic statement P — P. This could represent an English sentence like “If I don’t finish, then I don’t finish.” Note that this, sentence is obviously true, but it doesn’t convey any information. This is typically the case with such simple tautologies (One of the simplest and most important contradictions is the statement P \~ P. An English example would be “T love you and I don't love you.” Although this statement ‘might make sense in a psychological or emotional context, itis still a contradiction. That is, from a logical standpoint it cannot be true ‘The statement ~ P -* Q is propositionally equivalent to PV Q, as you can easily verify with tables. For instance, iI say, “If don’t finish this chapter this week, I'm in trouble,” this is equivalent to saying (and so has essentially the same meaning as), “I (aust) finish this chapter this week or I'm in trouble.” FFor the rest of this chapter, we use “equivalent” for the longer “propositionally equivalent.” Note that statements can be equivalent even if they don’t have the same set of propositional variables. For example, P -* (Q A ~ Q) is equivalent to ~P, as you can easily verify with truth tables2 Chapter? Propositional Logle Table 2.2 Truth tables of three symbolic statements ‘Truth Table of (PAQ)V~P PQ PAQ. ~P TT T F T TO OF F F F FOOT F T T FOF F T T Truth Table of P+ ([Q-*(PAQ)] PQ PAQ. _Q>RAQ P+/Q> (PA TT T T T T OF F T T FOOT F F r FOF F T T Truth Table of (P+ Q) (RAP) Po maa tals Sasansaa somal Sam “The ideas we have been discussing are quite straightforward as long as we restrict ourselves to symbolic statements, They become more challenging when they are applied to English or mathematical statements. Since logic is such a vital part of mathematics, every mathematics student should lear to recognize the logical structure of English and ‘mathematical statements and translate them into symbolic statements. With English statements, there is often more than one reasonable interpretation of their logical structure, but with mathematical statements there rarely is. Here are some examples of how this is done. Example 5: For each of the following statements, introduce a propositional variable for each of its atomic substatements, and then use these variables and connectives to write the most accurate symbolic translation of the original statement. (a) [like milk and cheese but not yogurt.2A The Basie of Propositional Logie 2 (b) Rain means no soccer practice. (©) The only number that is neither positive nor negative is zero. @ 242=4, Solution: (a) Don't be fooled by a phrase like “milk and cheese.” Connectives ‘must connect stafements, and a noun like “milk” is certainly not a statement. To understand its logical structure, the given statement should be viewed as an abbreviation for “I like milk and I like cheese, but I don’t like yogurt.” So we introduce the following propositional variables: P for “like milk.” Q for “like cheese.” R._ for “Tike yogurt.” ‘The only remaining difficulty is how to deal with the word “but.” This word conveys a different emphasis or mood from the word “and,” but the basic logical ‘meaning of the two words is the same. In other words, in statements where the word “but” could be replaced by “and” and still make sense grammatically, the right ‘connective for itis /\. So the best symbolic representation of the original statement is PAQA~R. (©) Once again, connectives must connect entire statements, not single words or noun phrases. So we write: P for “Itis raining.” Q for “There is soccer practice.” How should we interpret the word “means”? Although it would be plausible to think of itas “ifand only if,” the most sensible interpretation is that if it rains, there's no soccer practice. So we represent the given English statement as P + ~ Q (©) Since this statement involves an unspecified number, we can use a ‘mathematical variable like x to represent it ([t is possible to do this problem without using a letter to stand for the unspecified number, but the wording gets a bit awkward.) So we write: P for “xis positive Q for “wis negative.” R__ for “xis zero.” ‘Now we must interpret various words. A bit of thought should convince you that “neither P nor Q” has the logical meaning ~ (P VQ) or its propositional equivalent24 Chapter 2 Propositional Logie ~P A~ Q, The words “the only” in this statement require a quantifier to interpret precisely, but the gist of the statement seems to be that a number is neither positive nor negative if and only if the number is zero. So the statement can be represented symbolically as (~P A~Q)**R. If we allow ourselves mathematical symbols as well a5 connectives, we would. probably prefer to represent the statement in the form E@>O)A~ <0) x= or shorter still (er OAxeO)erx=0 (We use the standard convention that a slash through an equal sign, an inequality symbol, and so on, can be used instead of a negation symbol.) It should be noted that quantifiers are required for a totally accurate translation of this statement (@ This is sort of trick question. The statement contains no connectives, 50 it is atomic. Therefore, the only way to represent it symbolically is simply P, where P represents the whole statement! Tis very tempting just to assume that this simple equation isa tautology. But since its logical form is P, its not. It's certainly a true statement of arithmetic, and you might even claim that it's a law of arithmetic, but it’s nof a law of propositional logic. Even a statement like 1 = 1 is technically not a tautology! Exercises 2.1 (1) Construct the truth tables of the following statements: @) ~®AQ &) PH RVO (©) P>~P @ Po~P (©) P+Q+PAQ) ©) ~PAQ+CPA~Q (@) PAQAR) * @AQAR () [@VQ)R) > [P+R)AQ-+R] @) PAQVEPAR) Q) For each of the following, state whether it is a proposition, with a brief explanation. If you believe that a particular ease is borderline, provide brief pros and cons for whether it should be considered a proposition. For those which are propositions, determine which are true and which are false if possible. (2) 10s. prime number. (b) Are there any even prime numbers?24 The Basis of Propositional Loge 2s (©) Tum off that music or I'l seream, (@) Life is good. (@) 345. (f) The number mis bigger than 4. (g) Benjamin Franklin had many friends. (h). The Chicago Cubs will win the World Series in the year 2106. (@)_Llike olives but not very much. G)_ Goldbach’s conjecture is true. (This was described in Chapter 1.) (3) Determine whether each of the following is a tautology, a contradiction, or neither. If you can determine answers by commonsense logic, do so; otherwise, construct tith tables. @ -AQ = -FA-8 () ~PA~Q + ~PAQ (©) P+) + QP) @ @+Q) + Q>P) (©) [(@VQVR] + PVQVR) ©) [EVQAR] + PVQAR) (4) Determine whether each ofthe following pairs of statements are propositonally equivalent to each other. Ifyou can determine answers by commonsense logic, do so; otherwise, construct truth tables. (@) PAQ and QAP () P and ~~P (©) ~@VQ) and ~PV~Q (@) ~@VQ) and ~PA~Q (@) P-*Q and Q>P (f) ~@4Q) and ~P-+~Q (g) P++Q and PAQV~PV® G) PAQVR) and @AQVR @ PAQAR) and PAQVAR G) P-Q>R) and PIQRK ® Po QeR and PHOHRK (5) Match each statement on the left with a propositionally equivalent one on the right. As with the previous problem, see if you ean do this without writing out truth tables. (a) P>~ @ PA~P &) PHPAQ Gi) PQ PVOAPAQ i - PA @ Po~P Gv) Q>P © EVAWEAY — HP -Q (vi) ~ (vil) PQ (viii) QA~P26 Chapter2 Propositional Logie (©) Foreach ofthe following, replace the symbol # with a connective so thatthe resulting symbolic statement isa tautology. If you can, figure these out without using truth tables. @) -@#Q) + [PA~Q) ©) [P>QAR] + [P+ QAR] (©) [(P#QR]+ [PRA (Q+R)} @ (PAQSP] + PHO) © [P#Q RP QR) (1) Show, using a commonsense argument, that for two symbolic statements to be propositionally equivalent means precisely that they have the same truth value (both true or both false) for any truth values ofthe propositional variables in them. (8) Recall the discussion of the inclusive or and the exclusive or. Let the symbol [Link] the latter. (a) Construct the truth table for P VQ. (b) Write a statement using our five basic connectives that is equivalent to PvQ (©) Write a statement using only the connectives ~, A, and V that is equivalent toPVQ (4) Make up an English sentence in which you feel the word “or” should be interpreted inclusively. (€) Make up an English sentence in which you feel the word “or” should be interpreted exclusively. (f) Make up an English sentence in which you feel the word “or” can be interpreted either way. () Let P,Q, and R stand for “Pigs ar fish," “2 +2= 4," and “Canada is in Asia,” respectively. Translate the following symbolic statements into reasonable-sounding English. Also, determine whether each of them is true or false. @) PV~Q @) Qo-R (© ~Q>RA~P) (@ P>~P (10) Foreach ofthe following statements, introduce a propositional variable foreach of its atomic substatements, and then use these variables and connectives to write the ‘most accurate symbolic translation of the original statement. (a) Ineed to go to Oxnard and Lompoc. (©) Ifa number is even and bigger than 2, it's not prime, (©) You're damned if you do and damned if you don't. (@) Ifyou order from the dinner menu, you get a soup ora salad, an entree, and a beverage or a dessert. (Be careful with the word “or” inthis one.) (©) Ifitdoesn’t rain in the next week, we won't have vegetables or flowers, but if it does, we'll at least have flowers.2:2 Conditionals and Blcondiionals 7 (£) No shoes, no shirt, no service. (Of course, this is a highly abbreviated sentence, You have to interpret it properly.) (g) Men or women may apply for this job. (Be careful; this one’s a bit tricky.) (11) (@) Ifa symbolic statement has just one propositional variable (say P), how ‘many lines are in its truth table? (©) How many different possible truth functions are there for such a statement? ‘That is, in how many ways can the output column of such a truth table be filled in? Explain. (c), Repeat parts (a) and (b) for a symbolic statement with two propositional variables P and Q. Explain. *(é)_ On the basis of the previous parts of this problem, make conjectures that generalize them to a symbolic statement with an arbitrary number 1 of propositional variables. 2.2. Conditionals and Biconditionals ‘The connectives -* and +* are not only the most subtle ofthe five connectives; they are also the two most important ones in mathematical work. So it is worthwhile for us to discuss them at some length. We begin this section by considering the meaning of conditional statements. In the previous section, we linked the connective + to the word “implies,” but in ordinary language this word is not used very frequently. Probably the most common way of expressing conditionals in English is with the words “If... then ...” AS we see shortly, there are several other words or combinations of words that also express conditionals. Conditional and biconditional statements are often called implications and equivalences, respectively. However, there is a tendency to reserve these latter words for statements that are known to be true, For instance, “2 +3 = 5 if and only if pigs can fly” is a biconditional statement. But many mathematicians would not call it an equivalence, since it is false, Regardless of what words are used to represent conditionals, it takes some thought to understand the truth function for this connective. Refer back to Table 2.1 and note that the statement P ~* Q is false in only one of the four cases, specifically when P is, true and Q is false. Example 1: ‘The best way to understand why this makes sense is to think of a conditional as a promise. Not every conditional ean be thought of inthis way, but many can, So let's pick one at random, like “If you rub my back today, I'l buy you dinner tonight.” This is certainly a conditional; it can be represented as PQ, where P is “You rub my back today” and Q is “I'll buy you dinner tonight." Under what circumstances is or is not this promise kept?a Chapter? Propositional Logie ‘Two of the four entries in the truth table are clear-cut. Ifyou rub my back and I buy you dinner, I've obviously kept the promise, so the whole conditional is true. On the other hand, if you rub my back and I don’t buy you dinner, I've obviously broken my promise and the conditional must be considered false. It requires more thought to ‘understand the two truth table entries for which P is false. Suppose you don't rub my back and I don’t take you to dinner. Even though I haven't done anything, no one could say I've broken my promise. Therefore, we define PQ to be true if both P and Q are false. Finally, we get to the least intuitive case, Suppose you don’t rub my back but I go ahead and buy you dinner anyway. Have I broken my promise? If you reflect on this question, you will probably conclude that, although it’s unexpected for me to buy you inner afier you didn’t rub my back, it's not breaking my promise, To put it anoiher ‘way, although my promise might lead most people to assume that if you don’t rub my back, I won't buy you dinner, my statement doesn't say anything about what I'l do if you don’t rub my back. It is with these considerations in mind that the third entry in the ‘ruth table is also a T. A good way to understand these last two cases is to admit that if you don’t rub my back, my promise is true by default, because you haven't done anything to obligate me to act one way or the other regarding dinner. Now here’s some useful terminology. Definitions: In any conditional P -» Q, the statement P is called the hypothesis or antecedent and Q is called the eonelusion or consequent of the conditional Definitions: Given any conditional P+ Q, + the statement QP is called its converse. + the statement ~P-»~Q is called its inverse. + the statement ~ Q— ~P is called its contrapositive. ‘We now come to the first result in this text that is labeled a “theorem.” Since our serious study of proofs does not begin until Chapter 4, many of the theorems in this chapter and the next are presented in a very nonrigorous way. In other words, the proofs given for some of these theorems have more of the flavor of intuitive explanations than of mathematical proofs. Theorem 2.1: (a) Every conditional is equivalent to its own contrapositive. (©) A conditional is not necessarily equivalent to its converse of its inverse. (© However, the converse and the inverse of any conditional are equivalent to each other, (@) The conjunction of any conditional P -* Q and its converse is equivalent to the biconditional P + Q. Proof: This theorem is so elementary that we can prove it rigorously at this point. ‘The proof simply requires constructing several truth tables. For instance, to prove part (2) we only need to show that (P -* Q) ++ (~ Q-+ ~ P) isa tautology (Exercise 10).22 29 Example 2: Consider the conditional “If you live in California, you live in America.” This statement is true forall persons. lis converse is “If you live in America, ‘you live in California”; its inverse is “If you don’t live in Califomnia, you don’t live in “America.” These two statements are not true in general, so they are not equivalent to the original. However, they are equivalent to each other. The contrapositive of the original statement is “If you don’t live in America, you don’t live in California,” which has the same meaning as the original and is always true. By the way, it’s worth bearing in mind that implication is the only connective whose meaning changes when the two substatements being connected are switched, ‘That is, P A Q is equivalent to Q A P, and so on, Let's elaborate a bit on our earlier discussion of conditionals as promises. When someone says “If you rub my back today, I'll buy you dinner tonight,” many people would automatically read into it “And if you don't rub my back, I won't buy you dinner.” Note that this other promise is just the inverse of the original one. Now, there is no doubt that in ordinary language, when a person states a conditional, the inverse is sometimes also intended, And then again, sometimes itis not. This kind of fuzziness is a normal feature of spoken language, as we have already mentioned regarding the ambiguity ofthe word “or” (inclusive