Page 1 :
MOOCULUS, massive open online calculus, , sequences, and, series, THIS DOCUMENT WAS TYPESET ON APRIL 17, 2014.

Page 2 :
Copyright © 2014 Jim Fowler and Bart Snapp, This work is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike License. To view a copy of this license,, visit http://creativecommons.org/licenses/by-nc-sa/3.0/ or send a letter to Creative Commons, 543 Howard Street,, 5th Floor, San Francisco, California, 94105, USA. If you distribute this work or a derivative, include the history of the document., The source code is available at: https://github.com/kisonecat/sequences-and-series/tree/master/textbook, This text is based on David Guichard’s open-source calculus text which in turn is a modification and expansion of notes written by, Neal Koblitz at the University of Washington. David Guichard’s text is available at http://www.whitman.edu/mathematics/, , calculus/ under a Creative Commons license., The book includes some exercises and examples from Elementary Calculus: An Approach Using Infinitesimals, by H. Jerome Keisler,, available at http://www.math.wisc.edu/~keisler/calc.html under a Creative Commons license. In addition, the chapter, on differential equations and the section on numerical integration are largely derived from the corresponding portions of Keisler’s, book. Albert Schueller, Barry Balof, and Mike Wills have contributed additional material. Thanks to Walter Nugent, Nicholas J. Roux,, Dan Dimmitt, Donald Wayne Fincher, Nathalie Dalpé, chas, Clark Archer, Sarah Smith, MithrandirAgain, AlmeCap, and hrzhu for, proofreading and contributing on GitHub., This book is typeset in the Kerkis font, Kerkis © Department of Mathematics, University of the Aegean., We will be glad to receive corrections and suggestions for improvement at [email protected] or [email protected].

Page 3 :
Contents, , 1, , Sequences, , 13, , 1.1 Notation, , 13, , 1.2 Defining sequences, , 15, , 1.2.1, , Defining sequences by giving a rule, , 1.2.2, , Defining sequences using previous terms, , 1.3 Examples, , 15, , 17, , 1.3.1, , Arithmetic sequences, , 18, , 1.3.2, , Geometric sequences, , 18, , 1.3.3, , Triangular numbers, , 19, , 1.3.4, , Fibonacci numbers, , 20, , 1.3.5, , Collatz sequence, , 21, , 1.4 Where is a sequence headed? Take a limit!, 1.5 Graphs, , 16, , 24, , 1.6 New sequences from old, 1.7 Helpful theorems about limits, 1.7.1, , Squeeze Theorem, , 1.7.2, , Examples, , 29, , 28, , 25, 27, , 22

Page 4 :
4, , 1.8 Qualitative features of sequences, , 2, , 1.8.1, , Monotonicity, , 33, , 1.8.2, , Boundedness, , 34, , 32, , Series, , 39, , 2.1 Definition of convergence, 2.2 Geometric series, , 41, , 2.3 Properties of series, , 44, , 2.3.1, , Constant multiple, , 2.3.2, , Sum of series, , 44, , 47, , 2.4 Telescoping series, , 50, , 2.5 A test for divergence, 2.6 Harmonic series, , 39, , 51, , 54, , 2.6.1, , The limit of the terms, , 54, , 2.6.2, , Numerical evidence, , 54, , 2.6.3, , An analytic argument, , 2.7 Comparison test, , 55, , 56, , 2.7.1, , Statement of the Comparison Test, , 2.7.2, , Applications of the Comparison Test, , 2.7.3, , Cauchy Condensation Test, , 2.7.4, , Examples of condensation, , 2.7.5, , Convergence of p-series, , 59, 61, 63, , 56, 57

Page 5 :
5, , 3, , Convergence tests, 3.1 Ratio tests, 3.1.1, , Theory, , 3.1.2, , Practice, , 67, , 67, 68, 71, , 3.2 Integral test, , 74, , 3.2.1, , An example, , 74, , 3.2.2, , Harmonic series, , 3.2.3, , Statement of integral test, , 3.2.4, , p-series, , 3.2.5, , Integrating for approximations, , 75, 76, , 76, , 3.3 More comparisons, , 81, , 3.4 The mostly useless root test, , 4, , 85, , Alternating series, 4.1 Absolute convergence, 4.2 Alternating series test, , 5, , 78, , 89, 89, 93, , Another comparison test, , 97, , 5.1 Convergence depends on the tail, 5.2 Limit comparison test, , 98, , 101, , 5.2.1, , Proof of the Limit Comparison Test, , 101, , 5.2.2, , How to apply the Limit Comparison Test, , 102

Page 6 :
6, , 6, , Power series, 6.1 Definitions, , 105, 105, , 6.2 Convergence of power series, , 106, , 6.3 Power series centered elsewhere, 6.4 Calculus with power series, , 7, , Taylor series, 7.1 Finding Taylor series, 7.2 Taylor’s Theorem, , 110, , 114, , 117, 117, 126, , Answers to Exercises, , 141, , Index, , 147

Page 7 :
How to read do mathematics, , Reading mathematics is not the same as reading a novel—it’s more fun, and more, interactive! To read mathematics you need, , (a) a pen,, (b) plenty of blank paper, and, (c) the courage to write down everything—even “obvious” things., As you read a math book, you work along with me, the author, trying to anticipate, my next thoughts, repeating many of the same calculations I did to write this book., You must write down each expression, sketch each graph, and constantly think, about what you are doing. You should work examples. You should fill-in the details, I left out. This is not an easy task; it is hard work, but, work that is, I very much, hope, rewarding in the end., Mathematics is not a passive endeavor. I may call you a “reader” but you are, not reading; you are writing this book for yourself., , —the so-called “author”

Page 9 :
Acknowledgments, , This text is a modification of David Guichard’s open-source calculus text which was, itself a modification of notes written by Neal Koblitz at the University of Washington, and includes exercises and examples from Elementary Calculus: An Approach Using, Infinitesimals by H. Jerome Keisler. I am grateful to David Guichard for choosing, a Creative Commons license. Albert Schueller, Barry Balof, and Mike Wills have, contributed additional material. The stylesheet, based on tufte-latex, was, designed by Bart Snapp., This textbook was specifically used for a Coursera course called “Calculus Two:, Sequences and Series.” Many thanks go to Walter Nugent, Donald Wayne Fincher,, Robert Pohl, chas, Clark Archer, Mikhail, Sarah Smith, Mavaddat Javid, Grigoriy, Mikhalkin, Susan Stewart, Donald Eugene Parker, Francisco Alonso Sarría, Eduard, Pascual Saez, Lam Tin-Long, mrBB, Demetrios Biskinis, Hanna Szabelska, Roland, Thiers, Sandra Peterson, Arthur Dent, Ryan Noble, Elias Sid, Faraz Rashid, and, hrzhu for finding and correcting errors in early editions of this text. Thank you!, , —Jim Fowler

Page 11 :
Introduction, or. . . what is this all about?, , Consider the following sum:, 1, 2, , +, , 1, 4, , +, , 1, 8, , +, , 1, 16, , +···+, , 1, 2i, , +···, , The dots at the end indicate that the sum goes on forever. Does this make sense?, Can we assign a numerical value to an infinite sum? While at first it may seem, difficult or impossible, we have certainly done something similar when we talked, about one quantity getting “closer and closer” to a fixed quantity. Here we could, ask whether, as we add more and more terms, the sum gets closer and closer to, some fixed value. That is, look at, 1, , =, , 2, 3, , =, , 4, 7, , =, , 8, 15, , =, , 16, , 1, 2, 1, 2, 1, 2, 1, 2, , +, +, +, , 1, 4, 1, 4, 1, 4, , +, +, , 1, 8, 1, 8, , +, , 1, 16, , and so on, and ask whether these values have a limit. They do; the limit is 1. In, fact, as we will see,, 1, 2, , +, , 1, 4, , +, , 1, 8, , +, , 1, 16, , +···+, , and then, , lim 1 −, , i →∞, , 1, 2i, , 1, 2i, , =, , 2i − 1, 2i, , = 1 − 0 = 1., , = 1−, , 1, 2i

Page 12 :
12, , sequences and series, , This is less ridiculous than it appears at first. In fact, you might already believe, that, , 3, , 0.33333̄ =, , 3, , +, , 3, , +, , 3, , +, , +··· =, , 1, , ,, , 10, 100, 1000, 10000, 3, which is similar to the sum above, except with powers of ten instead of powers of two., And this sort of thinking is needed to make sense of numbers like π, considering, 3.14159 . . . = 3 +, , 1, , +, , 10, , 4, 100, , +, , 1, 1000, , +, , 5, , +, , 10000, , 9, 100000, , + · · · = π., , Before we investigate infinite sums—usually called series—we will investigate, limits of sequences of numbers. That is, we officially call, ∞, X, 1, i =1, , 2i, , 1, , =, , 2, , 1, , +, , 4, , +, , 1, 8, , +, , 1, 16, , +···+, , a series, while, 1, , ,, , 3, , ,, , 7, , 15, , ,, , 2i, , +···, , 2i − 1, , , ..., 2, 4, 8, 16, 2i, is a sequence. The value of a series is the limit of a particular sequence, that is,, ∞, X, 1, i =1, , 2i, , ,, , 1, , ...,, , = lim, , i →∞, , 2i − 1, 2i, , ., , If this all seems too obvious, let me assure you that there are twists and turns, aplenty. And if this all seems too complicated, let me assure you that we’ll be going, over this in much greater detail in the coming weeks. In either case, I hope that, you’ll join us on our journey.

Page 13 :
1 Sequences, , 1.1, , Notation, , A “sequence” of numbers is just a list of numbers. For example, here is a list of, numbers:, , 1,, , 1,, , 2,, , 3,, , 5,, , 8,, , 13,, , 21,, , ..., , Note that numbers in the list can repeat. And consider those little dots at the, end! The dots “. . . ” signify that the list keeps going, and going, and going—forever., Presumably the sequence continues by following the pattern that the first few “terms”, suggest. But what’s that pattern?, To make this talk of “patterns” less ambiguous, it is useful to think of a sequence, as a function. We have up until now dealt with functions whose domains are the, real numbers, or a subset of the real numbers, like f (x ) = sin(1/x )., A real-valued function with domain the natural numbers N = {1, 2, 3, . . .} is a, sequence., Other functions will also be regarded as sequences: the domain might include, 0 alongside the positive integers, meaning that the domain is the non-negative, integers, Z≥0 = {0, 1, 2, 3, . . .}. The range of the function is still allowed to be the, real numbers; in symbols, the function f : N → R is a sequence., Sequences are written down in a few different, but equivalent, ways; you might, , Maybe you are feeling that this formality is unnecessary, or even ridiculous; why can’t we just list off, a few terms and pick up on the pattern intuitively?, As we’ll see later, that might be very hard—nay,, impossible—to do! There might be very different—, but equally reasonable—patterns that start the same, way., To resolve this ambiguity, it is perhaps not so, ridiculous to introduce the formalism of “functions.”, Functions provide a nice language for associating, numbers (terms) to other numbers (indices).

Page 14 :
14, , sequences and series, , see a sequence written as, a1 ,, , a2 ,, , a3 ,, , ...,, , an, , (an )n ∈N ,, {an }∞, n =1 ,, , ∞, f (n ) n =1 ,, , or, , (f (n ))n ∈N ,, depending on which author you read. Worse, depending on the situation, the, same author (and this author) might use various notations for a sequence! In this, textbook, I will usually write (an ) if I want to speak of the sequence as a whole, (think gestalt) and I will write an if I am speaking of a specific term in the sequence., Let’s summarize the preceding discussion in the following definition., Definition A sequence (an ) is, formally speaking, a real-valued function with, domain, , {n ∈ Z : n ≥ N },, , for some integer N ., , Stated more humbly, a sequence assigns a real number to the integers starting, with an index N ., The “outputs” of a sequence are the terms of the sequence; the “n th term” is, the real number that the sequence associates to the natural number n, and, is usually written an . The n in the phrase “n th term” is called an index; the, plural of index is either indices or indexes, depending on who you ask. The, first index N is called the initial index., , Warning Usually the “domain” of a sequence is N and Z≥0 . But depending, on the context, it may be convenient for a sequence to start somewhere else—, perhaps with some negative number. We shouldn’t let the usual situation of N, or Z≥0 get in the way of making the best choice for the problem at hand., , Recall that the natural numbers N are the counting, numbers 1, 2, 3, 4, . . .. If we want our sequence to, start at zero, we use Z≥0 as the domain instead., The fancy symbols Z≥0 refer to the non-negative, integers, which include zero (since zero is neither, positive nor negative) and also positive integers (since, they certainly aren’t negative)., To confuse matters further, some people—especially, computer scientists—might include zero in the natural numbers N. Mathematics is cultural.

Page 15 :
sequences, , As you can tell, there is a deep tension between precise definition and a vague, flexibility; as instructors, how we navigate that tension will be a big part of whether, we are successful in teaching the course. We need to invoke precision when we’re, tempted to be too vague, and we need to reach for an extra helping of vagueness, when the formalism is getting in the way of our understanding. It can be a tough, balance., , 1.2, 1.2.1, , Defining sequences, Defining sequences by giving a rule, , Just as real-valued functions from Calculus One were usually expressed by a, formula, we will most often encounter sequences that can be expressed by a, formula. In the Introduction to this textbook, we saw the sequence given by the rule, ai = f (i ) = 1 − 1/2i . Other examples are easy to cook up, like, ai =, bn =, , i, i +1, 1, 2n, , ,, , ,, , cn = sin(nπ /6), or, di =, , (i − 1)(i + 2), 2i, , ., , Frequently these formulas will make sense if thought of either as functions with, domain R or N, though occasionally the given formula will make sense only for, integers. We’ll address the idea of a real-valued function “filling in” the gaps between, the terms of a sequence when we look at graphs in Section 1.5., Warning A common misconception is to confuse the sequence with the rule, for generating the sequence. The sequences (an ) and (bn ) given by the rules, an = (−1)n and bn = cos(π n ) are, despite appearances, different rules which, give rise to the same sequence. These are just different names for the same, , 15

Page 16 :
16, , sequences and series, , object., Let’s give a precise definition for “the same” when speaking of sequences., Definition Suppose (an ) and (bn ) are sequences starting at 1. These se-, , Compare this to equality for functions: two functions are the same if they have same domain and, codomain, and they assign the same value to each, point in the domain., , quences are equal if for all natural numbers n, we have an = bn ., More generally, two sequences (an ) and (bn ) are equal if they have the same, initial index N , and for every integer n ≥ N , the n th terms have the same value,, that is,, an = bn, , for all n ≥ N ., , In other words, sequences are the same if they have the same set of valid indexes,, and produce the same real numbers for each of those indexes—regardless of whether, the given “rules” or procedures for computing those sequences resemble each other, in any way., , 1.2.2, , Defining sequences using previous terms, , Another way to define a sequence is recursively, that is, by defining the later outputs, in terms of previous outputs. We start by defining the first few terms of the sequence,, and then describe how later terms are computed in terms of previous terms., Example 1.2.1 Define a sequence recursively by, a1 = 1,, , a2 = 3,, , a3 = 10,, , and the rule that an = an −1 − an −3 . Compute a5 ., Solution First we compute a4 . Substituting 4 for n in the rule an = an −1 −, an −3 , we find, a4 = a4−1 − a4−3 = a3 − a1 ., But we have values for a3 and a1 , namely 10 and 1, respectively. Therefore, a4 = 10 − 1 = 9., Now we are in a position to compute a5 . Substituting 5 for n in the rule, , You might be familiar with recursion from a computer, science course.

Page 17 :
sequences, , 17, , an = an −1 − an −3 , we find, a5 = a5−1 − a5−3 = a4 − a2 ., We just computed a4 = 9; we were given a2 = 3. Therefore a5 = 9 − 3 = 6., , 1.3, , You can imagine some very complicated sequences, defined recursively. Make up your own sequence, and share it with your friends! Use the hashtag, #sequence., , Examples, , Mathematics proceeds, in part, by finding precise statements for everyday concepts., , Tons of entertaining sequences are listed in the The, On-Line Encyclopedia of Integer Sequences., , We have already done this for sequences when we found a precise definition, (“function from N to R”) for the everyday concept of “a list of real numbers.” But, all the formalisms in the world aren’t worth the paper they are printed on if there, aren’t some interesting examples of those precise concepts. Indeed, mathematics, proceeds not only by generalizing and formalizing, but also by focusing on specific,, concrete instances. So let me share some specific examples of sequences., But before I can share these examples, let me address a question: how can I, hand you an example of a sequence? It is not enough just to list off the first few, terms. Let’s see why., Example 1.3.1 Consider the sequence (an ), a1 = 41,, , a2 = 43,, , a3 = 47,, , a4 = 53,, , ..., , What is the next term a5 ? Can you identify the sequence?, Solution In spite of many so-called “intelligence tests” that ask questions just, like this, this question simply doesn’t have an answer. Or worse, it has too, , This particular polynomial n 2 − n + 41 is rather interesting, since it outputs many prime numbers. You, can read more about it at the OEIS., , many answers!, This sequence might be “the prime numbers in order, starting at 41.” If, that’s the case, then the next term is a5 = 59. But maybe this sequence is the, sequence given by the polynomial an = n 2 − n + 41. If that’s the case, then, the next term is a5 = 61. Who is to say which is the “better” answer?, Now let’s consider two popular “families” of sequences., , Recall that a prime number is an integer greater, than one that has no positive divisors besides itself, and one.

Page 18 :
18, , sequences and series, , 1.3.1, , Arithmetic sequences, , The first family1 we consider are the “arithmetic” sequences. Here is a definition., Definition An arithmetic progression (sometimes called an arithmetic se-, , 1, , Mathematically, the word family does not have an, entirely precise definition; a family of things is a, collection or a set of things, but family also has a, connotation of some sort of relatedness., , quence) is a sequence where each term differs from the next by the same, fixed, quantity., , Example 1.3.2 An example of an arithmetic progression is the sequence (an ), which begins, a1 = 10,, , a2 = 14,, , a3 = 18,, , a4 = 22,, , ..., , and which is given by the rule an = 6 + 4 n. Each term differs from the, previous by four., In general, an arithmetic progression in which subsequent terms differ by m can, be written as, an = m (n − 1) + a1 ., Alternatively, we could describe an arithmetic progression recursively, by giving a, starting value a1 , and using the rule that an = an −1 + m., An arithmetic progression can decrease; for instance,, 17,, , 15,, , 13,, , 11,, , 9,, , ..., , is an arithmetic progression., , 1.3.2, , Geometric sequences, , The second family we consider are geometric progressions., , Why are arithmetic progressions called arithmetic?, Note that every term is the arithmetic mean, that, is, the average, of its two neighbors.

Page 19 :
sequences, , 19, , Definition A geometric progression (sometimes called a geometric sequence), is a sequence where the ratio between subsequent terms is the same, fixed, quantity., , Example 1.3.3 An example of a geometric progression is the sequence (an ), starting, a1 = 10,, , a2 = 30,, , and given by the rule an = 10 · 3, , a3 = 90,, , n −1, , a4 = 270,, , ..., , . Each term is three times the preceding, , term., In general, a geometric progression in which the ratio between subsequent terms, is r can be written as, an = a1 · r n −1 ., Alternatively, we could describe a geometric progression recursively, by giving a, starting value a1 , and using the rule that an = r · an −1 ., A geometric progression needn’t be increasing. For instance, in the following, geometric progression, 7, 5, , ,, , 7, 10, , ,, , 7, 20, , ,, , 7, 40, , ,, , 7, 80, , ,, , 7, 160, , ,, , ..., , the ratio between subsequent terms is one half, and each term is smaller than the, previous., , 1.3.3, , Triangular numbers, , The sequence of triangular numbers (Tn ) is a sequence of integers counting the, number of dots in increasingly large “equilateral triangles” built from dots. The term, Tn is the number of dots in a triangle with n dots to a side., There are a couple of ways of making this discussion more precise. Given an, equilateral triangle with n dots to a side, how many more dots do you need to build, the equilateral triangle with n + 1 dots to a side? All you need to do to transform, the smaller triangle to the larger triangle is an additional row of n + 1 dots placed, , Why are geometric progressions called geometric?, Note that every term is the geometric mean of its, two neighbors. The geometric, mean of two numbers, √, a and b is defined to be ab., Of course, that raises another question: why is, the geometric mean called geometric? One geometric, interpretation of the geometric mean of a and b is, this: the geometric mean is the side length of a, square whose area is equal to that of the rectangle, having side lengths a and b.

Page 20 :
20, , sequences and series, , T1 = 1, , T2 = 3, , T3 = 6, , T4 = 10, , T5 = 15, , T6 = 21, Figure 1.1: The first six triangular numbers, , along any side. Therefore,, Tn + 1 = Tn + ( n + 1 ) ., Since T1 = 1, this recursive definition suffices to determine the whole sequence., But there are other ways of computing Tn . Indeed, you may recall the explicit, formula, Tn =, , n · (n + 1), 2, , from Calculus One., , 1.3.4, , Fibonacci numbers, The Fibonacci numbers are interesting enough that, a journal, The Fibonacci Quarterly is published, four times yearly entirely on topics related to the, Fibonacci numbers., , The Fibonacci numbers are defined recursively, starting with, F0 = 0 and F1 = 1, and the rule that Fn = Fn −1 + Fn −2 . We can restate this formula in words, instead, of symbols; stated in words, each term is the sum of the previous two terms. So the, sequence of Fibonacci numbers begins, 0,, , 1,, , 1,, , 2,, , 3,, , 5,, , 8,, , 13,, , 21,, , 34,, , ..., , and continues., This is certainly not the last time we will see the Fibonacci numbers.

