## Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

**Advanced Calculus and Analysis MA1002
**

Ian Craw

ii

April 13, 2000, Version 1.3 Copyright © 2000 by Ian Craw and the University of Aberdeen All rights reserved. Additional copies may be obtained from: Department of Mathematical Sciences University of Aberdeen Aberdeen AB9 2TY DSN: mth200-101982-8

Foreword

These Notes

The notes contain the material that I use when preparing lectures for a course I gave from the mid 1980’s until 1994; in that sense they are my lecture notes. ”Lectures were once useful, but now when all can read, and books are so numerous, lectures are unnecessary.” Samuel Johnson, 1799. Lecture notes have been around for centuries, either informally, as handwritten notes, or formally as textbooks. Recently improvements in typesetting have made it easier to produce “personalised” printed notes as here, but there has been no fundamental change. Experience shows that very few people are able to use lecture notes as a substitute for lectures; if it were otherwise, lecturing, as a profession would have died out by now. These notes have a long history; a “ﬁrst course in analysis” rather like this has been given within the Mathematics Department for at least 30 years. During that time many people have taught the course and all have left their mark on it; clarifying points that have proved diﬃcult, selecting the “right” examples and so on. I certainly beneﬁted from the notes that Dr Stuart Dagger had written, when I took over the course from him and this version builds on that foundation, itslef heavily inﬂuenced by (Spivak 1967) which was the recommended textbook for most of the time these notes were used. A The notes are written in L TEX which allows a higher level view of the text, and simpliﬁes the preparation of such things as the index on page 101 and numbered equations. You will ﬁnd that most equations are not numbered, or are numbered symbolically. However sometimes I want to refer back to an equation, and in that case it is numbered within the section. Thus Equation (1.1) refers to the ﬁrst numbered equation in Chapter 1 and so on.

Acknowledgements

These notes, in their printed form, have been seen by many students in Aberdeen since they were ﬁrst written. I thank those (now) anonymous students who helped to improve their quality by pointing out stupidities, repetitions misprints and so on. Since the notes have gone on the web, others, mainly in the USA, have contributed to this gradual improvement by taking the trouble to let me know of diﬃculties, either in content or presentation. As a way of thanking those who provided such corrections, I endeavour to incorporate the corrections in the text almost immediately. At one point this was no longer possible; the diagrams had been done in a program that had been ‘subsequently “upgraded” so much that they were no longer useable. For this reason I had to withdraw the notes. However all the diagrams have now been redrawn in “public iii

iv domaian” tools, usually xfig and gnuplot. I thus expect to be able to maintain them in future, and would again welcome corrections.

Ian Craw Department of Mathematical Sciences Room 344, Meston Building email: Ian.Craw@maths.abdn.ac.uk www: http://www.maths.abdn.ac.uk/~igc April 13, 2000

Contents

Foreword Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Introduction. 1.1 The Need for Good Foundations 1.2 The Real Numbers . . . . . . . . 1.3 Inequalities . . . . . . . . . . . . 1.4 Intervals . . . . . . . . . . . . . . 1.5 Functions . . . . . . . . . . . . . 1.6 Neighbourhoods . . . . . . . . . 1.7 Absolute Value . . . . . . . . . . 1.8 The Binomial Theorem and other 2 Sequences 2.1 Deﬁnition and Examples . . . . 2.1.1 Examples of sequences . 2.2 Direct Consequences . . . . . . 2.3 Sums, Products and Quotients 2.4 Squeezing . . . . . . . . . . . . 2.5 Bounded sequences . . . . . . . 2.6 Inﬁnite Limits . . . . . . . . . . 3 Monotone Convergence 3.1 Three Hard Examples . . . . . 3.2 Boundedness Again . . . . . . . 3.2.1 Monotone Convergence 3.2.2 The Fibonacci Sequence 4 Limits and Continuity 4.1 Classes of functions . . . . . . . 4.2 Limits and Continuity . . . . . 4.3 One sided limits . . . . . . . . 4.4 Results giving Coninuity . . . . 4.5 Inﬁnite limits . . . . . . . . . . 4.6 Continuity on a Closed Interval iii iii 1 1 2 4 5 5 6 7 8 11 11 11 14 15 17 19 19 21 21 22 22 26 29 29 30 34 35 37 38

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Algebra

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . . . .

. . . . . .

. . . . . . v

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

vi 5 Diﬀerentiability 5.1 Deﬁnition and Basic Properties . . . 5.2 Simple Limits . . . . . . . . . . . . . 5.3 Rolle and the Mean Value Theorem 5.4 l’Hˆpital revisited . . . . . . . . . . . o 5.5 Inﬁnite limits . . . . . . . . . . . . . 5.5.1 (Rates of growth) . . . . . . . 5.6 Taylor’s Theorem . . . . . . . . . . .

CONTENTS 41 41 43 44 47 48 49 49 55 55 56 58 61 64 67 67 69 70 72 73 73 73 73 74 77 77 81 84 85 86 90 91 92

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

6 Inﬁnite Series 6.1 Arithmetic and Geometric Series . . . 6.2 Convergent Series . . . . . . . . . . . . 6.3 The Comparison Test . . . . . . . . . 6.4 Absolute and Conditional Convergence 6.5 An Estimation Problem . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

7 Power Series 7.1 Power Series and the Radius of Convergence . . . . . . . . . . . 7.2 Representing Functions by Power Series . . . . . . . . . . . . . 7.3 Other Power Series . . . . . . . . . . . . . . . . . . . . . . . . . 7.4 Power Series or Function . . . . . . . . . . . . . . . . . . . . . . 7.5 Applications* . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.5.1 The function ex grows faster than any power of x . . . . 7.5.2 The function log x grows more slowly than any power of 2 α 7.5.3 The probability integral 0 e−x dx . . . . . . . . . . . 7.5.4 The number e is irrational . . . . . . . . . . . . . . . . . 8 Diﬀerentiation of Functions of Several 8.1 Functions of Several Variables . . . . . 8.2 Partial Diﬀerentiation . . . . . . . . . 8.3 Higher Derivatives . . . . . . . . . . . 8.4 Solving equations by Substitution . . . 8.5 Maxima and Minima . . . . . . . . . . 8.6 Tangent Planes . . . . . . . . . . . . . 8.7 Linearisation and Diﬀerentials . . . . . 8.8 Implicit Functions of Three Variables . Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . x . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

9 Multiple Integrals 9.1 Integrating functions of several variables . 9.2 Repeated Integrals and Fubini’s Theorem 9.3 Change of Variable — the Jacobian . . . . References . . . . . . . . . . . . . . . . . . . . . Index Entries

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

93 . 93 . 93 . 97 . 101 101

. . 12 14 23 4. . . . . . . . Surface plot of z = x2 − y 2 . . . . . . . . . . . . . . . . . . . . . . . . . .3 4. . . .6 . . . . .1 A sequence of eye locations. . . . . .1 44 5. . . . . . . . . . . . . . . . . Graph of the function (x2 − 4)/(x − 2) The automatic graphing routine does not even notice the singularity at x = 2. . . . . . . The function which is 0 when x < 0 and 1 when x ≥ 0. . . . . . . . .1 8. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The lines y = x and y = −x are also plotted. .1 6. . . . . . . . . . . . . . . . . . . . . . . . . . . sin(1/x). . . . . . .2 3. the x . .4 4. . . . .1 2. . . . You can probably see how the discontinuity of sin(1/x) gets absorbed. . A string displaced from the equilibrium position . . . . . The accurate statement of this “obvious” observation is Rolle’s Theorem. . . . . . . . . . A monotone (increasing) sequence which is bounded above seems to converge because it has nowhere else to go! .2 46 57 58 64 78 78 79 80 85 89 6. Graph of a simple function of one variable . . . . . . . . . . Graph of the function sin(x)/x. . . . . .2 6. . . . . . . A picture of the deﬁnition of convergence . . . . . . . . . . . . . . Somewhere inside a chord.axis . . Comparing the area under the curve y = 1/x with the area of the rectangles above the curve . . . . . . . . . . . . . . . . . . . Sketching a function of two variables . . . .5 31 32 32 33 34 5. . . . . . . . . . . . . . . . . . . . .2 8. . . . . The accurate statement of this common-sense observation is the Mean Value Theorem. .2 4.1 4. . . . . . Graph of the function x. . . . . . . . . . . . Graph of the function sin(1/x). . . . . . . . . . . . . . .4 8. the plotting routine gives up near this singularity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 8. . . . . . . . Contour plot of the surface z = x2 − y 2 . . . . . . . . . . . .List of Figures 2. . . . . .5 8. . . . An upper and lower approximation to the area under the curve . . . . . . . . . . . . If f crosses the axis twice. . . . . . vii . . . . . . . . . . . . . . . . it has a jump discontinuity at x = 0. The missing points near are an artifact of the plotting program. . . . . . . . . . . . . . Comparing the area under the curve y = 1/x2 with the area of the rectangles below the curve . . . . . . . . . . . . the tangent to f will be parallel to the chord. somewhere between the two crossings. . . . . . . . . . . . . . . the function is ﬂat. . . . . Here it is easy to see the problem at x = 0. . . . . .3 8. . . . . . . . . . . . . . . . . . Again the automatic graphing routine does not even notice the singularity at x = 0. . . . . . . A dimensioned box . . .

The transformation from Cartesian to spherical Cross section of the right hand half of the solid a and inside the sphere of radius 2a . LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . polar co-ordinates. .4 Area of integration. . . . . . . . .viii 9. . . .1 9. . . . . . . Area of integration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 9. 95 96 99 99 . . . . . . . .2 9. . . . . . outside a cylinder of radius .

there are problems if the theory is accepted uncritically. For example the chain rule can be phrased as df df dy = . don’t panic. However if we consider the following result.Chapter 1 Introduction. dx dy dx and the “quick” form of the proof of the chain rule — cancel the dy’s — seems helpful. and to collect it in one place. because you will be using them all the time.1) a result which certainly does not appear “obvious”. It makes for a boring chapter. because naive arguments can quickly lead to errors. despite this. and we shall prove it towards the end of the course. 1. This chapter contains reference material which you should have met before. volume V and temperature T of an enclosed gas are related. and perhaps should have been headed “all the things you hoped never to see again”. So another aim of this chapter is to repeat the ideas. any of the books mentioned in the book list can give you more information. in which the pressure P . even though it is in fact true. And ask in tutorial if you don’t understand something here. and the ﬁrst tutorial sheet is designed to give you practice. you usually get a second chance. so you can easily look back and check things when you are in doubt. You are aware by now of just how sequential a subject mathematics is. However. 1 . However I am only emphasising things that you will be using in context later on. ∂V ∂T ∂P (1. and there is no real discussion about its viability as a theory.1 The Need for Good Foundations It is clear that the calculus has many outstanding successes. It is here both to remind you that you have. Indeed you will ﬁnd there are a number of ideas here which it is essential you now understand. If there is material here with which you are not familiar. we have ∂P ∂V ∂T = −1. If you don’t understand something when you ﬁrst meet it.

In this course you will start to see why this complication is necessary. ±n. .. R — the Reals are deﬁned in a much more complicated way. Another example comes when we deal with inﬁnite series. 2. the same set without 0. We hope to show that the complication of introducing the “extra” reals such as 2 is worthwhile because it gives simpler results. it is a real number that is not rational). . as you use the distinction between R and Q . it can be ignored. One point of this course is to illustrate the diﬀerence between Q and R . Contrast these with the positive integers. . However. }. ±2. INTRODUCTION. and explains why in the rest of this section we need to revise elementary notions. }. and as such can’t distinguish between the two √ sets. . Properties of R We summarise the properties of R that we work with. . . recall that 2 ∈ R \ Q (i.2 CHAPTER 1.√ and each inclusion is proper. It is this need for care. Addition: We can add and subtract real numbers exactly as we expect. and the usual rules of arithmetic hold — such results as x + y = y + x. that motivates much of this course.e. . to ensure we can rely on calculations we do. It is subtle: for example when computing. illustrates why we emphasise accurate argument as well as getting the “correct” answers. Z — the Integers are deﬁned as the set {0. . 1. Q — the Rational numbers are deﬁned as the set {p/q : p. . q ∈ Z... in increasing order of complication: N — the Natural numbers are deﬁned as the set {0. . . q = 0}.. n. . and this clearly adds up to half of the previous sum — or log(2)/2. . ±1.2 The Real Numbers We have four inﬁnite sets of familiar objects. . Note: We have a natural inclusion N ⊂ Z ⊂ Q ⊂ R . The only inclusion in any doubt is the last one. 2 3 4 5 6 7 8 9 10 adds up to log 2. an apparently simple re-arrangement gives 1 2 1 + 4 1 1 − 3 6 1 + 8 1 1 − 5 10 1− − − . . We shall see later on that the series 1− 1 1 1 1 1 1 1 1 1 + − + − + − + − . 1. because a computer always works with a rational approximation to any number.

so for us 0/0 is meaningless. Its eﬀect is that there are always “enough” numbers to do what we want.2. Note that we write a ≥ 0 if either a > 0 or a = 0. where we write a < 0 instead of the formally correct 0 > a. we write a > b whenever a − b > 0. even in the “funny” case when a = 0. negative or zero. the rest of your argument may need to be split into the two cases when x = 0 and x = 0. Here it is even possible to make the mistake with numbers. we are simply saying that a number is either positive. Formally these two properties say that (algebraically) R is a ﬁeld. multiplication and division behave as we expect. which in contrast is much more mysterious — it is complete. Normal algebraic manipulations can be done without comment. In fact there are (uncountably many) more . the same is not true of an inequality. Multiplication The order behaves as expected with respect to multiplication: if a > 0 and b > 0 then ab > 0. Completion The set R has an additional property. If you don’t know whether x = 0 or not. . the product of positives is positive. even those like x2 = 2 which can’t be solved in Q . If multiplying by an expression. certainly not rational. exactly one of a > 0. More generally. but two cases arise when more care is needed: Never divide by a number without checking ﬁrst that it is non-zero. although it is perfectly sensible to multiply an equality by a constant. Example. then of course 2x > 2y. So we have rules such as a(b + c) = ab + ac. we have (−2)x < (−2)y. Order As well as the algebraic properties. Never multiply an inequality by a number without checking ﬁrst that the number is positive. although it is not essential at this stage to know the terminology. One aim of the course is to emphasise accurate explanation. One reason for looking carefully at the properties of R is to note possible errors in manipulation. but if you divide by x. It is this property that distinguishes it from Q . Note that we can divide by any number except 0. are also in R. i.all the numbers like π. 1. usually written as “a > 0” or “≥”. R has an ordering on it. why x = 0.1. We explore this property during the course. We make no attempt to make sense of a/0. and interact with addition and subtraction in the usual way.e. you should always say. There are three parts to the property: Trichotomy For any a ∈ R. Of course we know that 2 is non zero.e. If x > y.1. Show that if a > 0 then −a < 0. in words. a = 0 or a < 0 holds. then again it may be necessary to consider diﬀerent cases separately. Addition The order behaves as expected with respect to addition: if a > 0 and b > 0 then a + b > 0. but in fact not even an algebraic number. the sum of positives is positive. THE REAL NUMBERS 3 Multiplication: In the same way. at least the ﬁrst time. Thus there are enough to solve any algebraic equation. so you don’t need to justify dividing by 2. However. i. and if a < 0 then −a > 0.

then a + (−a) is the sum of two positives and so is positive. whereas calculating directly can prove either impossible. Solution. and so bc − ac > 0 or bc > ac as required. Find {x ∈ R : (x − 2)(x + 3) > 0}.4 CHAPTER 1. Even at this simple level. Thus using the multiplication rule. Show that if a > b and c < 0. It often turns out that we can “give something away” and still get a useful result. This is not very interesting. So the only consistent possibility is that −a < 0. try rather to work through systematically. or (−a) > 0. Show that if a < 0 and b < 0. 1. and so we have to deal with inequalities. we can produce some interesting results. 1. in which case both x > 2 and x > −3. 2 . Proposition (Arithmetic . Assume the result is false. The hard bit about doing this is in Example 1. and is here to remind you that the order in which questions are asked can be helpful. If this latter holds. but is here to show how to use the properties formally. Find {x ∈ R : x2 − x − 2 < 0}. it is here to emphasise the need for care. The other part is essentially the same argument. Example. Solution. then ac < bc. If a ≥ 0 and b ≥ 0 then a+b √ ≥ ab. we see that −c > 0 and a − b > 0.6.4. This also isn’t very interesting. It may dissuade you from dipping into a sheet.1. We have just looked at the rules for manipulating the order relation. then by trichotomy. so x > 2. Note that if the product of two numbers is positive then either both are positive or both are negative. so x < −3. where the next question uses the result of the previous one. Suppose (x − 2)(x + 3) > 0. then ab > 0. we have (a − b)(−c) > 0.3. Solution. But a + (−a) = 0. 1. Thus {x : (x − 2)(x + 3) > 0} = {x : x > 2} ∪ {x : x < −3}.3 Inequalities One aim of this course is to get a useful understanding of the behaviour of systems.1 in the case a = −c. Example. Exercise. INTRODUCTION. Applying Example 1. Think of it as trying to see the wood. 1. or at best unhelpful.Geometric mean inequality). can we perhaps ignore a term because it is small and simplify things? In order to to this we need to estimate — replace the term by something bigger which is easier to handle. 1. when our detailed calculations tell us about individual trees. or x − 2 < 0 and x + 3 < 0.5. in which case both x < 2 and x < −3. This is an idea you will ﬁnd a lot in example sheets.2. −a = 0 (which is false because we know a > 0). 1. This section is probably all revision. Exercise. and by trichotomy 0 > 0 is false. So either x − 2 > 0 and x + 3 > 0. we may want to know roughly how a function behaves. For example.

it should be clear from the context whether this is being allowed. if you require in addition that I1 is centred on 1. We say that I ⊂ R is an open interval if I = (a. just for emphasis. b) ∪ {a} ∪ {b}. 3] = (1. show there is always an interval I with l ∈ I and m ∈ I. In contrast a closed interval contains both its end points. we have x2 ≥ 0 (why?). Write down an interval I with 2 ∈ I such that 1 ∈ I and 3 ∈ I. 5 a+b √ Since a ≥ 0 and b ≥ 0. so (a − b)2 ≥ 0. while the example (1. Exercise.7. and is of the form I = [a. .5 Functions Recall that f : D ⊂ R → T is a function if f (x) is a well deﬁned value in T for each x ∈ D.7. INTERVALS Solution. This is the arithmetic 2 . Of course we can easily get to more general subsets of R. 1. we have ≥ ab. T is the target space and f (D) = {f (x) : x ∈ D} is the range of f . 4) shows that the union of two intervals need not be an interval.8. 2) ∪ [2. b). the interval is sometimes called a ﬁnite interval.open intervals like (a. It is also sometimes useful to have half .geometric mean inequality. b] = {x ∈ R : a ≤ x ≤ b}. 1. b) = {x ∈ R : a < x < b}. It is sometimes convenient to allow also the possibility a = −∞ and b = +∞. 2 ∈ I2 and I1 ∩ I2 = ∅.4. Exercise. Thus an open interval excludes its end points. We study further work with inequalities in section 1. 3] shows that the union of two intervals may be an interval. The two end points a and b are points in R . Thus a2 − 2ab + b2 ≥ 0. 2) ∪ (3. Can you still do this.4 Intervals We need to be able to talk easily about certain subsets of R . b] = (a. Write down a pair of intervals I1 and I2 such that 1 ∈ I1 . It is trivial that [a. Can you ﬁnd the largest such interval? Is there a largest such interval if you also require that I is closed? Given l and m with l = m. So (1. taking square roots. 1. For any value of x. We say that D is the domain of the function. If these extensions are being excluded.1. b] and [a. but contains all the points in between. I2 is centred on 2 and that I1 and I2 have the same (positive) length? What happens if you replace 1 and 2 by any two numbers l and m with l = m? 1. (a + b)2 ≥ 4ab. a2 + 2ab + b2 ≥ 4ab.

b). we usually replace the . 2] → R deﬁned by f2 (x) = x2 are formally diﬀerent functions. Note ﬁrst that the deﬁnition says nothing about a formula. And the deﬁnition can be complicated. The point we have excluded. b). we are really looking at an interval (a − d. so changing the domain technically gives a diﬀerent function.] In the simplest examples. Exercise.6 Neighbourhoods This situation often occurs. to avoid dividing by zero. This illustrate our ﬁrst use of intervals. This distinction will start to be important in this course. we don’t want to worry about the singularity at x = −a when we are discussing the one at x = a (which is actually much better behaved). but sometimes it is useful. For example if f (x) = x−a x−a = x2 − a2 (x − a)(x + a) for x = a.9. such as f (x) = x. If we only look at the points distant less than d for a. since we cannot deﬁne the square root of a negative number. we call such an interval a neighbourhood of a. we can always restrict the domain of f to an interval I to get a new function. for example f (x) = 0 if x ≤ a or x ≥ b. Write down the natural domain of deﬁnition of each of the functions: f (x) = x2 x−2 − 5x + 6 g(x) = 1 .values.6 CHAPTER 1. and is zero elsewhere [and is usually called the characteristic function of the interval (a. So f1 : R → R deﬁned by f1 (x) = x2 and f2 : [−2. but even √ for relatively simple cases. We need to be able to talk about a function near a point: in the above example. then since x = a we can cancel to get f (x) = (x + a)−1 . Thus the function f (x) = 1 x−3 has domain {x ∈ R : x = 3}. This is of course a diﬀerent representation of the function. we need to restrict to a smaller domain. like f (x) = x2 the domain of f is the whole of R. 1 if a < x < b. in this case the domain D is {x : x ≥ 0}. Note that the domain is part of the deﬁnition of a function. sin x Where do these functions have singularities? It is often of interest to investigate the behaviour of a function near a singularity. INTRODUCTION. 1. 1. Given f : R → R . deﬁnes a function on the whole of R. a + d). just that the result must be properly deﬁned. Another natural situation in which we need to be careful of the domain of a function occurs when taking quotients. in the above case 3 is sometimes called a singularity of f . even though they both are “x2 ” Note also that the range of f2 is [0. Mostly this is trivial. at least if we want the function to have real . and provides an indication as to how f may be extended through the singularity at a — by giving it the value (2a)−1 . 4]. which has the value 1 on the open interval (a. so that T ⊂ R . For traditional reasons.