versus exclusive). But in mathematics and logic, connectives must have precise meanings. The most useful decision is to agree that conditionals in general should not include their own inverses, forthe simple reason that if they did, there would be no difference between conditionals and biconditionals (by ‘Theorem 2.1 (c) and (d)). In spoken language, conditionals aren’t always promises, but they almost always at least convey some Kind of causal connection between the antecedent and the consequent. When we say “P implies Q” or even “IfP then Q,” we normally mean that, the statement P, if true, somehow causes or forves the statement Q to be true. In ‘mathematics, most conditionals convey this kind of causality, but it s not a requirement. In logic (and therefore in mathematics), the truth or falsity of a conditional is based strictly on truth values, Example 3: The following three statements, although they may seem silly or even wrong, must be considered true: If 2+2=4, then ice is cold. If 2+2=3, then ice is cold. If 2+2=3, then ice is hot. On the other hand, the statement “If2 + 2= 4, then ice is hot” is certainly false, ‘There are quite a few ways of expressing conditionals in words, especially in ‘mathematics. Itis quite important to be familiar with all of them, so let’s talk about them. fora bit. You will find the most common ones listed in Table 2.3.30 Chapter? Propositional Logle ‘Table 2.3 The most common ways to express a conditional P+ Q in words (1) P implies Q. 2) If P then Q @ AQ @ QitP. () P onlyif Q (©) P issufficient for Q. (1D Q isnecessary for P. (8) Whenever P, Q. (9) Q whenever P. ‘Note that statements 1-4 of Table 2.3 contain nothing new—but pay attention to the word order in statement 4. For example, in the sentence “I'll buy you dinner if you rib my back,” the hypothesis consists of the last four words and the conclusion is the first four words. Now consider statement 5. An example of this construction is “You'll see the comet only if you look in the right spot.” What is this saying? The answer is open to debate, but the most likely meaning is “If you don’t lookin the right spot, you wont (or can't) see the comet,” which is the contrapositive of “If you (expect to) see the comet, you (have to) look in the right spot.” (The words in parentheses have been added to ‘make the sentence read better.) And this is what statement 5 says this sentence should ‘mean. On the other hand, it’s possible to believe that the sentence might also be saying, “if you do look in the right spot, you'll see the comet.” But we reject this interpretation because it would mean that “only if” would be a synonym for “if and only if.” We therefore follow the standard convention that “P only if Q* is the converse of “P if Q.” and neither of these means the same as “P if and only if Q.” ‘The pair of words “sufficient” and “necessary,” like the words “if” and “only if,” express conditionals in the opposite order from each othet. Suppose you are told, Passing the midterm and the final is sufficient to pass this course.” This appears to ‘mean that if you pass these exams, you will pass the course. But does it also mean that if you don't pass both these exams, you can’t pass the course? Again, that interpretation is possible, but the word “sufficient” seems to allow the possibility that there might be other ways to pass the course. So, as with the words “if” and “only if,” we reject this, other interpretation so that the word ‘sufficient” conveys the meaning of a conditional, not a biconditional. Now, suppose instead that you are told “Passing the midterm and the final is necessary to pass the course.” With only one word changed, this sentence has 2 completely different emphasis from the previous one. This sentence certainly does not say that passing the exams is any sort of guarantee of passing the course. Instead, it22 Conditionals and Biconaitonals rt appears to say that you must pass the exams to even have a chance of passing the course, ‘or, more directly, if you don’t pass the exams, you definitely won't pass the course. So, as statements 6 and 7 of Table 2.3 indicate, the word “necessary” is generally considered to express the converse of the word “sufficient.” ‘Statements 8 and 9 indicate that the word “whenever” often expresses a conditional. In the sentence “Whenever a function is continuous, it's integrable,” the ‘word “whenever” is essentially a synonym for “if.” English (and all spoken languages) has many ways of expressing the same thought, and even Table 2.3 does not include all the reasonable ways of expressing conditionals. It should also be pointed out that many statements that seem to have no connective in them are really conditionals. For instance, the important theorem, “A differentiable function is continuous,” is really saying that if a function is differentiable, it’s continuous. “Hidden connectives” are aso often conveyed by quantifies, as Section 3.2 demonstrates 1 — Without any doubt, the most frequent logical error made by mathematics students at all levels is confusing a conditional with its converse (or inverse) or assuming that if a conditional is true, its converse must also be true. Lear to avoid this confusion like the plague, and you will spare yourself much grief! Biconditionals ‘There are various ways to think of biconditionals, one of which was stated in Theorem 2.1): P +4 Q is equivalent to (P + Q) A (Q—* P). That is, when you assert both a ‘conditional and its converse, you're stating a biconditional. That's why the symbol for a biconditional is a double arrow. That's also why we use the phrase “if and only if” for biconditionals. (By the way, mathematicians often use the abbreviation “iff” for “if and only if") Table 2.4 shows this and other ways of expressing biconditionals, We have seen that the words “necessary” and “sufficient” also have converse meanings, and so the phrase “necessary and sufficient” is often used to express biconditionals. For example, if you read that “a necessary and sufficient condition for ‘a number to be rational is that its decimal expansion terminates or repeats,” that means that a number is rational if and only if its decimal expansion terminates or repeats. (The noun “condition” is often used in this way with the words “necessary” and/or “‘sufficient.") Another common way of expressing biconditionals in mathematics is with the word “equivalent.” For example, an alternate way of stating the same fact about ‘numbers that was just given would be “Rationality is equivalent to having a decimal expansion that either terminates or repeats.” (When mathematicians say that two statements are equivalent, it does not necessarily mean that they are propositionally equivalent. It just means that they can be proved to imply each other, using whatever axioms and previously proved theorems are available in the situation.) Finally, Table 2.4 indicates that the words “just in case” can also convey a biconditional, as in “A number is rational just in case its decimal expansion either terminates or repeats.”32 Chapter? Propositional Logle Table 2.4 The most common ways to express a biconditional P + Q in words (1) P ifand only if Q Q) P ismecessary and sufficient for Q. @) P isequivalent to Q. (4) P and Q are equivalent, (5) P (is tru) just in case Q (is) We have already mentioned that, in ordinary speech, statements that on the surface are just one-way conditionals are often understood to be biconditionals. This is partly because there are no fluid-sounding ways of expressing biconditionals in English. All the phrases in Table 2.4 sound fine fo a mathematician, but they are somewhat awkward ‘when used in ordinary conversation, If say “You'll pass this course if and only if you pass the midterm and the final,” I'm clearly stating a biconditional, but it sounds strange. Since people are not used to hearing the phrase “if and only if,” they might take this statement to mean a biconditional even if the words “if and” are left out. This interpretation could lead to some serious disappointment, since with these two words omitted I would only be stating a conditional. There are several useful ways of thinking of biconditionals. Most directly, biconditional represents a two-way conditional. Another way of looking at a biconditional P ++ Q is that if either P or Q is true, they both are. That is, either they're both true, of they're both false. So a biconditional between two statements says that they have the same sruth values. For this reason, the biconditional connective is very similar ‘to an equal sign, except that it is applied to statements rather than to mathematical quantities. To put it even more strongly, when mathematicians assert that two (or more) statements are equivalent, they are more or less saying that these statements are different ‘ways of saying the same thing. “We conclude this section with our first proof preview. These are called “previews” because they occur before our in-depth study of proof’. Thus they are not axiomatic or rigorous proofs. But each of them illustrates at least one important proof technique, and. ‘we see later that each of them can be fleshed out fo a more complete, rigorous proof. Furthermore, the relatively informal style of these proof previews is typical of the way ‘mathematicians write proofs in practice. In these proof previews, and occasionally elsewhere in proofs in this book, ‘comments in brackets and italics are explanations to the reader that would probably not be included under normal circumstances. Proof Preview 1 ‘Theorem: (2) An integer m is even if and only ifm + 1 is odd. (b) Similarly, » is odd if and only ifm +1 is even. Proof: (a) [We are asked to prove a biconditional. By Theorem 2.1(d), one way 10 do this—in fact, the most natural and common way—is fo prove two conditional2.2 Conditionals and B 33 statements: a forward direction, and a reverse (or converse) direction. Now, how should wwe try o prove a conditional statement? Well, a conditional statement has the form “If P. then Q.” That i, if P is true, Q is supposed to be true too. Therefore, the logical way 10 prove such a statement is to assume that P is true, and use this to derive the conclusion that Q is also true.] For the forward direction, assume that n is even. By definition of the word “even,” that means that mis ofthe form 2m, for some integer m. But from the equation n = 2m, ‘we can add 1 to both sides and obtain n + 1 = 2m + 1. Thus, n + 1 is odd (by the analogous definition of what it means to be odd]. Conversely, assume that n+ 1 is odd. That means n+ 1 is of the form 2m + 1, and by subtracting 1 from both sides of the equation n + 1 = 2m + 1, we obtain n = 2m. So nis even. (Biconditional (a) is now proved because we have proved both directions of it (©) For the forward direction, assume that mis odd, So n = 2m+ 1, for some integer m. From this equation, we get n + 1 = 2m + 2= 2m + 1). Therefore, n + 1 is even, ‘because it equals 2 times an integer. The reverse direction i left for Exercise 11, = ‘The only nonrigorous feature of the previous proofs that it does not properly deal with quantifiers (see Exercise 2 of Section 4.3). The proof is straightforward because of the definition of the word “odd” it uses. If “odd?” is defined to mean “not even,” this theorem becomes somewhat harder to prove. Exercise 12 covers a slightly different approach to this result. Exercises 2.2 (2) Consider a conditional statement P —* Q. Write the following symbolic statements. (Whenever you obtain two consecutive negation symbols, delete them). (a) The converse of the converse of the original statement (b) The contrapositive of the contrapositive of the original statement (©) The inverse of the contrapositive of the original statement (2) Restate each of the following statements in the form of an implication (using the words “If... then ..."): (a) Whenever a function is differentiable, it's continuous. (b) Accontinuous function must be integrable. (©) Apprime number greater than 2 can’t be even. (@) Annonnegative number necessarily has a square root. (©) Being nonnegative is a necessary condition for a number to have a square root. (© Aone-to-one function has an inverse function. (3) Write the contrapositve of the following statements. (Replace any substatement of the form ~~ P with P.) (@) If John’s happy, Mary's happy. (©) If Mary's not happy, John's happy.u Chapter? Propositional Logle (©) John’s not happy only if Mary’s not happy. (@) Mary's lack of happiness is necessary for Jobn’s happiness. (4) Write each ofthe following conditionals and its converse in the indicated forms from Table 2.3. Some answers might be difficult to express in sensible English, but do your best. For instance, statement (a) in form 9 could be “Whenever I read a good book, T’m happy all day,” and its converse in that form could be “Whenever I'm happy all day, {I must be reading a good book. (2) Reading 2 good book is sufficient to keep me happy all day. (Forms 3,5 and 7) (©) L will pay you if you apologize. (Forms 1, 3, and 5) (©) It’s necessary to give a baby nourishing food in order for it to grow up healthy. (Forms 2, 6, and 8) (5). Write each of the following biconditionals in the indicated forms from Table 2.4, Some answers might be difficult to express in sensible English, but do your best. (a) A triangle is isosceles if and only if it has two equal angles. (Forms 2 and 3) (b) Ml go fora hike today just in case I finish my paper this moming. (Forms Land 4) (©) The Axiom of Choice is equivalent to Zom’s lemma. (Foms 1 and 5) (@) Being rich is a necessary and sufficient condition to be allowed in that country club. (Forms 4 and 5) (6) Restate each of the following statements in the form of a conditional (with the words “If... then .."), a biconditional, or the negation of a conditional. If you think there’s more than one reasonable interpretation fora statement, you may give more than one answer. (@) Stop that right now or I'll call the police. (b) Ifyou clean your room, you can watch TV; otherwise you can’t, (©) You can’t have your cake and eat it too. (@) Thanksgiving must fall on a Thursday. (€) Youcan't get what you want unless you ask for it (6) This dog is fat but not lazy. (g) Am integer is odd or even, but not both. (h) In order to become president, it's necessary to have a good publicity firm, (@ A person can become a professional tennis player only by hard work. {_Lwon’t pay you if you don’t apologize. (©) Math professors aren’t boring. (1) Give an example of each of the following if possible: (2) A true (that is, necessarily tre) conditional statement whose converse is false (that is, not necessarily true) (b) ‘A false conditional statement whose contrapositive is true (©) A false conditional statement whose inverse is true (@) A false conditional statement whose converse is false23. Propositional Consequence; Introduction to Proofs 35 (8) Classify each of the following conditionals as necessarily true, necessarily false, or sometimes true and sometimes false (depending on which number or which person is being referred to). Also, do the same for the converse of each statement. Explain. (@) Ifice is cold, then 2+2=3, (b) Ifa number is divisible by 2, it’s divisible by 6. (©) Ifa person lives in Europe, then he or she lives in France. (d)_ Ifa person lives in Europe, then he or she lives in Brazil. (e) Ix> 0, then x>00or2+2 *(f) Ifx>0, thenx>Oand2+2=3, (9) Construct a truth table that you think best captures of the meaning of “p unless Q.” There may be more than one reasonable way to do this. To help you, you might want to consider a couple of specific examples, like “You can go swimming tomorrow unless you have a temperature.” Do you think that the word “unless” usually hnas the same meaning asthe exclusive or? (10) Prove Theorem 2.1, in the manner indicated in the text. (11) Prove the converse of part (b) of the theorem in Proof Preview 1. (12) Proof Preview 1 uses the definition that a number is odd iff itis of the form 2m + 1. I's just as correct to say that a number is odd iff i is of the form 2m ~ 1, Prove the same result, using this alternate definition, (13) Prove the following, in the manner of Proof Preview 1. Hint: You will need to use four