CALCULUS

A Clairer approach to Ritchstones

All rights reserved (1994)

Dedicated to my wife and to the memory of two friends professors Clair Gates and Martin Richstone.

Every year, I receive several new Calculus texts as well as updated editions of old ones. None seem to have a substantially novel point of view. Hence, one would think that the last thing we need is yet another book on this subject. To be worth my while, as well as to merit the readers attention, this book has to contain a signicantly fresh approach which is pedagogically more successful than books of the past. I hope that the reader agrees that this is exactly what has been accomplished.

THE NATURE AND TEACHING OF MATHEMATICS

This essay is intended for both teachers and students. If you are a student and you come upon a passage which assumes knowledge not yet available to you, then simply skim over it and try to lter out the point being made. It is astonishing that a subject, such as mathematics, to which everyone has had a great deal of exposure, is so misconstrued. Contrary to lore, mathematicians are not preoccupied with the manipulation of large numbers in their heads. Mathematicians tend to be good at ciphering, but this is a byproduct not the goal of their discipline. Whenever I ask students to examine classical mathematics texts, they are astonished by the dearth of numbers within their pages. At a time when we are told that we have achieved the highest levels of public education, the common picture of a mathematician is a person who is exceedingly good at balancing his checkbook. This is akin to believing that Seurat must have been very procient with a paint roller. Mathematics is not about nding numeric answers to problems. It is not even about solving problems. It is about logic and about viewing problems from different perspectives. Out of context, it may be impossible to determine the given and the required of a problem. Which of the following is an answer, which is the question?

1 2 /4 (/4)3 (/4)5 (/4)7 (/4)9 sin(45), , , cos(x)dx, + + -... , the ratio of one of the sides to the 4 3! 5! 7! 9! 2 2 0 hypotenuse of a right angled isosceles triangle, etc.
1 and the x-axis, over the interval [1,4], then we may x say that it is ln(4). If we were asked to nd the area between f(x) = sin(x2) and the x-axis, over the If we were asked to nd the area between f(x) =
1

interval [0,1], then we may respond that it is

sin(x
0

)dx . Some claim that we have solved the rst

problem, but not the second. This is not quite correct. About the most that can be said is that the number of ways of looking at the former is richer than the number of interpretations of the latter. If the desired result is a decimal approximation then, in both problems, we may need to use Riemann Sums or Taylor Series. In such a context, the solution ln(4) is hardly better than

sin(x
0

)dx .

Suppose f(x) = 2x. If the ancient Greeks had been aware of the Differential Calculus, then they might have preferred the formula f(x) = 2x f(0). We favour the formula f(x) = 2x ln(2). The Greeks believed that solutions to mathematical problems had to have a geometric interpretation. Having a factor which represents the slope of the function at x = 0, would have been pleasing to them. Our students consider f(x) = 2x f(0) to be a partial solution. To the mathematician, both forms are delightful. One conveys the fact that the derivative, at every point, is dependent on the specic derivative at x=0; the other form unveils a relation between the exponential and the logarithmic function. 5

For scientists, a mathematical problem is solved when a decimal solution is found. While an engineer may nd the equation = 3.1416 practical, the mathematician nds it repugnant. A mathematician never uses the equal symbol between two numbers which are not equal. In other words, = 3.1416 is objectionable because it is not true. Since all that is needed to make mathematics impotent is a single error, falsehoods are shunned at all costs. We are not being pedantic. = 3.1416 implies that is a Rational Number which it is not. Of course, engineers are not wrong. It is understood that, when they write =, they really mean ~, i.e. approximately equal. Unfortunately, often they fail to make this explicitly clear. 1 1 1 1 + - + -.., to be beautiful. Yet, it is unlikely to be 3 5 7 9 important in scientic applications, as the convergence is very slow. In mathematics, the goal is to change form not necessarily to obtain an efcient decimal approximation. Mathematicians consider the equality =1Of course, the scientists need for decimal results is perfectly understandable. All experimental equipment, tools and manufacturing machines speak the language of the Rational Numbers. Mathematicians are not so constrained. Mathematics is neither in the business of measurement, nor in the creation of tangible objects. Of course, I am talking about the strict discipline called mathematics, not about its practitioners. In reality, many of the discoveries of mathematics were motivated by observations about the universe and many of the greatest discoveries in mathematics were made by scientists. A common misconception about mathematics is that it is a subject in which one can have unconditional faith. Not only does mathematics lack a monopoly on truth, it is not even about truth. We dont know if our deductions reect the universes true essence. But, and this is an important but, we dont make this claim. Instead of making true statements, mathematicians make consistent statements. This is a kind of relative truth. We dont know if our conclusions are true, but we try to prevent contradictions. The difference between truth and consistency is very much like the difference between absolute and relative pitch. If a virtuoso violinist, having excellent relative pitch, tunes a violin and goes on to play a cadenza, then the tune will be pleasing to the ear. If the rest of the orchestra joins in, the result may be disastrous. It may turn out that, after all, the violin was mistuned. The relations between the notes were perfect, but the frequency of each note was not what is prescribed by impartial electronic instruments. (This explains why the soloist and the orchestra tune together.) Mathematics does not insist on perfect pitch, only on correct relative pitch. Though it may not be the absolute truth, the results are very pleasing to the ear. Ten different professors of English may not agree on the quality of a particular composition. But ten conductors will agree whether a person is or is not singing out of tune. Similarly, mathematicians agree on the validity of an argument. It is in this sense that mathematical arguments are rock solid not in the absolute sense. Mathematics is not one of the sciences. Its primary goal is neither to describe, nor to predict the exterior world. Mathematics has style, grace and form. It is an art. The conspicuous feature of mathematics, as opposed to the sciences, is the use of proofs instead of observations. Though the distant origins of mathematics lie in the need to explain physical phenomena, today, its main aim is to 6