For any x. given l = m there is some (suﬃciently small) δ such that we can ﬁnd disjoint δ . |x − y| ≥ |x| − |y| . so a δ . Proposition.7 in this language. −x if x < 0.11.7 Absolute Value Here is an example where it is natural to use a two part deﬁnition of a function.6. We sometimes call this “unwrapping” the modulus. y ∈ R . To see this. |x + y| ≤ |x| + |y|.1. Exercise. Show that an open interval contains a neighbourhood of each of its points. consider: 1. Since −|x| ≤ x ≤ |x|. y ∈ R . combining these we have −|x| − |y| ≤ x + y ≤ |x| + |y| and this is the same as the required result.). ABSOLUTE VALUE 7 distance d by its Greek equivalent. z ∈ R .neighbourhood of a consists of all points which are closer to a than δ. We can rephrase the result of Ex 1. Show that for any x. 1.13. Proof. we interpret |x − y| as the distance between x and y. 1. It’s particular use is in describing distances. If δ > 0 we call the interval (a − δ. For any x. Proposition (The Triangle Inequality.7.12. |x − z| ≤ |x − y| + |y − z|. and speak of a distance δ. We write |x| = x if x ≥ 0. it is simply necessary to show the corresponding pair of inequalities. Thus (a − δ. The signiﬁcance of a neighbourhood is that it is an interval in which we can look at the behaviour of a function without being distracted by other irrelevant behaviours. Note that we can always “expand out” the inequality using this idea.neighbourhoods of l and m. This is the absolute value or modulus of x. 1. a + δ) = {X ∈ R : |x − a| < δ}. . conversely. It usually doesn’t matter whether δ is very big or not. and the same holds for y. We use this result in Prop 2. √ An equivalent deﬁnition is |x| = x2 .10. we can rewrite this without a modulus sign as the pair of inequalities −k < x − y < k. in order to establish an inequality involving the modulus. y.neighbourhood) of a. a + δ) a neighbourhood (sometimes a δ . 1. Exercise. So if |x − y| < k.

k with corresponding special cases. Describe the set {x ∈ R : 1 ≤ x ≤ 3} using the absolute value function.. 1. Using 1. Multiplying this inequality by −1 and combining these we have −|x − y| ≤ |x| − |y| ≤ |x − y| and this is the required result. o gives |y| − |x| ≤ |x − y|. (1 + x)4 = 1 + 4x + 6x2 + 4x3 + x4 . Proof. INTRODUCTION. From one inequality we get 5x < −4 + 3. in fact it holds appropriately for more general exponents as we shall see in Chapter 7 Another simple algebraic formula that can be useful concerns powers of diﬀerences: a2 − b2 = (a − b)(a + b). . 1. Example. + x + . 1.. or x > 7/5. We have (1 + x)n = 1 + nx + n. (1 + x)5 = 1 + 5x + 10x2 + 10x3 + 5x4 + x5 . There is a more general form: (a + b)n = an + nan−1 b + n. .2. 1. .12 we have CHAPTER 1.16.(n − 1) 2 n. |x| = |x − y + y| ≤ |x − y| + |y| and so |x| − |y| ≤ |x − y|.. . or 5x − 3 > 4. Exercise. Formally this result is only valid for any positive integer n. .(n − 1). Thus {x ∈ R : |5x − 3| > 4} = (−∞.(n − 1) n−2 2 n. Describe {x ∈ R : |x + 3| < 1}..(n − 1). . . a3 − b3 = (a − b)(a2 + ab + b2 ). . or x < −1/5. a a 1. Unwrapping the modulus. the other inequality gives 5x > 4 + 3. and noting that |x| = | − x|.2 1. ∞). .k Recall in particular a few simple cases: (1 + x)3 = 1 + 3x + 3x2 + x3 .2 1.(n − k + 1) k x + . we have either 5x − 3 < −4.14. Exercise. a4 − b4 = (a − b)(a3 + a2 b + ab2 + b3 ) . the binomial theorem gives an expansion of (1 + x)n for any positive integer n. + b + . Describe {x ∈ R : |5x − 3| > 4}. + xn . . .8 Proof. Interchanging the rˆles of x and y.2. 1. . −1/5) ∪ (7/5.8 The Binomial Theorem and other Algebra At its simplest.(n − k + 1) n−k k b + .15. + bn .

1. . Note that we made use of this result when discussing the function after Ex 1. THE BINOMIAL THEOREM AND OTHER ALGEBRA and in general. + c− . . we have an − bn = (a − b)(an−1 + an−2 b + an−3 b2 + . + ab n − 1 + bn−1 ). And of course you remember the usual “completing the square” trick: b b2 ax2 + bx + c = a x2 + x + 2 a 4a =a x+ b 2a 2 9 +c− b2 4a b2 4a .9.8.

10 CHAPTER 1. INTRODUCTION. .

and in fact we usually start labeling a sequence from 1 rather than 0.1 Examples of sequences The most obvious example of a sequence is the sequence of natural numbers. − . 3 1. 3.14159. for example by enumerating them as 0. we usually write a1 . . 2. A (real inﬁnite) sequence is a map a : N → R Of course if is more usual to call a function f rather than a. 3 an = (−1)n+1 an = 3 A sequence doesn’t have to be deﬁned by a sensible “formula”. .1 Deﬁnition and Examples 2. even though we usually write functions in this way.141592 .Chapter 2 Sequences 2. −2 . 2. 3. 2 3 4 Limit does not exist (→ ∞) 0 does not exist (the sequence oscillates) 0 1 does not exist (the sequence oscillates) 3 an = (−1)n+1 an = (−1)n+1 an = n−1 n 1. If a : N → R . Deﬁnition. second member and so on. . Here is a sequence you may recognise:3. .14. − 2 3 4 3. 1 1 1 . . 3.1.1415. 3. . 3. 1. 1. 2. instead of the more formal a(1). . . 11 3. −1 1 1 1 1. it doesn’t really matter. 3. − 2 3 4 1 2 3 0. − . . 2 3 4 1 2 3 0.1. 1. although we can turn them into a sequence in many ways.141.1. . Here are some more sequences: Deﬁnition an = n − 1 an = 1 n 1 n n−1 n First 4 terms 0. a2 and so on. −1. What the deﬁnition is saying is that we can lay out the members of a sequence in a list with a ﬁrst member. a(2). −1. . Note that the integers are not a sequence. 3.

and interest in their limits.] 5 Note also that we need to specify the accuracy needed. Usually we are interested in what happens to a sequence “in the long run”. [You can check that a3 = 2 to 6 decimal places.1: A sequence of eye locations. In that case. ﬁnding a root of this equation is solving the equation x3 = 2. ﬁnding 2 In this case. f (a) Thus a better approximation than a to the root is a + h = a − f (a)/f (a). There is no single approximation √ to 3 2 or π which will always work. In order to use such a sequence of approximations. in Newton’s method. it is ﬁrst necessary to specify an acceptable accuracy. If we take f (x) = x3√ 2. If now we almost have a solution. so that f (a + h) = 0.2599 In fact the sequence does converge to 3 2. . so f (a) ≈ 0. whether we are measuring a ﬂower bed or navigating a satellite to a planet. we get the sequence deﬁned as follows 2 2 a1 = 1whilean+1 = an + 2 3 3an Note that this makes sense: a1 = 1. or what happens “when it settles down”. which is a true solution. we can try to perturb it to a + h.1) 2 2 etc. arise naturally in many situations. The interest is whether the sequence oscillates predictably.1 + 3 3. and we then often speak of an . Often we do this by specifying a neighbourhood of the limit. or in the limit of the sequence. a4 = 1. Sequences. a2 = if n > 1.neighbourhood. One such occurs when trying to solve equations numerically. (2. we have 0 = f (a + h) = f (a) + h. we get a2 = 1. we use the standard calculus approximation.2639. Calculating. In the examples above this was fairly easy to see. In Fig 2.2599 and a5 = 1.12 √ a3 = 1. . Of course we can graph a sequence. where we use (for error). SEQUENCES where the terms are successive truncates of the decimal expansion of π.f (a) and so h ≈ f (a) . rather than δ (for distance).1 we show a sequence of locations of (just the x coordinate) of a car driver’s eyes. by taking enough terms we can get an approximation that is as accurate as we need. So we are usually interested in what happens when n → ∞. that f (a + h) ≈ f (a) + h. 66 64 62 60 58 56 54 52 0 5 10 15 20 25 30 35 Figure 2. and it sometimes helps.12 CHAPTER 2. − 3 in other words.333.f (a).

eventually it will be 1994. −1. In Fig 2. |an − l| < > 0 eventually an is closer n ≥ N. that it will frequently be Friday.t. . 2. as long as it is still centred on the potential limit. A sequence which converges to some limit is a convergent sequence. Say a property P (n) holds eventually iﬀ ∃N such that P (n) holds for all n ≥ N . and eventually you will die.2. you may ﬁnd that it picks up as convergent some sequences you don’t want to be convergent. The sequence sin(n!π/17) is eventually zero. we can rephrase the deﬁnition of convergence as We say that an → l as n → ∞ iﬀ given any error to l then . The deﬁnition then says the sequence is convergent to the number we have shown as a potential limit. DEFINITION AND EXAMPLES 13 2. we give a picture that may help. Symbolically we have > 0 ∃N s.1. 2. I need to assume the sequence is inﬁnite. given > 0 there is some N such that |an − l| < whenever n ≥ N. there is some n ≥ N such that P (n) holds. You will ﬁnd. frequently be raining (or sunny).neighbourhood of the (potential) limit l is represented by the shaded strip.neighbourhood of l. Deﬁnition.3. talk it over with a tutor. 1. We call the n a witness. this time we use the idea of neighbourhood: We say that an → l as n → ∞ iﬀ given any (acceptable) error an is in the .4. A more diﬃcult one is whether Newton’s work will eventually be forgotten! Using this language. according to the deﬁnitions. Deﬁnition. } is eventually positive. In contrast. Deﬁnition. whenever Another version may make the content of the deﬁnition even clearer. Some examples using the language are worthwhile. but properly I am just interested in divergence at present. you can decide for yourself whether this is a philosophical statement. It holds frequently iﬀ given N . We sometimes speak of a sequence oscillating or tending to inﬁnity. provided the sequence is eventually in the shaded strip: and this must be true even if we redraw the shaded strip to be narrower. .2. Say that a sequence {an } converges to a limit l if and only if.2. The sequence {−2. a statement about the future of the universe. If you think you have an easier version. while individual members an of the sequence are shown as blobs. 0. it witnesses the fact that the property is true somewhere at least as far along the sequence as N . and even frequently February 29. > 0 eventually It is important to note that the deﬁnition of the limit of a sequence doesn’t have a simpler form. or just plain optimism! 1 . the sequence of natural numbers is frequently prime. It may help you to understand this language if you think of the sequence of days in the future1 . The . A sequence which is not a convergent sequence is divergent. 2. .

an is within a distance 1 of 2. We can argue this directly (so this is another version of this proof). • Let an = 1/n. thus n < l + 2 for all n. this is the required contradiction. 2. for if not. . Then an → 0 as n → ∞. Taking = 1. Then l = m.g.6. √ 2. we see that eventually (say after N ) . Let an → l as n → ∞ and assume also that an → m as n → ∞. • If an → 2 as n → ∞. Pick = |l − m|/2. Proof. we have −1 ≤ (n − 1) − l < 1. it has a unique limit. SEQUENCES Potential Limit n Figure 2. then (take = 1).7. eventually |an − m| < . So rather than always working directly. which is a contradiction. Show that the sequence an = (1/ n) → 0 as n → ∞. Here is a simple tool (or Proposition). In other words. then there is some l such that an → l as n → ∞. we argue by contradiction. We have 0≤ 1 1 ≤ < n N by choice of N . we choose disjoint neighbourhoods of l and m.5.2 Direct Consequences With this language we can give some simple examples for which we can use the deﬁnition directly. so this holds e. eventually. Suppose that l = m.14 CHAPTER 2. and note that since the sequence converges. Then eventually |an − l| < . we also use the deﬁnition to prove some general tools. Exercise.2: A picture of the deﬁnition of convergence 2. Using 1. pick with N > 1/ . and we are justiﬁed in talking about the limit of a sequence. showing this situation is impossible. and in particular. Also. that (n − 1) − l < 1 for all n ≥ N . most of the time it is too hard. and then use the tools to tell us about convergence or divergence. One consequence of this is that eventually an > 1 and another is that eventually an < 3. Proposition. if a sequence has a limit. for n ≥ N1 . Although we can work directly from the deﬁnition in these simple cases.. Now suppose that n ≥ N . To check this. > 0 and then choose N • The sequence an = n − 1 is divergent. eventually it lies in each of these neighbourhoods.

2. Show that eventually an > 0. and choose n ≥ N . for n ≥ N2 . (New Convergent sequences from old) Let an → l and bn → m as n → ∞.8. we are guaranteed that from some point onwards. and . Example. Then there is some N such that |an − l| < |l|/2 Now for all n ≥ N so |l| ≤ |l|/2 + |an |. the denominator to 1 + 0 and so the quotient to (1 + 0)/(1 + 0) = 1. Thus |l| ≤ |l − an | + |an |. and |l − m| = |l − an + an − m| ≤ |l − an | + |an − m| by the triangle inequality < + = |l − m| 2.9. Remember what this means. PRODUCTS AND QUOTIENTS 15 so this holds eg. Proof. and apply the deﬁnition of “an → l as n → ∞”. Then Sums: an + bn → l + m as n → ∞. SUMS. Proposition. and assume that l > 0. The proof is a variant of “if an → 2 as n → ∞ then eventually an > 1. l = l − an + an . Products and Quotients n+2 . 2. Then eventually an = 0. Show that an → 1 as n → ∞. n+3 Solution. But it is not obvious that our deﬁnition does indeed behave as we would wish. Here they are:2. we need rules to justify what we have done. Let an → l = 0 as n → ∞.7. There is an obvious manipulation here:2. use the ﬁrst method suggested above for l > 0. But we can do it all in one: Take = |l|/2.” One way is just to repeat that argument in the two cases where l > 0 and then l < 0. Theorem.3 Sums. In other words. we never have an = 0. n+3 1 + 3/n We hope the numerator converges to 1 + 0. Let an → l = 0 as n → ∞. and |an | ≥ |l|/2 = 0.2. Now let N = max(N1 . Exercise. Let an = an = n+2 1 + 2/n = .10. Products: an bn → lm as n → ∞. N2 ). Then both inequalities hold.3.

while the left hand side converges to l. using “product of convergents is convergent”. Using Theorem 2. A helpful manipulation is easy. 4 Using the corresponding result for sums shows that n2 − 7 → 0 − 7 as n → ∞. This gives: 2. In the equation 2 2 an+1 = an + 2 3 3an we now let n → ∞ on both sides of the equation.6. √ 2. we have |an + bn − (l + m)| < |an − l| + |bn − m| < /2 + /2 = . Assuming that {an } is convergent.12.(1/n) and 1/n → 0 as n → ∞. and in the same way. Example. an+1 = (4an + 2)/(an + 3) for n ≥ 1. Now because First pick > 0. Thus 2 2 l= l+ 2 3 3l and so l3 = 2.1 we derived a sequence (which we claimed converged to 3 2) from Newton’s method.10.14. the denominator → 1 as n → ∞. Since 1/n2 = (1/n). since the limit of the denominator is 1 = 0. Pick > 0. CHAPTER 2. 2. 2.13. Let an = an = 4 4 − 7n2 2 − 7 = n 3 . We can now show that provided the limit exists and is non zero. there is some N2 such that |bn − m| < /2 whenever n > N2 . Proof. Show that an → −7 as n → ∞. Exercise. Exercise. n2 + 3n Solution. We choose to divide both top and bottom by the highest power of n around. Proofs can be found in (Spivak 1967). an+1 = (2an + 2) for n ≥ 1. ﬁnd its limit. The other two results are proved in the same way. Deﬁne the sequence {an } by a1 = 1. SEQUENCES Note that part of the point of the theorem is that the new sequences are convergent. But they are 3 3l the same sequence. Proof. In the same way. n2 + 3n 1+ n We now show each term behaves as we expect. Since an → l as n → ∞. we see that 1/n2 → 0 as n → ∞. but are technically more diﬃcult.11. Thus by the “limit of quotients” result. we see that the 2 2 right hand side converges to l + 2 . Is the sequence convergent? . we must ﬁnd N such that |an + bn − (l + m) < when n ≥ N . N2 ). Assuming that {an} is convergent. there is some N1 such that |an − l| < /2 whenever n > N1 . ﬁnd its limit. Deﬁne the sequence {an } by a1 = 1. the quotient → −7 as n → ∞. 4 − 7n2 . and n > N . Example. In equation 2. so both limits are the same by Prop 2. √ the limit is indeed 3 2. Note ﬁrst that if an → l as n → ∞.16 Inverses: provided m = 0 then an /bn → l/m as n → ∞. then we also have an+1 → l as n → ∞. Then if N = max(N1 .

Lemma. √ √ n+1+ n = (n + 1) − n 1 √ √ = √ √ → 0 as n+1+ n n+1+ n n → ∞. this gives − < bn − l < or |bn − l| < as claimed.16. N2 ). we can’t take the last step yet. we can ﬁnd N1 such that |an − l| < for n ≥ N1 and since cn → l as n → ∞. and using only the middle and outer terms. but also shows why we need the dictionary. A simple piece of algebra gets us most of the way: √ √ √ √ n+1+ n an = n+1− n . note the result. then l ≤ m. 2. Now pick N = max(N1 . We really want a good dictionary of convergent sequences. and note that. SQUEEZING 2.4. and come back to the proof later. and suppose that an → l and cn → l as n → ∞.15. and bn → l as n → ∞. 17 Proof. Given that an → l and bn → m as n → ∞. and cn − l < . but you may still ﬁnd these techniques hard. Show that an → 0 as n → ∞. in particular. 2. Compare this with the next result. if so. . (The Squeezing lemma) Let an ≤ bn ≤ cn . Proof. The next results show that order behaves quite well under taking limits. The ﬁrst one is fairly routine to prove.4 Squeezing Actually. Exercise. Example.17. we have − < an − l Using the given order relation we get − < an − l ≤ bn − l ≤ cn − l < . and that an ≤ bn for each n. Let an = √ √ n + 1 − n.2. It is true and looks sensible. we can ﬁnd N2 such that |cn − l| < for n ≥ N2 . 2. Pick > 0. but it is another case where we need more results getting new convergent sequences from old. Then since an → l as n → ∞. The {bn } is convergent. where we can also deduce convergence.

Now 2 n n 1 → 0. Now 2 n n log n n 1 → 0 and n 1 → 0. n log n . We have 1+ 1 1 0≤ √ √ ≤ √ . and suppose that an → a as n → ∞. Let an = − 1 1 < an < 2 .5. This illustrates the need for a good bank of convergent sequences.18. it is clear we can state an “eventually” form of this result. 2. n2 so sin(n) →0 n2 as n → ∞. 2. We use the squeezing lemma: 2.21. 2.18 CHAPTER 2. but here is the results. We don’t need the inequalities to hold all the time. and 2 < e < 3. we always have −1 ≤ sin(n) ≤ 1. Thus n < n log n < n2 . Let f be a continuous function at a. Exercise. the result we proved in Exercise 2. Show that an → 0 as n → ∞. Then f (an ) → f (a) as n → ∞. n log n Solution. What do you deduce about the sequence an = exp (1/n) if you apply this result to the continuous function f (x) = ex ? 2. 2 n n+1+ n √ so we have our result since we showed in Exercise 2.22. n √ √ Note: We can now do a bit more with the n + 1 − n example. n2 Solution.5 that (1/ n) → 0 as n → ∞. Proposition. we can get such results in much more generality. Show that 1 → 1 as n → ∞. Let an = √ 1 for n ≥ 2. Show that an → 0 as n → ∞. so since n−1 → 0 as n → ∞. Let an = 1 for n ≥ 2. when n ≥ 3 and 1 1 1 < < . Note: This is another example of the “new convergent sequences from old” idea. Example.23. SEQUENCES Note: Having seen the proof. only eventually. We need the next section to prove it. we have n−1/2 → 0 as n → ∞. Exercise. 2. whatever the value of sin(n). The application is that f (x) = x1/2 is continuous everywhere on its domain which is {x ∈ R : x ≥ 0}. Note that 1 ≤ log n ≤ n if n ≥ 3. Example. Show that an → 0 as n → ∞. In fact we don’t have to use ad-hoc methods here. sin(n) . log is monotone increasing. because log(e) = 1.20.19. Note that. n2 so 1 → 0 as n log n n → ∞. Here we have used the “eventually” form of the squeezing lemma. Exercise.

Let {an } be an eventually bounded sequence.23]. This deﬁnition is correct.] 2. Then there is some N such that |an − l| < 1 whenever n ≥ N . and {bn } is a bounded sequence. Then by deﬁnition |a1 | ≤ L. our pre-declared error.30. Proposition. Exercise. Let an → 0 as n → ∞ and let {bn } be a bounded sequence. Give an example to show this. Say that {an } is a bounded sequence iﬀ there is some K such that |an | ≤ K for all n. Thus the sequence {an } is eventually bounded. Remember that if a deﬁnition contains a variable (like in the deﬁnition of convergence). We show how to use such a deﬁnition to get some results rather like those in 2. K}. Proposition. . . we show an = n2 + 5n →∞ 3n + 2 as n → ∞. then the deﬁnition is true whatever value you give to it — even if you use /2 (as we did in 2. [If an → l = 0 as n → ∞. 1 2. using the deﬁnitions. Let {an } be a convergent sequence.27.28. . 2. Exercise. to be 1.26. so in fact |an | ≤ L for all n. so there is some N . Now try: 2. the information you are given. Deﬁnition. In order to tackle simple proofs like this. |aN −1 |. For example.2. An eventually bounded sequence is bounded Proof.6 Inﬁnite Limits 2. Show that the sum of two bounded sequences is bounded. . and so is bounded by Prop 2.10. Here we have used the deﬁnition of convergence. BOUNDED SEQUENCES 19 2.10) or /K. Let an = √ for n ≥ 2. and indeed in the same way. 2. Finally see how you can get from the known information to what you need. Example. but not particularly useful at present. taking . This is just one deﬁnition. Say that an → ∞ as n → ∞ iﬀ given K.25. Show that an bn → 0 as n → ∞. And here is another result on which to practice working from the deﬁnition. Let L = max{|a1 |. then in general {an bn } is not convergent. for any constant K.29. you should start by writing down. Then write out (in symbols) what you wish to prove. and the sequence is actually bounded. [This n log n is the sequence of Exercise 2. Show that {an } is a bounded sequence. Deﬁnition. it does provide good practice in working with abstract formal deﬁnitions.24. 2. clearly you can have an → −∞ etc. |an | ≤ |an − l| + |l| ≤ 1 + |l| for all n ≥ N . with limit l say. |a2 . and a bound K such that |an | ≤ K for all n ≥ N .27. A convergent sequence is bounded Proof. |ak | ≤ L whenever k < N . there is some N such that an ≥ K whenever n ≥ N .5.5 Bounded sequences 2. But if n ≥ N then |an | ≤ K ≤ L. However. Then by the triangle inequality.

20 Pick some K.10. SEQUENCES n+5 = n. we see 3n + 2 that bn → 1/3 as n → ∞. We have an = n. we have an > n/6 > K. Hence an → ∞ as n → ∞.bn (say). .10. and so. Note: It is false to argue that an = n. CHAPTER 2. providing in addition n > 6K. bn > 1/6 (Just take = 1/6 to see this). you can’t let one part of a limit converge without letting the other part converge! And we refuse to do arithmetic with ∞ so can’t hope for a theorem directly like 2. Then for large enough n. eventually. Using results from 2.(1/3) → ∞.

Solving the quadratic gives √ 1± 5 an = . if it exists. 2 and so these are the only possible limits. or l2 − l − 1 = 0. There are better techniques than we have seen so far.1. we can eliminate the negative root. We have pn+2 pn =1+ pn+1 pn+1 and so an+1 = 1 + 1 . n n(n/e) Fibonacci Sequence: Deﬁne p1 = p2 = 1. we know that an+1 → l as n → ∞ (check this formally)! Using 2. we have an > 1 when n > 1. Thus by 2. then an → e as n → ∞. Consider the following examples: Sequence deﬁnition of e: Let an = Stirling’s Formula: Let an = √ 1+ 1 n n . and see that the limit. l ≥ 1. we see that since a1 = 1. Going back to equation 3. consider the third sequence above.1) Assume now that an → l as n → ∞. then an → 2π as n → ∞.1. then as before in 2. As another example of this. √ n! .16. dealing with more diﬃcult examples. although Stirling’s Formula takes quite a lot of calculation. √ pn+1 1+ 5 Let an = . we have l = 1 + 1/l. In fact. then an → as n → ∞.10 and equation 3.12. or when we can only ﬁnd the limit provided it exist.Chapter 3 Monotone Convergence 3. is unique. let pn+2 = pn+1 + pn .1 Three Hard Examples So far we have thought of convergence in terms of knowing the limit. all of the limits described above are within reach with some more technique. 21 . and for n ≥ 1. It is very helpful to be able to deduce convergence even when the limit is itself diﬃcult to ﬁnd.12 that knowing a sequence has a limit can help to ﬁnd the limit. pn 2 We already saw in 2. an (3.

we can distinguish between a strictly increasing sequence. a suitable value for K. while if n is odd. All we know from the deﬁnition is that it will be an upper bound. In this case.2. . We say that K is an upper bound for the sequence.1 a3 = 3. A sequence {an } is an increasing sequence if an+1 ≥ an for every n.141 a5 = 3. Note that an “arbitrary” sequence is not monotone (it will usually sometimes increase. which will take care of the three sequences described above and many more. and a (not necessarily strictly) increasing sequence. [Recall that | cos x| ≤ 1 for all x. So for any n.e. . 3. Probably the best way to show a sequence is bounded above is to write down an upper bound – i.24).14 a4 = 3. • If we need precision.5. Exercise. . Deﬁnition. Let an = 1 1 + cos . • There is a similar deﬁnition of a sequence being bounded below. the sequence a1 = 3 a2 = 3.1. Nevertheless. we claim K = 4 will do.22 CHAPTER 3. 3. Show that a sequence which is bounded above and bounded below is a bounded sequence (as deﬁned in 2. • The number K may not be the best or smallest upper bound. is how we often describe the decimal expansion of π. Deﬁnition. we have an ≤ 4 and 4 is an upper bound for the sequence {an }. 3. • There is a similar deﬁnition of a decreasing sequence. Monotone sequences are important because we can say something useful about them which is not true of more general sequences. and sometimes decrease). monotone sequences do happen in real life. For example. 3. Example. But there is a useful special case. Solution. But if n is even. MONOTONE CONVERGENCE 3. Show that {an } is bounded above and is bounded n n below. This is sometime called a nondecreasing sequence. then an = 2 + 1 = 3 ≤ 4. an = 2 − 1 = 1 ≤ 4. • What does it mean to say that a sequence is eventually increasing? • A sequence which is either always increasing or always decreasing is called a monotone sequence. Of course 3 is also an upper bound for this sequence. The sequence an = 2 + (−1)n is bounded above. To describe the result we need more deﬁnitions.2 Boundedness Again Of course not all sequences have limits.1 Monotone Convergence 3. Exercise. To check this we must show that an ≤ 4 for all n.1415 . A sequence {an } is bounded above if there is some K such that an ≤ K for all n.3.4.2.] 3. which describe extra properties of sequences.

