Ana Alves de Sá · Bento Louro

Sequences
and Series
Theory and Practice
Sequences and Series
Ana Alves de Sá • Bento Louro

Sequences and Series

Theory and Practice

Ana Alves de Sá Bento Louro
Department of Mathematics Department of Mathematics
Nova School of Science and Nova School of Science and
Technology-NOVA University Lisbon Technology-NOVA University Lisbon
(Faculdade de Ciências e (Faculdade de Ciências e
Tecnologia-Universidade Nova de Tecnologia-Universidade Nova de
Lisboa) Lisboa)
Lisboa, Portugal Lisboa, Portugal

ISBN 978-3-031-67201-9 ISBN 978-3-031-67202-6 (eBook)

https://doi.org/10.1007/978-3-031-67202-6

Mathematics Subject Classification: 97I30, 40A05, 40, 40A30, 40D15, 40C15

The translation was done with the help of an artificial intelligence machine translation
tool. A subsequent human revision was done primarily in terms of content.

Translation from the Portuguese language edition: “Sucessões e Séries: Teoria e Prática”
by Ana Alves de Sá and Bento Louro, © Ana Maria de Souza Alves de Sá and Bento
José Carrilho Miguens Louro 2014. Published by Escolar Editora. All Rights Reserved.

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer
Nature Switzerland AG 2024

This work is subject to copyright. All rights are solely and exclusively licensed by the
Publisher, whether the whole or part of the material is concerned, specifically the rights of
reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in
any other physical way, and transmission or information storage and retrieval, electronic
adaptation, computer software, or by similar or dissimilar methodology now known or
hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc.
in this publication does not imply, even in the absence of a specific statement, that such
names are exempt from the relevant protective laws and regulations and therefore free
for general use.
The publisher, the authors and the editors are safe to assume that the advice and in-
formation in this book are believed to be true and accurate at the date of publication.
Neither the publisher nor the authors or the editors give a warranty, expressed or implied,
with respect to the material contained herein or for any errors or omissions that may have
been made. The publisher remains neutral with regard to jurisdictional claims in published
maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland
AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

If disposing of this product, please recycle the paper.

Preface

This book is primarily intended for students of Science and Engineering.

Although the topics were not always addressed on the same subject, we
decided to group them here due to their scientific coherence. We have
strived to create a self-contained text, presenting theoretical exposition,
illustrated with various examples, solved exercises of different natures and
degrees of difficulty, and proposed exercises. Thus, the readers can test their
knowledge and level of understanding of the material.

We rigorously present the statements of definitions and results without delv-

ing too much into the theoretical exposition and without demonstrating a
large part of the Theorems. Many examples highlight aspects that could go
unnoticed by readers, mainly those inexperienced in reading mathematical
texts, for whom this book is primarily intended.

Any reader who has completed High School should be able to comprehend
Chap. 1, and the same holds true for Chap. 2, with the possible exception
of the Integral Test. For those unfamiliar with integral calculus, skipping
this section will not detract from their understanding of the rest of the
chapter. To understand Chap. 3, the reader must have basic knowledge of
differential and integral calculus. We go beyond the typical introduction
to function series by delving into the study of Fourier series. This topic
will prove to be especially useful when studying Differential Equations, but
no additional knowledge is required beyond what is already necessary for a
general understanding of function series.

At the end of each chapter, there are two lists of exercises: one titled ”Solved
Exercises” followed by the respective solutions, and another titled ”Proposed
Exercises,” which are intended for the reader’s practice and whose answers
can be found at the end of the book.

v
vi

We strongly encourage students to attempt the “Solved Exercises” on their

own before referring to the provided solution. It is important to note that the
written solution does not include the reasoning, drafts, or false starts made
in arriving at this point. By comparing their solution to the one presented in
the book, readers can more accurately assess their understanding. Reviewing
the solution without first attempting the problem independently may not
effectively facilitate learning.

The exercises featured in this book were used during practical classes and
exams for the subjects we taught. We extend our sincere appreciation to the
students whose inquiries aided in enhancing this publication, as well as to
our fellow educators who assisted us in teaching Calculus courses at Nova
School of Science & Technology. Their thorough review of the text and
copious suggestions were invaluable contributions.

In the margins of the text, we have provided concise biographies of the

mentioned mathematicians. For those interested in learning more about
these individuals, among many others, we suggest consulting the following
web page: https://mathshistory.st-andrews.ac.uk/Biographies/.

Despite our efforts toward meticulous writing and thorough proofreading,

we know some typos persisted. We extend our apologies to our readership
for any such errors and express our gratitude in advance for their assistance
in identifying and correcting them for future editions.

Lisboa, Portugal Ana Alves de Sá

Lisboa, Portugal Bento Louro
2024
Notes to the Reader

This section consists of a list of notations. They are quite common. Our aim is to avoid ambiguity and
to help the reader who, here or there, may be used to a different notation.

N = {1, 2, 3, 4, . . . }: the set of natural numbers

.

N0 = N ∪ {0}
.

Z = {. . . , −4, −3, −2, −1, 0, 1, 2, 3, 4, . . . }: the set of integers

.

Q = {x ∈ R : x = p/q, p, q ∈ Z, q = 0}: the set of rationals

.

R : the set of real numbers (remark that .N ⊂ Z ⊂ Q ⊂ R)

.

R+ = {x ∈ R : x > 0} : the set of positive real numbers

.

R+
.0 = R ∪ {0}
+

triangle inequality: .|x + y| ≤ |x| + |y|, ∀x, y ∈ R

intervals on .R: (suppose .a, b ∈ R, with .a < b) are the following sets:

[a, b] = {x ∈ R : a ≤ x ≤ b}; ] − ∞, b] = {x ∈ R : x ≤ b};

]a, b[ = {x ∈ R : a < x < b}; ] − ∞, b[ = {x ∈ R : x < b};

.
[a, b[ = {x ∈ R : a ≤ x < b}; [a, +∞[ = {x ∈ R : a ≤ x};
]a, b] = {x ∈ R : a < x ≤ b}; ]a, +∞[ = {x ∈ R : a < x};
[a, a] = {a}; ] − ∞, +∞[ = R

ε neighborhood of a: is the set .Vε (x) = ]a − ε, a + ε[ (where .a ∈ R and .ε > 0)

.

V is a neighborhood of a: there is an .ε > 0 such that .Vε (a) ⊂ V

vii
viii


n
. ai = an , .an ∈ R and .n ∈ N
i=n
n
. ai = am + am+1 + · · · + an , .ai ∈ R, i = m, m + 1, . . . , n and .m, n ∈ N with .n > m
i=m
factorial of .n ∈ N0 : .0! = 1, 1! = 1 and .n! = n × (n − 1)!
(in a suggestive way, .n! = n × (n − 1) × · · · × 3 × 2 × 1)
 
n n!
number of p-combinations of n distinct objects (.n, p ∈ N0 , .n ≥ p): . =
p p! × (n − p)!
n  
 n k n−k
Newton Binomial: if .a, b ∈ R, .n ∈ N then .(a + b)n = a b
k
k=0
Method of Mathematical Induction: to prove that a proposition .P (n) is true for all .n ∈ N
proceed as follows:
1) prove that .P (1) is true
2) prove hereditary, that is, assume that the Induction Hypothesis .P (n) is true, and then
prove the Induction Thesis .P (n + 1) is true
integer part of .x ∈ R+ : .Int(x), the largest integer less than or equal to x

upper bound of .X ⊂ R: a real number z such that .x ≤ z, ∀x ∈ X

X ⊂ R is bounded from above: there exists an upper bound of X

.

supremum of .X ⊂ R: a real number w such that w is an upper bound of X and .w ≤ z for all
z upper bound of X; we write .w = sup(X); if X is bounded from above, .sup(X) exists
maximum of .X ⊂ R: if .sup(X) ∈ X it is called the maximum of X and is denoted by .max(X)

lower bound of .X ⊂ R: a real number y such that .y ≤ x, ∀x ∈ X

X ⊂ R is bounded from below: there exists a lower bound of X

.

infimum of .X ⊂ R: a real number u such that u is a lower bound of X and .u ≥ y for all y
lower bound of X; we write .u = inf(X); if X is bounded from below, .inf(X) exists

minimum of .X ⊂ R: if .inf(X) ∈ X it is called the minimum of X and is denoted by .min(X)

X ⊂ R is bounded: X is bounded from below and from above

.

a function of the set X into the set Y is a relation between X and Y such that, to each
element .x ∈ X corresponds an unique element .y ∈ Y , and we will write .f : X → Y . X is the
domain of f and the set .{y ∈ Y : ∃x ∈ X, y = f (x)} is the range of f
 ix

if .a ∈ R+ , .f : [−a, a] → R is an odd function if .f (x) = −f (−x), ∀x ∈ [−a, a] and is an even

function if .f (x) = f (−x), ∀x ∈ [−a, a]

function of class .C 1 on .X ⊂ R: function with continuous derivative on X

function of class .C p on .X ⊂ R: function with all continuous derivatives until order p on X

0 ∞
indeterminate forms: . , , ∞ − ∞, 0 × ∞, 1∞ , 00 ; ∞0
0 ∞
L’Hôpital’s Rule: Let I be an interval, .a ∈ I, f and g two differentiable functions on I such
that:

a) .g  (x) = 0, ∀x ∈ I \ {a}.

b) . lim f (x) = lim g(x) = 0 or . lim f (x) = lim g(x) = +∞.

x→a x→a x→a x→a

f  (x) f  (x) f (x)

If there exists (real or .∞) . lim 
, then . lim  = lim .
x→a g (x) x→a g (x) x→a g(x)

L’Hôpital’s Rule (another version): Let .b ∈ R, f and g two differentiable functions on .]b, +∞[
such that:

a) .g  (x) = 0, ∀x ∈ ]b, +∞[.

b) . lim f (x) = lim g(x) = 0 or . lim f (x) = lim g(x) = +∞.

x→+∞ x→+∞ x→+∞ x→+∞

f  (x) f  (x) f (x)

If there exists (real or .∞) . lim , then . lim = lim .
x→+∞ g  (x) x→+∞ g  (x) x→+∞ g(x)
Contents

Preface v

Notes to the Reader vii

1 Sequences of Real Numbers 1

1.1 Sequences of Real Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Solved Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
1.3 Proposed Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

2 Numerical Series 67
2.1 Generalization of the Addition Operation . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.2 Definition of Series: Convergence – General Properties . . . . . . . . . . . . . . . . . 69
2.3 Alternating Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
2.4 Absolute Convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
2.5 Series of Nonnegative Terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
2.6 Products of Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
2.7 Solved Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
2.8 Proposed Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161

3 Series of Functions 167

3.1 Introduction: Sequences of Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
3.2 Pointwise and Uniform Convergence of Series of Functions . . . . . . . . . . . . . . . 172
3.3 Power Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
3.4 Taylor Series and Maclaurin Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
3.5 Introduction to Fourier Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
3.6 Solved Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
3.7 Proposed Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320

x
Contents xi

Answers to Proposed Exercises 327

Bibliography 337

Index 339
 Sequences of Real
Numbers 1

In everyday language, the word “sequence” refers to a collection of objects

or events arranged in a specific order. For instance, the daily stock quotes
of a particular company on the stock exchange market can be considered a
sequence. In this chapter, we shall explore sequences of real numbers and
establish precise definitions for the terms “sequence” and “convergence.”

1.1 Sequences of Real Numbers

Definition 1.1.1 A sequence of real numbers is a function of N into R.

Note: The elements of the range of a sequence u are called terms of

the sequence, and they are usually denoted by un , instead of the notation
u(n) which is generally used for functions. A sequence u is represented by
(un )n∈N or, more simply, (un ). As a function, its graph is the set formed
by ordered pairs of the form (n, un ), n ∈ N. For example, in Fig. 1.1 is
represented the graph of the sequence un = n/(n + 1). Of course, it is only
possible to represent part of a sequence graph.

Definition 1.1.2 The designatory expression that defines the sequence is

called general term of the sequence.

Example 1.1.1 The sequences of general terms an = n2 and bn = cos(n) Figure 1.1: The graph of the
sequence un = n/(n + 1)
are illustrated in Fig. 1.2.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 1

A. Alves de Sá, B. Louro, Sequences and Series,
https://doi.org/10.1007/978-3-031-67202-6 1
2 Sequences of Real Numbers

Notes:

1. We can define a sequence without specifying a general term. This is

the case for sequences defined recursively. A recursive relation is an
expression that relates the nth element of a sequence to the previous
element or elements. For example,
Leonardo Fibonacci, also
u1 = 1, u2 = 2, un+2 = un+1 + un , ∀n ∈ N (Fibonacci sequence).
known as Leonardo of Pisa
(1170–1250), was born in In this sequence, we obtain the term un as the sum of the two im-
Pisa, Italy and received his mediately preceding terms: u3 = u2 + u1 = 3, u4 = u3 + u2 = 5,
education in North Africa.
He is primarily known u5 = u4 + u3 = 8, . . .
for his work on number
theory and algebra. There 2. Sometimes only a few terms of the sequence are given, leading the
are copies of some of his reader to “infer” the rest, for example,
works, one of the most
important being the Liber 1, 1, 2, 1, 2, 3, 1, 2, 3, 4, . . .
.
Abaci, which introduced
Arabic numerals to Eu-
rope. It is also in this 3. There are sequences that are not defined for a finite number of values
1
book that the famous Fi- of n ∈ N. For instance, the sequence with general term un =
bonacci sequence appears: n−3
1, 1, 2, 3, 5, 8, 13, 21, 34, ..., is only defined for n = 3.
where each number is the
sum of the two previous
numbers. (Source of Definition 1.1.3 A sequence is bounded from above (or bounded
image: I benefattori
dell’umanità; vol. VI,
above) if the set of its terms has an upper bound; it is bounded from
Firenze, Ducci, 1850) below (or bounded below) if the set of its terms has a lower bound; and
it is bounded if the set of its terms is bounded, that is, if the set has an
upper bound and a lower bound.

Example 1.1.2 The sequence an = n2 is bounded below but not above

(see Fig. 1.2). Specifically, x = 1 is a lower bound; however, there are no
upper bounds since there is no number L such that n2 ≤ L, ∀n ∈ N.

Example 1.1.3 The sequence bn = cos(n) is bounded (see Fig. 1.2). It is

enough to note that −1 ≤ cos(x) ≤ 1, ∀x ∈ R.

Example 1.1.4 The sequence cn = −n is bounded above but not below

(see Fig. 1.3). In fact, x = −1 is an upper bound, but there are no lower
bounds because there is no number L such that −n ≥ L, ∀n ∈ N.
1.1. Sequences of Real Numbers 3

(a) The graph of an = n2 (b) The graph of bn = cos(n)

Figure 1.2: A bounded below sequence (a) and a bounded sequence (b)

Example 1.1.5 The sequence dn = (−1)n n is not bounded above or be-

low (see Fig. 1.3).

(a) The graph of cn = −n (b) The graph of dn = (−1)n n

Figure 1.3: A bounded above and unbounded below sequence (a) and a sequence not
bounded above or below (b)

Theorem 1.1.1 A sequence u is bounded if and only if there exists

M ∈ R+ such that |un | ≤ M , ∀n ∈ N.

Proof: Let u be a bounded sequence. According to Definition 1.1.3, u

has upper and lower bounds. Suppose α is a lower bound and β is an
upper bound. This means that α ≤ un ≤ β, ∀n ∈ N. Let us consider
M = max (|α|, |β|). Then

|un | ≤ M, ∀n ∈ N.
.
4 Sequences of Real Numbers

Conversely, suppose there exists M ∈ R+ such that

|un | ≤ M, ∀n ∈ N.
.

We have −M ≤ un ≤ M , ∀n ∈ N. Thus, −M is a lower bound and M

is an upper bound of the sequence. By Definition 1.1.3, u is a bounded
sequence.

n+2
Example 1.1.6 The sequence un = (−1)n is bounded (see Fig. 1.4):
n
 
 
n n + 2 2

.|un | = (−1) = 1 + ≤ 3, ∀n ∈ N.
 n  n

Figure 1.4: A bounded sequence

Definition 1.1.4 Given two sequences of real numbers, u and v, we call

sum, difference, and product of u and v to the sequences u + v, u − v,
and uv with general terms, respectively, un + vn , un − vn , and un vn .
If vn = 0, ∀n ∈ N, we define the quotient sequence of u and v as the
sequence u/v with general term un /vn .

Definition 1.1.5 A sequence u is increasing if un ≤ un+1 , ∀n ∈ N;

it is strictly increasing if un < un+1 , ∀n ∈ N; it is decreasing if
un ≥ un+1 , ∀n ∈ N; it is strictly decreasing if un > un+1 , ∀n ∈ N; it
is monotonic if it is increasing or decreasing; and it is strictly monotonic
if it is strictly increasing or decreasing.
1.1. Sequences of Real Numbers 5

2n
Example 1.1.7 The sequence un = is strictly increasing. We can
3n + 7
prove this by computing and simplifying the expression un+1 − un :

2(n + 1) 2n (2n + 2)(3n + 7) − 2n(3n + 10)

un+1 − un = − =
3(n + 1) + 7 3n + 7 (3n + 10)(3n + 7)
.
14
= > 0, ∀n ∈ N.
(3n + 10)(3n + 7)

5n
Example 1.1.8 The sequence un = is strictly decreasing. In fact,
n2 + 1
5(n + 1) 5n
un+1 − un = −
(n + 1)2 + 1 n2 + 1
5(n + 1)(n2 + 1) − 5n(n2 + 2n + 2)
. =
(n2 + 1)(n2 + 2n + 2)
5(n2 + n − 1)
= − < 0, ∀n ∈ N.
(n2 + 1)(n2 + 2n + 2)

Example 1.1.9 The sequence un = n2 is strictly increasing because

u
. n+1 − un = (n + 1)2 − n2 = n2 + 2n + 1 − n2 = 2n + 1 > 0, ∀n ∈ N.

Example 1.1.10 The sequence un = −n strictly decreases because

u
. n+1 − un = −(n + 1) + n = −n − 1 + n = −1 < 0, ∀n ∈ N.

Example 1.1.11 The sequence un = (−n)n is not monotonic. Indeed,

 n+1  
.un+1 − un = − (n + 1) − (−n)n = (−1)n+1 (n + 1)n+1 + nn ,
and this difference is positive if n is odd and negative if n is even.

n + (−1)n
Example 1.1.12 The sequence un = is not monotonic (see
n2
Fig. 1.5).
⎧
⎪
⎪ n+1−1 n+1 n3 − (n + 1)3
⎪
⎨ (n + 1)2 − = < 0, if n is even
n2 n2 (n + 1)2
.un+1 − un =
⎪
⎪ n+1+1 n−1 n2 + n + 1
⎪
⎩ − = 2 > 0, if n is odd.
(n + 1) 2 n2 n (n + 1)2
6 Sequences of Real Numbers

n + (−1)n
Figure 1.5: The sequence un =
n2

• Arithmetic and Geometric Sequences

We will now discuss two distinct types of sequences that hold significant
importance due to their properties: arithmetic sequences (also known as
arithmetic progressions) and geometric sequences (also known as geometric
progressions).

An arithmetic sequence .(an ) is a sequence defined by the recursive rela-

tion:
a1 = k,
.
an+1 = an + r, ∀n ∈ N,

where .k, r ∈ R, and r is called the common difference of the arithmetic

sequence.

According to this expression, it is easy to conclude that the arithmetic se-

quence is:

• Strictly increasing if .r > 0

• Strictly decreasing if .r < 0

• Constant if .r = 0

As an example, .1, 4, 7, 10, 13, . . . is an arithmetic sequence with .r = 3.

Let .(an ) be an arithmetic sequence with common difference r. Then

1.1. Sequences of Real Numbers 7

a2 = a1 + r
a3 = a2 + r = a1 + 2r
a4 = a3 + r = a1 + 3r
. ..
.
an = an−1 + r = a1 + (n − 1) r
..
.
The general term of the arithmetic sequence with common difference r
is given by the expression

a = a1 + (n − 1) r, ∀n ∈ N.
. n

At first glance, the sum of the first terms of an arithmetic sequence may
seem like a time-consuming problem to solve.
However, according to legend, Gauss, one of the greatest mathematicians of
all time, solved the question of calculating the sum of the first 100 natural
numbers in a few minutes when he was still a child. What reasoning would Carl Friedrich Gauss
(1777–1855) was a Ger-
have allowed him to arrive so quickly at the result?
man mathematician. He
If we write the integers from 1 to 100, we find that: worked in a wide variety of
areas in both Mathematics
and Physics. He was a
• The sum of the first and last numbers is 101. gifted child. His first
mathematical discoveries
• The sum of the second and penultimate numbers is 101. were made in his teens.
His magnum opus, Disqui-
• The sum of the third and third-to-last numbers is 101, and so on. sitiones Arithmeticae, was
completed in 1798, when
There are precisely 50 pairs of numbers whose sum is a constant of value Gauss was only 21 years
old. (Source of image:
101, so it is concluded that the sum of the first 100 natural numbers is Oil portrait by Gottlieb
.101 × 50 = 5050. Wilhelm Emil Biermann)

In general, .Sn , the sum of the first n consecutive terms of an arithmetic

sequence .(an ) with common difference r, is given by the formula

a1 + an
Sn =
. × n.
2
Let us prove this using the method of induction.
For .n = 1, the equality is trivial. The property is proven if it is hereditary.
If we admit that equality is valid for n, then
8 Sequences of Real Numbers

a1 + an
Sn+1 = Sn + an+1 = × n + an+1
2
a1 + a1 + (n − 1)r
= × n + a1 + nr
2
(n − 1)r
= n a1 + × n + a1 + nr
2
n−1
= (n + 1)a1 + + 1 × nr
. 2
n+1
= (n + 1)a1 + × nr
2
nr
= (n + 1) a1 +
2
a1 + a1 + n r
= (n + 1)
2
a1 + an+1
= (n + 1) ,
2
which confirms that the equality holds for .n + 1.

A geometric sequence .(an ) is the sequence defined by the recursive relation

a1 = k
.
an+1 = an × r, ∀n ∈ N,

where .k, r ∈ R\{0}, and r is known as the common ratio of the geometric
sequence.
1 1 1 1
For example, . , , , . . . is a geometric sequence with .r = .
10 100 1000 10
Similar to arithmetic sequences, it is possible to calculate any term of the
sequence knowing the first term and the common ratio. Considering the
definition of a geometric sequence with common ratio r, we obtain the
following:
a2 = a1 × r
a 3 = a2 × r = a 1 × r 2
a4 = a3 × r = a 1 × r 3
.
..
.
an = a1 × rn−1
..
.
1.1. Sequences of Real Numbers 9

The general term of a geometric sequence with ratio r is expressed as

follows:
.an = a1 × r , ∀n ∈ N.
n−1

According to this expression, it is easy to conclude that the monotonicity of

a geometric sequence is:

• Increasing if .a1 > 0 and .r > 1 or if .a1 < 0 and .0 < r < 1
• Decreasing if .a1 > 0 and .0 < r < 1 or if .a1 < 0 and .r > 1

• Constant if .r = 1

• Not monotonic if .r < 0

There is a well-known legend that a Hindu prince, fascinated with the chess
game, decided to reward its inventor, who asked for this reward to be given
in wheat, receiving one grain for the first square of the chessboard, two
grains for the second square, four grains for the third, and so on, doubling
the number of grains from the last square considered. We conclude that the
number of wheat grains is represented by the sum of the first 64 terms of a
geometric sequence with common ratio 2.

S64 = 1 + 2 + 22 + 23 + . . . + 263 .
.

How can we determine the value of this expression?

We can calculate the sum in the more general case of a geometric sequence,
which involves finding the sum of the first n terms. Let

Sn = a1 + a 2 + a 3 + . . . + a n
. (1.1)

be the sum of the first n terms. Multiplying .Sn by the ratio r, we obtain

r Sn = r a1 + r a2 + r a3 + . . . + r an = a2 + a3 + a4 + . . . + an+1 . (1.2)
.

Subtracting (1.2) from (1.1), we obtain

Sn − r Sn = (a1 + a2 + a3 + . . . + an ) − (a2 + a3 + a4 + . . . + an+1 ),

.

that is,
a1 − an+1 a1 − a1 r n 1 − rn
Sn =
. = = a1 .
1−r 1−r 1−r
10 Sequences of Real Numbers

Returning to the inventor of the game of chess, the requested reward is

S64 = 264 − 1 = 18 446 744 073 709 551 615

.

wheat grains (which not even the richest prince would be able to pay).

Example 1.1.13 Let us consider Fig. 1.6, and let .(An ) be the sequence of
the areas of the semicircles delimited by the lines .C1 , .C2 , .C3 , .. . .. We show
that .(An ) is a geometric sequence and express its general term as follows:

(a) (b) (c)

Figure 1.6: Semicircles

1
Each semicircle has an area . π r2 , where r is the radius of the respective
2
circumference. In Fig. 1.6a, the radius of the circumference is r; in Fig. 1.6b,
r
each semicircle has a radius of . ; in Fig. 1.6c, each semicircle has a radius
2
r
of . ; and so on. Therefore, we can write the first terms of the sequence
4
.An .

1
A1 = π r2
2
1 r 2 r2
. A2 = π ×2=π
2 2 4
1 r 2 r2
A3 = π ×4=π .
2 4 8

1
By induction, it can be proven that .An+1 = ×An , that is, it is a geometric
2
1
sequence with common ratio . , so the general term of the sequence is
2
r2
.An = π .
2n
1.1. Sequences of Real Numbers 11

• Limits of Sequences

Definition 1.1.6 A sequence u tends to +∞ (and we write un → +∞

or lim un = +∞ or lim un = +∞), if
n→+∞

∀ L ∈ R+ ∃ p ∈ N : n > p ⇒ un > L.
.

A sequence u tends to −∞ (and we write un → −∞ or lim un = −∞

n→+∞
or lim un = −∞), if

∀ L ∈ R+ ∃ p ∈ N : n > p ⇒ un < −L.

.

Example 1.1.14 We prove that un = n2 → +∞. Given L > 0, there

√
exists p ∈ N such that p > L; if n > p, then n2 > p2 > L. It is
evident that p depends on L; for instance, if L = 100, it is sufficient to
consider p = 10, but if L = 200, we must consider p = 14. This example is
illustrated in Fig. 1.7.

Figure 1.7: The sequence un = n2 tends to +∞

Similarly, it can be shown that vn = nα , α ∈ R+ , tends to +∞, that

yn = −n tends to −∞, and that if wn = (−n)n , then |wn | = nn → +∞.

Notes:

1. Sometimes the following concept is also used: A sequence (un ) has

general infinite limit if |un | → +∞. For example, the sequence
12 Sequences of Real Numbers

un = (−1)n n has general infinite limit despite un → +∞ and

un → −∞.

2. The fact that un → +∞ does not imply that u is increasing (nor

that there is an order from which it is increasing). For instance, the
sequence un = n + (−1)n , illustrated in Fig. 1.8, tends to +∞, but
it is not monotonic.
Figure 1.8: The sequence
un = n + (−1)n
3. If u is such that un → +∞, un → −∞, or |un | → +∞, then u is
unbounded. The converse is not true. For example, the sequence
⎧
⎨n, if n is even
.un =
⎩1, if n is odd
n

is unbounded and un → +∞, un → −∞, |un | → +∞.

From Definition 1.1.6, the next result follows immediately.

Theorem 1.1.2 Let u and v be sequences such that from a certain order,
un ≤ vn . Then,

a) un → +∞ ⇒ vn → +∞.

b) vn → −∞ ⇒ un → −∞.

Example 1.1.15 Let us consider the sequence

n
1 1 1 1
u =
. n √ = 1 + √ + √ + ··· + √ .
k 2 3 n
k=1

Figure 1.9: An illustration

of Theorem 1.1.2 Because we have inequality

1 1 1 1 √
1 + √ + √ + ··· + √ ≥ n × √ = n
.
2 3 n n
√
and n → +∞, we can state that un → +∞ (see Fig. 1.9).
1.1. Sequences of Real Numbers 13

Theorem 1.1.3 Let un → +∞, vn → +∞, and wn → −∞. Then

a) lim (un + vn ) = +∞.

b) lim (un vn ) = +∞.

c) lim (un wn ) = −∞.

d) lim (un )p = +∞, ∀p ∈ N.

e) lim |un | = lim |vn | = lim |wn | = +∞.

Definition 1.1.7 Let u be a sequence and a ∈ R. We say that u con-

verges to a (or that tends to a or, still, that the limit of the sequence is
a), and we write un → a or lim un = a or lim un = a, if
n→+∞

∀ ε > 0 ∃ p ∈ N : n > p ⇒ |un − a| < ε.

.

We say that a sequence u is convergent if there is a ∈ R such that u

converges to a; otherwise, u is divergent.

In other words, u is convergent if for every ε > 0, we can choose p such

that from the order p all terms un are in the interval ]a − ε, a + ε[.
5n
Taking the sequence un = , we illustrate this definition in Fig. 1.10
9n + 1
for two values of ε.

1 1 1 1 1 1
(a) If ε = 80 then a − 80
< un < a + 80
(b) If ε = 500
then a − 500
< un < a + 500
whatever
whatever n > 4 n > 30
Figure 1.10: The value of p varies with the value of ε
14 Sequences of Real Numbers

5
The limit of this sequence is a = . To prove this fact using the definition,
9
we proceed as follows: Let ε > 0; we must show that there exists a natural
number, p, such that
 
 5n 5 
.if n > p then  − < ε.
9n + 1 9 

Note that    
 5n 5   −5  5
 − 
.
 9n + 1 9   9(9n + 1)  = 9(9n + 1) .
=

Solving inequality
5
. <ε
9(9n + 1)
5 − 9ε
for n, we obtain n > .
81ε
5 − 9ε
If we choose a natural number p greater than or equal to , then
81ε
 
5 − 9ε  5n 5
.n > p ⇒ n > ⇒  −  < ε.
81ε 9n + 1 9

5 5
Note that if ε ≥ , we can choose p = 1. If ε < , we can choose
9 9
5 − 9ε 5 − 9ε
p = Int . This number p = Int is the smallest number
81ε 81ε
that makes the statement true.

1
Example 1.1.16 Let us prove that un = → 0. Given an arbitrary ε > 0,
n
we must show that there exists a natural number, p, such that
 
1 1
.if n > p then   = < ε.
n n

1
Let p be a natural number greater than or equal to . Then, for n > p, we
ε
1 1 1
have < < ε. If we choose p = Int , this will be the smallest order
n p ε
from which the proposition is true for a given value of ε (see Fig. 1.11).
1
Similarly, we can prove that the sequence vn = α , α ∈ R+ , converges to
n
zero.
1.1. Sequences of Real Numbers 15

Figure 1.11: If ε = 0.1, then −ε < 1

n
< ε if n > 10

Notes:

1. In the language of neighborhoods, Definition 1.1.7 is equivalent to

∀ ε > 0 ∃ p ∈ N : n > p ⇒ un ∈ Vε (a).

.

2. We could also write equivalently

∀ ε > 0 ∃ p ∈ N : |un − a| < ε, ∀n > p.

.

3. Consider the set R = R ∪ {−∞, +∞}, where −∞ and +∞ are two

distinct mathematical objects that are not real numbers. We can
introduce an order relation in this set as follows:

i) If x, y ∈ R, x < y on R if and only if x < y on R.

ii) −∞ < x < +∞, ∀x ∈ R.

The set R, with this order relation, is called extended real line.
We can extend the notion of neighborhood to R. Let ε > 0. If a ∈ R,
we call the ε neighborhood of a the set Vε (a) = ]a − ε, a + ε[ (which
 
coincides with the neighborhood in R). The set Vε (+∞) = 1ε , +∞
 
is called the ε neighborhood of +∞. The set Vε (−∞) = −∞, − 1ε
is called the ε neighborhood of −∞. With the definitions given pre-
viously, we can unify from a formal viewpoint Definitions 1.1.6 and
1.1.7: If a ∈ R,

u → a if and only if ∀ ε > 0 ∃ p ∈ N : n > p ⇒ un ∈ Vε (a).

. n
16 Sequences of Real Numbers

Theorem 1.1.4 (Uniqueness of the limit) Let a, b ∈ R. If un → a and

un → b, then a = b.

Proof: (By contradiction)

|b − a|
If a = b, we can consider ε = . By the definition of a convergent
3
sequence,
.∃ p ∈ N : n > p ⇒ |un − a| < ε

and
.∃ q ∈ N : n > q ⇒ |un − b| < ε.
Let m > max{p, q}. Then,

|b − a| |b − um + um − a| |b − um | + |um − a|
ε = = ≤
.
3 3 3
ε+ε 2ε
< = <ε
3 3
and we conclude that ε < ε, which is impossible.

Theorem 1.1.5 If u and v are convergent sequences, then:

a) lim(un + vn ) = lim un + lim vn .

b) lim(un · vn ) = lim un · lim vn .

c) lim(un )p = (lim un )p , p ∈ N.
un lim un
d) lim = , if vn = 0, ∀n ∈ N, and lim vn = 0.
vn lim vn
e) lim(un )1/p = (lim un )1/p (if p is even, it must be un ≥ 0, ∀n ∈ N).
 
f) lim (un )vn = (lim un )lim vn , if un > 0, ∀n ∈ N, and the limits of
the sequences are not both zero.

g) lim |un | = | lim un |.

h) (∃ p ∈ N ∀n ≥ p, un ≥ 0) ⇒ lim un ≥ 0.

i) (∃ p ∈ N ∀n ≥ p, un ≥ vn ) ⇒ lim un ≥ lim vn .

Proof: We will prove only items a) and g).

1.1. Sequences of Real Numbers 17

a) Let a and b be the limits of sequences (un ) and (vn ), respectively. Let
ε > 0 be arbitrary. By definition,
ε
.∃ p1 ∈ N : n > p1 ⇒ |un − a| <
2
and
ε
∃ p2 ∈ N : n > p2 ⇒ |vn − b| <
. .
2
Let p = max{p1 , p2 }; then, if n > p,

|(un + vn ) − (a + b)| = |un + vn − a − b| = |un − a + vn − b|

.
ε ε
≤ |un − a| + |vn − b| < + = ε.
2 2
We thus conclude that un + vn → a + b.

g) Let us denote a = lim un . Let ε be an arbitrary positive real number.

By definition,
.∃ p ∈ N : n > p ⇒ |un − a| < ε.

But ||un | − |a|| ≤ |un − a|, so ||un | − |a|| < ε, ∀n > p. We conclude that
|un | → |a|.

Definition 1.1.8 A sequence u is a null sequence if un → 0.

Notes:

1. It is evident from the definitions that (un ) converges to a if and only

if (un − a) is a null sequence.

2. Considering the definitions, it is easy to verify that if u is a null

sequence, |1/un | → +∞, and, reciprocally, if |vn | → +∞, then
1/vn → 0.

Theorem 1.1.6 If un → 0 and v is a bounded sequence, then

un · vn → 0.

Proof: Let M > 0 such that |vn | ≤ M, ∀n ∈ N. Given an arbitrary ε > 0,

ε
let p ∈ N, such that |un | < , ∀n > p. Then
M
|un · vn | < ε, ∀n > p.
.
18 Sequences of Real Numbers

Figure 1.12: An illustration of Theorem 1.1.6

Example 1.1.17 Let us calculate the limit of the sequence

−2 + 4 cos(n)
a =
. n .
n
We know that −1 ≤ cos(n) ≤ 1, ∀n ∈ N; therefore,

. − 6 ≤ −2 + 4 cos(n) ≤ 2, ∀n ∈ N.

In other words, the sequence bn = −2 + 4 cos(n) is bounded (see Fig. 1.12).

1
As Example 1.1.15 shows, the sequence with general term is a null se-
n
quence. By Theorem 1.1.6 we can assert that
−2 + 4 cos(n)
. lim = 0.
n

From Definition 1.1.7 it is easy to show the following theorem:

Theorem 1.1.7 Every convergent sequence is bounded.

Note: The converse is not true. For example, the sequence

u = cos(nπ) = (−1)n
. n

is bounded but does not converge.

1.1. Sequences of Real Numbers 19

Theorem 1.1.8 (Squeeze Theorem) If un → a, vn → a and, from a

certain order, un ≤ wn ≤ vn , then wn → a.

Proof: Let ε be any positive real number. Then

∃ p1 ∈ N : n > p1 ⇒ a − ε < un < a + ε,

. ∃ p2 ∈ N : n > p2 ⇒ a − ε < vn < a + ε,
∃ p3 ∈ N : n > p3 ⇒ u n ≤ w n ≤ v n .

Let p = max{p1 , p2 , p3 }. If n > p, then

a − ε < un ≤ wn ≤ vn < a + ε.
.

Figure 1.13: An illustration of the Squeeze Theorem

Example 1.1.18 Let us calculate the limit of the sequence

−2 + 4 cos(n)
a =
. n
n
from the previous example, now applying Theorem 1.1.8.
We know that −1 ≤ cos(n) ≤ 1, ∀n ∈ N, which implies that
6 −2 + 4 cos(n) 2
. − ≤ ≤ , ∀n ∈ N.
n n n
1 −2 + 4 cos(n)
Considering that → 0, we conclude that lim = 0 (see
n n
Fig. 1.13).
20 Sequences of Real Numbers

Theorem 1.1.9 Let P (x) = a0 xp +· · ·+ap and Q(x) = b0 xq +· · ·+bq be

two polynomial functions with real coefficients, p, q ∈ N, a0 = 0, b0 = 0,
and v be a sequence, vn → +∞. Then
⎧ a0
⎪
⎪ , if p = q
⎪
⎪ b0
⎪
⎪ a0
P (vn ) ⎨+∞, if p > q ∧ >0
. lim = b0
Q(vn ) ⎪ ⎪ −∞, if p > q ∧
a0
⎪
⎪ <0
⎪
⎪ b0
⎩
0, if p < q.

Proof: Let p = q. Dividing by vnp both members of the fraction

P (vn ) a0 vnp + a1 vnp−1 + · · · + ap−1 vn + ap

. = ,
Q(vn ) b0 vnp + b1 vnp−1 + · · · + bp−1 vn + bp

we obtain
a1 ap−1 ap
a0 + + · · · + p−1 + p
P (vn ) vn vn vn
. = .
Q(vn ) b1 bp−1 bp
b0 + + · · · + p−1 + p
vn vn vn
1
By hypothesis, vn → +∞; then → 0 for all s ∈ N. We conclude by
vns
Theorem 1.1.5 that
P (vn ) a0
. → .
Q(vn ) b0

Let p = q. Putting vnp in evidence in the numerator and vnq in the denomi-
nator, we obtain the equalities

P (vn ) a0 vnp + a1 vnp−1 + · · · + ap−1 vn + ap

=
Q(vn ) b0 vnq + b1 vnq−1 + · · · + bq−1 vn + bq
a1 ap−1 ap
vnp a0 + + · · · + p−1 + p
vn vn v n
=
q b1 bq−1 bq
.
v n b0 + + · · · + q−1 + q
vn vn vn
a1 ap−1 ap
a0 + + · · · + p−1 + p
v n v n v n
= vnp−q .
b1 bq−1 bq
b0 + + · · · + q−1 + q
vn vn vn
1.1. Sequences of Real Numbers 21

We saw earlier that

a1 ap−1 ap
a0 + + ··· + p−1 + v p
vn vn n a0
. → .
b1 bq−1 bq b0
b0 + + ··· + + q
vn vnq−1 vn

P (vn ) a0
If p > q, vnp−q → +∞, so the limit of will be +∞ if > 0 and
Q(vn ) b0
a0
−∞ if < 0.
b0
P (vn ) a0
If p < q, vnp−q → 0, so →0· = 0.
Q(vn ) b0

Theorem 1.1.10 Let a ∈ R. Then

⎧
⎪
⎪
⎨+∞, if a > 1
n
. lim a = 0, if |a| < 1
⎪
⎪
⎩1, if a = 1.

If a ≤ −1, the limit does not exist. If a < −1, |an | → +∞.

Proof: Let a ∈ R such that |a| > 1. We intend to prove that |a|n → +∞.
Consider an arbitrary L > 0. Since log(|a|) is positive when |a| > 1, we
have
log(L)
|a|n > L ⇔ en log(|a|) > elog(L) ⇔ n log(|a|) > log(L) ⇔ n >
. .
log(|a|)

Therefore, it is sufficient to consider in Definition 1.1.6, for example,

log(L)
p = Int log(|a|) + 1. Thus, we have shown that if |a| > 1, then
|an | → +∞.
If a > 1, |an | = an , so an → +∞. If a < −1, |an | → +∞ (note that, in
this case, an is negative when n is odd and positive when n is even).
If a = 0, it is obvious that an = 0 → 0.
Let a be such that |a| < 1, a = 0, and take an arbitrary ε > 0. As
log(|a|) < 0 because |a| < 1, we get

log(ε)
|an | = |a|n < ε ⇔ n log(|a|) < log(ε) ⇔ n >
. .
log(|a|)
22 Sequences of Real Numbers

If ε < 1, it is enough to take in Definition 1.1.7, for instance,

log(ε)
p = Int log(|a|) + 1; if ε ≥ 1, we can consider p = 1.

If a = 1, then an = 1, ∀n ∈ N, so an → 1.

If a = −1, then an = (−1)n , and this sequence does not converge.

Example 1.1.19 Let us calculate the limit of the sequence

n
4n
. .
2n + 1 n∈N

Given n ∈ N,

n n
4 4n 4n 4 4n
3n ≥ 2n + 1 ⇒
. = ≤ ⇒ ≤ .
3 3n 2n + 1 3 2n + 1

n
4
But by Theorem 1.1.10, lim = +∞. Considering the last inequality,
3
n
4n
we conclude by Theorem 1.1.2 that lim = +∞.
2n + 1

Example 1.1.20 Let us evaluate the limit of the sequence of general term

n
n
.(an )n = .
5n + 1

1
The sequence (an ) is monotonically increasing and has a limit of , so
5
n n n
1 n 1 1 n 1
. ≤ ≤ ⇒ ≤ ≤ .
6 5n + 1 5 6 5n + 1 5

n n
1 1
But by Theorem 1.1.10, lim = lim = 0. From the previous
5 6
n
n
inequalities, we conclude by Theorem 1.1.8 that lim = 0.
5n + 1
1.1. Sequences of Real Numbers 23

Theorem 1.1.11
u1 + · · · + un
a) Let a ∈ R. If un → a, then → a.
n
√
b) If b ∈ R+ , then n
b → 1.
un+1 √
c) Let c ∈ R. If un > 0, ∀n ∈ N, and → c, then n un → c.
un

Proof: a) Let a ∈ R and ε > 0 be arbitrary. We want to prove that

 
 u1 + u2 + · · · + un 
.∃ s ∈ N : n > s ⇒  − a  < ε,
 n 

that is,
 
 u1 − a + u2 − a + · · · + un − a 
.∃ s ∈ N : n > s ⇒   < ε.

n
We know that ∃ p ∈ N : n > p ⇒ |un − a| < 2ε . Additionally,
 
 u1 − a + u2 − a + · · · + up − a 
 ≤
 n 
.
|u1 − a| + |u2 − a| + · · · + |up − a| Mp
≤ ≤ ,
n n
where M = max{|u1 − a|, |u2 − a|, · · · , |up − a|}. Let q ∈ N such that
n > q ⇒ Mn p < 2ε . Let s = max{p, q}. If n > s, we have
 
 u1 − a + u2 − a + · · · + un − a 
 ≤
 n 
|u1 − a| + |u2 − a| + · · · + |up − a| |up+1 − a| + · · · + |un − a|
. ≤ +
n n
ε ε
(n − p)
≤ + 2 ≤ ε.
2 n
Thus, if a ∈ R, the statement is proved.
Now, we show the theorem for a = +∞. Let L be an arbitrary positive real
number. We know that there exists p ∈ N : n > p ⇒ un > 2 L. Given
that
u1 + u2 + · · · + up
. → 0,
n
then
u1 + u2 + · · · + up −L
.∃ q ∈ N : n > q ⇒ > .
n 2
24 Sequences of Real Numbers

Additionally,
up+1 + up+1 + · · · + un 2 L (n − p)
. > .
n n
As n−p
n → 1, ∃ r ∈ N : n > r ⇒ n−p 3
n > 4 . Let s = max{p, q, r}. If n > s,
we have
u1 + u2 + · · · + un −L 2 L (n − p) −L 3
. > + > + 2 L = L.
n 2 n 2 4
Thus, the statement is proved if a = +∞.
For a = −∞, the procedure is similar but with obvious adaptations.
√ √
b) Let b > 1. In this case n b > 1, so we can write n b = 1 + hn , where,
for each n, hn > 0. Then, by Newton’s Binomial,

n
n k
b = (1 + hn )n =
. h > 1 + n hn ,
k n
k=0

√ b − 1, from which 0 < hn < and we deduce that hn → 0,

b−1
so n hn < n ,
that is, n b → 1.
1
If b < 1, we take a = b > 1. Then

√ n 1 1 1
= √ → = 1.
n
. b=
a n
a 1
√
n
√
If b = 1, b = n 1 = 1 for all n, which is the limit of a constant sequence.

c) Let c ∈ R+ and 0 < ε < c be arbitrary. We know that

ε un+1 ε
∃p ∈ N : n > p ⇒ c −
. < <c+ .
2 un 2
Taking n > p, we obtain
ε up+2 ε ε up+3 ε ε un ε
.c− < < c+ , c− < < c+ , . . . , c− < < c+ ,
2 up+1 2 2 up+2 2 2 un−1 2
and multiplying term by term these inequalities, we obtain
ε n−p−1 un ε n−p−1
. c− < < c+ ,
2 up+1 2
which implies
up+1 ε n up+1 ε n
.   c− < un <   c+ .
ε p+1 2 ε p+1 2
c− 2 c+ 2
1.1. Sequences of Real Numbers 25

Applying roots,
 
up+1 ε √ up+1 ε
. n  p+1 c− < n un < n  p+1 c+ .
c − 2ε 2 c + 2ε 2

By item b),
 
up+1 up+1
. n  p+1 → 1 and n  p+1 → 1;
c − 2ε c + 2ε

therefore,
 
up+1 ε ε up+1 ε ε
. n  p+1 c− →c− and n  p+1 c+ →c+ .
c − 2ε 2 2 c + 2ε 2 2

As c − 2ε = c + 2ε , we cannot apply the Squeeze Theorem. However, by

applying the definition of limit, we know that there exists q ∈ N such that
if n > q, then

ε ε up+1 ε ε ε
.c − ε = c − − < n c− < c− + = c,
2 2 (c − 2ε )p+1 2 2 2
and there exists r ∈ N such that if n > r, then

ε ε up+1 ε ε ε
.c = c + − < n c+ < c+ + = c + ε;
2 2 (c + 2ε )p+1 2 2 2

taking s = max{p, q, r}, if n > s, then

 
up+1 ε √ up+1 ε
.c − ε < c− < un < n  p+1 c+ < c + ε,
n n
(c − 2ε )p+1 2 c+ ε 2
2
√
and it follows from the definition of limit that lim n un = c.
If c = 0, we follow the same reasoning, considering that un > 0 and,
√
therefore, n un > 0. We show that, from a certain order,

√ ε up+1
.0 <
n
un < n  p+1 .
2 ε
2
√
From this, we deduce that there is an order from which 0 < n u
n < ε and,
√
as before, we conclude that lim n un = 0.
Let c = +∞. Let L be an arbitrary positive number. We know that
un+1
.∃ p ∈ N : n > p ⇒ > 2L.
un
26 Sequences of Real Numbers

Taking n > p, we have

up+2 up+3 un
. > 2L, > 2L, . . . , > 2L,
up+1 up+2 un−1
and multiplying term by term these inequalities yields
un
. > (2L)n−p−1 ,
up+1
up+1
which implies un > up+1 (2L)n−p−1 = (2L)n . Applying roots,
(2L)p+1

√ up+1
.
n
un > 2L n .
(2L)p+1

up+1
By item b), n → 1; therefore,
(2L)p+1

up+1 1
.∃ q ∈ N : n > q ⇒ n > .
(2L)p+1 2
√
Let s = max{p, q}. If n > s, then n un > 2L 12 = L.

Notes:
√
1. It is easy to see, using item c) of the previous theorem, that n
n → 1.
2. The converse of item c) of the previous theorem is not true, that is,
√ un+1
n
un → c does not imply that → c (un > 0, ∀n ∈ N). For
un
example, consider the sequence un = e−n−(−1) :
n

√ (−1)n
.
n
un = e−1− n → e−1 ,
and
un+1 e−3 , if n is odd
= e−1+2(−1) =
n
.
un e, if n is even
has no limit.

Example 1.1.21 Let us calculate the limit of the sequence with general
√
term n 2n + 3n . The sequence un = 2n + 3n is positive and
n+1
2
+1
un+1 2n+1 + 3n+1 3
. lim = lim = lim n = 3.
n→+∞ un n→+∞ 2n + 3n n→+∞ 2 1 1
· +
3 3 3
1.1. Sequences of Real Numbers 27

By item c) of Theorem 1.1.11, we can conclude that

√
n
. lim 2n + 3n = 3.
n→+∞

Theorem 1.1.12 Every bounded monotonic sequence (un ) is convergent

and sup{un : n ∈ N}, if (un ) is increasing
. lim un =
inf{un : n ∈ N}, if (un ) is decreasing.

Proof: For example, suppose that the sequence (un ) is increasing. The set
of terms of the sequence is bounded; therefore, it has a supremum, which we
denote by S. Let ε > 0 be arbitrary. There is an element of the sequence,
up , such that S − ε < up ≤ S. However, the sequence is increasing;
therefore, up ≤ un , ∀n > p. We then have S − ε < un ≤ S, ∀n > p, so
un → S.

Note: The converse is not true; that is, there are convergent sequences that
1
are not monotonic. For example, the sequence un = (−1)n converges to
n
0 and is not monotonic (Fig. 1.14).