create something beautiful. Just as the photograph has freed the ne arts from the need to mimic reality, the mathematician is no longer compelled to explain nature. The idea that mathematics is an art is no radical theory. Every treatise on the subject, by authors such as Plato, Pascal, Gauss, Courant, Halmos, Coxeter, Erds, etc. adhere to this point of view. To quote Bertrand Russell: Mathematics possesses not only truth [i.e. consistency], but supreme beauty a beauty cold and austere, like that of sculpture, without appeal to any part of our weaker nature... sublimely pure, and capable of stern perfection such as only the greatest art can show. As an art, mathematics is unique because it affords little poetic license. As we descend through the hierarchy of its craftsmen, the freedom to express oneself is more and more restricted. Euler and Gauss had the liberty to choose both notation and axioms, while most of us have to be content with making a well known argument slightly more elegant. Luckily, there is more than a modest amount of pride to be gained from tweaking or just comprehending, a tricky proof. Trying to dene mathematics is as difcult as it is to dene the ne arts. Mathematicians are driven by such goals as simplication, categorization, organization, generalization, unication of different concepts and elegance of form. Since the eighteenth century, mathematicians have concentrated on non-applied, formal mathematics. Why is it that even this body of work is applicable? Perhaps, this is not a paradox. Perhaps, mathematics is an intrinsically natural subject which mirrors the organization of our brain, and by extension, our universe. Logicians tell us that mathematics is nothing but the development of rules which dictate what mathematical symbols may be juxtaposed and which sequences of symbols are illegal. Though technically this is correct, it is an arid point of view. All branches of mathematics consist of: (a)$ a formal content (b)$ an intuitive basis (c)$ (usually) applications FORMAL CONTENT The formal content of mathematics consists of relations among undened things. Every branch of mathematics starts with undened things to which is attached some properties and interrelationships, called denitions and axioms. Then, using the rules of mathematical logic, increasingly complex deductions are made about the properties of these undened objects. The crystallization of an intuitive concept into a rigorous denition often brings with it unforeseen consequences. Though there is great benet to be gained from taming a previously elusive idea by such means, the real dividends are those serendipitous side effects which are unleashed when we pursue the topic further than our original goals. For example, functions which are continuous only at a single point and functions which are continuous but nowhere differentiable are the surprising discoveries which make the subject exciting. The bureaucrat seeks the rule, the mathematician the exception. Mathematics is very much like the game of chess. The essence of a chess game is the rules which govern the actions of the pieces. It is useless to ask what a bishop is really like. The only real 7

aspect of a bishop is the rules governing its movement. Similarly, the formal aspects of mathematics does not concern itself with what the number 2 is really like. What is important is that 2 is positive, that it is greater than 1, etc. We learn the nature of the number 2 indirectly, from its relationships to other undened objects. As with any good game, the only formal requirement of a branch of mathematics is that its rules be consistent. It is perfectly possible to change the rules and invent a new branch of mathematics, as long as these new rules also are consistent. Mathematicians have entertained variants where the distance between two points is not given by the Pythagorean theorem, where no two lines are ever parallel, where areas which dont exist in one approach suddenly have a value in another, and many more. INTUITIVE CONTENT The reason that chess is so hard to learn is that the actions of the pieces have no basis in reality. Bishops of the Catholic church do not tend to run on diagonal lines; knights do not jump in L-shape forms. You will not improve your chess strategy by prolonged observation of mediaeval nobility. Luckily, the most elementary branches of mathematics contain undened objects whose behavior is predictable from ordinary experience. If you pretend that the number 2 really means a couple of pencils, then you will be able to anticipate many of the formal properties of the number 2. Our intuition, about the undened items of elementary mathematics, is so strong that it is our primary force in mathematical strategy. However, such isomorphisms between reality and mathematics have their limits. No amount of knowledge about a pair of pencils provides a hint that 2 is an Irrational Number and this is exactly the point at which true mathematics starts and drudgery of computation is left behind. You cant call yourself a chess acionado if all you know is the basic moves of the pieces. With a little experience, the novice chess player acquires chess intuition. Desirable patterns become quickly recognizable and those which are futile are abandoned. What was a bewildering mess of rules is slowly transformed into intuition. We are not born with this type of intuition. Intuition is built layer upon layer and sometimes, it may extend into extraordinary regions. There are mathematicians who, due to prolonged exposure, have acquired an intuition for 4 and even 5 dimensional objects. APPLICATIONS One of the greatest differences between chess and mathematics is that mathematics is a practical game. It is doubtful that even a game, such as Monopoly by Parker Brothers, translates to real life the way that mathematics does. For some people, the applicability of mathematics is its sole benet. For others, it is an insignicant byproduct. A balanced point of view is probably the healthiest. Any dismissal of Newtons proof, that planetary motion obeys Keplers laws, is certainly insensitive. On the other hand, engineers who are interested only in plugging into formulae may be boorish.