variables, not just two, in each of these proofs. (a) The sum of two even numbers must be even. (b) The sum of two odd numbers must be even. (©) The product of two odd numbers must be odd. (14) By experimentation, fill in each blank with a number that you believe yields a correct conjecture, Then prove the conjecture, in the manner of Proof Preview 1. (@) [fn is or____maore than a multiple of 10, then n?is 1 Jess than a multiple of 10. (©) Ifnis of ___ more than a multiple of 6 then there is no number m such that mn is i more than a multiple of 6 23. Propositional Consequence; Introduction to Proofs In Section 2.1 we defined the concepts of tautology and propositional equivalence. Now that we have discussed the various connectives individually, it's time to examine these concepts in more detail. Why are these notions important? Recall that a tautology is a statement that is always true because ofthe relationship or pattem of its connectives. Also recall that it's very easy to tell whether a given statement isa tautology; all that's required isa truth36 Chapter? Propositional Logie table, In other words, tautologies are absolute truths that are easily identifiable. So there is almost universal agreement thet all tautologies can be considered axioms in mathematical work. ‘As far as propositional equivalence is concemed, we have mentioned that if two statements are equivalent, they are essentially two different ways of saying the same thing. If that’s so, we should expect equivalent statements to be interchangeable; and in fact one simple but important tool in proofs is to replace one statement with another equivalent one. ‘Table 2.5 shows some of the more common and useful tautologies. Its certainly not a complete list. In fact there’s no such thing: there are an infinite number of tautotogies. At the same time, it’s important to realize that even Table 2.5 shows an infinite set of tautologies, in a certain sense; remember that our propositional variables can stand for any statement. So a single tautology like the law of the excluded middle ‘actually comprises an infinite numberof statements, including purely symbolic ones like (Q+~R)V~(Q-+~R), mathematical ones like “x + y=3 orx+y # 3,” and English ‘ones like “Either I'l finish or I won't.” To what extent should you know this list? Well, if there were only thirty tautologies in existence, it might be worthwhile to memorize them. But since there are tn infinite number of them, there's not much reason to memorize some finite list. It might be fruitful for you to go through Table 2.5 and try to see (without truth tables, as much as possible) why all the statements in it are tautologies. This would be one way to become familiar with these tautologies for future reference. Some of the statements, in Table 2.5, such as the law of the excluded middle and the law of double negation, are very simple to understand. Others, like numbers 26 and 27, are somewhat more ‘complex, and it might take some thought to realize that they are tautologies. ‘Notice the groupings of the entries in Table 2.5. Most useful tautologies are either implications or equivalences. Remember that an implication is a one-way street that says that ifthe left side is tre, the right side must also be. The usefulness of implications in proofs is based on this fact. For example, tautology number 3 seems to indicate that if wwe have proved a statement P /\ Q, we should then be allowed to assert the individual statement P, We will see that this type of reasoning is certainly allowed in proofs, (By the way, note that several of the tautologies in Table 2.5 are labeled “Basis for ...” ‘These tautologies are used to justify specific proof methods discussed in Chapter 4.) Equivalences are two-way streets asserting that if either side is true, the other must be. So the standard way that equivalences are used in proofs is to replace either side with the other. De Morgan's laws are particularly useful. For example, if you want to prove that a disjunction is false, tautology 18 says that you can do this by proving both the disjuncts false. Also, tautology 19 provides the most useful way of proving that a conditional statement is false. In general, knowing how to rewrite or simplify the negation of a statement is a very important skill (see Exercise 2). In Section 2.1 it was mentioned that it's not necessary to have five connectives. More precisely, there's quite a bit of redundancy among the standard connectives. For example, tautologies 20 and 22 provide ways of rewriting conditionals and biconditionals in terms of the other three connectives. Also, more equivalences of this sort can be obtained by negating both sides of tautologies 17 through 19. For example,23 Propositional Consequence; Introduction to Proots 7 Table 2.5 Some of the more useful tautologies () PV~P Law of the excluded middle Q) ~@A~P) ‘Law of noncontradiction ‘Some implications @) @AQ+P Basis for simp @) PAQ—$Q Basis for simplification () P+ VQ Basis for addition © Q7eVO Basis for addition 7) Q4+@+Q) (8) ~P + +Q) (9) PAP +Q]+Q Basis for modus ponens (10) [~QAP+Q)>~P Basis for modus tollens (QL) -PAPVQ)7Q (12) P>[Q> PAQ] 3) [PA QAQ>RI7 CR) ‘Transitivity of implication (4) PQ (PVR) QV) (15) PQ) (PAR) + QAR) 16) [PH QAQHRIAE HR Transitivity of equivalence Equivalences for rewriting negations (17) ~PAQ) + ~PV~Q De Morgan's law (18) ~@VQ) + ~PA~Q De Morgan's law (19) ~P+Q) + PA~Q Equivalences for replacing connectives (20) (PQ) + (~PVQ) (21) P+ Q) + [PF QAQ>P)] (22) (PQ) + [(PAQVGPA~Q) Other equivalences (3) ~~P + P Law of double negation 24) P+Q) > ©Q—~P) Law of contraposition Q5) (© F QAP) ~ (P> QAR) (26) (PR) AQ—>R)] + [(PV.Q)—R] Basis for proof by cases 27) [P+ (QR) + (PAQ>R) 28) [P+ (QA~Q)] « ~P Basis for indirect proof 29) [PAQVR)] © [(PAQVPAR)] _ Distributive law G0) [PVQAR)] + (PVQAPVR)] _ Distributive law38 (Chapter? Propositional Logie from the first De Morgan’s law we can construct the related equivalence (P AQ) ++ ~(-PV~Q). In other words, any conjunction can be rewritten in terms of negation and disjunction. In general, knowing when and how to rewrite a connective in terms of specific other ones is a very valuable skill in mathematics. Itis also often very useful to rewrite the negation of a given statement; tautologies 17-19 show how this is done. Exercises 11 through 17 are concemed with rewriting connectives and reducing the number of connectives. For the remainder of this book, references to “tautology number ..." refer to Table 2.5. For convenient reference, Table 2.5 is repeated as Appendix 3 at the end of the book, To conclude this chapter, we discuss a method that can be used to analyze everyday, nontechnical arguments for logical correctness. This method is really a simple (but incomplete) framework for doing proofs, so studying it will provide a good preview of Chapter 4. Definitions: A statement Q is said to be a propositional consequence of statements P,, P,P, iff the single statement (P, AP, /\... \P,) + Q is a tautology, (Un this section, the word “propositional” may be dropped when discussing this notion.) ‘The assertion that a statement Q is a consequence of some list of statements is called an argument. The statements in thelist are called the premises or hypotheses or givens of the argument, and Q is called the conelusion of the argument. If Q really is a consequence of the list of statements, the argument is said to be valid. Recall that if a conditional isa tautology, then whenever the hypothesis of that conditional is true, the conclusion must also be true. So the significance of having a valid argument is that whenever the premises are true, the conclusion must be too. In the definition of propositional consequence, itis possible that n = 1. So Qisa propositional consequence of P if P -» Qis a tautology. With this in mind, note that two statements are equivalent if and only if each is a consequence of the other. Example 1: Determine whether each of the following arguments is valid: (a) Premises: PQ =R~Q oR Conclusion: ~ By the way, this sort of diagram is commonly used for logical arguments, especially ones in which the statements involved are purely symbolic, (b) Premises: If I’m right, you're wrong. If you're right, I'm wrong. Conclusion: Therefore, at least one of us is tight,23° Propositi Consequence: Introduction to Proofs 39 (©) IfAl shows up, Betty won't. If Al and Cathy show up, then so will Dave. Betty or Cathy (or both) will show up. But Al and Dave won't both show up. Therefore, Al won't show up. Solution: (a) To determine whether this argument is valid, we just need to test whether [(P —* Q) A (~R + ~ Q) A~R] ~~ P is a tautology. We leave it to you (Exercise 3) to verify that itis, so the argument is valid. () It's not absolutely required, but such arguments are usually easier to analyze if they are translated into symbolic form. So let P stand for “I'm right” and Q stand for “You're right.” Let’s also make the reasonable interpretation that “wrong” means “not right.” The argument then has the form Premises: P->~Q QP Conclusion: PV Q The conditional [(P —* ~ Q) \ (Q-~P)] -+ P V Q) is not a tautology (Exercise 3), s0 this argument is not valid. By the way, this is an argument that I actually heard used in a real-life situation. Can you explain why the argument fails? The simplest explanation involves the relationship between the two premises. (©) Asin part (b), let's introduce propositional variables: A for “Al will show up” and similarly B, C, and D, for Betty’s, Cathy's and Dave’s showing up. It tums out that [(A+~B) (AAC+ D)ABVC)A~(AAD)] > ~A isa tautology (Exercise 3), so this argument is valid. Since this argument involves four propositional variables, the truth table required 1 validate it contains sixteen lines, which makes it somewhat unwieldy and tedious to construct. So we now introduce a “nicer” method for validating such arguments: ‘Theorem 2.2: Suppose the statement R is a consequence of premises Py, Py.» » P,, and another statement Q is a consequence of P,, Py, 5 P, and R. Then Qis a consequence of ust P, Py,» Py Proof: Let P be an abbreviation for (P, PA. \P,). So we are told that P + R and (PAR) + Q are both tautologies. Now consider what the truth table of P -* Q must look like, In every row where P is true, R- must be too, since P —» R is always true. But since (P AR) — Q is also always true, this guarantees that in every row where P is true, Q must be true too. And remember that when P is false, P+ Q is true by definition. In other words, P —+ Q must be a tautology; tis is what we wanted to show. ™7 Chapter? Propositional Logie ‘The practical significance of this theorem i that you can use intermediate steps to show an argument is valid, In other words, if you want to show a statement is a ‘consequence of some premises, you don’t have to test whether the entire conditional is 1 tautology. Instead, if you prefer, you can begin listing statements that are obvious consequences of some or all of the premises. Each time you find such a statement you can use it as a new premise to find more consequences. This method can lead easily to ‘the desired conclusion. (Unfortunately, it also can lead you nowhere, even if the argument is valid) ‘We now give altemate solutions to Examples 1(a) and 1(c), using this method of intermediate steps. If you have any experience with formal proofs (from high school ‘geometry, for example), you will recognize the similarity. In fact, the derivations that follow are perfectly good mathematical proofs, and except for the need to include principles involving quantifiers, mathematical proofs could be based entirely on propositional consequence. ‘Alternate Solution: Our solutions consist ofa sequence of statements, numbered for easy reference, beginning with the premises and ending with the desired conclusion. ch statement in the derivation, after the premises, is a consequence of the previous lines, Since constructing truth tables isso straightforward, there’s no need to explain or justify the steps in these derivations any further. But to help you develop the habit of ‘good proof-writing, we explain each step. Formal solution to Example 1(a): MP4 Premise Q)~R>~ Premise @)-R Premise @ ~Q From steps 2 and 3, by tautology 9 6) ~P From steps 1 and 4, by tautology 10 Formal solution to Example 1(e): () Aa~B Premise Q) AAC) Dd Premise @ BYC Premise (4) ~(AAD) Premise (5) ~B+C From step 3, by tautology 20, essentially @ Arc From steps 1 and 5, by tautology 13 ()AF(AAC) — Fromstep 6 @) A>D From steps 7 and 2, by tautology 13 (0) A4~D From step 4, by tautology 19, essentially (10) A+(DA~D) _ From steps 8 and 9, by tautology 25 (i) ~A From step 10, by tautology 2823° Propositional Consequence; In duction to Proofs ri Which is the easier solution to this problem: the sixteen-line truth table or the derivation just given? It's hard to say, but there’s no doubt that the derivation is more informative and better practice for leaning how to do proofs, ‘On the other hand, neither a sixteen-line truth table nor an eleven-step formal proof is particularly readable. One of the main themes of Chapter 4 is that formal proofs, although having the advantage of encouraging thoroughness and correctness in proofwriting, are cumbersome to write and to read. Mathematicians almost always prefer to write less formal proofs that communicate an outline or synopsis of the full formal proof. With that in mind, here is an informal solution to Example 1(c). Exercise 6 asks you to do the same for Example (a). Informal Solution to Example 1(c): We are given that Al and Dave won't both show up. Therefore, if Al shows up, Dave won’t (using tautology 19). ‘Now, let's assume Al shows up. Then we are told that Betty will not show up. But we also know that Betty or Cathy will show up. Therefore, Cathy must show up. But that means Al and Cathy show up, and we are told that if they both show up, then Dave must show up, So we have shown that if Al shows up, then Dave shows up. Putting both previous paragraphs together, we have shown that if Al shows up, then Dave will show up and Dave won't show up. That is, if Al shows up, something impossible occurs. Therefore, Al cannot show up (tautology 28). We close this chapter with two more proof previews. These are also written in an informal style but would not be difficult to tum into formal proofs. Each of them is based on one or two key tautologies from Table 2.5. Proof Preview 2 Theorem: Given sets 4, B, and C, if A c B and Bc C, then A « C. (The symbol c is read “is a subset of.” This notion is defined and discussed in Section 5.2, but we need 10 use its definition here to carry out this proof] Prooft [As in Proof Preview I at the end of Section 2.2, we are asked to prove a conditional statement. So, once again, we begin our proof by making an assumption. In the terminology ofthis section, we could say that Ac B and Bc Care the premises of this proof] Assume that A < B and Bc C. By the definition of c, this means that for any object x,x € A implies x € B, and x € B implies x € C. Therefore, x € A implies x= C because, by tautology 13, this later conditional statement is a consequence of the two in the previous sentence]. And this is exactly what Ac C means. ™ As with Proof Preview 1, this proof glosses over some points involving quantifiers (see Exercise 1 of Section 4.3) Proof Preview 3 ‘Theorem: For any real number x, [x| =x. Proof: Let the propositional variables Q, R, and P stand for x > 0, x <0, and |x| > x respectively. [Mathematicians would rarely introduce explicit propositional42 variables in this manner, but it can’t hurt to do so.] We know that x must be positive, zero, of negative; that is, we know QV R. Ifx > 0, we know that |x| =x (by definition of absolute value), which implies [x| 2 x. In other words, Q implies P. On the other hand, if x <0, then |x| > 0 > x, so we still can conclude |x| > x. In other words, R implies P. So we have shown that Q implies P, and R implies P. By tautology 26, we can conclude the equivalent statement (Q or R) implies P. But since we also know (Qo R), we obtain (by tautology 9) P; that is, [x| 2 x. a ‘The argument in Proof Preview 3 is a proof by cases, as we see in Section 4.2. Exercises 23 (1) Replace