(−1)n
Figure 1.14: The sequence un =
n
28 Sequences of Real Numbers

n
1
• The Sequence 1+
n n∈N
n
1
Let us consider the sequence with general term .un = 1+ . According
n
to Newton’s Binomial, it can be written as follows:
1
n 
n
n 1
p
un = 1+ =
n p=0
p n
0 1 2 n
n 1 n 1 n 1 n 1
= + + + ··· +
0 n 1 n 2 n n n
n! 1 n! 1 n! 1
= 1+ · + · 2+ · 3+
1!(n − 1)! n 2!(n − 2)! n 3!(n − 3)! n
n! 1 n! 1
+ ··· + · + ·
(n − 1)!(n − (n − 1))! nn−1 n!(n − n)! nn
1 n! 1 1 n! 1 1 n! 1
= 1+ · · + · · 2+ · · 3+
1! (n − 1)! n 2! (n − 2)! n 3! (n − 3)! n
.
1 n! 1 1 n! 1
+ ··· + · · + · ·
(n − 1)! (n − (n − 1))! nn−1 n! (n − n)! nn
1 1 n(n − 1) 1 n(n − 1)(n − 2)
= 1+ + · 2
+ · +
1! 2! n 3! n3
1 n! 1 n!
+ ··· + · + ·
(n − 1)! nn−1 n! nn
1 1 1 1 1 2
= 1+ + 1− + 1− 1− +
1! 2! n 3! n n
1 1 2 n−1
+ ··· + 1− 1− ··· 1 − .
n! n n n
p
Since .0 < 1 − < 1, for any .1 ≤ p ≤ n − 1, we can write
n
1 1 1
.2 ≤ un ≤ 1 + 1 + + + ··· + .
2 3! n!
It is easy to prove by induction that
1 1
. < n−1 , ∀n ≥ 3,
n! 2
which allows us to write, taking into account the formula for the sum of the
terms of a geometric sequence,
 n
1 1 1 1 − 12 1
.2 ≤ un ≤ 1 + 1 + + 2 + · · · + n−1 = 1 + < 1+ = 3.
2 2 2 1 − 12 1 − 12
1.1. Sequences of Real Numbers 29

We have just proved that the sequence is bounded. Let us see that it is
monotonic:

u
. n+1 − un =

1 1 1 1 1 2
= 1+ + 1− + 1− 1− +
1! 2! n+1 3! n+1 n+1 Leonhard Euler (1707–
1 1 2 n−1 1783) was a Swiss mathe-
+ ··· + 1− 1− ··· 1 − + matician who studied phi-
n! n+1 n+1 n+1 losophy and mathemat-
1 1 2 n ics at the University of
. + 1− 1− ··· 1 − − Basel. He was a professor
(n + 1)! n+1 n+1 n+1
 in St. Petersburg (Russia)
1 1 1 1 1 2 and Berlin. His contribu-
− 1+ + 1− + 1− 1−+ tions are important in al-
1! 2! n 3! n n
 most all areas of math-
1 1 2 n−1 ematics, namely, number
+ ··· + 1− 1− ··· 1 − .
n! n n n theory, differential equa-
tions, analysis, calculus
of variations, and rational
By associating terms, we obtain mechanics, as well as in
other scientific areas, such
  as calculation of planetary
1 1 1 orbits, artillery and bal-
un+1 − un = 1− − 1− +
2! n+1 n listics, shipbuilding, and
  navigation. In 1771, he
1 1 2 1 2
+ 1− 1− − 1− 1− + became blind, but this did
3! n+1 n+1 n n not prevent him from con-
 tinuing an extraordinary
1 1 2 n−1
. +··· + 1− 1− ··· 1 − − scientific production. Af-
n! n+1 n+1 n+1 ter his death, St. Peters-
 burg Academy continued
1 2 n−1
− 1− 1− ··· 1 − + to publish his unpublished
n n n works for 50 years. Eu-
1 1 2 n ler introduced the nota-
+ 1− 1− ··· 1 − . tion .f (x) for a function,
(n + 1)! n+1 n+1 n+1 e for the base of natural
logarithms, i for the root

p p of .−1, .π, . for summa-
Since .1 − > 1− > 0 and the last term is positive, we have tion, .Δy for finite differ-
n+1 n
.un+1 − un > 0, for all .n ∈ N. The sequence
ences, and many others.
He was the most prolific
n mathematician of all time.
1
u =
. n 1+ (Source of image: Oil por-
n trait by Jakob Emanuel
Handmann (1718–1781),
is bounded and monotonic. By Theorem 1.1.12 we conclude that it is Deutsches Museum)
convergent. The limit of this sequence is an irrational number denoted by
30 Sequences of Real Numbers

the letter e: n
1
. lim 1 + = e.
n
This number is usually referred to as Neper’s number or Euler’s constant.

x n
Theorem 1.1.13 Let .x ∈ R. Then .lim 1 + = ex .
n
John Napier (or Jhone
Neper) (1550–1617) was
a Scottish mathematician.
He has dedicated himself Theorem 1.1.14 Let .x ∈ R and u be a sequence such that .lim un = +∞
to studies in theology and un
x
mathematics. He is es- or .lim un = −∞. Then .lim 1 + = ex .
sentially known as the in- un
ventor of logarithms, al-
though he also has works
on spherical triangles. He
invented a calculating rule, Example 1.1.22 We know that .xn = n + 5 → +∞. For every .n ∈ N,
known as “Napier’s bones”
because it was made of   2n+1
2n+1 2n+1 n+5 n+5
ivory, which allowed mul- n+3 2 2
. an = = 1− = 1− .
tiplication, division, and n+5 n+5 n+5
the mechanical calculation
of square and cubic roots.
Using the previous theorem, we obtain the following:
(Source of image: Engrav-
ing by Samuel Freeman n+5
2
(1773–1857)) . lim 1 − = e−2 .
n+5

2n + 1
Since . → 2, then
n+5

. lim an = (e−2 )2 = e−4 .

• Subsequences

Let f and g be two functions. The composition .f ◦ g is only possible if

the range of g is contained in the domain of f . Sequences are functions of
a natural variable; thus, the composition of two sequences, .u ◦ v, is only
possible if the range of v is a subset of .N.
Consider two sequences u and v, where v is a sequence of natural numbers.
The composition .u ◦ v is still a sequence, of general term .uvn . For example,
1.1. Sequences of Real Numbers 31

if u is the sequence 1, 2, 1, 3, 1, 4, . . . and .vn = 2n − 1, then .uvn = 1; if

z = 2n, then .uzn = n + 1; and if .sn = 4, then .usn = 3.
. n

A subsequence of a sequence can be obtained by omitting some of its terms

and keeping the remaining terms in the original order. Let us now consider
a more formal definition.

Definition 1.1.9 Given two sequences u and w, we say that w is a sub-

sequence of u if there exists a strictly increasing sequence of natural
numbers v, such that .w = u ◦ v.

Example 1.1.23 From the sequences considered earlier, .u ◦ v and .u ◦ z

are subsequences of u, but .u ◦ s is not a subsequence of u.

Notes:

1. Every subsequence of a bounded sequence is bounded.

2. An unbounded sequence may have bounded subsequences. For exam-

ple,
⎧
⎨n, if n is even
.un =
⎩ 1 , if n is odd.
n

3. Every subsequence of a monotonic sequence is monotonic.

4. A non-monotonic sequence can have monotonic subsequences, as

shown by the sequence of Example 1.1.12. The same can be veri-
fied in the sequence whose graph is illustrated in Fig. 1.14.

Theorem 1.1.15 Every subsequence of a convergent sequence converges

to the same limit.

Proof: Let u be a sequence that converges to a. Let .ε > 0. By definition,

we know that
∃ p ∈ N : n > p ⇒ |un − a| < ε.
.
32 Sequences of Real Numbers

Let w be a subsequence of u. By Definition 1.1.9, there exists a sequence

of natural numbers, strictly increasing, v, such that .w = u ◦ v. Since
.vn ≥ n, ∀n ∈ N, then if .n > p, also .vn > p and

|wn − a| = |uvn − a| < ε,

.

which proves that .wn → a.

Note: This theorem is mainly used as a negative test: If two subsequences

with different limits are found, the sequence diverges.

Theorem 1.1.16 Let u be a sequence. Let .(uvn ) and .(uzn ) be two

subsequences such that the union of their indices is .N. If .(uvn ) and .(uzn )
converge to the same limit a, then u converges to a.

Proof: Let .ε > 0 be arbitrary. By definition, we know that

∃p1 ∈ N : vn > p1 ⇒ |uvn − a| < ε

.

and
∃p2 ∈ N : zn > p2 ⇒ |uzn − a| < ε.
.

Let .p = max{p1 , p2 }. Observing that every natural number is a term of

one of the sequences .(vn ) or .(zn ), then .|un − a| < ε if .n > p.

Note: Theorem 1.1.16 remains valid for a finite number of subsequences

with the same limit, provided that the union of the indices is .N.

Lemma 1.1.1 Every sequence of real numbers has monotonic subse-

quences.

Proof: Let
M = {p ∈ N : up < un , ∀n > p}.1
.

If .M is infinite, that is, if there exists .p1 < p2 < · · · < pn < · · · belonging
to .M,
.up1 < up2 < up3 < · · · < upn < · · · ,

1 If,for example, .(un ) is strictly increasing, .M = N, if .(un ) is decreasing, .M = ∅,

1
and if .un = (−1)n , .M is the set of odd numbers.
n
1.1. Sequences of Real Numbers 33

and the subsequence .(upn ) is monotonic (increasing).

If .M is finite or empty, we will show that there exists a decreasing sub-
sequence .(uqn ). If .M is empty, we consider .q1 = 1. If .M is finite, let
.q1 = max(M) + 1. It is clear that .q1 ∈ / M. By the definition of .M,
there exists .q2 > q1 such that .uq2 ≤ uq1 . Again, by the definition of .M,
there exists .q3 > q2 such that .uq3 ≤ uq2 . Similarly, for each .n ∈ N, there
exists .qn > qn−1 such that .uqn ≤ uqn−1 . Therefore, there exists a subse-
quence .(uqn ) satisfying .uq1 ≥ uq2 · · · ≥ uqn ≥ uqn+1 ≥ · · · , that is, the
subsequence .(uqn ) is decreasing.

Theorem 1.1.17 Every bounded sequence has convergent subsequences.

Proof: By Lemma 1.1.1, the sequence .(un ) admits at least one monotonic
subsequence .(upn ). Because .(upn ) is monotonic and bounded, then by
Theorem 1.1.12 it is convergent.

Definition 1.1.10 A number .a ∈ R is a sublimit of the sequence u if

there exists a subsequence of u that converges to a.

1
Example 1.1.24 Consider the sequence .un = (−1)n + . The sublimits
n
1
of .(un ) are .−1 and 1 because 1 is the limit of subsequence .u2n = 1 +
2n
1
and .−1 is the limit of subsequence .u2n−1 = −1 + (see Fig. 1.15).
2n − 1

1
Figure 1.15: Sublimits of sequence .un = (−1)n +
n
34 Sequences of Real Numbers

Notes: Let S be the set of sublimits of the sequence u.

1. By Theorem 1.1.17, if u is bounded, .S = ∅.

2. S can be empty. As an example, .un = n, which is not bounded, does

not have convergent subsequences.

3. If u is convergent, then S is a unit set, that is, it has only one element.

4. S can be a unit set, and u can be divergent; for example, we have

⎧
⎨ 1 , if n is even
.un = n
⎩n, if n is odd.

5. S can be an infinite set; for example, given the sequence

1, 1, 2, 1, 2, 3, 1, 2, 3, 4, 1, 2, 3, 4, 5, . . . ,
.

then .S = N.

Theorem 1.1.18 The set of sublimits of a bounded sequence has a max-

imum and a minimum.

Proof: Let .M > 0, such that .−M ≤ un ≤ M, ∀n ∈ N. Let S be the

set of sublimits of sequence u, which is not empty because the sequence
is bounded. By Theorem 1.1.5, if .a ∈ S, then .−M ≤ a ≤ M , so S is
bounded. Therefore, S has an infimum, .α, and a supremum, .β. We will
show that .β ∈ S, that is, .β = max S. By definition of supremum, there
exists .a1 ∈ S such that .β − 1/2 < a1 ≤ β; by definition of sublimit, there
exists .up1 such that .a1 − 1/2 < up1 < a1 + 1/2; and then,
1 1
β − 1 < a1 −
. < up1 < a1 + < β + 1.
2 2
Again, by definition of supremum, there exists .a2 ∈ S such that

β − 1/4 < a2 ≤ β;
.

by definition of sublimit, there exists .up2 such that

a − 1/4 < up2 < a2 + 1/4;

. 2

and then,
1 1 1 1
β−
. < a2 − < u p2 < a2 + < β + .
2 4 4 2
1.1. Sequences of Real Numbers 35

Repeating this reasoning, for each .n ∈ N, there exists .upn such that
1 1
β−
. < u pn < β + .
n n
We have constructed a subsequence .(upn ) that converges to .β, so .β ∈ S.
Similarly, it can be shown that .α ∈ S.

Definition 1.1.11 Let u be a bounded sequence and S be the set of

sublimits of u. The maximum of S is called the superior limit or upper
limit of u and is represented by .lim un = lim sup un = max(S). The
minimum of S is called the inferior limit or lower limit of u and is
represented by .lim un = lim inf un = min(S).
If u is not bounded above, we define .lim un = +∞. If u is not bounded
below, we define .lim un = −∞.
If .un → +∞, then we define .lim un = lim un = +∞. If .un → −∞, we
define .lim un = lim un = −∞.

Theorem 1.1.19 A bounded sequence is convergent if and only if

lim un = lim un .
.

Proof: If u is convergent, with limit a, every subsequence of u has limit a

(Theorem 1.1.15), which implies that .S = {a} and
lim un = min(S) = a = max(S) = lim un .
.

Suppose u is bounded and .lim un = lim un = a. If .un → a, then

∃ ε > 0 ∀ p ∈ N ∃ n > p : |un − a| ≥ ε.
.

Thus, there exists .n1 such that .|un1 − a| ≥ ε; there exists .n2 > n1 such
that .|un2 − a| ≥ ε; and there exists .nm > nm−1 such that .|unm − a| ≥ ε.
Because it is bounded, the subsequence .(unm ) has a convergent subsequence
.(unm ): .unm → b = a, since .|unm − a| ≥ ε for all terms. Therefore, we
s s s
have shown that S has an element b different from a, which is incompatible
with the hypothesis of .lim un = lim un = a.

Definition 1.1.12 A sequence u is a Cauchy sequence if

∀ε > 0 ∃ p ∈ N : m, n > p ⇒ |un − um | < ε.

.
36 Sequences of Real Numbers

1
Example 1.1.25 We prove that the sequence .un = is a Cauchy se-
n
quence. Let .m, n > p; then
 
1 1  1 1 1 1 2
.
n m ≤ n + m < p + p = p.
−

Let .ε denote an arbitrary positive number; to conclude, it is enough to take

2
.p > .
Augustin Louis Cauchy ε
(1789–1857) was a French
engineer and mathemati-
cian. He was a pioneer of
Note: In the definition of a convergent sequence, we introduced an element
Mathematical Analysis and external to the sequence, the limit. The sequence converges if, from a certain
initiated a project to ex- order, all elements of the sequence “are close” to the limit. In the definition
press and rigorously prove of Cauchy sequence, we compare only the elements of the sequence with
the theorems of calculus,
making significant contri- each other. We say that the sequence is Cauchy if, from a certain order, all
butions to Complex Anal- elements of the sequence “are close” to each other.
ysis. He wrote about 800
works covering all areas of
Mathematics and Mathe- Theorem 1.1.20 A real sequence is convergent if and only if it is a Cauchy
matical Physics. (Source sequence.
of image: Charles H. Reut-
linger, Smithsonian Insti-
tution Libraries) Proof: Let u be a convergent sequence and a be its limit. Let .ε > 0 be
arbitrary. By definition,
ε
∃ p ∈ N : |un − a| <
. , ∀n > p.
2
Let .m, n > p be arbitrary. Then,
ε ε
|um − un | = |um − a + a − un | ≤ |um − a| + |a − un | ≤
. + = ε,
2 2
so u is a Cauchy sequence.
We must now prove the converse, that is, every Cauchy sequence is conver-
gent.
Let u be a Cauchy sequence. We begin by demonstrating that u is bounded.
In the definition, we can take, for example, .ε = 1: There exists .p ∈ N such
that .|un − um | < 1, ∀m, n > p; in particular,

|un | − |up+1 | ≤ |un − up+1 | < 1, ∀n > p,

.

so
|un | < |up+1 | + 1, ∀n > p.
.
1.1. Sequences of Real Numbers 37

Considering .M = max{|u1 |, |u2 |, . . . , |up |, |up+1 | + 1}, we obtain

.|un | ≤ M, ∀n ∈ N.

Because sequence u is bounded, it has by Theorem 1.1.17 a convergent

subsequence .(uvn ), with limit a. Let .ε > 0 be arbitrary. There exists .q ∈ N
such that
ε
.|uvn − a| < , ∀ vn > q.
2
By definition of Cauchy sequence, there exists .r ∈ N such that
ε
|un − um | <
. , ∀m, n > r.
2
Let .p = max{q, r}. Let .n > p and .vn > p; then,
ε ε
.|un − a| = |un − uvn + uvn − a| ≤ |un − uvn | + |uvn − a| < + = ε.
2 2
We conclude, therefore, that .un → a.

Note: This theorem allows us to show that a sequence converges without

having to calculate its limit. Consider the following sequence:
1 1 1
u =1+
. n
2
+ 2 + ··· + 2.
2 3 n
We can take, without loss of generality, .n > m; thus, we obtain
 
 1 1 1 

|un − um | =  + + ··· + 2
(m + 1)2 (m + 2)2 n
1 1 1
= + + ··· + 2
(m + 1)2 (m + 2)2 n
1 1 1
. ≤ + + ··· +
m(m + 1) (m + 1)(m + 2) (n − 1)n
1 1 1 1 1 1
= − + − + ··· + −
m m+1 m+1 m+2 n−1 n
1 1 1
= − ≤ .
m n m
1
If .p > and .n ≥ m > p, we obtain .|un − um | < ε, so the sequence is
ε
Cauchy and therefore converges.
38 Sequences of Real Numbers

1.2 Solved Exercises

 n
1. Evaluate the limits of the following sequences and n2 + 3
h) earctan(n)
provide an argument for the calculations: 2n2 + 1

n √ 3. Find the limits of the following sequences and justify

a) + nn
n2
+3 the calculations:
n+1 3n sin(23n + 1)
b) a)
2n + 5 23n + 1
n3 + 3n2 + 5 1
c) b) · cos(n + 1) · log(n)
n2 + 2 n
35n7 + 1 n2 + 3  
d) c) √ · cos n3 + 2
3
71n7 + 5n6 + 1 n n +2
√
3 1 √
n2 + n + n d)
n
n!
e) √4 √ n
2 n4 + 1 + n   n4
√ √ n4
n3 + 2n4 + 1 − n e)
n
n2 e−n −
f) √ n4 + 1
−2n2 + 3 n2 + 3
√ √ n sin(n)
(1 + 2 n ) n + 1 f) √
g) √ 2n 5n3 + 1
n+ 3n 
√ g) n2 + 2n − n
3
1 − 27n3    
h) (n + 1)n+2 n 1
1 + 4n h) − · sin
 √  (n + 2)n+1 3 n
n (−1)n + n
i) √ 3n − 5
2 + n3 + 1 i)
√ 5n + 3
n 3 n2 + 2
j) 4. Find the limits of the following sequences and pro-
n2 + (−1)n n
vide a clear explanation of the reasoning behind each
2 n · e1/n calculation:
k) √ √
(−1)n + n2 + 5  sin2 (n)
n−1 n 3
2n
a) c) √
3
n2 + 3k2 n 4+k
2. Calculate the limits of the following sequences: k=1 k=1
 2 n
n −1 
n
5n
a) 2 b) √
n n4 + k
k=1
 n  n
4 −5 2 5. Evaluate the supremum and infimum of the set of
b)
4n + 3 terms, and the limit of the sequence
 
n + 2 n+1 1 
c) .an =n+ − n2 + 1.
n+4 2n
 
2+n n
d) 6. a) Find the limit of the sequence
5 + 5n
  √ √ 
3n + 1 n .an = 2n + 1 − 2n · cos2 (n)
e)
3n + 2
  and give a clear explanation of the answer.
2n + 1 4n−2
f) 3 − b) Determine the set of sublimits of the sequence
n
nπ
  bn = sin · arctan(n) and give a justifi-
2n + 5 n+4 2
g) cation for the answer provided.
2n + 1
1.2. Solved Exercises 39

7. Consider the sequence a) By induction show that xn > 0, ∀n ∈ N.

b) Show that the sequence is decreasing.
n n·(−1)n
.un = 1+2 . c) Show that the sequence is convergent and de-
termine its limit.
a) Write the subsequence of even-indexed terms
11. Let a ∈ R be a positive number. Consider the se-
and calculate its limit.
quence of real numbers defined recursively
b) Write the subsequence of odd-indexed terms
and calculate its limit. x0 = 0, x1 = a
.
c) Calculate lim un and lim un . xn+1 = xn + x2n−1 , ∀n ∈ N.

d) Given the previous items, what can we con-

a) Show that the sequence is increasing.
clude about the convergence of the sequence?
b) Show that xn > 0, ∀n ∈ N.
8. Consider the sequence defined recursively:
c) Show that if there exists b ∈ R such that
√ lim xn = b, then b = 0.
u1 = 2
. √ d) Given the previous items, calculate if it exists,
un+1 = 2 un , ∀n ∈ N.
lim xn .

a) By induction prove that 0 < un < 2, ∀n ∈ N. 12. Consider the sequence of real numbers defined re-
cursively
b) Prove that the sequence is increasing.
⎧
c) Prove that the sequence is convergent. ⎨ x1 = 2
. x 1
d) Evaluate the limit of the sequence. ⎩xn+1 = n + , ∀n ∈ N.
2 xn
9. Consider the sequence
a) Prove that xn > 0, ∀n ∈ N.
⎧ √
⎨a1 = √2 b) Prove that xn > 2, ∀n ∈ N.
⎩an+1 = √2 an ,
.
∀n ∈ N. c) Prove that (xn ) is monotonic.
d) Show that (xn ) converges.
a) By induction show that e) Evaluate the limit of (xn ).
√
. 2 ≤ an < 2, ∀n ∈ N. 13. Consider the sequence of real numbers defined re-
cursively
b) By induction show that the sequence (an ) is ⎧
⎪
⎨ x1 = 3
increasing.
.
⎪ x2 + 3
c) Show that there exists a ≤ 2 such that an → ⎩xn+1 = n , ∀n ∈ N.
2 xn
a.

10. Let a ∈ R be a positive number. Consider the se- a) Show by induction that
quence of real numbers defined recursively √
.xn − 3 > 0, ∀n ∈ N.
⎧
⎪
⎨x1 = a b) Show that the sequence (xn ) is decreasing.
. xn
⎪
⎩xn+1 = , ∀n ∈ N. c) Show that the sequence (xn ) converges.
2 + xn
d) Find the limit of the sequence (xn ).
40 Sequences of Real Numbers

14. Prove by definition the following limits: The middle segment of each of the four segments is
25n 25 removed and replaced by two segments of the same
a) lim = length. The resulting curve has a length of 16 .
3n + 1 3 9
√
n
b) lim √ =1
n+1
√
2 n + 5−n
c) lim √ = 2
n+1
 
sin(n)
d) lim 1+ =1
n
e) lim n3 = +∞
n2
f) lim = +∞
n+1 Continuing this process ad infinitum, the result is the
15. The Koch curve is created from a line segment of Koch curve.
length 1. This segment is divided into three equal
parts. We then remove the middle segment. Next,
we replace the removed segment with two segments
of length 13 . The resulting shape can be seen in the
figure below.

What is the length of the Koch curve?

1.2. Solved Exercises 41

SOLUTIONS

n ∞
1. a) Let an = . It is not possible to directly apply Theorem 1.1.5, as the indeterminate form ∞
occurs.
n2 + 3
To solve this, we can rewrite an as follows:

1 1
n2 ·
n n n
.an = =   = .
n2 + 3 3 3
n2 1 + 2 1+ 2
n n
 
1 3
We know, from Example 1.1.16, that is a null sequence, which by Theorem 1.1.5, implies that 2 → 0.
n n
Again by Theorem 1.1.5,
. lim an = 0.
√
n
Since lim n = 1 (see Note 1 after Theorem 1.1.11), we can conclude that
 
n √
n
. lim + n = 1.
n2 + 3

Note: The procedure described here is equivalent to dividing both terms of the fraction by the power of n
with the highest exponent that appears in the expression. Alternatively, we could use Theorem 1.1.9, with
vn = n, keeping in mind that p = 1 and q = 2.
n+1
b) The following sequence is given by an = . Taking into account the note from the previous item, we
2n + 5
divide both members of the fraction that defines the sequence (an ) by n, which is the highest power in the
expression.
n+1 1
1+
n+1 n n
.an = = = .
2n + 5 2n + 5 5
2+
n n
 
1
Since is a null sequence, we can apply Theorem 1.1.5 and conclude that
n
1
. lim an = .
2

n3 + 3n2 + 5
c) Let an = . Since n3 is the highest power of n that appears in the fraction that defines the
n2 + 2
sequence (an ), we can divide the numerator and denominator by n3 :

n3 + 3n2 + 5 3 5
1+ + 3
n3 +3n2 +5 n3 n n
.an = = = .
n2 + 2 n2 + 2 1 2
+
n3 n n3
Using Theorem 1.1.5, we can conclude that

. lim an = +∞,
 
1
because is a null sequence and an > 0, ∀n ∈ N.
n
42 Sequences of Real Numbers

35n7 + 1
d) Let an = . If we divide both the numerator and denominator of the fraction that defines
71n7 + 5n6 + 1
the sequence (an ) by n raised to the highest power, n7 , we obtain

35n7 + 1 1
35 + 7
35n7+1 n7 n
.an = = = .
71n7 + 5n6 + 1 71n7 + 5n6 + 1 5 1
71 + + 7
n7 n n
 
1
Knowing that is a null sequence, we can use Theorem 1.1.5, to conclude that
n
35
. lim an = .
71
√
3
n2 + n + n
e) Given an = √
4 √ , we can simplify the expression by dividing both the numerator and denomi-
2n4 + 1 + n
nator by n as shown below:
√  
3 2
n2 + n + n 3 n +n 3 1 1
√ +1 + 2 +1
3
n2 + n + n n n3 n n
.an = √4 √ = √ 4 √ =   =   .
2n4 + 1 + n 2n4 + 1 + n 4
4 2n + 1 n 4 1 1
n + 2+ 4 +
n4 n2 n n

Using Theorem 1.1.5, the conclusion is

1
. lim an = √
4
.
2
√
n3 + 2n4 + 1 − n
f) Let an = √ . We can rewrite an using the following equalities:
−2n2 + 3 n2 + 3
√  
n3 + 2n4 + 1 − n n3 + 2n4 + 1 1 1 1
+2+ 4 −
1
√ −
n + 2n + 1 − n
3 4 n 2 n 4 n n n n
.an = √ = √ =  =  .
−2n2 + 3 n2 + 3 −2n2 + 3 n2 + 3 n 2+3 1 3
n2 −2 +
3
−2 + 3
+ 6
n6 n4 n

By applying Theorem 1.1.5 to the above expression, we conclude that the limit is given by
√
2
. lim an = − .
2
 √ √
1+2 n n+1
g) Let an = √ . To simplify the expression of (an ), we can divide both the numerator and
n+ 3n
denominator of the fraction that defines it by n. This gives us
 √ √  √ √  
1+2 n n+1 1+2 n n+1 1 1
 √ √ √ √ √ +2 1+
1+2 n n+1 n n n n n
.an = √ = √ =  =  .
n+ 3n n+ 3n n 3 1
1+ 3 3 1+
n n n2
By applying Theorem 1.1.5, we can determine that

. lim an = 2.
1.2. Solved Exercises 43
√
3
1 − 27n3
h) Let an = . To find the limit of this sequence, we will manipulate the expression for an to
1 + 4n
simplify it.
First, we will divide both the numerator and denominator of the fraction that defines the sequence (an ) by
n, and next we simplify the expression. This gives us
√  
1 − 27n3 3 1 − 27n
3 3
1
√ 3
− 27
3
1 − 27n3 n n3 n3
.an = = = = .
1 + 4n 1 + 4n 1 1
+4 +4
n n n
Therefore, by Theorem 1.1.5,
3
=− .
. lim an
4
 √ 
n (−1)n + n
i) Consider the sequence (an ) given by an = √ . In this case, by dividing both the numerator
2 + n3 + 1
and denominator of the fraction that defines the sequence (an ) by n3/2 , we can simplify the expression as
follows:
 √  √
n (−1)n + n (−1)n + n (−1)n
 √  √ +1
n (−1)n + n n 3/2 n 1/2 n
.an = √ = √ =  =  .
2 + n3 + 1 2 + n3 + 1 2 n3 + 1 2 1
√ + √ + 1 +
n3/2 n3 n3 n3
n3
 
  1
Since (−1)n is bounded and √ is a null sequence, we can apply Theorem 1.1.6 to conclude that
n
(−1) n
√ → 0. Moreover, applying Theorem 1.1.5, we can compute the limit of the sequence and obtain
n
. lim an = 1.
√
3
n n2 + 2
j) Let an = . By dividing both the numerator and the denominator of the fraction that defines
n2 + (−1)n n
2
sequence (an ) by n , we obtain
√ √  
n 3 n2 + 2 3
n2 + 2 2
3 n +2 3 1 2
√
3 + 3
n n2 + 2 n2 n n3 n n
.an = = 2 = = = .
n2 + (−1)n n n + (−1)n n (−1)n (−1)n (−1)n
2
1+ 1+ 1+
n n n n

Therefore, according to Theorem 1.1.5, we arrive at the following result:

. lim an = 0.

2 n e1/n 2n
k) Let an = √ = √ · e1/n = bn · e1/n . By dividing both the numerator and
(−1)n+ n2 + 5 (−1)n + n2 + 5
denominator of the fraction that defines the sequence (bn ) by n, we obtain
2n 2 2 2
.bn = √ = √ =  =  .
(−1)n + n2 + 5 (−1)n + n2 + 5 (−1) n 2
n +5 (−1) n 5
+ + 1 +
n n n2 n n2
44 Sequences of Real Numbers

Hence,
. lim bn = 2.
1/n
Since lim e = 1, we can conclude that

2 n e1/n
. lim √ = 2.
(−1)n + n2 + 5

2. a) Consider the sequence (an ) defined as

 n  n  n2 1/n
n2 − 1 1 1
.an = = 1− = 1− , ∀n ∈ N.
n2 n2 n2

Using Theorem 1.1.14, we can show that

  n2
1
. lim 1− = e−1 .
n2

1
Since lim = 0, we can conclude by Theorem 1.1.5 that
n

. lim an = (e−1 )0 = 1.

b) Denoting by an the general term of the sequence, we can write

⎡  ⎤2 n ⎡ 22n ⎤1/2n
5 5
 2n 1− n
4n ⎢ 1− ⎥
4n − 5 ⎢ ⎥
=⎢ 
4
 ⎥ =⎢ ⎢
4n ⎥
, ∀n ∈ N.
.an =
4n + 3 ⎣ 3 ⎦ ⎣ 22n ⎥
⎦
4n 1 + n 3
4 1+
4n

Using Theorem 1.1.14, we can find the limit of the sequence:

 22n   n
5 5 4
. lim 1− = lim 1 − n = e−5
4n 4

and
 22n   n
3 3 4
. lim 1+ = lim 1 + n = e3 .
4n 4
1
But by Theorem 1.1.10, lim = 0; therefore,
2n
 0
e−5
. lim an = = 1.
e3

c) If an is the general term of the sequence being discussed, we have

⎡   ⎤n+1 ⎡ ⎤n+1  
2 2 2 n 2
 n+1 n 1+ 1+ 1+ 1+
n+2 ⎢ n ⎥ ⎢ n⎥ n n
.an = =⎢
⎣  ⎥⎦ =⎢
⎣
⎥
⎦ =  n · , ∀n ∈ N.
n+4 4 4 4 4
n 1+ 1+ 1+ 1+
n n n n
1.2. Solved Exercises 45

Using Theorem 1.1.13, we arrive at the following results:

 
2 n
. lim 1 + = e2
n

and  
4 n
. lim 1+ = e4 ;
n
therefore,
e2
. lim an = · 1 = e−2 .
e4
d) To find the limit of the sequence (an ), we can simplify the expression defining it by putting n in evidence
in the numerator of the expression that defines (an ) and 5n in the denominator. This gives us the following
expression:
⎡   ⎤n  
2 2 n
n 1+  n 1+
⎢ n ⎥ 1 n
.an = ⎢  ⎥ = ·  , ∀n ∈ N.
⎣ 1 ⎦ 5 1 n
5n 1 + 1+
n n
 n
1
We know from Theorem 1.1.10 that lim = 0. Additionally, by Theorem 1.1.13, we can find that
5
 
2 n
. lim 1 + = e2
n

and  n
1
. lim 1+ = e.
n
Therefore,  
2 n
 n 1+
1 n e2
. lim an = lim · lim   =0· = 0.
5 1 n e
1+
n
e) If we consider an as the general term of the sequence, we can write it as follows:
⎡   ⎤n ⎡ ⎤n  
1 1 1/3 n
 n 3n 1 + 1+ 1+
3n + 1 ⎢ 3n ⎥ ⎢ 3n ⎥ n
.an = =⎢
⎣  ⎥⎦ =⎢
⎣
⎥ =   , ∀n ∈ N.
3n + 2 2 2 ⎦ 2/3 n
3n 1 + 1+ 1+
3n 3n n

According to Theorem 1.1.13, we know that

 n
1/3
. lim 1+ = e1/3
n

and  n
2/3
. lim 1+ = e2/3 .
n
Therefore,
e1/3
. lim an = = e−1/3 .
e2/3
46 Sequences of Real Numbers

f) We have
    
1 4n 1 n 4
 4n−2 3−2− 1−
2n + 1 n n
.an = 3− =   =   , ∀n ∈ N.
n 1 2 1 2
3−2− 1−
n n
By Theorem 1.1.13, we know that  n
1
. lim 1− = e−1 .
n
Thus, we can conclude that  4
. lim an = e−1 = e−4 .

g) We have

⎡  ⎤n+4 ⎡⎛ ⎞2n ⎤1/2 ⎡ ⎤4

5 5 5
 n+4 2n 1 + ⎢⎜ 1 + ⎥ 1+
⎢ 2n ⎥ ⎢ 2n ⎟ ⎢ 2n ⎥
.an =
2n + 5
=⎢  ⎥ = ⎢⎜ ⎟ ⎥ ⎥ ·⎢ ⎥ , ∀n ∈ N.
2n + 1 ⎣ 1 ⎦ ⎣⎝ 1 ⎠ ⎦ ⎣ 1 ⎦
2n 1 + 1+ 1+
2n 2n 2n

Using Theorem 1.1.14, we can obtain

 2n
5
. lim 1+ = e5
2n

and  2n
1
. lim 1+ = e.
2n
Therefore, we can find that
 1/2
e5
. lim an = = e2 .
e
π
h) Knowing that lim arctan(x) = , we can conclude1 that lim earctan(n) = eπ/2 . Now, let us consider
 2 x→+∞
n 2
n +3 2
an = . Putting n in evidence in this expression yields
2n2 + 1

⎡   ⎤n    1/n
3 3 n 3 n2
 n n2 1+  n 1 +  n 1+ 2
n2 +3 ⎢ n 2 ⎥ 1 n 2 1 n
.an = =⎢
⎣  ⎥
⎦ = 2 · n = · 1/2n .
2n2 + 1 1 1 2 1 2n2
2n2 1 + 1 + 1+
2n2 2n2 2n2

Applying Theorem 1.1.14, we obtain

  2
3 n
. lim 1+ 2 = e3
n
and
 2n2
1
. lim 1+ = e.
2n2

1 We are using the following result: If lim f (x) = L and f (n) = un , n ∈ N, then lim un = L.
x→+∞
1.2. Solved Exercises 47
 n
1 1 1
Therefore, as lim = lim = lim = 0, we can conclude that
n 2n 2
(e3 )0
. lim an =0· = 0.
e0
Thus, we get  n
n2 + 3
. lim · earctan(n) = 0.
2n2 + 1
 
3n sin 23n + 1 3n  
3. a) Let an = = · sin 23n + 1 . We know that
23n + 1 23n + 1
. |sin(x)| ≤ 1, ∀x ∈ R;
%  3n %  
% %
therefore, sin 2 + 1 ≤ 1, ∀n ∈ N, that is, the sequence with general term sin 23n + 1 is a bounded
3n
sequence. We prove that the sequence with general term 3n is a null sequence.
2 +1
 n
3
3n 3n
8
. lim = lim n = lim  n = 0.
23n + 1 8 +1 1
1+
8
We conclude that the given sequence is a null sequence because it is the product of a null sequence and a
bounded sequence (see Theorem 1.1.6).
1
b) Let an = · cos(n + 1) · log(n). We know that −1 ≤ cos(x) ≤ 1, ∀x ∈ R. Then
n
. − 1 ≤ cos(n + 1) ≤ 1, ∀n ∈ N.
Therefore,
√  log(n) 1 log(n) √ 
.
n
− log
n =− ≤ · cos(n + 1) · log(n) ≤ = log n n , ∀n ∈ N.
n n n
√
Because we know that lim n n = 1, we have
√
n

. lim log n = 0.
It follows from the Squeeze Theorem that
. lim an = 0.
n2 + 3  
c) Let an = √ · cos n3 + 2 . For all n ∈ N, we have
n n3 + 2
%  %%
%
.0 ≤ %cos n3 + 2 % ≤ 1.

n2 + 3
Let us consider the sequence of general term √ . By dividing the numerator and denominator of
n n3 + 2
the fraction that defines this sequence by the highest power of n, we obtain
 
n2 + 3 n2 + 3 (n2 + 3)2 1 6 9
+ 3 + 5
n 5/2 n 5/2 n 5 n n n
. lim √ = lim √ = lim  = lim  = 0.
n n3 + 2 n3 + 2 n3 + 2 2
1+ 3
n5/2 n3/2 n3 n
We conclude that the sequence (an ) is a null sequence because it is the product of a null sequence and a
bounded sequence (see Theorem 1.1.6).
48 Sequences of Real Numbers

1 √
n n n! n!
d) Let an = n! = . Let bn = n . It is evident that bn > 0, ∀n ∈ N.
n nn n

(n + 1)!
 n
bn+1 (n + 1)n+1 (n + 1)! nn (n + 1) nn n 1
. lim = lim = lim = lim = lim = .
bn n! (n + 1)n+1 n! (n + 1)n+1 n+1 e
nn

1
By Theorem 1.1.11 we can conclude that lim an = .
e
√   n4
n4
and bn = n2 e−n . It is evident that bn > 0, ∀n ∈ N.
n
e) Let an = n2 e−n − 4
n +1

(n + 1)2  2
bn+1 2 n
. lim = lim en+1 = lim (n + 1) e = 1 lim n + 1 1
= .
bn n2 en+1 n2 e n e
en
√
n 1
By Theorem 1.1.11 we can conclude that lim n2 e−n = . Using Theorem 1.1.14 we obtain the following
e
result:

 n4
n4 1 1 1
. lim = lim 
4
n +1 n4 = lim   n4 = e .
n4+1 1
1+ 4
n4 n
1 1
We conclude that lim an =− = 0.
e e
 n
n sin(n) 1 n
f) Let an = √ = ·√ · sin(n). We know that
2n 5n3 + 1 2 5n3 + 1

. |sin(n)| ≤ 1, ∀n ∈ N;
 
  n
therefore, the sequence sin(n) is a bounded sequence. We show that the sequence √ is a
5n3 + 1
null sequence.

n 1 1
√ √
n n 3/2 n n
. lim √ = lim √ = lim  = lim  = 0.
5n3 + 1 5n3 + 1 5n3 + 1 1
5 +
n3/2 n3 n3

 n
1
Since lim = 0, we have
2
 n
1 n
. lim ·√ = 0.
2 5n3 + 1
We conclude that the given sequence is a null sequence because it is the product of a null sequence by a
bounded sequence (see Theorem 1.1.6).
1.2. Solved Exercises 49

g) It is not possible to directly apply Theorem 1.1.5, as the indeterminate form ∞ − ∞ occurs. Dividing and
multiplying the general term of the sequence by its conjugate, we obtain
√ √ 
  n2 + 2n − n n2 + 2n + n n2 + 2n − n2
lim n2 + 2n − n = lim √ = lim √
2
n + 2n + n n2 + 2n + n
2n
.
2n n 2 2
= lim √ = lim √ = lim  = lim  = 1.
n2 + 2n + n n2 + 2n + n 2
n + 2n 2
+1 1 + + 1
n n2 n

   
(n + 1)n+2 n 1
h) Let an = − · sin . Then
(n + 2)n+1 3 n
        
(n + 1)n+2 n 1 n + 1 n+2 n 1
lim n+1
− · sin = lim · (n + 2) − · sin
(n + 2) 3 n n+2 3 n
 n+2    
1 1 n 1
= lim 1− · (n + 2) · sin − · sin
n+2 n 3 n
 n+2    n+2    
. 1 1 1 1 n 1
= lim 1− · n · sin + 1− · 2 · sin − · sin
n+2 n n+2 n 3 n
⎛    ⎞
1 1
 n+2 sin  n+2   sin
⎜ 1 n 1 1 1 n ⎟
⎜
= lim ⎝ 1 − · + 1− · 2 · sin − · ⎟.
n+2 1 n+2 n 3 1 ⎠
n n
 n+2  
1 1 1
By Theorem 1.1.14, lim 1 − = e−1 . Since lim = 0, we conclude that lim sin = 0
  n+2 n n
1
sin
n 1
and lim = 1. Using Theorem 1.1.5, we arrive at the following result: lim an = e−1 − .
1 3
n
i) By dividing both the numerator and denominator of the fraction that defines the general term of the sequence
by the power of the largest base and exponent, we obtain, by Theorem 1.1.10,
 n  n−1
3n − 5 3 1
−
3 −5
n
5n 5 5
. lim = lim n = lim = 0.
5n + 3 5 +3 3
1+ n
5n 5

4. In this exercise, we apply the Squeeze Theorem. When dealing with sequences defined by summations, a useful
technique is to first identify the largest and smallest terms in the summation. Then, these terms can be used to
bound the sequence.

a) The given sequence has a general term an which is defined as the sum from k = 1 to k = n − 1 of
sin2 (n) sin2 (n)
2 2
. As we show below, the largest term is 2 (which corresponds to k = 1), and the smallest
n + 3k n +3
2
sin (n)
is 2 (corresponding to k = n − 1). Clearly,
n + 3(n − 1)2
50 Sequences of Real Numbers

.n
2
+ 3k2 > n2 , ∀ k, n ∈ N
and
.n
2
+ 3k2 ≤ n2 + 3(n − 1)2 , ∀ k, n ∈ N such that k ≤ n − 1.
Therefore, for all n and 1 ≤ k ≤ n − 1, we obtain
1 1 1
n2 < n2 + 3k2 ≤ n2 + 3(n − 1)2 ⇒ > 2 ≥ 2
n2 n + 3k2 n + 3(n − 1)2
.
sin2 (n) sin2 (n) sin2 (n)
⇒ > 2 ≥ 2 .
n 2 n + 3k 2 n + 3(n − 1)2

As (an ) is defined as the sum of n − 1 terms,

sin2 (n)  sin2 (n)

n−1
sin2 (n)
(n − 1) · ≤ < (n − 1) · , ∀n ∈ N
n2 + 3(n − 1)2 k=1
n 2 + 3k 2 n2

n−1  sin2 (n)

n−1
n−1
. ⇔ · sin2 (n) ≤ < · sin2 (n), ∀n ∈ N
n2 + 3(n − 1) 2
k=1
n 2 + 3k 2 n 2

 sin2 (n)
n−1  
n−1 1 1
⇔ · sin2 (n) ≤ < − 2 · sin2 (n), ∀n ∈ N.
n2 + 3(n − 1) 2
k=1
2
n + 3k 2 n n

n−1 n−1
Let bn = = . Dividing the numerator and denominator of the fraction that
n2 + 3(n − 1)2 4n2 − 6n + 3
2
defines the sequence by n , we get

1 1 1 1
− 2 − 2
n−1 n n n n
. lim = lim = lim = 0.
4n2 − 6n + 3 4n2 − 6n + 3 6 3
4 − +
n2 n n2
1 1
Let cn = − 2 . It is evident that lim cn = 0.
n n  
As we know that | sin2 (n)| ≤ 1, ∀n ∈ N, the sequence sin2 (n) is bounded. Since the product of a null
sequence by a bounded sequence is a null sequence, we can assert that the sequences of general terms
n−1
. · sin2 (n)
n2 + 3(n − 1)2
and  
1 1
. − 2 · sin2 (n)
n n
are null sequences. Finally, as the two limits are equal, the Squeeze Theorem allows us to conclude that

. lim an = 0.

5n
b) The general term an of the sequence is defined as the sum from k = 1 to k = n of √ . The largest
n4 + k
5n 5n
term is √ (which corresponds to k = 1), and the smallest is √ (corresponding to k = n), as
n4 + 1 n4 + n
shown below. We have 
. n4 + k > n , ∀ k, n ∈ N,
2
1.2. Solved Exercises 51

and  
. n4 + k ≤ n4 + n, ∀ k, n ∈ N such that k ≤ n.

Therefore, for all n and 1 ≤ k ≤ n, we have

  1 1 1
n2 < n4 + k ≤ n4 + n ⇒ > √ ≥ √
n2 n4 + k n4 + n
.
5n 5n 5n
⇒ > √ ≥ √ .
n2 n4 + k n4 + n

Because (an ) is defined as the sum of n terms, we obtain

5n 
n
5n 5n
n· √ ≤ √ <n· , ∀n ∈ N
n4 +n k=1
n4 + k n2
.
5n2 
n
5n 5n2
⇔ √ ≤ √ < = 5, ∀n ∈ N.
n4 +n k=1
n4 +k n2

5n2 n4
Let bn = √ = 5· . Dividing the numerator and denominator of the radicand of this
n4 + n n4 + n
4
sequence by n , we obtain &
' 1
. lim bn = 5 · lim '
( = 5.
1
1+
n3
Finally, as the two limits are equal, the Squeeze Theorem allows us to conclude that

. lim an = 5.
√
3
2n
c) The general term an of the sequence is defined as the sum from k = 1 to k = n of √3
. The largest
n 4+k
√
3
√
3
2n 2n
term is √
3
(which corresponds to k = 1), and the smallest is √ 3
(corresponding to k = n),
4
n +1 4
n +n
as shown below. We have √ √
3 3
. n4 + k > n4 , ∀ k, n ∈ N,
and √ √
3 3
. n4 + k ≤ n4 + n, ∀ k, n ∈ N such that k ≤ n.
Therefore, for all n and 1 ≤ k ≤ n, we obtain
√
3
√
3
√
3 1 1 1
n4 < n4 + k ≤ n4 + n ⇒ √
3
> √
3
≥ √
3
n4 n4 + k n4 + n
. √
3
√
3
√
3
2n 2n 2n
⇒ √
3
> √
3
≥ √
3
.
n4 n4 + k n4 + n
Because (an ) is defined as the sum of n terms, we obtain
√ √ √
3
2n 
n 3 3
2n 2n
n· √
3
≤ <n· √ √
3 3
, ∀n ∈ N
n4 + n k=1 n 4+k n4
. √ √ √
3
2n4 n 3
2n
3
2n4 √
3
⇔ √
3
≤ √3
< √3
= 2, ∀n ∈ N.
n4 + n k=1 n 4+k n 4
52 Sequences of Real Numbers

Dividing the numerator and denominator of the fraction that defines the sequence on the left side of the
inequality by n4/3 , we have
√
3
√ 2n4
√ √ √
3
2n4 3
n4
3
2 3
2 √
3
. lim √ = lim √ = lim n = lim = 2.
3
n4 + n
3
n4 + n 1+ √ 1
√ 3 1+ √
3 n4 3
n
n4

Finally, as the two limits are equal, the Squeeze Theorem allows us to conclude that
√
3
. lim an = 2.

1 
5. Let an = n + − n2 + 1. We can simplify this expression as follows:
2n
√ √
1  1 (n − n2 + 1)(n + n2 + 1) 1 n2 − (n2 + 1) 1 1
n+ − n2 + 1 = + √ = + √ = − √
2n 2n n+ n +12 2n n + n2 + 1 2n n + n2 + 1
. √ √ √ √
n + n2 + 1 − 2n n2 + 1 − n ( n2 + 1 − n)( n2 + 1 + n) 1
= √ = √ = √ = √ .
2n(n + n2 + 1) 2n(n + n2 + 1) 2n(n + n2 + 1)2 2n(n + n2 + 1)2

By taking the limit, we

√ can conclude that lim an = 0 and that the sequence (an ) is decreasing. Therefore,
sup(an ) = a1 = 32 − 2. According to Theorem 1.1.12, inf(an ) = 0.

6. a) We will use the fact that the cosine function is bounded. This means that

. |cos(n)| ≤ 1, ∀n ∈ N,

which implies that % %

. %cos (n)% ≤ 1, ∀n ∈ N.
2

In addition, √ √ √ √ 
√ √  2n + 1 − 2n 2n + 1 + 2n
lim 2n + 1 − 2n = lim √ √
2n + 1 + 2n
.
2n + 1 − 2n 1
= lim √ √ = lim √ √ = 0.
2n + 1 + 2n 2n + 1 + 2n

As a result of the above, we can conclude that (an ) is a null sequence since it is the product of a bounded
sequence and a null sequence (see Theorem 1.1.6).

b) The general term of the sequence can be rewritten as follows:

⎧
⎪
⎨−1, n = 4k − 1, k ∈ N
nπ
.an = sin = 0, n = 2k, k ∈ N
2 ⎪
⎩
1, n = 4k − 3, k ∈ N.

The sequence (an ) has three sublimits, namely, −1, 0, and 1, because it has subsequences that converge
π
to these real numbers. The sequence cn = arctan(n) has limit . Therefore, we can conclude that the
2
nπ π π
sublimits of the sequence bn = sin · arctan(n) are − , 0, and .
2 2 2
1.2. Solved Exercises 53

n (−1)n n
7. a) Let un = 1+2 . The even-indexed subsequence of (un ) is given by

2k (−1)2k 2k 2k 2k
.u2k = 1+2 = 1+2 , k ∈ N.
√
Let us consider the sequence an = n
1 + 2n . By Theorem 1.1.11, as 1 + 2n > 0, ∀n ∈ N, we can calculate
the limit of (an ) by
1 + 2n+1 1
1 + n+1
1+ 2n+1 2n+1 2
. lim = lim = lim = 2.
1 + 2n 1 + 2n 1 1
+
2n+1 2 2n+1

The limit of the sequence (an ) is 2; therefore, all its subsequences have this limit. In particular, the even-
indexed subsequence has limit 2. However, this subsequence is equal to the sequence (u2k ). We can affirm
that lim u2k = 2.
b) The odd-indexed subsequence of (un ) is given by

2k+1 (−1)2k+1 (2k+1) 2k+1 −(2k+1) 1
, k ∈ N,
2k+1
.u2k+1 = 1+2 = 1+2 = 1+
22k+1

and lim u2k+1 = 1.

k→+∞

c) Taking into account the results obtained in the previous items,

.lim un = 1 and lim un = 2.

d) Since lim un = 1 = lim un = 2, the sequence (un ) is divergent.