THE TEACHING OF MATHEMATICS

Perhaps, the biggest problem with the teaching of mathematics is that it is not guided by the above principles. Another problem is that, at the elementary level, most texts and curricula are being written by non-mathematicians. In Canada, the province of Quebec, for years has had in its high school texts some mathematical content which can be described only as nonsense. These books cost millions of dollars. One would think that they could afford a few dollars to have a professor of mathematics proofread them. Perhaps, the problem is that no professional mathematician can be found who agrees with the pedagogical approach of these texts. Mathematicians dont like texts on geometry, which lack geometry and texts on algebra which contain little or no algebra. Instead of talking in generalities, let us consider some specic examples which illustrate common problems. Most high school teachers, would reject the equation sin(45) = Some will give full marks for the expression sin(45) = The preferred result seems to be sin(45) = 0.707 . 1 unacceptable? Though I have yet to nd anyone who can provide a satisfactory 2 answer, my students nd this symbol so repugnant that not one of them would leave it as the answer to a question. On the other hand, the expression sin(45) = 0.707 is common, even though it is false. Why is the form Let us consider two simple exam questions. Problem 1. Write down the denition of the set of Rational Numbers. The following ve responses were provided by students: 1$ 2$ The Rational Numbers is the set {1,2,3,4, ... } The Rational Numbers is {x R | there exist a,b R where x = Q = {x R | there exist a,b Z where x = Q = {x R | there exist a,b Z where x = a }. b a and b 0}. b a }. b 2 . 2 1 . 2

3$

4$

5$

Q={

a | a, b Z and b 0} b

The teacher simply dismissed responses (1) and (2) with no explanation. Would it not have been more instructive to point out that the word Rational probably comes from the word ratio? Hence, response (1) is unlikely to be correct. As for response (2), we should note that, for any x x R, x = . Hence, response (2) implies the Rational Numbers must be the same as all the 1 Real numbers. Though this is not an impossibility, mathematicians rarely use synonyms. So, this is unlikely to be a good strategy. The teacher favored responses (4) and (5). She liked the fact that both students knew that, part of the denition is that 0 could not be a denominator. However, in response (4), the assertion that b 0, is superuous. The moment that we declare that a Rational Number x must be a Real a Number, we acknowledge that x , as any such form is not a Real Number. Denition (3) is 0 more elegant than response (4) and is perfectly correct. The greatest problem with response (3) is that it denes the Rationals in terms of the Reals, but the Reals form a superset of the Rationals. We do not dene the Naturals in terms of the Integers. We do it the other way around. (5) denes the Rational forms. Its advantage is that it does not dene Q in terms of R and therefore, it is necessary to add that the denominator is not 0. The disadvantage is that the Rational forms are not exactly the same as the Rational numbers. For instance, in 3 6 Combinatorics, we may want to count and as two distinct ratio forms. But in Algebra, we 4 8 3 6 do not want to claim that and are different points on the number line. Hence, the student 4 8 c e should have added the disclaimer: It is understood that if and represent the same point on b g the number line, then they represent only one Rational number. (I would not deduct any marks for the absence of this disclaimer as this is probably understood.) At any rate, it is up to a good teacher to point out these subtleties. Problem 2. What is the formula for the area of a circle? Three responses were: 1$ 2$ 3$ The area of a circle is . The area of a circle is 2r2. The area of a circle is 2r.