each of the following statements by an equivalent statement that is as short as possible (in number of symbols). In some cases, the answer may be the given statement. (@) PAP &) ~@>-Q (© QQ?) @ P+~P () PAQVEAR) @ PvOVR (@) P+Q+(Q>P) () P4Q>~P) (2) For each ofthe following statements, express its negation in as short and simple away as possible. You will probably want to use tautologies number 17 through 19 (and possibly others) from Table 2.5. (a) This function is continuous but not increasing. (b) Pigs are not biue or dogs are not green. (©) Ifx is positive, then x is positive. (@ Pigs are blue if and only if dogs are not green. (©) If set Ais finite, then set B is finite and not empty. (3) Construct the truth tables necessary to test the validity of the three arguments in Example 1. (4) Test each of the following arguments for validity, by directly applying the definition of propositional consequence. In other words, construct just one truth table for each argument, (a) Premises: P+ Q, P—+~R, Q “+R. Conclusion: ~ P. (b) Premises: P VQ“ ~PAR, RP. Conclusion: ~ (P VQWR). (o) Premises: PV Q, QVR+*~P. Conclusion: RV~Q (@ If Alice is wrong, then Bill is wrong, If Bill is wrong, then Connie is ‘wrong. Connie is wrong. Therefore, Alice is wrong.23. Propositional Consequence; Introduction to Prools a (©) Ifturtles can sing, then artichokes can fly. If artichokes can fly, then turtles can sing and dogs can't play chess. Dogs can play chess if and only if turtles can sing, ‘Therefore, turtles can’t sing (5)_ Show that each of the following arguments is valid, using the method employed in the alternate solutions given previously. Do not use any tautologies with more than, three propositional variables. Consult your instructor about whether to write formal or informal solutions. (@) Premises: QR, RVS—P,QVS . Conclusion: P. *(b) Premises: P+ (Q“+~R), PV~S,RS,~Q—~R. Conclusion: ~R. (©) Premises: Babies are illogical. A person who can manage a crocodile is not despised. logical persons are not despised. Therefore, babies cannot manage crocodiles. (This example was created by Lewis Carroll.) '*(4) If Toversleep, I will miss the bus. If T miss the bus, Il be late for work unless Sue gives me a ride. If Sue’s car is not working, she won’t give me a ride. If I'm late for work, I'll lose my job unless the boss is away. Sue’s car is not working. The boss is not away. Therefore, if I oversleep, I'll lose my job. (© Tum the formal altemate solution to Example 1(a) into an informal proof, similar to that given for Example 1(¢) (7) Two sets A and B are defined to be equal if they have exactly the same members, that is, ifx € 4 is equivalent to x € B, for any object x. Prove that A= B ifand only if (4 < B and Bc A).You may want to refer to Proof Preview 2 in this section, as, well as Proof Preview 1 in Section 2.2, to review how biconditionals are normally proved. But don't make this proof harder than it needs to be; there really isn’t much to it, (8) Prove that for any real number x, |x| 2 ~% (9) Prove that ifn is an integer, then n? + n must be even. Hint: You may assume that an integer must be even or odd, Then use the technique used in Proof Preview 3. (10) Prove that ifm is an integer which is not a multiple of 3, then n? is 1 more than a multiple of 3. Hint: To do this, you need to find a disjunction that is equivalent to the condition that n is not a multiple of 3. Do nor try to prove this equivalence; you may assume it, Exercises 11 through 17 are rather technical and are concemed with material that has not been directly discussed in the text “(11) A set of connectives is called complete if every truth function can be represented by it thats, given any truth function, there is a symbolic statement that uses only connectives in the set and has that truth function.44 CChapter2 Propositional Logie Show thatthe connectives /\,V, and ~ together form a complete set of connectives ‘Hint: First consider a trath function with exactly one Tin its final output column. Show that any such truth function can be represented by a conjunction of propositional variables and their negations. Then, any truth function at all can be represented by a disjunction of such conjunctions. The resulting statement is called the disjunctive normal form ofthe given ruth function. Don't try to make this a very rigorous proof. (12) Find the disjunctive normal form for each of the following statements: @ PQ &) ~PAQ) (© P#@Q>-R) @ ~PAQ>R) (13). Show that / and ~ together form a complete set of connectives. (14) Show that V and ~ together form a complete set of connectives. (15). Show that -* and ~ form a complete set of connectives. *(16) Show that A, V,~¥, and ++ do not form a complete set of connectives. (17) Define a connective |, called the Sheffer stroke, based on the words “not both.” “That is, P[Q is tre except when both P and Q are true. Show that the single connective | forms a complete set of connectives. ‘Suggestions for Further Reading: For a more thorough treatment of mathe- ‘matical logic at a level that is not much higher than the level of this text, see Copi and Cohen (1997), Hamilton (1988), or Mendelson (1987). For a more advanced treatment, see Enderton (1972) or Shoentfield (1967).Chapter 3 Predicate Logic 3.41 The Language and Grammar of Mathematies Propositional logic is important in mathematies, but itis much too limited to capture the fall power of mathematical language or reasoning. For one thing, although propositional logic deals with connectives and how they are used to build up statements, it does not concem itself with the structure of atomic statements, Remember that we call a statement atomic if it is not built up from any shorter statements. The goal of this section is to examine what atomic statements look like in mathematical language. Example 1: One important category of atomic mathematical statements are ‘equations such as x +y=3. As discussed in Section 2.1, a statement of this sort is called 1a predicate, since its truth depends on the values of variables. It may also be called an ‘open statement. You can see that it contains no connectives. Quantifiers are words like “all,” “every,” and “some” or symbols standing for those words; so our equation contains none of those either. And that makes it atomic. It’s important to see why neither x + y nor y= 3 can be considered a substatement ofx +y=3. The expression x + y isn’t even a sentence; it has no verb. The expression y = 3 is a perfectly good sentence, but it makes no sense to say that the equation 3+ y=3is built up grammatically from the equation y= 3, So this equation, and in fact ‘any equation, is atomic. In many branches of mathematics, equations and inequalities account for virtually all the atomic statements, In the equation we've been using as an example, the letters x and y are, of course, variables. Definitions: A mathematical variable is a symbol (or combination of symbols like x,) that stands for an unspecified number or other object. ‘The collection of objects from which any particular variable can take its values is called the domain or the universe of that variable. Variables with the same domain are said to be of the same sort. (It's generally assumed that the domain of a variable must be nonempty.) 4546 (Chapter 3. Predicate Logie You have undoubtedly been using variables to stand for numbers since junior high school, and you have probably also encountered variables representing functions, sets, points, vectors, and so on. These ate all mathematical variables. Example 2: Ifyou saw the equation f(x) = 3, you would probably read this as “f of x equals 3," because you recognize this as an example of function notation. You would probably also think of x as the only variable in this equation. But strictly speaking, this equation contains two variables: x, presumably standing for a number, and, “[Link] standing for a function. There is nothing that says what letters must be used to stand for what in ‘mathematics, but there are certain conventions or traditions that most people stick to avoid unnecessary confusion. In algebra and calculus, for example, the letters x,y, and 7 almost always stand for real numbers, whereas the letters fand g stand for functions. ‘The fact that almost everyone automatically interprets the equation f(x) = 3 in the same way shows how strong a cue is associated with certain letters. On the other hand, if someone wanted to let the letter Q represent an arbitrary triangle, it would be best to inform the reader of this unusual usage. In Chapter 2 we introduced the idea of a propositional variable—a letter used to stand for a statement. Propositional variables are not normally used in mathematics. ‘They are used primarily in the study of logic. EF The difference between propositional variables and mathematical variables is very important, and you should be careful not to confuse them. A. propositional variable always stands for a statement—spoken, written, mathematical, English, Swedish, or whatever—that could take on a value of true or false. A mathematical variable can stand for almost any type of quantity or object except a statement. Not every leter that stands for something in mathematies isa variable. Definition: A symbol (ora combination of symbols) that stands fora fixed number or other object is called a constant symbol or simply a constant. Example 3: The symbols stand e are constant symbols, not variables, since they stand for specific numbers, not unknown numbers. Constant symbols need not be letters: ‘numerals like 2, 73, and 5.3 are also constants. Starting with variables and constants, mathematicians use a variety of other symbols to build up mathematical expressions and statements. It is possible to describe the structure of mathematical language in great detail. Rather than do that, let's just make one vital point. We've already mentioned that equations and inequalities are two very common types of mathematical statements. Expressions like x + y and cos 32, on the other hand, are not statements at all because they take on numerical values, not truth values, when we substitute numbers for the mathematical variables in them. We call this, kind of mathematical expression, which represents a mathematical value ot object, a3A The Language and Grammar of Mathematics 47 term, (Throughout this book, our use of the word “term” is more general than its usual ‘meaning in high school algebra.) The simplest kind of term is a single variable or constant, ‘The distinction between statements and terms can be made more clear by drawing an analogy to English grammar. One of the first things taught in grammar is that a sentence must have a verb, This is just as true in mathematics as it is in English, The ‘word “equals” is a verb, and the word group “is less than” includes the verb “is” and functions as a verb. So if we say that one quantity equals another or is less than another, ‘we have a complete sentence or statement. Therefore, = and < should be regarded as ‘mathematical verbs that can be used to create symbotic statements. The technical name for such verb symbols is predicate symbols. In contrast, the word “plus” is not a verb and so cannot be used to form a statement. Since x + y stands for an object (specifically, ‘a number), it’s essentially a mathematical noun, I's no more a complete statement than the phrase “frogs and toads” is a complete English sentence. The technical name for ‘mathematical symbols like +, -, and v , which are used to form terms that denote objects, is function symbols or operator symbols. Example 4: Let's consider what could be the elements of a symbolic language for high school algebra. There would have to be at least two sorts of variables: real variables, that is, variables whose domain is the set of all real numbers, and function variables, that is, variables whose domain is the set of all real-valued functions. It might also be convenient to have variables whose domain isthe set of all integers. In addition, it is normal to have an infinite number of constant symbols (including numerals) representing particular real numbers. ‘The most basic operator symbols of algebra are the symbols +, -, x, and /. The minus sign can be used syntactically in two different ways: it can be put in front of a single term to make a new term, or it can be put between two terms to make anew term, Technically, there should be two different symbols for these two different operations, but it s standard to use the same one, Some other important operator symbols of algebra are the absolute value and radical symbols. Exponentiation represents a rather special case in the grammar of algebra. An expression like x” is certainly a term, built up from two simpler terms. But instead of using a symbol to show exponentiation, we show it by writing the second term to the ‘upper right of the first term. It would perhaps be better to have a specific symbol for exponentiation, but traditionally there isn’t one. However, note that most calculators and computer languages do have a specific key or symbol for exponentiation For more advanced work, one might want many other operator symbols, for things like logarithms, function inverses and compositions, trigonometric functions, and so on. It is much easier to list all the predicate symbols of algebra than all the operator symbols. The only atomic predicate symbols are = , <, and > . There are two other standard inequality symbols, < and >, but they are not atomic (their meaning includes an “ot”), Also since x> y means the same thing as y 0), the variable x is bound, so the statement is closed, In the statement Vx Jy (x - y = 2z),x and y are bound whereas z is free. So this satement is open; itis a propositional function of z Example 4: Strictly speaking, it's “legal” for the same variable to occur both ‘ound and free in the same statement. Consider x= y V 3x 2x=2). Then x is free in the first disjunct and bound in the second, But most people consider it very awkward and confusing to have the same variable bound and free in a single statement. Furthermore, this awkwardness can always be avoided, because a bound variable can be replaced by any new variable of the same sort, without changing the meaning of the statement, In the above example, the rewritten statement x= y V 3u (2u=2) would be more readable and would have the same meaning as the original, as long as uw and x have the same domain. Convention: This text follows the convention that the same variable should not ‘occur both bound and free in the same statement. You should, too. FF [tis important to develop an understanding of the difference between free and bound variables. A free variable represents a genuine unknown quantity—one ‘whose value you probably need to know to tell whether the statement is true or false. For example, given a simple statement like "S += 3,” you can’t determine whether it's true or false until you know the value of the free variable x. But a bound variable is quantified; this means that the statement is not talking about a single value of that variable, If you are asked whether the statement “3.x (5 + x= 3)” is true, it wouldn't ‘make sense to ask what the value of x is; instead, it would make sense to ask what the domain of xis. (Ifthe domain were all real numbers, the statement would be true; but if it were just the set of all positive numbers, the statement would be false.) In this way, ‘a bound variable is similar to a dummy variable, like the variable inside a definite integral: it doesn’t represent a particular unknown value. Notation: If is any propositional variable, itis permissible and often helpful to the reader to show some or all of its free (unquantified) mathematical variables in parentheses, So the notation P(x) (read “P of x") would imply that the variable xis free in P, whereas the notation P(x, y) would imply that both x and y are free in P. Some mathematicians follow the convention that all the free variables of a statement must be shown in parentheses in this manner, but we don’t. So, for example, when we write P(x), there could be other free variables besides x in P. ‘You may notice that this notation strongly resembles function notation f(x). The resemblance is deliberate. An