8. a) We will use the Principle of Mathematical Induction to demonstrate that the sequence (un ) satisfies

.0 < un < 2, ∀n ∈ N.

For the base case n = 1, we have the following trivial formula:

√ √ √
.0 = 0< 2 = u1 < 4 = 2.

Induction hypothesis: 0 < un < 2.

Induction thesis: 0 < un+1 < 2.

√
Proof: Using the induction hypothesis and the fact that the function f (x) = 2x is increasing, we have
√ √ √
.0 < un < 2 ⇒ 0 = 2·0< 2 un = un+1 < 2 · 2 = 2.

Therefore, the property is valid for n + 1.

By the Principle of Mathematical Induction, we can conclude that the inequality 0 < un < 2 holds for all
n ∈ N.
54 Sequences of Real Numbers

b) We will show that the sequence (un ) is increasing, that is,

.un+1 − un > 0, ∀n ∈ N.

For any natural number n, we have

√ √ 
√ 2 un − un 2 un + un 2 un − u2n un (2 − un )
.un+1 − un = 2 un − un = √ = √ = √ >0
2 un + un 2 un + un 2 un + un

because in item a), we have already established that un > 0 and 2 − un > 0. Therefore, the sequence is
increasing.
c) Since the sequence is bounded (as shown in item a)) and increasing (as shown in item b)), we can conclude
that the sequence (un ) is convergent because by Theorem 1.1.12, every bounded monotonic sequence is
convergent.
d) Let b ∈ R be the limit of the sequence. Since every subsequence of a convergent sequence is convergent to
the same limit, we have
. lim un+1 = b.

Hence,
√ √
.b = lim un+1 = lim 2 un = 2 b.
√
Therefore, b satisfies equation b = 2 b, which means that b2 − 2b = 0. Thus, b(b − 2) = 0, and we can
exclude the solution b = 0 because by item b), we have
√
.un ≥ u1 = 2, ∀n ∈ N.

Therefore, we conclude that the limit of u is b = 2.

√ x √ √
9. a) Let f be the function defined by f (x) = 2 = ex·log( 2) . Since 2 > 1, f is continuous and increasing
on R.
√ √
For n = 1, the formula is trivial: 2 ≤ a1 = 2 < 2.
√
Induction hypothesis: 2 ≤ an < 2.
√
Induction thesis: 2 ≤ an+1 < 2.

Proof: Assuming that the property is valid for n (induction hypothesis) and using the fact that f is an
increasing function, we have

√  √ √2 √ 
. 2 ≤ an < 2 ⇒ 2 = f 2 ≤ f (an ) = an+1 < f (2) = 2.

Using again the monotonicity of f , we obtain

√ √ √ √ √
.1 < 2 ⇒ f (1) = 2 ≤ f ( 2) = ( 2) 2 ,

which implies that the property is valid for the order n + 1. Therefore, the Principle of Mathematical
Induction is satisfied, and we conclude that
√
. 2 ≤ an < 2, ∀n ∈ N.
1.2. Solved Exercises 55

b) We will show, using the Principle of Mathematical Induction, that

.an < an+1 , ∀n ∈ N.

For n = 1, the formula is a consequence of the calculations in item a):

√ √ √2
. 2 = a1 < a2 = 2 .

Induction hypothesis: an < an+1 .

Induction thesis: an+1 < an+2 .

Proof: Assuming that the property is valid for n ∈ N (induction hypothesis), the validity of the property for
n + 1 is a direct consequence of the fact that f is increasing:

.an < an+1 ⇒ an+1 = f (an ) < an+2 = f (an+1 ).

Therefore, we can conclude that the sequence is increasing.

c) In item a), we saw that the sequence is bounded, and in item b), we showed that it is increasing. By
Theorem 1.1.12, every bounded monotonic sequence is convergent, so we can conclude that the sequence
with general term an converges. Let a ∈ R be its limit. Since an < 2, ∀n ∈ N, by Theorem 1.1.5, we come
to the conclusion that a ≤ 2.

10. a) For n = 1, we have x1 = a > 0 by hypothesis, which implies that x1 > 0.

Induction hypothesis: xn > 0.

Induction thesis: xn+1 > 0.

xn
Proof: We have xn+1 = . Since xn > 0 by induction hypothesis, we know that 2 + xn > 0. Thus,
2 + xn
xn+1 is the quotient of two positive quantities, so xn+1 > 0.

Therefore, by the Principle of Induction, we have established that xn > 0, ∀n ∈ N.

b) We want to prove that xn+1 − xn < 0 for any n ∈ N. We can simplify this expression as follows:

xn xn − 2xn − x2n −xn − x2n xn + x2n

.xn+1 − xn = − xn = = =− .
2 + xn 2 + xn 2 + xn 2 + xn

xn + x2n
Since by item a), xn > 0 for all n ∈ N, we have xn + x2n > 0, which implies that − < 0. Therefore,
2 + xn
xn+1 − xn < 0.
c) Since for all n ∈ N we have xn > 0 (item a)) and (xn ) is a decreasing sequence (item b)), we have
0 < xn ≤ x1 , that is, 0 < xn ≤ a, for all n ∈ N. In other words, the sequence of general term xn is
bounded. Because every monotone and bounded sequence is convergent, (xn ) is also convergent.
Let us assume that lim xn = b. Note that b ≥ 0 because xn > 0 for all n ∈ N.
xn
If the sequence converges to b, we also have lim xn+1 = b. Furthermore, the sequence is convergent,
2 + xn
since it is the quotient of two convergent sequences where the denominator never becomes zero and has a
limit different from zero.
56 Sequences of Real Numbers

Thus, we have

xn lim xn b
lim xn+1 = lim ⇔ lim xn+1 = ⇔b=
2 + xn lim(2 + xn ) 2+b

. b 2b + b2 − b b2 + b
⇔ b− =0⇔ =0⇔ = 0 ⇔ b2 + b = 0 (b = −2)
2+b 2+b 2+b

⇔ b(b + 1) = 0 ⇔ b = 0 ∨ b = −1.

However, by Theorem 1.1.5, b ≥ 0, which means that lim xn = 0.

11. a) We want to show that xn+1 − xn ≥ 0 for any n ∈ N0 . Now, if n = 0, we have x1 − x0 = a > 0. If n ≥ 1,
then
.xn+1 − xn = xn + x2n−1 − xn = x2n−1 ≥ 0.

b) As we saw in item a) that the sequence is increasing, we have xn ≥ x1 = a > 0, ∀n ∈ N.

c) Suppose there exists b ∈ R such that b = lim xn . Then, all its subsequences have limit b and

.b = lim xn+1 = lim (xn + x2n−1 ) = b + b2

from where we conclude that b = 0.

d) If the sequence is bounded, it will converge because it is monotonic. We have shown in the previous item
that if the sequence has a limit, it should be zero. This is not possible, however, since it is an increasing
sequence of positive numbers. Therefore, we can conclude that it is unbounded. Since the sequence is
increasing, it is bounded below by its first term. So, it has no upper bound. As a result, we can deduce that
lim xn = +∞.

12. a) For n = 1, x1 = 2 > 0.

Induction hypothesis: xn > 0.

Induction thesis: xn+1 > 0.

Proof: Using the given formula xn+1 = x2n + x1 , we can see that x2n and x1 are both greater than zero
n n
since xn is positive by the induction hypothesis. Therefore, the sum of these two quantities, which is xn+1 ,
is also greater than zero.

By the Principle of Induction, we have proven that xn > 0, ∀n ∈ N.

√
b) For n = 1, x1 = 2 > 2.
√
Induction hypothesis: xn > 2.
√
Induction thesis: xn+1 > 2.
√  √ 2
√xn 1 √ x2n + 2 − 2 2xn xn − 2
Proof: By definition, xn+1 − 2 = + − 2= = . As by induction
√ √ 2 xn 2xn √ 2xn
hypothesis xn > 2, then (xn − 2)2 > 0, and we conclude that xn+1 − 2 > 0.
√
Then, using the Principle of Induction, we proved that xn > 2, ∀n ∈ N.
1.2. Solved Exercises 57

c) Let us analyze the difference xn+1 − xn to determine whether the sequence is increasing or decreasing:

xn 1 xn 1 −x2n + 2
.xn+1 − xn = + − xn = − + = .
2 xn 2 xn 2xn
√
By item b), xn > 2, ∀n ∈ N; therefore, −x2n + 2 < 0, ∀n ∈ N. Then xn+1 − xn < 0, ∀n ∈ N, thus
proving that the sequence is decreasing.
first term is the maximum of the set of its terms; therefore, x1 = 2 ≥ xn ,
d) If a sequence is decreasing, its √
∀n ∈ N, and (xn ) is bounded: 2 < xn ≤ 2, ∀n ∈ N. Thus, we can conclude that (xn ) is convergent as it
is monotonic and bounded.
e) Let a = lim xn . All subsequences of (xn ) have limit a and

xn 1 a 1 a2 + 2
.a = lim xn+1 = lim + = + = .
2 xn 2 a 2a

a2 + 2 √ √ √
Solving the equation a = , we get a = − 2 or a = 2. As xn > 2, ∀n ∈ N, we conclude that
√ 2a
a = 2.
√
13. a) We will show by induction that xn − 3 > 0, ∀n ∈ N.
√
If n = 1, then x1 = 3 > 3, and the proposition is satisfied.
√
Induction hypothesis: xn − 3 > 0.
√
Induction thesis: xn+1 − 3 > 0.
√
√ x2n + 3 √ (xn − 3)2 √
Proof: By definition, xn+1 − 3 = − 3 = . We know that xn > 3; therefore,
√ 2 xn 2xn
xn+1 − 3 > 0.
√
Then, using the Principle of Induction, we proved that xn − 3 > 0, ∀n ∈ N.
b) We intend to show that the sequence is decreasing, that is, xn+1 − xn ≤ 0, ∀n ∈ N:

x2n + 3 −x2n + 3
.xn+1 − xn = − xn = .
2 xn 2xn
√
By item a), xn > 3, ∀n ∈ N; therefore, −x2n + 3 < 0, ∀n ∈ N. Then xn+1 − xn < 0, ∀n ∈ N, that is to
say, the sequence is decreasing.
c) If a sequence is decreasing, its first term is the maximum
√ of the set of terms of the sequence; therefore,
x1 = 3 ≥ xn , ∀n ∈ N. We have (xn ) bounded: 3 < xn ≤ 3, ∀n ∈ N. We can conclude that (xn ) is
convergent because it is monotonic and bounded.
d) Let a = lim xn . All subsequences of (xn ) have limit a and

x2n + 3 a2 + 3
.a = lim xn+1 = lim = .
2 xn 2a

a2 + 3 √ √ √
Solving the equation a = , we get a = − 3 or a = 3. As xn > 3, ∀n ∈ N, we conclude that
√ 2a
a = 3.
58 Sequences of Real Numbers

25n 25
14. a) In order to prove by definition that lim = , we must prove that the proposition
3n + 1 3
% %
% 25n 25 %%
.∀ε > 0 ∃ p ∈ N : n > p ⇒ %% − <ε
3n + 1 3 %

is true, that is, we need to show that for each positive real ε there is at least one order after which, for every
natural n, % %
% 25n 25 %%
. %% − < ε.
3n + 1 3 %
Let ε > 0. Solving the inequality for n, we obtain
% % % %
% 25n 25 %% % 75n − 75n − 25 % 25 25 − 3ε
. %% − % < ε ⇔ %% %<ε⇔
% <ε⇔n> .
3n + 1 3 3(3n + 1) 3(3n + 1) 9ε

25 − 3ε 25 − 3ε
If we choose p ∈ N such that p ≥ , then, as n > p, we get n > , that is,
9ε 9ε
% %
% 25n 25 %%
. %% − < ε.
3n + 1 3 %

Figure 1.16: If ε = 0.2, then −ε < 25n

3n+1
− 25
3
< ε if n > 13

Suppose, for instance, that we want to calculate the smallest order p from which
% %
% 25n 25 %%
. %% − < 0.2.
3n + 1 3 %

25 − 3ε
Substituting ε by 0.2 in the expression , we get the value 13.5; therefore, as p ∈ N, we can choose
9ε
p = Int(13.5) = 13. Thus, if n > 13, the desired result follows. This case is illustrated in Fig. 1.16.
√
n
b) In order to prove by definition that lim √ = 1, we should prove that the proposition
n+1
% √ %
% n %
.∀ε > 0 ∃ p ∈ N : n > p ⇒ %% √ − 1%% < ε
n+1
1.2. Solved Exercises 59

is true, that is, we need to show that for each positive real ε there is at least one order after which, for every
natural n, % √ %
% n %
. %% √ − 1%% < ε.
n+1
Let ε > 0. Solving the inequality for n, we obtain
% √ % %√ √ %
% n % % n − n − 1% √ 1−ε
. %% √ − 1%% < ε ⇔ %% √ %<ε⇔ √ 1 <ε⇔ n> .
n+1 n+1 % n+1 ε
If 1 − ε ≤ 0, the last inequality is true regardless of the choice of p ∈ N. If 1 − ε > 0, then the previous
inequalities are still equivalent to
 
1−ε 2
.n > .
ε
   
1−ε 2 1−ε 2
If we choose p ∈ N such that p ≥ , then, as n > p, it follows that n > , that is,
ε ε
% √ %
% n %
. %% √ − 1%% < ε.
n+1

√
n
Figure 1.17: If ε = 0.2, then −ε < √
n+1
− 1 < ε if n > 16

Suppose, for example, that we want to calculate the smallest order p from which
% √ %
% n %
. %% √ − 1%% < 0.2.
n+1
 
1−ε 2
Substituting ε by 0.2 in the expression , we get the value 16; therefore, we can choose p = 16.
ε
Thus, if n > 16, the desired result follows. This case is illustrated in Fig. 1.17.

√
2 n + 5−n
c) We intend to prove, using the definition, that lim √ = 2. For this, we must prove that the
n+1
proposition % √ %
% 2 n + 5−n %
.∀ε > 0 ∃ p ∈ N : n > p ⇒ %% √ − 2%% < ε
n+1
60 Sequences of Real Numbers

is true, that is, we need to show that for each positive real ε there is at least one order after which, for every
natural n, % √ %
% 2 n + 5−n %
. %% √ − 2%% < ε.
n+1
Let ε > 0.
% √ % % √ √ % % −n %
% 2 n + 5−n % % 2 n + 5−n − 2 n − 2 % % % −n
. %% √ − 2%% < ε ⇔ %% √ % < ε ⇔ % 5√ − 2 % < ε ⇔ 2√− 5 < ε.
n+1 n+1 % % n+1 % n+1

√
2 n+5−n
Figure 1.18: If ε = 0.45, then −ε < √
n+1
− 2 < ε if n > 11

However,
2 − 5−n 2
. √ < √ , ∀n ∈ N;
n+1 n+1
therefore, if
2
.√ < ε,
n+1
then
2 − 5−n
. √ < ε.
n+1
Solving the inequality for n, we obtain
2 √ 2−ε
.√ <ε⇔ n> .
n+1 ε

If 2 − ε ≤ 0, the last inequality is true regardless of the choice of p ∈ N. If 2 − ε > 0, then the previous
inequalities are still equivalent to
 
2−ε 2
.n > .
ε
   
2−ε 2 2−ε 2
If we choose p ∈ N such that p ≥ , then, as n > p, it results in n > , that is,
ε ε
% √ %
% 2 n + 5−n %
. %% √ − 2%% < ε.
n+1
1.2. Solved Exercises 61

Suppose, for example, that we want to calculate an order p from which

% √ %
% 2 n + 5−n %
. %%√ − 2%% < 0.45.
n+1


2−ε 2
Substituting ε by 0.45 in the expression , we get the value (3, 4)2  11.86; therefore, as p ∈ N,
ε
we can choose p = Int(11.86) = 11. Thus, if n > 11, the desired result is obtained. This case is illustrated
in Fig. 1.18.
 
sin(n)
d) In order to prove by definition that lim 1 + = 1, we must prove that the proposition
n
%  %
% sin(n) %
.∀ε > 0 ∃ p ∈ N : n > p ⇒ %% 1 + − 1%% < ε
n

is true, that is, we need to show that for each positive real ε there is at least one order after which, for every
natural n, %  %
% sin(n) %
. %% 1 + − 1%% < ε.
n

Let ε > 0. %  % % %
% % % sin(n) %
. %%
1+
sin(n)
− 1%% < ε ⇔ %% % < ε ⇔ |sin(n)| < ε.
n n % n

sin(n)
Figure 1.19: If ε = 0.05, then −ε < 1 + n
− 1 < ε if n > 20

But
|sin(n)| 1
. ≤ , ∀n ∈ N;
n n
therefore, if
1
. < ε,
n
we have
|sin(n)|
. < ε.
n
62 Sequences of Real Numbers

Solving the inequality for n, we obtain

1 1
. <ε⇔n> .
n ε
1 1
If we choose p ∈ N such that p ≥ , then, as n > p, it results in n > , that is,
ε ε
|sin(n)|
. < ε.
n

Suppose, for example, we want to calculate an order p from which

%  %
% sin(n) %
. %% 1 + − 1%% < 0.05.
n
1
Substituting ε by 0.05 in the expression , we get the value 20; therefore, as p ∈ N, we can choose p = 20.
ε
Thus, if n > 20, the desired result is obtained. This case is illustrated in Fig. 1.19.
e) To prove that lim n3 = +∞ by definition, we need to show that the following proposition is true:

.∀L ∈ R+ ∃ p ∈ N : n > p ⇒ n3 > L.

In other words, we must show that for each positive real L, there is at least one order after which, for every
natural n, n3 > L.
√
Let L > 0. √ Solving the inequality for n, we have n3 > L ⇔ n > 3 L. Therefore, if we choose p ∈ N such
3 3
that p > L, we have proven that lim n = +∞.
√
In Fig. 1.20 you can see that by choosing L = 1000, then from n > p = 3 1000 = 10, all terms of the
sequence are greater than L.

Figure 1.20: If L = 1000, then n3 > L if n > 10

n2
f) In order to prove by definition that lim = +∞, we must prove that the proposition
n+1

n2
.∀L ∈ R+ ∃ p ∈ N : n > p ⇒ >L
n+1
is true, that is, we must show that for each positive real number L, there is at least one order after which,
for every natural number n,
n2
. > L.
n+1
1.2. Solved Exercises 63

n2
Figure 1.21: If L = 10, then n+1
> L if n > 10

Let L > 0. We have

 √  √ 
n2 L− L2 + 4L L+ L2 + 4L
. > L ⇔ n2 − Ln − L > 0 ⇔ n− n− > 0.
n+1 2 2
√
L+ L2 + 4L
Because L > 0 and n ∈ N, we obtain from the previous inequality that n > . Therefore, if
 √  2
L + L2 + 4L n2
we choose p = Int , then for n > p, we get > L.
2 n+1
In Fig. 1.21, you can see that if we choose L = 10, then from
 √ 
10 + 140
.n > p = Int = 10
2

all the terms of the sequence are greater than L.

15. Let us denote Nn as the number of segments, Ln as the length of each segment, and ln as the length of the
curve in the nth iteration, that is, after repeating n times the process of constructing the curve. We will prove by
induction that Nn = 4n . If n = 1, we have four segments by construction, so N1 = 41 = 4. Suppose (induction
hypothesis) that Nn = 4n . Then Nn+1 = Nn × 4 since each segment originates four new segments. Substituting
the expression of Nn , we get Nn+1 = 4n × 4 = 4n+1 .
1
In each iteration, we divide each segment by 3. During the first iteration, the length of each segment is L1 = .
 n  n+1 3
1 1
Assuming (induction hypothesis) that Ln = , we can prove that Ln+1 = as follows:
3 3
 1 n  n+1
Ln 1
.Ln+1 = = 3 = .
3 3 3
 n  4  n
The length of the curve in the nth iteration is ln = Nn × Ln . We know that Nn = 4n , so ln = 4n 13 = 3 .
The length L of the Koch curve is determined by the limit of ln when n → +∞. Then, by Theorem 1.1.10,
 n
4
.L = lim = +∞.
3
64 Sequences of Real Numbers

1.3 Proposed Exercises

1. Prove by definition that lim un = +∞ in each item b) Prove that n! grows faster than en .
listed below: c) Arrange the following sequences in descending
√ order in terms of the speed of convergence:
a) un = n c) un = n
√
.2 n, 10 n, 2n , en , n!, log(n), n3 , nn .
b) un = n2 d) un = 2n

2. By definition, prove that the following sequences 6. Let (un ) and (vn ) be two null sequences such that
(un ) are null sequences, that is, that lim un = 0: vn = 0, ∀n ∈ N. We say that (un ) is of higher
un
order than (vn ) if lim = 0. Order the following
1 1 vn
a) un = c) un = √ null sequences:
n n
1 1 1 1 1 1 1 1
1 1 . , √ , , , , , , .
b) un = 2 d) un = n 2n 10 n 2n en n! log(n) n3 nn
n 2
3. Evaluate the limits of the following sequences and 7. Calculate the limit of the sequences with general
provide an argument for the calculations: terms:
    2
1−n n + 3 2n 5n + 5 n
a) a) c)
5n + 3 n+1 2n + 1
n2 + 2  n  
b) n+5 3 n
3n + 1 b) d) 1 − 2
2n + 1 n
n2 + 3n n2 − 1
c) − 8. Evaluate, if it exists:
n+2 n
3n 1 
d) a) lim n
(n + 1)!
4n3 + 1 2n
−n3 + 2 1 
e) b) lim n n(n + 1) · · · 2n
4n3 − 7 n

4. Find the limits of the following sequences and justify n!
the calculations: 9. Determine p ∈ R such that lim n
= 3.
(p n)n
√
n
a) 10. Find the limits of the following sequences and pro-
4n + 1
√ vide a clear explanation of the reasoning behind each
n calculation:
b) 1 √
2
− n  n
  a) cos(x) , x ∈ R
c) n2 + 1 − n2 − 1  
 1
√ √  1 b) cos2 (n) sin
d) n+1− n n+ n
2
n (n − 1) (n − 2) (n − 3)
1 1 1 c)
e) √ +√ + ··· + √ (n + 1) (n + 2) (n + 3)
2
n +1 2
n +2 2
n +n   
2 n
5. We say that the sequence (un ) grows faster than the d) n n!
un n
sequence (vn ) if → +∞.
vn 1 1 1
e) √ +√ + ··· + √
a) Prove that nn grows faster than n!. n2 + 1 n2 + 2 n2 + 2n + 1
1.3. Proposed Exercises 65
 n 
1 n+1 15. Consider the sequence of real numbers defined re-
1−
n
f)
n n cursively

g) n
(n + 1)! − n! ⎧
⎪
⎨ x0 = 1
1 1 1  
h) √ + √ + ··· + √ .
⎪
1 2
n n+1 2n ⎩xn+1 = xn + , ∀n ∈ N0 .
2 xn
1 1 1
i) + + ··· +
n2 (n + 1)2 (2 n)2
n n n a) Show by induction that
j) √ +√ + ··· + √
n4 + 1 n4 + 2 n4 + n √
. 2 ≤ xn ≤ 2, ∀n ∈ N.
11. For each item, provide, when possible, examples
of sequences u, v, and w such that un → +∞, b) Show that xn ≥ xn+1 , ∀n ≥ 1.
vn → −∞, and wn → 0.
c) Show that the sequence is convergent and cal-
a) un + vn → 1 culate its limit.

b) un + vn → −∞ 16. The sequences (un ) and (vn ) verify the following

c) un + wn → 1 conditions:

d) un × wn → 0 (i) ∀n ∈ N 0 < un < vn .

e) vn × wn → +∞ (ii) (vn ) is decreasing.

un
f) → −1 Decide whether the following statements are true or
wn
false and give a justification for the answers provided.
12. Let (xn ) and (yn ) be two convergent sequences of
real numbers such that xn → x and yn → y. Show a) (vn ) is convergent.
that the sequence of general term zn = min{xn , yn } b) (un ) is convergent.
converges and that zn → min{x, y}.
c) (un ) is decreasing.
13. Consider the sequence (un ) defined recursively by
⎧ 17. Determine the upper and lower limits of the se-
⎨u1 = 1, quences with the following general terms and provide
⎩un = un−1 − 1 ,
.
∀n > 1. a reason for the answers:
2 n
a) n(−1)
Explain why (un ) is convergent and evaluate its nπ
limit. b) cos
3
√ √
14. Consider the sequence (un ) defined recursively c) n − (−1)n n − 1
nπ
⎧ d) sin
⎨ u1 = 5 4
√ √
. 5un − 4 e) n − (−1)n n − 1
⎩un+1 = ∀n ≥ 1.
un 1 nπ nπ n
f) 2 cos + cos
n 10 2
(−1)n n2 + 3
a) Prove by induction that ∀n ∈ N, un > 4. g)
n+1
b) Prove that the sequence is convergent. nπ
h) sin +a , a∈R
2
c) Show that 4 is the infimum of the set of terms  n
1 1  
of the sequence. i) + + 2 n (−1)n 3 + 3
3 2n
66 Sequences of Real Numbers
 
(−1)n+3 − (−1)n n3 + 2 Hint:
j)
3n + 1 1
(i) Show that vn = x2n − 2 satisfies 0 ≤ vn ≤ .
18. Show that the following are Cauchy sequences in Q: 4n
x2n − x2m
    (ii) Use the relation xn − xm = .
1 1 xn + xm
a) b)
n2 2n 22. The Sierpinski carpet is a set constructed as follows:
19. Show that the sequence of general term Consider a square, R, with side 1. Divide R into nine
squares of equal area and remove the interior of the
1 1 middle square. Repeat the process in the remain-
.1 + + ··· +
2 n ing eight squares: Divide each one into nine squares
of equal area and remove the interior of the middle
is not a Cauchy sequence in Q.
square. Let Rn be the region that remains after this
20. Consider the sequence of general term process has been performed n times. The Sierpinski
carpet, S, is the set of all points in R that are not
n+1 removed by any of the previous operations, that is,
.un = .
n+2 S consists of the points in R that belong to Rn for
any n ≥ 1. The figure illustrates the first three steps
Check the sequence for convergence using the defi-
of this process.
nition of a Cauchy sequence.

21. Show that the sequence

⎧
⎪ 3
⎨x 1 = ,
. 2
⎪ x 1
⎩xn+1 = n + , ∀n ∈ N
2 xn

is a sequence in Q such that x2n → 2. Use this result

a) Calculate the area An of Rn , n ≥ 1.
to show that (xn ) is a Cauchy sequence in Q that
does not converge in Q. b) Show that lim An = 0.
 Numerical Series 2

The operation of adding two numbers extends without any difficulty to a

finite number of terms, even if this number is very large. However, consid-
ering an infinite number of terms raises more delicate issues addressed in
this chapter.

2.1 Generalization of the Addition Operation

Addition (or sum) is the operation that corresponds to each pair of real
numbers with another real number, according to specific rules. This opera-
tion satisfies certain properties and can be generalized to a finite number of
terms while maintaining all properties. The definition of the sum of a finite
number of terms is made recursively as follows:
⎧
⎪
⎪ a , if n = 1
n ⎨ 1
n−1 
. ai = 
⎪
⎪
i=1 ⎩ ai + an , if n > 1.
i=1
We can now consider generalizing the notion of sum to an infinite number
of terms, .a1 , a2 , . . . , an , . . .
If there is an order p from which all terms of the sequence are zero, the sum
of all terms equals the sum of the first p terms:
 
p
. ai = ai .
n∈N i=1

n
If there is a subsequence of nonzero terms, let us consider .Sn = ai , the
i=1
sum of the first n terms. It is natural to consider the sum of all terms as
the limit of the sequence .(Sn ) if it exists and is finite.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 67

A. Alves de Sá, B. Louro, Sequences and Series,
https://doi.org/10.1007/978-3-031-67202-6 2
68 Numerical Series

If the terms of the sequence .(an ) are all positive, it might seem at first
glance that .(Sn ) is not convergent. In fact, assuming that the sum of an
infinite number of positive terms is a real number is not an intuitive concept.
In this case, intuition fails precisely because we intend to generalize to infinity
a concept, that of sum, which we have intuitive for a finite number of terms.
It is common for intuition to deceive us in cases of “passing” from finite to
infinite.
Zeno of Elea (490–435
It is true that we will not always be able to find the limit of the sequence
B.C.) Greek philosopher .(Sn ), so, by this process, we will not be able to find the sum of an infinite

and disciple of Par- number of terms if it exists. Nevertheless, we are interested in knowing how
menides. Most of his the sequence .(an ) should be so that a real number, the sum of all its terms,
teaching records have
come to us through
is associated with this sequence.
secondary sources, notably
through Aristotle. Zeno’s One of the historical examples illustrating these difficulties in dealing with
great fame came from the concept of an infinite sum of terms is one of the famous Paradoxes of
his paradoxes. These Zeno, which we summarize as follows:
paradoxes aimed to defend
his master Parmenides’ A walker wants to move from one place to another at a constant speed
theory that reality was (Fig. 2.1). In T minutes, he covers half of the distance. He still has the
unique, immutable, and
immobile so that change,
other half to go. In .T /2 minutes, he will cover half of what remains (.1/4
movement, time, and of the total distance). Moreover, now he has .1/4 of the distance left. He
plurality would be nothing will cover half of the rest (.1/8 of the total) in .T /4 minutes. This argument
more than illusions. Par- may be repeated indefinitely. The walker spends
menides’ theory presented
conclusions that contra- T T T T
dicted what the senses T+
. + + + + ...
2 4 8 16
conveyed; Zeno’s goal was
not to present evidence minutes on his journey.
that reinforced the theory
Given that it is the sum of infinite positive terms, we are inclined to assume
itself but to show that
opposing theories also that the sum is infinite. The walker will not be able to reach his goal because
led to contradictions. it will take an infinite amount of time. This reasoning contradicts intuition
(Source of image: Jan and practical life, which is called a paradox.
de Bisschop, Paradig-
mata graphices variorum This is due to the lack of understanding of the concept of an infinite sum
artificum, Rijksmuseum.) of terms (more precisely, of series), which came to be fully understood only
in the 19th century. In this case, the walker will reach his goal in time 2T ,
as intuition tells us.

Figure 2.1: The walker’s path in Zeno’s Paradox

2.2. Definition of Series: Convergence – General Properties 69

2.2 Definition of Series: Convergence – General

Properties

Definition 2.2.1 Let (an ) be a sequence of real numbers. The series

generated by an is the sequence (Sn ) defined as follows:

S1 = a1
S2 = a1 + a 2
S3 = a1 + a 2 + a 3
. .. .
.
Sn = a1 + a2 + a3 + · · · + an
..
.

To represent a series, we can use any of the following notations:

∞
 
. an , an , a1 + a2 + a3 + · · ·
n=1

The numbers a1 , a2 , . . . , are called terms of the series, an is said to be

the general term of the series, and the sums S1 , S2 , . . . , are called partial
sums.

Definition 2.2.2 The series an is convergent if the limit


n
. lim Sn = lim ai
i=1

exists and is finite. If this limit does not exist or is not finite, the series is
divergent.

In the case of convergence, the sum of the series is the value S of the limit
of (Sn ), that is,
∞

.S = lim Sn = an .
n=1
∞
Note: The identification of a series with the symbol n=1 an is a misuse
of language because it is the identification of a series with its sum when it
exists. However, despite being a common misuse of language, it has been
found to be a valuable and harmless convention.
70 Numerical Series

Example 2.2.1 An important example is the geometric series. A geo-

metric series is a series generated by a geometric sequence: If (an ) is a
geometric sequence of ratio r = 1 and a1 = 0, then


n 
n
1 − rn
Sn =
. ai = a1 ri−1 = a1 · .
i=1 i=1
1−r

We know that (Sn ) is convergent if and only if |r| < 1, so the geometric
series is convergent if and only if the absolute value of the ratio of the
geometric sequence that generated it is less than 1. In this case
∞
 a1
. an = .
n=1
1−r

The series in Zeno’s Paradox (see Sect. 2.1) is geometric with a1 = T and
r = 1/2; it is convergent with the sum 2T .
If r = 1, the general term is constant; that is,
∞
 ∞

. an = a1 ,
n=1 n=1

and thus, Sn = na1 and, if a1 = 0, the series will diverge.

Example 2.2.2 The series

∞
 ∞ n
3n 1 3
. (−1)n = −
n=1
e2n+1 n=1
e e2

3 3
is geometric with ratio r = − 2 and first term − 3 . Since −1 < r < 1,
e e
the series converges and its sum is

− e33 −3
. = .
1 − (− e32 ) e (e2 + 3)

Example 2.2.3 We can write the rational number 2.325 = 2.3252525 . . .

in the form of a fraction using a geometric series.
23 25 25 25
2.325 = 2.3 + 0.025 + 0.00025 + . . . =
. + + 5 + 7 + ···
10 103 10 10
2.2. Definition of Series: Convergence – General Properties 71

1
After the first term, we have a geometric series with ratio r = 2 and first
10
25
term 3 . Therefore,
10
25
23 3 1151
2.325 =
. + 10 1 = .
10 1 − 102 495

∞
1
Example 2.2.4 Consider the series √ ; let us study the sequence of
n=1
n
its partial sums and the corresponding limit:

S1 = 1
1
S2 = 1 + √
2
1 1
S3 = 1 + √ + √
. 2 3
..
.
1 1 1
Sn = 1 + √ + √ + · · · + √
2 3 n
..
.

As
1 1 1 1 1 1 1 n √
1 + √ + √ + ··· + √ ≥ √ + √ + √ + ··· + √ = √ = n
.
2 3 n n n n n n
√
and lim n = +∞, we conclude by Theorem 1.1.2 that the sequence (Sn )
has limit +∞. The series under study is divergent.

∞
1
Example 2.2.5 Consider the series , commonly known as the har-
n=1
n
monic series. By mathematical induction, we can prove that the subse-
quence (S2n ) of the partial sums of this series satisfies the inequality
n
S 2n ≥ 1 +
. , ∀n ∈ N.
2
For n = 1, the proposition is true:
1 1
S2 = 1 +
. ≥1+ .
2 2
72 Numerical Series

Assuming that the proposition is valid for n, we prove that it is also valid
for n + 1.
1 1 n 1 n+1
S2n+1 = S2n +
. + · · · + n+1 ≥ 1 + + 2n n+1 = 1 + .
2n +1 2 2 2 2
Thus, the desired inequality is proved using the method of induction. By
Theorem 1.1.2, we can conclude that the sequence (S2n ) has limit +∞,
implying that the sequence of partial sums of the harmonic series does not
have a finite limit. Therefore, the harmonic series is divergent.

Figure 2.2 shows the sequence of partial sums of the harmonic series and its
general term. It illustrates how the sum of an infinite number of increasingly
smaller positive quantities can have a value as large as desired.

Figure 2.2: The sequences ∞

 1
1
an = n and Sn = 1 + · · · + n
1
Example 2.2.6 Consider the series . By observing that
n=1
n(n + 1)

1 1 1
. = − ,
n(n + 1) n n+1
we can write the sequence of partial sums as follows (see Fig. 2.3):

1
S1 = 1 −
2
1 1 1 1
S2 = 1 − + − = 1 −
2 2 3 3
Figure 2.3: The sequences 1 1 1 1
. S3 = 1 − + − =1−
1 1
an= n(n+1) and Sn= 1− n+1
3 3 4 4
..
.
1
Sn = 1 −
n+1
..
.
As lim Sn = 1, the series is convergent and its sum is 1:
∞
 1
. = 1.
n=1
n(n + 1)

Example 2.2.7 The sequence of partial sums of the series

∞
 ∞

n
. log = log(n) − log(n + 1)
n=1
n+1 n=1
2.2. Definition of Series: Convergence – General Properties 73

is the sequence

S1 = log(1) − log(2) = − log(2)

S2 = − log(2) + log(2) − log(3) = − log(3)
S3 = − log(3) + log(3) − log(4) = − log(4)
.
..
.
Sn = − log(n + 1)
..
.
As lim − log(n + 1) = −∞, the series diverges.

∞
 1
Example 2.2.8 The general term of the series can be ex-
n=1
n2 + 3n
1 1 1
pressed as − . The sequence of partial sums can be con-
3 n n+3
structed using this expression:

1 1
S1 = 1−
3 4
1 1 1 1 1 1 1 1 1
S2 = 1− + − = 1− + −
3 4 3 2 5 3 4 2 5
1 1 1 1 1 1 1
S3 = 1− + − + −
3 4 2 5 3 3 6
1 1 1 1 1 1
= 1− + − + −
3 4 2 5 3 6
.

1 1 1 1 1 1 1 1 1
S4 = 1− + − + − + −
3 4 2 5 3 6 3 4 7
1 1 1 1 1 1 1 1
= 1− + − + − + −
3 4 2 5 3 6 4 7
1 1 1 1 1 1
= 1+ − + − −
3 2 5 3 6 7
1 1 1 1 1 1 1 1 1
S5 = 1+ − + − − + −
3 2 5 3 6 7 3 5 8
74 Numerical Series

1 1 1 1 1 1 1 1
= 1+ − + − − + −
3 2 5 3 6 7 5 8
1 1 1 1 1 1
= 1+ + − − −
3 2 3 6 7 8
. ..
.
1 1 1 1 1 1
Sn = 1+ + − − −
3 2 3 n+1 n+2 n+3
..
.
1 1 1 11
As lim Sn = 1+ + = , the series is convergent and
3 2 3 18
∞
 1 11
.
2 + 3n
= .
n=1
n 18

The last three examples are particular cases of a type of series called tele-
scopic series. These are series whose general term an can be expressed in
the form αn − αn+k , with k ∈ N:
∞

. (αn − αn+k ), (2.1)
n=1
allowing us to determine the expression of the sequence of partial sums by
canceling consecutive terms and, consequently, in case the sequence con-
verges to easily calculate the sum of the series. These series are convergent
if and only if lim vn , where vn = αn+1 + · · · + αn+k , exists and is finite.
In the particular case of existing, finite, lim αn , we have:
∞
 
k
. (αn − αn+k ) = αi − k a, (2.2)
n=1 i=1
where a = lim αn . In fact, the sequence of partial sums is given by

n 
n 
n

. Sn = (αi − αi+k ) = αi − αi+k

i=1 i=1 i=1

= α1 + · · · + αk + αk+1 + · · · + αn
−(αk+1 + · · · + αn + αn+1 + · · · + αn+k )
. = α1 + · · · + αk − (αn+1 + · · · + αn+k )

k 
k
= αi − αi+n
i=1 i=1
2.2. Definition of Series: Convergence – General Properties 75

If (αn ) is convergent, then lim αi+n exists and lim αn = lim αi+n from
which it is concluded that
 

k 
k 
k
. lim Sn = αi − lim αi+n = αi − k a.
i=1 i=1 i=1

The following theorem establishes a necessary but not sufficient condition

for the convergence of a series. Therefore, it is primarily used to determine
whether a series is divergent. If the general term of a series is not a null
sequence, the series will be divergent.

Theorem 2.2.1 If the series an is convergent, then (an ) is a null

sequence.


n
Proof: Since the series is convergent, the sequence Sn = ai is also
i=1
convergent. This is true for Sn−1 as well, and we have lim Sn = lim Sn−1 .
Then

. lim an = lim (Sn − Sn−1 ) = lim Sn − lim Sn−1 = 0.

∞
 n n
Example 2.2.9 The series is divergent because lim = 1.
n=1
n + 1 n + 1

∞
1
Example 2.2.10 Consider the series √ . Although
n=1
n

1
. lim √ = 0,
n

we cannot use Theorem 2.2.1 to determine if the series converges or di-

verges. However, in Example 2.2.4, we have already proved that this series
is divergent.
76 Numerical Series

Theorem 2.2.2 Let an and bn be convergent series with sums A

and B, respectively, and λ ∈ R. Then:

a) The series (an + bn ) is also convergent, and its sum is A + B:

∞
 ∞ ∞
. (an + bn ) = an + bn .
n=1 n=1 n=1

b) The series λan is convergent, and its sum is λA:

∞
 ∞
. λan = λ an .
n=1 n=1

Proof:
a) Let (Sn∗ ) and (Sn∗∗ ) be the sequences of partial sums of the series an
and bn , respectively. Because they are convergent with sums A and B,
we have
∗
. lim Sn = A and lim Sn∗∗ = B.
Let (Sn ) be the sequence of partial sums of the series (an + bn ), that is,


n 
n 
n
Sn =
. (ai + bi ) = ai + bi = Sn∗ + Sn∗∗ .
i=1 i=1 i=1

Then

. lim Sn = lim(Sn∗ + Sn∗∗ ) = lim Sn∗ + lim Sn∗∗ = A + B,

that is, the series (an + bn ) is convergent and has sum A + B.

b) Let (Sn∗ ) be the sequence of partial sums of the series an . By hypoth-
esis, lim Sn∗ = A. Let (Sn ) be the sequence of partial sums of the series
λan . Then
n n
.Sn = λai = λ ai = λSn∗ .
i=1 i=1

Thus,
. lim Sn = lim λSn∗ = λ lim Sn∗ = λA,
that is, the series λan is convergent and has sum λA.
2.2. Definition of Series: Convergence – General Properties 77

Notes:

1. From the proof of item a), it is evident that the given series may
diverge, and yet the series (an + bn ) can still converge. The proof
also indicates that if the sequences of partial sums have infinite limits
of the same sign, that is, both series are divergent, and the sequence
of partial sums will be divergent. The same thing will happen if one
of the series is convergent and the other divergent. If (Sn∗ ) and (Sn∗∗ )
have infinite limits with opposite signs, the series (an + bn ) may
be convergent or divergent because an indeterminate form appears in
the calculation of the limit.

2. Based on the proof of part b), it can be concluded that if the value of
λ is not equal to 0, then the series λan will converge if and only if
the series an also converges. However, if the value of λ equals 0,
then the series λan will converge since all its terms are zero.

In order to better understand the previous theorem, we will examine an

example that involves the application of our knowledge about telescopic
series.

Example 2.2.11 Consider the series

∞
 1
. .
n=1
n(n + 3)(n + 6)

We can decompose the general term as follows:

1 1 1 1 1 1 1
. = − − − .
n(n + 3)(n + 6) 18 n n+3 18 n+3 n+6
The series
∞
 1 1
. −
n=1
n n+3
1
is telescopic with αn = and k = 3. As lim αn = 0, the series is conver-
n
1 1 11
gent, and its sum is α1 + α2 + α3 = 1 + + = .
2 3 6
The series
∞
1 1
. −
n=1
n+3 n+6
78 Numerical Series

1
is also telescopic with αn = and k = 3. In this case too, lim αn = 0;
n+3
1 1 1 37
therefore, the series is convergent and its sum is + + = .
4 5 6 60
According to the previous theorem, the given series is also convergent and
∞
 1
n=1
n(n + 3)(n + 6)
∞ ∞
1  1 1 1  1 1
.
= − − −
18 n=1 n n+3 18 n=1 n+3 n+6

1 11 1 37 73
= · − · = .
18 6 18 60 1080

Example 2.2.12 The series

∞ ∞ ∞ ∞
1  
n+1 n+1
5n+1 1 5 5 1
.
n+2
= · = and
n=1
8 n=1
8 8 8 n=1 8 n=1
n(n + 1)

5 52
are convergent. The first is geometric with ratio r = and first term 3 .
8 8
The second is telescopic, and as we saw in Example 2.2.6, it converges and
has a sum of 1. We can conclude that the series
∞
 5n+1 1
. +
n=1
8n+2 n(n + 1)

25 217
is convergent and its sum is +1= .
192 192

Theorem 2.2.3 (Cauchy’s Criterion) A series an converges if and

only if

∀δ > 0 ∃ p ∈ N : m > n > p ⇒ |an+1 + · · · + am | < δ.

.

Proof: As
m 
 
n 
 
.|an+1 + · · · + am | =  ai − ai  = |Sm − Sn |,
 
i=1 i=1
2.2. Definition of Series: Convergence – General Properties 79

∞

our aim is to prove that the series an converges if and only if
n=1

∀δ > 0 ∃ p ∈ N : m > n > p ⇒ |Sm − Sn | < δ,

.

that is, we want to prove that (Sn ) is a Cauchy sequence. However, by

definition, the series an converges if and only if (Sn ) is a convergent
sequence, and in R a sequence is convergent if and only if it is a Cauchy
sequence (see Theorem 1.1.20).

Example 2.2.13 In Example 2.2.5, we showed that the harmonic series,

1
n , is divergent. We will prove the same result using Theorem 2.2.3. If
the series is convergent, then for any given δ > 0, there would exist p ∈ N
such that |an+1 + · · · + am | < δ for all m > n > p. However, if we choose
m = n + n, then

|an+1 + · · · + am | = |an+1 + · · · + an+n |

1 1
. = + ··· +
n+1 n+n
1 1 1 1
≥ + ··· + =n· = .
n+n n+n 2n 2
1
Therefore, the condition of the theorem is not satisfied for δ < , which
2
means that the harmonic series is divergent.

Corollary 1 The convergence or divergence of a series does not depend

on the first p terms, whatever p ∈ N, that is, if an and bn are series
such that there exists p ∈ N satisfying an = bn , ∀n > p, then either both
series are convergent or both are divergent.

∞

Definition 2.2.3 The remainder of order p of the series an is the
n=1
series ∞
 ∞

r =
. p an+p = an .
n=1 n=p+1
80 Numerical Series

According to the previous corollary, the remainder of any order will also be
convergent if a series is convergent. The sum of the remainder of order p
of a convergent series gives us the error made by taking the partial sum Sp
as the approximate value of the sum of the series. This error is given by

∞
 ∞
 
p ∞

. an − Sp = an − an = an+p = rp .
n=1 n=1 n=1 n=1

Corollary 2 The convergence or divergence of a series does not change if

we suppress a finite, arbitrary number of terms.

The following theorem can be viewed as a generalization of the associative

property of addition to the case of convergent series.

Theorem 2.2.4 Let an be a convergent series and k1 , k2 , . . . , kn , . . .

be a strictly increasing sequence of elements of N. Let bn be the sequence
defined as follows:
⎧ k1
⎪ 
⎪
⎪
⎪
⎨ ai , if n = 1
i=1
.bn =
⎪
⎪ kn
⎪
⎪ ai , if n > 1.
⎩
i=kn−1 +1

∞
 ∞

Then, the series bn is convergent and bn = an .
n=1 n=1

Proof: The series an is convergent, then by Definition 2.2.2 there exists

and is finite the limit

n
. lim Sn = lim ai .
i=1

According to Theorem 1.1.15, every subsequence of (Sn ) is convergent and

has the same limit.
2.2. Definition of Series: Convergence – General Properties 81

∞
 
n

The series bn is convergent if and only if Sn = bi is convergent. But
n=1 i=1

 
n 
k1 
k2 
kn 
kn
Sn =
. bi = ai + ai + · · · + ai = a i = Sk n ,
i=1 i=1 i=k1 +1 i=kn−1 +1 i=1


that is, (Sn ) is a subsequence of (Sn ); therefore, it is convergent and has
the same limit:
∞
 ∞


. bn = lim Sn = lim Sn = an .
n=1 n=1

∞

Note: If the series an is convergent, then the theorem establishes that
n=1
the following “associative property” is valid:

a +a2 +· · ·+ak1 +· · ·+ak2 +· · · = (a1 +· · ·+ak1 )+(ak1 +1 +· · ·+ak2 )+· · ·

. 1

However, this property is not valid if the series is divergent. For instance,
∞

the series (−1)n is divergent because its general term does not tend to
n=1
zero. Nevertheless, (−1 + 1) + (−1 + 1) + · · · = 0.
82 Numerical Series

2.3 Alternating Series

Definition 2.3.1 A series is said to be alternating if its terms are alter-

natively positive and negative.

Gottfried Wilhelm Leib- Assuming that the first term of the series is positive, we can write
niz (1646–1716) was a
∞

German philosopher and
mathematician. He ob-
. (−1)n−1 an , an > 0, ∀n ∈ N.
tained a doctorate in Law n=1
when he was twenty years
old. He pursued a career
in Law and International
Theorem 2.3.1 (Leibniz’s Test) If (an ) is a decreasing sequence of pos-
Politics as an advisor to itive terms and lim an = 0, then the series (−1)n−1 an is convergent.
kings and princes. During
his countless travels,
Leibniz befriended the Proof: Consider the sequence (Sn ) of the partial sums of the series:
greatest intellectuals of
his time. He developed .Sn = a1 − a2 + a3 − · · · + (−1)n−1 an .
differential and integral
calculus simultaneously We will study the subsequences of even order terms and odd order terms.
and independently of the Let k ∈ N be arbitrary.
English mathematician
Isaac Newton and estab-
lished the foundations of
S2k = a1 − a2 + · · · + a2k−1 − a2k
.
dynamics. Leibniz was one S2k+1 = a1 − a2 + · · · + a2k−1 − a2k + a2k+1 .
of the last intellectuals to
master almost all of the We claim that the subsequence (S2k ) is increasing. This is because, as (an )
knowledge of his time.
is decreasing, we get:
(Source of image: Oil
portrait by Christoph
S2k+2 − S2k = a1 − a2 + · · · + a2k−1 − a2k + a2k+1 − a2k+2 −
Bernhard Francke (1665–
1729), Herzog Anton . −(a1 − a2 + · · · + a2k−1 − a2k )
Ulrich-Museum.)
= a2k+1 − a2k+2 ≥ 0.

Moreover, it is a bounded sequence since:

S2 ≤ S2k = a1 − [(a2 − a3 ) + (a4 − a5 ) + · · · + (a2k−2 − a2k−1 ) + a2k ] < a1 .

.

Hence, (S2k ) is monotonic and bounded, which implies that it is convergent.