Example. One way to check that a sequence is increasing is to show an+1 − an ≥ 0. Details will be given in third year. we have an ≥ aN ≥ K for all n ≥ N . An increasing sequence which is not bounded above tends to ∞ (see deﬁnition 2. we show eventually an > K. . Exercise.30). To prove this we need to appeal to the completeness of R. then {an } is a convergent sequence. 3. In other words. since the sequence is increasing monotonely. Let {an } be a decreasing sequence which is bounded below. then it seems as though the sequence can’t help converging — there is nowhere else for it to go. BOUNDEDNESS AGAIN 3.6. it is an interesting question whether or not it is bounded above. and we will see more ways later. then {an } is a convergent sequence Proof. Let {an } be an increasing sequence which is bounded above.3. Hence an → ∞ as n → ∞. n n If a sequence is increasing.1: A monotone (increasing) sequence which is bounded above seems to converge because it has nowhere else to go! In contrast.7. Theorem (The monotone convergence principle). Pick K. 3. Show that the sequence an = 23 n is increasing. But in that case.2. if there is no upper bound. or you can look in (Spivak 1967) for an accurate deduction from the appropriate axioms for R . Since K is not an upper bound for the sequence.2. 1 1 − 2 is decreasing when n > 1. there is some witness to this. there is some aN with aN > K. a second way is to compute an+1 /an .8. the situation is clear. Solution. Show that the sequence an = n+1 n − 2(n + 1) + 1 2n + 1 (2n2 + 3n + 1) − (2n2 + 3n) (2n + 3)(2n + 1) 1 > 0 for all n (2n + 3)(2n + 1) Upper bound n Figure 3. and assume it is not bounded above. Example. an+1 − an = = = and the sequence is increasing. Let the sequence be {an }. 3. 2n + 1 Solution. If an upper bound does exist. Here. as described in Section 1.9.

and lim + n→∞ 3 n→∞ 3 5 Solution. while if a > 1. we have l = a. Using the result that if |a| < 1. As before note that an+1 → l as n → ∞.5 a3 = 1. Given that k > 1 is a ﬁxed constant. a2 = 1.10 to the equation an+1 = a.l. Case 2 a > 1. . while if 0 < a < 1 then an+1 < an . we can change a ﬁnite number of terms in a sequence without changing the value of the limit. If a > 1 then an+1 < an . And we saw in 2. Since the sequence is monotone. MONOTONE CONVERGENCE This is a very important result.89 .12. to be just eventually bounded above? (Compare Proposition 2. We can calculate the ﬁrst few terms of the sequence as follows: a1 = 1. we must have l = 0. evaluate lim n→∞ 1− 1 k n .an . Suppose it is bounded above.11. by squeezing.2) . Example. and for n ≥ 1 deﬁne an+1 = .1.12 that just knowing a limit exists is sometimes enough to be able to ﬁnd its value. that (4/5)n → 0 as n → ∞. Example.76 a4 = 1. (3. the sequence {an } is bounded below by 0.27). Show that {an } is 7 + an convergent. it converges to a limit l satisfying l = a. and using 2. This means that the result must still be true of we only know the sequence is eventually increasing and bounded above. and ﬁnd its limit. then an → 0 as n → ∞. Evaluate lim . Note that the theorem only deduces an “eventually” property of the sequence. we deduce that (−2/3)n → 0 as n → ∞. Case 3 |a| < 1. it must tend to inﬁnity (as described in 3. Let f (x) = 6(1 + x) .an . since each outer limit → 0 by case 1. Write an = an .24 CHAPTER 3. and since a = 1.l. then since −|an | ≤ an ≤ |an |. 3. Let a be ﬁxed. an → l as n → ∞. Let a1 = 1. Exercise. so by the monotone convergence theorem. Then applying 2. Then an → 0 as n → ∞ if |a| < 1. the sequence {an } is increasing. This contradiction shows the sequence is not bounded above. we have an → 0 as n → ∞ whenever |a| < 1.13. 6(1 + an ) 3. an → ∞ as n → ∞. Solution. Solution. 7+x so f (an ) = an+1 .9). Case 1 0 < a < 1.10. and it looks as though the sequence might be increasing.10. It is the ﬁrst time we have seen a way of deducing the convergence of a sequence without ﬁrst knowing what the limit is. . in either case the sequence is monotone. What happens if we relax the requirement that the sequence is bounded above. 2 n −2 n 4 n 3. that the second limit is also 0. Example. then an+1 = a. then as in Case 1. How does your result compare with the sequence deﬁnition of e given in Sect 3. 3. and since |an | = |a|n → 0 as n → ∞.

Warning: There is a diﬀerence between showing that f is increasing. 2+ . What we know is that if f is increasing. in other words. we hope to deduce useful information about the sequence {an }. and since an ≥ 0 for all n. In the same way. (7 + x)2 ((7 + x)2 Recall from elementary calculus that if f (x) > 0. a3 > a2 . in other words. with limit l (say). then {an } is convergent.16.2 = 1+ . if it starts by decreasing. and ﬁnd it’s limit.14.15. BOUNDEDNESS AGAIN 25 By investigating f . so perhaps f (x) < 6 for all x. Show that {an } is convergent. If x is large. applying 2. and showing that the sequence is increasing. We show we have an increasing sequence which is bounded above. 3. it is precisely that f (an ) = an+1 . Thus we can only have limits 2 or −3.e. Proof. as in the above example. f (x) ≈ 6. There is of course a relationship between the function f and the sequence an . Show this for yourself. if b > a then f (b) > f (a).2 2.3. and the sequence {an } is increasing. Applying this argument inductively. then f is increasing. aN > l. we have an ≥ aN > l. 3 + ··· + n n 2! n 3! n n 1 1 ≤ 1 + 1 + + + ··· 2! 3! 1 1 1 ≤ 1+1+ + + + · · · ≤ 3. Thus l2 + l − 6 = 0 or (l + 3)(l − 2) = 0. the sequence will continue to decrease.16 to deduce that l ≥ aN > l which is a contradiction. n Solution. Since the sequence is increasing. We thus see that f is increasing. By the binomial theorem. Now apply 2. 3. Deﬁne the sequence {an } by a1 = 1.10 to the deﬁning equation gives l = . Let {an } be an increasing sequence which is convergent to l (In other words it is necessarily bounded above). We have f (x) = (7 + x). an+1 = (4an + 2)/(an + 3) for n ≥ 1. 7+l or l2 + 7l = 6 + 6l.6 − 6(1 + x) 36 = > 0.2. Example. we have f (a2 ) > f (a1 ). 3. we see that f (x) ≤ 6 whenever x ≥ 0. Then l is an upper bound for the sequence {an }. Clearly an ≥ 0 for all n. We argue by contradiction. Exercise. If l is not an upper bound for the sequence. Since a2 > a1 .2. Let an = 1+ 1+ 1 n n n n(n − 1) 1 n(n − 1)(n − 2) 1 1 + . 6(1 + l) Since also an+1 → l as n → ∞. necessarily l > 0. so f (an ) = an+1 ≤ 6 for all n. Proposition. if n ≥ N . there is some aN which witnesses this. i. and the sequence {an } is increasing and bounded above. Hence l = 2. 2 2. 1 n . then the sequence carries on going the way it starts. Hence {an } is convergent. 6 − f (x) = 6(7 + x) − 6 − 6x 36 = > 0 if 7+x 7+x x > −7 In particular. we see that an+1 > an for all n. then it continues to increase. if it starts by increasing.

and you should become familiar with both techniques. pn an (3. MONOTONE CONVERGENCE Thus {an } is bounded above. • a2n+1 is increasing and a2n is decreasing. in which we show the limit is actually e is given on tutorial sheet 3. We compute the ﬁrst few terms. 2! n 3! n n = 1+ from which it clear that an increases with n. and • an is convergent. we have an ≥ 1. let pn+2 = pn+1 + pn .6 6 8 1. and then try to prove them: • For all n. an and an ≤ 2. n pn an 1 1 1 2 1 2 3 2 1.Deﬁne p1 = p2 = 1. using our guess..67 5 5 1. The method we use could be very like that in the previous example. 3.. by deﬁnition. we make the following guesses. an+1 = pn+1 1 = =1+ . 2 − an+1 = 2 − 1 + 1 an =1− 1 ≥ 0.5 4 3 1. then an → as n → ∞.61 8 21 1.9. Recall the deﬁnition of the sequence which is the ratio of adjacent terms of the Fibonacci sequence as follows:. 2 Solution.26 CHAPTER 3. Note we are really behaving like proper mathematicians here. and for n ≥ 1. Thus we have an increasing sequence which is bounded above.17. in which case we already know it converges to .62 Formula pn+2 = pn + pn+1 an = pn+1 /pn On the basis of this evidence. Recall that. 1+ 1 n n n n(n − 1) 1 n(n − 1)(n − 2) 1 1 + . 2+ . we have 1 ≤ an ≤ 2. We show it is increasing in the same way. Note that we only have to show that this pn 2 √ 1+ 5 sequence is convergent.2 The Fibonacci Sequence 3.625 7 13 1. .3) Since pn+1 ≥ pn for all n. √ pn+1 1+ 5 Let an = . 3 + ··· + n n 2! n 3! n n 1 1 1 1 2 = 1+1+ 1− + 1− 1− + . in fact we choose to use induction more. Also. Either method can be used on either example. the aim is simply to use proof to see if earlier guesses were correct. Example. Another method. and so is convergent by the Monotone Convergence Theorem 3.2.

a2n forms a decreasing sequence. we can now use this information on each subsequence. Since a4 < a2 = 2. (We hope these subsequences are going to be better behaved than the sequence itself). bounded below by 1. we already know that the denominator is positive (and in fact is at least 4 and at most 9). Remember we are interested in the “every other one” subsequence. since a3 > a1 = 1. we have β = (1 + 5)/2. we have a5 > a3 and so on. bounded above by 2.4) We use this to compute how the diﬀerence between successive terms in the sequence behave.3 for these. Similarly. . by induction. the limit satisﬁes this equation as well. and get α = (1 + 5)/2. √ In exactly the same way. BOUNDEDNESS AGAIN 27 The next stage is to look at the “every other one” subsequences. Computing. 1 + an n an+2 = 1 + =1+ (3. α + α2 = 1 + 2α.2. 1 an+1 1 an 1 =1+ 1+a . and hence is convergent to some limit β.3. 1+α 2 √ Since all the terms are positive. First we get a relationship like equation 3. and both subsequences converge to the same limit. Thus √ α 1± 1+4 α =1+ α2 − α − 1 = 0 and α = . an+2 − an = = = an an−2 − 1 + an 1 + an−2 an + an an−2 − an−2 − an an−2 (1 + an )(1 + an−2 ) an − an−2 (1 + an )(1 + an−2 ) In the above. we can ignore the negative root.ask if you want to see the details) that the Fibonacci sequence itself converges to this limit. by induction a2n forms an increasing sequence. so as usual. It is now an easy deduction (which is left for those who are interested . Remember that adjacent terms of both of these sequences satisﬁed equation 3. This means that an+2 − an has the same sign as an − an−2 .4. and hence is convergent to some limit α. we have a6 < a4 and so on.

MONOTONE CONVERGENCE .28 CHAPTER 3.

we need to balance generality. it may be because the properties of that class are so weak that belonging may have essentially no useful consequences. However. We summarise this in the table: Strong restrictions Good behaviour Few examples Weak restrictions Bad behaviour Many examples It turns out that there are a number of “good” classes of functions which are worth studying. we shall see that sequences prove a useful tool both to investigate functions. and to give an idea of appropriate methods.1. the ﬁrst class that gives us much theory. or space. Which means the behaviour steadily improves.Chapter 4 Limits and Continuity 4. we study functions which have steadily more and more restrictions on them. then it can often be quite hard to show that a given example of such a function has the required properties. which has chaotic behaviour. we often insist that we can “get a good look” at the behaviour of the function at a given point. Conversely. and at the same time. Our main reason for being interested in studying functions is as a model of some behaviour in the real world. In selecting a suitable class of functions to study. if it is easy to show that the function belongs to a particular class. our earlier work wasn’t a diversion.1 Classes of functions We ﬁrst met a sequence as a particularly easy sort of function. In order to discuss functions sensibly. a + δ) ⊆ U . there is some open interval containing a which lies in U . Typically a function describes the behaviour of an object over time. Deﬁnition. In other words. A perfectly general function has essentially nothing useful that can be said about it. the number of examples steadily decreases. We now move to the more general study of functions. with good behaviour which occurs rarely. because there are strong restrictions on it. so we start by studying continuous functions. A subset U of R is open if given a ∈ U . In fact this is the same as saying that given a ∈ U . so typically we restrict the domain of the function to be well behaved. 4. rather than R. deﬁned on N . If a function has lots of good properties. In this chapter and the next ones. there is some δ > 0 such that (a − δ. a set is open if it contains a neighbourhood of each of its 29 .

Let f (x) = 0 if x < 0. and suppose f (a) = 2a. Note also that even a function like f (x) = 1/x is continuous. each of which is worse behaved than the previous one. as well as the “obvious” ones like polynomials and rational functions — those functions of the form p(x)/q(x). 1. We won’t spend time in this course discussing standard functions. We shall see later that this is quite a strong piece of information. the value of the function approaches 2a. Let f (x) = log(x). and so naturally has a restricted domain. where the function is not deﬁned. but rewriting the function 6. because. x In each case we are trying to study the behaviour of the function near a particular point. It thus seems very reasonable to extend the deﬁnition of f by deﬁning f (a) = 2a. 1.10 that an open interval is an open set. This deﬁnition has the eﬀect that if a function is deﬁned on an open set we can look at its behaviour near the point a of interest. Let f (x) = 5. it is continuous. In example 1. In particular. tan−1 x and sin−1 x. it is assumed that you know these are diﬀerentiable everywhere they are deﬁned. the function is well behaved everywhere. In fact. This is deﬁned for x > 0. tan x.2.5. it means they are examples of continuous functions. from both sides. x−a sin x x =1 when x = 0. It is assumed that you know about functions such as sin x. exp x.30 CHAPTER 4. Let f (x) = x2 − a2 when x = a.] 2. In example 3. log x. there are no problems. the function is still well behaved wherever it is deﬁned. x−a we see that as x approaches a.2 Limits and Continuity We discuss a number of functions. In particular. what we have observed is that x2 − a2 = lim (x + a) = 2a. but we had to restrict the domain. Our aim is to isolate an imprtant property of a function called continuity. and let f (x) = 1 for x ≥ 0. We saw in 1. In all of what follows. 4. Example. 1 when x = 0 and let f (0) = 0. 4. wherever it is deﬁned (ie on R \ {0}). In example 2. we will assume the domain of all of our functions is suitably restricted. if x = 0. Let f (x) = sin(x). 3. This is deﬁned for all x ∈ R . LIMITS AND CONTINUITY points. [Recall we use radians automatically in order to have the derivative of sin x being cos x. x→a x − a x→a lim . as promised in Sect. and so there is no need to pick out particular points for special care. where p and q are polynomials. Note also that the domain is an open set. 4. Let f (x) = sin x2 − a2 = x + a if x = a. cos x. the function is not deﬁned at a.

we note that the limit at 0 does not even exists.3. f is continuous at a iﬀ given > 0. It is closely connected with the existence of limits.4. and with very similar properties. LIMITS AND CONTINUITY We illustrate the behaviour of the function for the case when a = 2 in Fig 4.3. We sketch the function in Fig 4. which have an accurate deﬁnition. in order to deﬁne the function “properly” everywhere. Equivalently. we are forced to make the two part deﬁnition. Deﬁnition.4. that of having a deﬁnition at a point which is the “right” deﬁnition. there is some δ > 0 such that whenever |x − a| < δ. and again we see that the graph shows no evidence of a diﬃculty at x = 0 Considering example 5 shows that these limits need not always exist. there was a more natural deﬁnition of the function which gave the “right” answer. Say that f (x) tends to l as x → a iﬀ given such that whenever 0 < |x − a| < δ. then |f (x) − f (a)| < . .1: Graph of the function (x2 − 4)/(x − 2) The automatic graphing routine does not even notice the singularity at x = 2. Sketching the graph. very like the “sequence” ones. 4. near x = 0. but not at a. Deﬁnition.23.2. we can argue that the use of the (x2 − a2 )/(x − a) was perverse.1 6 5 4 3 2 1 0 -1 -2 -4 -3 -2 -1 0 1 2 3 4 31 Figure 4. The function is graphed in Fig 4. Say that f is continuous at a if limx→a f (x) = f (a). then |f (x) − l| < . So we again have to be careful near a particular point in this case. In fact this is not the worst that can happen.2. as can be seen by considering example 6. and the function has a jump discontinuity at 0. example 4. > 0. is the property of continuity. Our main interest in this deﬁnition is that we can now describe continuity accurately. In fact this is very similar to the deﬁnition we used for sequences. but diﬀer. there was no such deﬁnition. given how the function behaves near the point. 4. The crucial property we have been studying. In this example. But in the case of sin x/x. there is some δ > 0 Note that we exclude the possibility that x = a when we consider a limit. we describe this by saying that the limit from the left and from the right both exist. We prove this in more detail later in 4. we are only interested in the behaviour of f near a.

3: The function which is 0 when x < 0 and 1 when x ≥ 0. Say that f : U (open) → R is continuous if it is continuous at each point .2 -0.2: Graph of the function sin(x)/x. then f is trapped within of its value f (a). Again the automatic graphing routine does not even notice the singularity at x = 0. Deﬁnition. So you are expected to know the deﬁnition.32 1 0. y x Figure 4. Note that in the “epsilon . it has a jump discontinuity at x = 0. we shall state without proof a number of results which enable continuous functions both to be recognised and manipulated. and a few simply – δ proofs. all our work on limts and continuity of functions can be traced back to this deﬁnition. 4. we no longer need exclude the case when x = a.8 0.5.4 -8 -6 -4 -2 CHAPTER 4.2 0 -0. but you can apply (correctly . just as in our work on sequences. We shall not insist on this deﬁnition in the same way that the deﬁnition of the convergence of a sequence was emphasised.and always after checking that any needed conditions are satisﬁed) the standard results we are about to quote in order to do required manipulations. Rather than do this.delta” deﬁnition. everything could be traced back to the deﬁnition of a convergent sequence.4 0. there is some neighbourhood of a such that if we stay in that neighbourhood. Note also there is a good geometrical meaning to the deﬁnition. However. LIMITS AND CONTINUITY 0 2 4 6 8 Figure 4. Given an error .6 0.

Note: This is important. it looks as though it is going to be hard. directly from the deﬁnition. Show how to deﬁne f (4) in order to make f x−4 a continuous function at 4.4: Graph of the function sin(1/x). Here it is easy to see the problem at x = 0.6.15 0. but turns out to be quite possible. it is continuous. x3 − 8 for x = 2.05 0. and so. Sometimes. Exercise. Example. In the next example.7. Show how to deﬁne f (2) in order to make f x−2 .8.1 0. The function f (x) = 1/x is deﬁned on {x : x = 0}.5 -1 0 0. So deﬁning f (2) = 12 makes f continuous at 2.2 + 4) = 12 as x → 2. Solution. 1 x sin if x = 0. This is an example where the naive “can draw it without taking the pencil from the paper” deﬁnition of continuity is not helpful. LIMITS AND CONTINUITY 1 33 0. We cannot usefully deﬁne it on a larger domain. Let f (x) = for x = 4. 4.5 0 -0. [Can you work out why this has something to do with the derivative of f (x) = x3 at the point x = 2?] √ x−2 4. a ∈ U. Example.3 Figure 4.2. by the deﬁnition. We have x3 − 8 (x − 2)(x2 + 2x + 4) = = (x2 + 2x + 4) x−2 (x − 2) Thus f (x) → (22 + 2. 4.4. and is a continuous function. (and hence for all values of x). Let f (x) = a continuous function at 2. we can work out whether a function is continuous. x 0 if x = 0.2 0. Let f (x) = Then f is continuous at 0.25 0. the plotting routine gives up near this singularity.

We give one here and more in section 4.34 CHAPTER 4. . x/2 if x > 2.15 0.1 -0.05 -0. Pick > 0 and choose δ = [We know the answer. The graph of this function is shown in Fig 4.15 0.25 0.2 -0. for all x. The lines y = x and y = −x are also plotted. writen as limx→a+ f (x) = l 4. x sin 1 1 1 − 0 = x sin = |x|. but δ = /2.1 0. Example.3 Figure 4.4. 4.05 0. sin(1/x).5. or that f has a limit from the left iﬀ given > 0. LIMITS AND CONTINUITY Solution. it is usually helpful to have a larger range of techniques.2 0. Then if |x| < δ. Note that this is an example where the product of a continuous and a discontinuous function is continuous.25 0 0.9. or any value of δ with 0 < δ ≤ will do]. Deﬁne f (x) as follows: 3 − x if x < 2. sin ≤ |x| < δ ≤ x x x as required.1 0. We prove this directy from the deﬁnition. we have | sin(x) ≤ 1|. 0.05 0 -0. there is some δ > 0 such that whenever a − δ < x < a. then |f (x) − f (a)| < .3 One sided limits Although sometimes we get results directly.15 -0. 4. Deﬁnition.10.5: Graph of the function x. using that fact that. You can probably see how the discontinuity of sin(1/x) gets absorbed. Say that limx→a− f (x) = l. Calculate the left and right hand limits of f (x) at 2. There is a similar deﬁnition of “limit from the right”. f (x) = 2 if x = 2.

f g and (often) f /g are also functions. For example. and second. then k. in 4. As x → 2−. Let f (x) = |x|. Thus the left and right hand limits agree (and disagree with f (2).13. Also.15. . This splits the problem of ﬁnding whether a limit exists into two parts. 4. example 5. 4. both 1-sided limits exist. just as we did for sequences. Let f (x) = is continuous everywhere]. Proposition. so the right hand limit is 1. then f /g is continuous at a. so f is not continuous at 2). Exercise. Proposition. Note our convention: if f (x) → 1 and always f (x) ≥ 1 as x → 2−. Note that if x < 0 then |x| = −x and so is continuous. [In fact x cos x for x ≥ 0. 4. Instead we give a result that enables us to build new continuous functions from old ones. that we get the same answer. x→1+ so f is continuous at 1.14. lim f (x) = lim x2 = 1 while x→1− x→1+ lim f (x) = lim x = 1 = f (1). by the 4. so the left hand limit is 1. Note that f (1) = 1. Then f is continuous in R . we say that f (x) tends to 1 from above. sin x for x < 0. and let k be a constant. Solution. Also x→1− x2 for x ≤ 1.f . and check ﬁrst that we have a limit. From these identiﬁcations.4 Results giving Coninuity Just as for sequences. It remains to examine the function at 0. Solution. Let f (x) = Show that f is continuous at 0. As x → 2+. while if x > 0. 4. example 4] 4.f . If limx→a f (x) exists. Since 0+ = 0− = 0 = |0|. if g(a) = 0. then limx→a f (x) exists.12. [Recall the result of 4. we see that limx→0− |x| = 0+.2. then both one sided limts exist and are equal. while limx→0+ |x| = 0+. f (x) = 3 − x → 1+. RESULTS GIVING CONINUITY 35 Solution. f is continuous everywhere]. Example. f + g. then |x| = x and so also is continuous. 4. Note that if f and g are functions and k is a constant. Let f and g be continuous at a. building continuity directly by calculating limits soon becomes hard work. Conversely. Example.11. |x| is continuous at 0 4. [In fact f x for x ≥ 1. and write f (x) → 1+ etc.16. We use the above criterion. but are not equal. Show that f is continuous at 1.2. Proposition. if both one sided limits exits and are equal. we can look on either side. f + g and f g are continuous at f . (Continuity Test) The function f is continuous at a iﬀ both one sided limits exits and are equal to f (a).4.4. There is now an obvious way of checking continuity. f (x) = x/2 → 1+. Then k.12.