open sentence does define a function of its free variables,32 Quantiiers st namely a truth-valued function. This is why open sentences are also called “propositional functions.” (On the other hand, it’s important to distinguish between an open sentence and a mathematical function; the later is a mathematical object, not a statement.) “Another way that this new notation is similar to ordinary function notation involves substituting or “plugging in” for free variables. Suppose we introduce the notation P(x) for some statement. If we then write P(y) or P(2) or P(sin 3u), this means thet the term in parentheses is substituted for the free variable x throughout the statement P. Enough technicalities for now. It's time to talk about the meaning of the quantifiers and then look at some examples of how to use quantifier logic to represent English ‘words and statements symbolically. Definition: A statement ofthe form Vx P(x) is defined to be true provided P(x) is true for each particular value of x from its domain, Similarly, 3 P() is defined to be ‘rue provided P(x) is true for atleast one value of x from that domain, Perhaps you object to these definitions on the grounds that they are circular or just don’t say anything very useful. In a sense, this objection is valid, but there is no simpler ‘method (such as truth tables) to define or determine the truth of quantified statements. Note that this definition of the existential quantifier gives it the meaning of “there is atleast one.” There are also situations in which you want to say things like “There is exactly one real number such that ... .” It would be possible to introduce a third quantifier corresponding to these words, but it's not needed. Section 3.4 explains why. Also note that our interpretation of 3 is analogous to our interpretation of V as the inclusive or, since that connective means at least one disjunct is true, rather than exactly one disjunct is true. It is reasonable and often helpful to think of the existential and universal quantifiers as being closely related to disjunction and conjunction, respectively. Section 2.1 ended with a few examples of how to translate English statements into symbolic statements of propositional logic. When quantifiers are involved, these translations can be somewhat tricky to do correctly, but every mathematician needs to eam this skill. As in the earlier examples, the first step in these translations is to determine the atomic substatements of the given statement and then to assign a propositional variable to each of them. But when quantifiers are involved, it also ‘becomes very important to identify and show the free mathematical variables present. ‘This process is much easier if you remember some of the grammatical issues we've talked about: propositional variables stand for whole statements, each of which must contain a verb, The free mathematical variables of a given propositional variable should correspond to nouns or pronouns that appear in that statement. For instance, if you ‘wanted to symbolize a statement that talked about people liking each other, it would be reasonable to use a propositional variable L(x, y) to stand for the sentence “x likes y,” where it is understood that x and y represent people. The verb “likes” involves two nouns, so there are two free variables.2 Chapter3. Predicate Logie ES The following rule of thumb is also helpful: The symbolic translation of a statement must have the same free variables as the original statement. Example S: For each of the following, write a completely symbolic statement of predicate logic that captures its meaning, (@) All gorillas are mammals. (b) Some lawyers are reasonable. (©) No artichokes are blue. (@) Everybody has a father and a mother. (©) Some teachers are never satisfied. (B) (The number) x has a cube root. (g) For any integer greater than 1, there’s a prime number strictly between it and its double. Solution: (2) Certainly, the word “all” indicates a universal quantifier. But if you have never done such problems before, it might not be clear to you how to proceed, The key isto realize that what this proposition says is that if something is a gorilla, it must bea mammal, So within the universal quantifier, what we have is an implication. The logical structure of the statement is therefore ‘We (visa gorilla + x is a mammal) Of course, this is not a completely symbolic rendition of the original statement. If we want to make it completely symbolic, we have to introduce propositional variables for the atomic substatements. Let G(x) mean “x isa gorilla” and let M(x) mean “x is a mammal.” Then the original statement can be represented symbolically as vie (G@) + MQ). ‘We have not specified the domain of the variable x inthis solution. This is because wwe don't want any particular limitations on it. Since the implication inside the quantifier limits things to gorillas anyway, we might as well assume x ean stand for any thing whatsoever, or perhaps any animal. I's not uncommon to use a Variable whose domain might as well be unlimited, ‘Note that the given English statement has no free variables, and therefore neither does its symbolic translation. This is true forall the part of this example except part (f) Perhaps you see a shorter way of translating the given statement into symbols. Why not specify that the variable x stands for any gorilla, as opposed to a larger set like all animals? Then it appears that the given statement can be represented as Yr (cisa mammal) or vx M(x) ‘There is nothing wrong with this approach to the problem, and it does yield a shorter, simpler-looking answer. However, it's not necessarily helpful in mathematics to introduce variables with any old domain that’s considered convenient at the time.32 Quantiters 33 ‘Therefore, you should know the long way of doing this problem and especially that this, type of wording translates into an implication. (b) This time, because of the word “some,” the solution requires an existential quantifier, Notice that, except for replacing the word “all” by the word “some,” the structure of this statement seems the same as the structure ofthe previous statement. So ‘you might automatically think that an implication is involved here too. But if you give it some thought, you'll realize that this statement says that there is a person Who is @ lawyer and is reasonable. So it’s a conjunction, not an implication. With propositional variables L(x) and R(x) standing for “x is a lawyer” and “x is reasonable,” the correct symbolic translation is 3x (L(x) ARG). The same shortcut that was mentioned in part (a)—using a more specific variable—could also be applied to this problem. FS Pay close attention to the contrast between parts (a) and (b). Again, the deceptive thing is that the words seem to indicate that the only logical difference between the two isthe quantifier. Yet the “hidden connective” turns out to be different too. In general, the words “All. are ...” always represent an implication, whereas “Some ..s are.” always translates to @ conjunction, (©) Here we encounter the word “no,” which would seem to indicate a negation, perhaps combined with a quantifier. At first thought, it might seem that “No artichokes, are blue” is the negation of “All artichokes are blue.” But remember that the negation of a statement means that the statement is not true. And “No artichokes are blue” surely does not mean “It’s not true that all artichokes are blue.” Rather, it means “It’s not true that some artichokes are blue.” So one way to symbolize this statement is to first symbolize “Some artichokes are blue,” as in part (b), and then to stick ~ in front of it. ‘Another correct approach, perhaps less obvious, isto realize that the given statement ‘means the same thing as “All artichokes are nonblue” and to go from there. The details are lef for Exercise 2. This example illustrates some of the subtleties and ambiguities of English. “No artichokes are blue” definitely has a different meaning from "Not all artichokes are blue.” How about “All artichokes are not blue”? Do you think the meaning of this is clear, oris it ambiguous? (@ We can see that “everybody” means “every person.” So the symbolic form of this statement should begin with a universal quantifier, and it is convenient to use a variable whose domain is the set of all people. If we then write M(x) and F(x) to represent, respectively, “x has a mother” and “x has a father,” we can translate the given statement as vx (MG) AF(X)) ‘This solution isn’t wrong, but it can be eriticized as incomplete. A statement like “x has a mother” should not be considered atomic, because it contains a hidden54 Chapter3. Predicate Logie quantifier. That is, it really means “There is somebody who is x's mother.” So a better representation of the statement is obtained as follows: Let M(x, y) mean ‘y is x's mother” and F(x, y) mean “y is x’s father.” Then the statement can be symbolized as, vx Gy M(x») A 32 F(x 2)) where x, and are people variables. Note that there is a variable for each person under consideration—person x, mother y, and father z. But they ar all bound variables. (©) As before, letxbe a variable whose domain is the set of all people. Recall from part (b) that “Some teachers are .." should be thought of as “There exists someone wino is a teacher and who is...” But how do we say someone is never satisfied? This means that there is no time at which the person is satisfied. So we also need a variable ¢ whose domain is the set of all possible times. Let's define T(x) to mean “x is a teacher” and (4, #) to mean “x is satisfied at time ¢.” With this notation, the given statement can be represented as 3x (T(x) A~3e S(x, 9) (®) Inpart (e) we saw that words like “has a cube root” include a hidden quantifier. To say that a number has a cube root is to say that there is a number whose cube is the given number. So what we want is, By Oyy=x) or 3y(y?=2x) or 3y(y= Ve) ‘Note that, in all of these solutions as well as in the original, x is free whereas y is not. (g) Let m and n be variables whose domain is the set of all natural numbers (the positive integers 1, 2, 3, and so on). Then if we write P(n) for “n is a prime number,” we can translate the given statement as Ym [m > 1+ 3n (m OAy<0) (© VxPQ) 3x @ ~vr~Wy~ 2 2+2=0) (©) Ye [P@) > 32 QE) > vy RE, y))] 2) Write out both of the symbolic answers described in the solution to Example 5. (3) Translate each of the following into purely symbolic form. For the sake of uniformity, use the variables x, y, and z to stand for real numbers, and m,n, and k for integers. Initially, you may use only equations and inequalities as atomic statements. For instance, to express “n is a multiple of 10” symbolically, you could write “3m (n= 10m).” Then you can introduce new propositional variables as abbreviations for statements that you have written in symbolic form. For exemple, affer you do part (@, you can define a propositional variable, pethaps P(n), to stand for your answer to part (4) when you do parts (e) and (1). (a) 1s the smallest positive integer. (b) There is no largest integer. (©) mis an odd number. *(@) nis prime number. (e) Every prime number except 2 is odd. (8) There are an infinite number of prime numbers. Hints There’s no simple way to express this literally. Instead, say that there's no largest prime number. *(g) For any nonwhole real number x, there’s an integer strictly between x and x+ 1. Hint: The difficult part of this problem is that you may not use a variable whose domain is precisely the set of nonwhole real numbers. How can you express symbolically that x is not a whole number? (b) Between any two (different) real numbers there’s another one. (4) (@) Which of the statements in Exercise 3 are closed? (b) Name at least three of these closed statements that are true. (S) Asbefore, in the following statements, x,y, and z denote real numbers, and m, nad k denote integers. For each statement, frst identify its free variable(s); then find one set of values for its free variabl(s) that makes the statement true and one set that rakes the statement false. (Example: the statement 3 (m= 2) has only m asa free variable. The statement is true for m= 9, and false for m= 7.) Justify your answers (a) Sn(n>S Am? +k? =n?) () Vey @ y2) (©) Veapbe=yAyz=2A(ee Oy #x)] (@) Ve(et~ x2 m)56 Chapter Prediate Lope (© The following symbolic statements are true in the real number system. Rewrite each of them in reasonable-sounding English. (a) vx [x20 3y 0’ (b) Vebes 0-4~39(y © 2 Won” (@) Va,b [a + 0 3x (ax +b=0)) (1), Represent each ofthe following statements symbolically, starting with only the following atomic statements: P(x, y) for “xis a parent of y,” W(x) for “x is female,” and x=y (meaning x and y are the same person). All your Variables should have the set of all people as their domain, As in Exercise 3, you may introduce new propositional variables for statements that you have already written symbolically. Remember that it is OK to substitute for the free variables of a statement. For example, W(2) would mean that zis female, (@) xismale, (b) xisy's father. (©) xisy’s grandmother. (@) xis)’s sibling, (This means that x and y have the same mother and father, but they are not the same person.) (©) xisan only child, (That is,x has no siblings). (f) xisy’s first cousin, (g) x has no uncles. (bh) Some people have brothers but no sisters. (8) For each of the following statements, introduce a propositional variable (with free variables indicated) for each of its atomic substatements, and then write a totally symbolic translation of the given statement. You can define variables with any domain ‘you want. For instance, for part (a), you might let one of your propositional variables ‘be S(x), meaning “x likes spinach” (where x can be any person). (@) Not everyone likes spinach, and no one likes asparagus. (b) All crows are black, but not all black things are crows. (©) If someone kisses the frog, everyone will benefit. (@) There are people who like all vegetables. (©) It’s possible to fool all of the people some of the time and some of the people all of the time, but not all of the people all ofthe time. (8) everybody bothers me, I can't help anybody. (g) Anybody who bothers me won't be helped by me. (h) Every problem in this section is harder than every problem in Chapter 2 (i. No one is happy all the time, (Everybody loves somebody sometime. (It's not true in all cases that if one person likes another, the second likes the first. () There are days when everyone in my dorm cuts at least one class.33, Working with Quantifers 37 3.3 Working with Quantifiers In this section we examine some of the methods that mathematicians use to understand and simplify quantified statements. It was mentioned in Section 3.2 that quantifiers often occur in sequence. Usually, quantifiers of the same type (all 3s or all Vs) occurring in sequence are not difficult to understand or to work with, but alternations of quantifiers between 3 and V (in either order) can make statements confusing, In more advanced studies of the foundations of mathematics, the complexity of statements is measured by how many alternations of quantifiers they contain. (One well-known mathematical logician has expressed the opinion that three or four is the maximum number of alterations of quantifiers that the human brain can deal with.) Let’s begin this section by looking at sequences of quantifiers, paying particular attention to statements with a single alternation. Example 1: Let's assume that x and y are real variables and consider a simple atomic statement like x + = 0. One simple way to quantify this, with no alternations, is 3x Sy (x + y= 0). What does this quantified statement say, and is it true or false? ‘Technically, the statement says that thee isa value of x for which 5y (x += 0) is tue. But there's no need to split up the quantifiers in this way. In Section 3.2 it was ‘mentioned that this statement can be written as 3x,y (x+ y= 0), which would be read “There existx and y such that x +y= 0.” The point is thatthe statement simply says that there is some choice of values for the two variables that makes the equation x + y= 0 hold. Clearly, this is true; for example, we