Page 21 :
sequences, , 1.3.5, , 21, , Collatz sequence, , Here is a fun sequence with which to amuse your friends—or distract your enemies., Let’s start our sequence with a1 = 6. Subsequent terms are defined using the rule, , , , , , an −1 /2, an = , , , 3 an −1 + 1, , if an −1 is even, and, if an −1 is odd., , Let’s compute a2 . Since a1 is even, we follow the instructions in the first line, to find, that a2 = a1 /2 = 3. To compute a3 , note that a2 is odd so we follow the instruction, in the second line, and a3 = 3 a2 + 1 = 3 · 3 + 1 = 10. Since a3 is even, the first, line applies, and a4 = a3 /2 = 10/2 = 5. But a4 is odd, so the second line applies,, and we find a5 = 3 · 5 + 1 = 16. And a5 is even, so a6 = 16/2 = 8. And a6 is even,, so a7 = 8/4 = 4. And a7 is even, so a8 = 4/2 = 2, and then a9 = 2/2 = 1. Oh,, but a9 is odd, so a10 = 3 · 1 + 1 = 4. And it repeats. Let’s write down the start of, this sequence:, repeats, 6,, , 3,, , 10,, , 5,, , 16,, , 8,, , 4,, , 2,, , 1,, , 4,, , 2,, , 1,, , z }| {, 4, 2, 1,, , 4,, , ..., , What if we had started with a number other than six? What if we set a1 = 25 but, then we used the same rule? In that case, since a1 is odd, we compute a2 by finding, 3 a1 + 1 = 3 · 25 + 1 = 76. Since 76 is even, the next term is half that, meaning, a3 = 38. If we keep this up, we find that our sequence begins, 25,, , 76,, , 38,, , 19,, , 58,, , 13,, , 40,, , 20,, , 10,, , 5,, , 29,, 16,, , 88,, 8,, , 4,, , 44,, 2,, , 22,, 1,, , 11,, , 34,, , 17,, , 52,, , 26,, , ..., , and then it repeats “4, 2, 1, 4, 2, 1, . . . ” just like before., Does this always happen? Is it true that no matter which positive integer you, start with, if you apply the half-if-even, 3x + 1-if-odd rule, you end up getting stuck, in the “4, 2, 1, . . . ” loop? That this is true is the Collatz conjecture; it has been, , If you think you have an argument that answers, the Collatz conjecture, I challenge you to try your, hand at the 5x + 1 conjecture, that is, use the rule, , , , an −1 /2, an = , 5 a, , n −1 + 1, , if an −1 is even, and, , if an −1 is odd., , verified for all starting values below 5 × 260 . Nobody has found a value which doesn’t, return to one, but for all anybody knows there might well be a very large initial value, which doesn’t return to one; nobody knows either way. It is an unsolved problem2, in mathematics., , 2, , This is not the last unsolved problem we will encounter in this course. There are many things which, humans do not understand.

Page 22 :
22, , sequences and series, , 1.4, , Where is a sequence headed? Take a limit!, , We’ve seen a lot of sequences, and already there are a few things we might notice., For instance, the arithmetic progression, 1,, , 8,, , 15,, , 22,, , 29,, , 36,, , 43,, , 50,, , 57,, , 64,, , 71,, , 78,, , 85,, , 92,, , ..., , just keeps getting bigger and bigger. No matter how large a number you think of, if, I add enough 7’s to 1, eventually I will surpass the giant number you thought of., On the other hand, the terms in a geometric progression where each term is half, the previous term, namely, 1, 2, , ,, , 1, 4, , ,, , 1, 8, , ,, , 1, 16, , ,, , 1, 32, , ,, , 1, 64, , ,, , 1, 128, , ,, , 1, 256, , ,, , 1, 512, , ,, , 1, 1024, , ,, , ...,, , are getting closer and closer to zero. No matter how close you stand near but not at, zero, eventually this geometric sequence gets even closer than you are to zero., These two sequences have very different stories. One shoots off to infinity; the, other zooms in towards zero. Mathematics is not just about numbers; mathematics, provides tools for talking about the qualitative features of the numbers we deal, with. What about the two sequences we just considered? They are qualitatively very, different. The first “goes to” infinity; the second “goes to” zero., In short, given a sequence, it is helpful to be able to say something qualitative, about it; we may want to address the question such as “what happens after a while?”, Formally, when faced with a sequence, we are interested in the limit, , lim f (i ) = lim ai ., , i →∞, , i →∞, , In Calculus One, we studied a similar question about, , lim f (x ), , x →∞, , when x is a variable taking on real values; now, in Calculus Two, we simply want to, restrict the “input” values to be integers. No significant difference is required in the, definition of limit, except that we specify, perhaps implicitly, that the variable is an, integer., , If you were with us in Calculus One, you are perhaps, already guessing that by “goes to,” I actually mean, “has limit.”

Page 23 :
sequences, , 23, , Definition Suppose that (an ) is a sequence. To say that lim an = L is to say, n →∞, , that, for every ε > 0,, there is an N > 0,, so that whenever n > N ,, we have |an − L | < ε., , If lim an = L we say that the sequence converges. If there is no finite value L, n →∞, , so that lim an = L, then we say that the limit does not exist, or equivalently, n →∞, , that the sequence diverges., , Warning In the case that lim an = ∞, we say that (an ) diverges, or perhaps, n →∞, , more precisely, we say (an ) diverges to infinity. The only time we say that a, sequence converges is when the limit exists and is equal to a finite value., , One way to compute the limit of a sequence is to compute the limit of a function., , Theorem 1.4.1 Let f (x ) be a real-valued function. If an = f (n ) defines, a sequence (an ) and if lim f (x ) = L in the sense of Calculus One, then, x →∞, , lim an = L as well., , n →∞, , Example 1.4.2 Since lim (1/x ) = 0, it is clear that also lim (1/n ) = 0; in, x →∞, , n →∞, , other words, the sequence of numbers, 1, 1, , ,, , 1, 2, , ,, , 1, 3, , ,, , 1, 4, , ,, , 1, 5, , ,, , 1, 6, , ,, , ..., , get closer and closer to 0, or more precisely, as close as you want to get to zero,, after a while, all the terms in the sequence are that close., More precisely, no matter what ε > 0 we pick, we can find an N big enough, so that, whenever n > N , we have that 1/n is within ε of the claimed limit,, , The definition of limit is being written as if it were, poetry, what with line breaks and all. Like the best, of poems, it deserves to be memorized, performed,, internalized. Humanity struggled for millenia to find, the wisdom contained therein.

Page 24 :
24, , sequences and series, , zero. This can be made concrete: let’s suppose we set ε = 0.17. What is a, suitable choice for N in response? If we choose N = 5, then whenever n > 5, we have 0 < 1/n < 0.17., But it is important to note that the converse3 of this theorem is not true. To, show the converse is not true, it is enough to provide a single example where it fails., Here’s the counterexample4 ., Example 1.4.3 Consider the sequence (an ) given by the rule an = f (n ) =, , sin(nπ ). This is the sequence, sin(0π ),, , sin(1π ),, , sin(2π ),, , sin(3π ),, , 3, , The converse of a statement is what you get when, you swap the assumption and the conclusion; the, converse of “if it is raining, then it is cloudy” is the, statement “if it is cloudy, then it is raining.” Which, of those statements is true?, 4, An instance of (a potential) general rule being broken is called a counterexample. This is a popular, term among mathematicians and philosophers., , ...,, , which is just the sequence 0, 0, 0, 0, . . . since sin(nπ ) = 0 whenever n is an, integer. Since the sequence is just the constant sequence, we have, , lim f (n ) = lim 0 = 0., , n →∞, , n →∞, , But lim f (x ), when x is real, does not exist: as x gets bigger and bigger,, x →∞, , the values sin(xπ ) do not get closer and closer to a single value, but instead, y, , oscillate between −1 and 1., 2, , Here’s some general advice. If you want to know lim an , you might first think, , f (x ), , n →∞, , of a function f (x ) where an = f (n ), and then attempt to compute lim f (x ). If the, x →∞, , limit of the function exists, then it is equal to the limit of the sequence. But, if, , an, 1, , for some reason lim f (x ) does not exist, it may nevertheless still be the case that, x →∞, , lim f (n ) exists—you’ll just have to figure out another way to compute it., , n →∞, , x and n, 5, , 1.5, , 10, , 15, , Graphs, , It is occasionally useful to think of the graph of a sequence. Since the function is, , −1, , defined only for integer values, the graph is just a sequence of dots. In Figure 1.2 we, see the graph of a sequence and the graph of a corresponding real-valued function., There are lots of real-valued functions which “fill in” the missing values of a, sequence., , Figure 1.2: Plots of f (x ) = cos(π x ) + (4/5)x and, the sequence an = (−1)n + (4/5)n ., , 20

Page 25 :
sequences, , 25, , Example 1.5.1 Here’s a particularly tricky example of “filling in” the missing, values of a sequence. Consider the sequence, 1,, , 2,, , 6,, , 24,, , 120,, , 720,, , 5040,, , 25, , 40320,, , 362880,, , y, , ...,, 20, , where the n, , th, , term is the product of the first n integers. In other words an = n !,, , where the exclamation mark denotes the factorial function. Explicitly describe, , 15, , a function f of a real variable x, so that an = f (n ) for natural numbers n., 10, , Solution There are lots of solutions. Here is a solution:, , an, , f (x ) = bx c!., , 5, , f (x ), , In that definition, bx c denotes the “greatest integer less than or equal to x” and, is called the floor function. This is shown in Figure 1.3., , x and n, , On the other hand, there are much trickier things that you could try to do., If you define the Gamma function, , Γ (z ) =, , 1, , 2, , 3, , 4, , Figure 1.3: A plot of f (x ) = bx c! and an = n !. Recall, that, by convention, 0! = 1., , ∞, , Z, , t z −1 e −t dt ., , 0, , then it is perhaps very surprising to find out that g (x ) = Γ(x + 1) is a function, so that g (n ) = n ! for natural numbers n. A graph is shown in Figure 1.4., Unlike f , which fails to be continuous, the function g is continuous., It is hard to define the “greatest integer” function,, because they are all pretty great., , 1.6, , New sequences from old, , 25, , y, , Given a sequence, one way to build a new sequence is to start with the old sequence,, but then throw away a whole bunch of terms. For instance, if we started with the, , 20, , sequence of perfect squares, 15, , 1,, , 4,, , 9,, , 16,, , 25,, , 36,, , 49,, , 64,, , 81,, , ..., 10, , we could throw away all the odd-indexed terms, and be left with, 4,, , 16,, , 36,, , 64,, , 100,, , 144,, , 196,, , 256,, , 324,, , 400,, , 484,, , ..., , an, 5, , f (x ), , x and n, 1, , Z2, ∞, , Figure 1.4: Plots of f (x ) =, n !., , 0, , 3, t z e −t dt . and an =, , 4

Page 26 :
26, , sequences and series, , We say that this latter sequence is a subsequence of the original sequence. Here is, a precise definition., Definition Suppose (an ) is a sequence with initial index N , and suppose we, have a sequence of integers (ni ) so that, N ≤ n1 < n2 < n3 < n4 < n5 < · · ·, Then the sequence (bi ) given by bi = ani is said to be a subsequence of the, sequence an ., , Limits are telling the story of “what happens” to a sequence. If the terms of a, sequence can be made as close as desired to a limiting value L, then the subsequence, must share that same fate., Theorem 1.6.1 If (bi ) is a subsequence of the convergent sequence (an ), then, , lim bi = lim an ., , i →∞, , n →∞, , Of course, just because a subsequence converges does not mean that the larger, sequence converges, too., , We’ll see this again in more detail when we get to, , Example 1.7.6, but we’ll discuss it briefly now., Example 1.6.2 Find a convergent subsequence of the sequence (an ) given by, the rule an = (−1)n ., Solution Note that the sequence (an ) does not converge. But by considering, the sequence of indexes ni = 2 · i, we can build a subsequence, bi = ani = a2i = (−1)2i = 1,, which is a constant sequence, so it converges to 1.

Page 27 :
sequences, , There are other subsequences of an = (−1)n which converge but do not converge, to one. For instance, the subsequence of odd indexed terms is the constant sequence, cn = −1, which converges to −1. For that matter, the fact that there are convergent, subsequences with distinct limits perhaps explains why the original sequence (an ), does not converge. Let’s formalize this., Corollary 1.6.3, Suppose (bi ) and (ci ) are convergent subsequences of the sequence (an ),, but, , lim bi , lim ci ., , i →∞, , i →∞, , Then the sequence (an ) does not converge., , Proof Suppose, on the contrary, the sequence (an ) did converge. Then by, Theorem 1.6.1, the subsequence (bi ) would converge, too, and, , lim bi = lim an ., , i →∞, , n →∞, , Again by Theorem 1.6.1, the subsequence (ci ) would converge, too, and, , lim ci = lim an ., , i →∞, , n →∞, , But then lim bi = lim ci , which is exactly what we are supposing doesn’t, i →∞, , i →∞, , happen! To avoid this contradiction, it must be that our original assumption that, , (an ) converged was incorrect; in short, the sequence (an ) does not converge., , 1.7, , Helpful theorems about limits, , Not surprisingly, the properties of limits of real functions translate into properties, of sequences quite easily., , 27

Page 28 :
28, , sequences and series, , Theorem 1.7.1 Suppose that lim an = L and lim bn = M and k is some, n →∞, , constant. Then, , n →∞, , lim kan = k lim an = kL ,, , n →∞, , n →∞, , lim (an + bn ) = lim an + lim bn = L + M ,, , n →∞, , n →∞, , n →∞, , lim (an − bn ) = lim an − lim bn = L − M ,, , n →∞, , n →∞, , n →∞, , lim (an bn ) = lim an · lim bn = LM , and, , n →∞, , n →∞, , n →∞, , limn →∞ an, L, lim, =, = , provided M , 0., n →∞ bn, limn →∞ bn, M, an, , 1.7.1, , Squeeze Theorem, , Likewise, there is an analogue of the squeeze theorem for functions., Theorem 1.7.2 Suppose there is some N so that for all n > N , it is the case, that an ≤ bn ≤ cn . If, , lim an = lim cn = L, , n →∞, , n →∞, , , then lim bn = L., n →∞, , And a final useful fact:, Theorem 1.7.3, , lim |an | = 0 if and only if lim an = 0., , n →∞, , n →∞, , This says simply that the size of an gets close to zero if and only if an gets close, to zero., , Sometimes people write “iff” as shorthand for “if and, only if.”

Page 29 :
sequences, , 1.7.2, , 29, , Examples, , Armed with these helpful theorems, we are now in a position to work a number of, examples., Example 1.7.4 Determine whether the sequence (an ) given by the rule an =, n, converges or diverges. If it converges, compute the limit., n+1, Solution Consider the real-valued function, f (x ) =, , x, x +1, , ., , Since an = f (n ), it will be enough to find lim f (x ) in order to find lim an . We, x →∞, , compute, as in Calculus One, that, , n →∞, , (x + 1) − 1, x →∞ x + 1, x →∞, x +1, x + 1, , 1, = lim, −, x →∞ x + 1, x +1, , , 1, = lim 1 −, x →∞, x +1, lim, , x, , = lim, , = lim 1 − lim, x →∞, , x →∞, , = 1 − lim, , x →∞, , 1, x +1, , 1, x +1, , = 1 − 0 = 1., , We therefore conclude that lim an = 1., n →∞, , Example 1.7.5 Determine whether the sequence (an ) given by an =, converges or diverges. If it converges, compute the limit., Solution By l’Hôpital’s rule, we compute, , lim, , x →∞, , log x, x, , = lim, , x →∞, , 1/x, 1, , = 0., , log n, n, , And this is reasonable: by choosing n to be a large, n, enough integer, I can make, as close to 1 as I, n+1, 10000000000, would like. Just imagine how close, 10000000001, is to one.

Page 30 :
30, , sequences and series, , Therefore,, , lim, , log n, , n →∞, , n, , = 0., , Example 1.7.6 Determine whether the sequence (an ) given by the rule an =, (−1)n converges or diverges. If it converges, compute the limit., , I’m not too fond of l’Hôpital’s rule, so I would have, been happier if I had given a solution that didn’t, involve it; you could avoid mentioning l’Hôpital’s, rule in Example 1.7.5 if you used, say, the squeeze, √, theorem and the fact that log n ≤ n., , Solution Your first inclination might be to consider the “function” f (x ) =, , (−1)x , but you’ll run into trouble when trying to tell me the value of f (1/2)., How does the sequence an = (−1)n begin? It starts, −1,, , 1,, , −1,, , −1,, , 1,, , −1,, , 1,, , 1,, , −1,, , 1,, , ...,, , so the sequence isn’t getting close to any number in particular., Intuitively, the above argument is probably pretty convincing. But if you, want an airtight argument, you can reason like this: suppose—though we’ll, soon see that this is a ridiculous assumption—that the sequence an = (−1)n, did converge to L. Then any subsequence would also converge to L, by, Theorem 1.6.1 which stated that the limit of a subsequence is the same as the, limit of the original sequence. If I throw away every other term of the sequence, , (an ), I am left with the constant sequence, −1,, , −1,, , −1,, , −1,, , −1,, , −1,, , −1,, , ...,, , which converges to −1, and so L must be −1., On the other hand, if I throw away all the terms with odd indices and keep, only those terms with even indices, I am left with the constant subsequence, 1,, , 1,, , 1,, , 1,, , 1,, , 1,, , 1,, , ...,, , so L must be 1. Since L can’t be both −1 and 1, it couldn’t have been the case, that lim an = L for a real number L. In other words, the limit does not exist., n →∞, , I imagine that the “airtight argument” in the solution, to Example 1.7.6 is difficult to understand. Please, don’t worry if you find the argument confusing now—, we’ll have more opportunities for doing these sorts, of proofs by contradiction in the future.

Page 31 :
sequences, , 31, , Example 1.7.7 Determine whether the sequence an = (−1/2)n converges or, diverges. If it converges, compute the limit., Solution Let’s use the Squeeze Theorem. Consider the sequences bn =, , −(1/2)n and cn = (1/2)n . Then bn ≤ an ≤ cn . And lim cn = 0 because, n →∞, , lim (1/2)x = 0. Since bn = −cn , we have lim bn = − lim cn = −0 = 0., x →∞, n →∞, n →∞, Since bn and cn converge to zero, the squeeze theorem tells us that lim an = 0, , In this problem, you must be very careful to recognize, the difference between (−1/2)n and −(1/2)n . The, former flip-flops between being positive and being, negative, while the latter is always negative., , n →∞, , as well., , If you don’t want to mention the Squeeze Theorem, you could instead, apply Theorem 1.7.3. In that case, we would again consider the sequence, cn = |an | and observe that lim cn = 0. But then Theorem 1.7.3 steps in,, n →∞, , and tells us that lim an = 0 as well. Of course, a convincing argument for, n →∞, , why Theorem 1.7.3 works at all goes via the squeeze theorem, so this second, method is not so different from the first., , Example 1.7.8 Determine whether an =, , sin n, √ converges or diverges. If it, n, , converges, compute the limit., Solution Since −1 ≤ sin n ≤ 1, we have, , −1, sin n, 1, √ ≤ √ ≤ √ ,, n, , n, , n, , 1, and can therefore apply the Squeeze Theorem. Since lim √ = 0, we get, x →∞, x, , −1, 1, lim √ = lim √ = 0,, , n →∞, , n, , n →∞, , n, , and so by squeezing, we conclude lim an = 0., n →∞, , Example 1.7.9 A particularly common and useful sequence is the geometric, progression an = r n for a fixed real number r. For which values of r does this, sequence converge?, , You might be wondering why I love the Squeeze, Theorem so much; one reason is that the Squeeze, Theorem gets you into the idea of “comparing” one, sequence to another, and this “comparison” idea will, be big when we get to convergence tests in Chapter 3.

Page 32 :
32, , sequences and series, , Solution It very much does depend on r., If r = 1, then an = (1)n is the constant sequence, 1,, , 1,, , 1,, , 1,, , 1,, , 1,, , 1,, , 1,, , ...,, , so the sequence converges to one. A similarly boring fate befalls the case r = 0,, in which case an = (0)n converges to zero., If r = −1, we are reprising the sequence which starred in Example 1.7.6; as, we saw, that sequence diverges., If either r > 1 or r < −1, then the terms an = r n can be made as large as, one likes by choosing n large enough (and even), so the sequence diverges., If 0 < r < 1, then the sequence converges to 0., If −1 < r < 0 then |r n | = |r |n and 0 < |r | < 1, so the sequence {|r |n }∞, n =0, , converges to 0, so also {r n }∞, n =0 converges to 0., , That last example of a geometric progression is involved enough that it deserves, to be summarized as a theorem., Theorem 1.7.10 The sequence an = r n converges when −1 < r ≤ 1, and, diverges otherwise. In symbols,, , , , , 0, if -1<r<1,, , , , , , n, lim r = , 1, if r=1, and, , , n →∞, , , , does not exist if r ≤ −1 or r > 1., , 1.8, , Qualitative features of sequences, , Sometimes we will not be able to determine the limit of a sequence, but we still, would like to know whether or not it converges to some unspoken number. In many, cases, we can determine whether a limit exists, without needing to—or without even, being able to—compute that limit., , Your first exposure to mathematics might have been, about constructions; you might have been asked to, compute a numeric answer or to propose a solution, to a problem. But much of mathematics is concerned, with showing existence, even if the thing that is, being shown to exist cannot be exhibited itself.