10

The teacher treated each of these answers with equal scorn. I think that there is plenty of discrimination to be made here. By far, the worst answer is (1). It outrageous that a student believes that the area of a circle is independent of any of its dimensions. Apparently, all circles have the same area. Answer (3) is almost as bad. Such a student relies much more on poor memory than on logic. Since 2 and are unitless constants and r is a linear variable measured, say in cm, a unit check of response (3) would result in centimeters. But, area is measured in square centimeters. Hence, this could not be correct. A thinking student, who has no idea of the correct answer, would know that responses (1) and (3) are impossible. Without further knowledge, there is no way to tell whether answer (2) is right or wrong. For this reason, answer (2) should receive the most part marks. In most elementary texts, the denition of the Integers is as follows: I = { ..., -3, -2, -1, 0, 1, 2, 3, ... }. I have several objections to this notation. 1$ Historically, the symbol for the set of integers is not the letter I, but the letter Z. Though I is preferable, because the word Integer begins with the letter I, it is not up to us to change the symbol. If I could come up with an argument showing that the word asset is a much better word for things we sit on, than chair is, I could not ask someone to sit on their asset. 2$ You cant start a sentence with the phrase and so on. Yet, this is exactly what happens when we read the expression { ... , -3, -2, -1, 0, 1, 2, 3, ... }. 3$ This denition implies that Integers, like the Natural Numbers, are listable in order. This is no longer just a syntactical aw but a grave error. It attributes a property to the Integers which they lack. The classical denition of the Integers is Z = {0,1,-1,2,-2,3,-3,4,-4,...}. Why is this so objectionable? Who was the rst author who changed the denition, and why? Why did subsequent writers follow the lead? Whatever the answers are to these questions (if there are any answers), was it worth making the change? My preferred denition is inductive: Z = {0,1,-1,2,-2,3,-3,4,-4,...n,-n if n > 0 or -n+1 if n < 0, }. It is dangerous to encourage students to think that a nite number of examples are sufcient for the denition of an innite set. When proving a theorem about an innite set, we do not accept 3 or 4 examples as sufcient evidence that the whole set has the claimed property. An inductive denition is preferred to one based on a few examples. While reviewing the topic of curve sketching, I invited suggestions on how we could determine the shape of the function f(x) = x3-x2+x. Several students informed me that they were taught to solved such problems by consulting a graphing calculator. This response reminded me of an anecdote concerning Michelangelo: When asked how he carved the statue of David, he replied that he simply chipped away anything which did not look like David. 11

Surely, Michael Angelo was saying that his vision of David was so vivid that it guided his chisel with certainty. Using the graphing calculator to sketch a curve is akin to believing that David was literally present in the marble and any fool with a hammer could have cut away the superuous chips. In a Calculus course, the purpose of curve sketching is to give a terse geometric summary of the analysis which was conducted on a relation. I am not saying that making algebraic conclusions from geometric evidence has no place in mathematics. I am saying that it has no place in an introductory course on Calculus. In such a course, we dont really care what the graph looks like. What we care about is the cognitive ability to graph. And if this be so, what sense does it make to use a calculator to graph the function? One day, while driving home from school, I was listening to a radio station. There was a heated argument about the wisdom of allowing calculators in the college classroom. There were proponents who felt that it was sadistic to make students perform lengthy calculations when a calculator could relieve the drudgery. There were those who insisted that using a calculator was legalized cheating. In my opinion, both of these opinions were off the mark. What if the topic of discussion had been the use of the paint roller by Leonardo da Vinci? If, while painting the Mona Lisa, Leonardo had used the roller, would he have been cheating? The best that Leonardo could do with a roller is to expedite the coating of the raw canvas with a prewash and such automation would not detract from the craftsmanship of the painting. Surely, he could not have used the roller to apply the delicate strokes which make up the Mona Lisas face. If you have the proper goals in mind and ask appropriate questions on your exams, then having a calculator should neither help nor hinder a students effort. If, when curve sketching, you allocate 9 out of 10 marks for the actual graph and 1 out of 10 for the analysis of the function, then having a calculator provides an enormous edge. If you allocate marks the other way around, the calculator is almost useless. I rarely use a calculator in my classes and I do not care if my students have them during an exam. However, I do give them the following warning: 1$ A calculator almost always gives you the wrong answer. 2$ If you nd that you cant solve one of my problems without a calculator then surely, you are going about it incorrectly. The use of calculators, in an high school courses, is a totally different matter. It is true that a calculator frees us from the drudgery of elementary operations. Unfortunately, those who cant do elementary arithmetic, without a calculator, lack intuition about some of the most basic properties of numbers. If you need the calculator to multiply by 100, then how likely are you to understand the topic of binary arithmetic? How much intuition will you have about limits? To quote Albert Einstein: Everything should be made as simple as possible, but not simpler. I am not a fan of aphorisms, but this little witticism sums up all that is wrong with our educational system. I have no objection to the use of calculators by students who are procient at elementary Arithmetic. I do object to their use in lieu of such knowledge.