could take x= 3 and y=~3. A consequence of this analysis is that there is no difference in meaning between 31,y (r+y=0) and Byx(e+y=0). Example 2: Similarly, consider the statement Vx vy (x + = 0). Again, this can bbe rewritten as Vt,y (x +y'=0), with the practical consequence that the two quantifiers can be considered together. So this statement says thet for all choices of values for x and y, x+y =0. This is blatantly false; for example, it fails when x = y = 29. As in the previous paragraph, there is no difference in meaning between Vx, (x + y = 0) and ‘Wy,x (c+ = 0). This is a general fact: the order of the variables in a sequence of like ‘quantifiers is unimportant. Example 3: Now let's look at the more interesting cases with altemations of quantifiers. First, consider Vx Sy (x + y = 0). This says that for every value of x, the statement 3y (x+y = 0) holds. That is, for every choice ofa value for x, there must be a value for y that makes the equation hold. You can see that this is always so. When x= 7, y would be ~7; when x = ~2.68, y would be 2.68, and 50 on. Clearly, the correct ‘choice of y can be described in terms of x by the simple formula y= =x. This example also illustrates a general situation: fora statement of the form Vx 2y P(x, y) to be true, itmust be possible to choose y in terms of x (that is, a8 a function of x) so that the inner statement holds for all values of x when y is chosen according to that function.58 (Chapter 3. Predicate Logie Example 4: Now let's reverse the quantifiers and consider 3x (x +y=0). This says that there is a value for y that makes the statement Vx (x +y=0) hold. That is, there would have to exista single value of y, chosen independently of x, that makes the inner ‘equation work for all values of x. In ths situation, it’s not enough to define y in terms of x. You can see that there is no such value of y, and so the whole statement is false. ‘These examples illustrate several points. For one, they show that the order of quantifiers does matter when they are of opposite types. Also, in general, a statement of the form Sy Vx P(x, ») is harder to satisfy (that is, ess likely to be true) than the corresponding statement Vx 3y P(x, )). Additionally, the previous paragraph clarifies why the words “such that” are usually needed after an existential quantifier. If the statement Sy x (+= 0) were read “There is a y for every x...” it would seem to have the same meaning as “For every x there is ay...” which it doesn’t. The wording “There isa y such that, for every x, ..” helps reinforce the difference in meaning. ‘The next theorem generalizes the previous examples of how to decipher statements with altemating quantifiers. We omit the proof, since itis quite technical (but see Exercise 11). Theorem 3.1: Suppose a statement begins with a sequence of quantifiers, followed by some inner statement with no quantifiers. Then the statement is true provided each existentially quantified variable is definable as a function of some or all of the universally quantified variables to the left of it, in a way that makes the inner always true. (A function of no variables means a single, constant value, The a function of” in this theorem could be replaced by “in terms of.”) ‘We just saw how this theorem applies to statements of the form 'x Sy P(x, ») and Sy vx P(®,y). It can also help decipher statements with more alterations of quantifiers. Example $: Suppose we had to work with a monster like 3 Ww Sw Vx,y 3 (..) ur rule says that, to satisfy this, there must be a single value of w, a function defining 1w in terms of v, and a function defining z in terms of v, x, and y that guarantee that the inner statement is true. Knowing this probably won’t make the problem simple, but it cought to help. Proof Preview 4 Theorem: For any two real numbers, there is a real number greater than both of ‘them. Proof: In symbols, what we want to prove is Vx,y 32 (2>x A z>). By Theorem 341, to prove this is true, we must appropriately define 7 in terms of x and y. One concise way to do this is to let z = |x| + |y| + 1. We must then show that this makes the conjunction in parentheses true. (The rest ofthe proof uses numerous results from high school algebra, including basic properties ofthe absolute-value function, Most of these are proved in Appendix 2.] We know that |x| > x and |y| 2 0. Therefore |x| + |y| > x+0=x, and so [x| + |p|-+1>x. Similarly, [x] + |p| +1 >. (Mathematicians usually33° Working with Quant 59 omit part of a proof that is nearly identical to a previous part and instead make a comment like the previous sentence.] This completes the proof. Now let’s apply these ideas to determine the truth or falsity of various statements in various number systems. Example 6: For each statement, determine whether it’s true in each of these number systems: the set of all natural numbers (positive integers) N, the set of all integers Z, the sct of all real numbers R, and the set of all complex numbers C. @ Wey 3z@+2=y) @) 3rWe2) (c) Ifx. And even if this symbol is given its usual ‘meaning, the statement is false in a domain with a largest number, like the set of negative integers. (©) Even this isn’t a law of logic; interpreted! still depends on how the symbol < is (@ The definition says that the symbol = must be given its standard interpretation. “Therefore, this statement isa law of logic: if and y have the same value, and s0 do y and z, then clearly x and z must also. This statement is called the transitive property ‘of equality and is usually taken as an axiom of mathematics. By the way, this statement is nora tautology. (©) This statement says that ifa certain condition is true for all objects in a certain domain, it's true for at least one. Clearly, such an implication must always be true (see Exercise 1). So this is a law of logie (but not a tautology). In Chapter 4, with an axiom system at our disposal, we are able to solve more ‘complex problems of this type. In the meantime, you are welcome to peek ahead at Table 4.2, which lists some of the more useful laws of logic. Negations of Statements with Quantifiers We have just discussed at some length what has to happen in order for @ quantified statement to be true. We have not talked about what has to happen for a quantified statement to be false. It may not seem that this should require a separate treatment, but it does. Suppose that Pisa statement that begins with a sequence of quantifiers. We've said that P is true provided that certain functions and/or constants (corresponding to the existential quantifiers of P) exist. So we could say that P is false provided that not all these inetions andlor constants exist. However, often this view of the situation doesn’t help to figure out whether the statement is falsea Chapter 3. Predicate Loge ‘To say that Pis false is of course to say that ~ P is true, The statement ~ P begins ‘with a negation sign, followed by a sequence of quantifiers. It turns out to be useful to ‘be able to move the negation sign from outside the quantifiers (that is, in front of them) to inside the quantifiers. The key to doing this isthe following theorem, for which we just provide an informal, commonsense proof. ‘Theorem 3.2: For any statement P(x) (@ ~¥x P(x) is logically equivalent to 3x ~ P(2. (b) ~ 3x P(e) is logically equivalent to vx ~ P(x). Proof: (a) The statement ~ vx P(x) says that it’s not true that P(x) holds for every value of x in its domain, But this means that P(x) is false for at least one value of x, which is precisely what 5x ~ P(x) says. (©) This argument is similar and is left for Exercise 2, ‘Theorem 3.2 can be thought of as a direct parallel to De Morgan's laws. Recall that those tautologies say that you can distribute a negation into (or factor a negation out of) 1 conjunction or disjunction, but in doing so you have to change the inner connective from A to V, of vice versa. Similarly, Theorem 3.2 says you can move a negation across ‘a quantifier, in either direction, provided you reverse the quantifier from ¥ to 2, ot vice ‘versa. 1 like to call these quantifier equivalences De Morgan’s laws for quantifiers. Example 8: Simplify each of the following statements by moving negation signs inward as much as possible. (@) ~3xyve~ Su vw P (b) ~ Bx Ve [> 0+ 3d (d> 0A Yu (fe ui [fla) - FW)I OAVd (~d>0V Bu (lx - ul 0V ...as d>0—+ .... And if we are also emitted to use basic facts about the real number system, the statement ~ [fla) - flu)| 3t~ HG, 0). (6) It's not easy to put the solution to part (b) into smooth-sounding English, but the best try might be, “There are some people, all of whose friends are sometimes dishonest.” Perhaps you can do better than this. Of course, the original statement can6a Chapter 3 Predicate Logie easily be negated in words as “Not everybody has a friend who is always honest.” But that’s not what the problem asks us to do, Proof Preview 5 ‘Theorem: There is no smallest positive real number. Proof: “For convenience, let x and y be variables whose domain consists of all ‘positive real numbers. [This is perfectly legitimatel] In symbols, the statement that there is a smallest positive real number would be 2x Wy (x < »). So what we want to prove is ~ 3x vy & = 9). Now, by applying Theorem 3.2 to this, we can change it to Ve ay~(e 5»), ormore simply, vir Sy (y <2). To verify that this last statement is tue, we recall Theorem 3.1: we must define y as a function of x in such a way that the inequality ‘yP). . Br, , or >. Even though sets are not discussed in detail until Chapter 5, let’s introduce some abbreviations for a variable that is restricted to a set, since this notation is very similar to the notation just introduced, Notation: Let P be any statement, x any mathematical variable, and t any term that denotes a set. (So t could be a single letter standing for a set, or a more complicated ‘expression like A UB.) Then33. Working with Quanttiers 65 + WeetP is an abbreviation for Vx (xt P), . BxetP isan abbreviation for Ix(xetAP). Example 10: Write the following statements in symbolic form, using the abbreviations that have just been defined: (@) Every positive number has a positive cube root, and every negative number hhas a negative cube root. (b) For every nonnegative number x, there's an element of set B strictly between x and x+ 1. Solution: (a) vx>03y>0 =x) Avx<03y<0(* =x) (b) Wr205yeBQr) 52,08. Exercises 3.3 (1). What assumption must be made about the domain of the variable x for Example 7(@) to be correct? Has this assumption about domains been made in this chapter? (2) Prove Theorem 3.2(). @) Prove Theorem 3.3. Instead of doing this by mimicking the proof of Theorem 3.2, use the result ofthat theorem, the definitions of restricted quantifiers, and some tautologies to provide a more rigorous proof. (4) Write out each step of the transformation described in Example 8(b). (8) Determine whether each of the following statements is true or false if all variables have the set of real numbers as their domain. Explain briefly. @ ve R=) ©) Wy Sx@=y) (©) axwe+5=y) (@ ve vy 32 ¥u(etz=y+u) (©) ve vy 32 +y=2) (©) 3x[¥y Or =y) A~Vy Ox=y)]66 Chapter 3 Predicate Logie (©) Repeat Exercise 5 with all variables having the set of nonnegative integers as their domain. (7) Determine whether the following statements are laws of logic. Explain, (@) 3xP(o) > vx PG) () (Bx vy PC, y)] + [yy 3x PO »)] (©) (Wy 3x PG, »)] > [x Wy POs y)] @) [Bx PE) VQ) + [Bx P@) V 3x Q@)] (©) (3x PG) AQ@)] F [2x PG) A3x Q@)) (f) Va,yzu [r= yAz=u > (P(x, 2) POY) (8) Simplify each of the following statements by moving negation signs inward as much as possible. (2) ~vny 32 V~ vu Q) () ~C3xP > W~Q © ~ve~3y~¥z PA~Q) (9) Write each of the following statements in symbolic form, using the restricted quantifier notation introduced in this section. {a) Every number in set 4 has a positive square root, (©) Given any real number, there are integers bigger than it and integers smaller than it. (©) Every member of a member of 4 is a member of 4. (@) No positive number equals any negative number. (10) Prove the following. Your proofs can be based on the proof previews in this section, (@) For any real number x, there’s a number that is larger than both x and.x°, (©) Given any two unequal real numbers, there's a number between them. (©) There is no largest real number. (@) There is no largest negative real number. (11), Prove Theorem 3.1 for the special case of statements with only one existential ‘quantifier. Since we haven't studied functions yet, don’t expect to do this very rigorously. Just try to give a commonsense argument. 3.4. The Equality Relation; Uniqueness In Section 3.1, the idea of predicate symbols was introduced. Recall that these are symbols that act as mathematical verbs and are used to form atomic statements in predicate logic. Of course, different branches of mathematics require different predicate symbols, However, whatever differences may exist in the languages of different branches of mathematics, there is one predicate symbol common to all of them, and that is the equal sign. In other words, every branch of mathematics (as well as all of science and many other subjects) makes use of equations. Furthermore, the rules for working,34 The Equality Relation; Uniqueness 6 with equations do not change between different areas of mathematics and science, Because of this universality of the use of equations, the principles involved are usually included in the study of predicate logic. ‘You are familiar with equations and how to use them, and there will be no new tricks unleashed on you regarding them. In the next chapter, we begin using the axiom system contained in Appendix 1. But it won't hurt to take a look now at the standard axioms pertaining to equations, which are group III of the axioms, You can see that there are only four of them, and they are all very straightforward. The first one, reflexivity, says that anything equals itself. The second, symmetry, says that equations are reversible, (Symmetry is normally stated as a conditional, but it can be thought of as a biconditional. You might want to think about why this must be so.) The third, transitivity, says that “two things equal toa third thing are equal to each other.” Itis this axiom that allows you to write a long sequence of equations and then conclude that the first expression in the whole sequence equals the last one. The last axiom, substitution of equals, isa bit more involved. What it says is that if two things are equal, then they ate completely interchangeable. Its this axiom that also implies that it’s OK to change both sides of an equation, as long as the same thing is done to both sides. A more thorough discussion of these axioms and how to use them appears in Section 4.4. Uniqueness Recall thatthe existential quantifier has the meaning “there is at least one,” which ‘makes it analogous to the ‘“inclusive-or” meaning of the disjunction connective. In ‘mathematics we often want to say that there is exactly one mumber (or other object) satisfying a certain condition. In mathematics, the word “unique” is used to mean “exactly one.” Should we introduce a third quantifier with this meaning? There is nothing wrong with doing so, but it’s important to realize that this meaning can be captured with the symbols already defined, just as the exclusive or can be defined or ‘written in terms of the other connectives. ‘There are several different-looking but equivalent ways to say that there’s a unique object satisfying a certain condition, All these versions use the equality symbol; in fact, the desired meaning cannot be captured without it. For example, one way to express ‘uniqueness isto say “There's an object that satisfies the condition and that equals every object that satisfies the same condition.” Another way is “There's an object that satisfies the condition, and there are not two different ones satisfying it.” A third way, closely related tothe previous one, is “There’s an object that satisfies the condition, and if any two objects satisfy it, they must be equal.” Finally, a very concise way is “There's an object such that satisfying the condition is equivalent to being that object.” Let's state the content of this paragraph more formally. ‘Theorem 