From S2k+1 = S2k + a2k+1 , it follows that

. lim S2k+1 = lim S2k ,

k→+∞ k→+∞

since by hypothesis (an ) is a null sequence.

2.3. Alternating Series 83

As the subsequences of even and odd terms have the same limit, (Sn )
∞

converges. Therefore, the series (−1)n−1 an is convergent.
n=1

Note: Leibniz’s test is a particular case of a more general result known as

Dirichlet’s test. The Dirichlet’s test states the following:
n Peter Gustave Lejeune
If the sequence ( i=1 bi ) is bounded, and if (an ) is a decreasing sequence
∞ Dirichlet (1805–1859) was
of positive terms satisfying lim an = 0, then the series n=1 bn an is con- a German mathematician
vergent. who made valuable contri-
butions to number theory,
If we let bn = (−1)n in Dirichlet’s test, we obtain Leibniz’s test. analysis, and mechanics.
He was a professor at
the universities of Bres-
∞
 ∞
 lau (1827) and Berlin
n log(n) log(n)
Example 2.3.1 The series (−1) = (−1)n is alter- (1828–1855). In 1855, he
n n
n=1 n=2 succeeded Carl Friedrich
nating. Let us check the conditions for applying Leibniz’s test: Gauss at the University
of Göttingen. (Source of
log(n) log(x)
(i) lim = 0 (it is enough to note that lim is an indeter- image: Portrait Collection
n x→+∞ x of the German Museum,
∞
minate form of type ; we can calculate the limit using L’Hôpital’s Munich.)
∞
Rule: 
log(x) 1
. lim = lim = 0).
x→+∞ (x)  x→+∞ x
log(n)
(ii) > 0, ∀n > 1.
n
log(x)
(iii) The function f (x) = is a decreasing function on ]e, +∞[
x
because

log(x) 1 − log(x)
f  (x) =
. = < 0, ∀x > e.
x x2
log(n)
This implies that the sequence is decreasing for n ≥ 3.
n
Thus, we can conclude that the alternating series
∞
 log(n)
. (−1)n
n=3
n

is convergent. As this series differs from the initial series in one term, we
can affirm that the series under study is convergent.
84 Numerical Series

∞
 π
Example 2.3.2 The series cos(nπ) sin is alternating because
n=2
n
π
. cos(nπ) = (−1)n and an = sin > 0, ∀n > 1
n
π π
(note that 0 < ≤ , ∀n > 1). We can write the series in the form
n 2
∞ π
n
(−1) sin . Let us check the conditions of Leibniz’s test:
n=2
n
π
(i) lim sin = 0.
n
π
(ii) We have already seen that sin > 0, ∀n > 1.
n
π π
(iii) As we also justified earlier, 0 < ≤ , ∀n > 1; the sequence
π n 2  
π
is decreasing and the sine function is increasing on 0, , so
 n  π  2
sin is decreasing for n ≥ 2.
n
We can conclude that the alternating series is convergent.

∞
 1
Example 2.3.3 Consider the series (−1)nα
, α ∈ R.
n=1
n
If α ≤ 0, the series diverges because its general term does not tend to zero.
1
If α > 0, the series is convergent because an = α is a decreasing sequence
n
with positive terms and lim an = 0.

Not all alternating series meet the conditions of Leibniz’s test. In this
situation, the test is not applicable, and it is necessary to look for another
approach.

Example 2.3.4 Consider the alternating series

∞ ∞

(−1)n (−1)n 1 (−1)n
. √ 1+ √ = (−1)n √ + .
n=1
n n n=1
n n

1 (−1)n
Although an = √ + is a null sequence and an > 0, ∀n > 1, we
n n
cannot apply Leibniz’s test because (an ) is not decreasing. However, we can
2.3. Alternating Series 85

show that the given series is divergent because it is the sum of a convergent
 1 1
series, the series (−1)n √ , with a divergent series, the series
n n
(see Note 1 on page 77).

Theorem 2.3.2 Let (an ) be a decreasing sequence of positive terms such

that lim an = 0, and S be the sum of the series (−1)n−1 an . Then

0 ≤ (−1)n (S − Sn ) ≤ an+1 , ∀n ∈ N.
.

Proof: We know from the proof of the previous theorem that (S2k ) is
an increasing subsequence of (Sn ), and it has the same limit, S, as the
subsequence (S2k+1 ). Similarly, (S2k+1 ) is decreasing. This leads to the
following inequalities:

. S2k ≤ S and S ≤ S2k+1 , ∀k ∈ N.

From these inequalities, we can deduce that:

0 ≤ S2k−1 − S ≤ S2k−1 − S2k = a2k ,

.
0 ≤ S − S2k ≤ S2k+1 − S2k = a2k+1 .

Therefore:
0 ≤ S2k−1 − S ≤ a2k ,
.
0 ≤ S − S2k ≤ a2k+1 .
This implies that

0 ≤ (−1)2k−1 (S − S2k−1 ) ≤ a2k ,

.
0 ≤ (−1)2k (S − S2k ) ≤ a2k+1 .

From these last two inequalities, it follows that

0 ≤ (−1)n (S − Sn ) ≤ an+1 .
.

Corollary 1 Let (an ) be a decreasing sequence of positive terms such that

lim an = 0 and S be the sum of the series (−1)n−1 an . Then

|S − Sn | ≤ an+1 , ∀n ∈ N.
.
86 Numerical Series

Note: According to the previous corollary, it is evident that under the con-
ditions of Leibniz’s test, the absolute value of the error made in using a
partial sum as an estimation for the sum of an alternating series is, in abso-
lute value, always less than the absolute value of the first of the disregarded
terms.

Example 2.3.5 Consider the series

∞
 1
. (−1)n ,
n=1
n

which is known as the alternating harmonic series. According to Leibniz’s

test, this series is convergent, as an = n1 is a null decreasing sequence of
positive terms. If we use the partial sum S9 as an estimation of the sum of
the series, we will make an error that in absolute value is less than or equal
1
to , which is the value of a10 .
10
2.4. Absolute Convergence 87

2.4 Absolute Convergence

Theorem 2.4.1 If the series |an | is convergent, then the series an

is also convergent.

∞

Proof: By Theorem 2.2.3, the series |an | is convergent if and only if
n=1

∀δ > 0 ∃ p ∈ N : m > n > p ⇒ | |an+1 | + · · · + |am | | < δ.

.

Since
|an+1 + · · · + am | ≤ |an+1 | + · · · + |am |
.

and
| |an+1 | + · · · + |am | | = |an+1 | + · · · + |am |,
.

we have

∀δ > 0 ∃ p ∈ N : m > n > p ⇒ |an+1 + · · · + am | < δ,

.

∞

that is, the series an is convergent.
n=1

Note: It is worth noting that the converse of this theorem is not true. This
means that the series an can be convergent even if the series of modules,
|an |, is not. For instance, consider the harmonic series, which diverges,
and the alternating harmonic series, which converges. The harmonic series
is the series of modules of the alternating harmonic series.

Definition 2.4.1 A series an is absolutely convergent if the series

|an | is convergent. A series an is conditionally convergent if it is
convergent and the series |an | is divergent.

Definition 2.4.2 We say that the series bn is a rearrangement of the

series an , or that it is obtained by reordering its terms, if there exists
a bijection φ from N into N such that bn = aφ(n) .
88 Numerical Series

Theorem 2.4.2 If the series an is absolutely convergent, then any

series obtained by rearranging its terms is absolutely convergent and has
the same sum.

This theorem generalizes the commutative property of the usual addition to

absolutely convergent series. However, it should be noted that this property
is not valid for conditionally convergent series. By rearranging the terms of a
conditionally convergent series, we can obtain a series with a predetermined
sum and even a divergent series.

Theorem 2.4.3 Let an be a conditionally convergent series. Then:

a) There are bijections φ : N → N such that the series aφ(n) is

divergent.

b) For every real number k, there exists a bijection φ : N → N such

that the series aφ(n) is convergent and its sum is equal to k.

∞
 1
Example 2.4.1 Consider the alternating harmonic series, (−1)n−1
,
n=1
n
which we know to be conditionally convergent. Let us rearrange its terms
such that each positive term is followed by two negative terms. We will
obtain the following series:

1 1 1 1 1 1 1 1
1−
. − + − − + − − + ···
2 4 3 6 8 5 10 12
For this series, we have the following partial sums:

S1 = 1
 1
S2 = 1 −
2
 1 1 1 1 1 1 1
S3 = 1 − − = − = 1− = S2
. 2 4 2 4 2 2 2
 1 1 1
S4 = 1 − − +
2 4 3
 1 1 1 1
S5 = 1 − − + −
2 4 3 6
2.4. Absolute Convergence 89

 1 1 1 1 1 1 1 1 1 1
S6 = 1 − − + − − = 1− − + − −
2 4 3 6 8 2 4 3 6 8
1 1 1 1 1 1 1 1 1
= − + − = 1− + − = S4
2 4 6 8 2 2 3 4 2
..
.
  1 1 1 1 1 1 1 1 1 1
. S9 = S6 + − − = 1− + − + − −
5 10 12 2 2 3 4 5 10 12
1 1 1 1 1 1 1 1 1 1 1 1
= 1− + − + − = 1− + − + −
2 2 3 4 10 12 2 2 3 4 5 6
1
= S6
2
..
.

n
1
where Sn = (−1)i−1 .
i=1
i
1 
Using induction, we can demonstrate that S3n = S2n , which implies that
2
if we designate by S the limit of the sequence (Sn ), then we have
 1
. lim S3n = S.
2
  1   1 1
As S3n+1 = S3n + and S3n+2 = S3n + − , we have
2n + 1 2n + 1 4n + 2
  1
. lim S3n+1 = lim S3n + lim
2n + 1
and
  1 1
. lim S3n+2 = lim S3n + lim − lim ,
2n + 1 4n + 2
that is,
   1
. lim S3n+1 = lim S3n+2 = lim S3n = S.
2

Therefore, we conclude that lim Sn = 12 S. In other words, the series ob-
tained by this rearrangement of the terms of the alternating harmonic series
is convergent, and its sum equals half the original series’ sum.
90 Numerical Series

2.5 Series of Nonnegative Terms

In this section, we will establish convergence tests for a series of nonnegative
terms. These tests apply to investigating the absolute convergence of series
in general. We should note that convergence and absolute convergence are
equivalent for a series of nonnegative terms.

Theorem 2.5.1 Let . an be a series of nonnegative terms. The series

. an is convergent if and only if the sequence of its partial sums is
bounded.

Proof: Let .(Sn ) be the sequence of partial sums of the series . an . If

the series is convergent, then by definition, the sequence .(Sn ) converges.
Consequently, it is a bounded sequence (see Theorem 1.1.7).
Suppose that .(Sn ) is bounded. As .an ≥ 0, we have
.Sn+1 ≥ Sn , ∀n ∈ N,
that is, .(Sn ) is an increasing sequence. According to Theorem 1.1.12, the
two previous statements imply the convergence of .(Sn ), which is equivalent
to saying that the series . an is convergent.

Theorem 2.5.2 (Cauchy’s Condensation Test) Let .(an ) be a decreas-

ing sequence of positive numbers. The series . an converges if and only
if the series . 2n a2n converges.

Proof: Consider the partial sums of the series . an and . 2n a2n :


n 
n
Sn =
. ai and Tn = 2i a2i .
i=1 i=1

Note that the sequences .(Sn ) and .(Tn ) are increasing, with positive terms.
Suppose that the series . 2n a2n is convergent. By Theorem 2.5.1, the
sequence .(Tn ) is bounded. Then

S2n −1 = a1 + (a2 + a3 ) + (a4 + a5 + a6 + a7 ) + · · ·

+(a2n−1 + a2n−1 +1 + · · · + a2n −1 )
.
≤ a1 + 2 a2 + 4 a4 + · · · + 2n−1 a2n−1

= a1 + Tn−1 .
2.5. Series of Nonnegative Terms 91

As the sequence .(Tn ) is bounded, the sequence .(S2n −1 ) is also bounded.

Given that
.Sm ≤ S2n −1 , ∀m ≤ 2 − 1,
n

the sequence .(Sn ) is bounded; therefore, according to the previous theorem,

the series . an is convergent.
Conversely, suppose that the series . an is convergent. Then, using again
the previous theorem, the sequence .(Sn ) is bounded. We can observe that

S2n = a1 + a2 + (a3 + a4 ) + (a5 + a6 + a7 + a8 ) + · · · +

+(a2n−1 +1 + a2n−1 +2 + · · · + a2n )

≥ a1 + a2 + 2 a4 + 4 a8 + · · · + 2n−1 a2n
.

1
= a1 + (2 a2 + 4 a4 + 8 a8 + · · · + 2n a2n )
2
1
= a1 + Tn .
2
This implies that the sequence .(Tn ) is also bounded. Hence, the series
. 2n a2n is convergent.

∞
1
Example 2.5.1 Consider the series . , .p ∈ R, usually referred to as
n=1
np
the p-series.
If .p ≤ 0, the series is divergent because its general term does not tend to
zero.
1
If .p > 0, we can apply Theorem 2.5.2 because the sequence .an = p is
n
decreasing with positive terms. The series
∞
 ∞ ∞ n
1 2n 1
. 2n p = n =
n=1
(2n ) n=1
(2p ) n=1
2p−1

1 1
is geometric, with ratio . p−1 , and is convergent if and only if . p−1 < 1,
2 2
that is, if and only if .p > 1.
∞
1
Conclusion: The p-series . p
converges if and only if .p > 1.
n=1
n
92 Numerical Series

∞
 1
Example 2.5.2 Consider the series . α, where .α ∈ R. The
n=2
n log(n)
function
1
f (x) =
. α
x log(x)
is positive and continuous on .[2, +∞[. We can find the derivative of f to
determine whether it is increasing or decreasing. We have
α α−1
log(x) + α log(x)
f  (x) = 0 ⇔− 2α =0
x2 log(x)
α−1
. log(x) log(x) + α
⇔− 2α =0
x2 log(x)
⇔ log(x) + α = 0.
If .α ≥ 0, this equation has no root in .[2, +∞[, which implies that .f  (x) < 0,
.∀x ∈ [2, +∞[ .

If .α < 0, then .f  (x) < 0, .∀x > e−α , and this implies that .f  (x) < 0,
−α
.∀x ∈ ]e , +∞[ .
1
Then, the sequence .an = α is decreasing from the order q if
n log(n)
−α
.q ∈ N is such that .q ≥ max{2, e }. The series
∞
 ∞
 ∞
2n 1 1 1
. α = α = α α
n=q
2n log(2n ) n=q
nα log(2) log(2) n=q
n

is convergent if and only if .α > 1 as we saw in the previous example.

Remember that the convergence or divergence of the series does not change
if we omit a finite number of terms. Hence, we can conclude that the series
∞
 1
. α
n=2
n log(n)

is convergent if and only if .α > 1.

Note: In Theorem 2.5.2, the hypothesis that .(an ) is a decreasing sequence

cannot be removed. In fact, consider the series . an with
2.5. Series of Nonnegative Terms 93

⎧
⎪ 1
⎨ , if n = 2k
2k
.an =
⎪
⎩0, if n = 2k .
The series
∞
 ∞ ∞
2n
. 2n a2n = = 1
n=1 n=1
2n n=1

is divergent. However, we can write the original series as

∞
 ∞
 ∞
 k
1 1
. an = = ,
n=1
2k 2
k=1 k=1

which is a geometric series with ratio .1/2 and is therefore convergent.

Theorem 2.5.3 (Integral Test) Let .f : [1, +∞[ → R be a continuous,

positive, and decreasing function. For each .n ∈ N, let .an = f (n).
∞
  ∞
Then, the series . an and the improper integral . f (x) dx are both
n=1 1
convergent or both divergent.

Proof: Let us partition the interval .[1, n] into .n − 1 intervals of length 1, as

shown in Fig. 2.4. The total area of the rectangles inscribed in the figure is
.a2 + a3 + a4 + · · · + an . We have the following inequality:

 n
.Sn − a1 = a2 + a3 + a4 + · · · + an ≤ f (x) dx.
1

If we consider the area of the circumscribed rectangles (see Fig. 2.5), we Figure 2.4: Inscribed rectan-
have the inequality gles in the graph of f

 n
.Sn−1 = a1 + a2 + a3 + · · · + an−1 ≥ f (x) dx.
1
From the two previous inequalities, we can conclude that
 n
.Sn − a1 ≤ f (x) dx ≤ Sn−1 .
1
 n Figure 2.5: Circumscribed
If the integral is divergent, . lim f (x) dx = +∞ since f is positive. rectangles in the graph of f
n→+∞ 1
Then, by the inequality on the right, the limit of the sequence of the partial
94 Numerical Series

sums of the series is also .+∞; that is, the series diverges.
 n If the integral
converges, then there exists and it is finite . lim f (x) dx. Conse-
n→+∞ 1

n
quently, the sequence .Sn = ai is bounded. As the terms of the series
i=1
are positive, it is convergent by Theorem 2.5.1.  n
If the series is convergent, by the inequality on the right, . f (x) dx is
1
bounded, so the improper integral is convergent.
 n If the series is divergent,
based on the inequality on the left, . lim f (x) dx = +∞.
n→+∞ 1

Note: When we use the Integral Test, it is not necessary for the series or
the integral to start at .n = 1. Additionally, the function f does not need
to decrease on the interval .[1, +∞[, but only on some interval of the form
.[N, +∞[, .N ∈ N.

∞

en
Example 2.5.3 Consider the series . . The real-valued function
n=1
1 + e2n
ex
f of a real variable defined by .f (x) = is continuous and positive
1 + e2x
on .[1, +∞[. Let us study the monotonicity of f :

ex (1 + e2x ) − ex 2e2x ex (1 − e2x )

f  (x) =
.
2 = 2 < 0, ∀x ∈ [1, +∞[.
(1 + e2x ) (1 + e2x )

Thus, the function f decreases on .[1, +∞[. Furthermore,

 t
ex π
. lim dx = lim arctan(et ) − arctan(e) = − arctan(e).
t→+∞ 1 1 + e2x t→+∞ 2
 +∞
ex
Since the improper integral . dx converges, then, by the Inte-
1 1 + e2x
∞

en
gral Test, the series . also converges.
n=1
1 + e2n

∞
log(n)
Example 2.5.4 Consider the series . . The real-valued function
n=2
n
log(x)
f of a real variable defined by .f (x) = is continuous and positive on
x
2.5. Series of Nonnegative Terms 95

[2, +∞[. We can analyze the monotonicity of f by taking the derivative:

.

1
· x − log(x) 1 − log(x)

.f (x) =
x = < 0, ∀x ∈ ]e, +∞[.
x 2 x2
Thus, the function f decreases on .]e, +∞[. Now, we compute the integral:
 t
log(x) 1
. lim dx = lim (log(t))2 − (log(2))2 = +∞.
t→+∞ 1 x t→+∞ 2

∞
log(n)
According to the Integral Test, the series . diverges since the
n
 +∞
n=2
log(x)
improper integral . dx is also divergent.
1 x

Theorem 2.5.4 (General Comparison Test) Let . an and . bn be

two series of nonnegative terms such that .an ≤ bn , .∀n ∈ N:

a) If the series . bn is convergent, then the series . an is convergent.

b) If the series . an is divergent, then the series . bn is divergent.

Proof: Let .(Sn ) and .(Sn ) be the sequences of partial sums of series . an
and . bn , that is,

n
 
n
Sn =
. ai and Sn = bi .
i=1 i=1

Since .0 ≤ an ≤ bn , .∀n ∈ N, we have


0 ≤ Sn ≤ Sn , ∀n ∈ N.
. (2.3)
a) If the series . bn is convergent, then, by Theorem 2.5.1, the sequence of

its partial sums, .(Sn ), is bounded. By inequality (2.3), the sequence .(Sn )
is also bounded, that is, the series . an is convergent.
b) If the series . an is divergent, then by Theorem 2.5.1, the sequence .(Sn )

is not bounded. Inequality (2.3) implies that the sequence .(Sn ) is also not
bounded; thus, again by Theorem 2.5.1, the series . bn is divergent.

Note: Omitting a finite number of terms does not change the convergence
or divergence of the series, as we have seen. Therefore, the previous theorem
remains valid if there exists .p ∈ N such that .an ≤ bn , .∀n ≥ p.
96 Numerical Series

∞
 1
Example 2.5.5 Consider the series . . We know that
n=1
n!

1 1 1
0<
. = ≤ n−1 , ∀n ∈ N.
n! n(n − 1)(n − 2) . . . 2 2
∞
 1
Also, the series .
n−1
is convergent because it is geometric with ratio
n=1
2
1
. . By the General Comparison Test, we can conclude that the given series
2
is convergent.

∞
cos(n)
Example 2.5.6 The series . is not of positive terms. However,
n=1
n2
we can observe that the following inequality holds:
 
 cos(n)  1
.0 <  
 n2  ≤ n2 , ∀n ∈ N. (2.4)

∞
1
The series . is a p-series with .p = 2 > 1 and is, therefore, convergent.
n=1
n2
By applying the General Comparison Test and taking into
 account inequality
∞
 cos(n) 
(2.4), we can conclude that the series .  
 n2  is convergent. This
n=1
∞
 cos(n)
means that the series . is absolutely convergent.
n=1
n2

∞
 sin(n)
Example 2.5.7 Consider the series . (−1)n n. Let us analyze
n=1
log(10)
the series of modules:
∞ 
 
   ∞
| sin(n)|
 sin(n) 
. (−1)n n  = n.
 log(10)  log(10)
n=1 n=1

We notice that the following inequality holds for every natural number n:

| sin(n)| 1
0<
. n ≤ n.
log(10) log(10)
2.5. Series of Nonnegative Terms 97

∞
 1 1
The series . n is geometric with ratio .r = . Since
n=1
log(10) log(10)
1
10 > e ⇒ log(10) > 1 ⇒ 0 <
. < 1,
log(10)
we have .|r| < 1. Therefore, the geometric series converges. By applying
∞
| sin(n)|
the General Comparison Test, the series . n is convergent. This
n=1
log(10)
implies that the given series is absolutely convergent.

Corollary 1 Let . an and . bn be two series of positive terms, c be

a positive constant, and p be a natural number such that .an ≤ c bn ,
.∀n ≥ p:

a) If the series . bn is convergent, then the series . an is convergent.

b) If the series . an is divergent, then the series . bn is divergent.

Proof: Let .cn = c bn . By Theorem 2.5.4,

a) if the series . cn is convergent, then the series . an is also convergent;
b) if the series . an is divergent, then the series . cn is also divergent.
Since .c > 0, the series . cn and . bn are either both convergent or both
divergent. From this fact, we can conclude the desired result.

Corollary 2 Let . an and . bn be two series such that .an ≥ 0 and

an
.bn > 0, .∀n ∈ N. If .lim = k ∈ R+ , then both series are either
bn
convergent or divergent.

an
Proof: Suppose .lim = k. By Definition 1.1.7,
bn
 
 an 
.∀δ > 0 ∃ p ∈ N : n > p ⇒  − k  < δ.
 bn 
k
Let .δ = . From a certain order, we have
2
 
 an  k k an k k an 3
 
.
 bn − k  < 2 ⇔ − 2 < bn − k < 2 ⇔ 2 < bn < 2 k,
98 Numerical Series

which implies
3 k
a <
. n k bn and bn < a n .
2 2
From these inequalities and using Corollary 1, we get the intended result.

Corollary 3 Let . an and . bn be two series such that .an ≥ 0 and

an
.bn > 0, .∀n ∈ N. If .lim = 0, then:
bn
a) If the series . bn is convergent, then the series . an is also con-
vergent.

b) If the series . an is divergent, then the series . bn is also divergent.

an
Proof: Let .lim = 0. By Definition 1.1.7,
bn
 
 an 
.∀δ > 0 ∃ p ∈ N : n > p ⇒  
 bn  < δ.

From a certain order, we have

 
 an  an
 
.
 bn  = bn < δ,

since .an ≥ 0 and .bn > 0. Consequently, .0 ≤ an < δbn , and from Corollary 1
of Theorem 2.5.4, the result follows.

Corollary 4 Let . an and . bn be two series such that .an ≥ 0 and

an
b > 0, .∀n ∈ N. If .lim
. n = +∞, then:
bn
a) If the series . bn is divergent, then the series . an is also divergent.

b) If the series . an is convergent, then the series . bn is also con-

vergent.

an
Proof: Let .lim = +∞. By Definition 1.1.6,
bn
an
∀δ > 0 ∃ p ∈ N : n > p ⇒
. > δ.
bn
2.5. Series of Nonnegative Terms 99

From order p, we have

a > δbn > 0,
. n

since .bn > 0. The result is a consequence of Corollary 1 of Theorem 2.5.4.

Corollary 5 Let . an and . bn be two series such that .an > 0 and
b > 0, .∀n ∈ N. If there exists .p ∈ N such that
. n

an+1 bn+1
. ≤ , ∀n ≥ p,
an bn
then:

a) If the series . bn is convergent, then the series . an is convergent.

b) If the series . an is divergent, then the series . bn is divergent.

Proof: As .an > 0 and .bn > 0, we can deduce that:

an+1 bn+1 an+1 an

. ≤ ⇔ ≤ , ∀n ≥ p.
an bn bn+1 bn

an
This means that the sequence . is decreasing after order p. Therefore,
bn
ap an
there exists a constant k (we can take .k = ) such that . ≤ k. In other
bp bn
words:
a ≤ k bn , ∀n ≥ p.
. n

We can use Corollary 1 of Theorem 2.5.4 to arrive at the final result.

∞
1 + (−1)n
Example 2.5.8 The terms of the series . are nonnegative.
n=1
n2
As
1 + (−1)n 2
.0≤ ≤ 2 , ∀n ∈ N,
n2 n
∞
1
and the series . 2
is convergent, Corollary 1 of Theorem 2.5.4 allows us
n=1
n
to conclude that the given series is convergent.
100 Numerical Series

∞
2n2 + n
Example 2.5.9 Consider the series of positive terms . . Since
n=1
3n5 + 3
2n2 + n
3n5 + 3 2n5 + n4 2
. lim = lim = ,
1 5
3n + 3 3
n3
∞
 2n2 + n
by Corollary 2 of Theorem 2.5.4, the series . is convergent be-
n=1
3n5 + 3
∞
 1
cause . 3
converges.
n=1
n

∞
log(n)
Example 2.5.10 The series . is of nonnegative terms. Since
n=1
n3
∞
 1
.
2
is convergent and
n=1
n

log(n)
n3 log(n)
. lim = lim = 0,
1 n
n2
by Corollary 3 of Theorem 2.5.4, the original series is also convergent.

Example 2.5.11 Consider the series

∞
 ∞ ∞
1 × 3 × · · · × (2n − 1) 1
. bn = and .
n=1 n=1
2 × 4 × · · · 2n n=1
n
Both series are of positive terms, and the second is divergent. We have
1 × 3 × · · · × (2n − 1)(2n + 1)
bn+1 2 × 4 × · · · 2n(2n + 2) 2n + 1 an+1 n
. = = and = .
bn 1 × 3 × · · · × (2n − 1) 2n + 2 an n+1
2 × 4 × · · · 2n
n 2n + 1
It is easy to verify that ≤. , which allows us to conclude, by
n+1 2n + 2
∞

Corollary 5 of Theorem 2.5.4, that the series . bn is divergent.
n=1
2.5. Series of Nonnegative Terms 101

∞
arctan(n) + n
Example 2.5.12 The terms of the series . √ are positive.
n=1
n n+1
∞
 1
Let us consider the series . , which is divergent because it is a p-
n=1
n1/2
1
series with .p = . The limit
2
arctan(n) + n arctan(n)
√ 1/2 +1
n n+1 (arctan(n) + n) n n
. lim = lim √ = lim =1
1 n n+1 1
1 + 3/2
n1/2 n
is finite and different from zero. Therefore, by Corollary 2 of the General
∞
 arctan(n) + n
Comparison Test, the series . √ diverges.
n=1
n n+1

Theorem 2.5.5 (Ratio Test) Let . an be a series of positive terms:

an+1
a) If there exist .r < 1 and .p ∈ N such that . ≤ r < 1 .∀n ≥ p,
an
then the series . an is convergent.
an+1
b) If there exists .p ∈ N such that . ≥ 1, .∀n ≥ p, then the series
an
. an is divergent.

Proof:
a) The series . bn = rn is geometric with ratio r and is convergent since
.0 < r < 1. Moreover, we have

an+1 rn+1
. ≤ n = r, ∀n ≥ p,
an r
which implies by Corollary 5 of the General Comparison Test that the series
. an is convergent.
b) The series . bn = 1 is divergent. Additionally, we have
an+1 bn+1
. ≥1= , ∀n ≥ p.
an bn
By Corollary 5 of the General Comparison Test, we can conclude that the
series . an is divergent.
102 Numerical Series

Corollary 1 (D’Alembert’s Test) Let . an be a series of positive terms.

an+1
If there exists .lim = a . (a ∈ R+
0 or .a = +∞), then:
an
a) If .a < 1, the series . an is convergent.
Jean Le Rond D’Alembert b) If .a > 1, the series . an is divergent.
(1717–1783) French math-
ematician and philosopher
who was elected to the Proof: We know that if .a ∈ R,
Academy of Sciences of  
an+1  an+1 
Paris at the age of 24, . lim 
= a ⇔ ∀δ > 0 ∃p ∈ N : n > p ⇒  − a < δ.
and later to the Berlin an an
Academy. With Diderot,
a) If .a < 1, then .1 − a > 0. Let .δ be such that .0 < δ < 1 − a. There exists
he began publishing the
“Encyclopedia” in 1747, p ∈ N such that for all .n > p,
.
 
writing articles on mathe-  an+1  an+1 an+1
 
matics and literature. In .
 an − a < δ ⇔ −δ < an − a < δ ⇔ a − δ < an < a + δ.
1772, he was appointed
perpetual secretary of the Since .δ < 1 − a, then .a + δ < 1. Item a) of the Ratio Test allows us to
French Academy. As a
conclude that the series . an converges.
mathematician and scien-
tist, D’Alembert contin- b) If .a > 1, let .δ = a − 1. There exists .p ∈ N such that, for all .n > p,
ued the work of Newton  
and Leibniz and researched  an+1  an+1
 
differential equations with
.
 an − a < a − 1 ⇔ 1 < an < 2a − 1.
partial derivatives. He was
also responsible for stating By item b) of the Ratio Test, the series . an diverges.
the fundamental theorem If .a = +∞, there exists .p ∈ N such that
of algebra, which Gauss an+1
later proved. Addition- . > 1, ∀n > p.
ally, D’Alembert studied an
hydrodynamics and me- Again, by the Ratio Test, the series . an diverges.
chanics and formulated
what is now known as an+1
“D’Alembert’s principle.” Note: If .lim = 1, D’Alembert’s test is inconclusive as there are
This principle, when ap-
an
divergent and convergent series in this situation. For example, the harmonic
plied to the Earth’s move- ∞
 1
ment, explains the varia- series . is divergent and
tions of the rotation axis
n=1
n
of the globe. (Source
of image: Oil portrait
an+1 n
. lim = lim = 1,
by Maurice-Quentin de La an n+1
Tour (1704–1788), Louvre ∞
 1
Museum.) and the series . is convergent and
n=1
n2
2
an+1 n
. lim = lim = 1.
an n+1
2.5. Series of Nonnegative Terms 103

an+1
However, if .lim = 1 and the convergence is for values greater than 1,
an
an+1
that is, there is an order .p ∈ N from which . ≥ 1, then, by the Ratio
an
∞
Test, the series . an diverges.
n=1

∞
 k n n!
Example 2.5.13 Let .k > 0. The series . is of positive terms. As
n=1
nn

k n+1 (n + 1)!
n
(n + 1)n+1 k n+1 (n + 1)! nn n 1
. lim = lim = lim k · =k· ,
k n n! k n n! (n + 1)n+1 n+1 e
nn

k k
we can apply D’Alembert’s test: If . < 1, the series is convergent; if . > 1,
e e
the series is divergent.
n
k n+1
If = 1, D’Alembert’s test is inconclusive. However, as .
. is an
e n
n
n
increasing sequence with limit e, . is a decreasing sequence with
n+1
n
1 n
limit . , which implies that .e · is decreasing with limit 1, that is,
e n+1
an+1
. tends to 1 from above. Then, if .k = e, the series is divergent.
an

∞
 2
(n + 1)!
Example 2.5.14 The series . (−1)n is alternating. Let us
n=1
(2n)! 5n
∞ 2
(n + 1)!
start by studying the series of modules, . , using D’Alembert’s
n=1
(2n)! 5n
test. Let .an be the general term of the series of modules:
2
(n + 2)!
2
an+1 2(n + 1) ! 5n+1 (n + 2)! (2n)! 5n
lim = lim 2 = lim 2
an (n + 1)! 2(n + 1) ! 5n+1 (n + 1)!
.
(2n)! 5n
2
(n + 2)! (2n)!
= lim 2
(2n + 2)(2n + 1)(2n)! 5 (n + 1)!
104 Numerical Series

(n + 2)2 ((n + 1)!)2

= lim
(2n + 2)(2n + 1) 5 ((n + 1)!)2
.
(n + 2)2 1
= lim = .
5 (2n + 2)(2n + 1) 20
Since this value is less than 1, the series of modules is convergent; that is,
the original series is absolutely convergent.

∞
(n!)2 + n!
Example 2.5.15 The series . is of positive terms and
n=1
(4n)! + n4

(n!)2 + n! 2(n!)2
0<
.
4
< , ∀n ∈ N.
(4n)! + n (4n)!

∞
2(n!)2
Let us study the series . by D’Alembert’s test:
n=1
(4n)!

2
2 (n + 1)!
(4n + 4)! (n + 1)2
. lim = lim = 0 < 1.
2(n!)2 (4n + 4)(4n + 3)(4n + 2)(4n + 1)
(4n)!
∞
 2(n!)2
Therefore, the series . converges, so by the General Comparison
n=1
(4n)!
Test, the given series converges.

Example 2.5.16 Let us consider the series of positive terms

1 1 1 1 1 1 1 1 1
. + · + 2 · + 2 · 2 + 3 · 2 + ··· ,
2 2 3 2 3 2 3 2 3
1 1 1 1 1 1 1
that is, .a1 = , .a2 = · , .a3 = 2 · , .a4 = 2 · 2 , . . . ; in general,
2 2 3 2 3 2 3
⎧
⎪
⎪
⎪
⎪
1 1
⎨ 2 n2 · 3 n2 ,
⎪ if n is even
.an =
⎪
⎪ 1 1
⎪
⎪ n+1 · n−1 , if n is odd.
⎪
⎩2 2
3 2
2.5. Series of Nonnegative Terms 105

Then
⎧
⎪
⎪ 1 1
⎪
⎪ n+2 ·
n n
⎪
⎪
n
22 · 32 1
⎪
⎪ 2 2
3 2
= 2−1 = ,
⎪
⎪
= n+2 if n is even
⎪ 1 1 n
⎪ 2 2 · 32 2
an+1 ⎨ n2 · n2
. = 2 3
an ⎪
⎪ 1 1
⎪
⎪ n+1 ·
⎪
n+1 n−1
⎪
⎪2
n+1
2 2 ·3 2 1
⎪ −1
2 2
⎪ 3 = n+1
⎪ n+1 = 3 = , if n is odd.
⎪
⎪ 1 1 2 2 ·3 2 3
⎩ n+1 · n−1
2 2
2 3
an+1
Note that we cannot use D’Alembert’s test, as .lim does not exist.
an
However, it follows from the Ratio Test that the series converges because
an+1 1
. ≤ < 1, ∀n ∈ N.
an 2

Note: We saw in Theorem 2.4.1 that the convergence of the series of

modules implies the convergence of the series. Nevertheless, if the series of
modules is divergent, nothing can be concluded about the convergence of
the series. However, suppose we determine the divergence of the series of
modules using D’Alembert’s test. As this condition implies that the general
term of the series does not converge to zero, we can conclude the divergence
of the series.

∞
(n!)2
Example 2.5.17 Consider the series . (−1)n
. Let .an be the gen-
n=1
2n nn
eral term of the series of modules. Using D’Alembert’s test, we get

2
(n + 1)!
2
an+1 2n+1 (n + 1)n+1 (n + 1)! 2n nn
lim = lim = lim n+1
.
an (n!)2 2 (n + 1)n+1 (n!)2
n
2 n n
n
n+1 n
= lim = +∞.
2 n+1

The limit value is greater than 1, which means the series of modules is
divergent. Since we used D’Alembert’s test, the given series is also divergent.
106 Numerical Series

Theorem 2.5.6 (Root Test) Let . an be a series of nonnegative terms:

√
a) If there exist .r < 1 and .p ∈ N such that . n an ≤ r, ∀n > p, then
the series . an is convergent.

b) If there exist .p ∈ N and a subsequence, .(akn ), of .(an ) such that

√
. kn akn ≥ 1, ∀kn > p, then the series . an is divergent.

Proof:
√
a) If . n an ≤ r, ∀n > p, then .an ≤ rn < 1, .∀n ≥ p. The series . rn is
convergent because it is a geometric series with ratio r, where .0 < r < 1.
Therefore, the series . an is convergent.
√
b) If . kn akn ≥ 1, ∀kn > p, then .akn ≥ 1, .∀kn > p, so it does not tend to
zero. As a result, .(an ) is not a null sequence, which implies that the series
. an is divergent.

Corollary 1 Let . an be a series of nonnegative terms:

√
a) If .lim n an < 1, the series . an is convergent.
√
b) If .lim n an > 1, the series . an is divergent.

√
Proof: Let .a = lim n an .
a) Let r be such that .a < r < 1. We can affirm that
√
.∃ p ∈ N : n > p ⇒ an < r,
n

which implies by item a) of the theorem that the series . an converges.

√
b) By definition of upper limit, there is a subsequence of . n an with limit
.a > 1, so this sequence has an infinity of values greater than 1. By item b)

of the theorem, the series diverges.

Corollary 2 (Cauchy’s Root Test) Let . an be a series of nonnegative

√
terms. If there exists .lim n an = a . (a ∈ R+
0 or .a = +∞), then:

a) If .a < 1, the series . an is convergent.

b) If .a > 1, the series . an is divergent.

2.5. Series of Nonnegative Terms 107
√ √ √
Proof: If there exists .lim n
an = a, then .lim n an = lim n an = a. More-
over, Corollary 1 applies.

√
Note: If .lim n an = 1, the test is inconclusive, as there exist both conver-
gent and divergent series in this situation. For instance, the harmonic series
∞
1
. is divergent and
n
n=1 
n 1 1
. lim = lim √ = 1,
n n
n
∞
1
and the series . 2
is convergent and
n
n=1 
1 1
= lim √
n
. lim = 1.
n2 n
n2

∞
 n 2
n+1
Example 2.5.18 The series . is of positive terms. As
n
 n=1
n2 n
n n+1 n+1
. lim = lim = e > 1,
n n
by Cauchy’s Root Test, the series is divergent.

∞
 1
Example 2.5.19 Let us consider the series . .
(3 + (−1)n )n
⎧ n=1
⎪
⎪
 ⎪
⎪ 1
1 ⎨ , if n is even
n 4
. =
(3 + (−1)n )n ⎪
⎪ 1
⎪
⎪
⎩ 2 , if n is odd.

√ 1
Then . n an ≤ < 1, .∀n ∈ N, and by the Root Test, the series is convergent.
2

∞
 2n
∞ 2n
log(n) log(n)
Example 2.5.20 The series . = is of
n=1
nn n=2
nn
positive terms. We will study this series using the Cauchy’s Root Test:

2n 2
n log(n) log(n)
. lim = lim =0
nn n
108 Numerical Series

because using L’Hôpital’s Rule, we have

 
2 
log(x) 2 log(x)
. lim = lim = 0.
x→+∞ (x) x→+∞ x

Since the limit is less than 1, the series is convergent.

Example 2.5.21 Let us study the convergence of the series . an where

⎧
⎨(1 − √
n
n)n , if n is odd
a =
. n
⎩n2 e−n , if n is even.

It is an alternating series. Let us study the series of modules.

⎧ √
 ⎨ n |(1 − n n)n |, if n is odd
n
|an | = √
⎩ n n2 e−n , if n is even
.
⎧√
⎨ n n − 1, if n is odd
=
⎩e−1 √n
n2 , if n is even.

√ √ 
Since .lim( n n−1) = 0 and .lim e−1 n2 = e−1 , we obtain .lim n |an | = e−1 .
n

∞


As .lim n |an | < 1, the series . |an | is convergent. Therefore, the series
n=1
∞

. an is absolutely convergent.
n=1

Note: The Cauchy’s Root Test is more general than D’Alembert’s test.
This means that if Cauchy’s Root Test is inconclusive about a series, then
D’Alembert’s test will also be inconclusive. In fact, by Theorem 1.1.11
an+1 √
.lim
an .= a ⇒ lim
n a
n = a. This theorem is significant because if .a = 1,
then D’Alembert’s test is inconclusive, and the same happens for Cauchy’s
Root Test. However, it is essential to note that the reciprocal is not true.
It may be possible to draw conclusions through Cauchy’s Root Test, even if
D’Alembert’s test fails to do so.
2.5. Series of Nonnegative Terms 109

∞

2−n−(−1) . Using Cauchy’s
n
Example 2.5.22 Let us consider the series .
n=1
Root Test,

(−1)n 1
2−n−(−1) = lim 2−1 2−
n n
. lim n = < 1,
2

we conclude that the series is convergent, whereas D’Alembert’s test fails

to be applied. In fact,


2−(n+1)−(−1)
n+1
2, if n is even
= 2−n−1−(−1)
n+1 n
+n+(−1)
.
−n−(−1)n
= −3
2 2 , if n is odd.

Ernst Eduard Kummer

(1810–1893) was a
German mathematician.
Theorem 2.5.7 (Kummer’s Test) Let . an and . bn be two series of In 1855, he succeeded
positive terms, with . bn divergent. If there exists Dirichlet at the University
of Berlin, where he worked
1 an 1 with Weierstrass and
. lim · − = k, k ∈ R, Kronecker. In 1857, the
bn an+1 bn+1
Academy of Sciences of
then: Paris awarded him a prize
for his work on Fermat’s
a) If .k > 0, the series . an is convergent. Theorem. (Source of im-
age: Photo archive of the
b) If .k < 0, the series . an is divergent. Mathematical Research
Institute Oberwolfach.)

1 an 1
Proof: If . k ∈ R, .lim · − = k is equivalent to
bn an+1 bn+1

 
1 an 1 
∀δ > 0 ∃p ∈ N : ∀n > p,  ·
. − − k  < δ.
bn an+1 bn+1

But
 
1 an 1  1 an 1
 
.
 bn · an+1 − bn+1 − k  < δ ⇔ k − δ < bn · an+1 − bn+1 < k + δ.
110 Numerical Series

k
a) Let .k ∈ R+ and .δ = . There exists an order .n0 ∈ N from which we
2
have
k 1 an 1 k 1 an 1
k− < · − ⇔ < · −
2 bn an+1 bn+1 2 bn an+1 bn+1
2 1 an 1 2 1 an 1
. ⇔ 1< · − ⇔ an+1 < an+1 · −
k bn an+1 bn+1 k bn an+1 bn+1
2 an an+1
⇔ an+1 < − .
k bn bn+1

Adding the two members of the inequality from .n0 + 1 to .n + 1, we obtain


n+1 
n+1
2 ai−1 ai
ai < −
i=n0 +1 i=n0 +1
k bi−1 bi

n+1
2 an0 an +1 an +1 an +2 an an+1
. ⇔ ai < − 0 + 0 − 0 + ··· + −
i=n0 +1
k bn0 bn0 +1 bn0 +1 bn0 +2 bn bn+1

n+1
2 an0 an+1 2 an0
⇔ ai < − < .
i=n0 +1
k bn0 bn+1 k bn0

Therefore, the sequence of partial sums of the series . an is bounded be-

cause


n+1
2 an0
0 < Sn+1 =
. ai = Sn0 + an0 +1 + · · · + an+1 ≤ Sn0 + ,
i=1
k bn0

and by Theorem 2.5.1, the series . an converges.

If .k = +∞, let .α > 0 be arbitrary. There is an order .n0 ∈ N from which
we have
1 an 1 α
. · − > ,
bn an+1 bn+1 2
and we can apply the previous reasoning.
b) Let .k ∈ R− and .δ = −k. There is an order .n0 ∈ N from which we have

1 an 1 an bn an+1 bn+1
. · − <0⇔ < ⇔ > .
bn an+1 bn+1 an+1 bn+1 an bn

As the series . bn is divergent, Corollary 5 of the General Comparison Test

allows us to conclude that . an is divergent.
2.5. Series of Nonnegative Terms 111

If .k = −∞, there is also an order .n0 ∈ N from which we have

1 an 1
. · − < 0,
bn an+1 bn+1

and we finish in the same way.

Corollary 1 (Raabe’s Test) Let . an be a series of positive terms. Sup-

pose that
an
.lim n −1 =a:
an+1
Then
a) If .a < 1, the series . an is divergent.

b) If .a > 1, the series . an is convergent.

∞

1 1
Proof: In Kummer’s test, consider .bn = . The series . is divergent,
n n=1
n
and we have
1 an 1 an
lim · − = lim n · −n−1
bn an+1 bn+1 an+1
.
an
= lim n − 1 − 1 = a − 1. Joseph L. Raabe (1801–
an+1 1859), was a Swiss mathe-
matician and physicist. He
This demonstrates the corollary. was a professor at the
Zurich Polytechnic Insti-
tute. His name is mainly
Note: Often, cases that are inconclusive by D’Alembert’s test can be solved
associated with the con-
by Raabe’s test. vergence test for a series
of positive terms that ex-
tends D’Alembert’s Test.
He also studied various as-
Example 2.5.23 Consider the series pects of planetary move-
ments. (Source of im-
∞ ∞

1 × 3 × · · · × (2n − 1) 1 age: Lithographie by Carl
. · = an . Friedrich Irminger (1813–
n=1
2 × 4 × · · · × 2n n n=1
1863).)

We have
an+1 n(2n + 1)
. lim = lim = 1,
an (n + 1)(2n + 2)
112 Numerical Series

and therefore, D’Alembert’s test is inconclusive. As

an (n + 1)(2n + 2)
lim n −1 = lim n −1
an+1 n(2n + 1)
.
(n + 1)(2n + 2) − n(2n + 1) 3n + 2 3
= lim = lim = > 1,
2n + 1 2n + 1 2
Raabe’s test shows that the series is convergent.
2.6. Products of Series 113

2.6 Products of Series

∞
 ∞

Let . an andbn be two convergent series with sums A and B, re-
.
n=1 n=1 ∞  ∞ 
 
spectively. When thinking about the product . an × bn , it will
n=1 n=1
be natural to define it in such a way that the resulting series, if convergent,
has sum .A × B. We can define, for example,
∞  ∞  ∞
 ∞

   
. an × bn = an bk
n=1 n=1 n=1 k=1

obtaining
∞
 ∞
 ∞ ∞
   
. an bk = an · B = B · an = B × A.
n=1 k=1 n=1 n=1

However, one may ask if it is possible to form a series whose terms are
products of the form .an bk , in some order, so that the sum of this series is
.A × B. The answer to this question is given in the following theorem:

Theorem 2.6.1 Let . an and . bn be two convergent series, with sums

A and B, respectively. Let .φ be a bijective function, .φ : N2 → N,
.φ(i, j) = n. To each .φ, it corresponds to a series . cn , with

c = cφ(i,j) = ai × bj .
. n

The series . cn converges, regardless of the function .φ considered if and

only if . an and . bn are both absolutely convergent, and, in this case,
we have . cn = A×B, and the series . cn is also absolutely convergent.

Note: Saying that . cn converges for any function .φ considered is equiv-

alent to state that the product series converges regardless of the order in
which its terms are taken.

∞

Definition 2.6.1 The Cauchy product of two convergent series, . an
∞ ∞
  n=1
  
n
and . bn , is the series . ak bn−k+1 .
n=1 n=1 k=1
114 Numerical Series

Note: If .n ∈ N0 , then the Cauchy product is written as follows:

∞
 n 
 
. ak bn−k .
n=0 k=0

Corollary 1 If . an and . bn are absolutely convergent series with sums

A and B, respectively, their Cauchy product is absolutely convergent and
has sum .A × B.

∞
xn
Example 2.6.1 Consider the series . , .x ∈ R. As
n=0
n!
 n+1 
 x 
 
 (n + 1)!  |x|
. lim  n = lim = 0, ∀x ∈ R,
x  n +1
 
 n! 

the series is absolutely convergent for every .x ∈ R. Then, the Cauchy

product of two series of this type is absolutely convergent. Let us form the
product and verify that the obtained series is absolutely convergent.
∞  ∞  ∞
 n 
 xn  yn   xk y n−k
× = ·
n=0
n! n=0
n! n=0 k=0
k! (n − k)!
∞
 
 1 
n
n!
.= ·x y
k n−k

n=0
n! k!(n − k)!
k=0
∞
 (x + y)n
= .
n=0
n!

Note: The Cauchy product of two series that are not absolutely convergent
may lead to a divergent series.

∞
(−1)n
Example 2.6.2 The series .√ is conditionally convergent. Cal-
n=0
n+1
culating the Cauchy product of the series with itself, we get
2.6. Products of Series 115

∞
 n 
  (−1)k (−1)n−k
 ·
n=0 k=0 (k + 1) (n − k + 1)
∞
 n 
  (−1)n
.= √ √
n=0 k=0
k+1 n−k+1
∞
 n  ∞
  1 
= (−1) n
√ √ = (−1)n an ,
n=0 k=0
k + 1 n − k + 1 n=0

which is an alternating series. However, .(an ) is not a null sequence as


n
1  n
1
.an = √ √ ≥ √ √ = 1;
k=0
k + 1 n − k + 1 k=0 n + 1 n+1

thus, the product series is divergent.

Theorem 2.6.2 (Mertens) If at least one of the convergent series, . an

and . bn , is absolutely convergent, then their Cauchy product is conver-
gent, and its sum is the product of the sums of the given series.

Theorem 2.6.3 If the series . an and . bn are convergent, with sums

A and B, respectively, then if their Cauchy product is convergent, its sum
is .A × B.