Pick > 0. Proposition. it is enough to show that x→a lim (f (x) + g(x)) = f (a) + g(a). see (Spivak 1967) for proofs. we need to know that f /g is deﬁned in some neighbourhood of a. then |g(y) − g(f (a))| < .21. 4. Since −1 < g(x) < 1. and that f = g ◦ h. then |f (x)−f (a)| < /2.19. This gives the result.20. The function f : x −→ sin3 x is continuous. and show that f 1 − x2 is continuous at every point of its domain. Then |f (x) + g(x) − (f (a) + g(a))| ≤ |f (x) − f (a)| + |g(x) − g(a)| < /2 + /2 = . When we ﬁrst studied it. and let g be continuous at f (a). then g(f (x)) − g(f (a))| < .20 x2 − a2 . Combining these results gives the required inequality. Thus f is continuous. Since. Let f (x) = tan domain. The other results are similar. Show that f is continuous at every point of its x2 + a2 .36 CHAPTER 4. we can give a proof of Proposition 2. then |f (x) − f (a)| < δ1 .17. δ2 ). Solution. The function in Example 4. 4. Let δ = min(δ1 . Write g(x) = sin(x) and h(x) = x3 . Let g(x) = 1 + x2 . since each term is. x2 − a2 . Example. but rather harder. We ﬁnd δ using the given properties of f and g. Note: Just as when dealing with sequences. We show that f + g is continuous at a.8 is continuous everywhere. 4. we showed it was continuous at the “diﬃcult” point x = 0. then |g(x) − g(a)| < /2. LIMITS AND CONTINUITY Proof. Now we can deduce that it is continuous everywhere else. Solution. Then g ◦ f is continuous at a Proof. by deﬁnition. Example. Similarly there is some δ2 such that if |x − a| < δ2 . we have (f + g)(a) = f (a) + g(a). Now use the fact that f is continuous at a. Example. the function is properly deﬁned x2 + a2 for all values of x (whilst tan x is undeﬁned when x = (2k + 1)π/2 ). then there is some δ1 such that if |x−a| < δ1 . Exercise. Since g is continuous at f (a). Thus f is continuous. since f = tan ◦g. Let f (x) = exp As another example of the use of the deﬁnitions. This can be shown using a very similar proof to the corresponding result for sequences. and since x2 + a2 = 0 for any x. We must ﬁnd δ > 0 such that if |x − a| < δ. Note that each of g and h are continuous.18. Pick > 0. there is some δ1 > 0 such that if |y − f (a)| < δ1 . Write down the domain of f . and pick x with |x − a| < δ. 4. 4. and the quotient is continuous. so there is some δ > 0 such that if |x − a| < δ. Let f be continuous at a.

8. Deﬁnition.4.23. just by looking at a single sequence. Example. Proof. we see that the limit (if it exists) can only be l. We can consider the example f (x) = sin(1/x) with this tool. to get a contradiction. Thus x2 + 3 1 1 + 3/x2 → = 2 + 2x + 1 2 3x 3 + 2/x + 1/x 3 as x → ∞.25. and we must also have l = 1. The proof is a little harder than the one we have just given.8 that if an → 0 as n → ∞. if f is well behaved on each sequence which converges to a. and is left until next year. . sin(1/xn ) = sin((4n+1)π/2) = 1. consider the sequence xn = 2/(4n+1)π. We use the fact that 1/x → 0 as x → ∞. in other words.22 together with Example 4. as required. The idea here should be quite familiar from our sequence work. Thus. In this case. 4. INFINITE LIMITS 37 4. assume. then in fact f is continuous at a. Now since f is continuous at a. that the limit exists. 1 an →0 as n → ∞. then |f (x) − l| < . sin(1/xn ) = sin(nπ) = 0 → l as n → ∞. and so by assumption. Suppose that limx→0 sin(1/x) = l. 4. Pick > 0 we must ﬁnd N such that |an − f (a)| < whenever n ≥ N . then xn → 0 as n → ∞. We know from Prop 4.26. x→∞ 3x2 + 2x + 1 Solution. Example. there is some K such that x2 + 3 . sin ≤ |an | → 0 as an an n → ∞. and let an → a as n → ∞.5 Inﬁnite limits There are many more deﬁnitions and results about limits. 4. But instead. Say that limx→∞ f (x) = l iﬀ given whenever x > K. Combining these we see that if n ≥ N then |an − f (a)| < . then |f (x) − f (a)| < . First one that is close to the sequence deﬁnition: 4. Let f be a continuous function at a. then an sin Prove this directly using squeezing. We have 0 ≤ an sin 1 1 = |an |. Note: Sequences often provide a quick way of demonstrating that a function is not continuous. Example. like δ = /2 etc would work) such that |an − a| < δ whenever n ≥ N . Proposition. Thus l does not exist.22. Then f (an ) → f (a) as n → ∞. we can ﬁnd δ such that if |x − a| < δ. Solution. since an → a as n → ∞ there is some N (taking δ for our epsilon — but anything smaller. Also. while. The proof is essentially the same as the proof of Example 4. Let xn = 1/(πn). Evaluate lim > 0.5.24. so again xn → 0 as n → ∞. 4.

6 Continuity on a Closed Interval So far our results have followed because f is continuous at a particular point. 4. Indeed the results in this section are precisely why we are interested in discussing continuity in the ﬁrst place. Although some of the results are “obvious”. Corollary. The converse is equally trivial. Deﬁnition. Then there is a point c with a < c < b such that f (c) = h. there is some K such that whenever x > K. 4. limx→a+ f (x) = f (a). Let f be continuous on the compact interval [a. they only follow from the continuity property. Show there is at least one root of the equation x − cos x = 0 in the interval [0. It is also noted that each hypothesis is necessary. Proof. 4. 1]. indeed. and limx→b− f (x) = f (b). 1]. 4.29. y → ∞ as x → 0+ Proof. Then there is some point c with a < c < b such hat f (c) = 0. Let f be continuous on the compact interval [a. The function is continuous on that interval. simply take K = 1/ . Then y → 0+ as x → ∞. The reason for working on proofs from the deﬁnition is to be able to check what results of this type are trivially true without having to ﬁnd it in a book. we must ﬁrst be able to check our functions satisfy the hypothesis of continuity. 4. 1) such that f (c) = 0 as required. conversely.28. Show there is at least one root of the equation x − e−x = 0 in the interval (0. and assume there is some constant h such that f (a) < h and f (b) > h. So in order to be able to use the following helpful results. and assume that f (a) < 0 and f (b) > 0. but sketch a proof with a pair of sequences and a repeated bisection argument. Apply the Intermediate Value Theorem to f on the closed interval [0. and indeed we present counterexamples whenever that fails. Proof. For example 4. Theorem (Intermediate Value Theorem). Proposition. Exercise. Let y = 1/x. .30. while f (1) = 1 − cos(1) > 0. Say that limx→∞ f (x) = ∞ iﬀ given L > 0. That task is of course what we have just been concentrating on. b). 4.27. Then g(x) → 0+ as x → ∞ iﬀ f (x) → ∞ as x → ∞. We sometimes refer to f as being continuous on a compact interval. and f (0) = −1. b] iﬀ f is continuous on (a. then f (x) > L. Pick > 0. Let g(x) = 1/f (x).31.38 CHAPTER 4.32. Example. and if. then 0 < y < . LIMITS AND CONTINUITY 4. b]. in addition. We make no attempt to prove this formally. Thus there is some point c ∈ (0. 1). We show there is some K such that if x > K. Such an f has some very nice properties. Deﬁnition. In fact we get the best results from assuming rather more.33. Say that f is continuous on [a. b].

4. Note also that the hypotheses are all needed. this can give an inaccurate idea. Note that by considering −f + h we can cope with the case when f (a) > h and f (b) < h. 1]. b]. there is nothing to prove. This contradicts the fact that M was a least upper bound for f . and that these global extrema are actually attained. (Boundedness) Let f be continuous on the compact interval [a. Note also that the hypotheses are all needed. This uses the completeness of R . and again no proof is oﬀered. As we have seen with y = 1/x. Note: This theorem is the reason why continuity is often loosely described as a function you can draw without taking your pen from the paper. Theorem. This is clearly continuous. (Boundedness) Let f be continuous on the compact interval [a. This again uses the completeness of R . otherwise g(x) is deﬁned everywhere. Consider the function g(x) = (M − f (x))−1 . and is continuous. Then there are points x0 and x1 such that f (x0 ) < f (x) < f (x1 ) for all x ∈ [a. If there is some point at which f (y) = M . In other words. and follows in part from the previous result. Theorem. Proof. b].4. we are guaranteed that the graph of f is bounded both above and below. and hence bounded. b]. we are guaranteed that the graph of f is bounded both above and below. . If M is the least upper bound given by theorem 4.35.34. Apply the Intermediate Value Theorem to f − h on the closed interval [0. 4. In other words. b]. Proof. Then there are constants M and m such that m < f (x) < M for all x ∈ [a. CONTINUITY ON A CLOSED INTERVAL 39 Proof.6. it is in fact more akin to connectedness.34.

40 CHAPTER 4. LIMITS AND CONTINUITY .

the diﬀerence is that the graph of a diﬀerentiable function may not have any sharp corners in it.Chapter 5 Diﬀerentiability 5. The next most particular class of function we study after the class of continuous functions is the class of diﬀerentiable functions. Note that the Newton quotient is not deﬁned when x = a. if it exists.2. and prove that this is a smaller class: that every diﬀerentiable function is continuous. but that there are continuous functions that are not diﬀerentiable. and let f : U → R. so we move fairly quickly over the basics. We say that f is diﬀerentiable at a ∈ U iﬀ x→a lim f (x) − f (a) x−a or equivalently.1 Deﬁnition and Basic Properties In This section we continue our study of classes of functions which are suitably restricted. The limit. by introducing l’Hˆpital’s rule. We say that f is diﬀerentiable in U if and only if it is diﬀerentiable at each a in U . But the Newton quotient. Deﬁnition. our aim is to show that there are many attractive properties which hold for diﬀerentiable functions that don’t hold in general for continuous functions. Show. 41 . Although we don’t prove this. We take the view that much of this material has already been discussed last year. as calculating the value of the derivative. Example. Note also the emphasis on the existence of the limit.6. Compare this result with 4. Let U be an open subset of R . show how to get “new functions from old” in what by now is a fairly routine way. One we discuss in some detail is the ease with which certain limits can be evaluated. Let f (x) = x3 . can be extended to be a continuous function at a by deﬁning its value at that point to be f (a). 5. nor need it be for the deﬁnition to make sense. directly from the deﬁnition. is written as f (a). if it exists. We discuss the deﬁnition. As with continuous functions. Again we are passing from the general to the particular. Informally. Diﬀerentiation is as much about showing the existence of the derivative. 5.1. the corresponding results are o false for continuous functions. h→0 lim f (a + h) − f (a) h exists. that f (a) = 3a2 .

2. that f (a) = 1/2 a when a = 0.7. Proposition. DIFFERENTIABILITY This is just another way of asking about particular limits.4. cos and tan. and let k be a constant. and the quotient is diﬀerentiable. and exp and log. together with the known derivatives of elementary functions such as sin. we need results showing how to diﬀerentiate the usual class of functions.2x) . Thus we assume the rules for diﬀerentiation of sums products and compositions of functions. Let f and g be diﬀerentiable at a. . This is the same example we considered in 4. since each term is. Exercise. 5. than simply to be continuous. Show. . and calculate the derivative at each such point. Let f (x) = x. Then f is continuous at a. x + a2 Then g is properly deﬁned for all values of x. it is impractical to use this deﬁnition every time to compute derivatives. Let f be diﬀerentiable at a. x2 + a2 (x2 + a2 )2 x2 − a2 4a2 x . Exercise. Let f (x) = exp The ﬁrst point in our study of diﬀerentiable functions is that it is more restrictive for a function to be diﬀerentiable. their reciprocals sec. we have:f (x) = sec2 = x2 − a2 (x2 + a2 ). Finally to actually calculate the derivative. Show that f is diﬀerentiable at every point of its 1 − x2 domain. There we showed the domain x2 − a2 was the whole of R . then f /g is diﬀerentiable at a. In addition. and calculate the derivative at each such point.3. compute CHAPTER 5. f + g and f g are diﬀerentiable at f . Let f be diﬀerentiable at a.5. x→a x − a x→a x→a x−a lim √ √ 5. and we assume these are known from last year. What function do you have to consider in the particular case when a = 4? Just as with continuity.42 Solution. cosec and cot. Solution. like 4. Then k. sec2 2 + a2 )2 (x x2 + a2 1 + x2 . Example. Then g ◦ f is diﬀerentiable at a.20. Proposition. Let f (x) = tan x2 − a2 for a = 0.6. Show that f is diﬀerentiable at every x2 + a2 point of its domain.2x − ((x2 − a2 ). if g(a) = 0. we must x3 − a3 (x − a)(x2 + xa + a2 ) = lim = lim (x2 + 2xa + a2 ) = 3a2 . and we are assuming known that the elementary functions like tan are diﬀerentiable. 5. directly from the deﬁnition. Thus f is diﬀerentiable using the chain rule since f = tan ◦g. 5. Let g(x) = 2 . and let g be diﬀerentiable at f (a). and that the function was continuous everywhere. and since x2 + a2 = 0 for any x since a = 0. Also.f . 5. the usual rules for calculating these derivatives apply.

we must prove that limx→a f (x) = f (a). unless we get an indeterminate form 0/0 or ∞/∞ for the formal quotient. 5. Note ﬁrst that we cannot get the result trivially from 4. recall that |x| = x if x ≥ 0. .15 that |x| is continuous. since since g(a) = 0 and so we get the indeterminate form 0/0. Let f and g be functions such that o f (a) = g(a) = 0 while f (a) and g (a) both exits and g (a) = 0. we have for x = a. If f (a) and g (a) exist. x→0 x→0 x 1 lim as x → a. since f and g x→a g(x) must be continuous at a by Proposition 5.2 Simple Limits Our calculus of diﬀerentiable functions can be used to compute limits which otherwise prove troublesome. but are unequal.8. f (x) − f (a) = Hence f is continuous at a. x→0+ x→0+ x − 0 x−0 lim while f (x) − f (0) −x − 0 = lim = −1. Since f (a) = g(a) = 0. Show that lim = 2.16. However. 3x − sin x 5. We compute the Newton quotient directly. but not diﬀerentiable at 0. we have f (x) f (x) − f (a) f (x) − f (a) x−a f (a) = = → g(x) g(x) − g(a) x−a g(x) − g(a) g (a) where the last limit exists. x→0 x Solution. Thus f (x) − f (0) x−0 = lim = 1. Example.0 x−a as x → a.16. so the limit of the Newton quotient does not exist. 5. Then f (x) f (a) = . Solution.2. 5.10. although we need o to develop stronger forms. since g (a) = 0. We already know from Example 4. Example. x→0− x→0− x − 0 x−0 lim f (x) − f (a) . while |x| = −x if x < 0.5. Thus both of the one-sided limits exist. provided x = a. since x = 1 = 0.(x − a) → f (a). then f is continuous everywhere. Remark. Proposition (l’Hˆpital’s rule: simple form).7. Applying the rule gives o 5. Let f (x) = |x|. f (x) is easy by 4. To establish continuity. x→a g(x) g (a) lim Proof.11. Since the Newton quotient is known to converge. computing lim 3x − sin x 3 − cos x = lim = 2. SIMPLE LIMITS 43 Proof. In fact l’Hˆpitals rule helps in both cases.9. we are in a position to apply the simple form of l’Hˆpital.

15. Example. This is spurious because we need the limit to calculate the derivative in the ﬁrst place.13. Evaluate lim x→0 sin x = 1. Here is the diagram to make the point geometrically: f’(c) = 0 x a c b Figure 5.12. Note: The theorem guarantees that the point c exists somewhere. we are in a position to apply the simple form of l’Hˆpital. x Solution. and suppose that f (a) = f (b). b]. However. b). x→0 x Solution. but applying l’Hˆpital certainly gives the result. the function is ﬂat. b). somewhere between the two crossings. Exercise. Theorem (Rolle’s Theorem). x→0 x→0 x 1 log(1 + x) . Since f is continuous on the compact interval [a. It gives no indication of how to ﬁnd c. Then . it has both a global maximum and a global minimum. Let f be continuous on [a. 5.14. (Spurious. b]. Assume ﬁrst that the global maximum occurs at an interior point c ∈ (a. we pick h small enough so that c + h always lies in (a. Note again that we cannot get the result trivially from 4. DIFFERENTIABILITY 1+x−1 = 1/2.16. Example.1: If f crosses the axis twice. but helps to remember!) Show that lim 5.3 Rolle and the Mean Value Theorem We can combine our deﬁnition of derivative with the Intermediate Value Theorem to give a useful result which is in fact the basis of most elementary applications of the diﬀerential calculus. it is a global result. sin x x→0 5.44 √ 5. since this gives the indeterminate 0/0 form. Applying the rule gives o √ 1+x−1 2−1 (1 + x)−1/2 lim = lim = 1/2. and diﬀerentiable on (a. since x = 1 = 0. Then there is some c with a < c < b such that f (c) = 0. The accurate statement of this “obvious” observation is Rolle’s Theorem. b). Show that lim CHAPTER 5. because g(a) = 0. Proof. o 5. Like the results on continuous functions. In what follows. and so needs continuity and diﬀerentiability on a whole interval.

it follows that f is constant. and we have the result in this case. b) will do. Show that the equation x − e−x = 0 has exactly one root in the interval (0. b). Solution. it is diﬀerentiable on the open interval (a. and in the same way. since f (a) = f (b). b). b]. Looking as x → ∞ and as x → −∞ shows here are three roots of q. A similar argument applies if. let h(x) = f (b) − f (x) − f (b) − f (a) (b − x). Exercise. the global minimum occurs at the interior point c. we see that p has at most one root. b−a . does have a more general restatement. since f is.17. b−a Then h is continuous on the interval [a. Since p(0) = 1. ROLLE AND THE MEAN VALUE THEOREM If h > 0. Proof. 5. since we know the limit h→0+ h h f (c + h) − f (c) f (c + h) − f (c) ≥ 0. Hence p has exactly one root. there is a root of q between −1 and 1 by the Intermediate Value Theorem. Our version of Rolle’s theorem is valuable as far as it goes. while p(−1) = −3. Let f be continuous on [a. exists.18.3. and so lim ≥ 0. but the requirement that f (a) = f (b) is suﬃciently strong that it can be quite hard to apply sometimes. Since p (x) = 3(x2 + 1) > 0 for all x ∈ R . Example. and any c ∈ (a. b]. 5. 1). since we h→0+ h h know the limit exists. instead. We apply Rolle to a suitable function. Investigate the number of roots of each of the polynomials p(x) = x3 + 3x + 1 and q(x) = x3 − 3x + 1. f (b) − f (a) . We have q (x) = 3(x2 − 1) = 0 when x = ±1. Combining these. Similarly. Since q(−1) = 3 and q(1) = −1. Fortunately the geometrical description of the result — that somewhere the tangent is parallel to the axis. and so lim ≤ 0. 5. Then there is some c with a < c < b such that f (b) − f (a) = f (c) b−a or equivalently f (b) = f (a) + (b − a)f (c). if h < 0. The remaining situation occurs if both the global maximum and global minimum occur at end points. We can thus apply Rolle’s theorem to h to deduce there is some point c with a < c < b such that h (c) = 0.5. Thus we have 0 = h (c) = −f (c) + which is the required result. for if it had two (or more) roots there would be a root of p (x) = 0 between them by Rolle. f (b) = 0 and f (a) = 0. we see that f (c) = 0. 45 f (c + h) − f (c) f (c + h) − f (c) ≤ 0. and differentiable on (a. Theorem (The Mean Value Theorem).16. there is at least one root by the Intermediate Value Theorem. Also.

20. 1] to obtain some c with 0 < c < 1 such that f (1) − f (0) = f (c). Then there is some point c with a < c < b such that g (c) f (b) − f (a) = f (c) g(b) − g(a) . 5. (The Cauchy Mean Value Theorem) Let f and g be both continuous on [a. 1 and f (0) = 2. Theorem. and so 4 < 5 − x2 < 5.2 < f (1) < 2. DIFFERENTIABILITY B A a c b Figure 5. Exercise. Since for such x. From the given value of f (0).46 CHAPTER 5. We ﬁrst estimate the given derivative for values of x satisfying 0 < x < 1. Inverting we see that 1 1 < f (x) < 5 4 when 0 < x < 1.2: Somewhere inside a chord. b). Use the Mean 5 + sin x . The function f satisﬁes f (x) = theorem to estimate f (1). We now see that the accurate version of this replaces f (a) by f (c) for some c between a and a + h. b] and diﬀerentiable on (a. Note also that this gives meaning to the approximation f (a + h) ≈ f (a) + hf (a). Now apply the Mean Value theorem to f on the interval [0. Example. 5. The function f satisﬁes f (x) = Value theorem to estimate f (π/2). Use the Mean Value 5 − x2 Solution. we have 0 < x2 < 1. the function will behave linearly. The accurate statement of this common-sense observation is the Mean Value Theorem. Note the “common sense” description of what we have done. 1 and f (0) = 0.19. we see that 2. the tangent to f will be parallel to the chord.25 5. If the derivative doesn’t change much.21.

4 l’Hˆpital revisited o We can get a much more useful form of l’Hˆpital’s rule using the Cauchy Mean Value o Theorem. b]. we have f (b) − f (a) f (c) = . 5. Let h(x) = f (x) g(b) − g(a) − g(x) f (b) − f (a) . Note: Taking g(x) = x recovers the Mean Value Theorem. g(x) g(x) − g(a) g (c) f (x) f (c) f (x) = lim = lim . and diﬀerentiable on (a. and suppose that f and g are diﬀerentiable on an open interval I containing a. to ﬁnd c with a < c < x such that f (x) f (x) − f (a) f (c) = = . . Also h(a) = f (a)g(b) − g(a)f (b) = h(b). But h (c) = f (c) g(b) − g(a) − g (c) f (b) − f (a) Thus f (c) g(b) − g(a) = g (c) f (b) − f (a) This is one form of the Cauchy Mean Value Theorem for f and g.9. Pick x > a and apply the Cauchy Mean Value Theorem to the interval [a. If g (c) = 0 for any possible c. g(b) − g(a) g (c) giving a second form of the result. Then f (x) f (x) = lim . Proof. directly from the deﬁnition of the derivative. Thus we may apply Rolle to h. and apply Rolle’s theorem exactly as we did for the Mean Value Theorem.4. as we did in 5. g(b) − g(a) g (c) 47 Proof.ˆ 5. except perhaps at a. x→a g(x) x→a g (x) lim provided the second limit exists. Now repeat with x < a to get the result. 5. Let f and g be functions such o that f (a) = g(a) = 0. Then lim . L’HOPITAL REVISITED In particular. and that g (x) = 0. and so we can divide the above result to get f (b) − f (a) f (c) = . since we know the actual limit (not just the x→a+ g(x) c→a+ g (c) x→a+ g (x) one sided limit) exists. to deduce there is some point c with a < c < b such that h (c) = 0. Proposition (l’Hˆpital’s rule: general form). then the Mean Value theorem shows that g(b) − g(a) = 0. it follows that h has these properties. b). whenever the quotients make sense. Note ﬁrst that since both f and g are continuous on [a. rather than working.22. x].

the use of l’Hˆpital is justiﬁed the ﬁrst time.26. Use l’Hˆpital to establish the following: o √ 1 + x − 1 − x/2 1 lim =− . 2 x→0 x 2 CHAPTER 5. Evaluate lim Solution. one of the hypotheses of l’Hˆpital is that the ﬁrst quotient is of the 0/0 form. Evaluate lim x sin x→∞ 1 x . Proposition (l’Hˆpital’s rule: inﬁnite limits). g(x) x→∞ g (x) . note also that this gives o another way of ﬁnding an = (1 + 1/n)n . Exercise. Example. 2 x→0 x→0 2x x 2 lim where the use of l’Hˆpital is justiﬁed since the second limit exists. and still expect to get the correct answer. since we now o know the second limit exists. o 5.28. x→∞ y log(1 + y) = lim y→0+ y log(1 + y) = lim = 1. Note that you can’t o diﬀerentiate top and bottom again. o 5. and suppose that lim exists.27.25. We have √ 1 + x − 1 − x/2 (1/2)(1 + x)−1/2 − 1/2 (1/2)(−1/2)(1 + x)−3/2 −1 lim = lim = lim = .48 5. 5.24. 5. 1 5.28 to get results about inﬁnite limits. since the third limit exists. Exercise. DIFFERENTIABILITY 1 − cos x sin x 1 = lim = .5 Inﬁnite limits We can use Proposition 4. Evaluate lim x log 1 + . Let f and g be functions such o f (x) that limx→∞ f (x) = limx→∞ g(x) = ∞.23. since the ﬁnal limit exists by l’Hˆpital. 2 x→0 x 8 Solution. We have 1 − cos x 1 = . Example. Example. x→∞ x Solution. Evaluate lim x→0 1 1 − sin x x = lim x − sin x . 2 x→0 x→0 x→0 x 2x 2 8 The use of l’Hˆpital is justiﬁed the second time. x→0 x sin x 5. Then x→∞ g (x) x→∞ lim f (x) f (x) = lim . x→∞ lim x log 1 + 1 x = log(1 + y) writing y = 1/x. y→0 y lim The last step is valid.