3.4: The following four statements are equivalent, for any statement P(x) and any mathematical variables x and y. (@) 3x (PQ) Avy (PO) > x=y)) (@) Ax P(x) A~ 3x,y (PQ) APY) Ax #y)6 (Chapter 3. Predicate Logle (©) 3x PG) Avx,y (PG) APO) x=y) (d) 3x vy (PQ) y=2) Proof: We give a relatively informal proof of this theorem that is still rigorous enough to illustrate several ofthe proof methods that are introduced in the next chapter. To prove that three or more statements are equivalent, the most common procedure is to prove a cycle of implications. So we show that statement (a) implies statement (b), then that (b) implies (c), that (c) implies (4), and finally that (d) implies (a). (implies (o): Assume that statement (a) is true. Since there exists an x satisfying the statement after the first quantifier, let's say (for “definiteness”) that f is an object satisfying it. Then P(X) is true; this implies thatthe first conjunct of statement (b) holds. ‘Also, for any x andy, if P(2) and P() both hold, then we know that x= kand y= k. By ‘ransitvity, this implies x= y. So there cannot be two different objects satisfying P, and that is what the second conjunct of (b) says. (©) implies (c): See Exercise 1. (©) implies (4): Assume statement (c) is true. Since 3x P(x) holds, let's say that k is an object satisfying P(k). We are done if we can show that, for this k, ¥y PQ) > y=). Consider any y. Ify= k, then P(y) holds, since we know P(R). Conversely, if P()) holds, then we have both P(A) and P(y), and so by the second part of statement (c), y= k So we have established that P()) ++ y=, and since this is for any y, we are done, (@) implies (a): See Exercise 1. Notation: For any statement P and any mathematical variable x, we write Six P, read “There is a unique x such that P,” to stand for any one of the equivalent statements of Theorem 3.4 It is important to keep in mind that this notation, like the restricted-quantifier notation defined in Section 3.3, is just an abbreviation for a longer form. In particular, when you want to prove that there is a unique object satisfying some condition, you ‘must prove one of the forms listed in Theorem 3.4, Form (c) tends to be the easiest to ‘work with. Proof Preview 6 ‘Theorem: Ifa and b are real numbers with a ¥ 0, then the equation ax += 0 has ‘unique solution, Proof: Assume that a and b are real numbers and a ¢ 0. We must show that the equation ax +b = 0 has a unique solution. (We work with form (c) as given in Theorem 3.4, Think of P(x) as the equation ax + b = 0.) First we must prove existence—that there js at least one solution. Let x = bla. [Note that we do need the condition that a + 0.) {A litle elementary algebra makes it clear that this value satisfies the equation, Now we ‘must prove uniqueness—that if there are two solutions, they must be equal. So assume that x and y are both solutions. But if ax + b= O and ay +b =0, then ax +b=ay +, by the transitive property of equality. Subtracting b from both sides yields ax = ay, and then dividing both sides by a gives x= y. This completes the proof. #34 The Equality Relation; Uniqueness 9 ‘Uniqueness plays an important role with respect to definitions in mathematics. Normally, it makes no sense to define something in a permanent way unless we know the object being defined is unique. Example 1: To illustrate this, suppose we know that for every real number x, there isa larger number. It would be silly to write a definition that says, “Given x, let y be the ‘number that is larger than x,” because there are many such numbers. It would make ‘more sense to say “Given x, let y be some number that is larger than x.” This is fine as a temporary definition within a proof, in Chapter 4 we call this type of naming existential specification. But it’s not appropriate as a permanent definition. (On the other hand, suppose we know that for every real number x, there is a nique number y such that x + y=0. (Theorem A-3 in Appendix 2 proves ths.) Then it makes sense to have a permanent definition saying “Given x, let ~x be the number such that x+(-x)=0.” Note that having the variable x appear in the notation ~x conveys the fact that this number depends on x. Exercises 3.4 (1) @) Prove the (b) implies (¢) part of Theorem 3.4 (b) Prove the (4) implies (a) part of Theorem 3.4. Q) Write symbolic statements that say: (@) There are at least two objects such that P(x). (©) There are at least three objects such that P(x). (©) There are at least n objects such that P(x). Here, n is any unspeci positive integer. Since you don’t know its value, you need to use at least one “ your answer. @ Write symbolic statements that say: (@) ‘There are exacly wo objects such that PQ), (There are enacly three objects such that PG). *(@) There are exactly n objects such that Ps) (Se the comments for Exercise 2) (4) Using the method of Exercise 2, do you think it’s possible to write a single symbolic statement that says that there are an infinite number of objects such that P(x)? (8). Redo Exercise 5 of Section 3.3 replacing every 3 with 3! (© Redo Exercise 6 of Section 3.3 replacing every 3 with 3! (7) Translate each of the following into symbolic form, using the instructions for [Exercise 7 of Section 3.2. You can use the abbreviation 3}; in fact you should use this ‘quantifier (as opposed to 3) whenever you think it’s the intended meaning of the statement,0 Chapter Predicate Logle (2) Everybody has a father and a mother. (b)_ Not everybody has a sister. (©) Nobody has more than two grandmothers. (@) Some people are only children, (e) Some people have only one uncle. *(£) Two people can have a common cousin without being cousins (8) This exercise relates to Example 5 of Section 3.1. Translate each of the following into symbolic form, following the instructions of Exercise 7. You need to make frequent use the predicate symbol On. For uniformity, use the variables 4, B, and Cto represent points and L, M, and NV to represent lines, (a) Lines Land M are parallel (that is, they have no point in common). (b) Any two distinct lines meet in at most one point. (©) Given any two distinct points, there’s a unique line that they're both on. (@ lines Zand M are parallel, then any line that is parallel to Z (except for ‘Mj is also parallel to M. (©), Pythagoras’s theorem (use the symbols for angle and distance referred to in Section 3.1). (6) Given any line and any point not on that line, there's a unique line through that point that is parallel to the given line, (This is one version of the famous Parallel Postulate of plane geometry.) (g) Points 4, B, and C are collinear, and B is between A and C. (h)_ Cis the midpoint of the line segment 4B. (9) @ Which of the equality axioms remain true if the symbol = is replaced throughout with the symbol < (and all variables are assumed to represent real numbers)? (b) Repeat part (a) using the symbol <. (10) Which of the equality axioms remain true if all the variables are assumed to represent triangles, and the symbol = is replaced by the words “is similar to.” Recall that two triangles are called similar if they have the same angles. (11) Prove: (a) Ifx and y are real numbers, there is a unique number z such that 2 ~ x= *(b) Ifx and y are unequal real numbers, there is a unique number z such that le-x1=le-yl *(¢)_ I?m and n are unequal odd integers, there is @ unique integer k such that em] = k= a]. Tn parts (b) and (c), proving uniqueness requires extra care because ofthe absolute value signs. A picture might help you to see what's going on. (12) Prove the following. You need to use some standard results from first-year calculus. You also need to analyze the quantifier structure of the statement you are proving.34. The Eat Relation; Uniqueness n (@) Every graph of the form y = ar’ + bx + ¢, with a> 0, has a unique ‘minimum point, (b) Every graph of the form y= ar’ + bx? + ox +d, with a #0, has a unique point of inflection, Suggestions for Further Reading: The same references that were suggested at the end of Chapter 2 apply to this chapter as wel.Chapter 4 Mathematical Proofs Different Types of Proofs ‘Now that we have looked atthe basics of propositional and quantifier logic, we are ready to study mathematical proofs in depth. Recall, from Section 1.2, that the primary ‘means for establishing new results in mathematics is called the axiomatic method. Certain statements are taken as axioms; this means that they are accepted without proof. ‘A statement may be included as an axiom because it is obviously true and quite simple ot (in more specialized studies) because one or more mathematicians think it may have interesting consequences. Tn this book we do not use any specialized axioms, Rather, we confine ourselves, to standard, “obviously true” ones, that is, axioms almost all mathematicians and you ‘would agree are correct. Axioms fall into two categories. Those based on logic are called logical axioms, Axioms based on properties of a patticular type of mathematical object (integers, real numbers, sets, and so on) rather than on logic are called proper ‘axioms; in Euclid’s day they were called postulates. Recall also that logical deduction is the only acceptable way to prove new things in mathematics. Traditionally, mathematicians have not spelled out exactly what types of reasoning should be allowed in proofs. It was generally assumed that everyone doing ‘mathematics would have a feel for this that would be consistent with everyone else’s. ‘But in the past only a very small percentage of the population came into contact with advanced mathematics. In today’s technological society, it’s considered valuable to make mathematics comprehensible to a wider audience than ever before. Furthermore, for various kinds of theoretical studies, including those involving the use of computers to do mathematics, it's important to have an exact definition of what constitutes a proof. ‘Accordingly, this book not only presents 2 set of axioms but also explains numerous rules of inference that are commonly used in mathematics. A rule of Inference is a precise rule that describes how a new statement may be asserted in a proof on the basis of is relationship to previous statements in the proof. Usually, when ‘mathematicians refer to a proof method, they mean a rule of inference. Example 1: Here is a rule of inference that we call the conjunetion rule: if wo statements P and Q appeat as separate steps in a proof, then it’s allowable to conclude nR41 Different Types of Proofs B the single statement P Q in the proof. This is a particularly simple and obvious rule of inference, but many of the ones we introduce are not much more complex than this. Formal Proofs Definition: An axiom system consists of two parts: a list of statements that are tobe considered axioms, and a list of rules of inference. ‘The lists mentioned in this definition may be finite or infinite, But in either case, the axioms and rules of inference must be clearly and unambiguously defined, so that it’s always possible to determine whether any given statement is an axiom or follows, from certain other statements by a rule of inference. In the next three definitions, we assume that we have a particular axiom system in mind. Definition: A formal proof is a finite sequence of statements in which every statement (or step) is either (1) an axiom, (2) a previously proven statement, (3) @ definition, or (4) the result of applying a rule of inference to previous steps in the proof. Definition: A theorem is a statement that can be formally proved. That is, it’s a statement for which there's a formal proof whose last step is that statement. Remarks: (1) There are several other words with more or less the same meaning. ‘A relatively simple theorem may be called a proposition. (This usage of this word is clearly quite different from the meaning we defined for itn Chapter 2,) A theorem that isnot considered very important on its own but is useful for proving a more important result is usually called a lemma. And a theorem that is easily proved from another theorem is usually called a corollary to the other theorem. There are no hard-and-fast rules for which of these words to apply to a given result. Some important results in mathematics have been labeled propositions or lemmas, pethaps because their authors ‘were on the modest side. (2) Sometimes itis appropriate to begin a proof with one or more assumptior {also called hypotheses or premises or givens. This usage of the word “hypotheses’ is very different from its usage in the sciences, as described in Chapter 1). Its important to understand the distinction between an axiom and an assumption. An axiom is a statement that is agreed on and available for use in proofs permanently, at least within 1 particular subject. An assumption is a statement that is available for use only in the proof being attempted. Assumptions were made in several of the proof previews in Chapters 2 and 3, in which the goal was to prove an implication. In this chapter, we see that there are only a handful of situations in which it’s legitimate to make assumptions in proofs. In fact, proving implications is essentially the only situation that justifies, assumptions in mathematical proofs. It is extremely important to leam when it is appropriate to use the word “assume” and when itis not.4 Chapter 4 Mathematiea Proofs Example 2: In Section 2.3, whenever we verified that an argument was valid, we ‘were essentially doing proofs from premises in an axiom system. This axiom system is very simple, having just one rule of inference: you may assert any statement that is 2 propositional consequence of the previous statements in the proof. Technically, this ‘axiom system has no axioms, but practically speaking all tautologies are axioms, since every tautology is a propositional consequence of any set of statements. Remarks: (1) Our definition of a formal proof allows four types of steps. Two of those—previously proven statements and definitions—are never needed to prove anything. Quoting a previously proven theorem in a proof just saves the trouble of reproving it, so while it can be a substantial time-saver, it never allows you to prove anything that you couldn’t prove without quoting it. Unless specifically disallowed (which could occur on test or homework problems), it’s acceptable to save time in this ‘way when doing proofs in mathematics, ‘The role of definitions is more subtle, but a definition just introduces a shorthand or abbreviated way of saying something. So definitions also save time and can be very enlightening, but they don’t allow you to prove anything you couldn't prove without them. You can use two sorts of definitions in proofs. You can make your own definitions, which create abbreviations (temporary or permanent) for your own convenience. But you can also quote any definition that has been given (for example, in whatever text you are using), as if it were an axiom. So if we wanted to give 2 ‘bare-bones definition of what a formal proof is, we could have limited the possibilities toparts | and 4 of the definition Q) It is important to understand the difference between an axiom and a rule of inference. An axiom is a single statement that we agree to accept without proof and therefore may be asserted at any step in a proof. A rule of inference is never a single statement; rather, it describes some procedure for going from old statements to new ones. However, in the less formal proofs that mathematicians normally write, this distinction often gets blurred, as we soon see. Except for very specialized and unusual systems, rules of inference are always based on logic. The ones in this book are all based on logic and, like our axioms, are quite standard. ‘A General-Purpose Axiom System for Mathematics ‘Appendix 1 consists of a detailed axiom system that we refer to throughout this book. The first part of it (all the rules of inference and groups I, Il, and III of axioms) is based on logic and is logically complete and sound, meaning that its power to prove statements corresponds exactly fo logical consequence. The rest of the axiom system consists of the generally accepted axioms about sets, the real numbers, and the natural snumbers. Taken as a whole, our axiom system