∞
(−1)n
Example 2.6.3 The series . is conditionally convergent, and the
n=1
n
∞
 (−1)n
series . is absolutely convergent. By Mertens’s Theorem, the
n=1
n2
Cauchy product of the series, which is alternating, is convergent:

 ∞
 ∞  ∞
 n 
 (−1)n  (−1)n   (−1)k (−1)n−k+1
× = ·
n=1
n n=1
n2 n=1 k=1
k (n − k + 1)2
.
∞
 
 n
1
n+1
= (−1) .
n=1
k(n − k + 1)2
k=1
116 Numerical Series

2.7 Solved Exercises

∞   2
1. Show that the following series are convergent and 1
b) sin
find their sums: n=1
n

∞
∞
 log(n7 + 1)
1 c)
a) n2
n2 + 3n + 2 n=1
n=1
∞
 
∞  1 √
   d) n+ − n
b) arctan(n + 3) − arctan(n) n
n=1
∞ √
n=1
 √
∞   n+1− n
 1 1 e) √ √
c) n
− n+1 n=1
n+1+ n
n=0
e e
∞
∞   cos(n)
 1 1 f)
d) − n=1
n2 + 4
n=1
arctan(n) arctan(n + 1)
∞
 1
∞ 
  g)
1 1 n3 + 10 cos(n)
e) − n=1
22n−1 23n−2
n=1 ∞
1 − (−1)n
∞
 h) √
3n + 7n n
f) n=1
n=0
3n · 7n
5. Investigate the convergence of the following series:
∞
 1 2 3
g) · n · n+1 ∞
 log(n)
n=0
4n−1 5 6 a) (−1)n
n=1
n
∞
 log( n+2 ) ∞
h) n
 2
n=2
log(n) log(n + 2) b) (−1)n+1
n=1
en + e−n
2. Show that the series ∞
 1
c) (−1)n
∞
   n=1
log(3n)
1 1
. + n ∞ 1
(n + 1)2 π  sin n
n=2 n
d) (−1)
n=1
n
is convergent and find its sum, knowing that ∞
∞  3n
 1 π2 e) (−1)n
2
= . e2n+1
n 6 n=1
n=1
∞  
 sin nπ
3. Write the following repeating decimals as fractions: f) √ 2
n=1
n+1
a) 0, 5 c) 1, 345
∞
(−1)n+1
b) 0, 34 d) 0, 324101 6. Consider the series :
n=1
n
4. Test the convergence of the following series using a
comparison test: a) Study it for convergence.
b) Indicate a partial sum Sn that approximates
∞
 2 n2 − 1 the sum of the series with an error less than
a) 1
n=1
3 n5 + 2 n + 1 1000
.
2.7. Solved Exercises 117

c) Indicate an upper bound of the error commit- 9. Test for convergence or divergence the following se-
ted when S5 is taken as the sum of the series. ries. If convergence occurs, indicate whether it is
conditional or absolute:
7. Use the Ratio Test or D’Alembert’s Test to deter- ∞
 (−1)n
mine the convergence or divergence of the following a)
log(n n)
series: n=2
∞    
∞
nn n2 2 n
a) b) √ + 1+
π n n! n=1 n5 + 1 n
n=1
∞  
∞
 1
n 2n c) log 1 − 2
b) n
n=0
4 n3 + 1 n=2
∞
∞ sin(n) + 2n
(n + 1)! − n! d)
c) n + 5n
n=1
n! (n + 1)! n=1
∞

∞ en+1
2 · 4 · 6 · · · · · (2n + 2) e) (−1)n
d) nn
n=1
1 · 4 · 7 · · · · · (3n + 1) n=1
1
∞
 ∞ (−1)n +

(n + 1)! n
e) √ f)
n=1
nn 3n + 2 n=1
n
∞ ∞  
 3n + n! 1
f) g) log 1 + n
n! + nn n=1
2
n=1
∞
∞
 n!  √ n
g) h) 1− n
n
n=1
(2 n)! + 2n n=1
∞
∞
 n! + 3n (−1)n
h) i)
((n + 1)!)2 n=1
1 − (−1)n n2
n=0 
 (−1)n + 4  3 2n
∞
8. Use the Root Test or Cauchy’s Root Test to deter- j) +
n=1
n + 4n n+2
mine the convergence or divergence of the following
∞  
series: 1 n
k) 1−
∞ 
 n2 n=1
n
2
a) 1− ∞

n 1
n=3 l)  3
∞
 n=2 n log(n)
2
b) e−n ∞ n+2 1
n=0 m) dx
∞ n x2
 √
n n
n=1
c) 2−1 ∞
(−1)n
n=1 n) √ √
n+1+ 3n+1
∞   n n=1
π 1 ∞
d) cos +  1
n=1
6 n o) √ √ 
nn+ 1+ n
∞  n n=1
1 n ∞   
e) + (−1)n 1
n=0
2 4n + 1 p) (−1)n 1 − cos
n
∞  2 n n=1
 n −2 ∞  
f)  1
3n2 q) (−1)n 2n sin n
n=1 3
n=1
118 Numerical Series

∞
 √ ∞
  
n! n 1 1
r) (−1)n √ n) arcsin
n=1
nn n + 1 n=1
2n + 5 n
∞  
√ 1 11. Determine whether the following series are conver-
s) ( n + 1) sin
n=1
n2 gent or divergent. If convergence occurs, indicate if
∞
 it is conditional or absolute:
1
t)
∞  
n=2
log(n!) 1 1
a) sin n
Hint: Note that n! < nn n=1
n 2
∞ √
 3
n2
10. Assess the convergence or divergence of the given b) √
series. If the series converges, specify whether the n=1
n n + 2n2
convergence is absolute or conditional: ∞
cos(πn)
∞
c) √
 n cos(n π) n=2 n2 − 1
a) √
n3 + 1 ∞

n=0 (−1)n
∞
d) √
 n! n=2
n log(n)
b)
3 × 5 × · · · × (2 n + 1) ∞
n=1 4 + sin(n)
∞
e) √
 1 + cos2 (n) n=1
3
n+1
c) √
n ∞
n=1 n2n sin(n3 )
∞
f)
 3n (n + 1)! n=1
(3n2 + 5)n
d)
(n + 1)n ∞
  
n=0 n−1
∞   g) (−1)n arcsin
 (−1)n n2 + 1
n=3
e) sin
n ∞
n=1 4 × 7 × · · · × (3n + 1)
∞ √ h)
n n=1
8 × 11 × · · · × (3n + 5)
f)
n 2 + cos(n)
∞  
n=1 arctan(n3 ) 2n (2n)!
∞ i) √ +
 3n n + n2 3n (2n + 1)!
n=1
g)
2n + nn
n=1 ∞ sin
 √1
n
∞
1 + n(−1)n j) √
h) n=1
n+ n
1 + 2n3
n=1
∞     
1 1
∞
 2n k) n sin − (n + 2) sin
i) n=1
n n+2
n! + 1
n=1
∞
∞   2n + 3
 n! 1 l)
j) − n n=1
(n + 1)!
(2n)! 2
n=0
∞
∞
arctan(n + 1) − arctan(n)
 n (2n)! m)
k) n=1
n2
n=1
4n (n!)2
∞
 log(n)
∞
 sin(n) n)  
l) 1
n 3 log(n) n=1 n sin
n=2 n
∞   ∞
  
1 n  n n+1
m) (−1)n 1 − o) log(2) log
n=1
2n n=1
n
2.7. Solved Exercises 119

∞
 1 a) Show that Leibniz’s test cannot be applied.
p)  p , p ∈ R
 
n=3 n log(n) log log(n) b) Show that the series is divergent.

18. Let (an ) be a sequence of real numbers. Consider

12. a) Test for convergence or divergence the series
∞ the sequence (bn )⎧defined by
(n + 1)n
. ⎨a n+1 , if n is odd
n=1
3n n! .bn = 2
⎩−a n , if n is even.
b) Based on the previous item, find the limit of 2
(n + 1)n
the sequence with general term . Show that:
3n n!
13. Test for convergence or divergence the series an a) The series bn is convergent if and only if
where lim an = 0.
⎧  n2
⎪ b) The series bn is absolutely convergent if
⎪
⎪
n
, if n = 3k, k ∈ N
⎪
⎪ and only if the series an is absolutely con-
⎪
⎪ n+1
⎪
⎪
⎨1 vergent.
.an = , if n = 3k + 1, k ∈ N0
⎪
⎪ n! 19. Show that if (an ) is a decreasing sequence and an
⎪
⎪
⎪
⎪ (−1)n converges, then lim(n an ) = 0.
⎪
⎪ if n = 3k + 2, k ∈ N0 .
⎪
⎩ (n + 1)n
,
Hint: Relate the convergence of the series to the
If convergence occurs, indicate whether it is condi- fact that the sequence of partial sums is a Cauchy
tional or absolute. sequence (choose m and n conveniently.)

20. Let an be a convergent series of positive terms.

14. Investigate, by two distinct processes, the conver-  √a n 1
gence or divergence of the series an where Show that the series converges if p > .
np 2
⎧ n 2
⎪
⎪ , if n is even a2 + b 2
⎪ 2
⎨ Hint:Consider the inequality ab ≤ , ∀a, b ∈ R.
2
.an =
⎪
⎪ (n + 1)(n − 1)
⎪
⎩ , if n is odd. 21. Let an be a sequence of positive terms such that
22 an+1 1
≥ 1 − , ∀n ∈ N. Show that the series an
an n
If convergence occurs, indicate whether it is condi- is divergent.
tional or absolute.
22. Let g : R → R be a function of class C 2 . Suppose
15. Let (an )n∈N be a sequence of positive terms such there exists p ∈ N such that g, g  , (g  )2 −g g  are all
that the series nan is convergent. Prove that the positive functions on [p, +∞[. Show that the series
series (an )2 is convergent. ∞
 g  (n)
.
16. Show that, if an > 0, ∀n ∈ N, and the series an n=p
g(n)
converges, then the series log(1 + an ) converges.
is convergent if and only if lim g(x) exists and is
∞
 x→+∞
n−1 finite.
17. Consider the series (−1) an where
n=1
⎧ 23. Test for convergence or divergence the series
⎪
⎪
⎪1,
⎪ if n is odd ∞
⎨n 
.an = . arctan(vn )
⎪
⎪ 1
⎪
⎪
n=2
⎩ n2 , if n is even.
 
π
where v2 = k > 0 and vn+1 = vn sin , n ≥ 2.
n
120 Numerical Series

24. The Sierpinski triangle is a geometric figure con- to the sides of the triangle and to the circles that
structed as follows: A square is divided into four precede and follow it in the sequence. Find the sum
equal squares, and the upper right square is removed. of the areas of the circles.
Each of the remaining three squares is divided in the
same way, and the upper right square of each is re-
moved.

Continuing this process to infinity, we obtain a figure

called the Sierpinski triangle.

26. Consider a square inscribed in a circle of radius 1. In

this square, inscribe a circle and in it a new square,
and so on, as suggested by the figure. Find the area
of the colored region.

Suppose the length of the side of the initial square

is equal to 1. What is the total area of the squares
removed from the initial square? What can we con-
clude about the area of the Sierpinski triangle?

25. Consider an equilateral triangle with side length 1.

Construct a sequence of circles approaching the ver-
tices of the triangle such that each circle is tangent
2.7. Solved Exercises 121

SOLUTIONS

∞
 1
1. a) Let an be the general term of the series . Since n2 + 3n + 2 = (n + 1)(n + 2), we get
n=1
n2 + 3n + 2

1 A B
an = = +
n2 + 3n + 2 n+1 n+2
A(n + 2) + B(n + 1)
. =
(n + 1)(n + 2)
(A + B)n + 2A + B
= ,
(n + 1)(n + 2)

from which
 
A+B =0 A=1
. ⇔
2A + B = 1 B = −1.

Therefore,
1 1
.an = − .
n+1 n+2
1
The series is telescopic with αn = and k = 1, using the notation of formula (2.1) on page 74. In
n+1
this case, lim αn = 0, so we can conclude that the series converges and, using formula (2.2) on the page
mentioned above, its sum is
∞  
1 1 1
. − = α1 = .
n=1
n + 1 n + 2 2

Alternatively, we can write the sequence of partial sums and determine its limit as follows:

1 1
S1 = −
2 3
1 1 1 1 1 1
S2 = − + − = −
2 3 3 4 2 4
1 1 1 1 1 1
.
S3 = − + − = −
2 4 4 5 2 5
..
.
1 1
Sn = −
2 n+2
.
.
.

1 1
As lim Sn = , the series is convergent and its sum is :
2 2

∞
 1 1
. = .
n=1
n2 + 3n + 2 2
122 Numerical Series

∞
  
+∞
  
b) Let us consider the series arctan(n + 3) − arctan(n) = − arctan(n) − arctan(n + 3) . It is
n=1 n=1
telescopic with αn = arctan(n) and k = 3, using the notation of formula (2.1) on page 74. In this case,
π
lim αn = , therefore, we can conclude that the series converges, and, using the formula (2.2), its sum is
2
∞
 π 5π
. − (arctan(n) − arctan(n + 3)) = − α1 + α2 + α3 − 3 · = − arctan(2) − arctan(3).
n=1
2 4

An alternative path to solve this problem is to write the sequence of partial sums and find its limit:
π
S1 = arctan(4) − arctan(1) = arctan(4) −
4
π
S2 = arctan(4) − + arctan(5) − arctan(2)
4
π
S3 = arctan(4) − + arctan(5) − arctan(2) + arctan(6) − arctan(3)
4
π
S4 = arctan(4) − + arctan(5) − arctan(2) + arctan(6) − arctan(3) + arctan(7) − arctan(4)
4
π
= − + arctan(5) − arctan(2) + arctan(6) − arctan(3) + arctan(7)
4
π
. S5 = − + arctan(5) − arctan(2) + arctan(6) − arctan(3) + arctan(7) + arctan(8) − arctan(5)
4
π
= − − arctan(2) + arctan(6) − arctan(3) + arctan(7) + arctan(8)
4
π
= − − arctan(2) − arctan(3) + arctan(6) + arctan(7) + arctan(8)
4
.
.
.
π
Sn = − − arctan(2) − arctan(3) + arctan(n + 1) + arctan(n + 2) + arctan(n + 3)
4
.
..

π 3π 5π
As lim Sn = − − arctan(2) − arctan(3) + = − arctan(2) − arctan(3), the series is convergent
4 2 4
and

+∞
  5π
. arctan(n + 3) − arctan(n) = − arctan(2) − arctan(3).
n=1
4

∞  
1 1 1
c) The series n
− n+1
is telescopic, with αn = n and k = 1, using the notation of formula (2.1)
n=0
e e e
1
on page 74. As lim n = 0, the series is convergent. The sum of the series is
e
∞ 
 
1 1 1 1
. − = 0 − lim n = 1.
n=0
en en+1 e e
 
1 1
Another process to find the sum is to write the general term in the form 1− , highlighting the fact
en e
1 1
that this series is geometric with ratio r = . As 0 < < 1, the series converges and:
e e
∞   ∞     ∞  
1 1 1 1 1  1 1 1
. − = 1 − = 1 − = 1 − = 1.
n=0
en en+1 n=0
en e e n=0 en e 1− 1
e
2.7. Solved Exercises 123

∞ 
 
1 1 1
d) The series − is telescopic, with αn = and k = 1, using the
n=1
arctan(n) arctan(n + 1) arctan(n)
2
notation of formula (2.1) on page 74. In this case, lim αn = . Therefore, we can conclude that the series
π
converges and, using formula (2.2), has sum
∞ 
 
1 1 1 2 4 2 2
. − = − = − = .
n=1
arctan(n) arctan(n + 1) arctan(1) π π π π

Alternatively, we can write the sequence of partial sums and find its limit:
1 1 4 1
S1 = − = −
arctan(1) arctan(2) π arctan(2)
4 1 1 1 4 1
S2 = − + − = −
π arctan(2) arctan(2) arctan(3) π arctan(3)
4 1 1 1 4 1
S3 = − + − = −
. π arctan(3) arctan(3) arctan(4) π arctan(4)
.
..
4 1
Sn = −
π arctan(n + 1)
.
.
.

As  
4 1 4 2 2
. lim Sn = lim − = − = ,
π arctan(n + 1) π π π
the series is convergent, and its sum is
∞ 
 
1 1 2
. − = .
n=1
arctan(n) arctan(n + 1) π

∞
 ∞

1 1
e) The following two series and are geometric series. To see this, observe that:
n=1
22n−1 n=1
23n−2

∞
 ∞
 ∞ ∞
1 1 1 1
. = = 2 = 2
n=1
22n−1 n=1
22n · 2−1 n=1
22n
n=1
4n

and
∞
 ∞
 ∞ ∞
1 1 1 1
. = = 4 = 4 .
n=1
23n−2 n=1
23n · 2−2 n=1
23n
n=1
8n

1 1
The first series has ratio r = , and the second has ratio r = . The series are convergent as in both cases
4 8
|r| < 1. The sum of the series can be calculated as follows:
∞
 1 ∞
 1
1 2 1 4
. =2· 4
= and =4· 8
= .
n=1
22n−1 1− 1
4
3 n=1
23n−2 1− 1
8
7

∞ 
 
1 1 2 4 2
By Theorem 2.2.2 the series − is convergent, and its sum is equal to − = .
n=1
22n−1 23n−2 3 7 21
124 Numerical Series

∞ ∞   ∞ ∞
3n + 7n 1 1 1 1
f) The series n · 7n
can be written in the form n
+ n
. Both n
and n
are
n=0
3 n=0
7 3 n=0
7 n=0
3
1 1
geometric series with ratios r = and r = , respectively.
7 3
The series are convergent as in both cases |r| < 1. Using the formula for the sum of a convergent geometric
series, we get
∞ ∞
1 1 7 1 1 3
. = = and = = .
n=0
7n 1 − 1
7
6 n=0
3n 1 − 1
3
2

∞
3n + 7n
By applying Theorem 2.2.2, we can conclude that the given series is convergent, and its sum
n=0
3n · 7n
7 3 8
is equal to + = .
6 2 3
∞
1 2 3 1
g) The series n−1
· n · n+1 is geometric with ratio r = because
n=0
4 5 6 120

1 2 3 6·4 4
. · · = n n n = .
4n−1 5n 6n+1 4 ·5 ·6 ·6 120n
Thus, we can calculate the sum as follows:
∞
 ∞
1 2 3 1 1 480
. · · = 4 =4· = .
n=0
4n−1 5n 6n+1 n=0
120 n 1 − 1
120
119

h) Using the properties of logarithms, we can obtain the following equality:

∞
 ∞ 
 
log( n+2 ) 1 1
.
n
= − .
n=2
log(n) log(n + 2) n=2
log(n) log(n + 2)

1
Thus the series is telescopic, with αn = and k = 2, using the notation of formula (2.1) on page 74.
log(n)
1
The series is convergent because lim = 0, and
log(n)
∞ 
 
1 1 1 1
. − = + .
n=2
log(n) log(n + 2) log(2) log(3)

∞
 ∞
1 1
2. By reindexing the series, we can write = . Additionally,
n=2
(n + 1)2 n=3
n2

∞ ∞
1 1 1
.
2
=1+ + 2
.
n=1
n 4 n=3
n

∞ ∞
1 π2 1 π2 1 π2 5
Knowing that 2
= , then 2
= −1− = − , that is,
n=1
n 6 n=3
n 6 4 6 4

∞
 1 π2 5
.
2
= − .
n=2
(n + 1) 6 4
2.7. Solved Exercises 125

∞
1 1
The series n
is convergent because it is geometric with ratio . By Theorem 2.2.2, the sum of these two
n=2
π π
series is convergent and
∞ 
  ∞
 ∞ 1
1 1 1 1 π2 5 2 π2 5 1
. + n = + = − + π 1 = − + :
n=2
(n + 1)2 π n=2
(n + 1) 2
n=2
π n 6 4 1 − π
6 4 π(π − 1)

3. a) The rational number 0, 5 = 0, 55555 . . . can be written in the form of a fraction using geometric series. By
writing
5 5 5 5
.0, 5 = 0, 5 + 0, 05 + . . . = + 2 + 3 + 4 + ···
10 10 10 10
1 5
it is clear that we have a geometric series with ratio r = and first term . Therefore,
10 10
5
10 5
.0, 5 = = .
1− 1
10
9

b) As in the previous item, we can use geometric series to write the rational number 0, 34 = 0, 343434 . . . in
the form of a fraction. By writing

34 34 34 34
.0, 34 = 0, 34 + 0, 0034 + . . . = + 4 + 6 + 8 + ···
102 10 10 10
1 34
it is evident that we have a geometric series with ratio r = and first term . Therefore,
102 102
34
102 34
.0, 34 = = .
1− 1
102
99

c) We can write the rational number 1, 345 in the following form:

345 345 345 345

.1, 345 = 1 + 0, 345 + 0, 000345 + . . . = 1 + + 6 + 9 + 12 + · · ·
103 10 10 10
1 345
After the first term, we have a geometric series with ratio r = and first term . Therefore,
103 103
345
103 345 448
.1, 345 =1+ =1+ = .
1 − 1013 999 333

d) By writing the rational number 0, 324101 in the form

324 101 101 101

.0, 324101 = 0, 324 + 0, 000101 + 0, 000000101 + . . . = + 6 + 9 + 12 + · · ·
103 10 10 10
1 101
we see that after the first term, we have a geometric series with ratio r = and first term 6 . Therefore,
103 10
101
324 6 324 101 323777
.0, 324101 = + 10 1 = + = .
103 1 − 103 103 999 × 103 999000
126 Numerical Series

∞
 2 n2 − 1
4. a) The series is of positive terms. To study this series, we will compare it with the series
n=1
3 n5 + 2 n + 1
∞
1
, which is convergent because it is a p-series with p = 3:
n=1
n3

2 n2 − 1
3 n5 + 2 n + 1 (2 n2 − 1) n3 2 n5 − n3 2
. lim = lim = lim = .
1 5
3n + 2n + 1 3 n5 + 2 n + 1 3
n3

2
Since ∈ R+ , according to Corollary 2 of the General Comparison Test, we can conclude that the given
3
∞
2 n2 − 1
series 5 + 2n + 1
is convergent.
n=1
3 n
∞ 
  2  π
1 1
b) The series sin is of positive terms because ∈ ]0, 1] ⊂ 0, , ∀n ∈ N. To determine its
n=1
n n 2
∞
1
convergence, we compare it with the series 2
, which we know is convergent as it is a p-series with
n=1
n
p = 2. We calculate the limit
  2 ⎡   ⎤2
1 1
sin sin
n ⎢ n ⎥
. lim ⎢
= lim ⎣ ⎥ = 1,
1 1 ⎦
n2 n

which belongs to R+ . Therefore, by Corollary 2 of the General Comparison Test, the given series is conver-
gent.

∞
log(n7 + 1)
c) The series is a series of positive terms. We can analyze this series by comparing it with
n=1
n2
∞
1 3
the series 3/2
, which is convergent as it is a p-series with p = :
n=1
n 2

log(n7 + 1)
n2 n3/2 log(n7 + 1) log(n7 + 1)
. lim = lim = lim = 0.
1 n 2 n1/2
n 3/2

∞
 1
Since the series 3/2
is convergent, the fact that this limit is zero allows us to conclude, by Corollary 3
n=1
n
∞
log(n7 + 1)
of the General Comparison Test, that the series is also convergent.
n=1
n2

log(n7 + 1)
Note: For calculating the limit lim , we used L’Hôpital’s Rule applied to the calculation of
n1/2
log(x7 + 1) ∞
lim , where the indeterminate form of type ∞
arises:
x→+∞ x1/2
2.7. Solved Exercises 127

  7x6
√
log(x7 + 1) x7 +1 14x6 x
. lim = lim = lim = 0.
x→+∞ (x1/2 ) x→+∞ 1 x→+∞ x7 + 1
√
2 x

d) Let us begin by rewriting the general term of the series:

 
1 √ 1 √
 n+ − n n+ + n 1
1 √ n n
n
an = n+ − n=  = 
n 1 √ 1 √
n+ + n n+ + n
. n n
1 1 1
=  = ⎛ ⎞ = √ .
1 √ n2 +1 √ ⎠ n3 + n + n3
n n+ + n
n n⎝ + n
n
∞
 1
As an > 0, ∀n ∈ N, the series is of positive terms. Let us compare it with the series , which is
n=1
n3/2
3
convergent because it is a p-series with p = . The limit
2
1
√ √ √
n3 + n + n3 n3 1
. lim = lim √ =
1 n3 +n+ n3 2
n3/2

is finite
and different from zero; therefore, by Corollary 2 of the General Comparison Test, the series
∞
1 √
n + − n is convergent.
n=1
n

∞ √ √
 n+1− n
e) The series √ √ is of positive terms. Considering that
n=1
n+1+ n
√ √ √ √ √ √ 
n+1− n n+1− n n+1+ n 1
.√ √ = √ √ 2 = √ √ 2 ,
n+1+ n n+1+ n n+1+ n
∞
 1
we can write the series in the form √ √ 2 . We will analyze this series by comparing it with
n=1 n+1+ n
the harmonic series:
1
√ √ 2
n+1+ n n 1
. lim = lim √ √ 2 = .
1 n+1+ n 4
n
1
Since ∈ R+ , we can conclude, by Corollary 2 of the General Comparison Test, that the given series
√ 4 √
∞
n+1− n
√ √ is divergent.
n=1
n+1+ n
128 Numerical Series

∞
 cos(n)
f) The series 2+4
is not of positive terms. However, by bounding the modulus of the general term, we
n=1
n
can obtain the following inequality:
! !
! cos(n) ! 1 1
.0 < !! 2 !≤ ≤ 2 , ∀n ∈ N. (2.5)
n + 4! n2 + 4 n

∞
1
It is worth noting that the series 2
is convergent since it is a p-series with p = 2. Using the inequality
n=1
n
∞ ! !
! cos(n) !
(2.5), we can apply the General Comparison Test to show that the series ! !
! n2 + 4 ! converges. Therefore,
n=1
∞
cos(n)
the original series 2+4
is absolutely convergent.
n=1
n

g) Recall that −1 ≤ cos(n) ≤ 1, ∀n ∈ N. Using this fact, we obtain n3 + 10 cos(n) ≥ n3 − 10, ∀n ∈ N. Thus,
we have that
.n
3
+ 10 cos(n) ≥ n3 − 10 > 0, ∀n ≥ 3.

This inequality implies that

1 1
.0 < ≤ 3 , ∀n ≥ 3. (2.6)
n3 + 10 cos(n) n − 10

∞
 1
The series is of positive terms, and we can study its convergence by comparing it with the
n=3
n3 − 10
∞
1
series 3
, which we know to be convergent. Since
n=3
n

1
n3 − 10 n3
. lim = lim 3 =1
1 n − 10
n 3

∞

1
is a positive real number, the series 3 − 10
is also convergent. Using inequality (2.6) and the General
n=3
n
∞
1
Comparison Test, we can conclude that the series 3 + 10 cos(n)
is convergent. This series differs from
n=3
n
∞
1
the given series only in a finite number of terms; therefore, the series converges.
n=1
n3 + 10 cos(n)

∞
1 − (−1)n
h) Consider the series √ , where an is its general term. We can express an as follows:
n=1
n

⎧
⎪
⎨0, if n is even
.an = 2
⎪
⎩√ , if n is odd.
n
2.7. Solved Exercises 129

∞ ∞
1 − (−1)n 2
Hence, √ = √ . This is a series of positive terms. Let us now consider the series
n=1
n n=1
2n −1
∞
1 1
√ , which is divergent because it is a p-series with p = . The limit
n=1
n 2

2
√ √ 
2n − 1 2 n 4n √
. lim = lim √ = lim = 2
1 2n − 1 2n − 1
√
n

belongs to R+ ; therefore, by Corollary 2 of the General Comparison Test, the given series is divergent.
∞
 ∞
log(n) log(n)
5. a) The series (−1)n = (−1)n is alternating. To begin, let us study the series of modules
n=1
n n=2
n
∞
log(n)
, comparing it with the harmonic series. We can see that
n=2
n

log(n)
n
. lim = lim log(n) = +∞.
1
n

This result allows us to conclude, by Corollary 3 of the General Comparison Test, that the series of modules
is divergent. Since the initial series is alternating, let us check if we can apply Leibniz’s test:
log(n) log(x) ∞
(i) lim = 0 (notice that lim is an indeterminate form of type that can be solved
n x→+∞ x ∞
(log(x)) 1
using L’Hôpital’s Rule: lim = lim = 0).
x→+∞ (x) x→+∞ x
log(n)
(ii) > 0, ∀n > 1.
n
log(x) 1 − log(x)
(iii) The function f (x) = is decreasing on ]e, +∞[. Indeed, f  (x) = < 0 ⇔ x > e,
 x  x2
log(n)
which implies that is decreasing for n ≥ 3.
n
Therefore, we can state that the alternating series is convergent. As the series of modules is divergent, the
given series is conditionally convergent, taking into account Definition 2.4.1.

∞ ∞
log(n) log(n)
Note: The series and are both divergent since they differ only in a finite number of
n=2
n n=3
n
terms.
∞
 2 2
b) The series (−1)n+1n + e−n
is alternating, given that n > 0, ∀n ∈ N. To begin with, let us
n=1
e e + e−n
study the series of modules:

∞ ! ! ∞
! 2 ! 2
!(−1)n+1 != .
. ! e n + e −n ! e n + e−n
n=1 n=1
130 Numerical Series

We have

2 2
.0 < < n , ∀n ∈ N. (2.7)
en + e−n e
∞ ∞  n
1 1 1 1
The series n
= is geometric with ratio r = . It is convergent because |r| = < 1. By
n=1
e n=1
e e e
∞
2
Theorem 2.2.2, the series is also convergent. Using the General Comparison Test and considering
n=1
en
∞
2
inequality (2.7), the series n + e−n
is convergent. As this is the series of modules of the given series,
n=1
e
by Definition 2.4.1, we can conclude that the initial series is absolutely convergent.
∞
1 1
c) The series (−1)n is alternating because an = > 0, ∀n ∈ N. To study the convergence
n=1
log(3n) log(3n)
∞
1
of the series, let us first consider the series of modules, . We can compare this series with the
n=1
log(3n)
harmonic series: 1
log(3n) n
. lim = lim = +∞.
1 log(3n)
n
∞

1
It follows by Corollary 4 of the General Comparison Test that the series is divergent. Since the
n=1
log(3n)
initial series is alternating, we can check the conditions of Leibniz’s test to see if it converges:
1
(i) lim = 0.
log(3n)
1
(ii) > 0, ∀n ≥ 1.
log(3n)
 
  1
(iii) The sequence log(3n) is increasing; therefore, is a decreasing sequence.
log(3n)
According to Leibniz’s test, the alternating series is convergent. However, as the series of its modules is
divergent, the given series is conditionally convergent, considering Definition 2.4.1.

∞ 1 1
 sin sin 1 π
d) The series (−1)n n
is alternating, because an = n
> 0, ∀n ≥ 1 (recall that 0 < < ,
n n n 2
n=1
∞ 1 ∞
 sin  1
∀n ≥ 1). The series of modules is n
. Consider the convergent series 2
, which is a p-series
n=1
n n=1
n
with p = 2. The limit
1  
sin n 1
sin
n n
. lim = lim =1
1 1
n2 n
is finite and different from zero; therefore, by Corollary 2 of the General Comparison Test, the series of
modules is convergent; thus, considering Definition 2.4.1, we can conclude the given series is absolutely
convergent.
2.7. Solved Exercises 131

∞
 3n 3n
e) The series (−1)n 2n+1
is alternating because 2n+1 > 0, ∀n ∈ N. To analyze the absolute convergence
n=1
e e
of the series, we consider the series of modules:

∞ ! !
∞ ∞  n ∞  
!
!(−1)n 3
n !
!= 3n 1 3 1  3 n
. ! 2n+1 ! 2n+1
= 2
= 2
.
n=1
e n=1
e n=1
e e e n=1 e

∞ 
 
3 n 3
The series 2
is geometric with ratio r = 2 . Since |r| < 1, the series is convergent. We know by
n=1
e e
Theorem 2.2.2 that the product of a constant by a convergent series is convergent; therefore, the series of
modules is convergent. Then, the given series is absolutely convergent, again considering Definition 2.4.1.
f) Considering that ⎧
  ⎨0, if n is even
nπ
. sin =
2 ⎩(−1) n−1
2 , if n is odd,
∞  nπ 
 sin 2
we can write the series √ as
n=1
n+1

∞ ∞
(−1)n−1 1  (−1)n−1
. √ = √ √ .
n=1 2n 2 n=1 n

∞
 1
This series is alternating, and its series of modules is √ , which is divergent. We will now apply
n=1
n
Leibniz’s test to determine the convergence of our given series:
1
(i) lim √ = 0.
n
1
(ii) √ > 0, ∀n ∈ N.
n
 
1 1 1
(iii) √ − √ > 0, ∀n ∈ N, and therefore, the sequence √ is decreasing.
n n+1 n
Thus, the alternating series is convergent. However, since the series of absolute values is divergent, we know
the given series is conditionally convergent.

∞
 ∞
1 1
6. a) The series (−1)n+1 is alternating. The series of modules is the harmonic series, which is
n=1
n n=1
n
divergent.
To determine the convergence of the alternating series, let us apply Leibniz’s test:
1
(i) lim = 0.
n
1
(ii) > 0, ∀n ∈ N.
n
1 1 1
(iii) − = > 0, ∀n ∈ N, and therefore, the sequence is decreasing.
n n+1 n(n + 1)
Thus, the alternating series is convergent. As its series of modules is divergent, the given series is conditionally
convergent.
132 Numerical Series

b) From the Corollary of Theorem 2.3.2, we know that |S − Sn | ≤ an+1 . It is sufficient to require that
1
an+1 < , that is, n > 999. Therefore, the intended sum is S1000 .
1000
c) By the corollary referred to in the previous item, |S − S5 | ≤ a6 = 1
6
, and this value is an upper bound of
the error.

∞
nn
7. a) The series is of positive terms. Let an be the general term of the series. We can find the limit
n=1
π n n!
of the ratio of successive terms as follows:
(n + 1)n+1
an+1 π n+1 (n + 1)! (n + 1)n+1 π n n!
lim = lim n
= lim n n+1
an n n π (n + 1)!
.
π n n!
 
(n + 1)(n + 1)n n! 1 n+1 n e
= lim n
= lim = .
n π (n + 1) n! π n π
Since this value is less than 1, D’Alembert’s test guarantees that the series is convergent.
∞
 n 2n
b) The series 3+1
is of positive terms. Let an be the general term of the series. We will take the
n=0
4 n
limit of the ratio of successive terms of the series:
(n + 1)2n+1
an+1 4(n + 1)3 + 1 (n + 1) 2n+1 (4n3 + 1)
lim = lim n
= lim  
an n2 n 2n 4(n + 1)3 + 1
.
4n3 + 1
2(n + 1)(4n3 + 1) 2 (n + 1) (4n3 + 1)
= lim   = lim = 2.
3
n 4(n + 1) + 1 n (4n3 + 12n2 + 12n + 5)
As this limit has a value greater than 1, the series is divergent by D’Alembert’s test.

c) Noting that
∞ ∞ ∞
(n + 1)! − n! (n + 1)n! − n! n
. = = ,
n=1
n! (n + 1)! n=1
n! (n + 1)! n=1
(n + 1)!
let an be the general term of the series. As
n+1
an+1 (n + 2)! (n + 1)(n + 1)! n+1
. lim = lim = lim = lim = 0 < 1,
an n n(n + 2)! n(n + 2)
(n + 1)!
it follows from D’Alembert’s test that the series is convergent.

∞
2 · 4 · 6 · · · · · (2n + 2)
d) The series is of positive terms. Let an be the general term of the series. We
n=1
1 · 4 · 7 · · · · · (3n + 1)
have
2 · 4 · 6 · · · · · (2n + 2)(2n + 4)
an+1 1 · 4 · 7 · · · · · (3n + 1)(3n + 4) 2n + 4 2
. lim = lim = lim = .
an 2 · 4 · 6 · · · · · (2n + 2) 3n + 4 3
1 · 4 · 7 · · · · · (3n + 1)
2.7. Solved Exercises 133

As this value is less than 1, we can conclude by D’Alembert’s test that the series is convergent.
∞
 (n + 1)!
e) The series √ is of positive terms. Let an be the general term of the series. Evaluating
n=1
nn 3n + 2
an+1
lim , we obtain
n→∞ an
(n + 2)!
√ √
an+1 (n + 1)n+1 3n + 5 (n + 2)! nn 3n + 2
lim = lim = lim √
an (n + 1)! (n + 1)! (n + 1)n+1 3n + 5
. n
√
n 3n + 2
√   n
(n + 2) nn 3n + 2 n + 2 3n + 2 n 1
= lim n+1
√ = lim = .
(n + 1) 3n + 5 n + 1 3n + 5 n + 1 e
Since this value is less than 1, the series is convergent according to D’Alembert’s test.

 ∞
3n + n!
f) The series is of positive terms. It is easy to prove by induction that 3n < n!, ∀n ≥ 7, which
n=1
n! + nn
implies that
3n + n! 2 n!
. ≤ n , ∀n ≥ 7.
n! + nn n
∞
2 n!
Let an be the general term of the series . D’Alembert’s test confirms that the series converges
n=1
nn
since
2 (n + 1)!
 n
an+1 (n + 1)n+1 (n + 1)! nn n 1
. lim = lim = lim = lim = < 1.
an 2 n! n! (n + 1)n+1 n+1 e
nn
∞
3n + n!
By the General Comparison Test, we can conclude that the series also converges.
n=1
n! + nn
Note: Two series that differ only in a finite number of terms are both convergent or both divergent.
∞
 n!
g) The series is of positive terms. We can establish the inequality
n=1
(2 n)! + 2n

n! n!
. ≤ , ∀n ∈ N.
(2 n)! + 2n (2 n)!
∞
 n!
Let us investigate the convergence of the series . If an denotes the general term of this series,
n=1
(2 n)!
then
(n + 1)!
 
an+1 2 (n + 1) ! (n + 1)! (2 n)! n+1
. lim = lim = lim   = lim = 0.
an n! n! 2 (n + 1) ! (2n + 2)(2n + 1)
(2 n)!
By D’Alembert’s test, the series is convergent since this limit is less than 1. Applying the General Comparison
∞
n!
Test, the original series is convergent.
n=1
(2 n)! + 2n
134 Numerical Series

∞
 ∞ ∞
n! + 3n n!
h) The series  2 is of positive terms. Let us consider the series an =  2 and
n=0 (n + 1)! n=0 n=0 (n + 1)!
∞
 ∞
 3n
bn =  2 . If both series are convergent, it follows from Theorem 2.2.2 that the given series
n=0 n=0(n + 1)!
a
is convergent. Let us begin by finding the limit of n+1 a
:
n

(n + 1)!
 2  2
an+1 (n + 2)! (n + 1)! (n + 1)! n+1
. lim = lim
n!
= lim   2
= lim = 0 < 1.
an n! (n + 2)! (n + 2)2
 2
(n + 1)!

∞

As the limit is less than 1, we can use D’Alembert’s test to conclude that the series an is convergent.
n=0
∞

Similarly, the series bn is convergent:
n=0
3(n + 1)
 2  2
bn+1 (n + 2)! (n + 1) (n + 1)! n+1
. lim = lim
3n
= lim   2
= lim = 0 < 1.
bn n (n + 2)! n(n + 2)2
 2
(n + 1)!
Since both series are convergent, we can conclude that the given series is also convergent as it is the sum
of two convergent series.

∞ 
 n2
2
8. a) Consider the series of positive terms 1− and let an represent its general term. Since
n=3
n


  2  
√ 2 n 2 n 1
= e−2 = 2 < 1,
n
. lim n
an = lim 1− = lim 1 −
n n e

the series converges by Cauchy’s Root Test.

∞
 ∞  n

2
2 1
b) The series e−n = is of positive terms. We have
n=0 n=0
e

  n2  n
n 1 1
. lim = lim = 0.
e e

As this value is less than 1, Cauchy’s Root Test ensures the convergence of the series.

∞
 √
n n
c) The series 2−1 is of positive terms. We find that:
n=1

√
n n √
n 
. lim
n
2−1 = lim 2 − 1 = 0,

√
since lim n 2 = 1. As this value is less than 1, Cauchy’s Root Test allows us to conclude that the series is
convergent.
2.7. Solved Exercises 135

∞ 
  n
π 1 π 1 π
d) The series cos + is of positive terms because 0 < + < , ∀n ∈ N. Applying
n=1
6 n 6 n 2
Cauchy’s Root Test, we conclude that the series is convergent. In fact:
  n   √
n π 1 π 1 π 3
. lim cos + = lim cos + = cos = < 1.
6 n 6 n 6 2

∞  n
1 n
e) The series + (−1)n is of positive terms. The sequence of general term
n=0
2 4n + 1
 n
1 n 1 n
.
n
+ (−1)n = + (−1)n
2 4n + 1 2 4n + 1

3 1
has two sublimits: and . Since the upper limit is less than 1, we can apply Corollary 1 of the Root Test
4 4
to conclude that the series is convergent.

∞  2
 n ∞  2 n ∞  2 n
n −2 1 n −2 n −2
f) We can write + =− . The series is of positive terms
n=1
3n2 3 n=2 3n 2
n=2
3n 2

and differs from the given series in only one term. We can state that the two series are either both convergent
or both divergent. Using Cauchy’s Root Test, we can show that the series converges because
 n
n n2 − 2 n2 − 2 1
. lim = lim = < 1.
3n2 3n2 3

∞
(−1)n 1
9. a) The series n)
is alternating because an = n)
> 0, ∀n ≥ 2; its general term can be written
n=2
log(n log(n
∞ ! ! ∞

(−1)n ! (−1)n ! 1
in the form . Let us start by studying the series of modules, ! ! = ,
n log(n) ! n log(n) ! n log(n)
  n=2 n=2
1
applying the Cauchy’s Condensation Test. The sequence is obviously decreasing. The
n log(n) n∈N\{1}
series
∞ ∞ ∞
2n 1 1 1
.
n n
= =
n=2
2 log(2 ) n=2
n log(2) log(2) n=2 n

is divergent as it is the product of a constant by the harmonic series. Then, the series of modules is divergent
by Cauchy’s Condensation Test. As a result, if convergent, the series under study is conditionally convergent.
Since the series is alternating and satisfies the following conditions:
1
(i) >0
n log(n)
1
(ii) lim =0
n log(n)
 
1
(iii) As mentioned earlier, the sequence is decreasing
n log(n) n∈N\{1}
it follows from the Leibniz’s test that the alternating series is convergent. From the study of the series of
modules, we can state that the series is conditionally convergent, considering Definition 2.4.1.
136 Numerical Series

b) Let us evaluate the limit of the general term of the series:

  n   n 
n2 2 n2 2 n4
. lim √ + 1+ = lim √ + lim 1 + = lim + e2 = e2 .
n5 +1 n n5 + 1 n n5 + 1

As the general term of the series does not converge to zero, the series is divergent (refer to Theorem 2.2.1).
 
1 1
c) First, we note that 0 < 1 − 2
< 1, ∀n ≥ 2, which implies that log 1 − 2 < 0, ∀n ≥ 2. Thus, the
n n
series is of negative terms. Next, we consider the series of modules

∞ !  ! ∞   
! !
!log 1 − 1 ! = 1
− log 1 − 2 .
. ! n 2 ! n
n=2 n=2

∞
1
We can compare this series with the series 2
, which is convergent because it is a p-series with p = 2.
n=2
n
The limit
 
1
− log 1 − 2     2
n 1 1 n
. lim = − lim n2 log 1 − 2 = − lim log 1 − 2 = − log(e−1 ) = 1
1 n n
n2

is finite and different from zero. Therefore, the series of modules is convergent by Corollary 2 of the
General Comparison Test. Then, by Theorem 2.4.1, the given series is absolutely convergent, considering
Definition 2.4.1.

∞
sin(n) + 2n
d) The series is of positive terms because −1 ≤ sin(n) ≤ 1 and 2n > 1, ∀n ∈ N. The
n=1
n + 5n
previous inequalities allow us to obtain the following estimate:
 n  n
sin(n) + 2n 1 + 2n 1 2n 1 2
.0 < n
< n
= n + n = + , ∀n ∈ N.
n+5 5 5 5 5 5

∞  n
 ∞  n
1 1 2
The series is convergent because it is geometric with ratio r = , and the series is
n=1
5 5 n=1
5
2
convergent because it is geometric with ratio r = . Then, the series
5

∞  n  n 
1 2
. +
n=1
5 5

is convergent by Theorem 2.2.2. By the General Comparison Test, the original series is convergent. As it is
a series of positive terms, it is absolutely convergent.

∞
 en+1 en+1
e) The series (−1)nn
is alternating because an = > 0, ∀n ∈ N. To determine the convergence
n=1
n nn
∞
en+1
of the series, we first examine the series of modules, , using D’Alembert’s test:
n=1
nn
2.7. Solved Exercises 137

en+2
 n
an+1 (n + 1)n+1 en+2 nn e n 1
. lim = lim = lim = lim = 0 × = 0.
an en+1 en+1 (n + 1)n+1 n+1 n+1 e
nn
As this value is less than 1, the series of modules is convergent. Therefore, the given series is absolutely
convergent by Theorem 2.4.1, considering Definition 2.4.1.

1 1
∞ (−1)n +
 ∞ (−1)n +
f) To study the convergence of the series n = n , we will start by rewriting its
n=1
n n=2
n
general term:
1
(−1)n + n
.
n = (−1)n · n + (−1) .
n n2
n + (−1)n
Let an = . It is evident that an > 0, ∀n ∈ N \ {1}, and therefore the series under study is
n2
∞
n + (−1)n
alternating. Let us study the series of modules, , by comparison with the harmonic series.
n=1
n2
The limit
n + (−1)n
n2 n(n + (−1)n ) n + (−1)n
. lim = lim = lim =1
1 n 2 n
n
belongs to R+ . As the harmonic series is divergent, the series of modules is also divergent, by Corollary 2
of the General Comparison Test.
1
∞
(−1)n ∞
1 (−1)n + (−1)n 1
Let us consider the series and . Taking into account that n = + 2
n n 2 n n n
n=1 n=1
and that the two previous series are convergent (the first one is the alternating harmonic series, and the
1
∞ (−1)n +
second one is a p-series with p = 2), the series n is convergent by Theorem 2.2.2. From
n=1
n
the study of the series of modules, we can affirm that the series is conditionally convergent, according to
Definition 2.4.1.
 
n + (−1)n
Note: The Leibniz’s Test is not applicable because the sequence is not monotonic (see
n2
Fig. 2.6).
 
1 1
g) Bearing in mind that 1 + n > 1, ∀n ≥ 1, we conclude that log 1 + n > 0, ∀n ≥ 1. Thus, the
2 2
∞   ∞
1 1
series log 1 + n is of positive terms. Let us compare this series with the series n
, which is
n=1
2 n=1
2
1
convergent as it is a geometric series with ratio r = . The limit
2
 
1
log 1 + n     n
2 1 1 2
. lim = lim 2n log 1 + n = lim log 1 + n = log(e) = 1
1 2 2
2n
138 Numerical Series

n + (−1)n n + (−1)n
Figure 2.6: The sequences (−1)n and
n2 n2

is finite and different from zero; therefore, the series is convergent by Corollary 2 of the General Comparison
Test. It is absolutely convergent since it is a series of positive terms.

∞
 ∞

 √ n √ n
h) The series 1− n
n = (−1)n n n − 1 is an alternating series. Let us begin by studying the
n=1 n=1
series of modules using Cauchy’s Root Test:
" √ n √ 
. lim
n n
n − 1 = lim n n − 1 = 0

√
since lim n n = 1. As this value is less than 1, the series of modules is convergent. It follows by Theo-
rem 2.4.1 that the given series is convergent, and by Definition 2.4.1, it is absolutely convergent.

∞
 (−1)n
i) To determine the convergence of the series , we first rewrite its general term as
n=1
1 − (−1)n n2

(−1)n 1 −1
. = = 2 .
1 − (−1)n n2 (−1)n − n2 n + (−1)n+1

We can simplify the series as follows:

∞
 ∞ ∞
(−1)n −1 1
. = =− .
n=1
1 − (−1) n
n 2
n=1
n 2 + (−1) n+1
n=1
n 2 + (−1)n+1

Note that the general term of this last series is positive, whatever n ∈ N. We can compare the series with
∞
1
the series , which we know is convergent because it is a p-series with p = 2. The limit
n=1
n2

1
n2 + (−1)n+1 n2 1
. lim = lim 2 = lim =1
1 n + (−1)n+1 (−1)n+1
1 +
n2 n2

belongs to R+ . Therefore, by Corollary 2 of the General Comparison Test, the series is convergent, and the
convergence is absolute.
2.7. Solved Exercises 139

∞
 ∞ ∞ ∞  2n ∞
(−1)n + 4 3
j) Consider the series an = and b n = . The series an is of
n=1 n=1
n + 4n n=1 n=1
n+2 n=1
positive terms; in fact, the expression for an is as follows:
⎧
⎪ 5
⎪
⎨ if n is even
n + 4n
.an =
⎪
⎪ 3
⎩ if n is odd.
n + 4n
∞
5 5
Consequently, an ≤ , ∀n ∈ N. The series is a convergent geometric series because the ratio is
4n n=1
4n

∞
1 5
r = . Since an ≤ n , ∀n ∈ N, the General Comparison Test implies that an is convergent.
4 4 n=1
∞

We can use the Cauchy’s Root Test to study the series of positive terms, bn :
n=1
 2n  2
n 3 3
. lim = lim = 0 < 1.
n+2 n+2
∞

This implies that the series bn is convergent.
n=1
∞
 ∞

Since both an and bn are convergent, we can conclude that the given series is convergent by
n=1 n=1
Theorem 2.2.2. Since it is a series of positive terms, it is absolutely convergent.

∞  
1 n
k) The series 1− has a general term that is not a null sequence. Specifically,
n=1
n
 n
1
. lim 1− = e−1 .
n
Theorem 2.2.1 can be applied to establish that this series is divergent.