= 5. f (c) f (x) − f (a) = g (c) g(x) − g(a) for all x > K. + (n − 1)(n − 2)an−1 xn−3 + n(n − 1)an xn−2 . By Cauchy. Again. . . Then pick K such that if x > K.1 as x → ∞. This is explored further in the exercises. + (n − 1)an−1 xn−2 + nan xn−1 . • xα increases faster than any power of log x if α > 0.1 (Rates of growth) One interest in these results is to see how fast functions grow as x → ∞. . then g(x) − g(a) = 0. and have already observed that the Mean Value Theorem provides a more accurate version of this.Raphson method for solving an equation. Note that although c depends on x.] We have already met the approximation f (a + h) ∼ f (a) + hf (a) when we studied the Newton .1. and so f (0) = a1 . we have a statement that works whether h > 0 or h < 0. (Sketch for interest — not part of the course). Pick lim f (x) −l < g (x) 49 > 0 and choose a such that x→∞ for all x > a. 5. Note that f (0) = a0 . . Diﬀerentiating gives f (x) = a1 + 2a2 x + 3a3 x2 . g(x) − g(a) f (x) − f (a) g(x) → l. we always have c > a. TAYLOR’S THEOREM Proof. we have f (x) = 2a2 + 3. But important results are: • The function ex increases faster than any power of x.6. Now consider what happens when f is a polynomial of degree n f (x) = a0 + a1 x + a2 x2 + . + an−1 xn−1 + an xn . Then f (x) g(x) f (x) − f (a) f (x) g(x) − g(a) .6 Taylor’s Theorem We have so far explored the Mean Value theorem. . .5. [By writing the deﬁnition of c in this way. .5.2a3 x . . . which can be rewritten as f (a + h) = f (a) + hf (c) where c is some point between a and a + h.

+ x x . we get. and so getting a generalisation of the Mean Value Theorem. f (n+1) is deﬁned on [a. Then we can write f (x) = Pn (x) + Rn (x). this is a restatement of the Mean Value Theorem. this often called the Second Mean Value Theorem. • just as with Rolle. a] etc. • in general we can restate Taylor’s Theorem as f (x) = f (a) + (x − a)f (x) + . . 5. 2! n! where c is some point between a and x. f (n) (a) f (n+1) (c) (x − a)n + (x − a)n+1 . while after k diﬀerentiations. using its value. it is called Maclaurin’s Theorem. Theorem (Taylors Theorem . After the next diﬀerentiation. • the special case in which a = 0 has a special name. or the Mean Value Theorem. and assume that each of f . as f (x) = f (0) + f (0)x + f (0) 2 f (0) 3 f (n−1) (0) n−1 f (n) (0) n + x + x + . although the proof can be done with the tools we have at our disposal. We make no attempt to prove this. Some quick comments: • the theorem is also true for x < a. just restate it for the interval [x. Let f be continuous on [a. 2! 3! (n − 1)! n! This opens up the possibility of representing more general functions than polynomials in this way. are given by Pn (x) = f (a) + f (a)(x − a) + Rn (x) = f (n+1) (c) (x − a)n+1 .Lagrange form of Remainder). Thus we can rewrite the polynomial. f . where Pn (x). . .29. we have f (x) = f (a) + (x − a)f (x) + f (c) (x − a)2 2! for some c between a and x. f (k) (0) = k!ak .. we get f (0) = 3!a3 . and the value of its derivatives at 0. • if n = 1. the Taylor polynomial of degree n about a. DIFFERENTIABILITY and f (0) = 2a2 . .50 CHAPTER 5. there is no useful information about the point c. and Rn (x). the corresponding remainder.. we have f (x) = f (a) + (x − a)f (c) for some c between a and x. provided k ≤ n. 2! (n + 1)! . . + for some c between a and x. x]. • if n = 0. x]. . (n + 1)! f (a) f (n) (a) (x − a)2 + (x − a)n .

Find the Taylor polynomial of order 1 about a for f (x) = ex . thus |Rn (x)| = |x|n+1 xn+1 c e ≤ max(ex . and write down the corresponding remainder term. Solution.. TAYLOR’S THEOREM We now explore the meaning and content of the theorem with a number of examples. and so f (k) (0) = 1. 5. Solution. (n + 1)! We can actually say a little more about this example if we recall that x is ﬁxed. If x < 0. Example. + 2! 2! n! (n + 1)! for some point c between 0 and x. Solution. ec < 1. In particular. Find the Taylor polynomial of order n about 0 for f (x) = ex . Thus by Taylor’s theorem. P1 (x) = ea +(x − a) ea and R1 (x) = (x − a)2 c e . 1) → 0 as n → ∞.. then since c < x. the sequence Pn (x) → ex as n → ∞. 2! 2! n! and Rn (x) = xn+1 c e . ex = 1 + x + x 2 x2 xn xn+1 c e + + . 51 5. We have f (x) = (1 + x)3 f (x) = 3(1 + x)2 f (x) = 6(1 + x) f (x) = 6 f (n) (x) = 0 if n > 3. and write down the corresponding remainder term. Pn (x) = 1 + x + x 2 x2 xn + + . Find the Maclaurin polynomial of order n > 3 about 0 for f (x) = (1 + x)3 . and write down the corresponding remainder term.30.6.. We have ex = Pn (x) + Rn (x) = Pn (x) + xn+1 c e (n + 1)! We show that Rn (x) → 0 as n → ∞. (n + 1)! (n + 1)! We think of the limit of the polynomial as forming a series. In particular. 2! (x − a)2 c e 2! 5. There is no diﬃculty here in calculating derivatives — clearly f (k) (x) = ex for all k. so that (again for ﬁxed x). Example.5. by Taylor’s theorem.. Example. We study series (and then Taylor series) in Section 7. the Taylor series for ex . ex = ea +(x − a) ea + for some point c between a and x. while if x ≥ 1. we have ec < ex . Using the derivatives computed above.31.32. f (0) = 1 f (x) = 3 f (x) = 6 f (x) = 6 .

together with the corresponding remainder.33. DIFFERENTIABILITY 6 2 6 x + x3 . Find the Taylor polynomial of order n about 0 for f (x) = sin x.. . .34. but which is at least reassuring. .. 3! 5! (2n + 1)! x2 x4 x2n cosh x = 1 + + + .52 and so.. ex + e−x ex − e−x 5. 5.. by Taylor’s theorem (1 + x)3 = 1 + 3x + CHAPTER 5. Exercise. 2! 4! 2n! cos x = 1 − 5. + + . but the important point is that |R2n+1 (x)| ≤ x2n+3 →0 (2n + 3)! as n → ∞.. Now check the 2 2 shape of the following Taylor polynomials: x2 x4 x2n + + . Thus sin x = x − x 3 x5 + cos c 3! 5! x3 3! .. 2! 4! 2n! x3 x5 x2n+1 sinh x = x + + + ... + + .35. 2! 3! a result we could have got directly. There is no diﬃculty here in calculating derivatives — we have f (x) = sin x f (x) = cos x f (x) = − sin x f (x) = − cos x f (4) (x) = sin x and so on. sin x = x − x3 x5 x2n+1 + + . . Writing down the remainder term isn’t particularly useful. + (−1)n + . We use the Taylor polynomial for sin x of order 4 about 0. Find the maximum error in the approximation sin(x) ∼ x − given that |x| < 1/2.. Solution. + (−1)n+1 + .. Thus by Taylor’s theorem. Example. Solution. Example.. Recall that cosh x = . and that sinh x = . 3! 5! (2n + 1)! f (0) = 0 f (x) = 1 f (x) = 0 f (x) = −1.. and write down the corresponding remainder term.

= f (n) (0) = . This is another way to get the Binomial theorem described in Section 1. but we omit the proof. providing f is suitably diﬀerentiable. . TAYLOR’S THEOREM for some c with 0 < c < 1/2 or −1/2 < c < 0.. . x5 x5 1 1 cos c ≤ ≤ 5 ≤ . = 0. Pn tells us nothing useful about the function. n! f (0) = 1 f (x) = α f (x) = α(α − 1) f (x) = α(α − 1)(α − 2) f (n) = α(α − 1) ..36.5! 120. .xn = 0. .32 53 Warning: The Taylor polynomial always exists. . . . . + 0. In any case. . and so for every n.. . say α = n then |Rn (x)| = 0 and f (x) = Pn (x). provided |x| < 1. But it need not be useful.8. (1 + x)α = 1 + αx − α(α − 1) 2 α(α − 1)(α − 2) 3 x + x + 2! 3! α(α − 1) . (α − n + 1). it is well behaved everywhere else. . . The interest in f is at 0. since |x| < 1/2. Example.5. In fact.. Consider the example f (x) = exp(−1/x2 ) if x > 0. in fact |Rn (x)| → 0 when n → ∞. Clearly in this case.6. 5. 5! 5! 2 . + x + . (α − n + 1)(1 + x)α−n f Thus by Taylor’s theorem. Note also that if α > 0 is an integer. . It turns out that f (0) = f (0) = f (0) = . (α − n + 1) n . Solution. So the Taylor polynomial of degree n for f about 0 is Pn (x) = 0 + 0x + 0x2 + . Rn (x) = f (x). Find the Taylor polynomial of order n about 0 for f (x) = (1+x)α . . There is again no diﬃculty here in calculating derivatives — we have f (x) = (1 + x)α f (x) = α(1 + x)α−1 f (x) = α(α − 1)(1 + x)α−2 f (x) = α(α − 1)(α − 2)(1 + x)α−3 and in general (n) (x) = α(α − 1) . The remainder is not hard to deal with. and note that this gives a derivation of the binomial theorem. 0 if x ≤ 0. the remainder |Rn (x)| → 0 as n → ∞.

54 CHAPTER 5. DIFFERENTIABILITY .

6. It is more natural to think of adding additional terms to the polynomial. so 2Sn = (2a + nr) + (2a + nr) + · · · + (2a + nr).6. Sn = (a + nr) + (a + (n − 1)r) + · · · + (a + r) + a. and (a + (a + nr)) Sn = (n + 1). One motivation comes from section 5. in which we obtained a sequence of approximating polynomials {Pn } to a given function. Let Sn = a + (a + r) + (a + 2r) + · · · + (a + nr). so (1 − r)Sn = a(1 − r n+1 ). the sequence of partial sums. 2 Note that if r > 0 then Sn → ∞ as n → ∞.1 Arithmetic and Geometric Series Consider the sum a + (a + r) + (a + 2r) + · · · + (a + nr) We are trying to add up the terms in an arithmetic progression. those built by adding up more and more from a given collection of terms. while if r < 0 then Sn → −∞ as n → ∞. A small amount of notation makes the addition easy. However. then Sn → a as n → ∞.Chapter 6 Inﬁnite Series In this section we return to study a particular kind of sequence. rSn = ar + ar 2 + ar 3 + · · · + ar n+1 . there is a closely related sequence. We next consider a geometric progression (or series): Let Sn = a + ar + ar 2 + · · · + ar n . 1−r Note that if |r| < 1. and as such we are studying series. . and a(1 − r n+1 ) Sn = if r = 1. 1−r 55 .

5. note particular cases.4. . Find n=1 7 1 1 −4 n . say r = 1 + k. The series r n is convergent with sum 1/(1 − r).3. We can compute the partial sums explicitly: 6. the nth partial sum. + 2n 3n 6. n(n + 1) Solution. Sn → a/(1 − r) as n → ∞. We noted above that when |r| < 1. ∞ n=1 1 =1 2n or equivalently. we say that n→∞ an is a convergent series. n→∞ 6. provided that |r| < 1. 1 1 + 2n 3n = 1 1 1 1 1 1 1 + = + + + ··· + + + ··· 2n 3n 2 4 8 3 9 1 1 3 =1+ = . Example. Find n=1 Solution. is a divergent series. the series Solution. INFINITE SERIES (1 + k)n+1 − 1 k . and let If lim Sn exists.2. the series is divergent. Example. Exercise. 1 1 1 + + + · · · = 1. n 3 2 6. and write lim Sn = an . 6. Deﬁnition.56 If r > 1. (−1)n is divergent. 3 1 − 1/3 2 ∞ 6. a > 0. a 1 + (n + 1)k − 1 > a(n + 1) → ∞ if k 1 1 . 2 4 8 1 is convergent with sum 1. Example. Sn = a > ∞ CHAPTER 6. The limit of the sequence of partial sums is the sum of the series. Let {an } be a sequence.2 Convergent Series Sn = a1 + a2 + · · · + an . in particular. Thus a series is convergent if and only if it’s sequence of partial sums is convergent. The sum n Sn = k=1 1 = k(k + 1) n k=1 1 1 − k k+1 =1− 1 → 1 as n+1 n → ∞. For other values of r.1. A series which is not convergent.

Proposition.6. an ) − (a1 + a2 + . . and recall from our work on limits of sequences that Sn−1 → l as n → ∞.2. We show it is bounded above. but this is much n2 1/4 1/9 1/16 y = 1/x^2 1/n2 n-1 n 1 2 3 4 Figure 6. Since 1/n2 > 0. The sum Sn 1 1 1 1 1 1 + + + + + ··· + + 1 2 3 4 5 8 1 2 4 1 > 1+ + + >2 if n ≥ 15 2 4 8 2 1 2 4 8 > 1+ + + + > 3 if n ≥ 31 2 4 8 16 → ∞ as n → ∞.] 1 is convergent. x 1 n < n Thus Sn < 2 for all n. We estimate the partial sums: 6. and the series is convergent. an−1 ) = Sn − Sn−1 → l − l as n → ∞. Example. Then an → 0 as n → ∞. = 1 1 + ··· + 9 15 + ··· + 1 n 57 6. Write l = limn→∞ Sn . Proof. . 6. [Actually the sum is π 2 /6.8.1: Comparing the area under the curve y = 1/x2 with the area of the rectangles below the curve Solution. Then an = (a1 + a2 + . . n Solution. 1 1 1 + 2 + ··· + 2 2 2 3 n Sn dx . . whence by the Monotone Convergence Theorem (3. Let an be convergent. Example. . The sum harder. We estimate the partial sums.6.7. it is convergent. and so 2 1 x 1 n 1 < 1+ − ≤2− . clearly {Sn } is an increasing sequence. the sequence of partial sums is bounded above.9). CONVERGENT SERIES 1 is divergent. From the diagram.

Since 1/n > 0.] 6.12. we can deduce many more results by comparison as follows. Proposition. n Solution. we have Sn > log n > Sn − 1. This gives a necessarycondition for the convergence of a series. x 2 n 1 1 + · · · + .2: Comparing the area under the curve y = 1/x with the area of the rectangles above the curve From the diagram. Thus Sn → ∞ and the series is divergent. the diﬀerence Sn − log n → γ as n → ∞. or equivalently 1 + log n > 2 n Sn > log n for all n. Example. We estimate the partial sums. where γ is Euler’s constant. 6. an and bn be convergent. 1 6. Then an is convergent. The sum is divergent (Graphical method).11.3 The Comparison Test We have already used the Monotone Convergence Theorem in studying simple series. . This can be checked easily directly from the deﬁnition. 6. Deﬁne Sn = a1 + a2 + · · · + an . 1+ Writing Sn = 1 + 1 1 + ··· + > 2 n n 1 dx 1 1 > + ··· + . (The Comparison Test) Assume that 0 ≤ an ≤ bn for all n.10. We show it is not bounded above. and suppose that bn is convergent. it is in eﬀect the same proof that the sum of two convergent sequences is convergent etc.58 CHAPTER 6.9. it is not suﬃcient. even though 1/n → 0 as n → ∞. Remark. Theorem. When we know the behaviour of some simple series. whence by the note after 3. the sequence of partial sums → ∞ as n → ∞. Then (an + bn ) and c.an are Proof. y y = 1/x x 1 2 3 4 n-1 n Figure 6. Proof. INFINITE SERIES 6. For example we have seen that 1/n is divergent. Tn = b1 + b2 + · · · bn . clearly {Sn } is an increasing sequence.9. In fact it is a lot more useful. Let convergent. [There is a much better estimate.

Assume ﬁrst that bn is convergent. for if which it isn’t! log n 1 > = bn . THE COMPARISON TEST 59 Then by hypothesis. The conclusions of Theorem 6. we have an ≤ 2kbn . Hence an is convergent by 6.28.12. = 2 →1 bn n +1 1 n +1 as n → ∞. Suppose that an > 0 and an bn > 0. Remark. To get the converse. and compare with bn = . Then an is n→∞ bn convergent iﬀ bn is convergent. Solution. In particular.6. we replace the inequality by Sn ≤ Tn + a1 + a2 + · · · + aN and this then holds for all n. so is 6. which is convergent by 6. Let an = n2 n 1 and let bn = . For n ≥ 1. Note that if n ≥ 3. Thus Sn ≤ K for all n. Proof. . Then +1 n an is divergent by the limiting form of the comparison test. 6.13. Example. then log n > 1. log n 1 6. it follows that that 6. 3n3 − 1 2n an . for if it holds for n ≥ N . Sn+1 = Sn + an ≥ Sn and {Sn } is an increasing sequence. it is bounded above. 0 ≤ Sn ≤ Tn . We can thus use the “eventually” form of the comparison test. so 3n3 − 1 ≥ 2n3 . Since an /bn → k as n → ∞. and that there is some constant k such that lim = k > 0.14.15. Thus by the Monotone Convergence Theorem. Corollary (Limiting form of the Comparison Test). so {Sn } is a sequence that is bounded above. so there is some K such that Tn ≤ K for all n. Thus 6. so we can use the same argument with an and bn interchanged. an n n n2 = 2 . so we try to apply the limiting form of the comparison test directly. we have an = We deduce divergence. Example. n3 ≥ 1. an were convergent. it is a bounded sequence by Prop 2. Let an = an = Since we know that 2n 2n ≤ 3 = bn . n n bn was convergent. an is convergent bn is convergent. n n Solution.3. Then 3n3 − 1 n Solution.17. Note that the terms are all positive. Since {Tn } is a convergent sequence. eventually (take = k > 0). Example.11.12 remain true even if we only have an ≤ bn eventually. note that bn /an → 1/k as n → ∞. it is a convergent sequence . since an ≥ 0.16. Let an = . 2n 1 and let bn = 2 .

. 6. 6. and we see that an is divergent.20. Assume α > 1. This is all very well. From a diagram which is essentially the same as that of Fig 6. 2 3 n . if r > 1. Then (2n)! an is convergent. Let an = ison test to show that 2n + 7 and let bn = 3n − 1 an is convergent. 6. |an+1 | → |an | r as n → ∞. we need a few reference sequences before we can get further. and consider the graph of y = 1/xα . Let Sn = 1 + α + · · · + α . Example. 6. shaded area dx α 1 x 1 1 1 1 n + ··· + α < 2α n 1 − α xα−1 1 1 1 Sn − 1 < 1 − α−1 α−1 n 1 Sn < +1 α−1 < n Thus the sequence of partial sums is bounded above. 6. A proof follows by comparing with the corresponding geometric series with ratio r. 2 n noting that y is a decreasing function of x (which is where we use that fact that α > 1).22. the series is convergent. we estimate the partial sums.19. which we have already met. Proposition. The sum 1 is convergent when α > 1. the use of the limiting form of the comparison test is valid.1. Since 1/nα > 0. Details will be given in full in the third year course. by comparison with with 1/n 1/n. One set is the geometric series.21. nα Solution. by comparison 2 . Let an = 2n (n!)2 . Theorem (The Ratio Test). Let an = √ n 1 and let bn = 3/2 . while if r = 1. the test gives no information.18. Use the limiting form of the compar- We also know about 1/nα . Use the limiting form of the n +n+1 comparison test to show that an is convergent. Note also we need our work on sequences in Section 2 to evaluate the required limit. Let an be a series. Then if r < 1. and the series converges. INFINITE SERIES Since the limit is non-zero.60 CHAPTER 6. and assume that lim Proof. n5 We can consider the method of comparing with integrals as an “integral test” for the convergence of a series. note the method we have used. rather than state it formally. the series is divergent. at least when α ≥ 2 when it converges. and when α ≤ 1 when it diverges. clearly {Sn } is 1 1 an increasing sequence. but as with the “new sequences from old” programme. Exercise. Exercise.

Example.4. n2 ∞ Solution.25. We look the ratio of adjacent terms in the series (of positive terms). We have n=1 (−1)n+1 n2 = 1 (−1)n+1 and this is convergent by 6. So n2 n2 n=1 n=1 ∞ is absolutely convergent. 6. The series ∞ an is conditionally convergent if and only if the n=1 series ∞ an is convergent but not absolutely convergent. 6. and use the comparison test. it is usual to start a series with a positive term. which is the topic of this section. 6. An absolutely convergent series is convergent. Show that the series n=2 ∞ (−1)n is absolutely convergent.19. Indeed this is the easiest way to show a series is convergent if the terms are not all positive. The series is convergent.27. the series converges by the ratio test. Show that the series n=1 ∞ (−1)n+1 is absolutely convergent. Note: We choose to work with the sign (−1)n+1 rather than (−1)n simply for tidiness. [Hint: note that n2 n=1 | sin n| ≤ 1. Exercise. n2 log n 6.26.] Our interest in absolutely convergent series starts by observing that they are in fact all convergent. The easiest case occurs when the series really can be thought of as a series of positive terms.6. There is a good reason for this. ∞ n=1 an is absolutely convergent iﬀ the series ∞ n=1 |an | 6. so the coeﬃcient of a1 is chosen to be +.24. Deﬁnition. Exercise. ABSOLUTE AND CONDITIONAL CONVERGENCE Solution. Show that the series sin n is absolutely convergent.23. Proposition. n=1 ∞ 6.28. (n + 1)st term 2n+1 (n + 1)!(n + 1)! (2n)! an+1 = = an (2n + 2)! 2n n!n! nth term 2 2(n + 1) n+1 1 = = → (2n + 2)(2n + 1) 2n + 1 2 61 as n → ∞. Deﬁnition. Thus that of a2 must be − etc if the series is to alternate. there is very little we can say about series with mixed signs.4 Absolute and Conditional Convergence So far most of our work has been with series all of whose terms are positive. ∞ 6. Since the ratio of adjacent terms in the series tends to a limit which is < 1. . Indeed there is just one useful result at this level.

or the alternating series test. Try not to be confused by this popular but inaccurate usage. This gives one way of proving that a series is convergent even if the terms are not all positive. Write Sn for the nth partial sum of the series (−1)n+1 an . ∞ Proof. Warning: Note how we usually talk about the “Fibonacci series”. while |an | = a+ + a− n n and an = a+ − a− . Leibniz Theorem Let {an } be a decreasing sequence of positive terms such that an → 0 as n → ∞.1 to compare with the convergent series ∞ n=1 we see that each of a+ n ∞ and n=1 a− n is a convergent series of positive terms.62 Proof. .1. Using equation 6. and deﬁne if an > 0 if an ≤ 0 and a− = n |an | if an < 0 0 if an ≥ 0 an 0 The point of this deﬁnition is that 0 ≤ a+ ≤ |an | and n 0 ≤ a− ≤ |an | for all n.1) so we have two new series of positive terms.29. We give the proof because the argument is so like the proof of the convergence of the ratio of adjacent terms in the Fibonacci series 3.11. . Then the series ∞ (−1)n+1 an n=1 is convergent. we have s2n+1 = a1 − a2 + a3 − . n n ∞ n=1 |an |. Assume that a+ = n ∞ n=1 an CHAPTER 6. Theorem. which is a very special. By deﬁnition. . We show this sequence n=1 has the same type of oscillating behaviour that the corresponding sequence of partial sums in the Fibonacci example. . Thus ∞ ∞ an = n=1 n=1 a+ − a− = n n ∞ n=1 a+ + n ∞ n=1 a− n is also convergent using 6. or the theorem on alternating signs. even though it is a sequence rather than a series. There is essentially only one other way. + a2n−1 − a2n + a2n+1 s2n−1 = a1 − a2 + a3 − . + a2n−1 . but useful case known as Leibniz theorem. and so we can’t use the comparison test directly. n (6. INFINITE SERIES is absolutely convergent. 6.