is powerful enough to derive all currently accepted theorems of mathematics, even in the most advanced subjects ‘You might be surprised that it's possible to write an axiom system that encompasses all of mathematics but that is only three or four pages long. In fact, theDifferent Types of roofs 75 axioms for logic and set theory alone are sufficient for the development of all of mathematics. So if I wanted to be very economical, I could have omitted the entire sections of real number axioms and natural number axioms, and Appendix 1 would still encompass all of mathematics. These extra axioms have been included to make the system easier to use, since itis rather difficult to develop the theory of these number systems from logic and set theory alone. "You might also wonder whether this system is at all standard or if it’s just a personal creation of one author. As an experiment, you might ask your favorite mathematics professor what axioms she uses in her work. Her response will probably be a smile or a puzzled expression, with words to the effect, “I don’t use axioms when Ido mathematics. I just use intuition and deductive reasoning, plus a few well-known principles.” This type of response is probably honest, but don’t let it mislead you. By the time someone becomes a professional mathematician, she has had so much experience with the usual axioms and rules of inference of mathematics that they have ‘become second nature to her. Even ifshe has never had a course like this one, she is just about as comfortable using the principles discussed in this book as most people are driving a car or reciting the alphabet. Furthermore, just as these basic skills become ‘unconscious forall of us, an experienced mathematician may not even be aware that she uses a particular set of principles that has been subtly taught to her and is quite universal. But if you were to go beyond that first response and ask her to think in detail about how she does proofs, it would almost certainly tum out that she uses a combination of principles that are exactly equivalent to the list presented here. Informal Proofs ‘One major reason why this book discusses formal proofs is to help you understand that there is nothing mysterious or magical about what constitutes @ proof in mathematics It may take a great deal of ingenuity to find a formal proof of a given statement, but ‘once one is produced and written down, there would normally not be a controversy about whether it's correct. Any reasonable person who is willing to be very careful and take enough time ought to be able to check a formal proof for correctness. Better yet, computer programs can be written to check them. When mathematics is done formally, itbecomes a sort ofa game, with exact rules like chess or tic-ac-toe. However, although the rules of a game like chess are arbitrary and so must be leamed specifically, the rules of mathematics are directly based on logic and common sense, 50 that it should be a natural process to become fluent with them. ‘Now here comes the catch. In spite of the order and precision that could be brought to mathematics by sticking to formal proofs, this type of proof is almost never used by mathematicians, If you randomly went through a dozen mathematics books, you would probably not find a single formal proof. We'll soon see some formal proofs, and you'll casily see why they are avoided. They are often extremely long and tedious to write and even worse to read. A complete formal proof usually consists of pages of symbols, even if tis based on just one or two simple ideas. So if mathematicians don’t write formal proofs, what do they write? Naturally, we ‘may say that they write informal proofs. Unfortunately, i's not possible (o say exactly6 Chapter 4 Mathems 1 Proots what is meant by an informal proof. Furthermore, it’s inaccurate to think of formal proofs and informal proofs as two, clearly separate categories. The true situation is more like a whole spectrum. On one extreme are strictly formal proofs, On the other extreme are completely informal proofs. Informal proofs are not based on any specific axiom system, generally use English more than mathematical notation, skip and/or lump together many steps, and base most of their logical assertions on commonsense reasoning. They often are laced with words like “obviously,” “clearly,” “its easily seen that...” and in the case of one well-known mathematician, “tis intuitively obvious to the most casual observer that...” Informal proofs are much easier to write and read than formal ones, and a well-written informal proof conveys information better than any formal one can. The problem with completely informal proofs, especially when used by less experienced proof writers, is that they open the door to sloppy thinking and errors. To make all this clearer, here is a list of some of the ways in which most of the proofs written by mathematicians do not fit the definition of a formal proof: (1) Use of English: Normally, when an axiom system is precisely defined, the axioms and rules of inference are stated in symbols (that is, mathematical and logical notation), It would then follow that a formal proof in that system would consist of symbolic statements, not English ones. But most mathematics proofs flow better if there are words as well a5 symbols, and so most mathematicians use a liberal mixture of ‘words and symbols in their proofs. As long as these words strictly correspond to the axioms and valid logical principles, the use of English in a proof does not make the proof informal. But often, words are used to gloss over gaps in a proof, and in that case ‘the proof must be considered informal. (2) Lack of an Axiom System: We have already mentioned that most mathema- ticians do not consciously have an axiom system in mind when they write proofs; but ‘unconsciously, almost all of them do follow a system that is equivalent to the one given in Appendix 1. However, a mathematician may occasionally write a proof that is not based, even unconsciously, on a clear-cut list of axioms, Such a proof would have to be called informal. On the other hand, many mathematicians would say that a ronaxiomatic proof cannot be a correct mathematical proof. (3) Skipping Steps: Almost all mathematicians simply skip whatever steps they deem to be obvious. This is acceptable if the skipped steps really are obvious to whomever reads the proof. But this gets tricky: when you write a proof, how do you now who will be reading it and what will be obvious to whom? Something that would be obvious to most professional mathematicians would not necessarily be obvious to others, In practice, proofs are written differently for different audiences. How should you handle this subtle point? Under what circumstances and to what extent should you leave out obvious steps? There is no pat answer to this question. Your instructor and the remainder of this text will constantly try to give you a feel for this. In the meantime, here is a good rule of thumb:41. Different Types of Proots ” 8 Do not omit any steps in a proof unless you can see clearly how to fill in all the gaps completely. Nothing gets you into trouble more surely than skipping steps and calling them obvious, without knowing precisely how to carry out all the omitted steps. (4) Combining Steps: ‘This is a variant of skipping steps. Mathematicians often Jump several easy steps into a single sentence, glossing over them rather than leaving them out entirely. The same guidelines that were described for skipping steps also apply to this practice, (5) Reverse Proofs: This is a fairly common practice among mathematicians and one that can confuse the inexperienced reader. Technically speaking, when you prove a statement in mathematics, the statement you are proving should be the last step in the proof. That is, you start with things you are given (axioms and/or assumptions) and try to get to what you want, But sometimes the easiest way to figure out how to prove something is o do it in reverse, starting with what you want to prove, then looking for some statement that implies what you want, then looking for some statement that implies that statement, and finally reaching a known statement. A correct reverse proof ‘can always be turned into a formal proof by writing the steps in the standard, forward order. But often a mathematician decides that a proof reads better in reverse and so ‘keeps the final version that way. There is no harm in this informality if done properly. ‘But whenever you do a proof in reverse, make sure it works forward; otherwise, you're probably proving the converse of what you should be proving. This technique is discussed further in Section 4.6 under the heading “Reverse Reasoning,” and we will see many examples of this important idea. Good Proofs How formal or informal should your proofs be? There is no pat answer to this. The dangers of both extremes have already been pointed out. While you are learning to write proofs, it is probably better to play it safe by keeping your proofs relatively complete. As you start to gain confidence, you can start to write mote informally and skip a bit more, You should feel free to ask for guidance from your instructor and other experienced mathematicians regarding these issues, since it's quite hard to leam good proof writing without frequent feedback. Here’s another rule of thumb: A good proof should be a clearly written outline or summary ofa formal proof. This means that when you write a proof, each statement you ‘write (especially if it’s an English sentence) should describe or indicate one or more steps that you would include in a formal proof if you were writing a formal proof. You can't do this unless you see how the formal proof should go. Once you've done that, you need to outline the formal proof in such a way that any reasonably intelligent reader, including yourself, should be able to understand the outline well enough to reconstruct the formal proof from it. This outlining process can require considerable thought and is ‘what is meant by style in mathematics.B Chapter 4 Mathematical Proofs ‘We have already encountered two very different styles of proofs. In the “Alternate Solutions” to Example 1 of Section 2.3, formal proofs were given for parts (a) and (c). For contrast, a less formal proof of part (¢) was also included. All the proof previews in Chapters 2 and 3 are also written in good, nonformal style, as are the great majority of proofs from this point on. Occasionally we give a formal proof in the text, and there are several of them in Appendix 2. When you encounter these formal proof’, you should find it easy to understand their main advantage (that they encourage clear, correct step- by-step thinking) and their main disadvantage (that they are unwieldy, both to write and to read), ‘We occasionally use the term semiformal to describe a proof that directly parallels, and summarizes a particular formal proof, 4.2. The Use of Propositional Logic in Proofs Example 2 of Section 4.1 explained that the notion of propositional consequence introduced in Section 2.3 provides the basis for an important rule of inference. In fact, this one rule of inference is completely general with respect to propositional logic, in the sense that it includes every valid proof method based on propositional logic. Therefore, ‘we discuss this rule of inference first and view it as the basis for all the material in this section. Rule of Inference: Propositional Consequence In a proof, you may assert any statement that is a propositional consequence of previous steps in the proof. ‘We sometimes shorten “propositional consequence” to “prop. cons.” or simply “PC.” Equivalently, we often just say that a step in a proof follows from previous steps “by propositional lo Example 1: Suppose we are talking about a real number x. We know (from axiom ‘V-15) that either x > 0, x = 0, or x < 0. Suppose we also know, somehow, that x is nonzero. Then PC allows us to conclude that x > 0 or x <0. This use of PC is based on the tautology [(P VQ V R) A~Q)—+ (P VR). This exact tautology does not appear in Appendix 3, but it is quite close to tautology 11. At any rate, you should try to reach a point where you don't need to refer to Appendix 3 very often to check conclusions of this sort, because your own feel for logic makes it unnecessary,42° The Use ot Propositional Loge ln Proofs 9 Example 2: It isa theorem of calculus that if a function is differentiable, continuous. Suppose that we know this result, and we want to assert its contrapositive in a proof; that is, if function is not continuous, then itis not differentiable. The rule PC allows us to do this, using the simple tautology (P+ Q) ++ (~ Q-*~P). For some more substantial examples of proofs based on propositional consequence, refer back to the examples and proof previews in Section 2.3. ‘As we mentioned in the previous section, having the rule of inference PC essentially makes all tautologies axioms. We now make this explicit. All tautologies are axioms. Note that a tautology is, by definition, always true and it’s also straightforward to determine whether a given statement is a tautology. These are ideal characteristics of axioms, ‘Example 3: Suppose that we are trying to prove something about a real number x. Ifwe want to, we can assert the statement that either x= 0 or x + 0, since this is of the form PV ~ P (tautology 1, the law of the excluded middle). It might seem pointless to assert this disjunction, but this step might be used to set up a proof by cases, which could substantially simplify the proof. On the basis of what we said at the beginning of this section—that propositional consequence includes all valid proof methods based on propositional logic—we could technically end this section at this point. However, PC is too general to be very convenient in most situations. Instead, mathematicians commonly use at least a half dozen more specific rules of inference, So let's now examine some of the most important of these so-called derived rules of inference. Rule of Inference: Modus Ponens Ifyou have a step P and another step ofthe form P -* Q, you may then conclude the statement Q. This rule of inference can be diagrammed (in the style of Section 2.3) as follows:80 (Chapter 4 Mathematical Proofs P+Q Despite the Latin name (which means “method of affirming”), this isa very simple rule of inference. Hopefully, you can see that modus ponens is logically correct: if you know P and also that P implies Q, then Q must follow. This is more or les the definition of implication ‘A bit more formally, itis tautology 9 that justifies modus ponens. Exercise 8 asks you to derive modus ponens from propositional consequence. Example 4: One important theorem of calculus is that if a function is differentiable, it must be continuous. Another basic result is the derivative formula for polynomials, which guarantees that polynomial functions are differentiable. Applying modus ponens to these steps yields that any given polynomial function, such as 3x? - Gx + 2, must be continuous. Starting with the next example, we occasionally illustrate a method of proof by referring toa proof of a theorem in Appendix 2 at the end of this book. Even though we have not discussed Appendix 2, you need not be intimidated by these references, ‘because the proofs in Appendix 2 are based on the standard properties of the real number system, which are quite familiar to you from high school algebra. Example S: Every proof in Appendix 2 contains uses of modus ponens. A typical instance, although it’s not specifically mentioned, occurs in the proof of Theorem A-7. In that proof, we have the step z + 0 since it's an assumption in the proof. We also have, from axiom V-11, the implication z ¥ 0+ zz"'= 1, Modus ponens applied to these two statements yields the step 22" = Example 6: Let’s redo Example 1(a) of Section 2.3, using a formal proof from hypotheses, with tautologies as our only axioms and modus ponens as our only rule of inference: @ PQ Premise Q) -R>~Q Premise @) -R Premise @ ‘Modus ponens applied to steps 3 and 2 ~Q 6) @>Q+-Q>~P) —Tautology ©) ~Q>-P Modus ponens applied to steps 1 and $ ( ~P Modus ponens applied to steps 4 and 6