Page 33 :
sequences, , 1.8.1, , 33, , Monotonicity, , And sometimes we don’t even care about limits, but we’d simply like some terminology with which to describe features we might notice about sequences. Here is some, of that terminology., , For instance, how much money I have on day n is, a sequence; I probably hope that sequence is an, increasing sequence., , Definition A sequence is called increasing (or sometimes strictly increasing), if an < an +1 for all n. It is called non-decreasing if an ≤ an +1 for all n., Similarly a sequence is decreasing (or, by some people, strictly decreasing), if an > an +1 for all n and non-increasing if an ≥ an +1 for all n., , To make matters worse, the people who insist on saying “strictly increasing” may—, much to everybody’s confusion—insist on calling a non-decreasing sequence “increasing.” I’m not going to play their game; I’ll be careful to say “non-decreasing”, when I mean a sequence which is getting larger or staying the same., To make matters better, lots of facts are true for sequences which are either, increasing or decreasing; to talk about this situation without constantly saying, “either increasing or decreasing,” we can make up a single word to cover both cases., , Definition If a sequence is increasing, non-decreasing, decreasing, or nonincreasing, it is said to be monotonic., , Let’s see some examples of sequences which are monotonic., Example 1.8.1 The sequence an =, 1, 2, , ,, , 3, 4, , ,, , 2n − 1, 2n, 7, 8, , ,, , which starts, , 15, 16, , ,, , ...,

Page 34 :
34, , sequences and series, , is increasing. On the other hand, the sequence bn =, 2, 1, , ,, , 3, 2, , ,, , 4, 3, , ,, , 5, 4, , ,, , n+1, n, , , which starts, , ...,, , is decreasing., , 1.8.2, , Boundedness, , Sometimes we can’t say exactly which number a sequence approaches, but we can, at least say that the sequence doesn’t get too big or too small., Definition A sequence (an ) is bounded above if there is some number M so, that for all n, we have an ≤ M. Likewise, a sequence (an ) is bounded below if, there is some number M so that for every n, we have an ≥ M., If a sequence is both bounded above and bounded below, the sequence is, said to be bounded., If a sequence {an }∞, n =0 is increasing or non-decreasing it is bounded below (by a0 ),, and if it is decreasing or non-increasing it is bounded above (by a0 )., Finally, with all this new terminology we can state the most important theorem, of Chapter 1., Theorem 1.8.2 If the sequence an is bounded and monotonic, then lim an, exists., , n →∞, , In short, bounded monotonic sequences converge—though we can’t necessarily, describe the number to which they converge., We will not prove this theorem in the textbook.5 Nevertheless, it is not hard to, believe: suppose that a sequence is increasing and bounded, so each term is larger, than the one before, yet never larger than some fixed value M. The terms must then, get closer and closer to some value between a0 and M. It certainly need not be M,, since M may be a “too generous” upper bound; the limit will be the smallest number, , 5, , Proving this theorem is, honestly, the purview of, a course in analysis, the theoretical underpinnings, of calculus. That’s not to say it couldn’t be done in, this course, but I intend this to be a “first glance” at, sequences—so much will be left unsaid.

Page 35 :
sequences, , that is above6 all of the terms an . Let’s try an example!, Example 1.8.3 All of the terms (2i − 1)/2i are less than 2, and the sequence, , 35, , 6, This concept of the “smallest number above all the, terms” is an incredibly important one; it is the idea of, a least upper bound that underlies the real numbers., , is increasing. As we have seen, the limit of the sequence is 1—1 is the smallest, number that is bigger than all the terms in the sequence. Similarly, all of, the terms (n + 1)/n are bigger than 1/2, and the limit is 1—1 is the largest, number that is smaller than the terms of the sequence., We don’t actually need to know that a sequence is monotonic to apply this, theorem—it is enough to know that the sequence is “eventually” monotonic,7 that is,, that at some point it becomes increasing or decreasing. For example, the sequence, 10, 9, 8, 15, 3, 21, 4, 3/4, 7/8, 15/16, 31/32, . . . is not increasing, because among, , 7, , After all, the limit only depends on what is happening after some large index, so throwing away the, beginning of a sequence won’t affect its convergence, or its limit., , the first few terms it is not. But starting with the term 3/4 it is increasing, so if, the pattern continues and the sequence is bounded, the theorem tells us that the, “tail” 3/4, 7/8, 15/16, 31/32, . . . converges. Since convergence depends only on, what happens as n gets large, adding a few terms at the beginning can’t turn a, convergent sequence into a divergent one., Example 1.8.4 Show that the sequence (an ) given by an = n 1/n converges., Solution We might first show that this sequence is decreasing, that is, we, show that for all n,, n 1/n > ( n + 1 ) 1/ (n +1) ., , You may be worried about my saying that log 3 > 1., If log were the common (base 10) logarithm, this, would be wrong, but as far as I’m concerned, there, is only one log, the natural log. Since 3 > e, we may, conclude that log 3 > 1.

Page 36 :
36, , sequences and series, , But this isn’t true! Take a look, a1 = 1,, a2 =, a3 =, a4 =, a5 =, a6 =, a7 =, a8 =, a9 =, , √, 2 ≈ 1.4142,, , √, 3, √, 4, √, 5, √, 6, √, 7, √, 8, √, 9, , 3 ≈ 1.4422,, 4 ≈ 1.4142,, 5 ≈ 1.3797,, 6 ≈ 1.3480,, 7 ≈ 1.3205,, 8 ≈ 1.2968, and, 9 ≈ 1.2765., , But it does seem that this sequence perhaps is decreasing after the first few, terms. Can we justify this?, Yes! Consider the real function f (x ) = x 1/x when x ≥ 1. We compute the, derivative—perhaps via “logarithmic differentiation”—to find, f 0 (x ) =, , x 1/x (1 − log x ), x2, , ., , Note that when x ≥ 3, the derivative f 0 (x ) is negative. Since the function f is, decreasing, we can conclude that the sequence is decreasing—well, at least for, n ≥ 3., Since all terms of the sequence are positive, the sequence is decreasing and, bounded when n ≥ 3, and so the sequence converges., , Example 1.8.5 Show that the sequence an =, , n!, nn, , converges., , As it happens, you could compute the limit in Example 1.8.4, but our given solution shows that it, converges even without knowing the limit!

Page 37 :
sequences, , Solution Let’s get an idea of what is going on by computing the first few terms., a1 = 1,, a5 =, a7 =, , 24, , a2 =, , 1, 2, , ,, , a3 =, , ≈ 0.038400,, , 625, 720, , 117649, , 2, 9, , ≈ 0.22222,, , a6 =, , 5, 324, , 3, 32, , ≈ 0.093750,, , ≈ 0.015432,, , a8 =, , ≈ 0.0061199,, , a4 =, , 315, 131072, , ≈ 0.0024033., , The sequence appears to be decreasing. To formally show this, we would need, to show an +1 < an , but we will instead show that, an +1, an, , < 1,, , which amounts to the same thing. It is helpful trick here to think of the ratio, between subsequent terms, since the factorials end up canceling nicely. In, particular,, an +1, an, , (n + 1) ! n n, (n + 1 )n +1 n !, (n + 1) !, nn, =, n!, (n + 1 )n +1, , n, n+1, n, =, n+1 n+1, , =, , =, , , , n, n+1, , n, < 1., , Note that the sequence is bounded below, since every term is positive., Because the sequence is decreasing and bounded below, it converges., Indeed, Exercise 2 asks you to compute the limit., These sorts of arguments involving the ratio of subsequent terms will come up, again in a big way in Section 3.1. Stay tuned!, , 37

Page 38 :
38, , sequences and series, , Exercises for Section 1.8, (1) Compute lim x 1/x ., , à, , x →∞, , (2) Use the squeeze theorem to show that lim, , n →∞, , √, , (3) Determine whether { n + 47 −, the limit., , n!, nn, , = 0., , à, , √, , n }∞, n =0 converges or diverges. If it converges, compute, , à, , (, (4) Determine whether, limit., , à, (, , (, , à, , converges or diverges. If it converges, compute the, n =0, , )∞, , n + 47, , √, , à, , (6) Determine whether, , )∞, , (n + 1)2, , (5) Determine whether, limit., , n2 + 1, , n 2 + 3n, , 2n, n!, , converges or diverges. If it converges, compute the, n =1, , )∞, converges or diverges. If it converges, compute the limit., n =0

Page 39 :
2 Series, , We’ve only just scratched the surface of sequences, and already we’ve arrived in, Chapter 2. Series will be the main focus of our attention for the rest of the course. If, that’s the case, then why did we bother with sequences? Because a series is what, you get when you add up the terms of a sequence, in order. So we needed to, talk about sequences to provide the language with which to discuss series., Suppose (an ) is a sequence; then the associated series1, ∞, X, , ak = a1 + a2 + a3 + a4 + a5 + · · ·, , k =1, , I might be thinking that I feel just fine, but I have woken a terrible beast! What, does that innocuous looking “· · · ” mean? What does it mean to add up infinitely, many numbers? It’s not as if I’ll ever be done with all the adding, so how can I ever, attach a “value” to a series? How can I do infinitely many things, yet live to share, the answer with you?, I can’t., But I can do a large, but finite, number of things, and then see if I’m getting close, to anything in particular. In other words, I can take a limit., , 2.1, , Definition of convergence, , But a limit. . . of what? From a sequence, we can consider the associated series, and, associated to that series is a yet another sequence—the sequence of partial sums., Here’s a formal definition which unwinds this tangled web., , The Σ symbol may look like an E, but it is the Greek, letter sigma, and it makes an S sound—just like the, first letter of the word “series.” If you see “GRΣΣK,”, say “grssk.”, , 1

Page 40 :
40, , sequences and series, , Definition Suppose (an ) is a sequence with associated series, , ∞, X, , ak . The, , k =1, , sequence of partial sums associated to these objects is the sequence, sn =, , n, X, , ak ., , k =1, , Working this out, we have, s1 = a1 ,, s2 = a1 + a2 ,, s3 = a1 + a2 + a3 ,, s4 = a1 + a2 + a3 + a4 ,, s5 = a1 + a2 + a3 + a4 + a5 , and so on., , Instead of adding up the infinite sequence an , which we can’t do, we will instead, look at the sequence of partial sums, and ask whether that sequence of partial sums, converges. And if it converges to L, then we’ll call L the value of the series., , Definition Consider the series, of partial sums sn =, write, , n, X, , ∞, X, , ak . This series converges if the sequence, , k =1, , ak converges. More precisely, if lim sn = L, we then, n →∞, , k =1, , ∞, X, , ak = L, , k =1, , and say, “the series, , ∞, X, , ak converges to L.”, , k =1, , If the sequence of partial sums diverges, we say that the series diverges., , This might seem overly complicated, but it solves a, serious problem: we no longer are confronted with, the supertask of adding up infintely many numbers, but living to tell the tale. To take the limit of the, sequence of partial sums is to add up lots—but not, all!—of the terms in the original sequence to see if, we’re staying close to a particular number—the limit., That particular limiting value is then, by definition,, declared to be the result of adding up all the terms, in the original sequence in order., That the order matters will be a major theme in, Chapter 4.

Page 41 :
series, , Remember, infinity is not a number. So if it happens that lim, , n →∞, , we might write, , ∞, X, , n, X, , ak = ∞, then, , k =1, , ak = ∞ but nevertheless we still say that the series diverges., , k =1, , 2.2, , 41, , Sometimes, to emphasize that the series involves, adding up infinitely many terms, we will say “infinite series” instead of just “series.”, , Geometric series, , Armed with the official definition of convergence in general, we focus in on the specific, example: a sequence of the form an = a0 r n is called a geometric progression as, we learned back in Subsection 1.3.2. What happens when we add up the terms of a, geometric progression?, Definition (Geometric Series) A series of the form, ∞, X, , a0 r k, , k =0, , is called a geometric series., , We can’t simply “add” up the infinitely many terms in the geometric series. What, we can do, instead, is add up the a first handful of terms. Pick a big value for n,, and instead compute, , sn =, , n, X, k =0, , a0 r k = a0 + a0 r + a0 r 2 + a0 r 3 + · · · + a0 r n ., , A geometric series was used by Archimedes—who, lived more than two thousand years ago!—to compute, the area between a parabola and a straight line., Humans had the first inklings of calculus a very long, time ago.

Page 42 :
42, , sequences and series, , This is the n th partial sum. Concretely,, s0 = a 0 ,, s1 = a0 + a0 r ,, s2 = a0 + a0 r + a0 r 2 ,, s3 = a0 + a0 r + a0 r 2 + a0 r 3, , .., ., sn = a0 + a0 r + a0 r 2 + a0 r 3 + · · · + a0 r n ., In our quest to assign a value to the infinite series, , ∞, X, , a0 r k , we instead2 consider, , k =0, , lim sn = lim, , n →∞, , n, X, , n →∞, , 2, Replacing the actual “∞” by a limit (that is, a “potential” infinity) shouldn’t seem all that surprising;, we encountered the same trick in Calculus One., , a0 r k ., , k =0, , We can perform some algebraic manipulations on the partial sum. The manipulation, begins with our multiplying sn by (1 − r ) to cause some convenient cancellation,, specifically,, sn (1 − r ) = a0 (1 + r + r 2 + r 3 + · · · + r n ) (1 − r ), , = a0 (1 + r + r 2 + r 3 + · · · + r n ) 1 − a0 (1 + r + r 2 + r 3 + · · · + r n −1 + r n ) r, = a0 (1 + r + r 2 + r 3 + · · · + r n ) − a0 (r + r 2 + r 3 + · · · + r n + r n +1 ), = a0 (1 + r + r 2 + r 3 + · · · + r n − r − r 2 − r 3 − · · · − r n − r n +1 ), = a0 (1 − r n +1 )., , Dividing both sides3 by (1 − r ) shows, sn = a 0 ·, Therefore,, , ∞, X, k =0, , 1 − r n +1, 1−r, , k, , 3, Here, we tacitly assume r , 1. But can you see, what happens to the geometric series when r = 1, without going through this argument?, , ., , a0 r = lim sn = lim a0 ·, n →∞, , n →∞, , 1 − r n +1, 1−r, , !, .

Page 43 :
series, , The limit depends very much on what r is., Suppose r ≥ 1 or r ≤ −1. In those cases, lim r n +1 does not exist, and likewise, n →∞, , lim sn does not exist. So the series diverges if r ≥ 1 or if r ≤ −1. One quicker way, of saying this is that the series diverges when |r | ≥ 1., On the other hand, suppose |r | < 1. Then lim r n +1 = 0, and so, n →∞, , n →∞, , lim sn = lim a0, , n →∞, , 1 − r n +1, , =, , 1−r, , n →∞, , a0, 1−r, , ., , Thus, when |r | < 1 the geometric series converges to a0 /(1 − r ). This is important, enough that we’ll summarize it as a theorem., Theorem 2.2.1 Suppose a0 , 0. Then for a real number r such that |r | < 1,, the geometric series, ∞, X, , a0 r k, , k =0, , converges to, , a0, , ., 1−r, For a real number r where |r | ≥ 1, the aforementioned geometric series, , diverges., , Example 2.2.2 When, for example, a0 = 1 and r = 1/2, this means, ∞  k, X, 1, k =0, , 2, , =, , 1, 1−, , 1, 2, , = 2,, , which makes sense. Consider the partial sum, sn = 1 +, , 1, 2, , +, , 1, 4, , +, , 1, 8, , +, , 1, 16, , +···+, , 1, 2n, , ., , This partial sum gets as close to two as you’d like—as long as you are willing, to choose n large enough. And it doesn’t take long to get close to two! For, example, even just n = 6, we get, s6 = 1 +, , 1, 2, , +, , 1, 4, , +, , 1, 8, , +, , 1, 16, , +, , 1, 32, , +, , 1, 64, , =, , 127, 64, , 43

Page 44 :
44, , sequences and series, , 1, 524288, , 1, 131072, 1, , 1, , 262144, , 32768, 1, , 1, , 65536, , 8192, 1, , 1, , 16384, , 2048, , which is close to two., , 1, , 1, , 4096, , 512, 1, , 1, , 1024, , 128, 1, , 1, , 256, , 32, 1, , 1, , Example 2.2.3 Consider the series, , 64, , 8, , ∞, X, 1, , ., k, , 1, , 1, , 2, k =1, , 16, , This does not quite fit into the preceding framework, because this series starts, with k = 1 instead of k = 0., Nevertheless, we can work out what happens. The series, , ∞, X, 1, k =1, , ∞, X, 1, , 2k, , 2, , is just, , 1, 4, , except the former is missing an initial k = 0 term, which, 2k, k =0, is 1/20 = 1. So each partial sum of the former series is one less than the, like series, , corresponding partial sum for the latter series, so the limit is also one less, than the value of the geometric series. In symbols,, ∞, X, 1, ,  ∞, ,  X 1 , ,  − 1 = 1, = , n, n, 2, 2, 1−, n =1, n =0, , 1, 2, , − 1 = 1., , Properties of series, , Theorem 1.7.1 presented many properties of sequences. Since the value of a series, is the limit of the sequence of partial sums, properties of limits can be reformulated, into properties of series., , 2.3.1, , Constant multiple, , 1, , ∞, X, 1, n =1, , 2n, , = 1. Begin, , × 1 rectangle, and build each subsequent, 2, rectangle by cloning and halving the previous rectangle; all these rectangles fit together to fill up a unit, square., , with a, , If you don’t find this argument convincing, look at Figure 2.1, which displays, ∞, X, 1, a visual argument that the series, converges to one., k, 2, k =1, , 2.3, , Figure 2.1: Visual evidence that

Page 45 :
series, , Theorem 2.3.1 Suppose, Then, , ∞, X, , ∞, X, , ak is a convergent series, and c is a constant., , k =0, , c ak converges, and, , k =0, , ∞, X, , c ak = c, , k =0, , ∞, X, , ak ., , k =0, , Proof If you know just enough algebra to be dangerous, you may remember, that for any real numbers a, b, and c,, c (a + b ) = c a + c b ., This is the distributive law for real numbers. By using the distributive law, more than once,, c (a0 + a1 + a2 ) = c a0 + c a1 + c a2 ,, or more generally, for some finite n,, c (a0 + a1 + a2 + · · · + an ) = c a0 + c a1 + c a2 + · · · + c an ., So what’s the big deal with proving Theorem ? Can’t we just scream “distributive, law!” and be done with it? After all, Theorem amounts to, c a0 + c a1 + c a2 + · · · + c an + · · · = c (a0 + a1 + a2 + · · · + an + · · · )., But hold your horses: you can only apply the distributive law finitely many, times! The distributive law, without some input from calculus, will not succeed, in justifying Theorem ., Let’s see how to handle this formally. By hypothesis,, , ∞, X, k =0, , series, meaning its associated sequence of partial sums,, sn =, , n, X, k =0, , ak ,, , ak is a convergent, , 45

Page 46 :
46, , sequences and series, , converges. In other words, lim sn exists. Then, by Theorem ,, n →∞, , lim (c sn ) = c lim sn ., , n →∞, , n →∞, , But c sn is the sequence of partial sums for the series, , ∞, X, , c ak , because, , k =0, , c sn = c a0 + c a1 + c a2 + · · · + c an ., Consequently,, ∞, X, , c ak = lim (c sn ) = c lim sn ,, n →∞, , k =0, , n →∞, , which is what we wanted to prove., That theorem addresses the case of multiplying a convergent series by a constant, c; what about divergence?, , Example 2.3.2 Suppose that, , ∞, X, , ak diverges; does, , k =0, , Solution If c = 0, then, , ∞, X, k =0, , c ak =, , ∞, X, , c ak also diverge?, , k =0, , ∞, X, , 0 which does converge, to zero., , k =0, , On the other hand, provided c , 0, then, yes,, , ∞, X, , c ak also diverges. How, , k =0, , do we know?, We are working under the hypothesis that, the contrary, that, , ∞, X, , ∞, X, , ak diverges. Suppose now, to, , k =0, , c ak did converge; applying Theorem 2.3.1 (albeit with, , k =0, , ak replaced by c ak and c replaced by (1/c )), the series, ∞  , X, 1, k =0, , c, , c ak

Page 47 :
series, , 47, , converges, but that is ridiculous, since, ∞  , X, 1, k =0, , c, , c ak =, , ∞, X, , ak, , k =0, , and the latter, under our hypothesis, diverged. A series cannot both converge, and diverge, so our assumption (that, mistaken—it must be that, , ∞, X, , ∞, X, , c ak did converge) must have been, , k =0, , c ak diverges., , k =0, , But what about the case where c = 0? In that case, the series converged! Did we, make a mistake? In the argument for divergence, we multiplied by 1/c, which is, something we are not permitted to do when c = 0., , We can connect this discussion back to the discussion of geometric series in Section 2.2. If you, ∞, X, 1, believe that, rk =, , then by Theorem 2.3.3,, 1−r, , In light of this example, we have actually proved something stronger., , Theorem 2.3.3 Consider the series, stant. Then, , ∞, X, , ak and, , k =0, , ∞, X, , ∞, X, , k =0, , ak , and suppose c is a nonzero con-, , k =0, , you believe, rem 2.2.1., , ∞, X, , k =0, , c rk =, , c, 1−r, , , which is part of Theo-, , c ak share a common fate: either both series, , k =0, , converge, or both series diverge., Moreover, when, , ∞, X, , ak converges,, , k =0, , ∞, X, k =0, , 2.3.2, , c ak = c ·, , ∞, X, , ak ., , k =0, , Sum of series, , Suppose, , ∞, X, k =0, , ak and, , ∞, X, k =0, , bk are convergent series. What can be said of, , ∞, X, , (ak + bk )?, , k =0, , Addition is associative4 and commutative5 , so for any real numbers a, b, c,, , 4, , To say that addition is “associative” is, intuitively, to, say that how the expression is parenthesized doesn’t, matter; formally, “associativity” means that for any, a, b, and c, we have a + (b + c ) = (a + b ) + c., 5, , To say that addition is “commutative” is, intuitively,, to say that the order in which the adding is done, doesn’t matter; formally, “commutativity” means that, for a and b, we have a + b = b + a.

Page 48 :
48, , sequences and series, , and d,, , (a + b ) + (c + d ) = (a + c ) + (b + d )., More generally, for real numbers a0 , a1 , a2 , . . . , an and real numbers b0 , b1 , b2 ,, . . . , bn ,, , (a0 + a1 + a2 + · · · + an ) + (b0 + b1 + b2 + · · · + bn ) = (a0 + b0 ) + (a1 + b1 ) + (a2 + b2 ) + · · · + (an + bn )., But this finite statement can be beefed up into a statement about series. What we, want to prove is, ∞, X, , ak +, , k =0, , ∞, X, , bk =, , k =0, , ∞, X, , (ak + bk ) ., , k =0, , From above, we already know, n, X, , ak +, , k =0, , n, X, , bk =, , k =0, , n, X, , (ak + bk ) ., , k =0, , Take the limit of both sides., ,  n, , n, n, X, X,  X, , , lim , ak +, bk  = lim, (ak + bk ) ., , n →∞, , k =0, , n →∞, , k =0, , k =0, , But the limit of a sum is the sum of the limits6 , so, , 6, , ,  , , n, n, n, X, X, X, ,  , ,  lim, , , ak  +  lim, bk  = lim, (ak + bk ) ., n →∞, n →∞, n →∞, k =0, , k =0, , k =0, , Those three limits of partial sums can each be replaced by the series, which shows, ∞, X, k =0, , ak +, , ∞, X, , bk =, , k =0, , This can be summarized in a theorem., , ∞, X, k =0, , (ak + bk ) ., , I like to call this a chiastic rule, since it has the, rhetorical pattern of a chiasmus.