We have = and this is divergent by 6. Proposition (Re-arranging an Absolutely convergent Series).4. . . n n=1 6. Similarly s2 . < s2n−2 < s2n < s2n+1 < s2n−1 < . so an > 0. is a decreasing sequence which is bounded below (by s2 ). Thus we have a decreasing sequence s1 > s3 > s5 > . > s2n−1 > s2n+1 > . . Also s2n+1 = s2n + a2n+1 > s2n Thus s2 < s4 < s6 . s5 . . a2n > a2n+1 and so s2n+1 < s2n−1 . .30. Example. . . and ∞ ∞ bn = n=1 n=1 an . an+1 < an and an → 0 as n → ∞.19. thus the series is n n n=1 n=1 not absolutely convergent. subtracting. . . . Then ∞ n=1 bn is convergent. n n=1 ∞ ∞ (−1)n+1 1 Solution. . and so is conditionally convergent. Let ∞ an n=1 be an absolutely convergent series and suppose that {bn } is a re-arrangement of {an }.29 that this series is still convergent. s6 . is an increasing sequence which is bounded above (by s1 ). Write an = 1/n. 63 Since {an } is a decreasing sequence. . 6.6. . Show that the series ∞ (−1)n+1 is conditionally convergent. .9 is convergent to β (say) also s2n+1 − s2n = a2n+1 and so letting n → ∞ α−β =0 So α = β. . we have s2n+1 = s2n−1 − a2n + a2n+1 . . and all the partial sums are tending to α. s5 < s3 < s1 and the sequence s1 .31. We show using 6. ABSOLUTE AND CONDITIONAL CONVERGENCE and so. . . and so by 3.9 is convergent to α (say). Similarly s2n > s2n−2 and we have an increasing sequence s2 < s4 < s6 . . Thus all the conditions ∞ (−1)n+1 of Leibniz’s theorem are satisﬁed. s4 . . so the series converges. < s2n−2 < s2n < . and so the series is convergent. s3 . . and so by 3.

the point here is that we need absolute convergence before series behave in a reasonable way.64 CHAPTER 6. or (Spivak 1967). to . n2 n=1 Again we are going to use geometrical methods for this.1). See next year. Our geometric statement follows from the diagram. There is a result which is an extreme form of this: Pick x ∈ R. and let ∞ an be a conditionally convergent series.5 An Estimation Problem This section shows how we can use a lot of the earlier ideas to produce what is often wanted in practice — results which are an approximation. which is less than the area of the rectangles which contain the curve. and is the assertion that the area of the rectangles below the curve is less than the area under the curve. then there is n=1 a re-arrangement {bn } of {an } such that ∞ bn = x! n=1 In other words. we can re-arrange to get any answer we want! 6. y = 1/x2 N N+1 N+2 M-1 M Figure 6. Find how accurate the approximation obtained by just using the ﬁrst ten terms ∞ 1 is. INFINITE SERIES Proof. together with an indication of how good the approximation is. Warning: It is not useful to re-arrange conditionally convergent series (remember the rearrangement I did in section 1.3: An upper and lower approximation to the area under the curve Writing this out geometrically gives the statement: M n=N +1 1 ≤ n2 M N dx ≤ x2 M −1 n=N 1 n2 We can evaluate the middle integral: M N dx 1 = − 2 x x M = N 1 1 − . N M .

which also holds.5. . N +1 N 1 ≤ S − SN . so SM → S.6. and 1/M → 0. We have S − SN ≤ 1 ≤ S − SN −1 N Replacing N by N + 1. we have S − SN +1 ≤ 1 1 ≤ S − SN ≤ ≤ S − SN −1 . AN ESTIMATION PROBLEM For convenience. as 1 1 ≤ S − SN ≤ . N +1 In particular. we deﬁne S= 1 n2 n=1 ∞ 65 and SN = 1 . 10 11 110 Our conclusion is that although S10 is not a very good approximation. namely S − SN +1 ≤ and combining these two. N +1 N To make the point that this is a useful statement. Then 1 1 ≤ S − S10 ≤ 11 10 or 0 ≤ S − S10 + 1 11 ≤ 1 1 1 − = . gives another inequality. we now specialise to the case when N = 10. let M → ∞. n2 n=1 N We can now express our inequality in these terms: SM − SN ≤ 1 1 − ≤ SM −1 − SN −1 N M Next. we can describe the error well enough to get a much better approximation. we have both upper and lower bounds for S − SN .

66 CHAPTER 6. INFINITE SERIES .

We have more eﬀective ways of showing that such a series converges — we can use that ratio test.. 5! (2n + 1)! x4 x2n + + .. + (−1)n + . Deﬁnition.36. 2! 2! n! x2n+1 x5 + + . which we can use to deﬁne functions. for some functions.. + . and x is a real variable.6. . We now recognise this as showing that certain series converge. not that it converges to the function used to generate it in the ﬁrst place. A real power series is a series of the form real numbers. . We shall see in this section that this class of functions are really just “grown up polynomials”.. the remainder Rn (x) → 0 as n → ∞ for each ﬁxed x. We even saw in 5. we met the idea of writing f (x) = Pn (x) + Rn (x). To summarise the results we had in Section 5.. + + . The corresponding functions are the best behaved of all the classes of functions we meet in this course.. they are real power series. where the an are . to express a function in terms of its Taylor polynomial..Chapter 7 Power Series 7. Proposition. 7. + + .. . . But note that such a test will only show that a series converges.. and that almost any manipulation valid for polynomials remains valid for this larger class of function.2.. together with a remainder. 5! (2n + 1)! x2n x4 + + . 4! 2n! x5 x2n+1 + + ...6. 7. We saw an example of such a problem in the Warning before Example 5.1.1 Power Series and the Radius of Convergence In Section 5.. indeed are as well behaved as could possibly be expected. 4! 2n! These are all examples of the subject of this section.30 that. + (−1)n+1 + . 67 an xn .. The following series converge for all values of x to the functions shown: ex = 1 + x + x3 3! x2 cos x = 1 − 2! x3 sinh x = x + 3! x2 cosh x = 1 + 2! sin x = x − x 2 x2 xn + + ...

7. 7. and then we have a genuine circle of convergence. so we look at the ratio of the moduli. It turns out that a power series is usually best investigated using the ratio test.4. one for each diﬀerent value of x. Find the radius of convergence of the series . By allowing R = 0 and R = ∞.68 CHAPTER 7. Deﬁnition.3. The number R described above is called the radius of convergence of the power series. Theorem 6. Theorem (Radius of Convergence). we are guaranteed that when |x| > R. Our hope is that there is some coherence. And the behaviour of power series is in fact very coherent. No statement is made in the third case about what happens when x = R.21. or an xn converges absolutely for all x. Suppose one of the following happens: • • an xn converges only when x = 0. It is characterised by the fact that the series converges (absolutely) inside this interval and diverges outside the interval. Example.5. and guaranteed divergence whenever |z| > R. the interval of convergence.2 2 = = . Example. POWER SERIES We are thus dealing with a whole collection of series. ∞ 7. with convergence for all (complex) z with |z| < R. Only on the circle of convergence is there ambiguity. nxn 7. it can’t even converge “accidentally” for a few x s with |x| > R. where the power series is guaranteed to converge. the series diverges. • Note the power of the result. or an xn is a power series. and diverges for all x with |x| > R. 2n+1 Solution. The geometric series n=0 xn is another example of a power series we have already met. • The word “radius” is used.6. n. that the behaviour of series for diﬀerent values of x are related in some sensible way. |(n + 1)st term| |nth term| (n + 1)|x|n+1 2n+1 2n+2 n|x|n (n + 1)|x| |x| → as n → ∞. We saw this series is convergent for all x with |x| < 1. Then • there is some number R > 0 such that an xn converges absolutely for all x with |x| < R. We sometimes call the interval (−R. This regularity of behaviour makes it easy to investigate the radius of convergence of a power series using the ratio test. Recall that the ratio test only applies to series of positive terms. because in fact the same result is true for complex series. Thus every power series has a radius of convergence. we can consider every power series to have a radius of convergence. R).

and diverges when x = 2. We now deﬁne a function f on this open interval I as follows: ∞ f (x) = n=1 an xn for x ∈ I. REPRESENTING FUNCTIONS BY POWER SERIES 69 Thus the given series diverges if |x| > 2 and converges absolutely (and so of course converges) if |x| < 2. (−1)n n!xn . .2 Representing Functions by Power Series Once we know that a power series has a radius of convergence. These results are all easy to check by direct substitution. Example. Consider the following power series.2. nn Solution. Again we look at the ratio of the moduli of adjacent terms. R). Hence it has radius of convergence 2.8. Exercise. we can use it to deﬁne a function. (n + 1)n+1 n!|x|n (n + 1)(n+1) n nn n |x| = |x| (n + 1)n n+1 |x| |x| → as n → ∞.|x|. while the third converges both when x = 2 and when x = −2. ∞ 7.29. This one is a little more subtle than it looks.7. Find the radius of convergence of the series |(n + 1)st term| |nth term| nn (n + 1)!|x|n+1 nn = (n + 1). 7.7. Hence it has radius of convergence e. on the circle of convergence. Find the radius of convergence of the series n=0 xn n2 + 1 We noted that the theorem gives no information about what happens when x = R. n2 2n The ﬁrst is divergent when x = 2 and when x = −2. 7. although we have met the limit before.1 to note that 1+ 1 n n → e as n → ∞.e. all of which have radius of convergence 2. ∞ n=1 xn 2n ∞ n=1 xn n2n ∞ n=1 xn . There is a good reason for this — it is quite hard to predict what happens. i. Suppose the power series an xn has radius of convergence R > 0. and let I = (−R. Thus the given series diverges if |x| > e and converges absolutely (and so of course converges) if |x| < e. and using Theorem 6. the second converges when x = −2. e n+1 n n = = = Here we have of course used the result about e given in Section 3.

. and show that even when we add an inﬁnite number of errors. . then this can be done for all x. and so we get the series ∞ log(1 + x) = n=0 (−1)n xn+1 n+1 valid for |x| < 1. 1−x we replace x by −x to get 1 = 1 − x + x2 − x3 + . Quite a lot harder than it looks. Proof. 7. . If R = ∞. for |x| < 1. Let f and I be deﬁned as 7. Corollary. provided we interpret the open interval I as R . . . Let an xn be a power series with radius of convergence R > 0... + (−1)n + . (7. Let I be the open interval (−R. in (Spivak 1967).g. 1+x Integrating both sides then gives log(1 + x) = K + x − x2 x3 xn+1 + + . Starting with the geometric series 1 = 1 + x + x 2 + x3 + . Then nan xn−1 has radius of convergence R. .9.9 to G to see that G (x) = f (x). which is the required result. Apply 7. 7. + (−1)n xn + . and ∞ f (x) = n=1 nan xn−1 for x ∈ I.by term everywhere inside the circle of convergence. Proof. (7.1.70 CHAPTER 7. We summarise this result by saying that we can diﬀerentiate a power series term . they don’t add up to too much. . and then use the Mean Value Theorem to estimate diﬀerences. . Theorem. . R). POWER SERIES It turns out that this is the last. we need to be able to re-arrange power series.1) where K is a constant of integration. 7. Then f has an indeﬁnite integral deﬁned on I. + x n + . It can be found e. and best behaved of the classes of functions we study in this course.3 Other Power Series We now derive some further power series to add to the collection described in 7. given by ∞ G(x) = n=0 an xn+1 (n + 1) for x ∈ I. f is diﬀerentiable on I.2) . In fact all of what we say below remains true when R = ∞.10.9. for |x| < 1. 2 3 n+1 for |x| < 1. . and deﬁne f (x) = an xn for x ∈ I. Evaluating both sides when x = 0 shows that K = 0.

1. [You may assume. and ∞ f (x) = n=0 an xn for |x| < R. We return to equation 7. so the given series is the Taylor (or Maclaurin) series for f n! . + (−1)n + . for some constant of integration K. Again putting x = 0 shows that K = 0. and thus f (x) = log(1 + x) for |x| < 1. without proof.4 n(n − 1) By diﬀerentiation or otherwise.12. Then an+1 xn+1 n(n − 1) = → |x| an xn n(n + 1) as n → ∞.3. . Thus the new series has radius of convergence 1. we have arctan(x) = K + x − x3 x5 x2n+1 + + .. Let deﬁne an xn be a power series.2 4. Example. Denote its sum by f (x). 1 + x2 Again integrating both sides. . .by .3) 7. + (−1)n x2n + . Hence f (x) = log(1 + x) dx = K + x log(1 + x) − x+1−1 dx 1+x (7. Proposition. it is permissible to diﬀerentiate term . we have still found the Taylor coeﬃcients. Find the radius of convergence R of the power series x2 x3 x4 x5 (−1)n xn − + − ··· + + ··· 2 3. 3 5 2n + 1 (−1)n x2n+1 2n + 1 n=0 ∞ for |x| < 1. Inside the circle of convergence. OTHER POWER SERIES 71 Note: It is easy to get this result directly from the Taylor Series. We have now been able to derive a power series representation for a function without working directly from the Taylor series. and so arctan(x) = valid for |x| < 1. and replace x by −x2 to get 1 = 1 − x2 + x4 − x6 + . where K is a constant of integration. .3 5. (7. The next one is not quite so easy. and doing the diﬀerentiations — which can often prove very awkward. 7. Putting x = 0 shows that K = 0 and so f (x) = (1 + x) log(1 + x) − x. ﬁnd the sum of the series for |x| < R. Then an = f (n) (0) .4) (7. Apply the ratio test to the given power series.7. that log(1 + x) dx = K + x log(1 + x) − x + log(1 + x). since they have the same power series. .5) = K + x log(1 + x) − x + log(1 + x).11.] Solution.term. Nevertheless. .. deﬁned for |x| < 1. for |x| < 1. with radius of convergence R > 0.

about 3. 1−x 2 Writing this in terms of x gives 1 1 =− 1−x 2 1+ x − 3 (x − 3)2 (x − 3)n + . + y n /2n + . We can diﬀerentiate n times by 7. 1 . Also. but totally explicable when it is realised that the two points i and −i are stopping the expansion from being valid in a larger circle.13. and we can manipulate it by manipulating.14. . Clearly f (n) (0) = 0 unless n is a multiple of 3. + xn + . More investigation (quite a lot more . in the obvious way. We now read oﬀ the various derivatives.. It should be no surprise that this is the Taylor series for the same function about the point 3. e.. . . And it is in fact not an accident that the radius of convergence of the new series is 2.4 Power Series or Function We have now seen that when a power series is used to deﬁne a function. Let f (x) = 1 = 1 + x3 + x6 + x9 + . . the corresponding power series. 7. . . 1−x To expand about a diﬀerent point. We use the binomial theorem to get a power series expansion about 0. And that is why it is of interest to sometimes consider power series as complex power series. We can already do this about 0 by the Binomial Theorem. . Example.12. Calculate f (n) (0). giving the uniqueness result. 1 7. + 2 4 2n for |y/2| < 1. calculating exactly as in the start of Section 5. . . 7. 1 − x3 Solution. Then 1 1 1 1 1 = = =− . This seems implausible viewed with real spectacles. A function has one deﬁnition which works everywhere it makes sense (at least for simple functions). + x3n + .mainly for complex functions) shows the radius of convergence is always that of the largest circle that can be ﬁtted into the domain of deﬁnition of the function. 1−x Solution. . whereas the power series corresponding to a function depends also on the point about which the expansion is happening. . . for |x| < 1. we have 1 1 = − 1 + y/2 + y 2 /4 + y 3 /9 + . However there are snags.6. we have: 1 = 1 + x + x2 + x3 + . .. An example will probably make this clearer than further discussion. then that function is very well behaved. Give the power series expansions for the function f (x) = . for |x − 3| < 2. 1 − x3 valid for |x| < 1. . write y = x − 3. 1−x 1 − (y + 3) −2 − y 2 1 − y/2 and again using the Binomial Theorem on the last representation.9 and we still get a series with the same radius of convergence. The power series expansion for (1 + x2 )−1 has radius of convergence 1. while f (3k) (0) = (3k)! by 7.g. we see that the derivatives satisfy f (n) (0) = n!an .72 CHAPTER 7.. Example. POWER SERIES Proof. + + .

Thus y/β = log x.5. 7. but is here to show how we can get more interesting results from power series.1 The function ex grows faster than any power of x Speciﬁcally. 2 It turns out that this integral cannot be evaluated using the usual tricks — substitution. giving the result claimed.10 can help.5. Put y = β log x. xN −α x ≥ N! N! if x > 1. since N is ﬁxed. In particular. we claim that for any β > 0. 7. so y βy −1 eY → ∞.e β when y = β log x. We are interested in what happens when x → ∞. . Thus we have xβ = ey . n! n=0 ∞ Given α. limx→∞ x−β log x = 0. We have ex = xn n! n=0 ∞ and so x −α x e = xn . N! 7. we claim that for any α ∈ R. or equivalently. x = ey/b . so we will restrict to the situation when x > 0. and so x−β log x = y −y . we can always ﬁnd some N such that N − α ≥ 1. integration by parts etc. in so doing one is often led to consider an integral of the form α I= 0 e−x dx. limx→∞ x−α ex = ∞. e−y → 0 as y → ∞. perhaps for computing risks and premiums on life insurance.5. APPLICATIONS* 73 7. so y → ∞. noting this is possible since x > 0.3 The probability integral α −x2 e 0 dx The normal distribution is a very common model throughout the whole of science for the situation when things occur “at random”.7.5 Applications* This section will not be formally examined.2 The function log x grows more slowly than any power of x Speciﬁcally. But from our previous result.6. which is the same as saying that . x−α ex > In particular.6) Since β > 0. Next note that provided x > 0. So in particular. each term in the series for x−α ex is positive. this is the required β result by equation 7. x → ∞ as x → ∞. and hence the sum is greater than any given term.5. as y → ∞. (7. But a power series representation and Corollary 7. probability theory attempts to predict what will happen “on average”. as x → ∞.

to get our contradiction... . to give α 0 e− x2 dx = (−1)n α2n+1 . by Corollary 7.74 Thus ∞ CHAPTER 7.10. b . so it can be written in the form e = a/b. a + 1 (a + 1)(a + 2) (a + 1)(a + 2)(a + 3) ∞ n=0 ∞ (−1)n . that e is rational.. as was done to √ show that 2 was irrational. to get e−x dx = K + 2 (−1)n x2n+1 . b (−1)n = e−1 = a n! n=0 and so. The left hand side is an integer. (2n + 1)n! n=0 ∞ Performing a deﬁnite integral removes the constant of integration. where a and b are integers. 1! 2! 3! a! a! a! (−1)a a! = a! − + − + . even thought we can’t “do” the integral. so we have a practical method of evaluating the integral. .. n! . From the exponential series. So assume. together with a proof by contradiction.4 The number e is irrational We use a power series argument. and converge quite quickly. POWER SERIES e = e −x2 n=0 ∞ x xn n! n (−x 2 )n ∞ = (−1) n=0 n! = (−1)n n=0 x2n n! and we can integrate this term-by term. + 1! 2! 3! a! 1 1 1 + (−1)a+1 − + − . (2n + 1)n! n=0 ∞ The partial sums of the power series on the right can be computed. as is each of the terms in the sum a! − a! a! a! (−1)a a! + + .a! .a! = e−1 = a Thus.5.. multiplying through by a!. we have b(a − 1)! = a! − a! a! a! + − + . Note that this means e−1 = b/a. 7.. + . 1! 2! 3! a! .

so (−1)a+1 1 1 1 − + − .29. The 1 1 ﬁrst partial sum is while the second is (check this). which decreases to 0.. and so by 6. this contradiction demonstrates our initial a+1 a+2 assertion. So there must be an integer a+1 a+2 1 1 between and .5.. . But this is an alternating sequence of positive terms. is convergent to a sum which lies between the ﬁrst two partial sums.7. Since there is not. a + 1 (a + 1)(a + 2) (a + 1)(a + 2)(a + 3) 75 is an integer. APPLICATIONS* it follows that the rest of that equation is an integer.

76 CHAPTER 7. POWER SERIES .

enclosed by a rectangle with sides x and 10 − x. or the maximum possible area of the rectangle. t). which is 77 . V = πr 2 h. T ). • why it is interesting. We shall mainly be concerned with diﬀerentiation and integration of functions of more than one variable. x. In this chapter we concentrate on diﬀerentiation. and in the last one. When studying the real world. as you know. and • how to interpret the process geometrically. so we have y = f (x. Note the trick I have just used. and indeed. it is unusual to have functions which depend solely on a single variable. typically written as y = f (x). or the pressure of a gas may depend on its volume V and temperature T . you did a signiﬁcant amount of work studying functions. For example you might have been interested in the height y after a given time x. in terms of applications. 8. Of course the single variable situation is a little simpler to study. the volume V of a cylinder depends on the radius r of the base and its height h.Chapter 8 Diﬀerentiation of Functions of Several Variables We conclude with two chapters which are really left over from last year’s calculus course. We describe • how each process can be done.1 Functions of Several Variables Last year. For example the height y of a particle may depend on the position x and the time t. so P = P (V. which represented the variation that occurred in some (dependent) variable y. We are going to do the same thing now for functions of several variables. In the earlier part of the course. as another (independent) variable. or the area y. and for the function itself: we don’t always need separate symbols as in the y = f (x) example. move on to integration. changed. the usual rules of calculus could be applied. and which should help to remind you of the techniques you met then. we have extended this work by taking a more rigorous approach to a lot of the same ideas. it is often convenient to use P both for the (defendant) variable. and results such as the time when the particle hits the ground. Once the function was known. could be calculated.

we shall usually have a “standard” function name. rather than one (independent) variable. y).1. and to see how we could get the same result using calculus. We then used such a graph to pick out points such as the local minimum at x = 3/2. DIFFERENTIATION OF FUNCTIONS OF SEVERAL VARIABLES Graph of $y = x^2 . We know the surface will be a plane. y). . . Working with two or more independent variables is more complicated. So we represent a function as a surface rather than a curve.2) y (0.1: Graph of a simple function of one variable why we started with it last year. Thus it is enough to plot three points that the plane passes through. xn ). Graphing functions of Several Variables One way we tried to understand the function y = f (x) was by drawing its graph. y. Sketch the surface given by z = 2 − x/2 − 2y/3 Solution. as shown in Fig 8. There are graphical .0) (4. since most of the extra complications occur when we have two. 8.1.0) x Figure 8. and then try to sketch the resulting surface in three dimensions. . or even y = f (x1 .0. 8. And just as last year. Example. but the ideas are familiar. 10 8 6 4 2 0 -2 -4 -2 -1 0 1 2 3 4 Figure 8. instead of y = f (x). and we don’t need to consider more general cases like w = f (x.2. because z is a linear function of x and y. x2 . z (0. we often work with z = f (x. y) we think of z as the height of the function f at the point (x. This gives Fig.3.3x$.0. .2: Sketching a function of two variables Of course it is easy to sketch something as simple as a plane.78 CHAPTER 8. z). To plot z = f (x.