∞
1
l) We will use the Cauchy Condensation Test to determine the convergence of the series  3 .
n log(n)
 n=2
1
It is clear that the sequence  3 is decreasing with positive terms. We can apply the Cauchy
n log(n)
Condensation Test by considering the series
∞
 ∞
 ∞
2n 1 1 1
.  3 =  3 =  3 3
,
n=2 2n log(2n ) n=2 n3 log(2) log(2) n=2
n

∞
1
which is convergent as it is the product of a constant by a convergent p-series, . Therefore, by the
n=2
n3
∞
 1
Cauchy Condensation Test, the series  3 is convergent. Furthermore, since it is a series of
n=2 n log(n)
positive terms, it is absolutely convergent.
140 Numerical Series

1
m) If > 0 for all x = 0, we can say that the integral is positive, which implies that the series consists of
x2
positive terms. As
# $
n+2 1 1 n+2 1 1
. dx = − = − ,
n x2 x n n n+2
we can write the series as
∞
 n+2 ∞  
1 1 1
. dx = − .
n=1 n x2 n=1
n n+2

1
This series is telescopic with αn = and k = 2, using the notation of formula (2.1) on page 74. The
n
∞  
1 1
telescopic series converges if (αn ) is a convergent sequence. As αn → 0, the series − is
n=1
n n+2
convergent. Furthermore, it is absolutely convergent since it consists of positive terms.
∞
 ∞
(−1)n 1
n) The series √ √ is alternating. The series of modules is the series √ √ ,
n=1
n+1+ 3n+1 n=1
n+1+ 3n+1
∞
1 1
which we can compare with the series 1/2
, which is divergent because it is a p-series with p = . As
n=1
n 2
the value of the limit
1
√ √ √
n+1+ 3n+1 n
. lim = lim √ √ =1
1 n+1+ 3n+1
n1/2
belongs to R+ , Corollary 2 of the General Comparison Test guarantees that the series of modules is divergent.
Let us study the alternating series by applying the Leibniz’s test:
1
(i) lim √ √ = 0.
n+1+ 3n+1
1
(ii) √ √ > 0, ∀n ∈ N.
n+1+ 3n+1
 
1 √ √ 
(iii) The sequence √ √ is decreasing, because n + 1 + 3 n + 1 is obviously increas-
n+1+ 3n+1
ing.

The test conditions are satisfied; therefore, the series is convergent. As the series of modules is divergent,
we conclude that the alternating series is conditionally convergent.

∞
 1
o) The series √ √  is of positive terms. Let us study this series by comparing it with the
n=1
n n+1+ n
∞
 1 3
series , which we know to be convergent, as it is a p-series with p = . The limit
n=1
n3/2 2

1
√ √ 
n n+1+ n n3/2 1
. lim = lim √ √  =
1 n n+1+ n 2
n3/2

belongs to R+ . It follows from Corollary 2 of the General Comparison Test that the given series is convergent.
It is absolutely convergent since it is a series of positive terms.
2.7. Solved Exercises 141

∞
     
1 1
p) The series (−1)n 1 − cosis alternating, because 1 − cos > 0, ∀n ≥ 1. The series of
n=1
n n
∞   
1
modules is the series 1 − cos . As the value of the limit
n=1
n

 
1
1 − cos
n 1
. lim =
1 2
n2

∞
1
belongs to R+ . As the series 2
is convergent because it is a p-series with p = 2, then by Corollary 2 of
n=1
n
the General Comparison Test, the series of modules is convergent. Therefore, by Theorem 2.4.1, the given
series is convergent, and it is absolutely convergent (see Definition 2.4.1).

∞
    
1 1
q) The series (−1)n 2n sin is alternating, because an = 2n sin > 0, ∀n ≥ 1 (note that
n=1
3n 3n
∞  
1 π 1
0 < n
< , ∀n ≥ 1). The series of modules is 2n sin n . Considering that if x > 0, then
3 2 n=1
3
sin(x) ≤ x, we get
   n  n
1 1 2
n
.2 sin ≤ 2n = , ∀n ∈ N.
3n 3 3
∞  n
 2 2
The series is geometric with ratio r = ; therefore, it is convergent. According to the General
n=1
3 3
Comparison Test, the series of modules is convergent, so the given series is absolutely convergent.

∞
 √ √
n! n n! n
r) The series (−1)n√ is alternating because √ > 0, ∀n ∈ N. Let us study the series
n=1
nn n + 1 nn n + 1
∞ √ √
n! n n! n
of modules, √ , using D’Alembert’s test. Taking a n = √ :
n=1
nn n + 1 nn n + 1

√
(n + 1)! n + 1
√
an+1 (n + 1)n+1 n + 2 (n + 1)! nn (n + 1)
lim = lim √ = lim
an n! n n! (n + 1)n+1 n(n + 2)
. √
nn n + 1
 n
(n + 1)2 nn n+1 n 1
= lim = lim = .
(n + 1)n+1 n(n + 2) n(n + 2) n + 1 e

Since this value is less than 1, the series is convergent. As it is the series of modules of the initial series, we
∞ √
n! n
can affirm that the series (−1)n n √ is absolutely convergent.
n=1
n n+1

∞  
√ 1 1 π
s) The series ( n + 1) sin 2
is of positive terms because 0 < 2 < . Let us consider the series
n=1
n n 2
142 Numerical Series

∞
 1 3
, which we know to be convergent because it is a p-series with p = . The limit
n=1
n3/2 2
   
√ 1 1
( n + 1) sin sin √
n2 n2 n+1
. lim = lim · √ =1
1 1 n
n3/2 n2

belongs to R+ , and therefore, by Corollary 2 of the General Comparison Test, the series is convergent. As
it is of positive terms, it is absolutely convergent.

∞
 1
t) The series is of positive terms. Knowing that n! < nn and that the logarithmic function is
n=2
log(n!)
increasing, we have
1 1 1
.0 < = < , ∀n ≥ 2.
log(nn ) n log(n) log(n!)
∞
 1
We proved in exercise 9a) that the series is divergent. By the General Comparison Test, the
n=2
n log(n)
∞
 1
series is divergent.
n=2
log(n!)

∞ ∞
n cos(nπ) n cos(nπ) n
10. a) The series √ = √ is alternating because cos(nπ) = (−1)n and √ > 0,
n 3+1 n 3+1 n 3+1
n=0 n=1
∞
n
∀n ∈ N. First, we will examine the series of modules, √ , by comparing it with the series
n=1 n3 + 1
∞
1 1
1/2
, which is divergent because it is a p-series with p = . The limit
n=1
n 2

n √
√
n3 + 1 n3
. lim = lim √ =1
1 3
n +1
n1/2
is finite and different from zero. Therefore, by Corollary 2 of the General Comparison Test, the series
∞
n
√ is divergent.
n 3+1
n=1
Next, we will apply Leibniz’s test to the alternating series:
n
(i) lim √ = 0.
n3 + 1
n
(ii) √ > 0, ∀n ∈ N.
n3 + 1
(iii) To verify that it is a decreasing sequence, consider the real function of real variable
x −x3 + 2
f (x) = √ . The derivative of f is f  (x) = √ , and therefore, f  (x) < 0,
√ x3 + 1 2(x3 + 1) x3 + 1
∀x > 3 2. The sequence then decreases for n ≥ 2.
Since the test conditions are satisfied, we can conclude that the alternating series is convergent. As the
series of modules is divergent, the given series is said to be conditionally convergent.
2.7. Solved Exercises 143

∞
 n!
b) The series is of positive terms. Let us study it using D’Alembert’s test. Let
n=1
3 × 5 × · · · × (2 n + 1)
an be the general term of the series:

(n + 1)!
an+1 3 × 5 × · · · × (2 n + 1)(2 n + 3) n+1 1
. lim = lim = lim = .
an n! 2n + 3 2
3 × 5 × · · · × (2 n + 1)

Since this limit is less than 1, it follows from D’Alembert’s test that the series is convergent. The series is
absolutely convergent, as it is of positive terms.
∞
1 + cos2 (n)
c) The series √ is of positive terms. We have the following inequality
n=1
n
1 1 + cos2 (n)
.√ ≤ √ , ∀n ∈ N.
n n
∞
 1 1
Since the series √ is divergent as it is a p-series with p = , the General Comparison Test allows us
n=1
n 2
to conclude that the given series is divergent.
∞
3n (n + 1)!
d) The series is of positive terms. Let us study it using D’Alembert’s test. Let an be the
n=0
(n + 1)n
general term of the series:

3n+1 (n + 2)!
 
an+1 (n + 2)n+1 n+1 n 3
. lim = lim n = 3 lim = .
an 3 (n + 1)! n+2 e
(n + 1)n

This limit value is greater than 1; therefore, by D’Alembert’s test, the series is divergent.

∞
  
(−1)n
e) We know that the function sin(x) is odd; therefore, the series sin is alternating and can be
n=1
n
∞   ∞  
1 1
written in the form (−1)n sin . Let us begin by studying the series of modules, sin by
n=1
n n=1
n
comparing it with the harmonic series. The limit
 
1
sin
n
. lim =1
1
n

belongs to R+ . It follows from Corollary 2 of the General Comparison Test that the series of modules is
divergent.
Let us study the alternating series by applying Leibniz’s test:
 
1
(i) lim sin = 0.
n
 
1
(ii) sin > 0, ∀n ∈ N.
n
144 Numerical Series
 
1 π
(iii) The sequence is decreasing, the sine function is increasing on [0, 2
], and therefore, the sequence
   n
1
sin is decreasing.
n
We conclude that the alternating series is convergent. Since the series of modules is divergent, the alternating
series is conditionally convergent.

∞
 √
n
f) The series is of positive terms. The limit
n=1
n2 + cos(n)
√
n
n2 + cos(n) n2
. lim = lim 2 =1
1 n + cos(n)
n3/2
∞
 1 3
is finite and different from zero. The series 3/2
is convergent, since it is a p-series with p = .
n=1
n 2
The General Comparison Test ensures that the original series is convergent. As it is of positive terms, it
converges absolutely.

∞
 3n 3n 3n
g) The series is of positive terms, and the inequality ≤ n holds in N. Let us study
n=1
2n + nn 2n + nn n
∞
3n
the series n
using the Cauchy’s Root Test. Let an be the general term of the series:
n=1
n

√ 3
. lim n an = lim = 0.
n

Since this limit is less than 1, the series is convergent. Using the General Comparison Test, we can affirm
that the series under study is convergent and it is absolutely convergent since it is of positive terms.

∞
1 + n(−1)n n + (−1)n
h) The series 3
is alternating, since we can write its general term as (−1)n . Let
n=1
1 + 2n 1 + 2n3
∞
n + (−1)n
us start by studying the series of modules, . The limit
n=1
1 + 2n3

n + (−1)n
1 + 2n3 n3 + (−1n )n2 1
. lim = lim =
1 2n3 + 1 2
n2

∞
 1
is finite and different from zero. Since the series 2
is a p-series with p = 2, it converges. Then,
n=1
n
∞
n + (−1)n
Corollary 2 of the General Comparison Test allows us to conclude that the series is con-
n=1
1 + 2n3
vergent. Therefore, by Theorem 2.4.1, the alternating series is convergent, and it converges absolutely (see
Definition 2.4.1).
2.7. Solved Exercises 145

∞
 2n 2n 2n
i) The series is of positive terms, and the inequality ≤ holds in N. Let us study the
n=1
n! + 1 n! + 1 n!
∞
2n
series using D’Alembert’s test. Let an be the general term of the series:
n=1
n!

2n+1
an+1 (n + 1)! 2
. lim = lim = lim = 0.
an 2n n+1
n!

This limit has a value less than 1; therefore, according to D’Alembert’s test, the series is convergent. By
the General Comparison Test, the initial series is convergent and absolutely convergent as it is of positive
terms.
∞
 ∞

n! 1
j) Let us consider the series and n
. The first one, which is of positive terms, can be studied
n=0
(2n)! n=0
2
using D’Alembert’s test. Let an be the general term of the series:

(n + 1)!
an+1 (2n + 2)! (n + 1)!(2n)! n+1
. lim = lim = lim = lim = 0.
an n! n!(2n + 2)! (2n + 2)(2n + 1)
(2n)!

This limit is less than 1; therefore, by D’Alembert’s test, the series is convergent. As it is of positive terms,
it is absolutely convergent.
1
The second series is geometric with ratio r = ; therefore, it is convergent. Again, as it is of positive terms,
2
it converges absolutely.
∞ ∞
n! 1
The original series is the sum of the series and − n , so it is convergent and absolutely
n=0
(2n)! n=0
2
convergent (consider the inequality: |a − b| ≤ |a| + |b|, ∀a, b ∈ R).

∞
n (2n)!
k) The series n (n!)2
is of positive terms. Let an be the general term of the series,
n=1
4

(n + 1)(2n + 2)!
 2
an+1 4n+1 (n + 1)! (n + 1)(2n + 2)! 4n (n!)2
lim = lim = lim  2
an n(2n)! n(2n)! 4n+1 (n + 1)!
4n (n!)2
. (n + 1)(2n + 2)(2n + 1) 2(n + 1)2 (2n + 1)
= lim 2
= lim
4 n(n + 1) 4 n(n + 1)2

2n + 1
= lim = 1,
2n
 
2n + 1
and therefore, D’Alembert’s test is inconclusive. The sequence tends to 1 by values greater than
2n
1, from which we conclude, by the Ratio Test, that the series is divergent.
146 Numerical Series

Note: We can also, alternatively, study the series using Raabe’s test (a common procedure when D’Alembert’s
test is inconclusive). In this case,
   
an 2n −n 1
. lim n − 1 = lim n − 1 = lim =− .
an+1 2n + 1 2n + 1 2

As this value is less than 1, the series is divergent, as we already saw.

∞
 1
l) The modulus of the general term of the given series is bounded by the general term of the series ,
n=2
n3 log(n)
since ! !
! sin(n) ! 1
. !! !≤ , ∀n ∈ N \ {1}.
n3 log(n) ! n3 log(n)
∞
 1
As the series is convergent as it is a p-series with p = 3 and
n=1
n3

1
n3 log(n) 1
. lim = lim = 0,
1 log(n)
n3

∞

1
Corollary 3 of the General Comparison Test allows us to conclude that the series is convergent.
n3 log(n) n=2
Consequently, the series of modules is convergent, making the given series absolutely convergent.

∞
  n
1
m) The general term of the series (−1)n 1− is not a null sequence. In fact,
n=1
2n

 n  2n 1
2
1 1
. lim 1− = lim 1− = e−1/2 ,
2n 2n
 n
1
which implies that the sequence of general term (−1)n 1− does not have a limit. From Theo-
2n
rem 2.2.1, it follows that the series is divergent.

∞
   ∞
1 1 1
n) The series · arcsin is of positive terms. Consider the series 2
, which we know to be
n=1
2n + 5 n n=1
n
convergent as it is a p-series with p = 2. The limit
   
1 1 1 1
· arcsin arcsin
2n + 5 n 2n + 5 n 1
. lim = lim · =
1 1 1 2
n2 n n

∞
  
1 1
belongs to R+ . By Corollary 2 of the General Comparison Test, the series · arcsin is
n=1
2n + 5 n
convergent. As it is of positive terms, it is absolutely convergent.
2.7. Solved Exercises 147

∞  
1 1 1 π
11. a) The series sin n consists of positive terms because 0 < n < , ∀n ∈ N. Consider the series
n=1
n 2 2 2
∞
1 1
. This series is convergent since it is a geometric series with r = . To show that the first series is
n=1
2n 2
also convergent, we can use Corollary 3 of the General Comparison Test. Specifically, we can take the limit
   
1 1 1
sin n sin n
n 2 1 2
. lim = lim · = 0 × 1 = 0.
1 n 1
2n 2n

Since the limit value equals zero, we can conclude that the series under consideration is convergent. More-
over, as it is of positive terms, it is absolutely convergent.
∞ √ ∞
 3
n2  1
b) The series √ 2
is of positive terms. Consider the series 4/3
, which is convergent as it is
n=1
n n + 2n n=1
n
4
a p-series with p = . The limit
3
√3
n2
√
n n + 2n2 n2 1
. lim = lim √ =
1 n n + 2n2 2
n4/3

is finite √
and different from zero; therefore, by Corollary 2 of the General Comparison Test, the series
∞ 3
n2
√ is convergent. As it is a series of positive terms, it is absolutely convergent.
n=1
n n + 2n2

∞
 cos(πn) 1
c) The series √ is alternating because cos(nπ) = (−1)n and √ > 0. We will study the
n=2 n2 − 1 n2 − 1
∞
 1
series of modules, √ , by comparing it with the harmonic series. The limit
n 2−1
n=2

1 √
√
n2 − 1 n2
. lim = lim √ =1
1 n −1
2

is finite and different from zero; therefore, by Corollary 2 of the General Comparison Test, the series

+∞
1
√ is divergent.
n 2−1
n=2
Let us apply Leibniz’s test to study the alternating series:
1
(i) lim √ = 0.
n2 − 1
1
(ii) √ > 0, ∀n ∈ N \ {1}.
n2 − 1
 
√ 1
(iii) The sequence ( n2 − 1) is increasing so √ is decreasing for n ≥ 2.
n2 − 1
148 Numerical Series

The conditions for applying the test are satisfied; therefore, the alternating series is convergent. As the
series of modules is divergent, the given series is conditionally convergent.

∞
 (−1)n 1
d) The series √ is alternating because an = √ > 0, ∀n ≥ 2. Let us start by studying
n=2
n log(n) n log(n)
the series of modules
∞ !

!

! (−1)n ! +∞ 1
!√ ! √
. ! n log(n) ! = n log(n)
.
n=2 n=2

Given the limit

1
√
n log(n) n3/4 n1/4
. lim = lim √ = lim = +∞,
1 n log(n) log(n)
n 3/4

∞
 1 3
and the fact that the series 3/4
is divergent because it is a p-series with p = , by Corollary 4 of the
n=2
n 4
General Comparison Test, the series of modules is divergent. Consequently, if convergent, the original series
will be conditionally convergent. Since the series is alternating, let us check if we are in the conditions of
Leibniz’s test:
1
(i) √ > 0.
n log(n)
1
(ii) lim √ = 0.
n log(n)
 
√ 1
(iii) Since ( n log(n)) is clearly increasing, the sequence √ is decreasing.
n log(n) n∈N\{1}
As we are in the conditions of Leibniz’s test, we can conclude that the alternating series is convergent. From
the study on the series of modules, we can affirm that the series is conditionally convergent.

∞
 4 + sin(n) 3 4 + sin(n)
e) The series √
3
is of positive terms. We can observe that 0 < √ 3
≤ √3
, ∀n ∈ N.
n=1
n + 1 n + 1 n+1
∞
1 1
As the series √
3
is a p-series with p = , it is divergent. Since the limit
n=1
n 3

3
√ √
3
n+1 33n
. lim = lim √
3
=3
1 n+1
√
3
n

is finite and different from zero, Corollary 2 of the General Comparison Test allows us to conclude that the
∞ ∞
3 4 + sin(n)
series √3
is divergent. Hence, the General Comparison Test ensures that the series √
3
n=1
n + 1 n=1
n+1
is divergent.

∞
n2n sin(n3 )
f) We begin by analyzing the series of modules, as the series is not of positive terms. The
n=1
(3n2 + 5)n
inequality ! 2n !
! n sin(n3 ) ! n2n
.0 ≤ !! 2 n
!≤
!
(3n + 5) (3n + 5)n
2
2.7. Solved Exercises 149

holds true in N.
We evaluate the limit 
n n2n n2 1
. lim = lim = ,
(3n2 + 5)n 3n2 + 5 3
∞

n2n
which is less than 1. Consequently, by Cauchy’s Root Test, the series 2 + 5)n
is convergent. By
n=1
(3n
∞ ! !
 ! n2n sin(n3 ) !
the General Comparison Test, the same happens to the series ! !
! (3n2 + 5)n !. The original series is
n=1
absolutely convergent since the series of modules is convergent, considering Definition 2.4.1.
 
n−1 n−1
g) Let an = arcsin 2
. Since 0 < 2 < 1, ∀n > 2, we have that an > 0, and therefore, the series
n +1 n +1
∞
 
+∞
(−1)n an is alternating. Let us start by studying the series of modules an . The limit
n=3 n=3

 
n−1
arcsin
n2 + 1
. lim =1
n−1
n2 + 1

 ∞
n−1
belongs to R+ , and the series 2+1
is divergent (compare it with the harmonic series). We can then
n=3
n
conclude that the series of modules is divergent. Consequently, if convergent, the series under study is
conditionally convergent. Since the series is alternating, let us check if we are in the conditions of Leibniz’s
test:
 
n−1
(i) arcsin > 0, ∀n > 2.
n2 + 1
 
n−1
(ii) lim arcsin 2
= 0.
n +1
 
x−1
(iii) Let f be the real function of a real variable defined by f (x) = arcsin . This function is
x2 + 1
decreasing on [3, +∞[, because

 −x2 + 2x + 1
.f (x) = √ < 0, ∀x ∈ [3, +∞[.
(x2 + 1) x4 + x2 + 2x

Therefore, the sequence   

n−1
. arcsin
n2 + 1 n∈N\{1,2}

is decreasing.
As we are in the conditions of Leibniz’s test, we can conclude that the alternating series is convergent. The
study done on the series of modules shows that the series is conditionally convergent.

∞
4 × 7 × · · · × (3n + 1)
h) The series is of positive terms. Let us use D’Alembert’s test to study it. If an
n=1
8 × 11 × · · · × (3n + 5)
is the general term of the series, then:
150 Numerical Series

4 × 7 × · · · × (3n + 1)(3n + 4)
an+1 8 × 11 × · · · × (3n + 5)(3n + 8) 3n + 4
. lim = lim = lim = 1.
an 4 × 7 × · · · × (3n + 1) 3n + 8
8 × 11 × · · · × (3n + 5)

Since this limit is 1, we cannot draw any conclusions using D’Alembert’s test. Therefore, let us use Raabe’s
test, which is a common procedure when D’Alembert’s test is inconclusive. In this case,
   
an 3n + 8 4n 4
. lim n − 1 = lim n − 1 = lim = .
an+1 3n + 4 3n + 4 3

As this value is greater than 1, we can conclude that the series is convergent. Moreover, since it is a series
of positive terms, it converges absolutely.

∞ ∞
arctan(n3 ) 2n (2n)!
i) Consider the two series √ 2
and n (2n + 1)!
. The first one is of positive terms, and we
n=1
n + n n=0
3
∞
1
compare it with the p-series 2
, which we know to be convergent. Indeed, the limit
n=1
n

arctan(n3 )
√
n + n2 n2 π
. lim = lim √ arctan(n3 ) =
1 n + n2 2
n2

∞
arctan(n3 )
is finite and different from zero, so, by Corollary 2 of the General Comparison Test, the series √
n=1
n + n2
is convergent.
The second series is also of positive terms. We can study it using D’Alembert’s test. Let an be the general
term of the series:

2n+1 (2n + 2)!

an+1 3n+1 (2n + 3)! 2n+1 (2n + 2)! 3n (2n + 1)! 2(2n + 1) 2
. lim = lim n = lim = lim = .
an 2 (2n)! 2n (2n)! 3n+1 (2n + 3)! 3(2n + 3) 3
3n (2n + 1)!

This limit is less than 1; therefore, by D’Alembert’s test, the second series is convergent.
The original series is the sum of the two series studied, so it is convergent. Since it is of positive terms, the
series is absolutely convergent.

∞ sin
 √1
n 1 π
j) The series √ consists of positive terms, since 0 < √ < , ∀n ∈ N. To check its convergence,
n=1
n+ n n 2
∞
1
let us compare it with the series , which we know to be convergent because it is a p-series with
n=1
n3/2
2.7. Solved Exercises 151

3
p= . Taking the limit, we have
2

sin √1 sin √1
n n
√ √ sin √1n
n+ n n+ n n
. lim = lim = lim · √ = 1.
1 1 1 n+ n
√ √
n3/2 n n n

Since the limit is finite and different from zero, by Corollary 2 of the General Comparison Test, the series
∞ sin √1
n
√ is convergent. Moreover, as a series of positive terms, it converges absolutely.
n=1
n + n

∞     
1 1
k) The series n sin − (n + 2) sin is telescopic. We can write the sequence of partial
n=1
n n+2
sums:
 
1
S1 = sin(1) − 3 sin
3
     
1 1 1
S2 = sin(1) − 3 sin +2 sin −4 sin
3 2 4
         
1 1 1 1 1
S3 = sin(1) − 3 sin +2 sin −4 sin + 3 sin − 5 sin
3 2 4 3 5
     
1 1 1
= sin(1) + 2 sin −4 sin −5 sin
2 4 5
.          
1 1 1 1 1
S4 = sin(1) + 2 sin −4 sin −5 sin + 4 sin − 6 sin
2 4 5 4 6
     
1 1 1
= sin(1) + 2 sin −5 sin −6 sin
2 5 6
.
..
     
1 1 1
Sn = sin(1) + 2 sin − (n + 1) sin − (n + 2) sin
2 n+1 n+2
.
..

As  
1
  sin
1 n
. lim n sin = lim =1
n 1
n
we have  
1
. lim Sn = sin(1) + 2 sin − 2.
2
The series is convergent and

∞       
1 1 1
. n sin − (n + 2) sin = sin(1) + 2 sin − 2.
n=1
n n + 2 2
152 Numerical Series

To determine whether it is absolutely convergent,  we will

 study the sign of the general  term.
 For this, we
1 1
will prove that the sequence of general term n sin is increasing. Let f (x) = x sin , x ≥ 1. We
n x
have    
 1 1 1
.f (x) = sin − cos
x x x
   
1 1 1 1
and, therefore, f  (x) = 0 ⇔ tan = . However, this equation has no zeros. In fact, tan > ,
x x   x x
1
∀x ≥ 1. Then f  (x) > 0, ∀x ≥ 1, which implies that f is increasing. As f (n) = n sin , we can conclude
n
∞ ∞

that the sequence is increasing and, therefore, the series is of negative terms. Since |an | = − an ,
n=1 n=1
the given series is absolutely convergent.

∞
2n + 3
l) The series is of positive terms. We can establish the following inequality:
n=1
(n + 1)!

2n + 3 2n
. ≤2 , ∀n ≥ 2.
(n + 1)! (n + 1)!

∞
 2n
To determine the convergence of the series , we can apply D’Alembert’s test. Let an be the
n=1
(n + 1)!
general term of the series:
2n+1
an+1 (n + 2)! 2
. lim = lim = lim = 0.
an 2n n+2
(n + 1)!
This limit is less than 1, indicating that the series is convergent according to D’Alembert’s test. Using the
General Comparison Test, we can affirm that the series under study is convergent and, hence, absolutely
convergent.

∞
 arctan(n + 1) − arctan(n)
m) The series is of positive terms because arctan(x) is an increasing function
n=1
n2
∞
1
on R. Let us compare this series with the series , which we know is convergent as it is a p-series
n=1
n2
with p = 2. By Corollary 3 of the General Comparison Test, given the value of the limit

arctan(n + 1) − arctan(n)
n2
. lim = lim (arctan(n + 1) − arctan(n)) = 0,
1
n2

we can conclude, as we are comparing with a convergent series, that the series under study is convergent.
As it is a series of positive terms, it converges absolutely.

∞
 log(n)
n) The general term of the series  1  is not a null sequence. In fact,
n=1
n sin n
2.7. Solved Exercises 153

1
log(n) n
. lim   = lim log(n) ·   = +∞.
1 1
n sin sin
n n
Therefore, by Theorem 2.2.1, the series diverges.

∞
  
 n n+1 n+1
o) Consider the series log(2) log . It consists of positive terms because > 1, ∀n ∈ N.
n=1
n n
∞
  n
Let us analyze the series log(2) . It converges because it is a geometric series with 0 < r = log(2) < 1.
n=1
The limit  
 n n+1
log(2) log  
n n+1
. lim  n = lim log =0
log(2) n
allows us to use Corollary 3 of the General Comparison Test to conclude that the series under study is
convergent because we are comparing it with a convergent series. As the series is of positive terms, it is
absolutely convergent.
    p
p) Given that n ≥ 3 implies log(n) > 1, then log log(n) > 0 so n log(n) log log(n) > 0, ∀p ∈ R.
Therefore, the series is of positive terms.
 
Let us assume that p ≤ 0. If n > ee (for example, if n ≥ 33 = 27), then log log(n) > 1. Thus,
  −p
log log(n) ≥ 1, with equality holding for p = 0. Consequently, we can write:
  −p
1 log log(n) 1
.  p = ≥ .
n log(n) log log(n) n log(n) n log(n)


+∞
1
Since the series is divergent as seen in Example 2.5.2, we can apply the General Com-
n=27
n log(n)

+∞
1
parison Test to deduce that the series   p is divergent. Therefore, the series
n=27
n log(n) log log(n)

+∞
1
  p is also divergent.
n=3
n log(n) log log(n)

Now, suppose that p > 0. With the application of the Integral Test in mind, let us consider the function
1
.f (x) =   p
x log(x) log log(x)
  p
and let g(x) = x log(x) log log(x) . We know that g(x) > 0, ∀x ∈ [3, +∞[. Additionally, since the
logarithm function is continuous and increasing, g is continuous and increasing on [3, +∞[, so f is positive,
+∞
continuous, and decreasing on [3, +∞[. We can now study the improper integral f (x) dx.
3
If p = 1,
# $y
y 1 1   −p+1
  p dx = log log(x)
3 x log(x) log log(x) −p + 1 3
.
1   −p+1   −p+1
= log log(y) − log log(3) .
−p + 1
154 Numerical Series

If p = 1,
y 1     y      
.   dx = log log log(x) = log log log(y) − log log log(3) .
3 x log(x) log log(x) 3

We can then determine that

⎧
⎪
⎪ if 0 < p ≤ 1
y 1 ⎨+∞,
lim   p =
1   −p+1
.
y→+∞ x log(x) log log(x) ⎪
⎪
3 ⎩ log log(3) if p > 1.
p−1
+∞
This means that the improper integral f (x) dx is convergent if p > 1 and divergent if 0 < p ≤ 1. By
3
the Integral Test, the series under consideration is convergent if p > 1 and divergent if p ≤ 1.

+∞
1
Conclusion: The series   p is convergent if and only if p > 1.
n=3
n log(n) log log(n)

∞
(n + 1)n
12. a) The series is of positive terms. To analyze this series, we will use D’Alembert’s test. Let an
n=1
3n n!
be the general term of the series:

(n + 2)n+1
 
an+1 3n+1 (n + 1)! 1 n+2 n+2 n e
. lim = lim n = lim = .
an (n + 1) 3 n+1 n+1 3
3n n!
Since this limit is less than 1, according to D’Alembert’s test, the series converges.
∞
(n + 1)n
b) In the previous item, we proved that the series is convergent, which implies, by Theorem 2.2.1,
n=1
3n n!
(n + 1)n
that its general term is a null sequence. Therefore, lim = 0.
3n n!

∞

13. We will study the convergence of the series of modules, |an |. We will use the Root Test or one of its corollaries,
n=1
to do this. We have
⎧
⎪  n2  n
⎪
⎪ n n
⎪
⎪
n
= , if n = 3k, k ∈ N
⎪
⎪ n+1 n+1
⎪
⎪
⎨ 1
⎪
1
.
n
|an | = n
= √ , if n = 3k + 1, k ∈ N0
⎪ n!
⎪
n
n!
⎪
⎪ 
⎪
⎪ !! !
⎪
⎪ (−1)n !! 1
⎪
⎪ n !
if n = 3k + 2, k ∈ N0 .
⎩ ! (n + 1)n ! = n + 1 ,

1 1
The sublimits of n
|an | are
and 0; therefore, lim n |an | = . Since this value is less than 1, we can affirm, by
e e
∞
 ∞

Corollary 1 of the Root Test, that the series |an | is convergent. Consequently, the series an is absolutely
n=1 n=1
convergent.
2.7. Solved Exercises 155

∞

14. One initial approach is to analyze the general term of the series an . Since lim a2n = lim a2n+1 = +∞, then
n=1
lim an = +∞, and by Theorem 2.2.1, the series diverges.
Another approach: ⎧
⎪
⎪
n+2
an+1 ⎨ , if n is even
n
. =
an ⎪
⎪ n+1
⎩ , if n is odd, n > 1.
n−1
an+1
We find that > 1, ∀n ∈ N \ {1}, and therefore by the Ratio Test, the series diverges.
an
∞

15. If (an ) is a sequence of positive terms such that the series nan is convergent, then the sequence (an ) is
n=1
bounded. We can prove this by reductio ad absurdum: If (an ) was
not bounded, then lim (nan ) would not be
∞

zero, and in this case, the series nan would not be convergent. It follows from the fact that (an ) is bounded
n=1
that
a2n an
. lim = lim = 0.
nan n
∞

By Corollary 3 of the General Comparison Test, the series a2n is convergent.
n=1

∞

16. As the series an converges, lim an = 0. By hypothesis, an > 0, ∀n ∈ N, so log(1 + an ) > 0. Thus, we have
n=1
two series of positive terms. The limit
log(1 + an )
. lim =1
an
∞

is finite and different from zero; therefore, by Corollary 2 of the General Comparison Test, the series log(1+an )
n=1
also converges.

log(1 + x) 0
Note: The limit lim is an indeterminate form of type . Applying L’Hôpital’s Rule, we get
x→0 x 0
 
log(1 + x) 1
. lim = lim =1
x→0 (x) x→0 1+x

log(1 + x) log(1 + an )
therefore, lim = 1, which implies that lim = 1.
x→0 x an
∞

17. a) The series (−1)n−1 an is alternating. Leibniz’s test cannot be applied because the sequence (an ) does
n=1
not decrease. In fact, if n is even,

1 1 n2 − n − 1
.an+1 − an = − 2 = 2 > 0.
n+1 n n (n + 1)
156 Numerical Series

∞

b) Let us prove that the series (−1)n−1 an is divergent. If it was convergent, then, by Theorem 2.2.4, any
n=1
∞

series obtained from it by association of its terms would be convergent. Consider the series bn , where
n=1
(bn ) is the sequence defined by

1 1 (2n)2 − 2n + 1 4n2 − 2n + 1
.bn = a2n−1 − a2n = − = = .
2n − 1 (2n)2 (2n)2 (2n − 1) 4n2 (2n − 1)
The limit
4n2 − 2n + 1
4n2 (2n − 1) 4n2 − 2n + 1 1
. lim = lim =
1 4n(2n − 1) 2
n
is finite and different from zero; therefore, based on Corollary 2 of the General Comparison Test, the series
∞ ∞
bn diverges because the harmonic series is divergent. However, we obtained the series bn by
n=1 n=1
association of the terms of the original series. Hence, the original series must be divergent.
∞

18. a) Suppose that the series bn is convergent. Then, its general term is a null sequence, which implies
n=1
that all its subsequences have limit zero. Therefore, we have lim b2n−1 = 0, or equivalently, lim an = 0.
∞

Conversely, let lim an = 0 and (Sn ) be the sequence of partial sums of the series bn . We can see that
n=1
S2n = 0 and S2n−1 = an for all n ∈ N. Since lim an = 0, it follows that lim Sn = 0. By definition, if the
sequence of partial sums of a series has a finite limit, the series converges. As a result, it can be inferred
∞

that the series bn converges.
n=1
∞
 ∞

∗ denote the general terms of the sequences of partial sums of the series
b) Let Sn and Sn |bn | and |an |,
n=1 n=1
respectively. We can observe that

S1 = |b1 | = |a1 | = S1∗

S2 = |b1 | + |b2 | = |a1 | + |a1 | = 2S1∗
S3 = |b1 | + |b2 | + |b3 | = |a1 | + |a1 | + |a2 | = S1∗ + S2∗
. .
S4 = |b1 | + |b2 | + |b3 | + |b4 | = |a1 | + |a1 | + |a2 | + |a2 | = 2S2∗
S5 = |b1 | + |b2 | + |b3 | + |b4 | + |b5 | = |a1 | + |a1 | + |a2 | + |a2 | + |a3 | = S2∗ + S3∗
..
.

So, we can infer that

∗ ∗ ∗
.S2n = 2Sn and S2n+1 = Sn + Sn+1 .
∞

If the series bn converges absolutely, then the limits lim S2n and lim S2n+1 exist, are finite, and equal.
n=1
∞

∗ =
Since lim Sn 1
lim S2n , we can conclude that the series an converges absolutely.
2
n=1
2.7. Solved Exercises 157

∞

Reciprocally, if the series ∗ exists and is finite. This implies that
an converges absolutely, then lim Sn
n=1
∞

∗ + S∗
lim S2n+1 = lim(Sn ∗
n+1 ) = lim 2Sn = lim S2n , and we can state that the series bn converges
n=1
absolutely.

∞

19. If the series an is convergent, then the sequence of its partial sums, (Sn ), is also convergent. This implies
n=1
that all its subsequences are convergent and have the same limit. Since the sequence (an ) is decreasing, we have

.S2n − Sn = a2n + · · · + an+1 ≥ a2n + · · · + a2n = n a2n ,

and as lim (S2n − Sn ) = 0 and an ≥ 0, we can conclude that lim n a2n = 0, which implies that lim 2 n a2n = 0.
Similarly, we have

.S2n+1 − Sn+1 = a2n+1 + · · · + an+2 ≥ a2n+1 + · · · + a2n+1 = n a2n+1 .

As lim (S2n+1 − Sn+1 ) = 0 and (an ) is a sequence of nonnegative terms, we conclude that lim n a2n+1 = 0.
This also implies that lim (2n + 1) a2n+1 = 0 because lim an = 0 since, by hypothesis, the series is convergent.
We have shown that the subsequence of even order terms of (n an ) has the same limit as its subsequence of odd
order terms, which is zero. Hence lim n an = 0.

a2 + b 2
20. We can use the inequality ab ≤ , ∀ a, b ∈ R, along with the fact that the sequence (an ) is of positive
2
terms, to establish the following expression:

1  
1 √ n2p
+ an 1 1
.0 ≤ an ≤ = + an .
np 2 2 n2p

∞
 ∞
1 1
By hypothesis, the series an is convergent, and if p >
, then 2p > 1, which implies that the p-series 2p
n=1
2 n=1
n
is convergent. The sum of these two series is convergent, and given the previous inequality, we can state by the
∞ √
an
General Comparison Test the convergence of the series .
n=1
np

∞
 1 1
21. Consider the series , which is divergent as it is the harmonic series. Let bn = . Then
n=2
n−1 n−1

1
bn+1 n n−1 1 an+1
. = = =1− ≤ , ∀n ≥ 2.
bn 1 n n an
n−1

∞

By Corollary 5 of the General Comparison Test, we can determine that the series an is divergent.
n=1
158 Numerical Series

g
22. The functions g and g  are, by hypothesis, continuous and positive on [p, +∞[, so is continuous and positive.
g
    2
g (x) g  (x)g(x) − g  (x) g
In addition, =  2 < 0, so is decreasing on [p, +∞[. We are in the conditions of
g(x) g(x) g
the Integral’s Test. But
y g  (x)    
. dx = log g(y) − log g(p) ;
p g(x)
g  (x)
+∞  
therefore, the improper integral dx is convergent if and only if there exists and is finite lim log g(x) .
p g(x)  
x→+∞
By the continuity of the logarithm function, lim log g(x) exists finite if and only if there exists finite and dif-
x→+∞
ferent from zero the lim g(x). But, by hypothesis, g is positive and increasing (g  is positive), so it cannot
x→+∞
have a limit 0 when x tends to +∞.

+∞
g  (n)
Conclusion: The series is convergent if and only if lim g(x) exists and is finite.
n=p
g(n) x→+∞

23. We can prove by induction that (vn ) is a sequence of positive terms. Applying D’Alembert’s test, we find that
∞

the series vn is convergent because
n=2
 
vn+1 π
. lim = lim sin = 0 < 1.
vn n

Furthermore,
arctan(vn )
. lim = 1,
vn
∞

which implies by Corollary 2 of the General Comparison Test that the series arctan(vn ) is convergent.
n=2

24. Let us look at the numbers of the construction of the Sierpinski triangle in a table.

Number Number of squares Length Area of each Area of the squares

of squares removed of the edges square removed

1 0 1 1 0

 2  2
1 1 1
3 1 1×
2 2 2

 2  2
1 1 1
32 3 3×
22 22 22

 2  2
1 1 1
33 32 32 ×
23 23 23
. . . . .
. . . . .
. . . . .
2.7. Solved Exercises 159

So the total area of the squares removed is the sum of the geometric series
∞
  2 ∞  
1 1  3 n
.A = 3n = = 1.
n=0
2n+1 4 n=0 4

Since the initial square has area 1, we conclude that the area of the Sierpinski triangle is zero.

∞

25. The sum of the areas of the circles is given by the series πr12 + 2
3πrn , where r1 is the radius of the largest
n=2
circle, r2 is the radius of the circles tangent to it, etc.

π
We need to find the expression of rn as a function of r1 . In the figure, the angle ∠BAC measures radians and
3
π r1 1 r1 1 r2
cos = , and therefore, = . Similarly, = . Then |AB| = 2r1 and |DB| = 2r2 . As a
3 |AB| 2 |AB| 2 |DB|
1
result, we have the equality 2r1 = r1 + r2 + 2r2 , from which we conclude that r2 = r1 . Similarly, we can show
 n−1 3
1 1 1
that r3 = r2 = r1 and, more generally, rn = r1 , n ≥ 2.
3 9 3
The series now takes the form
∞ ∞  2  ∞
  1  1 11 2
2 2
.πr1 + 3πrn = πr12 + 3π n−1
r 2
1 = πr 2
1 1+3 n−1
= πr1 .
n=2 n=2
3 n=2
9 8

1 π |BC| 1
What is the value of r1 ? Considering that |BC| = , |AB| = 2r1 , and sin = , we have r1 = √ .
2 3 |AB| 2 3
Finally, the sum of the areas of the circles is
∞

2 2 11π
.πr1 + 3πrn = .
n=2
96

∞

26. The area of the colored region can be written as A = (An − Bn ), where An is the area of the circle and Bn
n=1
is the area of the square inscribed in that circle. Let rn be the radius of the nth circle and ln be the length of the
side of the inscribed square.
160 Numerical Series

2 + l2 = (2r )2 , that is, l2 = 2r 2 . We can write the series in the

From Pythagoras’s Theorem, it follows that ln n n n n
form
∞
 ∞

.A =
2
(πrn − ln
2
)= (π − 2) rn2
.
n=1 n=1

What is the relationship between r1 and rn ? We know that 2r2 = l1 , so 4r22 = l12 . As l12 = 2r12 , we have that
1 1
r22 = r12 . By induction, it is easy to show that rn
2
= n−1 r12 . Thus,
2 2
∞
 ∞
 1
.A = (π − 2) rn
2
= (π − 2) r 2 = 2(π − 2) r12 .
n=1 n=1
2n−1 1

As r1 = 1, we conclude that A = 2(π − 2).

2.8. Proposed Exercises 161

2.8 Proposed Exercises

∞

1. Find the general term and the sum of each of the c) (−1)n x2n+1
following series: n=0

1 1 1 1 ∞

a) + + + + ··· 5. Show that if an = A ∈ R, then
3 8 15 24
n=0
1 1 1
b) + + + ···
1×2×3 2×3×4 3×4×5 ∞

1 1 1 . (an−1 + an + an+1 ) = 3A − a1 − 2a0 .
c) + + +· · · n=1
1×2×3×4 2×3×4×5 3×4×5×6
6. Test for convergence or divergence the following se-
2. Find the sum of the series: ries. If convergence occurs, indicate whether it is
∞
 conditional or absolute:
2n + 1
a) (−1)n ∞
n(n + 1)  (−1)n+1
n=1
a)
∞ n
1 a n=1
b) tan n , a ∈ R \ {kπ, k ∈ Z}. ∞
2n 2  (−1)n−1
n=1 b)
x x n2
Hint: tan = cot − 2 cot(x) n=1
2 2 ∞
∞ √ √  n2
 n+1− n c) (−1)n
c) √ 1 + n2
2
n +n n=1
n=1
∞ ∞
 (−1)n − 8 (−1)n
d) 7. Consider the series :
3n n=1
(n + 1)2
n=1
∞
   n+1
 1 n 1 a) Verify that it is convergent.
e) 1− − 1−
n=1
n n+1 b) Calculate its sum with an error less than
1
∞
 .
 
f) 2−n − 2−3n 10000
n=1
∞ 8. Let an and bn be two convergent series, cn
 arctan(n) + arctan(−n + 2)
g) and dn be two divergent series, and α = 0 be a
n=3
arctan(n) arctan(n − 2) real number. What can be said about the conver-
∞ gence of the following series?
  
3. Let an be a convergent series. Show that the a) (an + bn ) e) (an cn )
n=1
∞
 a3 + 5n  
series √n is divergent. b) (an bn ) f) (αcn )
n=1 n2 + 1
 
4. Find the values of x for which the following series c) (αan ) g) (cn + dn )
converge and, when possible, evaluate the sum of  
the series: d) (an + cn ) h) (cn dn )
∞
 8n 9. Test the convergence of the following series using a
a)
(x + 1)3n comparison test:
n=0
∞
 ∞

 n 1
b) |x| − 1 a)
n=1 n=1
n3 + 3
162 Numerical Series

∞  ∞
 n+1  n!
b) b)
n=1
n2 + 1 n=1
3n

∞
 ∞
1 n3
c) c)
n=1 n(n2 + 1) n=1
n!

∞ √ ∞
n n nn
d) √ d)
3 n3 + 1 (2n)!
n=1 (n + 1) n=1
∞ ∞
   n!
e) n− n2 − 1 e) 2
n=1
n
n=1
∞ ∞

 π 2 n n+1
f) sin f) 2−n+(−1) 3−n+(−1)
n=1
n n=1

∞ √ ∞
 n!
n+ n g)
g)
n2 − n n=1
3 × 5 × 7 × · · · × (2n − 1)
n=2
∞ ∞

 n2 + 1 2 × 4 × 6 × · · · × (2n)
h) h)
n3 + 1 n=1
2 × 5 × 8 × · · · × (3n − 1)
n=1

12. Use the Root Test or Cauchy’s Root Test to deter-

mine the convergence or divergence of the following
10. Using the Cauchy Condensation Test or the Integral
series:
Test, determine the convergence of the following se-
ries: ∞
 1
a)
∞
 (n + 1)n
1 n=1
a)  2 ∞
n log(n)  nn/2
n=2
b)  n
∞ log(n)
log(3n) n=2
b) ∞
n2  kn
n=2
c) ,k∈R
∞
 n=1
n!
2
c) n e−n ∞  

2
n=2 n+1 n
d)
∞ n
arctan(n) n=1
d) ∞
n2 + 1  n
n=2
e) 2−2n+(−1) n
∞
 1 n=1
e) √ ∞  n
n+1−1  1 1
n=1
√
f) − 1
∞  
 n n=1
nn n
1
f) ∞
2  π n
n=1 g) sin
n=1
n

11. Use the Ratio Test or By the continuity of the log-

arithm function to determine the convergence or di- 13. Test the convergence of the following series using
vergence of the following series: Raabe’s test:

∞ ∞

n n
a) n
a)
n=1
3 n=1
(2n + 1)!
2.8. Proposed Exercises 163

∞
 ∞

1
b) n) ne log(n)
n=1
n(n + 1)(n + 2) n=2
∞ √ ∞
 n−1 2n
c) o)
n n=0
(2n + 1)!
n=1
∞
∞
 2 cos(nθ)
(2n + 1)! p) ,θ∈R
d) n5/2
n=1
n × n! n=1
∞
3n n2n
q)
14. Investigate the convergence or divergence of the n=0
(2n)!
following series. If convergence occurs, indicate ∞
log(n!) + n!
whether it is conditional or absolute: r)
n=1
nn + 2n
∞
 ∞ √
1  3
n2 + 1
a) √ s) (−1)n
n=0 n2 + 1 n+3
n=0
∞
 1
b)
n=2 n(n2 − 1) 15. Examine the convergence or divergence of the follow-
∞ ing series. If convergence occurs, indicate whether
en n n
c) it is conditional or absolute:
n=1
n!
∞

∞
sin(n + 1)
 2 a) 2 log(n + 1)
d) ,p∈R n=1
n
n2 + p 2
n=1 ∞  
1 (n!)2
∞
 b) +
cos(n π2 ) 3 n (2n)!
e) n=3
n2 ∞
n=1  1
∞ c)
 52n (n + 1)(n + 2)
f) n=1
(n + 1)! ∞
n=0 e
d) cos(nπ) tan
∞
 n
π n n=3
g) tan ∞
n=3
n  1 + (−1)n n
e)
∞ n2 + 5
1 nπ n=1
h) cos ∞  3n−1
n=1
n 2 2n
f)
∞ 4n + 1
 1 n=1
i) ∞  
1)
(1+ n n! n 1
n=1 n g) 2 +
nn n2 + n
∞ n=1
n2 + 2n + 1 ∞
j)  (n + p)!
n=1
3n2 + 2 h) , p, q ∈ N
n!(n + q)!
∞   n n=1
 sin 3π ∞
k) 2  n!
n2 + 1 i)
n=1
n=1
(π + 1)(π + 2) . . . (π + n)
∞
 1 ∞  
l) (−1)n 1
nn log(n) j) +√
n=2 n=1
n(n + 1) 3n
∞
 ∞

1 2n + 3
m) 
1
 k) (−1)n
1+ log(n) (n + 1)(n + 2)
n=2 n n=0
164 Numerical Series

∞ 
 n
a 19. Knowing that an is convergent and of positive
l) , a ∈ R \ {−3}
n=0
a+3 terms and bn > 0, ∀n ∈ N, what can be said about
an
∞ the convergence of the series ?
1 1 + bn
m) n+a
, a ∈ R+
0
n=0
2
∞ !  ! 20. Knowing that an and bn are convergent se-
! 1 !!
n) log !!log ! ries of positive terms, study the convergence of the
n=2
n!
following series:
∞
 (3n)!
o)  1 1

n=0
27n (n!)2 a) +
an bn
1 1 1×3 1 1×3×5 1  n+1
p) 1 + · + · + · + ···
2 3 2×4 5 2×4×6 7 b) an
∞ √ n
 2n − 1 log(4n + 1)
q)
n=1
n(n + 1)
∞
21. Let an be a divergent series, an ≥ 0, and sn be
 3 × 5 × 7 × · · · × (2n + 1)
r) the sum of its first n terms. Show that the series
2n (2n − 1) n!  √
n=1 √ 
∞
 . sn+1 − sn
(−3)n
s)
n=1
3 × 5 × 7 × · · · × (2n + 1) is divergent.
∞
2 + sin3 (n + 1) 0a np +···+a
p
t) 22. Prove that the series b0 nq +···+bq
in which a0 ,
n=1
2n + n2
. . . , ap , b0 , . . . , bq are real numbers and a0 > 0,
∞
1 b0 > 0, is convergent if and only if q − p > 1.
u)
n=1
2n − 1 + sin2 (n3 )
23. Study the conditional and absolute convergence of
the series:
16. Indicate for which values of α the following series ∞
 an
are conditionally or absolutely convergent: a) ,a>0
n=1
(1 + a)(1 + a2 ) . . . (1 + an )
∞
  n ∞
a) 1 + sin(α)  (α + 1)(α + 2) . . . (α + n)
b) , β ∈ R \ Z− :
n=1 (β + 1)(β + 2) . . . (β + n)
∞ n=1
 1
b) (−1)n i. If α ∈ R \ Z−
(n + 1)α
n=0 ii. If α ∈ Z−
17. a) Test for convergence the series
∞   n2 un+1 2 1
2n n 24. Let un > 0 and ≤ 1 − + 2 ∀n ∈ N \ {1}.
e . un n n
n=1
n+3 Show that un is convergent.
b) Based on the previous item, indicate the limit
   n2
n 25. Consider the series
of the sequence e2n .
n+3 ∞ ∞
(−1)n 1
. and √ :
18. Let an and bn be two convergent series of pos- n=0
n! n=0
(n + 1) n+1
√
itive terms. Show that the series an bn also
converges.
a) Calculate the partial sum of order three of
an + b n
Hint: Prove that ≥ an b n . Cauchy’s product of the two series.
2
b) Study the convergence of the product series.
2.8. Proposed Exercises 165

26. The figure shows a sequence of cubes constructed connect the midpoints of the sides of this square to
as follows: The edge of the larger cube has length 1, form another square inside it. Repeat this process
and the length of the edge of each of the following with the new square, connecting its midpoints to
cubes is half the length of the edge of the previous form another square. Keep repeating this process
cube. Find the sum of the volumes of the cubes. infinitely, and we will obtain infinite squares within
the first one.
Determine the sum of the areas of all the squares.