Page 49 :
series, , Theorem 2.3.4, ∞, X, , Suppose, , ∞, X, , ak and, , k =0, , ∞, X, , bk are convergent series., , Then, , k =0, , (ak + bk ) is convergent, and, , k =0, , ∞, X, k =0, ,  ∞,   ∞, ,  X   X, , , , , (ak + bk ) = , ak  + , bk  ., k =0, , k =0, , That covers sums of convergent series., , Example 2.3.5 Now suppose that, , ∞, X, , ak and, , k =0, , diverge?, , ∞, X, , bk diverge; does, , k =0, , ∞, X, k =0, , diverge. But, , (ak + bk ) =, , k =0, , ∞, X, , ((1) + (−1)) =, , k =0, , (ak + bk ), , k =0, , Solution Not necessarily. Let ak = 1 and bk = −1, so, , ∞, X, , ∞, X, , ∞, X, , ak and, , ∞, X, , bk, , k =0, , 0 = 0., , k =0, , This is not to say that the term-by-term sum of divergent series necessarily, converges, either. It is entirely possible that, , k =0, , (ak + bk ) =, , (ak + bk ) will also diverge. For, , k =0, , istance, if ak = bk = 1, then, ∞, X, , ∞, X, , ∞, X, , (1 + 1) =, , k =0, , ∞, X, , 2, , k =0, , also diverges. So the term-by-term sum of divergent series might converge or, might diverge, depending on the situation., , 49

Page 50 :
50, , sequences and series, , 2.4, , Telescoping series, , For most of this course, we will be happy if we can show that a series converges or, that a series diverges; we will not, usually, be too concerned with finding the value, of a series. Why not? Usually it is just too hard to determine the value; we would if, we could, but since it is often too hard, we don’t bother., Nevertheless, there is one family of series for which we can calculate the value, with relative ease: the telescoping series. Here is a first example which suggests, what we mean by “telescoping.”, n, X, , Example 2.4.1 Compute, , k =1, , 1, k · (k + 1), , ., You might know about a method called partial frac1, 1, tions to rewrite, as a combination of, k, k · (k + 1), 1, ., and, k+1, , Solution Note that, 1, k, , −, , 1, , =, , k+1, , =, , k+1, k · (k + 1), k +1−k, k · (k + 1), , −, , =, , k, k · (k + 1), 1, k · (k + 1), , Consequently,, n, X, k =1, , 1, k · (k + 1), , =, =, =, , 1, 1, 1, 1, 1, 1, , −, , 1, 1+1, , , , + −, −, , , , +, , 1, , 1+1, 1, , 1, , +, , 2, , −, , , 1, 2, , 1, , , , 2+1, , , + −, , +, , 1, 2+1, , 1, 3, , +, , −, , 1, 3+1, , , 1, 3, , , , +···+, , +···+ −, , 1, n, , −, , 1, , , , n+1, , 1, , (n − 1) + 1, , +, , 1, , !, , n, , n+1, , since most of these terms end up canceling., In general, we say that a series telescopes if, after some simplification, there, is a formula for the sequence of partial sums with a fixed number of terms. The, name suggests the way the cancellation happens: just as the nesting rings in an, expandable spyglass fit together, so too do the neighboring terms in a telescoping, series fit together and collapse., , −, , 1, n+1

Page 51 :
series, , Armed with a formula for the sequence of partial sums, we can attack the, corresponding infinite series., , Example 2.4.2 Compute, , ∞, X, , 1, , k · (k + 1), k =1, , ., , Solution We just computed that, n, X, , 1, , k · (k + 1), k =1, , = 1−, , 1, n+1, , ., , Rewriting the infinite series as the limit of the sequence of partial sums yields, ∞, X, , 1, , k · (k + 1), k =1, , = lim, , n, X, , 1, , k · (k + 1), k =1, , , 1, = lim 1 −, n →∞, n+1, 1, = lim 1 − lim, n →∞, n →∞ n + 1, n →∞, , = 1 − 0 = 1., , So the value of this series is 1., , 2.5, , A test for divergence, , Usually, the sequence of partial sums sn = a0 + a1 + · · · + an is harder to understand and analyze than the sequence of terms ak . It would be helpful if we could say, something about the complicated sequence sn by studying the easier-to-understand, sequence ak ., Specifically, if the sequence sn converges, what can be said about the sequence, ak ? If adding up more and more terms from the sequence ak gets closer and closer, to some number, then the size of the terms of ak had better be getting very small., Let’s make this precise., , 51

Page 52 :
52, , sequences and series, , Theorem 2.5.1 If, , ∞, X, , ak converges then lim an = 0., n →∞, , k =0, , Proof Intuitively, this should seem reasonable: after all, if the terms in the, sequence an were getting very large (that is, not converging to zero), then adding, up those very large numbers would prevent the series, , ∞, X, , ak from converging., , k =0, , We can put this intuitive thinking on a firm foundation. Say, , ∞, X, , ak con-, , k =0, , verges to L, meaning lim sn = L. But then also lim sn −1 = L, because that, n →∞, n →∞, sequence amounts to saying the same thing, but with the terms renumbered. By, Theorem ??,, , lim (sn − sn −1 ) = lim sn − lim sn −1 = L − L = 0., , n →∞, , n →∞, , n →∞, , Replacing sn with a0 + · · · + an , we get, sn − sn −1 = (a0 + a1 + a2 + · · · + an ) − (a0 + a1 + a2 + · · · + an −1 ) = an ,, and therefore, lim an = 0., n →∞, , The contrapositive of Theorem 2.5.1 can be used as a divergence test., , Theorem 2.5.2 Consider the series, , ∞, X, , ak . If the limit lim an does not exist, , k =0, , n →∞, , or has a value other than zero, then the series diverges., , We’ll usually call this theorem the “n th term test.”, Warning The converse of Theorem 2.5.1 is not true: even if lim an = 0, the, series could diverge., , n →∞

Page 53 :
series, , This is a very common mistake: you might be tempted to show that a series, converges by showing lim an = 0, but that doesn’t work. The n th term test, n →∞, , either says “diverges!” or says nothing at all. It is not possible to show that, anything converges by using the n th term test., For an example, see Section 2.6., One analogy that can be helpful is think about weather: whenever it is raining, it, is cloudy. Yet it is possible for there to be clouds, even on a rainless day. 7, Likewise, whenever the series, , ∞, X, , ak converges (“it is raining”), the sequence, , k =0, , ak converges to zero (“it is cloudy”). If it isn’t cloudy, then we can be sure it isn’t, raining—and this is the statement of Theorem 2.5.2. Let’s use this “divergence test”, to show that a particular series diverges., , Example 2.5.3 Show that, , ∞, X, n =1, , n, n+1, , diverges., , Solution We apply the n th term test: all we need to do is to compute the limit, , lim, , n, , n →∞, , n+1, , = 1 , 0., , Since the limit exists but is not zero (i.e., “it is not cloudy”), the series must, diverge (“it can’t be raining.”)., Looking at the first few terms perhaps makes it clear that the series has no, chance of converging:, 1, , +, , 2, , +, , 3, , +, , 4, , +···, , 2, 3, 4, 5, will just get larger and larger; indeed, after a bit longer the series starts to, look very much like · · · + 1 + 1 + 1 + 1 + · · · , and if we add up many numbers, which are very close to one, then we can make the sum as large as we desire., , 7, , We first introduced this idea on Page 24., , 53

Page 54 :
54, , sequences and series, , 2.6, , Harmonic series, , The series, , ∞, X, 1, n =1, , n, , has a special name., , Definition The series, ∞, X, 1, n =1, , n, , =, , 1, 1, , +, , 1, 2, , +, , 1, 3, , +, , 1, 4, , +···, , is called the harmonic series., , The main question for this section—indeed, the question which should always be, our first question anytime we see an unknown series—is the following question:, Does the harmonic series converge. . . or diverge?, , How can we begin to explore this question?, , 2.6.1, , The limit of the terms, , In general, the easiest way to prove that a series diverges is to apply the n th term, test from Theorem 2.5.2. What does the n th term test tell us for the harmonic series?, We calculate, , lim, , n →∞, , 1, n, , When confronted with the question of whether a, series diverges or converges, the first thing to check, is the limit of the terms—if that limit doesn’t exist,, or does exist but equals a number other than zero,, then the series diverges., , = 0,, , so the n th term test is silent; since the limit of the terms of the series exists and is, equal to zero, the n th term test does not tell us any information., The harmonic series passed the first gauntlet—but that does not mean the, harmonic series will survive the whole game. All we know is that the harmonic does, doesn’t diverge for the most obvious reason, but whether it diverges or converges is, yet to be determined., , 2.6.2, , Numerical evidence, , If you have the fortitude8 to add up the first hundred terms, you will find that, , 8, , lacking fortitude, software will suffice

Page 55 :
series, , 100, X, 1, n =1, , n, , =, =, , 1, 1, , +, , 1, , 1, , +···+, , 2, , 100, , 14466636279520351160221518043104131447711, 2788815009188499086581352357412492142272, , ≈ 5.19., , If we add up the first thousand terms, we will find that, 1000, X, n =1, , 1, n, , 1, , =, , 1, , +, , 1, 2, , +···+, , 1, 1000, , ≈ 7.49., , If we add up the first ten thousand terms, we will find that, 10000, X, n =1, , 1, n, , =, , 1, 1, , +, , 1, 2, , +···+, , 1, 10000, , ≈ 9.79, , The partial sums are getting bigger, but not very quickly. Maybe the series converges., Maybe it converges to. . . about ten?, , 2.6.3, , An analytic argument, , That numeric evidence might have made us think otherwise, but the harmonic, series diverges., But in fact the partial sums do get arbitrarily large; they just get big very, very, slowly. Consider the following:, 1+, 1+, , 1, 2, , +, , 1, 3, , +, , 1, 4, , 1+, , +, 1, 2, , 1, 5, , +, , +, 1, 3, , 1, 2, 1, 6, , +, +, , 1, 3, 1, 7, , +, +, , +···+, , 1, 4, 1, 8, 1, , 16, , > 1+, > 1+, > 1+, , 1, 2, 1, 2, 1, 2, , +, +, +, , 1, 4, 1, 4, 1, 4, , +, +, +, , 1, 4, 1, 4, 1, 4, , = 1+, +, +, , 1, 8, 1, 8, , +, , 1, 2, 1, 8, , +, +, , 1, 2, 1, , +···+, , 8, 1, 8, , +, +, , 1, 8, , = 1+, 1, , 16, , 1, 2, , +, , +···+, , 1, , 2, 1, , 16, , and so on. By swallowing up more and more terms we can always manage to add, at least another 1/2 to the sum, and by adding enough of these we can make the, partial sums as big as we like. In fact, it’s not hard to see from this pattern that, 1+, , 1, 2, , +, , 1, 3, , +···+, , 1, 2n, , > 1+, , n, 2, , ,, , +, , 1, 2, , = 1+, , 1, 2, , +, , 1, 2, , +, , 1, 2, , +, , 1, 2, , 55

Page 56 :
56, , sequences and series, , so to make sure the sum is over 100, for example, we’d add up terms until we get to, around 1/2198 , that is, about 4 · 1059 terms., , 2.7, , Comparison test, , A bounded, monotonic sequence necessarily converges (Theorem 1.8.2). How does, this fact about sequences relate to series? When is the sequence of partial sums, monotonic? If the terms of a series are non-negative, then the associated sequence, of partial sums is non-decreasing., , Corollary 2.7.1 Consider the series, , ∞, X, , ak . Assume the terms ak are non-, , k =0, , negative. If the sequence of partial sums sn = a0 + · · · + an is bounded, then, the series converges., , So we can show that a series of positive terms converges, provided we can bound, the sequence of partial sums., , 2.7.1, , Statement of the Comparison Test, , But how can we manage to do that? One way to ensure that the sequence of partial, sums is bounded is by comparing the series to another series. Consider two series, ∞, X, , ak and, , k =0, , ∞, X, , bk ., , k =0, , Suppose, for all k, that bk ≥ ak ≥ 0. Then, a0 + a1 + · · · + an ≤ b0 + b1 + · · · + bn ., Suppose that, , ∞, X, , bk converges to L. Then, , k =0, , a0 + a1 + · · · + an ≤ b0 + b1 + · · · + bn ≤ L ,

Page 57 :
series, , so the sequence of partial sums sn = a0 + a1 + · · · + an is bounded. But we just won, the game: each term ak is nonnegative, so the sequence of partial sums sn =, , n, X, , ak, , k =0, , is increasing. Theorem 1.8.2 guarantees that the sequence (sn ) converges., Let’s summarize what just happened: if a series with positive terms is, termwise,, less than a convergent series, it converges. We have just proved half of the following, theorem., Theorem 2.7.2 Suppose that an and bn are non-negative for all n and that,, for some N , whenever n ≥ N , we have an ≤ bn ., If, , ∞, X, , bn converges, so does, , n =0, , If, , ∞, X, , ∞, X, , an ., , n =0, , an diverges, so does, , n =0, , ∞, X, , bn ., , n =0, , This is usually called the Comparison Test; we might summarize it like this:, , • A non-negative series, overestimated by a convergent series, converges., • A non-negative series, underestimated by a divergent series, diverges., , Warning Being less than a divergent series does not help: the comparison test, is silent in that case., Similarly, being larger than a convergent series does not help. The Comparison Test only says something when a series (with non-negative terms!) is less, than a convergent series, or greater than a divergent series., , 2.7.2, , Applications of the Comparison Test, , Like the n th term test (Theorem 2.5.2), we can use the Comparison Test (Theorem 2.7.2) to show that a series diverges., , 57

Page 58 :
58, , sequences and series, , Example 2.7.3 Does the series, , ∞, X, log n, , converge?, , n, , n =2, , Solution Our first inclination might be to apply the n th term test, but in this, case,, , lim, , log n, , n →∞, , n, , = 0,, , so the n th term test is silent in this case. As far as we know at this point, the, series may diverge or converge., Instead, we’ll try the Comparison Test. Set an =, that whenever n ≥ 3, we have, , 1, n, , and bn =, , log n, n, , . Note, , 0 ≤ an ≤ bn ,, but the series, , ∞, X, 1, , diverges, and so by the Comparison Test, the given series, n, (which is even bigger!) must likewise diverge., n =3, , Recall that the n th term test cannot be used to prove that a series converges; if, the n th term test does not answer “diverges!” then the test is silent. In wonderful, contrast, the Comparison Test can be used to show that a series converges., , Example 2.7.4 Does the series, , ∞, X, sin2 n, n =1, , Solution Yes. Set, an =, , sin2 n, 2n, , Note that 0 ≤ an ≤ bn . But the series, , 2n, , converge?, , and bn =, ∞, X, , 1, 2n, , bn converges, since it is a geometric, , n =1, , series with common ratio 1/2, as in Example 2.2.2. Therefore, the series, ∞, X, , n =1, , an converges by the comparison test.

Page 59 :
series, , 2.7.3, , 59, , Cauchy Condensation Test, , Remember in Section 2.6 when we considered the harmonic series? We showed, ∞, X, 1, that it diverged by comparing it with the divergent series, , but we couldn’t, 2, n =1, make that comparison right away—first we had to group together the terms in a, somewhat complicated seeming way., We can generalize that “grouping together” trick; this is called the Cauchy, Condensation Test., Theorem 2.7.5 Suppose (an ) is a non-increasing sequence of positive numbers. The series, , ∞, X, , an converges if and only if the series, , n =1, , The series, series, , ∞, X, , ∞, X, , ∞, X, , (2n a2n ) converges., , n =0, , (2n a2n ) is often called the condensed series associated to the, , n =0, , an ., , n =1, , Proof Let’s suppose that, ∞, X, , ∞, X, , (2n a2n ) converges; the goal then is to show that, , n =0, , an also converges., , n =1, , Since the sequence (an ) is decreasing, we have that, a2 + a3 ≤ a2 + a2, a4 + a5 + a6 + a7 ≤ a4 + a4 + a4 + a4, a8 + · · · + a15 ≤ 8 a8, , .., ., a2n + · · · + a2n +1 −1 ≤ 2n a2n ., , If you have already seen some convergence tests, before—perhaps you have already been through Calculus Two!—you might be wondering why “condensation” is making an appearance. It is perhaps less, popular than other tests, but I like it. Pedagogically,, it is nice to see that the “trick” in the harmonic series, can be generalized and applied to lots of other series., In particular, condensation permits the study of pseries without going through the usual route of the, Integral Test.

Page 61 :
series, , What we’ve shown is a bit stronger than simply that the original series and the, condensed series share the same fate—converging or diverging together. In fact, we, have an estimate on the value of the original series, in terms of the value of the, condensed series. We have shown that, ∞, X, , k, X, , an ≤, , n =1, , 2.7.4, , (2n a2n ) ≤ 2 ·, , n =0, , ∞, X, , an ., , n =1, , Examples of condensation, ∞, X, 1, , Example 2.7.6 Does the series, , n =1, , converge?, , n2, , This is not a geometric series: we already know that, ∞, X, 1, , ∞, X, 1, n =1, , 2n, , converges, but this, , ., n2, Solution To get some intuition for what is going on, let’s do some numerical, , is asking about something very different, namely, , n =1, , calculations., 10, X, 1, n =1, , n2, , =, =, , 1, 12, , +, , 1, 22, , +, , 1968329, 1270080, , 1, 32, , +···+, , 1, 102, , ≈ 1.5498,, , or going out a bit farther,, 100, X, 1, n =1, , n2, , 1000, X, , =, 1, , n2, n =1, , 1, 12, , =, , +···+, 1, 12, , 1, 1002, , +···+, , ≈ 1.6350 and, 1, , 10002, , ≈ 1.6439., , 61

Page 62 :
62, , sequences and series, , From this numerical evidence, it certainly looks like this series converges. And, indeed, it does—quite surprisingly,, ∞, X, 1, n =1, , n2, , =, , π2, 6, , ., , This is the so-called Basel problem., We do not yet have the tools necessarily to show that the value of the series, is π 2 /6, but do we have the tools needed to show that the series converges., ∞, X, 2n, converges. But, By condensation, it suffices to show that, n 2, n =1 (2 ), ∞, X, , 2n, , n =1, , (2n )2, , =, , ∞, X, , 1, , (2n ), n =1, , = 1,, , and since the “condensed” series converges, so too must the original series, converge., , Example 2.7.7 Does, , ∞, X, | sin n |, n =2, , n2, , converge?, , Solution We can’t apply Cauchy condensation here, because the terms of this, series are not decreasing. But we can apply the Comparison Test. Moments, ∞, X, 1, ago, we saw that, converges, and, 2, n, n =1, , | sin n |, n2, , ≤, , 1, n2, , ,, , because | sin n | ≤ 1. The partial sums are non-decreasing and bounded above, by, , ∞, X, n =1, , 1/n 2 = L, so the series converges.

Page 63 :
series, , 63, , Convergence of p-series, , 2.7.5, , ∞, X, 1, , . Such a series is called a p-series. Does a p-series, np, converge? Diverge? It depends on p., Let us consider the series, , n =1, , Example 2.7.8 Let p ≤ 1. Does the series, , ∞, X, 1, n =1, , np, , converge?, , diverge in Section 2.6., ∞, X, 1, , diverges whenever p ≤ 1. We will, , np, , n =1, , show this by comparing to a harmonic series. Since p ≤ 1, then n p ≤ n, and so, 1, np, But the harmonic series, ∞, X, 1, n =1, , np, , ∞, X, 1, n =1, , n, , ≥, , 1, n, , ., , diverges, and so by comparison, the series, , diverges., , Example 2.7.9 Let p > 1. Does the series, , ∞, X, 1, n =1, , np, , converge?, , Solution It converges. For this, we use Cauchy condensation: consider the, “condensed” series, , ∞, X, , 2n ·, , n =1, , 1, , (2n )p, , ., , But this series simplifies to, ∞, X, n =1, , 2n ·, , 1, , (2n )p, , =, , ζ (p ) =, , ∞, X, 1, n =1, , Solution When p = 1, this series is the harmonic series we already proved to, , But more generally, the series, , If we think of this as a function of p, then we have, the Riemann zeta function, that is,, , ∞, X, , 1, , n =1, , (2p−1 )n, , ,, , np, , ., , The Riemann zeta function is quite important: it, plays a key role in number theory via the Riemann, hypothesis and also has applications in physics., Something that connects the physical world to number theory must be pretty incredible.

Page 64 :
64, , sequences and series, , which converges.

Page 67 :
3 Convergence tests, , It is generally quite difficult—indeed, often impossible—to determine the value of a, series exactly. Even if we can’t compute the value of a series, in many cases it is, possible to determine whether or not the series converges. We will spend most of, our time on this problem., , 3.1, , Ratio tests, , Does the series, , ∞, X, n5, , converge? It is possible, but a bit unpleasant, to approach, 5n, this with the comparison test., n =0, , Example 3.1.1 The series, , ∞, X, n5, n =0, , 5n, , converges., , Solution As long as n ≥ 23, we have 2n ≥ n 5 . Therefore, as long as n ≥ 23,, we have, , n5, 5n, , But the geometric series, , ≤, , 2n, 5n, , =, , ∞  n, X, 2, ,  2 n, 5, , converges, since the ratio between subse5, quent terms is less than one. So by comparison, the smaller series, n =23, , ∞, X, n5, n =23, , 5n, , Mathematics is more than just about getting answers; a goal of mathematics is not only to find, truth, but to package the resulting “truth” in a format that permits another human being to understand the reasons for its being true. Arguments like, this—which are perfectly convincing but seem entirely unmotivated—are, arguably, missing the point., What has been gained if we find something is true but, the reason for its being true remains inscrutable?, This idea—that the first handful of terms do not, affect convergence at all—will be discussed formally, in Section 5.1.