1. so form the lines y = x and y = −x. One approach is to draw a contour map of the surface. And if there are three or more independent variables. sketch the surface given by z = x2 + y 2 . or otherwise. Sketch the surface given by z = x2 − y 2 . We now describe how to looking at similar examples without such a program.8.3: Surface plot of z = x2 − y 2 . . We this get the alternative representation as a contour map shown in Fig. Example. so x2 −y 2 = 1. 8. then z = a2 − y 2 . So far we have looked at the intersection of the planes z = k with the surface z = x2 − y 2 for diﬀerent values of the constant k. By looking at the curves where z is constant. For the surface z = x2 − y 2 . at the value a. and look at the points where z = 1. which make sketching and visualisation rather harder than for functions of one variable.2.3. We can continue in this way. z = a2 .3. A ﬁnal way to conﬁrm that you have the right view of the surface is to section it in diﬀerent planes.3 shows the resulting surface. 8. This is one of the hyperbolae shown Fig. 8. We can represent the surface directly by drawing it as shown in Fig. 20 15 10 5 0 -5 -10 -15 -20 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 Figure 8.4. Exercise. Each of these curves is a parabola with its vertex upwards. 8. 8.4. there are some tricks. If instead we ﬁx x. Such a representation is easy to create using suitable software and Fig. but with diﬀerent constants. the points where z = 0 lie on x2 = y 2 . FUNCTIONS OF SEVERAL VARIABLES 79 diﬃculties when dealing with more complicated functions. ﬁxing z at diﬀerent values shows the contours (lines of constant height or z value) are all the same shape. at the point y = 0. But for just two independent variables. Solution. indeed. 8. and then use the usual tricks to visualise the surface. there is really no good way of visualising the behaviour of the function directly.

DIFFERENTIATION OF FUNCTIONS OF SEVERAL VARIABLES 4 3 2 1 0 -1 -2 -3 -4 -4 -3 -2 -1 0 1 2 3 4 Figure 8. . so that throughout the process. The missing points near the x . and we have now seen that it doesn’t.80 CHAPTER 8. 0) = 2x. Consider ﬁrst the case when x → 0 along the x-axis. Continuity As you might expect. y0 ) if x→x0 . the usual “putting together” theorems show that f is continuous everywhere else. x2 + 0 Next consider the case when x → 0 and y → 0 on the line y = x. so f is not continuous at (0. y0 ). Investigate the continuity of f (x. Example.4: Contour plot of the surface z = x2 − y 2 . x) = 2x2 =1→1 x2 + x 2 as x → 0. 0). x + y2 Solution. we say that a function f of two variables is continuous at (x0 . We illustrate with a simple example.0 = 0 → 0 as x → 0. y) = f (x0 . so we are looking at the special case when x = y.4. y) = 2 at the point (0.y→y0 lim f (x. We have f (x. Of course f is only continuous if it has the same limit however x → 0 and y → 0.axis are an artifact of the plotting program. We have f (x. 2xy 8. Although we won’t go into it. 0). The only complication comes when we realise that there are many diﬀerent ways if which x → x0 and y → y0 . y = 0.

y0 ). and its partial derivatives.2. Treating ﬁrst y and then x as constants. ∂x f1 (x0 . y0 ). such as (x0 . y0 ) and D1 F (x0 . It is for this reason that a new symbol for diﬀerentiation is introduced. and diﬀerentiate with respect to y.5. We shall meet a number of examples of such a partial diﬀerential equation later. to get ∂f 2 cos x (2y + cos x). We thus often write the partial derivative as ∂f (x0 .2 = . . All of the following mean the same thing:∂f (x0 . Example. but we now require to treat the variables one at a time.2 − 2y. Note: This equation is an equation satisﬁed by the function we started with. Compute x ∂z 1 = cos ∂x y thus x ∂z ∂z x +y = cos ∂x ∂y y x y − x cos y x y = 0. to get a partial derivative ∂f 2y sin x = ∂x (2y + cos x)2 where we have diﬀerentiated with respect to x as usual. y0 ). Or we can treat x as a constant. ∂x There are a number of diﬀerent notations in use to try to help understanding in diﬀerent situations. y0 ) − f (x0 . x y and ∂z −x = 2 cos ∂y y x y . y) = 2y y + cos x We can consider y ﬁxed. PARTIAL DIFFERENTIATION 81 8.2 Partial Diﬀerentiation The usual rules for diﬀerentiation apply when dealing with several variables. keeping the others constant. ∂z ∂z +y . y0 ) = lim ∂x δx→0 δx provided of course that the limit exists. Let z = sin(x/y). which involves both the function. y0 ). ∂x ∂y Solution. we often want to evaluate it at some ﬁxed point. and so treat it as a constant. fx (x0 .8. we have 8. Consider the function f (x. y0 ). Note also that there is a simple deﬁnition of the derivative in terms of a Newton quotient:∂f f (x0 + δx. = 2 ∂y (2y + cos x) (2y + cos x)2 Although a partial derivative is itself a function of several variables.

y) + f (x0 . x0 = x(t0 ) etc. With our next result. DIFFERENTIATION OF FUNCTIONS OF SEVERAL VARIABLES ∂z ∂z +y = 0. We state a typical example df when t = π/2. Then F is diﬀerentiable and dF ∂f dx ∂f dy = + . 8. these examples are quite typical. y) = f (x/y) satisﬁes this partial diﬀerential equation. df ∂f dx ∂f dy = + = −y(t) sin t + x(t) cos t = −1. dt ∂x dt ∂y dt The chain rule easily extends to the several variable case. pronounced “Xi” is the Greek letter “x”. Let F (t) = f (x(t). and y and y0 respectively. and note that in this case. Write x = x(t). y) − f (x0 . But there are variants. the function z(x. y0 ) ∂f ∂f = (ξ. Note that ξ. dt .18) to write ∂f (ξ. y) = F (t) − F (t0 ) ∂f (y − y0 ) (x − x0 ) ∂f = + (ξ. dt ∂x dt ∂y dt Proof. and since ξ and η are trapped between x and x0 . y0 ) = f (x. x → x0 and y → y0 . y) − f (x0 . y) (x0 .8. and let x = cos t. y = sin t. Let f (x. pronounced “Eta” is the Greek letter “y”. η)(y − y0 ) ∂x ∂y Here we have used the Mean Value Theorem (5. Let z = log(x/y). The fact that the last two ∂x ∂y function satisfy the same diﬀerential equation is not a co-incidence. Here is how we diﬀerentiate compositions. Exercise. and that x = x(t) and y = y(t) are themselves diﬀerentiable functions of t. ∂x ∂y Because the deﬁnitions are really just version of the 1-variable result. Example. where f is suitably diﬀerentiable. y)(x − x0 ) + (x0 . most of the usual rules for diﬀerentiation apply in the obvious way to partial derivatives exactly as you would expect. Show that x +y = 0. y) − f (x0 . then also ξ → x0 and η → y0 . y(t)). Show that x 8. in the same way η.9. Exercise. only the notation is complicated. sin(π/2) = −1. y)(x − x0 ) ∂x for some point ξ between x and x0 . y) = xy. Theorem. y) − f (x0 . η) t − t0 ∂x t − t0 ∂y t − t0 Now let t → t0 . and have argued similarly for the other part. Compute Solution. Assume that f and all its partial derivatives fx and fy are continuous. From the chain rule. Then we calculate the Newton quotient for F . we can see that for any suitably diﬀerentiable function f .6.7. Let z = f (x/y). The result then follows from the continuity of the partial derivatives. Thus f (x. F (t) − F (t0 ) = f (x. ∂z ∂z 8. 8.82 CHAPTER 8.

= 1. perhaps to deﬁne a new function g(x.−1= − ∂u ∂v ∂w ∂u ∂w ∂f ∂f ∂f ∂f ∂f . = 0.0 = − + ∂u ∂v ∂w ∂u ∂v ∂f ∂f ∂f ∂f ∂f . Adding then gives the result claimed. v). We apply the chain rule. . = 0.8. y. v). y. y. we have ∂u ∂x ∂u ∂y ∂u ∂z Then ∂F ∂x ∂F ∂y ∂F ∂z = = = ∂f ∂f ∂f ∂f ∂f . = −1.0 + . z). z. v)). and let f be a function deﬁned on a subset U ∈ R 3 . Let F (x. although strictly speaking. But note the confusion if you ever want to talk about the value f (y. v. these names are not bound to the corresponding places. w) has continuous partial derivatives. v) and z = z(u. Proposition.0 + . z) have continuous partial derivatives This has the advantage that you are reminded of the names of the variables on which f acts. We often do it more quickly by saying Let f (x.−1+ . = 1. v) = f (x(u. y. x). 8.11. y = y(u. ∂F ∂F ∂F + + = 0. and that u = x − y. ∂w ∂x ∂w ∂y ∂w ∂z = −1. PARTIAL DIFFERENTIATION 83 8.1 = − + ∂u ∂v ∂w ∂v ∂w = 1.10. v). simply to show it was a function of three variables is clumsy. z) = f (u(x. ∂x ∂y ∂z Solution. v). z)).2. y(u.1 + . z). y. v(x. y. Then ∂F ∂f ∂x ∂f ∂y ∂f ∂z = + + ∂u ∂x ∂u ∂y ∂u ∂z ∂u and ∂F ∂f ∂x ∂f ∂y ∂f ∂z == + + . ∂v ∂x ∂v ∂y ∂v ∂z ∂v The introduction of the domain of f above. and suppose that all the partial derivatives of f are continuous. noting ﬁrst that from the change of variable formulae. Show that v =y−z w = z − x.−1+ . Let x = x(u. w(x. = −1. This is an example where we adopt the notation which is very common in engineering maths. z).1 + . Example. z(u. Assume that f (u. ∂v ∂x ∂v ∂y ∂v ∂z = 0. Write F (u.

there are four derivatives of second order.84 CHAPTER 8. when f has mild restrictions. and so further partial derivatives can be calculated. ∂2f ∂2f ∂2f ∂2f = +2 + 2 2 2 ∂u ∂x ∂x∂y ∂y and v2 ∂2f ∂f ∂2f ∂f ∂2f ∂2f = − + 2 +2 + 2 ∂v 2 ∂y ∂x ∂x ∂x∂y ∂y You may assume that all second order derivatives of f exist and are continuous. ∂x2 ∂2f . we have ∂f ∂f ∂x ∂f ∂y ∂f ∂f = + = + ∂u ∂x ∂u ∂y ∂u ∂x ∂y and ∂f ∂f ∂x ∂f ∂y 1 ∂f 1 ∂f = + = − . ∂y 2 Fortunately.13. = ∂y ∂y ∂y 2 This notation generalises to more than two variables. with the usual convention. we have ∂2f ∂v 2 = = ∂ 1 ∂f 1 ∂f 1 ∂f 1 ∂ ∂f 1 ∂f 1 ∂ ∂f − =− 2 + + 2 − ∂v v ∂x v ∂y v ∂x v ∂v ∂x v ∂y v ∂v ∂y 2f 2f 2f 1 ∂f 1 ∂ 1 1∂ 1 ∂f 1 1 ∂ 1 ∂2f = − 2 − + + 2 − − v ∂x v v ∂x2 v ∂y∂x v ∂y v v ∂x∂y v ∂y 2 2f 2f 2f 1 ∂f ∂ ∂f ∂ ∂ = − + 2 −2 + 2 . ∂v ∂x ∂v ∂y ∂v v ∂x v ∂y ∂2f ∂2f ∂2f ∂2f + + + 2. as follows: ∂2f . DIFFERENTIATION OF FUNCTIONS OF SEVERAL VARIABLES 8. There is a complication that does not occur when dealing with functions of a single variable.e. Show that. ∂x ∂y ∂x∂y ∂ ∂f ∂2f = . y) is written in terms of u and v where x = u + log v and y = u − log v. Using the chain rule. ∂x∂y ∂y∂x 8. Example. ∂x2 ∂x∂y ∂y∂x ∂y Thus using both these and their operator form. Assume that all second order derivatives of f exist and are continuous. the order in which the diﬀerentiation is done doesn’t matter. ∂x∂y ∂2f ∂y∂x and ∂2f . ∂2f ∂2f = . We write ∂ ∂f ∂2f . Solution. Suppose that f (x. we have ∂2f ∂ ∂f ∂f ∂ ∂ ∂f ∂f = + = + + 2 ∂u ∂u ∂x ∂y ∂x ∂y ∂x ∂y while diﬀerentiating with respect to v. Proposition. Then the mixed second order partial derivatives of f are equal. 2 v ∂y ∂x ∂x ∂x∂y ∂y . i. and to more than two derivatives in the way you would expect.12. = ∂x ∂x ∂x2 ∂ ∂f ∂2f = . 8. ∂y ∂x ∂y∂x ∂ ∂f ∂2f .3 Higher Derivatives Note that a partial derivative is itself a function of two variables.

∂t2 ∂x f(x. and allowed to move under the elastic forces of the string. would normally be written as f ’s By the chain rule ∂ = ∂x ∂ ∂u . SOLVING EQUATIONS BY SUBSTITUTION 85 8.(−c) + ∂ ∂v . . at least one space variable. We illustrate with just one example. 8.15. all the F ’s that appear below. Example. the function f satisﬁes the wave equation ∂2f ∂2f = c2 2 . because the two variables appear inde∂F pendently. b where h is also an arbitrary (diﬀerentiable) function. t). v) = g1 (v)dv + h(u) = g(v) + h(u). Assume the displacement of a length of string at time t from its rest position is described by the function f (x.t) a Figure 8. v) and now in principle confuse F with f .8.4. t) = F (u. Example. Write f (x. The laws of physics describe how the string behaves when released in this position.4. Rewrite the wave equation using co-ordinates u = x − ct and v = x + ct. Solve the equation ∂2F = 0. but note that in a practical case. together with time. So = g1 (v). Thus we can integrate with ∂u∂v respect to v to get F (u.g. where g1 is an arbitrary (diﬀerentiable) function. We move away from 1-variable results as soon as we have properties which depend on e.1 + ∂ ∂v .5: A string displaced from the equilibrium position 8.4 Solving equations by Substitution One of the main interests in partial diﬀerentiation is because it enables us to write down how we expect the natural world to behave. In practice we use diﬀerent symbols to help the learning process.1 and ∂ = ∂t ∂ ∂u .14.c. This is illustrated in Fig 8. designed to whet the appetite for the whole subject of mathematical physics. Solution. Solution. ∂u∂v Such a function is easy to integrate. since when ∂v ∂2F diﬀerentiated with respect to u we are given that = 0. so we can tell them apart only by the names of their arguments.

Show that ∂2F ∂2F − c2 2 = 0. Thus solutions to the wave equation are of the form f (u) + g(v) for any (suitably diﬀerentiable) functions f and g. b) is that it have a critical point there. one use for derivatives in several variables is in calculating maxima and minima. t) = log(2x + 2ct) for x > −ct. the principle remains the same. ∂u∂v Substituting in the wave equation. DIFFERENTIATION OF FUNCTIONS OF SEVERAL VARIABLES diﬀerentiating again. ∂u∂v an equation which we have already solved.17. b) if and only if ∂f ∂f (a. b) = (a. we thus get 4c2 ∂2F = 0.5 Maxima and Minima As in one variable calculations.86 CHAPTER 8. 8. then it has a global maximum and a global minimum. Exercise.16. ∂x ∂y It is clear by comparison with the single variable result. while in R 2 . We refer to this as the ﬁrst derivative test. we note that these must occur either at a local maximum or minimum. ∂t2 ∂x Note that this is simply checking a particular case of the result we have just proved. And we can test for local maxima and minima in the same way as for one variable. it must be constant when x = ct. 8. For example we may have sin(x − ct). or else on the boundary of the region. ∂2f ∂t2 = c ∂ ∂ −c ∂v ∂u c ∂F ∂F −c ∂v ∂u ∂2F ∂2F ∂2F ∂2F − c2 − c2 + c2 2 ∂v 2 ∂u∂v ∂v∂u ∂u ∂2F ∂2F ∂2F − 2c2 = c2 + ∂v 2 ∂u2 ∂u∂v = c2 and similarly ∂2f ∂x2 = ∂2F ∂2F + ∂v 2 ∂u2 +2 ∂2F . Say that f (x. y) has a critical point at (a. where c is a ﬁxed constant. Of course in R . . although that is not a suﬃcient condition. Note that this is not just any function. that a necessary condition that f have a local extremum at (a. Deﬁnition. and using the operator form of the chain rule as well. we shall rely on the theorem that if f is continuous on a closed bounded subset of R2 . the boundary of the region usually consisted of a pair of end points. And again as before. Let F (x. However. b) = 0. the situation is more complicated. Again as for one variable. 8.

8. We have f (a + h. 8. where we have written A= ∂2f .18. ∂x∂y and C= ∂2f . Of course we know it has a global minimum there. We give an indication of how the theorem can be derived — or if necessary how it can be remembered. ∂y 2 . b) + h ∂f ∂f 1 (a. Then 2 • If. Let f (x. then f has a local minimum at (a. b). we shall use ∂x fxx instead of ∂2f . 2 The test is inconclusive at (a. so for any pair (h.5. 0). at (a. see section 5. but here goes with the test: Solution. Note that the discriminant is easily remembered as ∆= fxx fxy fyx fyy 2 = fxx fyy − fxy A number of very simple examples can help to remember this. 2 • If. b). b)+k (a. fxy = 0. Recall that we gave a number of diﬀerent notations for partial derivatives. the result of the test should work on things where we can do the calculation anyway! 8. at (a. b) + ∂x ∂y 2 h2 ∂2f ∂2f ∂2f + 2kh + k2 2 ∂x2 ∂x∂y ∂y where we have actually taken an expansion to second order and assumed the corresponding remainder is small. k). 2 • If. then f has a local maximum at (a. ∂x2 B= ∂2f . Theorem (Second Derivative Test). b) = ∂x ∂y 0 and everything hinges on the behaviour of the second order terms. b) is a critical point for f . Show there is a unique critical point. It is thus enough to study the behaviour of the quadratic Ah2 + 2Bhk + Ck2 . and this is the 2 only critical point. b). b + k) ∼ f (a. b) if fxx fyy − fxy = 0. fy = 2y. so fx = fy precisely when x = y = 0. Show that f (x. MAXIMA AND MINIMA 87 We can get more information by looking at the second derivative.6.20. ∂f ∂f We are looking at a critical point. we have fxx > 0 and fxx fyy − fxy > 0. b). We start with the two dimensional version of Taylor’s theorem. we have fxx < 0 and fxx fyy − fxy > 0. Exercise. This idea extends to higher derivatives. We have fxx = fyy = 2. 0). which is a saddle point Proof. Assume that (a. We have fx = 2x. After all. ∂x2 and fxy instead of ∂2f ∂x∂y etc.8. at (a.19. we have fxx fyy − fxy < 0. we have h (a. and in what follows we use fx rather ∂f than the more cumbersome etc. and the investigation has to be continued some other way. y) = x2 + y 2 has a minimum at (0. b) + k (a. b). b). Example. so ∆ = fxx fyy − fxy = 4 > 0 and there is a local minimum at (0. y) = xy. then f has a saddle point at (a.

Solution. Thus the discriminant is ∆ = −6. Exercise. and • if ∆ < 0 then the coeﬃcients of the two squared terms have opposite signs. 0).(12x − 12) − 36. since x3 → ∞ as x → ∞. y) = xy − x2 − y 2 − 2x − 2y + 4.23. so by going out in two diﬀerent directions. .22. What is the maximum value of the surface area? [You may assume that the maximum exists. we have h = 30 − 2(x + y). • if A > 0 and ∆ > 0 we have a local minimum.21. and that the corresponding dimensions of the tank are strictly positive. Find the extrema of f (x. To see there is no global maximum. Substituting. Diﬀerentiating again.] Solution. 0) = 2x3 (1 − 3/x) → ∞ as x → ∞. Note also that we could have completed the square in the same way. show that f does not achieve a global maximum. −1). we have a local maximum at (0. When x = 0. 8. An open-topped rectangular tank is to be constructed so that the sum of the height and the perimeter of the base is 30 metres. the quadratic may be made either to increase or to decrease. fxx = 12x − 12. rather than the h term. ∆ = −36 < 0. DIFFERENTIATION OF FUNCTIONS OF SEVERAL VARIABLES Assuming that A = 0 we can write Ah2 + 2Bhk + Ck2 = A h + =A h+ Bk A Bk A 2 + C− + ∆ A B2 A k2 k2 2 where we write ∆ = CA − B 2 for the discriminant. or otherwise. but starting from the k term. Let the area of the surface of the material be S. so there is a saddle at (1. 8. Example. We have fx = 6x2 − 12x− 6y and fy = −6y − 6x. Thus critical points occur when y = −x and x2 − x = 0. It is thus clear that • if A < 0 and ∆ > 0 we have a local maximum. 0) and (1. y) = 2x3 − 6x2 − 3y 2 − 6xy. Find the dimensions which maximise the surface area of the tank. Then S = 2xh + 2yh + xy. Example. and since. from our restriction on the base and height. note that f (x. ∆ = 36 > 0 and since fxx = −12. When x = 1. and so at (0.88 CHAPTER 8. Find and classify the critical points of f . 30 = 2(x + y) + h. 0). fyy = −6 and fxy = −6. Let the dimensions of the box be as shown. we have S = 2(x + y) 30 − 2(x + y) + xy = 60(x + y) − 4(x + y)2 + xy. Let f (x. By considering f (x. −1). so the result could just as easily be stated in terms of C instead of A 8. We have thus expressed the quadratic as the sum of two squares.

this must be it.8. we have x = 0 or y = 0 or x + y = 15. In this case exactly as in the one variable case. y) = 2 + 2y − y 2 . or x = y = 4. If y = 0.] Sometimes a function necessarily has an absolute maximum and absolute minimum — in the following case because we have a continuous function deﬁned on a closed bounded subset of R 2 . S is deﬁned for x ≥ 0. On the boundary of ∂x ∂y the domain of deﬁnition of S. Example. Thus h = 14 and the surface area is S = 16(−4 − 4 − 7 + 15 + 15) = 240 square metres. but the critical proportions are still the same. If y = 9 − x. which in fact reduce to 1-variable methods) and the critical points in the interior. investigate f (x.hoc methods. the area is doubled. or at a critical point. using our ability to ﬁnd local maxima. = −8x − 7y + 60 = 0. y ≥ 0 and x + y ≤ 15. while if x = 0. when = = 0. 0) = 2 + 2x − x2 . 8. y = 0 and y = 9−x Solution. [If both sides of the surface are counted. y) = 2 + 2x + 2y − x2 − y 2 on the triangular plate in the ﬁrst quadrant bounded by the lines x = 0. we need only search the boundary (using ad . We know there is a global maximum. Now S ∂S ∂x ∂S ∂y = −4x2 − 4y 2 − 7xy + 60x + 60y. and so the analogue of 4. A global maximum (which we are given exists) can only occur on the boundary of the ∂S ∂S domain of deﬁnition of S. investigate f (0. Since we are given that a maximum exists.24. investigate f (x. at a critical point. because the function is continuous on a closed bounded subset of R 2 . so Subtracting gives x = y and so 15x = 60. = −8y − 7x + 60 = 0.5. in which case h = 0.35 holds. 9 − x) = 2 + 2x + 2(9 − x) − x2 − (9 − x)2 . Find the absolute maximum and minimum values of f (x. We are given that we may ignore these cases. or on the boundary.6: A dimensioned box and for physical reasons. Thus the absolute max will occur either in the interior. MAXIMA AND MINIMA 89 h x y Figure 8.