27. The figure shows a sequence of squares constructed

as follows: Start with a square of side length 4. Now,
 Series of Functions 3

It is possible to express various functions as a combination of “simpler”

functions, such as a series of monomials (known as Taylor series) or trigono-
metric functions (known as Fourier series). These function series are useful
not only in approximating other functions but also in other areas of mathe-
matics, such as differential equations.

3.1 Introduction: Sequences of Functions

In many Analysis issues, it is interesting to consider sequences of functions of

the form .f1 (x), .f2 (x), . . . , .fn (x), . . ., and naturally the question of passing
to the limit arises.

Definition 3.1.1 Let .(fn ) be a sequence of functions, .fn : D ⊂ R → R.

The sequence .(fn ) converges at a point .a ∈ D if the numerical sequence
 
. fn (a) is convergent.

If the sequence .(fn ) converges at all points of D, we can define a function

.f : D → R by

.f (x) = lim fn (x),

n→+∞

which is called the limit of .(fn ) on D. We also say that .(fn ) converges
pointwise to f on D.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 167
A. Alves de Sá, B. Louro, Sequences and Series,
https://doi.org/10.1007/978-3-031-67202-6 3
168 Series of Functions

 x n
Example 3.1.1 The sequence of functions . 1 + , defined on .R, con-
n
verges whatever .x ∈ R (see Fig. 3.1). The limit function is .f (x) = ex :
 x n
. lim 1 + = ex ∀x ∈ R.
n

Figure 3.1: The function .f (x) = ex and the first terms of the sequence

Example 3.1.2 Consider the functions .fn (x) = xn , .n ∈ N, on the interval

.[0, 1]. They are continuous, and the limit function exists, but it is not

continuous: ⎧
⎨0, if 0 ≤ x < 1
n
.f (x) = lim x =
⎩1, if x = 1

(see Fig. 3.2).

Figure 3.2: Sequence of continuous functions converging to a discontinuous function

3.1. Introduction: Sequences of Functions 169

Based on Example 3.1.2, we can observe that a sequence of continuous

functions may converge pointwise to a discontinuous function. We will
see later that even if all the integrals in question exist, the integral of the
limit function may differ from the limit of the integrals of the sequence’s
functions. This means that the limit does not always commute with the
integral. However, there is a stronger form of convergence known as uniform
convergence, which guarantees both the continuity of the limit of a sequence
of continuous functions and the ability to interchange limit and integration
symbols.

Definition 3.1.2 The sequence of functions .(fn ) converges uniformly

to f on .D ⊂ R (.D = ∅) if

∀δ > 0 ∃ p ∈ N : n > p ⇒ |fn (x) − f (x)| < δ, ∀x ∈ D.

.

This condition is equivalent to

∀δ > 0 ∃ p ∈ N : ∀n > p, sup |fn (x) − f (x)| < δ,

.
x∈D

that is,
. lim sup |fn (x) − f (x)| = 0.
n→+∞ x∈D

It is evident that if a sequence of functions .(fn ) converges uniformly to f on

D, then it also converges pointwise to f on D. The converse is not true. If
we take the sequence .(fn ) from Example 3.1.2, which converges pointwise
to f on .[0, 1], we have
. sup |fn (x) − f (x)| = 1, ∀n ∈ N,
x∈[0,1]

so .(fn ) does not converge uniformly to f on .[0, 1].

Note that if .(fn ) converges uniformly to f on D, then .(fn ) converges

uniformly to f on any subset of D.

The definition of uniform convergence means that, for every fixed .δ > 0,
there is an order from which the images of all functions are in the .δ neigh-
borhood of .f (x) for any .x ∈ D, that is, there is an order from which the
images of all functions are in the region of the plane defined by .f (x) − δ
and .f (x) + δ, in the set D. This is illustrated in Fig. 3.3.
170 Series of Functions

Figure 3.3: An illustration of uniform convergence

Obviously, there are sequences of continuous functions that converge point-

wise to continuous functions, but not uniformly, as shown in the following
example.

Example 3.1.3 Let .α ∈ R. Consider the sequence of functions of general

term
α −nx
.fn (x) = x n e , x ∈ R+
0.

This sequence converges pointwise to the function (Fig. 3.4)

f (x) = 0, ∀x ∈ R+
. 0.

However, this convergence is not uniform if .α ≥ 1: Note that

1 nα−1
. sup |fn (x) − f (x)| = max+ fn (x) = fn = ,
x∈R+ x∈R0 n e
0

which implies that

⎧
⎪
⎪+∞, if α > 1
⎪
⎨
nα−1 1
. lim sup |fn (x) − f (x)| = lim = , if α = 1
e ⎪e
⎪
x∈R+ 0 ⎪
⎩
0, if α < 1.
Figure 3.5 illustrates the uniform convergence of the sequence of functions
f (x) = x n1/2 e−nx .
. n

Remark the variation of .p ∈ N in relation to .δ > 0.

3.1. Introduction: Sequences of Functions 171

 
Figure 3.4: The sequence of functions . x nα e−nx , .α > 1, .α = 1, and .α < 1

 
Figure 3.5: The sequence of functions . x n1/2 e−nx
172 Series of Functions

3.2 Pointwise and Uniform Convergence of Series of

Functions
The concepts of pointwise and uniform convergence extend to series of
functions.

Definition 3.2.1 Let .(fn ) be a sequence of functions, .fn : X ⊂ R → R.

The series of general term .fn is the sequence of functions .(Sn ) defined
by
.Sn (x) = f1 (x) + f2 (x) + · · · + fn (x), ∀x ∈ X;

∞
the series is also represented by . fn or by . fn .
n=1

Definition 3.2.2 The series . fn converges at a point .a ∈ X if the

numerical series . fn (a)is convergent.
If the series is convergent at all points of .D ⊂ X, we can define a function
.f : D → R that corresponds to the sum of the series . fn (x) at each
point .x ∈ D; the function f is called the sum function of the series.
We also say that . fn (x) converges pointwise to f on D.

∞
x2
Example 3.2.1 Consider the series . .
n=0
(1 + x2 )n
If .x = 0, all terms are zero; therefore, the series is convergent.
If .x = 0, we can write
∞ ∞ ∞ n
x2 1 1
.
2 )n
= x2 2 )n
= x2 ,
n=0
(1 + x n=0
(1 + x n=0
1 + x2
1
and this is a geometric series with ratio .r = ; as .|r| < 1, the series
1 + x2
is convergent and
∞
x2 1
. = x2 · = 1 + x2 .
(1 + x 2 )n 1
n=0 1−
1 + x2
The sum function is
⎧
⎨1 + x2 , if x = 0
f (x) =
.
⎩0, if x = 0.
3.2. Pointwise and Uniform Convergence of Series of Functions 173

∞
xn
Example 3.2.2 Consider the series . . We can use a numerical series
n=0
n!
test to study the pointwise convergence of series of functions. In this case,
we will apply D’Alembert’s test to study the series

∞  n
x 
 .
.
 n! 
n=0

 n+1 
 x 
 
 (n + 1)!  |x|
. lim  n = lim = 0, ∀x ∈ R.
x  n +1
 
 n! 

Thus, we conclude that the original series is absolutely convergent .∀x ∈ R,

defining a function f on .R. We will see later that .f (x) = ex , that is,

∞
xn
. = ex , ∀x ∈ R.
n=0
n!

∞
Example 3.2.3 Consider the series . x (1 − x)n , x ∈ [0, 1].
n=0
If .x = 0, all terms are zero. Therefore, the series is convergent.
∞
If .x = 0, as the series . (1 − x)n is geometric with ratio .r = 1 − x and
n=0
|r| < 1 since .x ∈ ]0, 1], the series converges. In this case,
.

∞
1
. x (1 − x)n = x · = 1.
n=0
1 − (1 − x)

∞
Then, the series . x (1 − x)n , .x ∈ [0, 1], converges pointwise to the
n=0
function
⎧
⎨1, if 0 < x ≤ 1
f (x) =
.
⎩0, if x = 0.
174 Series of Functions

Definition 3.2.3 We say that a series . fn converges uniformly to the

function f on .D ⊂ R, .D = ∅, if
 
 n 
 
.∀δ > 0 ∃ p ∈ N : n > p ⇒ f (x) − fi (x) < δ, ∀x ∈ D.
 
i=1

This condition is equivalent to

∀δ > 0 ∃ p ∈ N : n > p ⇒ sup |f (x) − Sn (x)| < δ,

.
x∈D

that is,
. lim sup |f (x) − Sn (x)| = 0.
n→+∞ x∈D

Note: Uniform convergence implies pointwise convergence, but the opposite

is not necessarily true. This means that if a series of functions converges
uniformly, it will also converge pointwise. However, if a series of functions
converges pointwise, it does not necessarily converge uniformly.

∞
x2
Example 3.2.4 We saw that the series . is pointwise conver-
n=0
(1 + x2 )n
gent to the function f defined by
⎧
⎨1 + x2 , if x = 0
.f (x) =
⎩0, if x = 0.

However, this series is not uniformly convergent on .[−1, 1]. In fact,

lim sup |f (x) − Sn (x)|

n→+∞ x ∈ [−1,1]
 
 1 
= lim sup 1 + x2 − (1 + x2 ) 1 − 
n→+∞  2
(1 + x ) n+1 
x ∈ [−1, 1]
.
x = 0
 
 1  1
= lim sup  
n→+∞  (1 + x2 )n  = n→+∞
lim sup
(1 + x2 )n
= 1.
x ∈ [−1, 1] x ∈ [−1, 1]
x = 0 x = 0

The function f and some partial sums are illustrated in Fig. 3.6.
3.2. Pointwise and Uniform Convergence of Series of Functions 175

f (x)

Karl Theodor Wilhelm

Weierstrass (1815–1897)
was a German mathe-
matician often referred
to as the “father of
modern analysis.” His
work had a huge influence
on the mathematics of the
nineteenth and twentieth
centuries, and many of
Figure 3.6: An illustration of the sum function and some partial sums of Example 3.2.4
his results are part of
any Calculus course,
such as, for example,
the Weierstrass’ test for
The following theorem allows us to test the uniform convergence of a series series and his example
of functions without knowing its sum function. of a continuous function
that is not differentiable
at any point. (Source of
Theorem 3.2.1 It is a necessary and sufficient condition for the series
image: Dibner Library of
. fn to be uniformly convergent on .D ⊂ R that the History of Science and
  Technology, Smithsonian
 m 
  Libraries)
.∀δ > 0 ∃ p ∈ N : m > n > p ⇒  fr (x) < δ, ∀ x ∈ D.
 
r=n+1

Theorem 3.2.2 (Weierstrass’ Test) If there exists a convergent numer-

ical series with positive terms, . an , such that

|fn (x)| ≤ an , ∀ x ∈ D, ∀ n ∈ N,
.

then the series . fn is uniformly convergent on D.

Proof: We know by Theorem 2.2.3 that . an converges if and only if

∀δ > 0 ∃ p ∈ N : m > n > p ⇒ |an+1 + · · · + am | < δ.

.

Let .δ > 0 be arbitrary. We have

176 Series of Functions

     
fn+1 (x) + · · · + fm (x) ≤ fn+1 (x) + · · · + fm (x)

.
≤ an+1 + · · · + am ∀x ∈ D

= |an+1 + · · · + am |, since an > 0, ∀n ∈ N.

Then
∃ p ∈ N : m > n > p ⇒ |fn+1 (x) + · · · + fm (x)| < δ,
.

or still,  
 m 
 
.∃ p ∈ N : m > n > p ⇒  fr (x) < δ.
 
r=n+1

From Theorem 3.2.1 comes the desired result.

∞
Example 3.2.5 Let k be a constant such that .|k| < 1. The series . fn (x),
n=1
where
1 × 3 × · · · × (2n − 1) 2n  2n
f (x) =
. n ·k sin(x) ,
2 × 4 × · · · × 2n
is uniformly convergent on .R. In fact,
 
     
fn (x) =  1 × 3 × · · · × (2n − 1) · k 2n sin(x) 2n 
 2 × 4 × · · · × 2n 
.
1 × 3 × · · · × (2n − 1) 2n
≤ · k , ∀x ∈ R,
2 × 4 × · · · × 2n
and the numerical series
∞
1 × 3 × · · · × (2n − 1) 2n
. ·k
n=1
2 × 4 × · · · × 2n

is convergent. To verify this, we apply D’Alembert’s test:

1 × 3 × · · · × (2n − 1)(2n + 1) 2n+2

·k
2 × 4 × · · · × 2n(2n + 2) 2n + 1 2
. lim = lim · k = k 2 < 1.
1 × 3 × · · · × (2n − 1) 2n 2n + 2
·k
2 × 4 × · · · × 2n
3.2. Pointwise and Uniform Convergence of Series of Functions 177

  ∞
 sin(nx)  1 1

Example 3.2.6 Since .  ≤ 2 , .∀x ∈ R, and the series . is
n2  n n=1
n2
∞
sin(nx)
convergent, the series . is uniformly convergent on .R.
n=1
n2

Note: The Weierstrass’ test is a sufficient but not necessary condition

for the uniform convergence of a series of functions: there exist uniformly
convergent series whose general term cannot be bounded as required by the
Weierstrass’ test. It is important to note that this test implies absolute
convergence of the series of functions.

∞
x2 + n
Example 3.2.7 For each .x ∈ R, the series . (−1)n is alternating,
n=1
n2
and using Leibniz’s test, we can prove that it is convergent. However, it is
not absolutely convergent because
 
 2
n x + n
 x2 + n 1

. (−1) = ≥ , ∀x ∈ R,
 n2  n2 n
∞
1
and the series . is divergent. This makes it impossible to use the Weier-
n=1
n
strass’ test to determine the uniform convergence of the series. Instead, we
have to study it directly.
For each .x ∈ R, let .Sn (x) be the partial sum of order n of the series
∞
x2 + n
.S(x) = (−1)n . By Corollary 1 of Theorem 2.3.2,
n=1
n2

x2 + n + 1
|S(x) − Sn (x)| ≤
. .
(n + 1)2

If .X ⊂ R is a bounded set, that is, if there exists .M > 0 such that .|x| ≤ M ,
∀x ∈ X, then
.

M2 + n + 1
|S(x) − Sn (x)| ≤
. , ∀x ∈ X;
(n + 1)2

therefore .(Sn ) converges uniformly to S on X, that is, the given series

converges uniformly on any bounded subset of .R.
178 Series of Functions

The importance of uniform convergence lies in the fact that a series that
converges uniformly behaves similarly to the sum of a finite number of func-
tions. For instance, when we add a finite number of continuous functions,
the result is also a continuous function. Similarly, with a series of functions
that converge uniformly, we obtain the following outcome:

Theorem 3.2.3 If the functions .f1 , f2 , . . . , fn , . . . are continuous on D

and the series . fn converges uniformly to f on D, then f is continuous
on D.

Proof: Let .x0 be an arbitrary point of D. We want to prove that

∀δ > 0 ∃ ε > 0 : |x − x0 | < ε ⇒ |f (x) − f (x0 )| < δ.

.

We can write

f (x) − f (x0 ) = f (x) − Sn (x) + Sn (x) − Sn (x0 ) + Sn (x0 ) − f (x0 ),

.

which implies that

|f (x) − f (x0 )| ≤ |f (x) − Sn (x)| + |Sn (x) − Sn (x0 )| + |Sn (x0 ) − f (x0 )|.
.

Let .δ > 0. As the series converges uniformly to f , we know that

δ
∃ p ∈ N : ∀n > p, |f (x) − Sn (x)| <
. , ∀x ∈ D.
3
As .f1 , f2 , . . . , fn , . . . are continuous, .Sn is a continuous function, so
δ
∃ ε > 0 : |x − x0 | < ε ⇒ |Sn (x) − Sn (x0 )| <
. .
3
Then
∃ ε > 0 : |x − x0 | < ε ⇒ |f (x) − f (x0 )| < δ.
.

The result follows from the arbitrariness of .x0 .

∞
cos(nx)
Example 3.2.8 Consider the function .f (x) = √ . The terms of
n=1
n3 + 1
the series are functions with domain .R. The function f is the sum of the
series; therefore, its domain is the subset of .R where the series converges.
We have the inequality
 
 cos(nx)  1

. √ 
 n3 + 1  ≤ √n3 + 1 , ∀x ∈ R, (3.1)
3.2. Pointwise and Uniform Convergence of Series of Functions 179

since .| cos(x)| ≤ 1, .∀x ∈ R.

∞
1
Let us study the series of positive terms . √ , comparing it with
3
n +1
n=1
∞
1
the series . 3/2
, which we know is convergent. As the limit
n=1
n

1 
√
3
n +1 n3
. lim = lim =1
1 3
n +1
n3/2
∞
1
is finite and different from zero, we conclude that the series . √
+1 n3
 
n=1
∞
 cos(nx) 
is convergent. By inequality (3.1), the series . √ 
 n3 + 1  converges,
n=1
∞
cos(nx)
∀x ∈ R, so the series
. √
. is convergent, .∀x ∈ R, that is, the
n=1
n3 + 1
domain of f is .R. To conclude that f is continuous on .R, it is sufficient
that the following two conditions are satisfied:

cos(nx)
(i) The functions .fn (x) = √ are continuous on .R.
n3 + 1
∞
cos(nx)
(ii) The series . √ is uniformly convergent to f on .R.
n=1
n3 + 1

The functions .cos(nx) are continuous on .R, .∀n ∈ N, so the first condition is
true. The second condition is a consequence of the inequality (3.1) and the
Weierstrass’ test. As the series converges to f , this function is continuous.

Note: If the sum of a series of functions is not continuous, this indicates

that either the individual functions .f1 , .f2 , .. . . , .fn , . . . are not continuous
or the series is not uniformly convergent. Therefore, if .f1 , f2 , . . . , fn , . . .
are continuous functions and f is not continuous, we can be sure that the
convergence is not uniform.

∞
x2
Example 3.2.9 Let us consider the series . , on the interval
n=0
(1 + x2 )n
.[−a, a], .a > 0. We proved in Example 3.2.1 that this series converges
180 Series of Functions

pointwise to the function

⎧
⎨1 + x2 , if x = 0
. f (x) =
⎩0, if x = 0.

x2
Since f is discontinuous at .x = 0 and .fn (x) = is continuous,
(1 + x2 )n
∀n ∈ N, the series is not uniformly convergent.
.

Theorem 3.2.4 Let .a, b ∈ R, .a < b. If the functions .f1 , f2 , . . . , fn , . . .

∞
are continuous on .[a, b] and the series . fn converges uniformly to f on
n=1
[a, b], then
.
 b ∞  b
. f (x) dx = fn (x) dx.
a n=1 a

(The series is integrable term by term.)

Proof: By Theorem 3.2.3, f is continuous on .[a, b]; therefore, it is inte-

grable on .[a, b]. By hypothesis, the functions .f1 , f2 , . . . , fn , . . . are contin-
uous on .[a, b], which implies that they are integrable on this interval. We
intend to prove that the sequence .(Sn∗ ) of the partial sums of the series
∞  b  b
. fn (x) dx has limit . f (x) dx, that is,
n=1 a a

  b 
 
 ∗ 
.∀δ > 0 ∃ p ∈ N : n > p ⇒ Sn − f (x) dx < δ.
 a 

Let .δ > 0 be arbitrary.

  b     
   n b b 
 ∗   
. S n − f (x) dx =  fi (x) dx − f (x) dx
 a   a a 
i=1

    
 b n b 
 
= fi (x) dx − f (x) dx
 a a 
i=1
3.2. Pointwise and Uniform Convergence of Series of Functions 181
      
 b n  b n 
   
= fi (x) − f (x) dx ≤  fi (x) − f (x) dx
 a  a  
i=1 i=1

 b
= |Sn (x) − f (x)| dx.
a

∞
But the series . fn (x) converges uniformly to f on .[a, b]; therefore,
n=1

δ
∃ p ∈ N : n > p ⇒ |f (x) − Sn (x)| <
. , ∀x ∈ [a, b],
b−a
which implies that
 b  b
δ
∃p ∈ N : n > p ⇒
. |Sn (x) − f (x)| dx < dx = δ.
a a b−a

∞
e−nx
Example 3.2.10 Let us consider the series . , on .[0, 1].
n=1
2n
 −nx 
e  1 1
. 
 2n  = enx 2n ≤ 2n , ∀x ∈ [0, 1].

∞
1 1
The series .
n
is geometric with ratio . ; therefore, it is convergent.
n=1
2 2
According to Weierstrass’ test, the original series is uniformly convergent on
.[0, 1]. By Theorem 3.2.4,

 1 ∞ ∞ 
e−nx 1
e−nx
dx = dx
0 n=1
2n n=1 0 2n
.
∞  −nx 1 ∞
1 e 1 − e−n
= − = .
n=1
2n n 0 n=1 n2n

∞
xn
Example 3.2.11 The series is uniformly convergent on every in-
.
n=0
n!
terval .[a, b], .a, b ∈ R, because on this interval
182 Series of Functions

 n
x  Mn  
. 
 n!  ≤ n! , where M = max |a|, |b| ,

∞
Mn xn
and the series . is convergent. As .fn (x) = is continuous .∀n ∈ N,
n=0
n! n!
∞
xn
the series . is integrable term by term and
n=0
n!
 b ∞ n ∞  b n
x x
dx = dx
a n=0 n! n=0 a
n!
.
∞  b ∞
1 xn+1 bn+1 − an+1
= = .
n=0
n! n + 1 a n=0 (n + 1)!

Note: A series can be integrable term by term on an interval .[a, b] even if

it is not uniformly convergent on .[a, b].

∞
Example 3.2.12 The series .x + (xn − xn−1 ) is convergent on .[0, 1] to
n=2
the function f defined by
⎧
⎨0, if 0 ≤ x < 1
.f (x) =
⎩1, if x = 1.

Since .fn is continuous on .[0, 1], .∀n ∈ N, and f is discontinuous on this

 1
interval, the series is not uniformly convergent. However, . f (x) dx = 0
0
and
∞  1  1 ∞  1
fn (x) dx = x dx + (xn − xn−1 ) dx
n=1 0 0 n=2 0

 
2 1 ∞  
n 1 ∞
x xn+1 x 1 1 1
. = + − = + −
2 0 n=2
n+1 n 0 2 n=2 n+1 n

∞
1 1 1
= − − = 0.
2 n=2 n n+1
3.2. Pointwise and Uniform Convergence of Series of Functions 183

Corollary 1 Let .a, b ∈ R, .a < b. If the functions .f1 , f2 , . . . , fn , . . . are

∞
continuous on .[a, b] and the series . fn converges uniformly to f on
n=1
[a, b], then
.
 x ∞  x
. f (t) dt = fn (t) dt, ∀x ∈ [a, b].
a n=1 a

(The series is integrable term by term.)

∞
Corollary 2 Let .a, b ∈ R, .a < b. If the series . fn converges pointwise
n=1
on the interval .[a, b] to a function f , if on this interval the derivatives .fn
∞
exist and are continuous, and if the series . fn converges uniformly on
n=1
[a, b], then f is differentiable on .[a, b] and
.
∞
f  (x) =
. fn (x), ∀x ∈ [a, b].
n=1
(The series is differentiable term by term.)

∞
Proof: Let .g(x) = fn (x), .x ∈ [a, b]. By Corollary 1,
n=1

 ∞  ∞  x ∞
x x  
. g(t) dt = fn (t) dt = fn (t) = fn (x) − fn (a) ,
a a a
n=1 n=1 n=1

that is,
 x
. g(t) dt = f (x) − f (a);
a

therefore,
 x
f (x) =
. g(t) dt + f (a),
a

∞
which implies that .f  (x) = g(x), that is, .f  (x) = fn (x).
n=1
184 Series of Functions

∞
Example 3.2.13 Consider the series . xn . It is a geometric series of
n=0
ratio x. The series converges if and only if .|x| < 1 and, in this case,
∞
1
. xn = .
n=0
1−x

∞
Let .0 < r < 1. Then .|xn | ≤ rn , .∀x ∈ [−r, r]. As . rn is a convergent
n=0
∞
n
numerical series, then . x is uniformly convergent on .[−r, r]. Further-
n=0
more
 r  r ∞ ∞  r
1
dx = xn dx = xn dx
−r 1−x −r n=0 n=0 −r

 r ∞  n+1
r
x
⇔ − log |1 − x| =
−r
n=0
n+1 −r
.
∞
rn+1 − (−r)n+1
⇔ − log |1 − r| + log |1 + r| =
n=0
n+1
∞ ∞
1+r r n+1  r2n+1
⇔ log = 1 − (−1)n+1 = 2 .
1−r n=0
n+1 n=0
2n + 1

∞ ∞
Differentiating term by term the series . xn , we obtain . (n + 1) xn .
n=0 n=0
Since  
(n + 2)xn+1  (n + 2)|x|
. lim  
(n + 1)xn  = lim n + 1 = |x|,

by D’Alembert’s test, the series converges if .|x| < 1 and diverges if .|x| > 1;
∞ ∞
if .|x| = 1, we have the divergent series . (n + 1) and . (−1)n (n + 1).
n=0 n=0
∞
n
The series . (n + 1) x is uniformly convergent on every interval .[−r, r] if
n=0
∞
 
0 < r < 1 because .(n + 1)xn  ≤ (n + 1)rn and the series .
. (n + 1) rn is
n=0
3.2. Pointwise and Uniform Convergence of Series of Functions 185

convergent. We can then write

∞  ∞

1
= x n
= (n + 1) xn , |x| < 1
1−x n=0 n=0
.
∞
1
⇔ = (n + 1) xn , |x| < 1.
(1 − x)2 n=0
186 Series of Functions

3.3 Power Series

Definition 3.3.1 Let x0 ∈ R be arbitrary. A series of the form

∞
. an (x − x0 )n
n=0

with an ∈ R, ∀n ∈ N, is called a power series of x − x0 .

Notes:

1. By making y = x − x0 , a power series can always be reduced to the

∞
form an y n .
n=0

2. To ensure accuracy, the power series should be written as

+∞
a +
. 0 an (x − x0 )n .
n=1

At x = x0 , the sum of the series is a0 (all terms are 0 for n ≥ 1).

This is the meaning of the series that appears in the definition, which
we write in this way to simplify. When evaluating the indeterminate
form a0 00 at x = x0 , it should be interpreted as equal to a0 .

1
Theorem 3.3.1 Let  = r. If r ∈ R+ , the power series
lim n
|an |
∞
an x is absolutely convergent at each point x ∈ ] − r, r[ and di-
n

n=0
vergent at each point x ∈ ] − ∞, −r[ ∪ ]r, +∞[. If r = +∞, then the
power series is absolutely convergent whatever x ∈ R. If r = 0, the series
converges if x = 0 and diverges if x = 0.

∞
Proof: Consider the series |an xn |. Let r ∈ R+ . Using the fact that
n=0

 
lim
.
n
|an xn | = |x| lim n
|an |,
3.3. Power Series 187

we can apply Corollary 1 of the Root Test to state that if |x| lim n
|an | < 1
∞
(i.e., if |x| < r), the series converges; therefore, the series an xn con-
n=0
verges absolutely.

If |x| lim n |an | > 1 (i.e., if |x| > r), then, by the reasoning used in
 
Corollary 1 of the Root Test, there exists a subsequence of |an xn | that
takes values greater than or equal to 1, which implies that the sequence
   
|an xn | does not tend to zero; thus the sequence an xn does not tend
∞
to zero, and the series an xn diverges.
n=0

If r = 0, then lim |an | = +∞. It follows that, whatever x ∈ R, x = 0,
n

 ∞
|x| lim n |an | = +∞ > 1, and as before, the series an xn diverges.
n=0
Obviously, if x = 0, all terms are zero; thus, the series is convergent.

If r = +∞, then lim n |an | = 0. Therefore, whatever x ∈ R,
 ∞
|x| lim n |an | = 0 < 1, and we can conclude that the series an x n
n=0
converges absolutely.

Definition 3.3.2 Under the conditions of Theorem 3.3.1, r is called the

radius of convergence of the series, and the interval ] − r, r[ is the
interval of convergence.

 
 an 

Corollary 1 If lim   = r is a positive real number, zero, or +∞,
an+1 
then the radius of convergence of the power series is r.

Proof: It is sufficient to note that by Theorem 1.1.11,

 
 an  |an | 1 1
 
. lim
 an+1  = lim |an+1 | = lim 
n
|an |
= 
lim |an |
n
= r.


|an |; that
n
Note that there always exists lim  is, every power series has a
 an 
radius of convergence. However, lim   may not exist, as illustrated in
an+1 
the following example.
188 Series of Functions

∞ ∞
 n
Example 3.3.1 Consider the series an x n = 3 + (−1)n xn . As
n=0 n=0
⎧
an ⎨2n−1 , if n is even
=
⎩ 1 ,
.
an+1 if n is odd,
2n+2
 
 an  1
1

lim   does not exist, but r =  . =
an+1  lim |an | 4
n

1 1
The interval of convergence of the series is − , . The series converges
  4 4   
1 1 1 1
absolutely on the interval − , and diverges on −∞, − ∪ , +∞ .
4 4 4 4

Example 3.3.2 Let us calculate the radius of convergence of the series

∞
xn
:
n=1
nn
1 1 1
r=
.  =  = +∞.
=
lim n
|an | 1 1
lim lim
n

nn n
The radius of convergence of the series is infinite; therefore, the series is
absolutely convergent ∀x ∈ R.

∞
Example 3.3.3 The series n! xn has a radius of convergence r = 0:
n=0
 
 an  n! 1
.r = lim  
 an+1  = lim (n + 1)! = lim n + 1 = 0,

that is, the series only converges at x = 0.

∞
xn
Example 3.3.4 Consider the series . Taking into account that
n=1
n

1
 
 an  n n+1
.r = lim  
 an+1  = lim 1
= lim
n
= 1,

n+1
3.3. Power Series 189

the interval of convergence is ] − 1, 1[. The series converges absolutely on

the interval ] − 1, 1[ and diverges on ] − ∞, −1[ ∪ ]1, +∞[.

∞
(−1)n
Example 3.3.5 Consider the series n
(x + 1)n . Let y = x + 1.
n=0
2
The power series of y
∞
(−1)n n
. y
n=0
2n
has radius of convergence
 
   (−1)n 
 an   
.r = lim   = lim  2n 
 = 2.
 an+1   (−1) n+1 
 
2n+1

Then the power series of y converges absolutely if y ∈ ] − 2, 2[ and diverges

if y ∈ ] − ∞, −2[ ∪ ]2, +∞[. The power series of x + 1 converges absolutely
if x ∈ ] − 3, 1[ and diverges if x ∈ ] − ∞, −3[ ∪ ]1, +∞[.

∞
x2n+1
Example 3.3.6 Consider the series (−1)n . Let y = x2 . The
n=0
(2n + 1)!
∞
yn
series (−1)n has radius of convergence
n=0
(2n + 1)!
 
 (−1)n 
 
 (2n + 1)!  (2n + 3)!
.r = lim  
 (−1)n+1  = lim (2n + 1)! = lim (2n + 3)(2n + 2) = +∞;
 
 
(2n + 3)!

therefore, the series is absolutely convergent for all y ∈ R+

0 , and the original
series is absolutely convergent for all x ∈ R (note that it is enough to
multiply the power series of x by x).

Note: The preceding theorem omits any reference to the convergence or

divergence of the power series at the endpoints of the convergence interval
] − r, r[, r ∈ R+ . It is possible for the series to converge at both endpoints,
converge at one and diverge at the other, or diverge at both. Therefore, we
must always analyze the series at x = r and x = −r.
In the case of Example 3.3.4, the convergence interval is ] − 1, 1[:
190 Series of Functions

∞
(−1)n
– If x = −1, we obtain the series , which is conditionally con-
n=1
n
vergent.
∞
1
– If x = 1, we obtain the series , which is divergent.
n=1
n

We conclude that the series diverges on ] − ∞, −1[∪[1, +∞[ and converges

on [−1, 1[.

∞
Theorem 3.3.2 If the radius of convergence of the series an xn is
n=0
r > 0 and if 0 < ρ < r, then the series is uniformly convergent on
[−ρ, ρ].

Proof: By hypothesis, |an xn | ≤ |an | ρn = |an ρn |, ∀x ∈ [−ρ, ρ]. The series

∞
|an ρn | is convergent by Theorem 3.3.1, since ρ ∈ ] − r, r[.
n=0
∞
According to Weierstrass’ test, the series an xn is uniformly convergent
n=0
on [−ρ, ρ].

Corollary 1 A power series is uniformly convergent on every closed interval

[a, b] contained within its interval of convergence and
 b ∞ ∞
bn+1 − an+1
. an xn dx = an .
a n=0 n=0
n+1

Proof: If [a, b] ⊂ ] − r, r[, then there exists ρ > 0 such that

[a, b] ⊂ [−ρ, ρ] ⊂ ] − r, r[ .
.

By the theorem, the series is uniformly convergent on [−ρ, ρ] and also on

[a, b]. Then we can integrate the series term by term on [a, b]:
 b ∞ ∞  b ∞  b
n n
an x dx = an x dx = an xn dx
a n=0 n=0 a n=0 a
.
∞
bn+1 − an+1
= an .
n=0
n+1
3.3. Power Series 191

 1 ∞
x2n
Example 3.3.7 Let us evaluate f (x) dx, where f (x) = (−1)n .
0 n=0
(2n)!
∞
yn
Let y = x2 . The series (−1)n has an infinite radius of convergence:
n=0
(2n)!
 
 (−1)n 
 
 (2n)!  (2n + 2)!
.r = lim  
 (−1)n+1  = lim (2n)! = lim (2n + 2)(2n + 1) = +∞,
 
 
(2n + 2)!
which implies that the original series converges for all x in R; therefore, it is
uniformly convergent on [0, 1] and integrable term by term on that interval:
 1 ∞ ∞  1
x2n n x2n
(−1) dx = (−1)n dx
0 n=0 (2n)! n=0 0
(2n)!
.
∞  2n+1 1 ∞
1 x (−1)n
= (−1)n = .
n=0
(2n)! 2n + 1 0 n=0 (2n + 1)!

Theorem 3.3.3 Every power series with radius of convergence r > 0 is

term by term differentiable on the interval of convergence, that is,
∞  ∞
. an x n = n an xn−1 , ∀x ∈ ] − r, r[.
n=0 n=1

∞
Proof: Consider the series an xn . It is pointwise convergent on ] − r, r[;
n=0
the functions (an xn ) = nan xn−1 are continuous on ] − r, r[, ∀n ∈ N.
∞
Moreover, nan xn−1 is a power series with radius of convergence r:
n=0

1 1 1
.  = √  =  = r;
lim |nan |
n
lim n |an |
n n
lim |an |
n

therefore, it is uniformly convergent on [a, b] ⊂ ] − r, r[.

Thus, by Corollary 2 of Theorem 3.2.4,
∞  ∞
. an x n = n an xn−1 , ∀x ∈ ] − r, r[ .
n=0 n=1
192 Series of Functions

∞
Note: If the power series n=0 an xn has radius of convergence r, then
both the series of derivatives and the series of indefinite integrals have the
same radius of convergence, r.

∞
(x − 5)n
Example 3.3.8 Let us consider the power series (−1)n+1 .
n=1
n 5n
∞
yn
Let y = x − 5. The series (−1)n+1 n has radius of convergence
n=1
n5
 
 (−1)n+1 

  n+1
.r = lim 
n 5n  = lim (n + 1) 5 = 5,
 (−1)n+2  n 5n

 
(n + 1) 5n+1
which implies the absolute convergence of the original series on the interval
]0, 10[.
∞
−1
– If x = 0, we obtain the series , which is divergent.
n=1
n
∞
(−1)n+1
– If x = 10, we obtain the series , which is conditionally
n=1
n
convergent.

Then, the original series converges absolutely on the interval ]0, 10[, con-
verges conditionally at x = 10, and diverges on ] − ∞, 0] ∪ ]10, +∞[.

The series of derivatives is

∞  ∞
n+1 (x − 5) (x − 5)n−1
n
(−1) = (−1)n+1 n
n=1
n 5n n=1
n 5n
.
∞
(x − 5)n−1
= (−1)n+1 .
n=1
5n

The interval of convergence of this series is ]0, 10[.

∞
1
– If x = 0, we obtain the series which is divergent.
n=1
5
∞
(−1)n+1
– If x = 10, we obtain the series which is divergent.
n=1
5
3.4. Taylor Series and Maclaurin Series 193

3.4 Taylor Series and Maclaurin Series

Let I be an interval and .f : I ⊂ R → R a function of class .C n on I. Let

.x0 ∈ I. We know that there exists .0 < θ < 1 such that

.f (x) = f (x0 ) + f  (x0 )(x − x0 )

(x − x0 )2 (x − x0 )n−1
+f  (x0 ) + · · · + f (n−1) (x0 ) + Rn (x), Brook Taylor (1685–1731)
2! (n − 1)!
was an English mathe-
  (x − x0 )n matician. His work ti-
where .Rn (x) = f (n) x0 +θ (x−x0 ) . This is the Taylor’s formula tled Methodus Incremen-
n! torum Directa et Inversa
for f , of order n, with Lagrange remainder around the point .x0 .
(London, 1715) added a
Suppose that .f ∈ C ∞ (I). The power series new branch to Mathemat-
ics known as “calculus of
∞
f (n) (x0 ) finite differences.” Among
. (x − x0 )n other applications, he used
n=0
n!
it to determine the shape
of the movement of a vi-
is called the Taylor series of f at .x0 .
brating string, reducing it
If .x0 = 0 ∈ I, the Taylor series is called Maclaurin series and is written as to mechanical principles.
follows: This work also contains the
∞
f (n) (0) n famous formula known as
. x . Taylor’s Formula. (Source
n=0
n!
of image: Line engrav-
ing after Richard Earlom
Example 3.4.1 Let us determine the Maclaurin series of .f (x) = sin(x). (1743–1822))
 nπ 
We know that .f ∈ C ∞ (R) and .f (n) (x) = sin x + , .∀n ∈ N. Then
 nπ  2
(n)
.f (0) = sin , and therefore, the Maclaurin series of f is
2
∞
x3 x5 x7 x2n+1
x−
. + − + ··· = (−1)n .
3! 5! 7! n=0
(2n + 1)!

We saw in Example 3.3.6 that this series is convergent on .R.

Example 3.4.2 Consider the function .f (x) = (1 + x)α , .α ∈ R, .x > −1.

By the rules of differentiation, .f ∈ C ∞ (] − 1, +∞[) and

f (n) (x) = α(α − 1) . . . (α − n + 1)(1 + x)α−n , ∀n ≥ 1.

.

Therefore, .f (n) (0) = α(α − 1) . . . (α − n + 1), and its Maclaurin series is

α(α −1) 2 α(α −1) . . . (α − n + 1) n
1 + αx +
. x + ··· + x + ··· =
2! n!
194 Series of Functions

∞
α(α − 1) . . . (α − n + 1) n
. =1+ x .
n=1
n!
If .α ∈ N, the series is reduced to the development of Newton’s binomial.
Suppose that .α ∈ N0 . Then let us study the convergence of the series. The
radius of convergence is
 
 α(α − 1) . . . (α − n + 1) 
Collin Maclaurin (1698–  
 
1764) was a Scottish lim  n!  = lim (n + 1)! 1
α(α − 1) . . . (α − n + 1)(α − n)  n! |α − n|
mathematician. In 1717,  
at the age of 19 years, he
.
 (n + 1)! 
was appointed professor n+1
in Aberdeen. In 1719, = lim = 1.
|α − n|
he published the work
Organic Geometry , a text Therefore, the series is absolutely convergent on .] − 1, 1[ and divergent on
that can be considered
] − ∞, −1[ ∪ ]1, +∞[.
.
the most important of
his works and contains, This series is usually referred to as the binomial series.
among others, an original
method of generating
A fundamental question in the Taylor series development of an indefinitely
conics. He developed
Newton’s work in calculus, differentiable function is as follows:
geometry, and gravitation.
(Source of image: Pencil
Is there a neighborhood V of .x0 such that
and chalk on paper by
David Steuart Erskine
f (x) = f (x0 ) + f  (x0 )(x − x0 )
.

(1742–1829), Scottish (x − x0 )2 (x − x0 )n−1

+f  (x0 ) + · · · + f (n−1) (x0 ) + · · · , ∀x ∈ V ?
National Gallery) 2! (n − 1)!
In other words, does the Taylor series expansion of f at .x0 converge for all
x ∈ V and sum up to .f (x)?
.

In reality, the mere existence of the derivatives .f (n) (x0 ) for all natural val-
ues of n, although it allows writing the Taylor series of f at .x0 , does not
guarantee that, in some neighborhood of .x0 , the equality is satisfied:
∞
f (n) (x0 )
f (x) =
. (x − x0 )n , (3.2)
n=0
n!

as can be seen in the following example.

Example 3.4.3 Consider the function

⎧
⎨e−1/x2 , if x = 0
.f (x) =
⎩0, if x = 0.
3.4. Taylor Series and Maclaurin Series 195

As .f (n) (0) = 0, .∀n ∈ N, the Maclaurin series of f is written as follows:

0 + 0x + 0x2 + · · · ,
.

which converges to the zero function on .R. Therefore, the only point where
f equals the sum of the series is 0, since .f (x) = 0 if .x = 0.

To ensure equality (3.2), what additional conditions must be imposed on a

function f that is assumed to be indefinitely differentiable on a neighborhood
of .x0 ? We can easily answer this question, by considering Taylor’s formula
for f . Specifically, if .Sn (x) denotes the partial sum of order n of the Taylor
series of f at .x0 ∈ I, then we have .Rn (x) = f (x) − Sn (x), which satisfies
the following result:

Theorem 3.4.1 It is a necessary and sufficient condition for an indefinitely

differentiable function, .f : I → R, to be the sum of its Taylor series in a
neighborhood, V , of .x0 ∈ I, that

. lim Rn (x) = 0, ∀x ∈ V.

In practice, sufficient conditions are used:

Theorem 3.4.2 Let .f : I → R be an indefinitely differentiable function,

and suppose that there are constants .M, k ≥ 0 such that, in a neighbor-
hood, V , of .x0 ,
 
 (n) 
. f (x) ≤ M k n , ∀x ∈ V, ∀n ∈ N.

Then, f is the sum of its Taylor series on V .

Proof: We know that the expression of the Lagrange remainder, .Rn (x), is
  (x − x0 )n
Rn (x) = f (n) x0 + θ (x − x0 )
. , 0 < θ < 1.
n!
Then  n
k|x − x0 |
. |Rn (x)| ≤ M , x∈V;
n!
 n
k|x − x0 |
since the series of general term . is convergent, this sequence
n!
has limit 0. The desired result is an immediate consequence of Theo-
rem 3.4.1.
196 Series of Functions

Corollary 1 If there exists .M ≥ 0 such that on the neighborhood V of

.x0 ,  
 (n) 
. f (x) ≤ M ∀x ∈ V, ∀n ∈ N,

then f is the sum of its Taylor series on V .

Example 3.4.4 Consider the function .f (x) = sin(x). In Example 3.4.1,

we found that the Maclaurin series of f converges absolutely on .R. We
know that  
sin θx + nπ
.Rn (x) =
2
xn , 0 < θ < 1,
n!
from which
 
| sin θx + nπ
| |x|n
.0 ≤ |Rn (x)| = |x|n ≤
2
.
n! n!

|x|n
But .lim = 0, .∀x ∈ R, as it is the general term of a convergent series,
n!
which implies that
∞
x2n+1
. sin(x) = (−1)n , ∀x ∈ R.
n=0
(2n + 1)!

Figure 3.7: Some Taylor polynomials of the function .f (x) = sin(x)

In Fig. 3.7, we can see how the partial sums are increasingly better ap-
proximations of the function .sin(x), thus illustrating the convergence of
n
x2k+1
the series. The polynomial . (−1)k is denoted by .P2n+1 in the
(2k + 1)!
k=0
figure. The black line represents the graph of the function .sin(x).
3.4. Taylor Series and Maclaurin Series 197

Example 3.4.5 If .f (x) = ex , we obtain .f (n) (x) = ex , .∀n ∈ N. Then its

Maclaurin series development is
∞ ∞
f (n) (0) n xn
. x = .
n=0
n! n=0
n!

We know that this series is absolutely convergent on .R defining a function

g on .R. We intend to prove that .f (x) = g(x), .∀ x ∈ R. For this purpose,
we will show that the Lagrange remainder of the Maclaurin formula of the
function f tends to 0 on .R.
f (n) (θx) n eθ x n
Rn (x) =
. x = x , 0 < θ < 1,
n! n!
which implies that, taking into account that .eθ x ≤ e|x| since f is an in-
creasing function,
e|x| n
.0 ≤ |Rn (x)| ≤ |x| .
n!
e|x| n
The series of general term . |x| is convergent, .∀x ∈ R; therefore,
n!
e|x| n
. |x| = 0, ∀ x ∈ R,
lim
n!
which allows us to conclude that
∞
xn
ex =
. , ∀ x ∈ R.
n=0
n!

The Maclaurin series developments in the first two examples were obtained
directly from the formula:
f  (0) 2
f (0) + f  (0) x +
. x + ··· ,
2!
by evaluating all derivatives of the function at zero. Since this process
is quite laborious, it is not commonly employed. Instead, well-known de-
velopments are often utilized in practice. Furthermore, the next theorem
guarantees that these developments are indeed the Taylor series.

Theorem 3.4.3 Every power series of .x − x0 is the Taylor series (around

x ) of the function defined by it. In particular, the power series develop-
. 0

ment of .x − x0 is unique.
198 Series of Functions

∞
Proof: By hypothesis, .f (x) = an (x − x0 )n in a neighborhood V of .x0 ,
n=0
which implies that .f (x0 ) = a0 . Differentiating,
∞

f (x) =
. n an (x − x0 )n−1 ,
n=1

and therefore, .f  (x0 ) = a1 . The nth order derivative is

f (n) (x) = n! an + (n + 1) · · · 2 an+1 (x − x0 )

.

+(n + 2)(n + 1) · · · 3 an+2 (x − x0 )2 + · · · ,

and so .f (n) (x0 ) = n! an , .n ∈ N. Therefore, we have

f (n) (x0 )
a =
. n , ∀n ∈ N.
n!

1
Example 3.4.6 Consider the function .f (x) = . Taking into ac-
2 + 3x
count that
1 1 1
. = ·  
2 + 3x 2 1 − − 23 x
and that
∞
1
. = xn , ∀x ∈ ] − 1, 1[,
1 − x n=0
we can deduce that
∞ n
1 1 1 3
. ·  3 = − x ,
2 1 − −2 x 2 n=0 2
 
 3  2
and this equality is valid as long as .− x < 1, that is, .|x| < . Then, the
2 3
Maclaurin series of f is
∞
3n 2
. (−1)n xn , |x| < .
n=0
2n+1 3

1 1
Example 3.4.7 Let .f (x) = = . We have
x2 − x − 6 (x − 3)(x + 2)
 
1 1 1 1 1 1 1 1
.f (x) = − = − · − ·   .
5 x−3 x+2 5 3 1 − x3 2 1 − − x2
3.4. Taylor Series and Maclaurin Series 199

After taking into account that

1
∞  x n
. = , |x| < 3
1− x
3 n=0
3

and
1  x n ∞  x n ∞
.  x = − = (−1)n , |x| < 2,
1 − −2 n=0
2 n=0
2
we can write the Maclaurin series of f as follows:
 ∞  n

1 1 x 1
∞  n
n x
f (x) = − − (−1)
5 3 n=0 3 2 n=0 2
.
∞
1 (−1)n+1 1
= − n+1 xn , |x| < 2.
n=0
5 2n+1 3

Example 3.4.8 In Example 3.4.2 we developed the function

f (x) = (1 + x)α , α ∈ R,
. x > −1,

into a Maclaurin series, thereby obtaining

∞
α(α − 1) . . . (α − n + 1) n
. 1+ x ,
n=1
n!

convergent on the interval .] − 1, 1[. Let

∞
α(α − 1) . . . (α − n + 1) n
g(x) = 1 +
. x , |x| < 1.
n=1
n!

Next, we prove that .f (x) = g(x), .∀x ∈ ] − 1, 1[, that is, f is the sum of its
Maclaurin series in that interval.
Since a power series is differentiable term by term on the interval of conver-
gence, we obtain
∞
α(α − 1) . . . (α − n + 1) n−1
g  (x) =
. x ,
n=1
(n − 1)!

and multiplying by x
∞
α(α − 1) . . . (α − n + 1) n
x g  (x) =
. x .
n=1
(n − 1)!
200 Series of Functions

Then

g  (x) + x g  (x) =
.