Page 68 :
68, , sequences and series, , must also converge. The first handful of terms when n < 23 doesn’t affect, convergence at all, so we are justified to conclude that the original series, ∞, X, n5, converges., 5n, n =0, , That worked—but it invoked an unmotivated fact: how do I know that 2n ≥ n 5, whenever n ≥ 23? Invoking that fact seems a bit random—yes, yes, a proof, but, perhaps not a proof that conveys exactly what is going on. It is a valid argument,, but missing some motivation., , 3.1.1, , Theory, , Instead, consider what happens as we move from one term to the next term in this, series, that is, consider two neighboring terms, , ···+, , n5, 5n, , +, , (n + 1)5, 5n +1, , +··· ., , The denominator goes up by a factor of 5, 5n +1 = 5 · 5n , but the numerator goes up, by much less: (n + 1)5 = n 5 + 5n 4 + 10n 3 + 10n 2 + 5n + 1, which is much less, than 5n 5 when n is large, because 5n 4 is much less than n 5 . (This sort of thinking, is why it was worth comparing n 5 to an exponential 2n for n large.), So we might guess that in the long run—when n is very large—it begins to look as, if each term is about 1/5 of the previous term. We have seen series that behave, like this, namely the geometric series, ∞, X, 1, n =0, , 5n, , =, , 5, 4, , ., , We are beginning to see why it made sense to compare the given series to a geometric, series as in the initially very unmotivated argument above., But we can do better! Instead of an ad hoc argument which compared n 5 to 2n ,, we can try to make rigorous the idea that each term is “eventually” about a fifth as, big as the previous term. The key is to notice that, , lim, , n →∞, , an +1, an, , = lim, , n →∞, , (n + 1)5 5n, 5n +1, , n5, , = lim, , n →∞, , (n + 1)5 1, n5, , 5, , = 1·, , 1, 5, , =, , 1, 5, , .

Page 69 :
convergence tests, , 69, , This is a more formal version of what we noticed about the ratio of subsequent, terms: in the long run, each term is one fifth of the previous term. Pick some, 1, 1, number just slightly bigger than ; let’s call that number + ε., 5, 5, Because, an +1, 1, , lim, , =, , an, , n →∞, , 5, , ,, , then by choosing n big enough, say n ≥ N for some N , we can guarantee that, is as close as we’d like to, , 1, , , say within ε of, , 5, N so that whenever n ≥ N , we have, , an +1, an, , <, , 1, 5, 1, 5, , 1, 5, , +ε, , , + ε an ., , We can say the same thing when n is replaced by N + 1, meaning, aN +2 <, , 1, 5, , , + ε aN + 1 ,, , which together with the previous statement when n is N implies, aN +2 <, , 1, 5, , 2, + ε aN ., , And we can repeat this again! Since, aN +3 <, , 1, 5, , we also learn that, aN +3 <, , 1, 5, , , + ε aN + 2 ,, 3, + ε aN ,, , or in general,, , 1, , +ε, , k, , aN ., 5, This is setting up a comparison. The geometric series, aN +k <, , ∞ , X, 1, k =0, , 5, , +ε, , k, , an, , . More succinctly, there must be some, , Multiplying both sides by an , we find whenever n ≥ N that, an +1 <, , an +1, ,  ∞ , k ,  X 1, , , aN = , + ε  aN, k =0, , 5, , The symbol ε is the Greek letter epsilon, and, conventionally, denotes a small but positive number.

Page 70 :
70, , sequences and series, , converges as long as ε had been chosen so small that, converges, then the series, ∞, X, , 1, 5, , + ε < 1. But if that series, , aN +k = aN + aN +1 + aN +2 + · · ·, , k =0, , also converges, because each term of that series is smaller than the corresponding, term of the convergent geometric series. But if that series converges, then, ∞, X, , ak = a0 + a1 + a2 + · · ·, , k =0, , converges since we’ve just added the number a0 + a1 + · · · + aN −1 to the convergent, series, , ∞, X, , aN +k ., , k =0, , Under what circumstances could we do this? What was crucial was that the limit, of an +1 /an , say L, was less than 1 so that we could pick a value ε so that L + ε < 1, and then compare to the convergent series, ∞, X, , (L + ε )k · aN ., , k =0, , That’s really all that is required to make the argument work. We also made use of, the fact that the terms of the series were positive. Let’s summarize the situation, when this works., , Theorem 3.1.2 (The Ratio Test) Consider the series, , ∞, X, , an where each term, , n =0, , an is positive. Suppose that, , lim, , n →∞, , If L < 1 the series, , ∞, X, n =0, , test is inconclusive., , an +1, an, , = L., , an converges. If L > 1 the series diverges. If L = 1 this, , In general, we can consider instead the absolute, values of the terms, and end up testing for absolute, convergence. We’ll discuss this topic in Section 4.1.

Page 71 :
convergence tests, , 71, , Proof Example 3.1.1 essentially proves the first part of this, if we simply replace, 1/5 by L and 1/2 by r. So when L < 1, the series converges., Let’s consider the other situation. Suppose that L > 1, and pick r so that, 1 < r < L. Then for n ≥ N , for some N ,, , |an +1 |, > r and |an +1 | > r |an |., |an |, This implies that |aN +k | > r k |aN |, but since r > 1 this means that lim |aN +k | , 0,, k →∞, , which means also that lim an , 0. By the divergence test, the series diverges., n →∞, , Finally, when L = 1, the test truly is inconclusive—not because we aren’t, clever enough, but because there are both convergent and divergent series with, L = 1. For example, the p-series, ∞, X, , ∞, X, , 1/n 2 converges, and the harmonic series, , n =1, , 1/n diverges—but in both cases, L = 1., , n =1, , 3.1.2, , Practice, , The ratio test is particularly useful for series involving the factorial function., Example 3.1.3 Does the series, ∞, X, , “It works in practice—but does it work in theory?”, The ratio test is awfully useful in practice, but I, think the theory behind it—that we’re taking a limit, of the ratio between neighboring terms in order to, compare with a geometric series—is quite lovely., , 5n /n !, , n =0, , converge or diverge?, Remember our convention that 0! = 1. Despite, appearances, we are not dividing by zero.

Page 72 :
72, , sequences and series, , Solution Let’s name the terms an = 5n /n ! so we are considering, , ∞, X, n =0, , apply the ratio test, we consider, , lim, , n →∞, , an +1, an, , 5n +1, , n!, (n + 1)! 5n, n!, 5n +1, = lim, n →∞ 5n (n + 1) !, , = lim, , n →∞, , = lim 5, n →∞, , 1, , (n + 1), , Since 0 < 1, the series converges by the ratio test., , = 0., , an . To

Page 73 :
convergence tests, , Exercises for Section 3.1, , (1) Compute lim |an +1 /an | for the series, n →∞, , ∞, X, 1, n =1, , (2) Compute lim |an +1 /an | for the series, n →∞, , n2, , ∞, X, 1, n =1, , n, , ., , ., , à, à, , Determine whether each of the following series converges or diverges., (3), , ∞, X, , (−1)n, , n =0, , (4), , ∞, X, n!, n =1, , (5), , ∞, X, n5, n =1, , (6), , nn, , nn, , nn, , 5n, , à, , à, à, , ∞, X, (n ! )2, n =1, , 3n, , à, , 73

Page 74 :
74, , sequences and series, , 3.2, , Integral test, , If all of the terms an in a series are non-negative, then clearly the sequence of partial, sums sn is non-decreasing. This means that if we can show that the sequence, of partial sums is bounded, the series must converge. We know that if the series, converges, the terms an approach zero, but this does not mean that an ≥ an +1 for, every n. Many useful and interesting series do have this property, however, and, they are among the easiest to understand. Let’s look at an example., , 3.2.1, , y, 1, , An example, , a1, , 0.8, , Example 3.2.1 Show that, , ∞, X, 1, n =1, , n2, , converges., , 0.6, f (x ) = 1/x 2, , Solution We have actually already seen this very series—it appeared in Exam-, , 0.4, , ple 2.7.6, where we attacked this series via condensation. Here, we consider a, different approach, namely integration., , a2, , 0.2, , 2, , 2, , The terms 1/n are positive and decreasing, and since lim 1/x = 0, the, , a4, , a3, , x →∞, , a5, , 2, , terms 1/n approach zero. We seek an upper bound for all the partial sums,, that is, we want to find a number N so that sn ≤ N for every n. The upper, bound is provided courtesy of integration., Figure 3.1 shows the graph of y = 1/x 2 together with some rectangles that, lie completely below the curve and that all have base length one. Because the, heights of the rectangles are determined by the height of the curve, the areas, of the rectangles are 1/12 , 1/22 , 1/32 , and so on—in other words, exactly the, terms of the series. The partial sum sn is simply the sum of the areas of the, first n rectangles. Because the rectangles all lie between the curve and the, x-axis, any sum of rectangle areas is less than the corresponding area under, the curve, and so of course any sum of rectangle areas is less than the area, under the entire curve, that is, all the way to infinity. There is a bit of trouble, at the left end, where there is an asymptote, but we can work around that, , x and n, 1, , 2, , 3, , 4, , Figure 3.1: Plot of f (x ) = 1/x 2 alongside boxes, representing an = 1/n 2 ., , 5

Page 75 :
convergence tests, , 75, , easily. Here it is:, sn =, , 1, 12, , +, , 1, 22, , +, , 1, , +···+, , 32, , 1, n2, , n, , Z, < 1+, 1, , 1, x2, , ∞, , Z, dx < 1 +, 1, , 1, x2, , dx = 1 + 1 = 2,, , recalling how to compute this improper integral. Since the sequence of partial, sums sn is increasing and bounded above by 2, we know that lim sn = L < 2,, n →∞, , and so the series converges to some number less than 2., , In fact, it is possible, though difficult, to show that, , ∞, X, 1, n =1, , 3.2.2, , n2, , = π 2 /6., , Harmonic series, , We already know that, , X, , 1/n diverges. What goes wrong if we try to apply this, , technique to it? We find, sn =, , 1, 1, , 1, , +, , 2, , +, , 1, 3, , +···+, , 1, n, , n, , Z, < 1+, 1, , 1, x, , Z, dx < 1 +, 1, , ∞, , 1, x, , dx, , Z, , ∞, , 1, dx does, x, not converge, and claiming that sn is bounded by something divergent is to make, but this amounts to saying nothing, because the improper integral, , 1, , no claim atX, all: every real number is less than infinity! In other words, this does not, prove that, , 1/n diverges; it is just that this particular calculation fails to prove, , that it, converges. A slight modification, however, allows us to prove in a second way, X, that, , y, 1.2, , 1/n diverges., , f (x ) = 1/x, , a1, , 1, , Example 3.2.2 Consider a slightly altered version of Figure 3.1, shown in, Figure 3.2. Explain how to use the figure to see that the harmonic series, , 0.8, , diverges., 0.6, , a2, , Solution The rectangles this time are above the curve, that is, each rectangle, completely contains the corresponding area under the curve. This means that, , 0.4, , a3, a4, , sn =, , 1, 1, , +, , 1, 2, , +, , 1, 3, , +···+, , 1, n, , Z, >, 1, , n +1, , 1, x, , n +1, , dx = ln x, , = ln(n + 1)., , a5, , 0.2, , 1, , x and n, 1, , 2, , 3, , 4, , 5, , Figure 3.2: Plot of f (x ) = 1/x alongside boxes, representing an = 1/n., , 6

Page 76 :
76, , sequences and series, , As n gets bigger, ln(n + 1) goes to infinity, so the sequence of partial sums sn, must also go to infinity, so the harmonic series diverges., , 3.2.3, , Statement of integral test, , The important fact that clinches this example is that, , Z, , n +1, , lim, , n →∞, , which we can rewrite as, , Z, 1, , ∞, , 1, x, , 1, , 1, x, , dx = ∞,, , dx = ∞., , So these two examples taken together indicate that we can prove that a series, converges or prove that it diverges with a single calculation of an improper integral., This is known as the integral test, which we state as a theorem., Theorem 3.2.3 (Integral test) Suppose that f (x ) > 0 and is decreasing on, the infinite interval [k , ∞) (for some k ≥ 1) and that an = f (n ). Then the series, ∞, X, , Z, , n =k, , ∞, , f (x ) dx converges., , an converges if and only if the improper integral, k, , 3.2.4 p-series, The two examples we have seen are examples of p-series, which we first encountered, in Subsection 2.7.5. Recall that a p-series is any series of the form, , ∞, X, , 1/n p ., , n =1, , Theorem 3.2.4 A p-series converges if and only if p > 1., , We already proved this theorem using condensation in Example 2.7.8 and, Example 2.7.9. Nevertheless, we provide a second proof, using the integral test.

Page 77 :
convergence tests, , Proof We use the integral test; the case p = 1 is that of the harmonic series,, which we know diverges, so without loss of generality, we may assume that, p , 1. If p ≤ 0, lim 1/n p , 0, so the series diverges by Theorem 2.5.2. So we, n →∞, , may also assume that p > 0., Then we compute, , Z, 1, , ∞, , 1, xp, , dx = lim, , N →∞, , x 1− p, 1−p, , N, , = lim, 1, , N →∞, , N 1−p, 1−p, , −, , 1, 1−p, , ., , If p > 1 then 1 − p < 0 and lim N 1−p = 0, so the integral converges. If 0 < p < 1, N →∞, , then 1 − p > 0 and lim N 1−p = ∞, so the integral diverges., N →∞, , Example 3.2.5 Show that, , ∞, X, 1, n =1, , n3, , converges., , Solution We could of course use the integral test, but now that we have the, theorem we may simply note that this is a p-series with p = 3 > 1., , Example 3.2.6 Show that, , ∞, X, 5, n =1, , Solution We know that if, , ∞, X, , n4, , converges., , 1/n 4 converges then, , n =1, , by Theorem 2.3.1. Since, also., , ∞, X, , 5/n 4 also converges,, , n =1, 4, , 1/n is a convergent p-series,, , n =1, , Example 3.2.7 Show that, , ∞, X, , n =1, , ∞, X, 5, √ diverges., n =1, , ∞, X, , n, , 5/n 4 converges, , 77

Page 78 :
78, , sequences and series, , Solution This also follows from Theorem 2.3.4. Since, ∞, X, 5, with p = 1/2 < 1, it diverges, and so does, √ ., n =1, , 3.2.5, , ∞, X, 1, √ is a p-series, n =1, , n, , n, , Integrating for approximations, , Since it is typically difficult to compute the value of a series exactly, a good approximation is frequently required. In a real sense, a good approximation is only as, good as we know it is, that is, while an approximation may in fact be good, it is, only valuable in practice if we can guarantee its accuracy to some degree1 . This, guarantee is usually easy to come by for series with decreasing positive terms., ∞, X, , Example 3.2.8 Approximate, , 1/n 2 to two decimal places., , 1, , After all, π ≈ 17 just with very bad error bounds., It is better to make a statement like |π − 17| < 14,, which is not only saying that π is “close” to 17, but is, quantifying exactly how close (within 14—so perhaps, not all that close)., , n =1, , Solution Referring to Figure 3.1, if we approximate the sum by, , N, X, , 1/n 2 ,, , n =1, , the error we make is the total area of the remaining rectangles, all of which, lie under the curve 1/x 2 from x = N out to infinity. So we know the true, value of the series is larger than the approximation, and no bigger than the, approximation plus the area under the curve from N to infinity. Roughly, then,, we need to find N so that, , Z, , ∞, , 1, , N, , x2, , Z, , ∞, , dx < 1/100., , We can compute the integral:, , N, , 1, x2, , dx =, , 1, N, , ,, , so N = 100 is a good starting point. Adding up the first 100 terms gives approximately 1.634983900, and that plus 1/100 is 1.644983900, so approximating, the series by the value halfway between these will be at most 1/200 = 0.005, , ∞, X, 1, , π2, =, , so your approxima2, 6, n, n =1, tion of this series will also—in a roundabout way—, yield an approximate value for π—and one which will, be better than 17., It turns out that

Page 79 :
convergence tests, , in error. The midpoint is 1.639983900, but while this is correct to ±0.005, we, can’t tell if the correct two-decimal approximation is 1.63 or 1.64. We need to, make N big enough to reduce the guaranteed error, perhaps to around 0.004 to, be safe, so we would need 1/N ≈ 0.008, or N = 125. Now the sum of the first, 125 terms is approximately 1.636965982, and that plus 0.008 is 1.644965982, and the point halfway between them is 1.640965982. The true value is then, 1.640965982 ± 0.004, and all numbers in this range round to 1.64, so 1.64 is, correct to two decimal places., , Since, , ∞, X, 1, n =1, , n2, , =, , π2, 6, , , our estimate yields, , 1.63 <, , π2, 6, , < 1.65,, , and so, 9.78 < π 2 < 9.90., which means that, 3.127 < π < 3.147,, which is better than π ≈ 3. And we have explicit bounds on the error., , 79

Page 80 :
80, , sequences and series, , Exercises for Section 3.2, Determine whether each series converges or diverges., , (1), , ∞, X, n =1, , (2), , (4), , n, n2, , +1, , ∞, X, ln n, n =1, , ∞, X, , à, , n π /4, , ∞, X, n =1, , (3), , 1, , n2, , (5), , ∞, X, 1, n =1, , à, , (6), , ∞, X, n, n =1, , à, , (7), , 1, , à, , (9) Find an N so that, , (8), , ∞, X, 1, n =1, , (10) Find an N so that, , ∞, X, 1, n =0, , (11) Find an N so that, , en, , ∞, X, , N, X, 1, n =1, , is between, , n2, , n4, , N, X, 1, n =0, , ∞, X, ln n, n =1, , (12) Find an N so that, , is between, , n4, , is between, , en, , n =1, , 1, , n (ln n )2, n =2, , is between, , 1, , 1, , n (ln n )2, n =2, N, X, 1, n =1, , and, , n2, N, X, , à, , n ln n, , ∞, X, , and, , N, X, ln n, , à, , en, , ∞, X, n =2, , n2 + 1, n =1, , à, , en, , n4, , N, X, 1, n =0, , and, , en, , + 0.005., , + 10−4 ., , N, X, ln n, n =1, , 1, , n (ln n )2, n =2, , à, , n2, , and, , à, à, , + 0.005., N, X, , à, , 1, , n (ln n )2, n =2, , + 0.005., , à

Page 81 :
convergence tests, , 3.3, , 81, , More comparisons, , Armed with the ratio test and the integral test, it can be tempting to apply them, ∞, X, , frequently. But when faced with a new series, , an , what should we do? First,, , n =1, , consider lim an and if that limit is nonzero, then by Theorem 2.5.2 the series, n →∞, , diverges. If the series passes that first test, then it is worth considering other tests, like the ratio test, Z (perhaps if an has factorials and powers) or the integral test (if, an = f (n ) and, , f (x ) dx is not too hard to compute)., , But the ratio test and integral test are not the only tools in our toolbox. The, comparison test, which we discussed in Section 2.7, is extremely useful., , Example 3.3.1 Does, , ∞, X, n =2, , 1, n2, , log n, , converge?, When I write log, I mean the natural logarithm. I, prefer writing log n to writing ln n., , Solution The obvious first approach, based on what we know, is the integral, test. Unfortunately, we can’t compute the required antiderivative. But looking, at the series, it would appear that it must converge, because the terms we are, adding are smaller than the terms of a p-series, that is,, 1, n 2 log n, , <, , 1, n2, , ,, , when n ≥ 3. Since adding up the terms 1/n 2 doesn’t get “too big”, the new, series “should” also converge. Let’s make this more precise., ∞, ∞, X, X, 1, 1, The series, converges, if, and, only, if, converges—all, 2, 2, n log n, n log n, n =2, n =3, we’ve done is dropped the initial term. We know that, , n =3, , at two typical partial sums:, sn =, , 1, 32, , log 3, , +, , 1, 42, , log 4, , +, , 1, 52, , ∞, X, 1, , log 5, , +···+, , 1, n2, , log n, , <, , 1, 32, , n2, , +, , converges. Looking, , 1, 42, , +, , 1, 52, , +···+, , 1, n2, , = tn ., , Since the p-series converges, say to L, and since the terms are positive, tn < L., Since the terms of the new series are positive, the sn form an increasing

Page 82 :
82, , sequences and series, , sequence and sn < tn < L for all n. Hence the sequence {sn } is bounded and, so converges., Sometimes, even when the integral test applies, comparison to a known series, is easier, so it is a good idea to think about doing a comparison before doing the, integral test., Like the integral test, the comparison test can be used to show both convergence, and divergence. In the case of the integral test, a single calculation will confirm, whichever is the case. To use the comparison test we must first have a good idea as, to convergence or divergence and pick the sequence for comparison accordingly., , Example 3.3.2 Does, , ∞, X, , √, , n =2, , 1, n2, , converge?, , −3, , Solution We observe that the −3 should have little effect compared to the n 2, inside the square root, and therefore guess that the terms are enough like, , √, , 1/ n 2 = 1/n that the series should diverge. We attempt to show this by, comparison to the harmonic series. We note that, , √, , 1, , 1, 1, = ,, > √, n, n2 − 3, n2, , so that, sn = √, , 1, 22 − 3, , + √, , 1, 32 − 3, , +···+ √, , 1, , n2 − 3, , >, , 1, 2, , +, , 1, 3, , +···+, , 1, n, , = tn ,, , where tn is 1 less than the corresponding partial sum of the harmonic series, (because we start at n = 2 instead of n = 1). Since lim tn = ∞, lim sn = ∞, as well., , n →∞, , n →∞, , So the general approach is this: If you believe that a new series is convergent,, attempt to find a convergent series whose terms are larger than the terms of the, new series; if you believe that a new series is divergent, attempt to find a divergent, series whose terms are smaller than the terms of the new series. This is more of an, art than a science, which is part of what makes these sorts of problems so much, fun to do.