12

For one of my assignments, students had to nd out how often their parents used Calculus on their jobs and in their daily lives. Of course, the answer was very rarely or never. I am not suggesting that Calculus is useless. On the contrary, both directly and indirectly it has advanced our technology and understanding of the universe. But, besides mathematicians, the people who need a detailed understanding of Calculus form a very small percentage of the population. For too many students, mathematics appears to be more of a bureaucratic obstacle than a desirable subject. They know that the main reason for taking mathematics is to make it easier for university administrators to reject the large number of applicants vying for few positions. Mathematics has become the whipping boy of modern education. Mathematics is a very desirable subject and should be marketed for what it is. The main purpose of a Calculus course is for the student to learn abstract and rigorous thinking. Ironically, in the world of quantum theory, string theory and the absolute speed of light, reality in the macro and micro universe have become as abstract as any mathematics. Mathematics is precisely the subject that allows us to navigate in this intangible reality. Unfortunately, we tend to push pseudo applicability of mathematics. Does mathematics become more palatable because farmer Brown can use quadratic equations to gure out how much fencing he needs? Do baseball players really think about instantaneous velocity when they strike a ball? Applications can be the icing on the cake of mathematics. The problem with applications is that the genuine ones tend to be difcult. The unfortunate result is that a lot of the applications, seen by our students, are perceived to be bogus. Over the last 35 years, we have seen a great shift in the focus of mathematical education. Deductive geometry has become just geometry. Mathematics without proofs is like music without sound. The argument that the average student cant understand proofs either demonstrates that the average student should not take mathematics, or that its teaching is poor, or both. There is a narrow window of opportunity for the maturing mind to be exposed to the type of disciplined thinking which Geometry imparts. If we miss this critical point of maturation, then what should have been easy may become unattainable. When I arrived in Canada, my father was 45 years old and spoke 4 languages uently, English not being one of them. In contrast, I was a linguistic ignoramus, aged 11. My father acquired English within a month or so. I took two years to make myself barely understood. Nevertheless, I learned to speak without an accent, while my father forever had a heavy Hungarian drone. No abundance of intellect could make up for the fact that the part of my fathers brain, which allowed natural assimilation of language inection, was closed. I am convinced that the same process applies to common logic. There are crucial times, during the minds maturation, when a students brain is ready to absorb certain complex phenomena. If you miss exposure at these times then there may be no recourse. By abandoning deductive geometry, we are crippling our students. Why have we forsaken geometry for pie charts? Which is more likely: A student trained in Euclidian geometry can be self-taught pie charts or a student trained in pie charts can be self-taught elementary geometry?

13

It is very likely that the same quantity and quality of neuron synapses must be established to learn the proper version of a language as its vernacular. The consequences of this philosophy are rather obvious. Do not make concessions to the mathematical and linguistic immaturity of the present day student. The dumbing down of modern texts perpetuates illiteracy. Experience shows that the average person will perform slightly below what is expected and of what they are capable, this margin being quite inelastic. In other words, if we lower our standards, with the expectation that marks will improve, we will be caught in a descending spiral exactly our present state. Since it costs no more, we might as well maintain the proper standards. Some of the major difculties of teaching mathematics are analogous to those of teaching a foreign language. Consequently, great effort should be extended in guiding the student toward a proper mathematical language. Students should be required to write their mathematics papers in full and coherent sentences. A large portion of their difculties with mathematics stems from their slapping down mathematical symbols and formulae isolated from a meaningful sentence and appropriate quantiers. Papers should be neat and legible. They should not look as if a stoned y had just crawled out of an inkwell. Insisting on reasonable standards of expression will result in drastic improvements of mathematical performance. The systematic use of words such as let, there exists, for all, if ... then, therefore should be taught and expected. I do not believe that Prof. Serge Lang, who is being paraphrased in the above paragraph, is a neat freak. His pedagogical point of view is that productive thinking and proper presentation are linked. A student who does not use quantiers and proper logical syntax does not think about quantiers and probably lacks a reliable logical foundation. It is possible to be a great writer and spell very poorly, but not in mathematics. If you write (ab + c) instead of a(b+c) then not only is your spelling wrong, but your actual meaning is also incorrect. When I graduated from high school, in Quebec, in the early 1960s, an average grade of about 80% would guarantee an acceptance to most any North American university or college. Though, I and my friends did not have to write SAT tests, we wrote them. Unlike our US cousins, we did not practice for them. Yet, we all placed within the top 3 percentile. The education system was far from perfect. Yet, we seem to have achieved some goals. Our nal grades were not statistically laundered. The introduction of statistical treatment of marks has been a disaster. It punishes excellence and rewards mediocrity. When I was a student, letters of recommendation meant very little. They were viewed as not objective and possibly corrupt. Today, recommendations are more important than grades. CEGEPs inate their marks and have vastly different standards. A grade of 95% has less of a value than 80% had in the 1960s. In view of this, administrators are forced to resort to letters of recommendation. I have had students audit my courses though they were enrolled, in the same subject, in other classes. When I asked why they would not transfer, they replied that they enrolled in a poor class on purpose so that their standardized grade would appear as high as possible. Apparently, how much you know is unimportant. What counts is how much more you know than your classmates. Standardization of marks means that we have lost sight of the aims of the curricula. We must devise a priory, a method of testing to see if we have achieved what we had set out to accomplish. Despite the overwhelming evidence that our present methods are disastrous, there are educators who cling to them. When tests seem to show a failure to absorb knowledge, they scramble to change these tests 14