then for any other point (x. which lies on the surface. and so in the tangent. Find the equation of the tangent plane to the surface F (x. z − z0 ) and F are perpendicular. Example. z0 ) is given by ∂F ∂F ∂F (x − x0 ) + (y − y0 ) + (z − z0 ) = 0. 8. y.90 CHAPTER 8. y(t). writing this in terms of vectors. 1) and f (1. z0 ) lies on the surface. 2.e. y. z(t)). dx dy dz . to F ). Theorem. 3. and so must be normal to the normal to the curve (i. . DIFFERENTIATION OF FUNCTIONS OF SEVERAL VARIABLES for an absolute maximum. Then since the curve lies in the surface. 8. that it is a local maximum.v(t) = ∂F ∂F ∂F . f takes the values −41/2. we see that the left hand vector must be the normal to the curve. or (9/2. 0) or (0. The tangent to the surface F (x. y(t). so this must be the global maximum. z) in the tangent plane. Since the RH vector is the velocity of a point on the curve. . and suppose that f and F have continuous partial derivatives. we have F. say φ(t) = (x(t). ∂x ∂z ∂z . perhaps as z = f (x. y) = (0. 1) or (1. z) = c at the point (x0 . the vector (x− x0 . −61. 0) or (9. y). Note that we have deﬁned the gradient vector F associated with the function F by F = ∂F ∂F ∂F . 9). and so.26. z − z0 ) must lie in the tangent plane. y0 . y − y0 . ∂x ∂y ∂z Proof. y. At these points. z) = c. dt ∂x dt ∂y dt ∂z dt or. so (x. we have F (x(t). 9/2).6 Tangent Planes Consider the surface F (x. we have dF ∂F dx ∂F dy ∂F dz = + + = 0. y − y0 . and that requirement is the equation which gives the tangent plane. y) = (1. Can check also using the second derivative test. . applying the chain rule. Given that (x0 .25. This is a simple example of the use of vector geometry. Next we seek critical points in the interior of the plate. 8. 0) or (0. fx = 2 − 2x = 0 and fy = 2 − 2y = 0. . dt dt dt = 0. 2. y0 . z) = x2 + y 2 + z − 9 = 0 at the point P = (1. y. Thus (x − x0 . In fact extreme may occur when (x. z(t)) = c. 1) = 4. ∂x ∂z ∂z . 4). Suppose now we have a smooth curve on the surface.

z) = z − f (x. we can use it to see how a function changes when its independent variables are subjected to small changes. ∂x ∂y ∂f ∂f (x0 . We have F |(1. y0 ) + We call the right hand side the linear approximation to f at (x0 . y0 ). Solution.h0 ) Thus the volume is 10 times as sensitive to errors in measuring r as it is to measuring h. or that f (x. Then V = πr 2 H. ∂x ∂y ∂z and writing the derivatives in terms of f . y0 )(y − y0 ) + (−1)(z − f (x0 . y). y0 )) = 0. 8. or F (x. y0 ) + ∂f ∂f (x0 . Example. yo . y0 ) ≈ h ∂f ∂f (x0 . the height of the tangent plane above the ground plane. Just as in one variable. Show that the tangent plane to the surface z = 3xy − x3 − y 3 is horizontal only at (0. Try this with a short fat tank! . y0 ). y0 )(x − x0 ) + (x0 . y) ≈ f (x0 .4) = (2. ∂x ∂y or. We can rewrite this with h = x − x0 and k = y − y0 . 91 8. 1).y0 ) This has applications. ∂x ∂y Our assumption that the tangent plane lies close to the surface is that z ≈ f (x. y). How sensitive is the volume of the tank to small variations in the radius and height.2. f (x0 .y0 ) ∂f ∂y (x0 . and the equation of the tangent plane is 2(x − 1) + 4(y − 2) + (z − 4) = 0. z = f (x0 . y) − f (x0 . Exercise. so we shall use the tangent plane to approximate the surface deﬁned by a function of two variables. Note that the tangent plane to this surface at the point (x0 .h0 ) ∂h = 250πdr + 25πdh.27. 4. y) = 0. y0 )(y − y0 ). we used the tangent line to approximate the graph of a function.7 Linearisation and Diﬀerentials We obtained a geometrical view of the function f (x. dh. y0 ) + k (x0 . A cylindrical oil tank is 25 m high and has a radius of 5 m. and so dV = ∂V ∂V dr + ∂r (r0 . y0 )(x − x0 ) + (x0 . 0. y0 )(x − x0 ) + (x0 . 1). y. to get f (x. (r0 . 0) and (1. writing in terms of z.7. 8. The equation of our tangent plane is ∂F ∂F ∂F (x − x0 ) + (y − y0 ) + (z − f (x0 . y0 )) = 0. we have ∂f ∂f (x0 . Let V be the volume of the cylindrical tank of height h and radius r. ∂x ∂y or df ≈ h ∂f ∂x +k (x0 . LINEARISATION AND DIFFERENTIALS Solution. y0 )) lies close to the surface itself. 1. y) by considering the surface z = f (x.8. y0 )(y − y0 ).28.

∂x ∂F ∂F ∂z 0= + ∂x ∂z ∂x ∂F ∂z = − ∂x ∂F ∂x ∂z ∂ ∂F ∂x ∂F ∂y ∂F ∂z F (x. x). z) and y = y(z.29. y. and h measured with an accuracy of 0. z(x. with x = x(y.(0. DIFFERENTIATION OF FUNCTIONS OF SEVERAL VARIABLES 8. Then form three such equations. The volume of a cylindrical oil tank is to be calculated from measured values of r and h. if r is measured with an accuracy of 2%. and the height an error of 2%. Assume we have variables x. y)) = + + .5%. The volume V of a cone is given by V = πr 2 h/3. Then assuming that Fz = 0. + .8 Implicit Functions of Three Variables Finally in this section we discuss another application of the chain rule. ∂x ∂x ∂x ∂y ∂x ∂z ∂x and Now assume in the same way that Fx = 0 and Fy = 0. and h is the height. related by the equation F (x. Exercise.03).(0. ∂F ∂F ∂F ∂z ∂x ∂y ∂y ∂z = −1! . . y. y). Doing this gives F (x.30.06 + 0. What is the percentage error in the volume. Now diﬀerentiate both sides partially with respect to x. Example. so we can get two more relations like this. = − ∂x ∂F ∂F ∂F ∂x ∂y ∂z ∂z ∂x ∂y We met this result as Equation 1. 8. when it seemed totally counter-intuitive! . so we can “solve for z”. z) = 0. 8. Thus dV 2 1 πrh dr + πr 2 dh 3 3 πr 2 h πr 2 h = 2. The radius has a measurement error of 3%.1 in Chapter 1. where r is the radius of the cone. there is a version of the implicit function theorem which means we can in principle.02) = V (0. A cone is measured. What is the error in measuring the volume? Solution. write z = z(x. We get 0= and so since ∂y = 0.92 CHAPTER 8. y. z(x.02) 3 3 = Thus there is an 8% error in measuring the volume. y and z. y)) = 0.

while the fundamental theorem of the calculus enables us to compute this using the process of “anti-diﬀerentiation” — undoing the diﬀerentiation process. the order does not matter when calculating the answer.2 Repeated Integrals and Fubini’s Theorem As might be expected from the form. yj )dxi dyj = lim Smn . 93 . where the ﬁrst sum is thought of as a limiting case. in which we can sum over the elementary rectangles dx dy in any order. adding up the areas of a number of rectangles each of height f (xi ).Chapter 9 Multiple Integrals 9. We write such a double integral as R f (x. y) as representing the height of f at the point (x. This gives rise to the concept of the repeated integral. We think of the area as f (xi ) dxi = f (x) dx. y) dy dx. y) and area dxi dyj . y) dx R dy or R f (x. y) in the plane. There are two important orders — where we ﬁrst keep x constant and vary y. 9. and interpret the integral as the sum of the volumes of a number of small boxes of height z = f (x.1 Integrating functions of several variables Recall that we think of the integral in two diﬀerent ways. and width dxi . which we write as f (x. This leads to the natural generalisation to several variables: we think of the function z = f (x. y) lying above a certain region R in the plane leads to integrals of the form n m = R i=1 j=1 f (xi . Thus the volume of the solid of height z = f (x. and the opposite way round. In one way we interpret it as the area under the graph y = f (x). and then vary x. y) dA.

Let f (x. since we have no direct way of calculating a double (as opposed to a repeated) integral. and sketch the region of integration. y) dA = R a g(x) dx c h(y) dy. Exercise. Then b g2 (x) f (x. 3 2 0 9. y) dy dx Note that this is something like an inverse of partial diﬀerentiation. we recognise V as volume under the plane z = 4 − x − y which lies above {(x. y) be continuous on the rectangular region R : a ≤ x ≤ b. y) dA = R c h1 (y) f (x. h1 (y) ≤ x ≤ h2 (y). Evaluate 0 (4 − y 2 ) dy dx. g1 (x) ≤ y ≤ g2 (x). y) dA = R c a f (x. MULTIPLE INTEGRALS Our result that the order in which we add up the volume of the small boxes doesn’t matter is the following. c ≤ y ≤ d. Then d b b d f (x. In fact Fubini’s theorem is valid for more general regions than rectangles. y) dx dy . using Fubini’s theorem. In doing the ﬁrst inner (or repeated) integral. y) dx dy = a c f (x. since b d f (x. Integrate z = 4 − x − y over the region 0 ≤ x ≤ 2 and 0 ≤ y ≤ 1. 9. say f (x. Theorem (Fubini’s theorem — Stronger Form). From our interpretation of the integral as a volume.1. y) be continuous on a region R • if R is deﬁned as a ≤ x ≤ b. then it doesn’t matter which order we do the integration. Of course if f is a particularly simply function. We calculate the integral as a repeated integral. y) dy dx • if R is deﬁned as c ≤ y ≤ d. 9.4. Then d h2 (y) f (x. Let f (x. we keep y constant. which also formally shows that we evaluate a double integral as any of the possible repeated integrals. y) dA = R a g1 (x) f (x. Hence calculate the volume under the plane z = 4 − x − y above the given region.94 CHAPTER 9. 0 ≤ y ≤ 1}. Then we integrate with respect to y. 9. 2 1 y=0 2 x=0 2 V = x=0 (4 − x − y) dy dx = (4y − xy − y 2 /2) 1 0 dx = x=0 (4 − x − 1/2) dx etc.3.2. y) = g(x)h(y). and integrate with respect to x. Theorem (Fubini’s theorem for Rectangles). Here is a pair of statements which extend its validity. Solution. We use the Fubini theorem to actually evaluate integrals. y) | 0 ≤ x ≤ 2. Example.

9.2. REPEATED INTEGRALS AND FUBINI’S THEOREM

95

2

y

1

x 1 2

Figure 9.1: Area of integration. Proof. We give no proof, but the reduction to the earlier case is in principle simple; we just extend the function to be deﬁned on a rectangle by making it zero on the extra bits. The problem with this as it stands is that the extended function is not continuous. However, the diﬃculty can be ﬁxed. This last form enables us to evaluate double integrals over more complicated regions by passing to one of the repeated integrals. 9.5. Example. Evaluate the integral

2 1 x 2

y2 dy dx x2

as it stands, and sketch the region of integration. Reverse the order of integration, and verify that the same answer is obtained. Solution. The diagram in Fig.9.1 shows the area of integration. We ﬁrst integrate in the given order.

2 1 x 2

y2 dy dx = x2 =

2 1

y3 3x2

2

2

dx =

x 2 x2 1 1

8 x − 2 3x 3

dx = 5 . 6

−

8 − 3x 6

=

4 2 − − 3 3

8 1 − − − 3 6

**Reversing the order, using the diagram, gives
**

2 1 1 y

y2 dx dy = x2 =

2 1

y2 − x y2 2 −

y

2

dy =

1

−

1 3 2 y

−y + y 2 dy 8 3 1 1 − − + 2 3 = 5 . 6

3

=

1

−2 +

Thus the two orders of integration give the same answer. Another use for the ideas of double integration just automates a procedure you would have used anyway, simply from your knowledge of 1-variable results.

96

CHAPTER 9. MULTIPLE INTEGRALS

1

y

x 1

Figure 9.2: Area of integration. 9.6. Example. Find the area of the region bounded by the curve x2 + y 2 = 1, and above the line x + y = 1. Solution. We recognise an area as numerically equal to the volume of a solid of height 1, so if R is the region described, the area is

1 √ y= 1−x2 1

1 dx dy =

R 0 y=1−x

dy dx =

0

1 − x2 − (1 − x) dx = . . . .

And we also ﬁnd that Fubini provides a method for actually calculating integrals; sometimes one way of doing a repeated integral is much easier than the other. 9.7. Example. Sketch the region of integration for

1 0 y 1

x2 exy dx

dy.

Evaluate the integral by reversing the order of integration. Solution. The diagram in Fig.9.2 shows the area of integration. Interchanging the given order of integration, we have

1 0 y 1

x2 exy dx

1

x

dy =

0 1

x2 exy dy dx

1

dx

=

0 1

0 2 exy x x

x

2

0

=

0

x ex dx −

1 0

x dx

0

= 9.8. Exercise. Evaluate the integral

1 x 2 x2 e − 2 2

=

1 (e −2) . 2

√ ( x − y 2 ) dy dx,

R

where R is the region bounded by the curves y = x2 and x = y 4 .

9.3. CHANGE OF VARIABLE — THE JACOBIAN

97

9.3

Change of Variable — the Jacobian

Another technique that can sometimes be useful when trying to evaluate a double (or triple etc) integral generalise the familiar method of integration by substitution. Assume we have a change of variable x = x(u, v) and y = y(u, v). Suppose that the region S in the uv - plane is transformed to a region S in the xy - plane under this transformation. Deﬁne the Jacobian of the transformation as ∂x J(u, v) = ∂u ∂y ∂u ∂x ∂v = ∂(x, y) . ∂y ∂(u, v) ∂v

It turns out that this correctly describes the relationship between the element of area dx dy and the corresponding area element du dv. With this deﬁnition, the change of variable formula becomes: f (x, y) dx dy =

S S

f (x(u, v), y(u, v)) |J(u, v)| du dv.

Note that the formula involves the modulus of the Jacobian. 9.9. Example. Find the area of a circle of radius R. Solution. Let A be the disc centred at 0 and radius R. The area of A is thus

A

dx dy. We

evaluate the integral by changing to polar coordinates, so consider the usual transformation x = r cos θ, y = r sin θ between Cartesian and polar co-ordinates. We ﬁrst compute the Jacobian; ∂x = cos θ, ∂r Thus ∂x ∂r J(r, θ) = ∂y ∂r We often write this result as dA = dx dy = r dr dθ Using the change of variable formula, we have dx dy =

A A

∂y = sin θ, ∂r

∂x = −r sin θ, ∂θ

∂y = r cos θ. ∂θ

∂x cos θ −r sin θ 2 2 ∂θ ∂y = sin θ r cos θ = r(cos θ + sin θ) = r. ∂θ

|J(r, θ)| dr dθ =

R 0 0

2π

r dr dθ = 2π

R2 . 2

We thus recover the usual area of a circle. Note that the Jacobian J(r, θ) = r > 0, so we did indeed take the modulus of the Jacobian above. 9.10. Example. Find the volume of a ball of radius 1.

and so we instead do a change of variable ﬁrst. so we are able to do the dr and dθ parts of the integral at the same time. and the Jacobian is easily calculated as ∂(x. and so V =2 D 1 − x2 − y 2 dx dy where the region of integration D consists of the unit disc {(x. inside the sphere of radius 2a. The moment of inertia of a solid occupying the region R. Example.98 CHAPTER 9.9. Although we can try to do this integration directly. the integrand is a product. we show that the same ideas work in 3 dimensions. and above the x − y plane. 9. One of these. The second useful co-ordinate system is spherical polars with transformation x = r sin φ cos θ.3. y = r sin θ. It can be thought of as twice the volume enclosed by a hemisphere of radius 1 in the upper half plane.11. z = r cos φ. y = r sin θ. The ball is the set {(x. And ﬁnally. z) =r ∂(r. It is easy to check that Jacobian of this transformation is given by dV = r 2 sin φ dr dφ dθ = dx dy dz. As in 9. cylindrical polars is given by the transformation x = r cos θ. if we write x = r cos θ. θ. z) so dV = dx dy dz = r dr dθ dz. . There are (at least) two co-ordinate systems in R 3 which are useful when cylindrical or spherical symmetry arises. z = z. The transformation is illustrated in Fig 9. the natural co-ordinates to use are plane polars. Let V be the required volume. 3 Note that after the change of variables. y. z) | x2 + y 2 + z 2 ≤ 1}. y) | x2 + y 2 ≤ 1}. when rotated about the z . Thus V =2 D 1 − x2 − y 2 dx dy = 2 = 2 0 ( 1 − r 2 ) r dr dθ D 2π 1 dθ 0 1 − r 2 dr 1 0 (1 − r 2 )3/2 = 4π − 3 = 4π .axis is given by the formula I= R (x2 + y 2 )ρ dV. Calculate the moment of inertia about the z-axis of the solid of unit density which lies outside the cylinder of radius a. MULTIPLE INTEGRALS Solution. we have dx dy = r dr dθ. y. y = r sin φ sin θ.

. 9.4: Cross section of the right hand half of the solid outside a cylinder of radius a and inside the sphere of radius 2a Solution. Let I be the moment of inertia of the given solid about the z-axis. θ.3. A diagram of a cross section of the solid is shown in Fig 9. We thus have a single integral.4.3: The transformation from Cartesian to spherical polar co-ordinates. z).12. Using the substitution u = 4a2 − r 2 . z a 2a r x Figure 9. and then by doing the z integration ﬁrst and using plane polars. Exercise.9. CHANGE OF VARIABLE — THE JACOBIAN 99 z r ϕ y θ x Figure 9. We use cylindrical polar co-ordinates (r. Show that z 2 dxdydz = 4 π 15 (where the integral is over the unit ball x2 + y 2 + z 2 ≤ 1) ﬁrst by using spherical polars. the Jacobian gives dx dy dz = r dr dθ dz. you can check that the √ integral evaluates to 22 3πa5 /5. so 2π 2a √ 4a2 −r 2 I = 0 dθ a 2a a r dr 0 r 2 dz = 2π r3 4a2 − r 2 dr.

100 CHAPTER 9. MULTIPLE INTEGRALS .

30. 48 l’Hˆpital’s rule: simple form. A. 86 from above. 5 Ratio Test. 7 101 (or Monotone Convergence Principle. 5 geometric progression series). 6 Newton quotient. M. 91 local maximum. 90 half . 7 absolutely convergent.geometric mean inequality. 61 continuity. 87 Second Mean Value Theorem. 55 singularity. 22 bounded below. 5 Comparison Test. 67 properties of R. 62 limit from the left. 3 partial diﬀerential equation. 35 Fubini’s Theorem. (1967). 60 rational numbers. 59 completeness of R. 50 Mean Value Theorem. 68 range. 2 real numbers. 23 natural numbers. 67 repeated integral. 26 ﬁrst derivative test. 4 integers. 22 inequalities. Calculus. 87 Maclaurin’s Theorem. 92 increasing. 2 neighbourhood. 9 conditionally convergent. 80 convergent series. 34 linear approximation. 2 power series. 31 continuous. Index Entries absolute value. o 43 Leibniz Theorem. 3 completing the square.open. W. 5 double integral. 38 interval of convergence. 2 open. 5 ordering of R . 29. 68 intervals. 81 positive integers. 62 arithmetic . 5 Jacobian. 45 modulus.Bibliography Spivak. 86 cylindrical polars. 87 second derivative test. 44 saddle point. 32. 97 l’Hˆpital’s rule: o general form. 56 domain. 6 . 93 Fibonacci sequence. 8 bounded above. 94 Rolle’s Theorem. 61 alternating series test. 60 Intermediate Value Theorem. 5 implicit functions. 47 l’Hˆpital’s rule: inﬁnite limo its. 2 integral test. 2 radius of convergence. 93 function. 56 critical point. Benjamin. 41 numbers. 98 divergent series. 87 local minimum. 50 series. 55 Binomial Theorem. 55 gradient. 29 open interval. 2 real power series. 22 closed interval. 5 arithmetic progression.

5 Taylor series. 51 Taylor’s Theorem. 50 teacher redundant. 90 target space.102 spherical polars. 3 upper bound. 78 tangent planes. iii BIBLIOGRAPHY tends to. 49. 31 Triangle Inequlaity. 7 trichotomy. 22 . 56 surface. 98 sum of the series.

- Graduate Studies in Mathematics Volume 65
- Mathematics of the Discrete Fourier Transform
- Darling - Differential Forms
- Infinitesimal Calculus
- Vector Calculus - Theodore Voronov
- Solutions to Modern Algebra (Durbin, 5E)
- Algebra
- Singer Thorpe - Lecture Notes on Elementary Topology and Geometry
- Electro Magnetism Problems
- calculus
- Calculus I Complete
- CalcII_Complete
- Bystrom Applied Mathematics
- calculus
- The Fourier Transform and Its Applications
- Fermat's Last Theorem
- The Jordan Curve Theorem Via the Brouwer Fixed Point Theorem
- Viro, Ivanov, Netsvetaev Kharlamov - Elementary Topology Problem Textbook (2008)
- Junior Problem Seminar
- Applied Analysis Mathematical Methods in Natural Science
- Vinberg, A Course in Algebra
- (ebook-pdf) - Mathematics - Alder - Multivariate Calculus
- Calculus II Complete
- Markov Chain Random Walk
- Lowell W. Beineke & Robin J. Wilson & Peter J. Cameron - Topics in Algebraic Graph Theory
- Elementary Differential Equations Sixth Edition Edwards and Penney
- tomas hungerford

Skip carousel

- Laconic study on Incident / Accident Investigation technique -
- tmp6AC3.tmp
- Applications of algebra to a problem in topology - by Michael Hopkins
- A Survey of Modern Data Classification Techniques
- tmp6819.tmp
- UT Dallas Syllabus for ee6349.521.07u taught by Mohammad Saquib (saquib)
- UT Dallas Syllabus for ee6349.501.08s taught by Mohammad Saquib (saquib)
- tmp2D02.tmp
- tmpA940.tmp
- A Mathematical Theory of Communication
- tmp5573.tmp
- tmpA253.tmp
- frbrich_wp08-4.pdf
- On a Result of Fixed Point Theorems for Commuting and Weakly Compatible Maps
- tmpBF3
- UT Dallas Syllabus for ee6349.001.07s taught by Mohammad Saquib (saquib)
- Analysis of Residue Number System based PN sequence in AWGN channel
- tmp48DC.tmp
- A Comparison of Computation Techniques for DNA Sequence Comparison
- A Survey of Sequential Rule Mining Techniques
- An Enhanced Approach for Detecting User's Behavior Applying Country-Wise Local Search
- As 60068.2.61-2003 Environmental Testing Tests - Test Z ABDM- Climate Sequence
- Tmp 2330
- tmp4BA8.tmp
- 201450 Pap
- tmpA14.tmp
- Common Fixed Point of Mappings Using Sequentially Weak Compatible Mapping

- FIDIC (Conditions of Particular Application With Guidelines)
- Proje Tasdigi[1]
- Ced Prosedürü
- Multivariate Calculus - Alder
- FIDIC In Bahrain
- ÇED YÖNETMELİĞİ
- LİDERLİK
- Çevresel Etki Değerlendirmesi (ÇED)
- Ced Rapor Yazim Rehberi
- İnşaat Sektöründe Risk yonetimi
- Standart Teklif Mektupları
- Yapı (İnşaat) Ruhsatı
- İNŞAAT İŞLERİNE UYGUN TEKLİF BEDELİNİN REGRESYON ANALİZİ İLE BELİRLENMESİ
- Mimarlar için Tesisat El Kitabı
- İnşaat Ruhsatı Almak İçin Gerekli Belgeler
- İstikamet röleve
- Zemin Ve Temel Etüdü Raporunun Hazirlanmasina ilişkin Esaslar
- Meslekİ Sorumluluk sİgortasi Genel Şartlari
- Hakediş Nedir Nasıl düzenlenir
- Belediye Ruhsat Başvurusu
- İskan yap Kullanma İzni,İnşaat Ruhsatı Almak,İmar Hizmetleri Alma Koşulu
- Projeler İçin Belediyelerden İmar Ruhsatı Nasıl Alinir
- Yalitim hesaplarında Yapay sinir agi
- Kat Mulkiyeti Yasasi
- Yapi Maliyetinin Yapay sinir agi ile Analizi
- Mühendisin Hayatı (Fevzi Akkaya)
- Problemli Zeminlerde Geoteknik Çözümler

Sign up to vote on this title

UsefulNot usefulRead Free for 30 Days

Cancel anytime.

Close Dialog## Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

Close Dialog## This title now requires a credit

Use one of your book credits to continue reading from where you left off, or restart the preview.

Loading