Page 83 :
convergence tests, , Example 3.3.3 Does, , ∞, X, , √, , n =1, , 1, n2, , +3, , converge?, , Solution Just as in the last example, we guess that this is very much like the, harmonic series and so diverges. Unfortunately,, , √, , 1, n2, , +3, , <, , 1, n, , ,, , so we can’t compare the series directly to the harmonic series. A little thought, leads us to, , √, , 1, , n2, , so if, , X, , (1/2), , 1, 1, > √, =, ,, 2, 2, 2n, +3, n + 3n, , 1/(2n ) diverges then the given series diverges. But since, , X, , 1/n, theorem ?? implies that it does indeed diverge., , X, , 1/(2n ) =, , 83

Page 85 :
convergence tests, , 3.4, , 85, , The mostly useless root test, , There is another convergence test called the root test, which can be justified with an, argument not so different2 from that which justified the ratio test in Subsection 3.1.1., The root test is very occasionally easier to apply, but usually not as good as choosing, to use the ratio test., With those disparaging remarks out of the way, let us now state the root test., , Theorem 3.4.1 (The Root Test) Consider the series, , ∞, X, , an where each term, , n =0, , an is positive. Suppose that lim (an ), , 1/n, , n →∞, , • if L < 1 the series, , X, , = L. Then,, , an converges,, , • if L > 1 the series diverges, and, • if L = 1, then the root test is inconclusive., Let’s apply the root test to analyze the convergence of a series., , Example 3.4.2 Analyze, , ∞, X, 5n, n =0, , nn, , ., , Solution Usually, the ratio test is a good choice when the series involves n th, powers; in this case, the ratio test turns out to be a bit difficult on this series,, since we have to calculate, , lim, , n →∞, , 5n +1 /(n + 1)n +1, 5n / n n, , 2, , Indeed, justifying the root test makes a good exercise for you, the reader, so it is included among the, exercises for this section.

Page 86 :
86, , sequences and series, , and that may not be entirely obvious. So we, begrudgingly, apply the root test,, which asks us to calculate, 5n, , L = lim, , !1 / n, , nn, , n →∞, , (5n )1/n, n →∞ (n n )1/n, , = lim, , = lim, , n →∞, , 5, n, , = 0., , Since L = 0 < 1, we may conclude that the given series converges., The root test is frequently useful when n appears as an exponent in the general, term of the series—though the ratio test is also useful in that case. Technically,, whenever the ratio test is conclusive (i.e., whenever lim an +1 /an = L , 1), so is, n →∞, , the root test—but not vice versa. In other words, the root test does work on some, series that the ratio test fails on., Example 3.4.3 Find a series for which the ratio test is inconclusive, but the, root test determines that the series converges., , Solution Here is such a situation. Try using the ratio test on, , ∞, X, , an where, , n =1, , (an ) is a sequence which “stutters” like, , , n /2, , , if n is even, and, 1/2, an = , , (, n, +, 1, ), /, 2, , 1/2, if n is odd., Note that a1 = a2 and a3 = a4 and a5 = a6 , so an +1 /an is often 1, which, messes up the ratio test—indeed, lim an +1 /an does not exist in this case., Nevertheless,, , n →∞, , , , 1/2, , , if n is even, and, √, 1 / 2, n, an = , , , 1/2(n +1)/(2n ) if n is odd,, √, √, and so lim n an = 1/ 2 < 1, which means the sequence converges by the, n →∞, , root test. Admittedly, we didn’t need the root test: this series is just a geometric

Page 87 :
convergence tests, , series where the terms repeat, so it definitely converges. Still, it proves the, point that the ratio test can fail while the root test succeeds., , 87

Page 88 :
88, , sequences and series, , Exercises for Section 3.4, , à, X, , (1) Prove theorem 3.4.1, the root test., 1/n, , (2) Compute lim |an |, n →∞, , 1/n, , (3) Compute lim |an |, n →∞, , for the series, for the series, , X, , 1/n 2 ., 1/n., , à, , à

Page 89 :
4 Alternating series, , 4.1, , Absolute convergence, , Roughly speaking there are two ways for a series to converge: As in the case of, ∞, X, , 1/n 2 , the individual terms get small very quickly, so that the sum of all of, , n =1, , them stays finite, or, as in the case of, enough (, , ∞, X, , ∞, X, , (−1)n +1 /n, the terms don’t get small fast, , n =1, , 1/n diverges), but a mixture of positive and negative terms provides, , n =1, , enough cancellation to keep the sum finite. You might guess from what we’ve seen, that if the terms get small fast enough to do the job, then whether or not some, terms are negative and some positive the series converges., , Theorem 4.1.1 If, , ∞, X, n =1, , |an | converges, then, , ∞, X, n =1, , an converges.

Page 90 :
90, , sequences and series, , Proof Note that 0 ≤ an + |an | ≤ 2|an | so by the comparison test, , ∞, X, , (an + |an |), , n =1, , converges. Now, ∞, X, , (an + |an |) −, , n =1, , ∞, X, , |an | =, , n =1, , ∞, X, , (an + |an | − |an |), , n =1, , =, , ∞, X, , an, , n =1, , converges by Theorem 2.3.4., , So given a series, ask whether, , ∞, X, , ∞, X, , an with both positive and negative terms, you should first, , n =1, , |an | converges. This may be an easier question to answer, because, , n =1, , we have tests that apply specifically to series with non-negative terms. If, converges then you know that, still may be true that, , ∞, X, , ∞, X, , ∞, X, , |an |, , n =1, , an converges as well. If, , X, , |an | diverges then it, , n =1, , an converges—you will have to do more work to decide, , n =1, , the question. Another way to think of this result is: it is (potentially) easier for, ∞, X, , an to converge than for, , n =1, , ∞, X, , |an | to converge, because the latter series cannot, , n =1, , take advantage of cancellation., If, ∞, X, , ∞, X, , |an | converges we say that, , n =1, , ∞, X, , an is absolutely convergent; to say that, , n =1, , an converges absolutely is to say that any cancellation that happens to come, , n =1, , along is not really needed, as the terms already get small so fast that convergence, is guaranteed by that alone. If, , ∞, X, n =1, , an converges but, , ∞, X, n =1, , |an | does not, we say that, , With this terminology, Theorem 4.1.1 is saying that, an absolutely convergent series is also a plain old, convergent series.

Page 91 :
alternating series, , ∞, X, , an converges conditionally. For example,, , n =1, , while, , ∞, X, , ∞, X, , (−1)n −1, , n =1, , (−1)n −1, , n =1, , 1, n, , 1, n2, , converges absolutely,, , converges conditionally., , Example 4.1.2 Does, , ∞, X, sin n, n =2, , n2, , converge?, , Solution In Example 2.7.7, we saw that, , ∞, X, | sin n |, n =2, , n2, , converges, so the given, , series converges absolutely., , Example 4.1.3 Does, , ∞, X, n =1, , (−1)n, , 3n + 4, 2n 2 + 3n + 5, , Solution Taking the absolute value,, to, , ∞, X, 3, n =1, , 10n, , ∞, X, n =1, , converge?, , 3n + 4, 2n 2, , + 3n + 5, , diverges by comparison, , , so if the series converges it does so conditionally. It is true, , that lim (3n + 4)/(2n 2 + 3n + 5) = 0, so to apply the alternating series, n →∞, , test we need to know whether the terms are decreasing. If we let f (x ) =, , (3x + 4)/(2x 2 + 3x + 5) then f 0 (x ) = −(6x 2 + 16x − 3)/(2x 2 + 3x + 5)2 ,, and it is not hard to see that this is negative for x ≥ 1, so the series is, decreasing and by the alternating series test it converges., , 91

Page 93 :
alternating series, , 4.2, , 93, , Alternating series test, , Next we consider series with both positive and negative terms, but in a regular, pattern: the signs alternate, as in the alternating harmonic series for example:, ∞, X, (−1)n +1, n =1, , n, , =, =, , 1, 1, 1, 1, , −1, , +, −, , +, , 2, 1, 2, , +, , 1, 3, , 1, 3, , −, , +, 1, 4, , −1, 4, , +···, , +··· ., , In this series the sizes of the terms decrease, that is, |an | forms a decreasing, sequence, but this is not required in an alternating series. As with positive term, series, however, when the terms do have decreasing sizes it is easier to analyze the, series, much easier, in fact, than positive term series. Consider pictorially what, is going on in the alternating harmonic series, shown in Figure 4.1. Because the, sizes of the terms an are decreasing, the partial sums s1 , s3 , s5 , and so on, form a, decreasing sequence that is bounded below by s2 , so this sequence must converge., Likewise, the partial sums s2 , s4 , s6 , and so on, form an increasing sequence that is, bounded above by s1 , so this sequence also converges. Since all the even numbered, partial sums are less than all the odd numbered ones, and since the “jumps” (that is,, the ai terms) are getting smaller and smaller, the two sequences must converge to, the same value, meaning the entire sequence of partial sums s1 , s2 , s3 , . . . converges, as well., 0, , 1, , 7, , 2, , 12, , 47 5, 60 6, , 1, , There’s nothing special about the alternating harmonic series—the same argument, works for any alternating sequence with decreasing size terms. The alternating, series test is worth calling a theorem., , Figure 4.1: Partial sums of the alternating harmonic, series

Page 94 :
94, , sequences and series, , Theorem 4.2.1 Suppose that (an ) is a decreasing sequence of positive numbers and lim an = 0. Then the alternating series, n →∞, , ∞, X, , (−1)n +1 an converges., , n =1, , Proof The odd numbered partial sums, s1 , s3 , s5 , and so on, form a decreasing sequence, because s2k +3 = s2k +1 − a2k +2 + a2k +3 ≤ s2k +1 , since, a2k +2 ≥ a2k +3 . This sequence is bounded below by s2 , so it must converge,, say lim s2k +1 = L. Likewise, the partial sums s2 , s4 , s6 , and so on, form, , We have considered alternating series with first index, 1, and in which the first term is positive, but a little, thought shows this is not crucial. The same test, applies to any similar series, such as, ∞, X, , k →∞, , an increasing sequence that is bounded above by s1 , so this sequence also, converges, say lim s2k = M. Since lim an = 0 and s2k +1 = s2k + a2k +1 ,, n →∞, , k →∞, , L = lim s2k +1 = lim (s2k + a2k +1 ) = lim s2k + lim a2k +1 = M + 0 = M ,, k →∞, , k →∞, , k →∞, , k →∞, , so L = M, the two sequences of partial sums converge to the same limit, and, this means the entire sequence of partial sums also converges to L., , We have shown more than convergence: if we are careful about thinking about, the previous argument, we can find error bounds. Let’s see how. Suppose that, L=, , ∞, X, , (−1)n +1 an, , n =1, , and that we approximate L by a finite part of this sum, say, L≈, , N, X, , (−1)n +1 an ., , n =1, , Because the terms are decreasing in size, we know that the true value of L must be, between this approximation and the next one, that is, between, N, X, n =1, , (−1)n +1 an and, , N, +1, X, , (−1)n +1 an ., , n =1, , Depending on whether N is odd or even, the second will be larger or smaller than, the first. This is important enough that it deserves to be highlighted as a theorem., , n =1, , (−1)n an ,, , ∞, X, n =17, , ∞, X, n =0, , (−1)n an , etc., , ( −1 ) n a n ,

Page 95 :
alternating series, , Theorem 4.2.2 Suppose that {an }∞, n =1 is a decreasing sequence of positive, numbers and lim an = 0. By Theorem 4.2.1, we then know that, n →∞, , ∞, X, , (−1)n +1 an, , n =1, , converges to some value, say L. Moreover, L is between, N, X, n =1, , (−1)n +1 an and, , N, +1, X, , (−1)n +1 an ., , n =1, , Example 4.2.3 Approximate the alternating harmonic series to one decimal, place., Solution We need to go roughly to the point at which the next term to be, added or subtracted is 1/10. Adding up the first nine and the first ten terms, we get approximately 0.746 and 0.646. These are 1/10 apart, but it is not, clear how the correct value would be rounded. It turns out that we are able to, settle the question by computing the sums of the first eleven and twelve terms,, which give 0.737 and 0.653, so correct to one place the value is 0.7., , 95

Page 96 :
96, , sequences and series, , Exercises for Section 4.2, Determine whether the following series converge or diverge., , (1), , (2), , ∞, X, (−1)n +1, 2n + 5, n =1, , à, , (3), , ∞, X, (−1)n +1, √, n−3, n =4, , à, , (4), , (5) Approximate, , ( −1 ) n + 1, , n =1, , ∞, X, , ( −1 ) n + 1, , n =1, , ∞, X, , (−1)n +1, , n =1, , (6) Approximate, , ∞, X, , ∞, X, n =1, , (−1)n +1, , 1, n3, 1, n4, , to two decimal places., , à, , to two decimal places., , à, , n, 3n − 2, , ln n, n, , à, , à

Page 97 :
5 Another comparison test, , We’ve covered a ton of material thus far in this course; there is one more comparison, test that comes in quite handy—the Limit Comparison Test—which we will meet in, Section 5.2. The purpose of this chapter, however, runs deeper than “just” another, comparison test., The emphasis on series has been almost entirely on the question of their convergence; we have not paid much heed to the value of the series, but we’ve developed, a lot of techniques to analyze their convergence. The question is always “Does it, converge?” and the answer is “yes, it converges!” or “no, it does not converge.”, Considering how qualitative our answer is, we might hope that there are equally, qualitative methods for analyzing series. If convergence is just a yes-or-no matter, one might hope that the methods for analyzing series are equally loose and, qualitative., But that hasn’t been our experience. Convergence is a tricky business, requiring, precision and careful analysis. There have been hints, though, that things are easier, than they seem: the comparison test is perhaps the best example of that. In your, past mathematical life, you’ve probably been given “expressions” or “equations” to, which you apply various rules in order to derive an answer. With the comparison, test, the situation is less about rules, and more about creatively ignoring parts of, the expression in order to find a useful bound. More than being rule-based, testing, convergence via the comparison test requires some guesswork, and a willingness to, ignore the parts of the expression that don’t matter, in order to get at the part that, does., Let me be more precise. Suppose we wanted to analyze the convergence of a

Page 98 :
98, , sequences and series, , series such as, , ∞, X, n =52, , n 4 − 3n + 5, 2n 5 + 5n 3 − n 2, , This is a complicated series, but the given expression is the ratio between a fifth, degree polynomial and a fourth degree polynomial, so this series “is more or less”, the same as the series, , ∞, X, n4, n =52, , n5, , =, , ∞, X, 1, n =52, , n, , which is the harmonic series, and diverges! This is how we’d like to think, but we, need to justify the concept of “is more or less.” This is what Section 5.2 will teach, us to do., , 5.1, , Convergence depends on the tail, , The harmonic series diverges, but so does the series, ∞, X, 1, n =100, , n, , =, , 1, 100, , +, , 1, 101, , +, , 1, 102, , +, , 1, 103, , +···, , Convergence doesn’t depend on the beginning of the series; whether or not I include, the first 99 terms, , 1, , +, , 1, , +, , 1, , +···+, , 1, , +, , 1, , +, , 1, , 1, 2, 3, 97, 98, 99, does affect whether or the sum of all the terms diverges. In short, convergence, depends not on how a series begins, but on how a series ends. The end of a series, is sometimes called the tail of the series., , Definition Let N > 1 be an integer, and consider a series, we get by removing the first N − 1 terms, namely, ∞, X, n =N, , an, , ∞, X, n =1, , an . The series

Page 99 :
another comparison test, , is called a tail of the given series., Here is the theorem that describes how tails relate to convergence., , Theorem 5.1.1 Let N > 1 be an integer. The series, only if, , ∞, X, , ∞, X, , an converges if and, , n =1, , an converges., , n =N, , This could be shortened to “The series converges iff a tail of the series converges,” or, even just to the slogan that convergence depends on the tail., Proof Suppose, , ∞, X, , an converges to L, meaning, , n =1, , lim, , M →∞, , M, X, , an = L ., , n =1, , In that case, applying limit laws reveals, ,  M,   −1 ,  X  NX, , , lim, an = lim , an  − , an , M →∞, M →∞, n =1, n =1, n =N, N −1 ,  M, ,  X ,  X , = lim , an  − lim , an , M →∞, M →∞, n =1, n =1,  M,   −1 , ,  X  NX, = lim , an  − , an , M, X, , M →∞, , n =1, , N −1 ,  X , = L − , an  ,, n =1, , n =1, , 99

Page 100 :
100, , sequences and series, , which means the series, , M, X, , an converges., , n =N, , The other direction is left to you, the reader., , ∞, X, , Example 5.1.2 Does the series, , 1, , n =153, , n2, , converge?, , Solution Yes! This series is a tail of the convergent p-series, , ∞, X, 1, n =1, , case, p = 2., , n2, , ; in this, , You might recall Example 3.1.1, which we’ll redo here., , Example 5.1.3 Show that, , ∞, X, n5, n =0, , 5n, , converges by using the comparison test and, , Theorem 5.1.1., Solution The given series converges iff, ever n ≥ 23, we have, , n =23, , n5, 5n, , The series, , ∞  n, X, 2, n =23, , 5, , ∞, X, n5, , ≤, , 2n, 5n, , =, , 5n, , by Theorem 5.1.1. But when-, ,  2 n, 5, , converges, since it is a geometric series with common, , ratio 2/5, so by comparison, the smaller series, , ∞, X, n5, n =23, , 5n, , also converges., , In light of Theorem 5.1.1, many textbooks will choose to write, , X, n, , an

Page 101 :
another comparison test, , instead of, , ∞, X, , an when discussing convergence. Whether or not the series converges, , n =1, , doesn’t depend on the initial index, so if we want to state theorems about convergence, we can avoid potentially distracting details by simply not speaking about, where the series begins., , 5.2, , Limit comparison test, , This test is usually called the limit comparison test,, Theorem 5.2.1 Suppose an ≥ 0 and bn ≥ 0. Then if, , lim, , n →∞, , the series, , ∞, X, , an, bn, , = L > 0,, , an converges if and only if, , n =1, , ∞, X, , bn converges., , n =1, , The idea is that this is basically a comparison test, but instead of asking for, something on the nose, like 0 ≤ an ≤ bn , I am instead asking only that an and bn, be similar in size., , 5.2.1, , Proof of the Limit Comparison Test, , Assuming lim, , n →∞, , an, bn, , = L > 0, then we may also conclude that lim, , n →∞, , Theorem 5.2.2 Suppose an ≥ 0 and bn ≥ 0. Then if, , lim, , n →∞, , the series, , ∞, X, n =1, , an, bn, , = L > 0,, , an converges if and only if, , ∞, X, n =1, , bn converges., , bn, an, , = 1/L > 0., , 101

Page 102 :
102, , sequences and series, , 5.2.2, , How to apply the Limit Comparison Test, , Example 5.2.3 Does the series, ∞, X, , n 4 − 3n + 5, , n =52, , 2n 5 + 5n 3 − n 2, , converges or diverge?, Solution We will invoke the Limit Comparison Test, that is, Theorem 5.2.2., Set, an =, , n 4 − 3n + 5, 2n 5, , +, , 5n 3, , − n2, , 1, , and bn =, , n, , ,, , so in this case,, , lim, , n →∞, , an, bn, , = lim, , n →∞, , = lim, , n →∞, , = lim, , n →∞, , =, =, =, and so, , ∞, X, , an and, , n =52, , too does, , ∞, X, , ∞, X, n =52, , n 4 − 3n + 5, 2n 5, , +, , 5n 3, , − n2, , ·, , n, 1, , n 5 − 3n 2 + 5n, 2n 5 + 5n 3 − n 2, n 5 − 3n 2 + 5n, 2n 5 + 5n 3 − n 2, , ·, , 1/n 5, 1/n 5, , 1 − n33 + n54, lim, n →∞ 2 + 5 − 1, 2, n3, n, , 3, limn →∞ 1 − n 3 + n54, , , limn →∞ 2 +, 1, 2, , 5, n2, , −, , 1, n3, , , , > 0,, , bn share the same fate. But, , ∞, X, , bn diverges, and so, , n =52, , an ., , n =52, , Of course, we could have just used a direct comparison test; the big benefit here is, that it is relatively painless to compute the limit, but it requires significant creativity

Page 103 :
another comparison test, , to set up a direct comparison test. Think back, for instance, to Example 3.1.1., , 103

Page 105 :
6 Power series, , 6.1, , Definitions, , At first, we studied the geometric series, ∞  n, X, 1, n =0, , but then we replaced, , 2, , ,, , 1, , with a variable, so we were able to analyze all geometric, 2, series simultaneously. In the end, we discovered that, ∞, X, n =0, , kx n =, , k, 1−x, , if |x | < 1,, , and that the series diverges when |x | ≥ 1. At the time, we thought of x as an, unspecified constant, but we could just as well think of it as a variable. A number, which depends on another number is just a function, so we may write, f (x ) =, , ∞, X, , kx n, , n =0, , and then observe that it is also the case that f (x ) = k /(1 − x ), as long as |x | < 1., While, X k /(1 − x ) is a reasonably easy function to deal with, the more complicated, , kx n does have its attractions: it appears to be an infinite version of one of the, , simplest sorts of functions—a polynomial. Do other functions have representations, as series? Is there an advantage to viewing them in this way?, , You may recall that we used r instead of x; I think r, was a good choice back then, since r evokes “ratio”, and stood for the common ratio between the terms, in a geometric series. Now we use x, which I think, is a good choice now. Why the change? To me, x, suggests “independent variable of a function” much, more generally than just a parameter in a geometric, series, and we will be thinking very broadly about, how quite general functions can be represented as, a power series. One may regard power series as a, natural generalization of polynomials, and you may, already be comfortable thinking of x as the variable, in a polynomial.

Page 106 :
106, , sequences and series, , Usually the coefficients aren’t all the same in a polynomial; the geometric series, is somewhat unusual in that all the coefficients of the powers of x are the same,, namely k. We will need to allow more general coefficients if we are to get anything, other than the geometric series., Definition Let (an ) be a sequence of real numbers starting with a0 . Then the, power series associated to (an ) is, ∞, X, , an x n ., , n =0, , Sometimes people are confused when considering, say,, ∞, X, , (sin x ) · x n ., , n =0, , Despite appearances, we will not be regarding such series as “power series.” Since, , (an ) is a sequence of real numbers, the coefficients an cannot depend on x., , 6.2, , Convergence of power series, , Example 6.2.1 Consider the power series, f (x ) =, , ∞, X, xn, n =1, , n, , ., , In Example 6.2.1, note that we are no longer considering convergence of a single series; instead, by, regarding x as a parameter, we are considering convergence for a whole family of series—namely all, those of the form, , ∞, X, , x n /n. It often happens in, , n =1, , For which x does this converge?, Another way to think of this question is this: we are being asked to determine, the domain of f ., Solution We can investigate the convergence of this series using the ratio test., The difficulty now is that, instead of a single limit, we must now compute a, limit than involves a variable x which is not the variable with respect to which, we are taking a limit., Another issue is that x may be negative, so we will actually check absolute, , mathematics that it is wiser to study a family of, objects simultaneously than to study a single object, in isolation.