so that their point of view may be vindicated. Or, they standardize the results so that a certain pass rate is guaranteed. This is neither a logical nor a proper scientic approach. Recently, a speaker at an educational conference claimed, in all sincerity, that if we eliminated all tests then students would not perform poorly on them. Was this speaker acquainted with the notion of a tautology? Was she familiar with the properties of the empty set? In her opening remarks, the chairman of this event entreated the audience thus: to improve our educational methods until all of our students are above average. I wonder how this person denes the notion of average?

15

INTRODUCTION

Often, teachers are convinced that the best approach to a subject is, by an amazing coincidence, the same as the one they were exposed to as students. This fosters the prejudice that any departure from established dogma must make the subject more difcult. Of course, this is not so. From the perspective of a three hundred year development, it is possible to approach the Calculus from an intrinsic, rather than a chronological point of view. I have tried, without strain, to make my exposition unied. Indeed, most every concept is directly or indirectly related to the two fundamental ideas of Order of Magnitude and Mean Value. The Mean Value concept is nothing new. The notion of Order of Magnitude is also old, but it has disappeared from introductory texts. Certainly, my almost total reliance on it is, as far as I know, is novel. The consequences of the above approach are too numerous to list here. Almost every topic is handled in an unorthodox manner. Of course, just because a proof is original it does not mean that it is better. However, I believe that my arguments are heuristic, more elegant, and more uniform than what we see in present texts. Again, the consequences of my approach are too numerous to list here as they are evident in almost every page of the text. Let me just highlight two examples. The notion of Order of Magnitude allows the derivative to be developed from the point of view of a best linear approximator, rather than from the geometric notion of a tangent. It is much better to anchor an analytic concept in numerical analysis than in geometry. Further, this approach makes the introduction of Taylor Polynomials much more natural and convincing. We prove that, (in most cases) as we increase the degree of the Taylor Polynomial, we lower the Order of Magnitude of the error of estimate. We completely bypass the usual, vague argument rooted in geometric intuition, that if two functions, at a point, have the same tangent, concavity, etc. then they must be very close to each other, at the point. The word integrate is usually interpreted to mean add, or to establish the whole as the sum of its parts. Hence, most texts begin the Integral Calculus with the Riemann Sum of a continuous function. They ask the student to believe that the limit of any such sum converges to a unique number, and that this number represents area. Another, legitimate interpretation of the word integrate is to nd, among a large set of data, an average condition. Using this interpretation, we prove, rather quickly, that antiderivatives are area measuring functions. The rest of the theory, including the Fundamental Theorem, is easily and rigorously developed. It is quite a stretch to assume that all Riemann Sums, of a continuous function, converge to a unique number. Further, Riemann sums are the most computationally demanding part of the course hardly the place to begin. Only trivial examples may be discussed, as only the limits of such examples are within the grasps of our students. With our approach, any function which is antidifferentiable has an area associated with it. Hence, we have an abundance of functions whose area we can nd. When, in traditional courses, students are asked to comment on Riemann sums, they usually say: The Riemann sum is used to torture us until the teacher lets the Fundamental Theorem out of the bag. In the long run, I cant think of any good reason to study the Riemann sum. I am not a fan of such sadism and I dont like leaving my students with the conclusion that the Riemann sum is useless. We introduce Riemann sums toward the end of the Integral Calculus. Its 16