Page 107 :
power series, , convergence; in other words, we are actually considering, ∞, X, xn, , f (x ) =, , n =1, , n, , ., , Bravely,, , L=, , |x |n +1, n +1, lim, n, n →∞ |x |, n, , |x |n +1 1, · |x |n, = lim, n →∞ n + 1, n, , |x |n +1 n, = lim, ·, n →∞ n + 1 |x |n, |x |n +1 · n, = lim, n →∞ (n + 1) · |x |n, , = lim |x |, n →∞, , n, n+1, , = |x | · lim, , n →∞, , n, n+1, , = |x |., , So when L = |x | < 1, the ratio test says that the series converges absolutely., When |x | > 1, the series does not converge absolutely—in fact, when |x | > 1, we, have that, , lim, , n →∞, , xn, n, , ,0, , and so the series diverges., So when |x | < 1 the series converges and when |x | > 1 it diverges, which, leaves only two values in doubt. When x = 1, the series is the harmonic series, and diverges; when x = −1, it is the alternating harmonic series (rather, the, negative of the “usual” alternating harmonic series) and converges., In other words, we may regard f as a function with domain [−1, 1)., We analyzed this power series by invoking the ratio test, but that was no accident., We will see that the ratio test applied to a power series will always have the same nice, form. Feeling confident from our display of bravery before the preceding example,, , 107

Page 108 :
108, , sequences and series, , let’s attack the general case of a power series, f (x ) =, , ∞, X, , an x n ., , n =0, , When does this series converge absolutely? Applying the ratio test again, we find, , |an +1 | · |x |n +1, n →∞, |an | · |x |n, |an +1 |, = lim |x | ·, n →∞, |an |, |an +1 |, = |x | · lim, ., n →∞ |an |, , L = lim, , So the series converges absolutely whenever L < 1, but in this case, L depends in a, rather uncomplicated way on |x |. The whole story is controlled by a limit that does, not depend on x, namely, 1/R = lim, , n →∞, , |an +1 |, ,, |an |, , which you see I have presciently related to a hitherto unmentioned variable, R. So, the series converges absolutely whenever |x |/R < 1, meaning the series converges, absolutely whenever |x | < R. When |x | > R, the series diverges1 ., Once again, only the two values x = ±R require further investigation. In any, case, if we begin with a sequence (an ) for which, 1/R = lim, , n →∞, , the associated power series, , ∞, X, , |an +1 |, ,, |an |, , an x n will converge for values of x in the interval, , n =0, , (−R , R ), and it may even converge (or not, depending on the specific situation) at, x = R or at x = −R, so sometimes the interval on which the series makes sense will, be [−R , R ) or (−R , R ] or even [−R , R ]., , 1, , This is a consequent of the ratio test, but perhaps, we haven’t emphasized it enough; if L > 1 in the, ratio test, then the limit of the terms is not zero, so, not only does the series not converge absolutely—the, original series diverges, too, by the limit test.

Page 109 :
power series, , Definition The set of values of x for which the series, , ∞, X, , an x n converges is, , n =0, , the interval of convergence., , As a consequence of the above discussion, the interval of convergence of a power, series always have a nice form., Theorem 6.2.2 For a power series, the interval of convergence is, in fact, an, interval. It has the form (−R , R ) or [−R , R ) or (−R , R ] or [−R , R ]. In short, it is, centered around 0., , Because the interval of convergence is, indeed, centered around 0, and because, we often don’t care about what happens at the endpoint, it is often convenient to, just describe R., Definition In the interval of convergence of a power series, the value R is, called the radius of convergence of the series., , Two special cases deserve mention. If, , lim, , n →∞, , |an +1 |, = 0,, |an |, , then we might say “R = ∞” since no matter what x is, the product, , |x | lim, , n →∞, , |an +1 |, = 0,, |an |, , and therefore for all values of x, the power series, , ∞, X, , an x n converges. So the radius, , n =0, , of convergence, R, is infinite., The opposite may happen as well. It may happen that, , lim, , n →∞, , |an +1 |, = ∞,, |an |, , 109

Page 110 :
110, , sequences and series, , in which case the only way that, , |x | lim, , n →∞, , |an +1 |, <1, |an |, , is when |x | = 0. The interval of convergence is just the single point {0}. And in this, case, we say R = 0., Warning People often confuse “radius of convergence” with “interval of convergence.” For starters, the radius of convergence is a single number, while, the interval of convergence is, well, an interval (though it was defined as just a, set—the fact that the set where the series converges is an interval was a theorem, and was certainly not obvious—perhaps the first of many lovely surprises with, power series)., There is another difference between the radius of convergence and the, interval of convergence; it is not simply that they are the same information, in different packages. The radius of convergence cannot distinguish between,, say, (−R , R ) versus [−R , R ]. The interval of convergence contains the extra, information about what is happening at the endpoints. So a homework question, which asks you to find the radius R is much easier than a homework question, asking you to find the interval., , 6.3, , Power series centered elsewhere, , If we are speaking of the “radius” of convergence, you might wonder about the “center”, of convergence. Thus far, our intervals of convergence have been centered around, the origin, but by changing the series, we can move the interval of convergence., Let’s see how., Consider once again the geometric series,, f (x ) =, , ∞, X, n =0, , xn =, , 1, 1−x, , for x ∈ (−1, 1)., , Whatever benefits there might be in using the series form of this function are only

Page 111 :
power series, , 111, , available to us when x is between −1 and 1. We can address this shortcoming by, modifying the power series slightly., Example 6.3.1 Find a series representation for, , (−5, 1)., , 1, 1−x, , valid on the interval, , Solution Consider that the series, ∞, X, (x + 2)n, n =0, , 3n, , =, , ∞ , , X, x +2 n, n =0, , 3, , 1, 3, =, ,, 2, 1−x, 1 − x+, 3, , =, , because this is just a geometric series with x replaced by (x + 2)/3. Multiplying, both sides by 1/3 gives, ∞, X, (x + 2)n, n =0, , 3n +1, , =, , 1, 1−x, , ,, , the same function as before. For what values of x does this series converge?, Since it is a geometric series, we know that it converges when, , |x + 2|/3 < 1, |x + 2| < 3, −3 < x + 2 < 3, −5 < x < 1., , So we have a series representation for 1/(1 − x ) that works on a larger interval, than before, at the expense of a somewhat more complicated series. The, endpoints of the interval of convergence now are −5 and 1, but note that those, two endpoints can be described as −2 ± 3. We say that 3 is the radius of, convergence, and we now say that the series is centered at −2., Let’s capture this in a definition., , It is worth contrasting Definition 18 with Definition 15, our original description of power series. In, particular,, zero., , ∞, X, n =0, , an x n is a power series centered at

Page 112 :
112, , sequences and series, , Definition Let (an ) be a sequence of real numbers starting with a0 . Then the, power series centered at c and associated to (an ) is the series, ∞, X, , an (x − c )n ., , n =0, , In Example 6.3.1, we formed a series that involved (x + 2)n , meaning that in, that case c = −2., You are now in a position to try your hand at finding the radius of convergence, and sometimes even the interval of convergence for some power series. I encourage, you to cook up your own power series, and then see what you can say about the, radius and interval of convergence.

Page 113 :
power series, , Exercises for Section 6.3, Find the radius and interval of convergence for each series. In exercises 3 and 4, do not, attempt to determine whether the endpoints are in the interval of convergence., (1), , ∞, X, , nx n, , à, , n =0, , (2), , (3), , ∞, X, xn, n!, n =0, ∞, X, n!, n =1, , (4), , ∞, X, n!, n =1, , (5), , nn, , xn, , à, , (x − 2)n, , ∞, X, (n ! )2, n =1, , (6), , nn, , à, , nn, , à, , (x − 2)n, , ∞, X, (x + 5)n, n (n + 1), n =1, , à, , à, , (7) Find a power series with radius of convergence 0., , à, , 113

Page 114 :
114, , sequences and series, , 6.4, , Calculus with power series, , Now we know that some functions can be expressed as power series, which look, like infinite polynomials. Since calculus, that is, computation of derivatives and, antiderivatives, is easy for polynomials, the obvious question is whether the same is, true for infinite series. The answer is yes!, , Theorem 6.4.1 Suppose the power series f (x ) =, , ∞, X, , an (x − c )n has radius, , n =0, , of convergence R. Then, f 0 (x ) =, , ∞, X, , nan (x − c )n −1 ,, , n =1, , Z, f (x ) dx = C +, , ∞, X, n =0, , an, n+1, , (x − c )n +1 ,, , for values of x in the interval (c − R , c + R ). The two new series have radius of, convergence R, just like the original series., , Example 6.4.2 Find an alternating series for log, , 3, , to two decimal places., Solution Begin with the geometric series, namely, 1, 1−x, , =, , ∞, X, n =0, , xn., , 2, , . Approximate this value

Page 115 :
power series, , Then integrate term-by-term to find, , Z, , 1, 1−x, , dx = − log |1 − x | =, , ∞, X, n =0, , log |1 − x | =, , ∞, X, n =0, , −, , 1, n+1, , 1, n+1, , x n +1, , x n +1, , for real numbers x such that |x | < 1. To compute log(3/2), we should choose, x = −1/2. In that case,, , log(3/2) = log(1 − −1/2), ∞, X, 1, 1, =, (−1)n, n, n + 1 2 +1, n =0, and so, , log(3/2) ≈, , 1, 2, , −, , 1, 8, , +, , 1, 24, , −, , 1, 64, , +, , 1, 160, , −, , 1, 384, , +, , 1, 896, , =, , 909, 2240, , ≈ 0.406., Because this is an alternating series with decreasing terms, we know that the, true value is between 909/2240 and 909/2240 − 1/2048 = 29053/71680 ≈, .4053, so log(3/2) is, to two decimal places, 0.41., , Example 6.4.3 What about log(9/4)? Find an approximate value for it., Solution Since 9/4 is larger than 2 we cannot use the series directly, but, , log(9/4) = log((3/2)2 ) = 2 log(3/2) ≈ 0.82,, so in fact we get a lot more from this one calculation than first meets the eye., To estimate the true value accurately we actually need to be a bit more careful., When we multiply by two we know that the true value is between 0.8106, and 0.812, and so, rounded to two decimal places, the true value is 0.81., , 115

Page 116 :
116, , sequences and series, , Exercises for Section 6.4, , (1) Find a series representation for log 2., , à, , (2) Find a power series representation for 1/(1 − x )2 ., (3) Find a power series representation for 2/(1 − x )3 ., , à, à, , (4) Find a power series representation for 1/(1 − x )3 . What is the radius of convergence?, , à, Z, (5) Find a power series representation for, , log(1 − x ) dx., , à

Page 117 :
7 Taylor series, , We have seen that some functions can be represented as series. But thus far, our, only examples have been those that result from manipulation of our one fundamental, example, the geometric series. We might start with, , 1, 1−x, , =, , ∞, X, , x n when |x | < 1,, , n =0, , and then, say, integrate term-by-term to get a formula for a logarithm, as we did in, Example 6.4.2. Instead of starting with a series representing a function, and then, messing around with the series to find more functions represented by series, we, should start with a function, and then try to find a series that represents it—if that, is possible!, , 7.1, , Finding Taylor series, , The easiest case is when the function f is given to us as a power series already!, Suppose that f (x ) =, , ∞, X, , an x, , n, , on some interval of convergence, say, when |x | <, , n =0, , R. Then we know, by Theorem 6.4.1, that we can compute derivatives of f by, , It might seem that considering this case is pointless:, who cares about representing a function by a power, series if the function is given to us as a power series? The point is not to come up with a new power, representation: the point is to relate the derivatives, of the function to the coefficients of the power series,, which is the sort of thing that generalizes. By studying a case we already understand completely, we are, seeking insights which will apply even in cases we, don’t fully understand.

Page 118 :
118, , sequences and series, , differentiating the series term-by-term. Let’s look at the first few derivatives., f 0 (x ) =, , ∞, X, , nan x n −1 = a1 + 2a2 x + 3a3 x 2 + 4a4 x 3 + · · ·, , n =1, , f 00 (x ) =, , ∞, X, , n (n − 1)an x n −2 = 2a2 + 3 · 2a3 x + 4 · 3a4 x 2 + · · ·, , n =2, , f 000 (x ) =, , ∞, X, , n (n − 1)(n − 2)an x n −3 = 3 · 2a3 + 4 · 3 · 2a4 x + · · ·, , n =3, , By examining these derivatives, we can discern the general pattern. The k th, Recall that, for a function f , we write its k th derivative by writing f (k ) . So the k th derivative evaluated, at x is written f (k ) (x )., , derivative must be, f (k ) ( x ) =, , ∞, X, , n (n − 1)(n − 2) · · · (n − k + 1)an x n −k, , n =k, , = k (k − 1)(k − 2) · · · (2)(1)ak + (k + 1)(k ) · · · (2)ak +1 x +, + (k + 2)(k + 1) · · · (3)ak +2 x 2 + · · · ,, , but we can write this more easily by using factorials, namely, f (k ) ( x ) =, , ∞, X, n =k, , n!, , (n − k ) !, , an x n −k = k !ak + (k + 1)!ak +1 x +, , (k + 2) !, ak + 2 x 2 + · · · ., 2!, , Now, substituting x = 0 yields, f (k ) (0) = k !ak +, , ∞, X, , n!, an 0n −k = k !ak ,, (, n, −, k, ), !, n =k +1, , and solving for ak gives, ak =, , f (k ) ( 0 ), , ., , k!, So if a function f can be represented by a series, we know just what series it is! We, know the series because we know the derivatives of f ., , Note that ak = f (k ) (0)/k ! makes sense even when, k = 0, since in that special case, we have a0 =, f (0) (0)/0! = f (0) because the zeroth derivative of f, is f itself, and 0! = 1.

Page 119 :
taylor series, , Definition Given a function f , the series, ∞, X, f (n ) ( 0 ) n, x, n!, n =0, , is called the Maclaurin series for f , or often just the Taylor series for f, centered around zero, since it is a power series centered around zero in the, sense of Section 6.3., , Let me warn you that, in order to write down this power series, you had better be, able to take first, second, third—indeed, n th —derivatives at zero, because the given, formula involves f (n ) (0). And there is a worse warning., Warning Even if f is infinitely differentiable at zero (meaning f (n ) (0) makes, sense), even if I then write down, ∞, X, f (n ) ( 0 ) n, x ,, n!, n =0, , does not mean that I have a power series for f valid on any open interval! All I, have done is written down the power series that must represent f assuming it, has a power series representation at all! Nobody is promising you that some, function you find—even if it is infinitely differentiable—actually has a power, series representation. In other words, I am not claiming that there is an R , 0, so that, f (x ) =, , ∞, X, f (n ) ( 0 ) n, x for x ∈ (−R , R )., n!, n =0, , All I am claiming is that if f has such a power series, then I can find it by taking, derivatives., , Let’s see this worked out in an example we already understand, namely, for the, function f (x ) = 1/(1 − x )., , 119

Page 120 :
120, , sequences and series, , Example 7.1.1 Find a Taylor series for f (x ) = 1/(1 − x ) centered around, zero., Solution We need to compute the derivatives of f , and then hope to spot a, pattern. Here we go:, f ( x ) = ( 1 − x ) −1 ,, f 0 ( x ) = ( 1 − x ) −2 ,, f 00 (x ) = 2(1 − x )−3 ,, f 000 (x ) = 6(1 − x )−4 ,, f (4) = 4!(1 − x )−5 ,, , .., ., f (n ) = n !(1 − x )−n −1 ., , I see the pattern!, f (n ) ( 0 ), , =, , n !(1 − 0)−n −1, , n!, n!, and the Taylor series centered around zero is, ∞, X, n =0, , 1 · xn =, , ∞, X, , = 1,, , xn,, , n =0, , which we already knew—it is just the geometric series., So given a function f , we may be able to differentiate it around zero, spot a, pattern, and thereby compute the Taylor series, but that does not mean we have, found a series representation for f . Worse, we don’t even know if the series we wrote, down converges anywhere, let alone converges to the function f ! The miracle is that, for many popular functions1 the Taylor series does converge to f on some interval., 2, , But this is certainly not true of all functions—or even true of most functions!, As a practical matter, if we are interested in using a series to approximate a value, of a function at some point, we will need some finite number of terms of the series., , 1, , Arguably, this is not so surprising: maybe these, are popular functions precisely because they have, nice properties. That the sky is sky-colored is not so, surprising as the fact that it is blue., 2, , Most functions are probably not even continuous,, let alone infinitely differentiable, let alone having the, property of having a power series representation!, The latter mouthful is usually termed real analytic.

Page 121 :
taylor series, , Even for functions with terrible looking derivatives, we can compute the initial terms, using computer software like Sage. If we want to know the whole series—that is, the, n th term in the series for an arbitrary n—then we need a function whose derivatives, fall into a pattern that we can discern. A few of the most popular functions have, nice patterns. Let’s see some now!, Example 7.1.2 Find a Taylor series for sin x., Solution The derivatives are quite easy, namely, f 0 (x ) = cos x ,, f 00 (x ) = − sin x ,, f 000 (x ) = − cos x ,, f (4) (x ) = sin x ,, and then the pattern repeats. We want to know the derivatives at zero, so, those are, f 0 (0) = cos 0 = 1, f 00 (0) = − sin 0 = 0, f 000 (0) = − cos 0 = −1,, f (4) (0) = sin 0 = 0,, and then the pattern repeats, namely it goes—starting from the zeroth derivative—, like this:, 0,, , 1,, , 0,, , −1,, , 0,, , 1,, , 0,, , −1,, , 0,, , ...., , And so, the Taylor series is, x−, , x3, 3!, , +, , x5, 5!, , −··· =, , ∞, X, n =0, , (−1)n, , x 2n +1, , (2n + 1)!, , ., , Sometimes people are confused by the fact that the exponent in this power, series is 2n + 1 instead of just n. It turns out that setting it up this way,, with x 2n +1 , nicely manages to kill all the terms x even number which have a zero, , 121

Page 122 :
122, , sequences and series, , coefficient, since differentiating f an even number of times results in ± sin x,, which is zero when x = 0., But there is more to be anxious about. Before worrying whether this series, converges to sin x, we should ask the prior question: does this series converge, anywhere? Let’s determine the radius of convergence by using the ratio test,, which asks us to consider, , |x |2n +3 (2n + 1)!, n →∞ (2n + 3) ! |x |2n +1, |x |2, = lim, = 0,, n →∞ (2n + 3)(2n + 2), , L = lim, , and so the series converges regardless of what x is. It will turn out later, by, applying Theorem 7.2.1, that this series does converge to sin x., Sometimes the formula for the n th derivative of a function f is difficult to discover,, but a combination of a known Taylor series and some algebraic manipulation leads, easily to the Taylor series for f ., Example 7.1.3 Find the Taylor series for f (x ) = x sin(−x ) centered at zero., Solution To get from sin x to x sin(−x ) we substitute −x for x and then multiply, by x. Let’s do the same thing to the series for sin x, namely, x, , ∞, X, n =0, , (−1)n, , ∞, X, (−x )2n +1, x 2n +1, (−1)n (−1)2n +1, =x, (2n + 1)!, (2n + 1)!, n =0, , =, , ∞, X, n =0, , ( −1 ) n + 1, , x 2n +2, , (2n + 1)!, , .

Page 123 :
taylor series, , Of course, we could have differentiated f (x ), which would have yielded, f 0 (x ) = −x cos (−x ) + sin (−x ) = 0,, f 00 (x ) = −x sin (−x ) − 2 cos (−x ) = −2,, f 000 (x ) = x cos (−x ) − 3 sin (−x ) = 0,, f (4) (x ) = x sin (−x ) + 4 cos (−x ) = 4,, , but maybe it would have been harder to pick up on the pattern that way., As we have seen in Section 6.3, a general power series can be centered at a point, other than zero, and the method that produces the Taylor series can also produce, such series., Definition Given a function f , the series, ∞, X, f (n ) ( c ), (x − c )n, n!, n =0, , is called the Taylor series for f centered around c., , Let’s see an example, reminiscent of Example 6.3.1 but attacked with a different, method., Example 7.1.4 Find a Taylor series centered at −2 for 1/(1 − x )., , Solution If the series is, , ∞, X, , an (x + 2)n then looking at the kth derivative:, , n =0, , k !(1 − x )−k −1 =, , ∞, X, n =k, , n!, , (n − k ) !, , an (x + 2)n −k, , 123

Page 124 :
124, , sequences and series, , and substituting x = −2 we get k !3−k −1 = k !ak and ak = 3−k −1 = 1/3k +1 , so, the series is, , ∞, X, (x + 2)n, n =0, , 3n +1, , .

Page 125 :
taylor series, , Exercises for Section 7.1, For each function, find the Taylor series centered at c, and the radius of convergence. Do, not worry about whether or not the series converges to the given function—that will be our, concern in Section 7.2., , à, e around c = 0, à, 1/x around c = 5 à, log x around c = 1 à, log x around c = 2 à, 1/x 2 around c = 1 à, , (1) cos x around c = 0,, (2), (3), (4), (5), (6), , x, , √, , (7) 1/ 1 − x around c = 0, , à, , (8) Find the first four terms of the Taylor series for tan x centered at zero. By “first four terms”, I mean up to and including the x 3 term., , à, , (9) Use a combination of Taylor series and algebraic manipulation to find a series centered, at zero for x cos(x 2 )., , à, , (10) Use a combination of Taylor series and algebraic manipulation to find a series centered, at zero for xe −x ., , à, , 125

Page 126 :
126, , sequences and series, , 7.2, , Taylor’s Theorem, , In Section 7.1, we were finding Taylor series without worrying at all that what, we were finding actually related to the original function f . We now remedy this, oversight., Our whole mission in this, Chapter 7, was to start with a function, then write, down a power series converging to f , with the plan that we might finally use that, power series to, say, approximate values of the function, among other things. What, we have done thus far is assumed a function had such a power series representation,, and then deduced that the coefficients are somehow related to the derivatives of f ., So does the power series converge to the function?, And if the power series we got in Section 7.1 does converge to f , how good is the, approximation? I mean, if I just consider the first N terms of the power series, am, I close to f at all? How close? Remember our experience with alternating series?, For alternating series, Theorem 4.2.2 not only gave a test for convergence: it also, bounded how far the N th partial sum could be from the true value of the series., Let’s see a similar sort of “error estimate” for power series., Theorem 7.2.1 (Taylor’s theorem) Suppose that f is defined on some open, interval I = (a − R , a + R ) around a and suppose the function f is (N + 1)times differentiable on I, meaning that f (N +1) (x ) exists for x ∈ I. Then for each, x , a in I there is a value z between x and a so that, f (x ) =, , N, X, f (n ) ( a ), f (N +1) ( z ), (x − a )n +, (x − a )N +1 ., n!, (, N, +, 1, ), !, n =0, , The upshot here is that, by bounding the function f (N +1) on the interval between, x and a, we manage to bound the difference between f (x ) and the partial sum., See Example 7.2.2 for an example. But before we get to the example, let’s prove, Theorem 7.2.1; the proof perhaps seems unmotivated, since we’ll be “clever” in, setting things up, but I hope you will be able to follow the argument, even if you, don’t trust that you could have created the proof ex nihilo.