purpose will be to approximate area when the antiderivative is difcult or impossible to express in terms of elementary functions. At the very end of the section, we use the Riemann sum to prove that all continuous functions are antidifferentiable and hence, have an area associated with them. By far, the greatest advantage of our approach is seen in applications. The denite integral, which solves a particular application, is set up with much more ease than is traditional. We never resort to the odious notion of negligible quantities. Every attempt has been made to give this work mathematical integrity. Nevertheless, a completely axiomatic treatment of the Calculus, at this level, is neither practical nor essential. For instance, we admit that the most elegant introduction to transcendental functions is through innite series or denite integrals. Neither is feasible here. We will relegate to an appendix, a short but completely Analytic introduction to the transcendental functions. We shall make clear that occasionally, we are justifying axioms not proving them. There is a tendency to gloss over these details to such an extent that the student believes that something has been demonstrated when it has been only dened. For instance, we shall avoid circular arguments which prove that the area of the circle is r2 based on the antiderivatives of Trigonometric functions, knowing that the proofs of these antiderivatives are rooted in the assumption that the area of a circle is r2. Most of what is novel here is pedagogical in nature. If you are a mathematician, then the most likely response to my proofs will be a slight smile. If you do not specialize in mathematics than you may have to divest your self of some prejudices. I assure you that my students found this approach far more digestible than the standard courses which I was teaching. Of course, they had no prejudices of any sort. One of the most unorthodox aspects of this text is the use of Socratic dialogues. When I was a youngster, this method made a lasting impression on me. It has a way of illuminating the mathematical way of thinking and anticipating the most common mistakes made by students. Another advantage of a dialogue is that it readily lends itself to an heuristic approach. Since Polya, everyone has paid lip service to this method of teaching but very few have delivered on it. At the unfortunate price of being somewhat verbose, I have attempted to convey how the actual process of mathematical thinking takes place. The appendices can be skipped completely, used as a reference or as a source for extra material. Covering the whole text would make an excellent honors course. If you view the book as an unfolding story, then the appendices cant be skipped. Often, we reveal the most exciting aspects of the Calculus in these sections. I use exercises, not only to hone technique, but to advance the theory. This is consistent with my belief that the facts learned in this course are relatively unimportant. What is paramount is the method of thinking. By advancing the theory through problems, the student is challenged and receives the same sort of satisfaction as an explorer who uncovers new territory. I provide the solutions to every exercise. These solutions are very elaborate. They are meant to convey proper form, the detailed thinking behind each solution, as well as the actual answer. If the student lacks discipline and will simply look at my solutions, without attempting his/her own, then surely the results will be disastrous. A fundamental aspect of doing exercises is to exorcise the demon of errors. If the student does not do this, then she/he will make her/his rst errors on the exam. Who has not been 17

shown a card trick and has turned around to demonstrate it to a friend? Even if the theory of the trick was well understood, without practice, the demonstration was a failure. In the same way, problem solving must be experienced rsthand. It is not enough for a mathematics text to contain all the right theorems. In a rather Zen-like way, it must also tell an exciting story. It should be a Dickensian novel with heroes, full of subplots and interwoven relationships culminating in a unied conclusion. It is up to the reader to decide which sections of this story represent periods of tension, which are relatively tranquil. I can assure you that, while writing this book, I enjoyed such emotions. This book contains actual mathematics and its treatment is far from supercial. I not only describe creative reasoning, I expect it. I believe that my exposition is logically transparent. Sadly, we live in an age when logical economy and clarity are of not much value. If I had gone through the present educational system, I would have hated this subject which is so dear to me. Hopefully, this whole exercise achieves more than the easing of my conscience. $ P. Hajdu, Dec. 27, 1994

18

CONVENTIONS LEXICAL CONVENTIONS. We use Arabic numerals (they should be called Indian numerals) for chapter numbers and Roman numerals for appendix numbers Convention for enumerating theorems:$ x.y x is the chapter or appendix number, and y is the theorem number within the chapter. Examples:$ 10.5 and IV.2 These are respectively Chapter 10 Theorem 5 and Appendix IV Theorem 2. Convention for enumerating problems:$ x.y.z... x is the chapter number. y is the exercise set within the chapter. z is the problem within the exercise set. Examples:$ 10.5.3c and IV.2.1 The rst example refers to chapter 10, exercise set 5, problem 3. Apparently, problem 3 has several parts, and this is part c. The second example refers to Appendix IV, Exercise Set 2, Problem 1. Sometimes, we may refer to, say Problem (b). When this is done, it is assumed that we are referring to the present problem, part b, in the present exercise set. If * follows a problem number, then the problem is referred to in the rest of the text. Usually, this means that the solution of the problem is important for further comprehension. If ** follows a problem, then the problem is essential for further comprehension. Convention for enumerating examples:$ x.y x is the chapter number, y is the example number within the chapter. Examples:$ 7.4b and IV.3 The rst refers to chapter 7 example 4 part b. The second refers to example 3, in appendix IV. Convention for enumerating diagrams:$ x.y x is the chapter number, y is the diagram number within the chapter. Examples:$ 3.4a and IV.1 The rst example refers to Diagram 4a, in Chapter 3. The second example refers to Diagram 1 in Appendix IV. Convention for enumerating formulae or line numbers:$ x.y Line numbers, which may or may not contain formulae, are always in bold. x is the chapter number, and y is the line number within the chapter. Example:$ 5.3 or V.4 The rst example refers to line number 3, in chapter 5. The second is line 4 in appendix V. Convention for referring to a theorem, a problem or a line number: We place all such reference in square brackets. Example: [Theorem 2.3] or [Problem (5.2.6)] or [5.3] The rst example is a reference to Theorem 3, in chapter 2. The second refers to Problem 6, in exercise set 2, of chapter 5. The last example is a reference to line 3 in chapter 5.