Page 127 :
taylor series, , Proof Define the function F (t ) by, F (t ) =, , N, X, f (n ) ( t ), (x − t )n + B (x − t )N +1 for t between a and x., n!, n =0, , Here we have replaced a by t in the first N + 1 terms of the Taylor series, and, added a carefully chosen term on the end, with B to be determined. Note that, we are temporarily keeping x fixed, so the only variable in this equation is t, and, we will be interested only in t between a and x. If we set t = a, then we get, F (a ) =, , N, X, f (n ) ( a ), (x − a )n + B (x − a )N +1 ., n!, n =0, , Set this equal to f (x ):, f (x ) =, , N, X, f (n ) ( a ), (x − a )n + B (x − a )N +1 ., n!, n =0, , Since x , a, we can solve this for B, which is a “constant”—it depends on x and, a but those are temporarily fixed. Now we have defined a function F (t ) with, the property that F (a ) = f (x ). Consider also F (x ): all terms with a positive, power of (x − t ) become zero when we substitute x for t, so we are left with, F (x ) = f (0) (x )/0! = f (x ). So F (t ) is a function with the same value on the, endpoints of the interval [a , x ]. By Rolle’s theorem, we know that there is a, value z ∈ (a , x ) such that F 0 (z ) = 0. Let’s look at F 0 (t ). Each term in F (t ),, except the first term and the extra term involving B, is a product, so to take the, derivative we use the product rule on each of these terms. It will help to write, out the first few terms of the definition:, F (t ) = f (t ) +, , +, , f (1) ( t ), 1!, f (N ) ( t ), N!, , (x − t )1 +, , f (2) ( t ), 2!, , (x − t )2 +, , (x − t )N + B (x − t )N +1 ., , f (3) ( t ), 3!, , (x − t )3 + · · ·, , 127

Page 128 :
128, , sequences and series, , Now take the derivative:, , , ,  f (1) (t ), , f (2) ( t ), 0, 1, F (t ) = f (t ) + , (x − t ) (−1) +, (x − t ) , 1!, 1!, , , (, 2, ), (, 3,  f (t ), , f ) (t ), 1, 2, + , (x − t ) (−1) +, (x − t ) , 1!, 2!, , , (, 4, (, 3, ), ,  f (t ), f ) (t ), 2, 3, , (x − t ) (−1) +, (x − t )  + · · · +, + , 2!, 3!, , , (, N, ), ,  f (t ), f (N +1) ( t ), N, N −1, , + , (x − t ) , (x − t ) (−1) +, N!, (N − 1) !, 0, , 0, , + B (N + 1)(x − t )N (−1)., , Now most of the terms in this expression cancel out, leaving just, F 0 (t ) =, , f (N +1) ( t ), N!, , (x − t )N + B (N + 1)(x − t )N (−1)., , At some z, F 0 (z ) = 0 so, 0=, B (N + 1)(x − z )N =, B=, , f (N +1) ( z ), N!, f (N +1) ( z ), N!, f (N +1) ( z ), , (N + 1) !, , (x − z )N + B (N + 1)(x − z )N (−1), (x − z )N, ., , Now we can write, F (t ) =, , N, X, f (n ) ( t ), f (N +1) ( z ), (x − t )n +, (x − t )N +1 ., n!, (N + 1) !, n =0

Page 129 :
taylor series, , Recalling that F (a ) = f (x ) we get, f (x ) =, , N, X, f (n ) ( a ), f (N +1) ( z ), (x − a )n +, (x − a )N +1 ,, n!, (, N, +, 1, ), !, n =0, , which is what we wanted to show., It may not be immediately obvious that this is particularly useful, so let’s look at, some examples of Theorem 7.2.1 in action., Example 7.2.2 Suppose x ∈ [−π /2, π /2]. Find a polynomial approximation, for sin x accurate to ±0.005., Solution Note that if we can compute sin x for x ∈ [−π /2, π /2], then we can, compute sin x for all x. So what we are asking for here is actually quite general!, Once we figure out how to calculate sin x for x ∈ [−π /2, π /2], we can shift any, other value of x into that interval, and thereby compute sin x., Let’s get started. From Theorem 7.2.1, we have that, , sin x =, , N, X, f (N +1) ( z ), f (n ) ( a ), (x − a )n +, (x − a )N +1 for some z between a and x., n!, (, N, +, 1, ), !, n =0, , What can we say about the size of that “error” term? In other words, how big, could, , f (N +1) ( z ), , (x − a )N +1, , (N + 1) !, , possibly be when z is between x and a? Every derivative of sin x is either ± sin x, or ± cos x, so |f (N +1) (z )| ≤ 1. The factor (x − a )N +1 is a bit more difficult, since, x − a could be quite large. To simplify matters, let’s pick a = 0. We need to, pick N so that, , x N +1, , (N + 1) !, , < 0.005., , Since we are only considering the case when x ∈ [−π /2, π /2], we have, x N +1, , (N + 1) !, , <, , 2N +1, , (N + 1) !, , ., , 129

Page 130 :
130, , sequences and series, , The quantity on the right decreases with increasing N , so all we need to do is, find an N so that, , 2N +1, , (N + 1) !, , < 0.005., , A little trial and error shows that N = 8 works, and in fact 29 /9! < 0.0015., Consequently,, , sin x =, , 8, X, f (n ) ( 0 ) n, x ± 0.0015, n!, n =0, , =x−, , x3, 6, , +, , x5, 120, , −, , x7, 5040, , y, 1, , ± 0.0015., , f (x ) = sin x, , 0.5, , Figure 7.1 shows the graphs of sin x and and the approximation on [0, 3π /2]., As x gets larger, the approximation heads to negative infinity very quickly,, , x, , −4, , since it is essentially acting like −x 7 ., We can extract a bit more information from this example. If we do not restrict, , −2, , 2, , −0.5, , the value of x to lie in the interval x ∈ [−π /2, π /2], we still know that, f (N +1) ( z ), , (N + 1) !, , x N +1 ≤, , x N +1, , (N + 1) !, , because the derivative f (N +1) (z ) is either ± sin z or ± cos z. We can use this to prove, the following result., Theorem 7.2.3 For all real numbers x,, , sin x =, , ∞, X, , (−1)n 2n +1, x, ., (2n + 1)!, n =0, , Proof If we can show that, , lim, , N →∞, , x N +1, , (N + 1) !, , =0, , −1, Figure 7.1: A plot of f (x ) = sin x with a thick line,, placed alongside the dotted line y = x, a dashed, x3, plot of y = x −, , and a thin solid plot of y =, 6, x3, x5, x−, +, . Note how successively higher partial, 6, 120, sums of the Taylor series are hugging the graph of, sin x increasingly well., , 4

Page 131 :
taylor series, , for each x, then we know that the error term, f (N +1) ( z ), , (N + 1) !, , x N +1, , x N +1 ≤, , (N + 1) !, , ,, , is, in the limit, zero, and so we may conclude that, , sin x =, , =, , ∞, X, f (n ) ( 0 ) n, x, n!, n =0, ∞, X, , (−1)n, , n =0, , x 2n +1, , (2n + 1)!, , ., , In other words, we will conclude that the sine function is actually equal to its, Taylor series around zero for all x. But how can we prove that the limit is zero?, Suppose that N is larger than |x |, and let M be the largest integer less than |x |., If M = 0, then x N +1 is small, and even smaller after dividing by (N + 1)!. On, the other hand, if M > 0, then just a bit more work is called for. We compute that, , |x N +1 |, |x | |x | |x |, |x | |x | |x |, |x | |x |, ···, ···, =, N +1 N N −1, M +1 M M −1, 2 1, (N + 1) !, |x |, |x | |x |, |x | |x |, ≤, ·1·1···1·, ···, N +1, M M −1, 2 1, |x | |x |M, ., =, N + 1 M!, The quantity |x |M /M ! is a constant, so, , |x | |x |M, =0, N →∞ N + 1 M !, lim, , and by squeezing via Theorem 1.7.2,, , lim, , N →∞, , x N +1, , (N + 1) !, , =0, , as desired., Essentially the same argument works for cos x, which yields the following., , 131

Page 132 :
132, , sequences and series, , Theorem 7.2.4 For all real numbers x,, , cos x =, , ∞, X, (−1)n 2n, x ., (2n )!, n =0, , And similarly, the function e x is real analytic, meaning that the Taylor series, for e x converges to e x ., Theorem 7.2.5 For all real numbers x,, ex =, , ∞, X, 1 n, x ., n!, n =0, , Example 7.2.6 Find a polynomial approximation for e x near x = 2 accurate, to ±0.005., Solution From Taylor’s theorem:, ex =, , N, X, e2, ez, (x − 2)n +, (x − 2 )N +1 ,, n!, (, N, +, 1, ), !, n =0, , since f (n ) (x ) = e x for all n. We are interested in x near 2, and we need to, control the size of |(x − 2)N +1 |, so we may as well specify that |x − 2| ≤ 1,, meaning x ∈ [1, 3]. Also, ez, , (N + 1) !, , ≤, , e3, , (N + 1) !, , ,

Page 133 :
taylor series, , so we need to find an N that makes e 3 /(N + 1)! ≤ 0.005. This time N = 6, makes e 3 /(N + 1)! < 0.004, so the approximating polynomial is, e x = e 2 + e 2 (x − 2) +, , e2, 2, , (x − 2)2 +, , e2, 6, , (x − 2)3 +, , e2, 24, , (x − 2)4 +, , e2, 120, , (x − 2)5 +, , e2, 720, , This presents an additional problem for approximation, since we also need to, approximate e 2 , and any approximation we use will increase the error, but we, will not pursue this complication., Note well that in these examples we found polynomials of a certain accuracy only, on a small interval, even though the series for sin x and e x converge for all x. This, is usually how things work out. To get the same accuracy on a larger interval, what, can you do? Use more terms!, , (x − 2)6 ± 0.004., , 133

Page 134 :
134, , sequences and series, , Exercises for Section 7.2, (1) Find a polynomial approximation for cos x on [0, π ], accurate to ±10−3, , à, , (2) How many terms of the series for log x centered at 1 are required so that the guaranteed, error on [1/2, 3/2] is at most 10−3 ? What if the interval is instead [1, 3/2]?, , à, , (3) Find the first three nonzero terms in the Taylor series for tan x on [−π /4, π /4], and, compute the guaranteed error term as given by Taylor’s theorem. (You may want to use, Sage or a similar aid.), , à, , (4) Prove Theorem 7.2.4, that is, show that cos x is equal to its Taylor series for all x by, showing that the limit of the error term is zero as N approaches infinity., , à, , x, , (5) Prove Theorem 7.2.5, that is, show that e is equal to its Taylor series for all x by showing, that the limit of the error term is zero as N approaches infinity., , à

Page 135 :
Review, , These problems require various techniques, and are in no particular order. Many, problems may be done in more than one way., I also encourage you to try your hand at crafting your own problems: mathematics, is more than answer-getting—it is also about problem-posing! Building your own, problems will give you a strong sense of the limitations of the machinery we have, built in this book.

Page 136 :
136, , sequences and series, , Exercises for Section 7.2, Determine whether the series converges., (1), , (2), (3), , (4), , ∞, X, , 1, 1·2, , +, , ∞, X, n!, , 3, 4, , ∞, X, , (11), , 8, , 7·8, , 12, , 1, n2, , ∞, X, n, , +, , 9, 16, , +···, , ∞, X, , à, , n!, , 1, , 1, 2·3·4, , à, , n, , +, , 2, 3·4·5, , +, , 3, 4·5·6, , (12), , ∞, X, 1 · 3 · 5 · · · (2n − 1), (2n )!, n =1, , (13), , ∞, X, 6n, n!, n =0, , (14), , à, , 1 · 3 · 5 · · · (2n − 1), n =0, , √, , à, , ∞, X, (−1)n −1, √, n =1, , à, , à, , en, , n, n =1, , à, , à, , +4, , n2, , ∞, X, , +···, , à, , 7, , −, , ∞, X, sin3 (n ), , n =0, , (10), , 5, , +, , √, , 1, , +, , 5·6, , à, , 8n, , n =0, , (9), , 1, , +, , n, , n =0, , (8), , 1, , (n 2 + 4)2, n =0, , (5) 1 −, , (7), , à, , 3·4, , ∞, X, , n =0, , (6), , n, , n2 + 4, n =0, , n, , à, , à, , +, , 4, 5·6·7, , +···, , à

Page 137 :
taylor series, , (15), , ∞, X, 2n 3n −1, n!, n =1, , (16) 1 +, (17), , ∞, X, , 52, , 54, , +, , 22, , à, , (2 · 4)2, , +, , 56, , (2 · 4 · 6)2, , +, , 58, , (2 · 4 · 6 · 8)2, , +···, , à, , à, , sin(1/n ), , n =1, , Find the interval and radius of convergence; you need not check the endpoints of the intervals., (18), , (19), , ∞, X, 2n n, x, n!, n =0, ∞, X, n =0, , (20), , (21) x +, , (24), , à, , n3n, 1 x3, 2 3, n2, , +, , xn, , ∞, X, (−1)n, n =1, , à, , 1 + 3n, , ∞, X, n!, n =1, , (23), , xn, , ∞, X, xn, n =1, , (22), , à, , n 2 3n, , 1 · 3 x5, 2·4 5, , +, , 1 · 3 · 5 x7, 2·4·6 7, , +···, , à, , à, x 2n, , ∞, X, (x − 1)n, n!, n =0, , à, à, , Find a series for each function, using the formula for Maclaurin series and algebraic, manipulation as appropriate., (25) 2x, , à, , à, ln, à, 1−x, √, 1+x à, 1, 1 + x2 à, arctan(x ) à, , (26) ln(1 + x ), (27), (28), (29), (30), , 1 + x , , (31) Use the answer to the previous problem to discover a series for a well-known mathematical, constant., , à, , 137

Page 139 :
Epilogue (or. . . what happens to Harry?), , The worst part about reading a great3 novel is that last page. The book gets thinner, , 3, , Or even a bad novel., , 4, , Or even., , and thinner, and then, poof! Not just the characters, but the whole world that the, author has crafted for them is gone! And how frequently I want to stay in that world, just a bit longer., Humanity has found an antidote to novel-endings; this antidote is the sequel., With a name like “Calculus Two” you might be getting the idea that there is a, Calculus Three, and who knows. . . maybe!, A-hundred-and-some-odd4 -pages ago, I pointed out that reading mathematics, is not the same as reading a novel. This book is done, but you are not. There is, more mathematics yet to learn about, and more mathematics yet to create. And, I don’t mean to say that you should write go and write fan fiction. I’m no author,, and you are no mere reader. You have worked through the exercises, you have, thought about this material in your own way—so you are the author of your own, understanding, and you must keep writing., , —the so-called “author”

Page 141 :
Answers to Exercises, , Answers for 1.8, 1. 1, , 3. 0, , 4. 1, , 5. 1, , 6. 0, , Answers for 2.7, 1., , lim n 2 /(2n 2 + 1) = 1/2, , n →∞, , ∞, X, 1, , 3, , n =1, , n, , 4. −3/2, , diverges, , 2., , lim 5/(21/n + 14) = 1/3, , n →∞, , 5. 11, , 3., , ∞, X, 1, n =1, , 6. 20, , 7. 3/4, , 8. 3/2, , n, , diverges, so, , 9. 3/10, , Answers for 3.1, 1. 1, , 2. 1, , 3. converges, , 4. converges, , 5. converges, , 6. diverges, , Answers for 3.2, 1. diverges, 2. diverges, 3. converges, 4. converges, 5. converges, 6. converges, 7. diverges, 8. converges, 9. N = 5, 10. N = 10, 11. N = 1687, 12. any integer greater than e 200, , Answers for 3.3, 1. converges, 7. converges, , 2. converges, 8. diverges, , 3. converges, 9. converges, , 4. diverges, 10. diverges, , 5. diverges, , 6. diverges

Page 142 :
142, , sequences and series, , Answers for 3.4, Answers for 4.1, 1. converges absolutely, 2. diverges, absolutely, 5. converges conditionally, converges conditionally, , 3. converges conditionally, 4. converges, 6. converges absolutely, 7. diverges, 8., , Answers for 4.2, 1. converges, , 2. converges, , 3. diverges, , 5. 0.90, , 4. converges, , 6. 0.95, , Answers for 6.3, 2. R = ∞, I = (−∞, ∞), 3. R = e, I = (−e , e ), 4. R = e,, 1. R = 1, I = (−1, 1), I = (2 − e , 2 + e ), 5. R = 0, converges only when x = 2, 6. R = 1, I = [−6, −4], 7., There are many choices—for instance, see Exercise 5—but, , ∞, X, , n ! · x n works., , n =0, , Answers for 6.4, , ∞, X, (n + 1)(n + 2), , 2, , n =0, , ∞, X, , 2., , 1. the alternating harmonic series, , (n + 1)x n, , ∞, X, , 3., , n =0, , xn, R = 1, , 5. C +, , (n + 1)(n + 2)x n, , 4., , n =0, , ∞, X, , −1, x n +2, (, n, +, 1, )(, n, +, 2, ), n =0, , Answers for 7.1, 1., , ∞, X, , (−1)n x 2n /(2n )!, R = ∞, , n =0, , 5, ∞, X, , 4., , ∞, X, , ∞, X, , x n /n !, R = ∞, , n =0, , (−1)n −1, , (x − 1)n, , n =1, , n, , , R = 1, , n =0, , 8. x + x 3 /3, , 5., , log(2) +, , 3., , ∞, X, , ∞, X, , (−1)n, , n =0, , (−1)n −1, , n =1, , (x − 2)n, n2n, , (x − 5)n, 5n +1, , , R =, , , R = 2, , 6., , ∞, ∞, X, X, 1 · 3 · 5 · · · (2n − 1) n, (2n − 1)!, x, =, 1, +, xn,, n, 2n, −, 1 (n − 1) ! n !, n !2, 2, n =1, n =1, ∞, ∞, X, X, n 4n +1, n n +1, 9., ( −1 ) x, /(2n )!, 10., ( −1 ) x, /n !, , (−1)n (n + 1)(x − 1)n , R = 1, , R=1, , 2., , n =0, , 7. 1 +, , n =0

Page 143 :
taylor series, , Answers for 7.2, 1. 1 −, , x2, 2, , +, , x4, 24, , −, , x6, 720, , x 12, , +···+, , 12!, , 3. x +, , 2. 1000; 8, , x3, 3, , +, , 2x 5, 15, , , error ±1.27., , Answers for 7.2, 1. diverges, 2. converges, 3. converges, 4. diverges, 5. diverges, 6. diverges, 7. converges, 8. converges, 9. converges, 10. converges, 11. converges, 12. converges, 13. converges, 14. converges, 15. converges, 16., converges, 17. diverges, 18. (−∞, ∞), 19. (−3, 3), 20. (−3, 3), 21. (−1, 1), √ √, 22. radius is 0—it converges only when x = 0, 23. (− 3, 3), 24. (−∞, ∞), ∞, ∞, ∞, X ( −1 ) n, X, X (ln(2))n, 2, 25., xn, 26., x n +1, 27., x 2n +1, 28. 1 + x /2 +, n!, n+1, 2n + 1, ∞, X, n =2, , ∞, X, , n =0, , n =0, , n =0, , 1 · 3 · 5 · · · (2n, ( −1 ) n + 1, 2n n !, , ( −1 ) n, , 4, 2n + 1, , − 3), , x, , n, , 29., , ∞, X, n =0, , n =0, , n 2n, , (−1) x, , 30., , ∞, X, (−1)n 2n +1, x, 2n + 1, n =0, , 31. π =, , 143

Page 147 :
Index, , p-series, 63, alternating harmonic series, 93, arithmetic progression, 18, associativity, 47, , infinite series, see series, integral test, 76, interval of convergence, 109, limit comparison test, 101, limit of a sequence, 23, , Basel problem, 62, Maclaurin series, 119, Cauchy condensation test, 59, chiastic rule, 48, commutativity, 47, comparison test, 57, limit, 101, condensed series, 59, convergent sequence, 23, convergent series, 40, distributive law, 45, divergence test, 52, divergent sequence, 23, divergent series, 40, geometric series, 41, geometric progression, 19, harmonic series, 54, alternating, 93, , p-series, 76, partial sums, 39, power series, 106, radius of convergence, 109, sequence, 12, 14, bounded, 34, bounded above, 34, bounded below, 34, convergent, 23, decreasing, 33, divergent, 23, equality, 16, increasing, 33, index, 14, initial, 14, monotonic, 33, non-decreasing, 33, , non-increasing, 33, of partial sums, 39, subsequence, 26, term, 14, series, 12, p-series, 76, absolute convergence, 90, alternating harmonic, 93, condensed, 59, conditional convergence, 91, convergent, 40, divergent, 40, geometric, 41, harmonic, 54, integral test, 76, interval of convergence, 109, Maclaurin, 119, radius of convergence, 109, Taylor, 123, telescoping, 50, value of, 40, subsequence, 26, Taylor series, 123, telescoping series, 50, value of series, 40