19

20

NOTATIONAL CONVENTIONS

Standard and semi-standard notation. [a,b] represents the closed interval {x | x R, and a x b}. We do allow the closed interval [a,a] = {a}. We do not allow [a,b] where a > b. In general, we assume that an interval of any type is non-empty. If (a,b) represents an interval (as opposed to an ordered pair), then it means {x | x R, and a < x < b}. Since (a,a) is an empty interval, we do not allow it. Similarly, we do not allow (a,b) where a > b. x, z, w, shall represent an increment in the respective variable x, z, w, In other words, the expression (z+z) is best conceptualized as a number which is z away from the number z. z may be thought of as a directed distance. If z > 0, then (z+z) is at a distance of |z|, to the right of z. If z < 0, then (z+z) is at a distance |z| to the left of z. (Absolutely nothing wrong will happen if you view z as just a variable. However, thinking of it as an increment is usually the best paradigm.) Suppose that a function f, is a subset of the Cartesian product DXR. We call D the domain of the function. Every element of D must appear, as a rst coordinate, in one and only one ordered pair of f. R is the range of the function. The collection C, of second coordinates of the ordered pairs of f is the codomain of f. Note that C R. Some text reverse our meaning of range and codomain. We prefer our notation, because it is more intuitive. When cows are allowed to graze on a certain range, it is not implied that they will eat every blade of grass on that range. The whole range represents a potential feeding ground. Similarly, we make the range of f large enough that it contains all the second elements of ordered pairs. Since the domain gives rise to the set of second elements, it is most natural to call this collection the codomain. (I list this among the semi-standard notations, because the number of texts which take the same approach as I do is in a minority.) The function f, is said to be onto or surjective if and only if its codomain and range are equal. Using the above notation, C = R. f is said to be one-to-one (1-1) or injective if and only if no second coordinate is repeated. If f is both surjective and injective, then it is bijective or a 1-1 correspondence. I favor the expression injective, over the expression one-to-one, because oneto-one is easily confused with one-to-one correspondence. In general, because of their brevity and consistency, I prefer injective, surjective and bijective to their counterparts. (Again, I did not invent these concepts, but they are rarely seen in introductory texts.) If L is some subset of the domain of a function f, then f(L) = {y | y = f(x) for at least one x L}. Put simply, f(L) is the collection of all codomain elements which correspond to the domain elements of L. (Note that, to a beginner, codomain seems more reasonable to use here than range.) When a function is expressed by some sort of equality relation and no explicit domain is stated, then we assume that the domain is the largest possible subset of R on which the relation is well dened. If no range is stated, then we assume that the codomain is the range, i.e. the function is surjective. 21

~ will mean approximately. For example, ~ 3.14. The word small shall mean close to 0. Though -10000 is smaller than -0.04, we shall refer to numbers, such as -10000, as negatively large. Since -0.04 is close to 0, we may say that it is small. In other words, small will be an absolute value concept. Some nonstandard notation. There were many improvements and additions, to standard notation, which I was tempted to consider. However, I think it would be highly arrogant of me to introduce anything unless there was a blatant need for it. With this in mind, I have introduced just two new notations. In the Calculus, we speak often of a number, say c, being between two other numbers, say b and e. Unfortunately, this simple concept can have many mathematical interpretations. If b < e then we amy mean b c e or b < c < e. If b > e, the we may mean b > c > e or b c e. Not only is this awkward, but often, it forces a proof to branch into two or more forks. To avoid all this, we shall use the symbol #, as a variable over the set of symbols {<, , >, , =}. As an example, suppose that a and c are end points of an interval and b is in the interval. Then, we may write a # b # c. It is the context which will make it clear whether a # b # c represents a < b < c or a > b > c or a b < c etc. The rst will be appropriate if the interval is (a,c), the second if the interval is (c,a), the third if it is [a,c) etc. A similar problem, to the above, appears in interval notation. Often, we are interested in a closed interval whose, end points are say, z and z+z. If z < 0, the interval is [z+z,z]. Otherwise, it is [z,z+z]. Again, these two cases may force arguments to have two, nearly identical branches. To avoid this, we shall use the notation ]a,b[. Specically, if a b then ]a,b[ = [a,b]. If a b, then ]a,b[ = [b,a]. We shall use the notation ]a,b [ when all that is important is that we have a closed interval with end points a and b, but we do not care (or we do not know) which is the rst end point and which is the second. Of course, we introduce a parallel notation to the above, for open intervals. )a,b( represents the open interval (a,b) when a < b and it is the interval (b,a) if a > b. PLEASE NOTE THAT WE DO NOT USE ]a,b[ TO REPRESENT AN OPEN INTERVAL. In this text, square brackets always indicate closed intervals. Open intervals are denoted by round brackets.

22