∞ ∞
α(α − 1) . . . (α − n + 1) n−1 α(α − 1) . . . (α − n + 1) n
= x + x
n=1
(n − 1)! n=1
(n − 1)!
∞ ∞
α(α − 1) . . . (α − n) n α(α − 1) . . . (α − n + 1) n
= x + x
n=0
n! n=1
(n − 1)!
∞
α(α − 1) . . . (α − n) α(α − 1) . . . (α − n + 1)
=α+ + xn
n=1
n! (n − 1)!
∞
. α(α − 1) . . . (α − n + 1) α−n
=α+ + 1 xn
n=1
(n − 1)! n
∞
α(α − 1) . . . (α − n + 1) α n
=α+ · x
(n − 1)! n
 n=1
∞

α(α − 1) . . . (α − n + 1) n
=α 1+ x
n=1
n!

= α g(x),
that is,
(1 + x) g  (x) = α g(x).
. (3.3)
g(x)
Let us consider the function . and calculate its derivative:
(1 + x)α

g(x) g  (x)(1 + x)α − α(1 + x)α−1 g(x)
=
(1 + x)α (1 + x)2α
.  
(1 + x)α−1 (1 + x) g  (x) − α g(x)
= .
(1 + x)2α
By equality (3.3) the numerator of this fraction is zero; therefore,

g(x)
. = 0,
(1 + x)α
which implies that the function defined by
g(x)
.
(1 + x)α
is constant on .] − 1, 1[, that is, .g(x) = c (1 + x)α , for some constant c.
Since .g(0) = 1, we obtain .c = 1, and we have
g(x) = (1 + x)α , ∀x ∈ ] − 1, 1[ .
.
3.5. Introduction to Fourier Series 201

3.5 Introduction to Fourier Series

Taylor series are not the only series used to approximate and/or represent
functions. The Fourier series, which we introduce in this section, are another
example of how series are applied to the study of functions.
First, we establish some definitions and results concerning periodic functions
that will be useful in studying Fourier series (Fig. 3.8).

Definition 3.5.1 We say that a function .f : R → R is periodic with

period .T > 0 if .f (x + T ) = f (x), .∀x ∈ R.

Figure 3.8: A periodic function with period T

The following two propositions are direct consequences of the definition.

Proposition 3.5.1 Let .f : R → R be a periodic function with period T .

Then f is periodic with period kT , .∀k ∈ N.

Proposition 3.5.2 Let f and g be periodic functions with period T , and

.α, .β ∈ R. Then, the function .αf + βg is periodic with period T .

Consider the functions

a cos(kx) + bk sin(kx) , k = 1, 2, . . . ,
. k

where .ak , .bk .∈ R, ∀k ∈ N. These are periodic functions with period

2π
.Tk = , .k = 1, 2, . . .. Therefore, .T = 2π = kTk is a common period
k
to all. As a result, sums of the form
n
 
s (x) = A +
. n ak cos(kx) + bk sin(kx) ,
k=1
202 Series of Functions

where A is a constant, are periodic functions with period .T = 2π. These

sums are usually referred to as trigonometric polynomials.

Definition 3.5.2 A trigonometric series is a series of the form

∞
a0  
. + an cos(nx) + bn sin(nx) .
2 n=1

The constants .a0 , .an , and .bn , .n ∈ N, are the coefficients of the trigono-
metric series.

Two questions arise at this point:

1. What conditions ensure the convergence of a series of this kind?
2. What conditions must a function meet to be the sum of a trigono-
metric series, and how can the coefficients be determined?
Let us begin by answering the final part of the second question, remarking
that if the series converges, its sum is a periodic function with period .2π.
Suppose that the trigonometric series converges and its sum is a function
f , that is,
∞
a0  
.f (x) = + an cos(nx) + bn sin(nx) . (3.4)
2 n=1
The objective is to find the coefficients .a0 , .an , and .bn , .n ∈ N. Suppose the
function f is integrable and the series involved in this process is integrable
term by term. By integrating the previous equality from .−π to .π, we obtain
 π  π  π ∞
a0  
f (x) dx = dx + an cos(nx) + bn sin(nx) dx
−π −π 2 −π n=1
∞  π
 
. = πa0 + an cos(nx) + bn sin(nx) dx
n=1 −π
∞  π  π
= πa0 + an cos(nx) dx + bn sin(nx) dx .
n=1 −π −π

But  π  π
1
. cos(nx) dx = sin(nx) =0 (3.5)
−π n −π
and  π  π
1
. sin(nx) dx = − cos(nx) = 0; (3.6)
−π n −π
3.5. Introduction to Fourier Series 203

therefore,  π
. f (x) dx = πa0 .
−π

Solving for .a0 , we get  π

1
a =
. 0 f (x) dx.
π −π

To calculate .an , for .n ≥ 1, we multiply both sides of equality (3.4) by

.cos(mx) (.m ∈ N) and integrate term by term between .−π and .π:

 π
f (x) cos(mx) dx =
−π
 ∞
π
a0  
= + an cos(nx) + bn sin(nx) cos(mx) dx
−π 2 n=1
 π ∞  π
a0  
.= cos(mx) dx + an cos(nx) + bn sin(nx) cos(mx) dx
2 −π n=1 −π
 π
a0
= cos(mx) dx+
2 −π
∞  π  π
+ an cos(nx) cos(mx) dx + bn sin(nx) cos(mx) dx .
n=1 −π −π

From the trigonometric equalities

1    
. cos(nx) cos(mx) = cos (n + m)x + cos (n − m)x (3.7)
2
1    
sin(nx) cos(mx) = sin (n + m)x + sin (n − m)x , (3.8)
2
it follows 
 π
0, if n = m
. cos(nx) cos(mx) dx = (3.9)
−π π, if n = m
and  π
. sin(nx) cos(mx) dx = 0. (3.10)
−π

From the previous equalities, we obtain

 π
. f (x) cos(mx) dx = am π.
−π
204 Series of Functions

Solving for .am and considering that .m = n,

 π
1
.an = f (x) cos(nx) dx.
π −π

To evaluate .bn , for .n ≥ 1, we multiply both sides of equality (3.4) by

.sin(mx) (.m ∈ N) and integrate term by term between .−π and .π:

 π
f (x) sin(mx) dx =
−π
  ∞

π
a0  
= + an cos(nx) + bn sin(nx) sin(mx) dx
−π 2 n=1
 ∞  π
a0 π  
.= sin(mx) dx + an cos(nx) + bn sin(nx) sin(mx) dx
2 −π n=1 −π
 π
a0
= sin(mx) dx+
2 −π
∞  π  π
+ an cos(nx) sin(mx) dx + bn sin(nx) sin(mx) dx .
n=1 −π −π

From the trigonometric equality

1    
. sin(nx) sin(mx) = cos (n − m)x − cos (n + m)x , (3.11)
2

we conclude that

 
π
0, if n = m
. sin(nx) sin(mx) dx = (3.12)
−π π, if n = m,

and we obtain
 π
. f (x) sin(mx) dx = bm π.
−π

Solving for .bm and taking into account that .m = n,

 π
1
.bn = f (x) sin(nx) dx.
π −π
3.5. Introduction to Fourier Series 205

Definition 3.5.3 Let f be a real-valued function of a real variable, pe-

riodic with period .2π, for which the coefficients .a0 , .an , and .bn , .n ∈ N,
can be calculated as follows:

1 π
a0 = f (x) dx
π −π
 π
1
. an = f (x) cos(nx) dx
π −π
 π Joseph Fourier (1768–
1 1830) was a mathemati-
bn = f (x) sin(nx) dx,
π −π cian and physicist born
in Auxerre, France. He
and define the trigonometric series. This series is called the Fourier series studied at the École
Royale Militaire of the
of f . The coefficients are referred to as Fourier coefficients of f .
Benedictine Order, where
he became a professor.
Notes: During 1795, Fourier was
a student of Lagrange,
1. We will use the notation Laplace, and Monge at
∞
a0   the École Normal in Paris,
f (x) ∼
. + an cos(nx) + bn sin(nx) and in that same year,
2 n=1 he became a professor at
the École Polytechnique.
to mean that the series is the Fourier series of f , but we still do not In 1797, he succeeded
know if it converges to f . Lagrange in the chair of
Analysis and Mechanics.
2. The Fourier series is well determined by the values of the coefficients
His most well-known work
.a0 , .an , .bn , .n ∈ N, calculated earlier. If we change the value of the is Théorie analytique de
function f at a finite number of points, the integrals that define the la chaleur, published in
coefficients do not change. In particular, the value of the function 1822. Due to his research
in partial differential equa-
at isolated points, including the endpoints of the interval, does not tions, Fourier stated that
matter. As a result, we can define the function on the closed interval, “any” function could be
the open interval, or the interval that is closed at one endpoint and represented by a series of
open at the other. elementary trigonometric
functions—sines and
We selected the interval .[−π, π] for convenience. In fact, as the trigono- cosines. This statement
metric functions involved are periodic with a period of .2π, any interval of was so astonishing that
scientists of his time did
amplitude .2π can be used. It is essential to keep in mind Proposition 3.5.3,
not believe it. (Source
which can be easily proved by a change of variables. of image: Engraving by
Amédée Félix Barthélemy
Geille (1803–1843) after
Proposition 3.5.3 Let .f : R → R be a periodic function with period .2π.
Julien-Léopold Boilly
Then, for every a, .b ∈ R, (1796–1874))
 a+2π  b+2π
. f (x) dx = f (x) dx.
a b
206 Series of Functions

In the case where f is explicitly defined on the interval .[0, 2π] instead of
[−π, π], it may be more convenient to compute the coefficients using the
.

following formulas (see Example 3.5.6):

 2π
1
a0 = f (x) dx
π 0
 2π
1
. an = f (x) cos(nx) dx
π 0
 2π
1
bn = f (x) sin(nx) dx.
π 0

A problem that often appears in applications is to develop a function f in

a Fourier series with the function only defined on the interval .[−π, π] and
without reference to the periodicity of f . As the formulas to compute the
coefficients involve only the interval .[−π, π], we can calculate the Fourier
series of the function. The trigonometric functions intervening in the Fourier
series are periodic with period .2π. If the series is convergent, the sum
function, g, will also be periodic with period .2π:

g(x + 2π) = g(x), ∀x ∈ R.

.

When calculating the Fourier series of a function defined on .] − π, π], we

are, in fact, determining the Fourier series of its periodic extension, whose
definition we introduce next.

Proposition 3.5.4 Let .f : ]−π, π] → R. There exists a unique function .f˜

periodic with period .2π, called the periodic extension of f , that satisfies
.f˜(x) = f (x), for all .x ∈ ] − π, π].

Given the expression of .f (x), the function .f˜ is well defined by

f˜(x) = f (x), ∀x ∈ ] − π, π]
.
f˜(x + 2π) = f˜(x), ∀x ∈ R.

In some situations, specifying the analytical expression of .f˜ may be useful.

3.5. Introduction to Fourier Series 207

Example 3.5.1 The periodic extension of .f (x) = x, defined on .] − π, π],

is the function .f˜ represented in Fig. 3.9.

Figure 3.9: Periodic extension of .f (x) = x, .x ∈ ] − π, π]

The analytical expression of .f˜ is

f˜(x) = x − 2kπ, x ∈ ](2k − 1)π, (2k + 1)π], k ∈ Z.

.

Example 3.5.2 The periodic extension of .f (x) = x2 , defined on the in-

terval .] − π, π], is the function .f˜ represented in Fig. 3.10.

Figure 3.10: Periodic extension of .f (x) = x2 , .x ∈ ] − π, π]

Its analytical expression is

f˜(x) = (x − 2kπ)2 , x ∈ ](2k − 1)π, (2k + 1)π], k ∈ Z.

.

Note: If .f (−π + ) = f (π) and assuming that f is continuous on .] − π, π], its

extension by periodicity is a continuous function on .R. If .f (−π + ) = f (π),
the periodic extension is not continuous.
208 Series of Functions

Before we study the properties of the Fourier series, let us examine some
examples of how these series are calculated.

Example 3.5.3 Consider the real-valued function of a real variable

⎧
⎨1, if − π ≤ x < 0
.f (x) =
⎩x, if 0 ≤ x ≤ π.

The graph of f is presented in Fig. 3.11. The Fourier coefficients of f can

be calculated using integration by parts.

 π  0  π
1 1
a0 = f (x) dx = 1 dx + x dx
Figure 3.11: The graph of π −π π −π 0
the function of Example 3.5.3   0  2 π 
1 x π
= x + =1+ ;
π −π 2 0 2
 0  π
1
an = cos(nx) dx + x cos(nx) dx
π −π 0
 0  π  π 
1 sin(nx) sin(nx) sin(nx)
= + x − dx
π n −π n 0 0 n
 π
1 cos(nx) 1 cos(nπ) cos(0) (−1)n − 1
. = = − = ;
π n2 0 π n2 n2 πn2
 0  π
1
bn = sin(nx) dx + x sin(nx) dx
π −π 0
 0  π  
π
1 cos(nx) cos(nx) cos(nx)
= − + −x + dx
π n −π 0 n 0 n
 π
1 1 cos(−nπ) cos(nπ) sin(nx)
= − + −π +
π n n n n2 0

1 1 (−1)n π (−1)n (1 − π)(−1)n − 1

= − + − = .
π n n n πn

The Fourier series of f is

∞
1 π (−1)n − 1 (1 − π)(−1)n − 1
. + + cos(nx) + sin(nx) .
2 4 n=1 π n2 πn
3.5. Introduction to Fourier Series 209

With this example, it is easy to verify that the sum of the series is not equal
to the function at all points of the interval .[−π, π]. In fact, at .x = −π and
at .x = π, we obtain the numerical series
∞
1 π 2
. + + ,
2 4 n=1 π (2n − 1)2
and .f (−π) = 1 = f (π) = π.

Example 3.5.4 Let us consider the function .f (x) = x defined on .[−π, π]

(Fig. 3.12). The Fourier coefficients of f can be computed using integration
by parts and equalities (3.5) and (3.6).
  π
1 π 1 x2 Figure 3.12: The graph of
a0 = x dx = = 0; the function of Example 3.5.4
π −π π 2 −π
  π  π 
1 π 1 sin(nx) sin(nx)
an = x cos(nx) dx = x − dx = 0;
π −π π n −π −π n
.
  π  π 
1 π 1 cos(nx) cos(nx)
bn = x sin(nx) dx = −x + dx
π −π π n −π −π n

1 cos(nπ) cos(−n π) 2
= −π −π = (−1)n+1 .
π n n n
∞
2
The Fourier series of f is . (−1)n+1 sin(nx).
n=1
n

Example 3.5.5 Let us consider the function .f (x) = x2 defined on .[−π, π]

(Fig. 3.13). Using the calculations from the previous example, we have

1 π 2 2π 2
a0 = x dx = ;
π −π 3
  π  π 
1 π 2 1 2 sin(nx) sin(nx)
.
an = x cos(nx) dx = x − 2x dx
π −π π n −π −π n
4
= (−1)n 2 ;
n
 Figure 3.13: The graph of
1 π 2 the function of Example 3.5.5
bn = x sin(nx) dx
π −π
.  π  π 
1 2 cos(nx) cos(nx)
= −x + 2x dx = 0.
π n −π −π n
210 Series of Functions

Thus, the Fourier series of f is

∞
π2 4
. + (−1)n 2 cos(nx).
3 n=1
n

In the following example, we have a periodic function with period .2π, defined
on .]0, 2π[ (see Proposition 3.5.3).

Example 3.5.6 Consider the real-valued function of a real variable


sin(x), if 0<x<π
.f (x) =
0, if π ≤ x < 2π.

The graph of f is shown in Fig. 3.14 Using equalities (3.8) and (3.11), we
can calculate the Fourier coefficients as follows:
 
1 2π 1 π 2
a0 = f (x) dx = sin(x) dx = ;
π 0 π 0 π
 
1 2π 1 π
Figure 3.14: The graph of an = f (x) cos(nx) dx = sin(x) cos(nx) dx
the function of Example 3.5.6 π 0 π 0
⎧
⎪
⎨0, if n is odd
= 2
. ⎪
⎩ , if n is even;
π(1 − n2 )
 
1 2π 1 π
bn = f (x) sin(nx) dx = sin(x) sin(nx) dx
π 0 π 0
⎧
⎨0, if n = 1
= 1
⎩ , if n = 1.
2

Thus, the Fourier series of f is

∞
1 1 2
. + sin(x) + cos(2nx).
π 2 n=1
π(1 − 4n2 )
3.5. Introduction to Fourier Series 211

Let us examine the issue of convergence of Fourier series. Generally, this

topic can be quite complex. Even if we know that the series does converge,
the issue of determining the sum function still remains. For instance, in
Example 3.5.4, the series cannot converge to .f (x) = x for all values of x.
When .x = π, each term in the series is zero, causing the series to converge
to zero, which is different from .f (π) = π.

We now introduce some concepts that will allow us to approach the study
of the convergence of the Fourier series, which is an essential problem of
this theory.

Definition 3.5.4 A function f is piecewise continuous on an interval

.[a, b] if it is continuous on that interval or continuous except at a finite
number of points .x1 ≤ x2 ≤ · · · ≤ xn , .xi ∈ [a, b], .∀i ∈ {1, . . . , n}, and
exist, at each point of discontinuity, the left- and right-hand limits

.f (x−
k ) = lim− f (x) and f (x+
k ) = lim+ f (x).
x→xk x→xk

A function f is piecewise continuous on .R if it is piecewise continuous on

every bounded interval of .R.

In Fig. 3.15 we can see an illustration of the definition above.

Figure 3.15: A piecewise continuous function

212 Series of Functions

Definition 3.5.5 A function f is piecewise of class .C1 , or simply piece-

wise .C1 , on an interval .[a, b] if it is continuously differentiable on that
interval, or continuously differentiable except at a finite number of points
.x1 ≤ x2 ≤ · · · ≤ xn , .xi ∈ [a, b], .∀i ∈ {1, . . . , n}, and exist, at each point

.xk , the left- and right-hand limits

f (x−
.
k ) = lim f (x) and f (x+
k ) = lim f (x)
x→x−
k x→x+
k

and
f  (x−
.

k ) = lim f (x) and f  (x+ 
k ) = lim f (x).
x→x−
k x→x+
k

A function f is piecewise .C 1 on .R if it is piecewise .C 1 on every bounded

interval of . R.

Theorem 3.5.1 If the function f , periodic with period .2π, is piecewise

C 1 on .[−π, π], its Fourier series converges at all points. At points of
.

continuity of f , the sum of the series is equal to the value of the function.
At each point of discontinuity of f , the sum of the series is equal to the
arithmetic mean of the left- and right-hand limits of the function at that
point.

Example 3.5.7 In Example 3.5.4, we obtained the Fourier series of the

function .f (x) = x, .x ∈ ] − π, π]
∞
2
. (−1)n+1 sin(nx).
n=1
n

In Example 3.5.1, we extended f by periodicity, defining a function .f˜. From

the previous theorem, we obtain
⎧
∞ ⎪
⎨f˜(x), if x = (2k + 1)π, k ∈ Z
2(−1)n+1
. sin(nx) = ˜ + ˜ −
n ⎪
⎩ f (x ) + f (x ) = 0, if x = (2k + 1)π, k ∈ Z.
n=1
2

Figure 3.16 shows some partial sums of the Fourier series of the function .f˜.
3.5. Introduction to Fourier Series 213

(a) The partial sums S1 (x) and S3 (x)

(b) The partial sums S5 (x) and S10 (x)

Figure 3.16: Partial sums of the Fourier series of the function from Example 3.5.7

Example 3.5.8 Consider the real-valued function of a real variable f de-

fined by (Fig. 3.17)
⎧
⎨0, if − π < x < 0
.f (x) =
⎩3, if 0 < x < π.

The Fourier coefficients are

 π  π
1 1
a0 = f (x) dx = 3 dx = 3;
π −π π 0
 π  π
1 1 Figure 3.17: The graph of
an = f (x) cos(nx) dx = 3 cos(nx) dx = 0;
. π −π π 0 the function of Example 3.5.8
⎧
  ⎪ 6
1 π
1 π ⎨ , if n is odd
bn = f (x) sin(nx) dx = 3 sin(nx) dx = nπ
π −π π 0 ⎪
⎩0, if n is even.
214 Series of Functions

The Fourier series is

∞
3 6  
. + sin (2n − 1)x
2 n=1 (2n − 1)π

and converges pointwise to the function

⎧
⎪
⎪ 0, if (2n − 1)π < x < 2nπ, n ∈ Z
⎪
⎨
.f˜(x) = 3, if 2nπ < x < (2n + 1)π, n ∈ Z
⎪
⎪
⎪
⎩3, if x = nπ, n ∈ Z.
2

Figure 3.18 shows some partial sums of the Fourier series of the function f .

Example 3.5.9 In Example 3.5.3, we obtained the Fourier series for the
real-valued function of a real variable
⎧
⎨1, if − π ≤ x < 0
.f (x) =
⎩x, if 0 ≤ x ≤ π.

The Fourier series is given by

∞
1 π (−1)n − 1 (1 − π)(−1)n − 1
. + + 2
cos(nx) + sin(nx) .
2 4 n=1 πn πn

The series converges pointwise to the function

⎧
⎪
⎪ 1, if (2k − 1)π < x < 2kπ, k ∈ Z
⎪
⎪
⎪
⎪
⎪
⎨1, if x = 2kπ, k ∈ Z
.f˜(x) = 2
⎪
⎪x − 2kπ, if 2kπ < x < (2k + 1)π, k ∈ Z
⎪
⎪
⎪
⎪
⎪
⎩π + 1, if x = (2k + 1)π, k ∈ Z.
2
The coefficients .an of the function .f (x) = x are all zero. This is a conse-
quence of the fact that f is odd.
Let us consider the following two results, which can be easily shown (see
Figs. 3.19 and 3.20).
3.5. Introduction to Fourier Series 215

(a) The partial sums S1 (x) and S3 (x)

(b) The partial sums S5 (x) and S7 (x)

Figure 3.18: Partial sums of the Fourier series of the function from Example 3.5.8

Proposition 3.5.5 The sum of two even functions is an even function,

and the sum of two odd functions is an odd function. The product of two
even functions or two odd functions is an even function. The product of
an even function and an odd function is an odd function.

Figure 3.19: An odd func-

Proposition 3.5.6 Let .a ∈ R+ and .f : R → R be an integrable function tion
on .[−a, a].
 a  a
a) If f is even, then . f (x) dx = 2 f (x) dx.
−a 0
 a
b) If f is odd, then . f (x) dx = 0.
−a

Let f be an even function. Taking into account that the functions .cos(nx)
are even and .sin(nx) are odd, for all .n ∈ N, by Proposition 3.5.5, we find Figure 3.20: An even func-
that .f (x) cos(nx) are even and .f (x) sin(nx) are odd, for all .n ∈ N. tion
216 Series of Functions

By Proposition 3.5.6, the Fourier coefficients of f are

 
1 π 2 π
a0 = f (x) dx = f (x) dx;
π −π π 0
 
1 π 2 π
an = f (x) cos(nx) dx = f (x) cos(nx) dx;
. π −π π 0

1 π
bn = f (x) sin(nx) dx = 0.
π −π

We can state the following proposition:

Proposition 3.5.7 Let f be an even function defined on .[−π, π]. Then

∞
a0
f (x) ∼
. + an cos(nx),
2 n=1

2 π
where .an = f (x) cos(nx) dx, .n ∈ N0 , that is, f is represented by
π 0
a Fourier cosine series.

Example 3.5.10 Consider the even function .f (x) = |x| on the interval
.[−π, π] (Fig. 3.21). The Fourier coefficients of f are
 
2 π 2 π
a0 = f (x) dx = x dx = π;
π 0 π 0
 
2 π 2 π
an = f (x) cos(nx) dx = x cos(nx) dx
. π 0 π 0
⎧ 4
⎨− , if n is odd
= n2 π
⎩
0, if n is even, n = 0.
The analytical expression of the periodic extension of f to .R is
f˜(x) = |x − 2kπ|, x ∈ [(2k − 1)π, (2k + 1)π] k ∈ Z.
.

The function is continuous on .R; therefore by Theorem 3.5.1,

∞
π 4  
f˜(x) = −
. cos (2n − 1)x , ∀x ∈ R.
2 n=1 (2n − 1) π
2
3.5. Introduction to Fourier Series 217

Figure 3.21: The graph of the extension of .f (x) = |x|

Let f be an odd function. Considering that .cos(nx) are even and .sin(nx)
are odd for all .n ∈ N, by Proposition 3.5.5, we obtain that .f (x) cos(nx)
are odd and .f (x) sin(nx) are even for all .n ∈ N. By Proposition 3.5.6, the
Fourier coefficients of f are

1 π
a0 = f (x) dx = 0;
π −π

1 π
. an = f (x) cos(nx) dx = 0;
π −π
 π 
1 2 π
bn = f (x) sin(nx) dx = f (x) sin(nx) dx.
π −π π 0
We can state the following proposition:

Proposition 3.5.8 Let f be an odd function defined on .[−π, π]. Then

∞
f (x) ∼
. bn sin(nx),
n=1

2 π
where .bn = f (x) sin(nx) dx, .n ∈ N, that is, f is represented by a
π 0
Fourier sine series.

Example 3.5.11 Consider the function f defined by

⎧
⎨−π − x, if − π ≤ x < 0
.f (x) =
⎩π − x, if 0 ≤ x < π.

It is an odd function (Fig. 3.22). The Fourier coefficients of f are

 
2 π 2 π 2
.bn = f (x) sin(nx) dx = (π − x) sin(nx) dx = .
π 0 π 0 n
218 Series of Functions

Figure 3.22: Graph of the extension of the function from Example 3.5.11

The analytical expression of the periodic extension of f to .R is

f˜(x) = −x + (2k + 1)π, x ∈ [2kπ, (2k + 2)π[ k ∈ Z.

.

It is a piecewise .C 1 function on .R; therefore by Theorem 3.5.1,

⎧
⎪ ˜
∞
2 ⎨f (x), if x ∈ ]2kπ, (2k + 2)π[, k ∈ Z
⎪
. sin(nx) =
n ⎪
⎪ 0, if x = 2kπ, k ∈ Z.
n=1 ⎩

Sometimes, we must solve the problem of determining the development in a

Fourier series of sines or cosines of a function defined on the interval .[0, π].
To develop f in a Fourier cosine series, we consider an even extension of f
from the interval .[0, π] to the interval .[−π, π] (see Fig. 3.23):
⎧
⎨f (−x), if − π < x < 0
f
. even (x) =
⎩f (x), if 0 ≤ x < π.

(a) (b) (c)

Figure 3.23: (a) The function f . (b) Even extension of f . (c) Odd extension of f
3.5. Introduction to Fourier Series 219

Then, we apply everything that has been stated for extensions as an even
function. Therefore, the Fourier coefficients can be computed by the for-
mulas of Proposition 3.5.7.
To develop f in a Fourier sine series, we take an odd extension of f from
the interval .[0, π] to the interval .[−π, π] (see Fig. 3.23):
⎧
⎨−f (−x), if − π < x < 0
.fodd (x) =
⎩f (x), if 0 ≤ x < π.

Next, we apply the previous knowledge of extensions as an odd function.

Consequently, the formulas of Proposition 3.5.8 can be used to determine
the Fourier coefficients.

Example 3.5.12 Let us determine the Fourier sine series and the Fourier
cosine series of the function f defined by .f (x) = x3 −x, on .[0, π[ (Fig. 3.24).
The analytical expression of the even extension to the interval .] − π, π[ is
⎧
⎨−x3 + x, if − π < x < 0
.feven (x) =
⎩x3 − x, if 0 ≤ x < π.

(a) (b) (c)

Figure 3.24: (a) The function f . (b) Even extension of f . (c) Odd extension of f
220 Series of Functions

The coefficients of the Fourier cosine series are

  π
2 π 3 2 x4 x2 π3
a0 = (x − x) dx = − = − π;
π 0 π 4 2 0 2
 
2 π 2 π 3
. an = f (x) cos(nx) dx = (x − x) cos(nx) dx
π 0 π 0
2 3π 2 6 + n2 
= (−1)n 2 + (1 − (−1)n ) ,
π n n4
and the series is
∞
π3 π 2 3π 2 6 + n2  
. − + (−1)n 2
+ 4
1 − (−1)n cos(nx).
4 2 π n=1
n n

Note that the periodic extension of .feven is

⎧
⎨−(x − 2kπ)3 + (x − 2kπ), if (2k − 1)π < x < 2kπ, k ∈ Z
.f˜even (x) =
⎩(x − 2kπ)3 − (x − 2kπ), if 2kπ ≤ x < (2k + 1)π, k ∈ Z

whose graph can be seen in Fig. 3.25.

Figure 3.25: Periodic extension of the function .feven

Since .f˜even is continuous on .R, Theorem 3.5.1 allows us to affirm that the
following equality is satisfied, whatever .x ∈ R:
∞
π3 π 2 3π 2 6 + n2  
f˜
. even (x) = − + (−1)n 2
+ 4
1 − (−1)n cos(nx).
4 2 π n=1
n n

The analytical expression of the odd extension to the interval .] − π, π[ is

f˜
. odd (x) = x3 − x, x ∈ ] − π, π[ .
3.5. Introduction to Fourier Series 221

The coefficients of the Fourier sine series are


2 π 1 − π2 6
.bn = f (x) sin(nx) dx = 2 (−1)n + 3 ,
π 0 n n

and the series is

∞
1 − π2 6
2
. (−1)n + 3 sin(nx).
n=1
n n

The periodic extension of .f˜odd has the analytical expression

f˜
. odd (x) = (x − 2kπ)3 − x − 2kπ, x ∈ ](2k − 1)π, (2k + 1)π[, k ∈ Z,

and its graph is shown in Fig. 3.26.

Figure 3.26: Periodic extension of the function .fodd

The function is continuous on .R \ {(2k + 1)π, k ∈ Z}, but it is piecewise

C 1 on .R. Theorem 3.5.1 allows us to affirm that the equality is satisfied:
.

⎧
∞
1−π 2
6 ⎨f˜odd (x), if x = (2k + 1)π
.2 (−1)n + 3 sin(nx) =
n n ⎩0, if x = (2k + 1)π,
n=1

for all .k ∈ Z.
222 Series of Functions

Thus far, we have dealt with periodic functions of period .2π. Let us now
focus on functions f with period .T = 2l, .l > 0.
lt lt
Let .x = . The function .g(t) = f has period .2π. In fact,
π π

l(t + 2π) lt lt
g(t + 2π) = f
. =f + 2l =f = g(t).
π π π

The Fourier series of g is

∞
a0  
. + an cos(nt) + bn sin(nt) ,
2 n=1

where  π
1
a0 = g(t) dt;
π −π
 π
1
. an = g(t) cos(nt) dt;
π −π
 π
1
bn = g(t) sin(nt) dt.
π −π

lt
Then the Fourier series of .f (x) = f is
π

a0  ∞  nπx   nπx 
. + an cos + bn sin ,
2 n=1
l l

where
 π  l 
1 1 π 1 l
a0 = g(t) dt = f (x) ·
dx. = f (x) dx ;
π −π −l π l l −l
   nπx  π
1 π 1 l
an = g(t) cos(nt) dt = f (x) cos · dx
π −π π −l l l
 l  
1 nπx
. = f (x) cos dx ;
l −l l
   nπx  π
1 π 1 l
bn = g(t) sin(nt) dt = f (x) sin · dx
π −π π −l l l
 l  
1 nπx
= f (x) sin dx.
l −l l
3.5. Introduction to Fourier Series 223

Proposition 3.5.9 If f is a periodic function with period 2l, the coeffi-

cients of its Fourier series are given by
 l
1
a0 = f (x) dx
l −l
 l  nπx 
1
. an = f (x) cos dx
l −l l
 l  nπx 
1
bn = f (x) sin dx.
l −l l

Example 3.5.13 Consider the function f , periodic with period 4, defined

by ⎧
⎨0, if − 2 ≤ x < 0
.f (x) =
⎩2, if 0 ≤ x < 2.

The graph of f is shown in Fig. 3.27. The coefficients of its Fourier series
are given by
 
1 2 1 2
a0 = f (x) dx = 2 dx = 2;
2 −2 2 0
 2  nπx   2  nπx 
Figure 3.27: The graph of
1 1 the function of Example
an = f (x) cos dx = 2 cos dx = 0; 3.5.13
2 −2 2 2 0 2
.   nπx    nπx 
1 2 1 2
bn = f (x) sin dx = 2 sin dx
2 −2 2 2 0 2
⎧
⎨0, if n is even
= 4
⎩ , if n is odd.
nπ
The Fourier series of f is

a0  ∞  nπx   nπx 
+ an cos + bn sin
2 n=1
2 2
.
∞
4 (2n − 1)πx
= 1+ sin .
n=1
(2n − 1)π 2
224 Series of Functions

Example 3.5.14 Consider the function f defined by

⎧
⎨x, if 0 ≤ x < 1
.f (x) =
⎩2 − x, if 1 ≤ x < 2.

We want to develop f into a Fourier sine series. For this, we take an odd
extension of f , which has period .2l = 4 (Fig. 3.28).

Figure 3.28: Odd extension of function f

The coefficients of its Fourier series are given by

  nπx   2  nπx 
1 2
bn = f (x) sin dx = f (x) sin dx
2 −2 2 0 2
8  nπ 
= sin
. (nπ)2 2
⎧
⎪
⎨0, if n is even
= 8
⎪
⎩ (−1)(n−1)/2 , if n is odd.
(nπ)2

The Fourier sine series of f is

∞  nπx  8
∞
(−1)n+1 (2n − 1)πx
. bn sin = 2 sin .
n=1
2 π n=1
(2n − 1) 2 2

We saw in Theorem 3.5.1 sufficient conditions for the convergence of the

Fourier series of a periodic function with period .2π. If f has period 2l, the
theorem can be easily generalized:
3.5. Introduction to Fourier Series 225

Theorem 3.5.2 If the function f , periodic with period 2l, is piecewise .C 1

on .[−l, l], its Fourier series converges at all points. At points of continuity
of f , the sum of the series is equal to the value of the function. At each
point of discontinuity of f , the sum of the series is equal to the arithmetic
mean of the left- and right-hand limits of the function at that point.

We often want to differentiate and integrate term by term the Fourier series
and, as we saw in Sect. 3.2, sufficient conditions to guarantee this procedure
involve the concept of uniform convergence. We have the following results.

∞
 
Theorem 3.5.3 If the series . |an |+|bn | converges, then the trigono-
n=1
metric series
a0
∞   nπx   nπx 
. + an cos + bn sin
2 n=1
l l
converges absolutely and uniformly on .R, and its sum is a continuous
function. The trigonometric series is the Fourier series of the sum func-
tion.

Proof: It is a direct consequence of the Weierstrass’ test because

  nπx   nπx    nπx    nπx 
     
an cos + bn sin  ≤ an cos  + bn sin 
. l l l l
≤ |an | + |bn |, ∀x ∈ R.

Theorem 3.5.4 Let f be a periodic function with period 2l, continuous,

piecewise .C 1 . Then, its Fourier series converges absolutely and uniformly
to f .

Theorem 3.5.5 If the Fourier series of a continuous function f is con-

vergent, then the sum of the series is f .

Theorem 3.5.6 Let f be a periodic function with period 2l, piecewise

.C 1 on .[−l, l]. Let .x1 ≤ x2 ≤ · · · ≤ xn , .xi ∈ ] − l, l[, .∀i ∈ {1, . . . , n}, be
the points of discontinuity of f . Let .[a, b] ⊂ ] − l, l[ such that .xi ∈ / [a, b],
.∀i ∈ {1, . . . , n}. Then, its Fourier series converges uniformly to f on

.[a, b].
226 Series of Functions

Theorem 3.5.7 Let f be a piecewise continuous function on .[−l, l[ and

periodic with period 2l. Then, the Fourier series of f is integrable term
by term. More precisely, if
∞   nπx   nπx 
a0
.f (x) ∼ + an cos + bn sin
2 n=1
l l

and .a, b ∈ R, then

 b
a0
f (x) dx = (b − a)
a 2
   
.
∞ b  nπx  b  nπx 
+ an cos dx + bn sin dx .
n=1 a l a l

It should be noted that the previous theorem remains valid even if the
Fourier series is not pointwise convergent for the function f . By integrating
1
term by term, a factor of . is introduced in the series, which results in its
n
convergence.

Theorem 3.5.8 Let f be a continuous function on the interval .[−l, l],

with .f (−l) = f (l), periodic with period 2l and such that .f  is piecewise
1 
.C on .[−l, l]. Then, the Fourier series of .f can be obtained by differenti-

ating term by term the Fourier series of f . In addition, the Fourier series
of .f  converges to .f  at the points of continuity of .f  , and at each point
of discontinuity, the sum of the series is equal to the arithmetic mean of
the left- and right-hand derivatives of the function f at that point.

Note that the theorem regulating term by term differentiation is more de-
manding than that of term by term integration. Recall what we saw in
Sect. 3.2: Term by term differentiation of a series of functions can only be
guaranteed if the series of derivatives is uniformly convergent.
The previous theorems allow us to operate on the Fourier series similarly to
what was done with the Taylor series.

Example 3.5.15 We saw in Example 3.5.4 that

∞
2
.x= (−1)n+1 sin(nx), −π < x < π,
n=1
n
3.5. Introduction to Fourier Series 227

that is,
∞
x sin(nx)
. = (−1)n+1 , −π < x < π.
2 n=1 n
We can integrate this series term by term, obtaining
 x ∞  x
t sin(nt)
dt = (−1)n+1 dt, −π < x < π
0 2 n=1 0 n
.
∞ ∞
x2 (−1)n+1 cos(nx)
⇔ = 2
− (−1)n+1 , −π < x < π.
4 n=1
n n=1
n2

The value of the numerical series can be computed by integrating this result
between .−π and .π:
 π 2 ∞ ∞ 
x (−1)n+1 (−1)n+1 π
dx = 2π − cos(nx) dx
−π 4 n=1
n2 n=1
n2 −π
. ∞
(−1)n+1
= 2π .
n=1
n2

Therefore,
∞
π3 (−1)n+1
. = 2π ,
6 n=1
n2
that is,
∞
π2 (−1)n+1
. = .
12 n=1 n2
Thus, the Fourier series of .f (x) = x2 is
∞
π2 cos(nx)
x2 =
. +4 (−1)n , −π < x < π,
3 n=1
n2

as we saw in Example 3.5.5.

The hypothesis about the continuity of the function in Theorem 3.5.8 is

essential, as can be seen in the following example.

Example 3.5.16 Consider the development in Fourier series of the func-

tion .f (x) = x defined on .[−π, π], obtained in Example 3.5.4. We saw, in
Example 3.5.7, that
∞
2
x=
. (−1)n+1 sin(nx), −π < x < π.
n=1
n
228 Series of Functions

We have .f  (x) = 1, but

∞  ∞
2 2
(−1)n+1 sin(nx) = (−1)n+1 n cos(nx)
n=1
n n=1
n
. ∞
= 2 (−1)n+1 cos(nx),
n=1

and this series diverges whatever .x ∈ R (the general term is not a null
sequence).
This result does not contradict Theorem 3.5.8 because .f (−π) = f (π),
which shows the need for the condition of equality. The extension by peri-
odicity of f is discontinuous (see Example 3.5.1). We can observe that, in
Theorem 3.5.8, the continuity of f on .[−l, l] and the condition .f (−l) = f (l)
ensure the continuity of the extension by periodicity of f .
3.6. Solved Exercises 229

3.6 Solved Exercises

1. Consider the real-valued function f of a real variable, 6. Consider the real-valued function f of a real variable,
which is defined by which is defined by
∞
sin(n2 x) ∞
e−nx
.f (x) = √5
. .f (x) = .
n=1 n7 + 3 n 4+5
n=0

 1
a) Find the domain of f . Find f (x) dx.
0
b) Check if f is continuous on its domain.
7. Consider the series of functions defined, for x ≥ 0,
2. Consider the real-valued function f of a real variable, ∞   x
1 n+ n
which is defined by by x+ .
n=1
n
∞
 1
.f (x) = . a) Determine the values of x for which the series
n=0
1 + (3x)n
is convergent.
b) Let 0 < α < 1. Show that the series con-
a) Determine the domain of f .
verges uniformly on [0, α].
b) Show that f is continuous on [1, +∞[.
8. Let (an ) be a sequence of positive terms such that
3. Consider the real-valued function f of a real variable,
which is defined by an+1 1
.lim = .
∞
 an 2
x
.f (x) = .
n=1
nx If, for each n ∈ N, fn : R → R is a function such
that
1
.|fn (x)| ≤ , ∀ x ∈ R,
a) Find the domain of f . n
b) Show that f is continuous on ]1, +∞[. ∞

show that the series of functions an fn (x) con-
4. Consider the real-valued function f of a real variable, n=1
which is defined by verges uniformly on R.
∞
 1 9. Let fn : [0, 1] → R, n ∈ N, and f : [0, 1] → R be
.f (x) = √ .
(n x + 1) n3 + 1 continuous functions such that
n=1
 2 ∞

Calculate f (x) dx. .f (x) = fn (x), ∀x ∈ [0, 1].
1 n=1
5. Consider the real-valued function f of a real variable,
Assuming that
which is defined by
 1
∞ 1 1
cos(nx) (i) fn (x) dx = −
.f (x) = . n n+1
n2 0
n=1  1
Show that (ii) f (x) dx = 0
0
 π/2 ∞
 (−1)n
. f (x) dx = . show that the series of functions does not converge
0 (2n + 1)3
n=0 uniformly on [0, 1].
230 Series of Functions

∞
10. Consider the real-valued function f of a real variable, 3n log(n) n
c) x
which is defined by n=1
7n n2
∞
∞ 2 × 5 × · · · × (3n − 1) n
arccot(nx) d) x
.f (x) = √ . (n + 2)!
n=1 n5 n=1
∞
log(n) n
a) Determine the domain of f . e) x
n=1
n
b) Demonstrate that f is continuous on its do- ∞
 3 n2
main. f) xn
n=0
4n (n3 + 2)
c) Show that f is differentiable on its domain.
∞
1 × 4 × · · · × n2 n
g) x
11. Let (an ) be a sequence of positive terms. Show (2 n)!
∞ n=1
that an cos(nx) converges uniformly on R if, ∞   n
1
n=0 h) nn sin 2
xn
∞
 n=1
n
and only if, an converges. ∞
 1
n=0 i)   xn
  n=1
2n n arctan(n) − π
2
12. Consider a sequence of functions fn , continu- ∞
n∈N
 3 × 6 × · · · × 3n n
ous on R, such that j) x
n=1
(n + 3)!
.|fn (x)| ≤ an , ∀n ∈ N, ∀x ∈ R, ∞
 n+1
k) (−1)n √ xn
where an is the general term of a convergent series n=0 e2n + 5
of positive terms. Let (bn )n∈N be a sequence such ∞
1 × 3 × 5 × · · · × (2n − 1) 1 n
∞ l) · x
that the series (bn − bn+1 ) converges. Show n=1
2 × 4 × 6 × · · · × 2n n
n=1
∞

that the series of functions bn fn (x) defines a 15. Find the real values of x for which the following se-
n=1 ries are absolutely convergent, conditionally conver-
continuous function on R.
gent, and divergent. Explain the reasoning behind
∞
 the response.
13. a) Show that the series (−1)n (1 − x)xn con-
n=0 ∞ 
 2n
n
verges uniformly on [0, 1]. a) (x − 1)n
n=1
2n + 1
∞

(1 − x)xn converges ∞

b) Show that the series (x − 2)n
n=0 b) (−1)n
n log(n)
pointwise, but not uniformly, on [0, 1]. n=2
∞
 n
14. Find the real values of x for which the following series c) (x + 4)n
n=1
(−3)n (2n2 − 1)
are absolutely convergent, conditionally convergent,
∞

and divergent. Give a justification for the answers en
d) (x − e)n
provided. log(n)
n=2
∞ ∞ π 
 π n xn  cos 2n
a) e)
n=2
n log2 (n) n=1
(2n − 1)2 (x − 1)n
∞
 ∞
2n nn (2 x2 − 5)n
b) xn f)
n=1
(n + 1)n+1 n=1
n 3n+1
3.6. Solved Exercises 231

∞ √ √ ∞
 n+1− n 
g) √ √ (x + 2)n 20. Assume that the series an 5n converges. Is the
n=1
n+1+ n n=1
 n ∞

∞
log2 (n) 2x series an (−2)n convergent? Provide a justifi-
h)
n=1
n2 x2 + 1 n=1
cation for the answer.
∞
 1 1 ∞

i) √ · n
2n n + 1 x 21. Can the power series an (x − 3)n converge at
n=1
n=0
∞  
2 1 n the point x = 0 and diverge at the point x = 1?
j) + (2x − 3)n
3 2n
n=1 ∞

∞  2 bn xn be a power series whose radius of con-
 23n (n − 1)! 22. Let
k) x2n n=0  
n=2
(2n)! vergence is R > 1. Let fn n∈N be a sequence of
∞ 
  functions such that
2n − 1 2n
l) (x − 3)n
n=0
3n + 1 .|fn (x)| ≤ |bn |, ∀x ∈ R.
∞ π
 sin n What can be said about the uniform convergence of
m) (x − 1)2n ∞
4n (n + 1) 
n=1 the series of functions fn ?
n=0
16. Find the values
 of x for which the series
∞ 23. Let (an ) be a sequence of positive terms such that
x2n x2 ∞
√ log 1 + √ is convergent. Justify the 
n=1
n n the series an (x − 1)n is conditionally conver-
answer. n=1
gent at x = −1.
17. Compute the radius of convergence of the series a) What is the radius of convergence of the power
∞  π n ∞

cos + 2 − (−1)n x2n and indicate the series an (x − 1)n ? Justify.
n=1
n
n=1
largest open interval on which the series converges.
b) Show that x = −1 is the only point at which
∞
 the series is conditionally convergent.
4n
18. Consider the power series n + 5n
xn . Know-
2 ∞

n=1 1
ing that the radius of convergence of this series is 24. Consider the series (x − 2)n .
5 n=1
n 5n
, indicate, without resorting to additional calcu-
4 a) Find the real values of x for which the series
lations, whether the series converges or diverges at
points 1 and 2. Investigate the convergence of the is absolutely convergent, conditionally conver-
5 gent, and divergent. Justify.
series at the point .
4 b) Let f be the sum of the series. Find the set
19. Find the real values of x for which the series of points where f is differentiable and deter-
∞
 mine the analytical expression of f  . Justify
an xn is absolutely convergent, conditionally
the answer given.
n=0
convergent, and divergent, where ∞
(−1)n  n+1
25. Consider the series log(x) .
n=0
n + 1
1, if n is even
.an =
2, if n is odd. a) Find the real values of x for which the series
is absolutely convergent, conditionally conver-
Justify the answer. gent, and divergent. Justify.
232 Series of Functions

5x − 1 x
b) Using the series of derivatives, calculate, for a) g) cos(t2 ) dt
x2 − x − 2 0
each x, the sum of the series.
 x
b)
2
et dt x + x2
h)
26. Consider the real-valued function f of a real variable, 0 (1 − x)2
which is defined by  1 1 − e−tx i) x arctan(x)
c) dt
0 t
∞
 2x
(−1)n 
.f (x) = x2n . x j)
22n (n!)2 d) arctan(t) dt (1 + x2 )2
n=0
0
k) log(2 + x3 )
1
e) 
(2x + 5)2 l) log 1+x
a) Determine the domain of f . 1−x
2
b) Show that x2 f  (x) + xf  (x) + x2 f (x) = 0. f) + e2x m) log(4 − x2 )
3 + 2x
Justify the answer given.
30. a) Determine Maclaurin series of the function

27. Let (an ) be a sequence of real numbers such that x

∞ .f (x) = .
 (1 − x)2
the series an 2n is conditionally convergent.
n=1
Indicate for which values of x the development
is valid. Justify the answer.
∞

a) Test for convergence the series an and Hint: Use term by term differentiation.
n=1
∞
 b) Using item a), calculate the sum of the series
∞
an 3n . Give a reason for the answer. n
n=1 n
.
n=0
3
sin(an )
b) Calculate lim . Justify.
an 31. Let f be a real-valued function of a real variable
defined by
28. Let f be a real-valued function of a real variable, ⎧
defined by ⎨ sin(x) , if x = 0
∞ .f (x) = x
 ⎩
.f (x) = an xn , 1, if x = 0.
n=0

such that f  (x) = f (x), ∀x ∈ R, and f  (0) = 1.

a) Expand the function f in a Maclaurin series.
Indicate for which values of x the development
a) For each n ∈ N, compute an . is valid. Explain the reasoning behind the re-
Hint: Use the method of Mathematical Induc- sponse.
tion. b) Use item a) to obtain the value of f (n) (0), for
all n ∈ N.
b) Using item a), identify the function f .

32. Consider the real-valued function f of a real variable

29. Develop the following functions in a Maclaurin series defined by
and indicate the interval on which the expansion is
2
valid. Explain the findings for each question and .f (x) = ex + cos(2x).
provide a reason for the answers.
3.6. Solved Exercises 233

a) Compute the Maclaurin series of f , indicating 36. Represent in a power series of x − 3 the real-valued
the values of x for which the development is function f of a real variable defined by
valid. 1
.f (x) = + log(4 − x).
b) Using the expansion obtained in item a), indi- 4−x
cate the value of f (17) (0). Determine the values of x for which the development
is valid. Give a clear explanation of the answer.
33. Using only the Maclaurin expansions of the functions
1
f (x) = sin(x), g(x) = ex , and h(x) = , find 37. Expand in a power series of x − 3 the real-valued
1−x function f of a real variable defined by
the sum of the following series:
1
∞
 .f (x) = .
1 n x2
a) (−1)
(2n + 1)!
n=0 Determine the values of x for which the development
∞ is valid. Provide a clear explanation of the answer.
2n
b)
n=0
n! Hint: Use term by term differentiation.
∞
 1 38. Develop in a power series of x + 1 the real-valued
c) (−1)n
n=1
2n n function of a real variable f defined by
1
34. Consider the function f : R → R defined by .f (x) = .
x2 + x − 6

.f (x) = log(x2 + 1) + cos(2x). Indicate the values of x for which the expansion is
valid. Justify.

39. Consider the function

 x
a) Find the Maclaurin series of f and indicate the log(t) − t + 1
.f (x) = dt.
values of x for which the development is valid. 1 t−1

b) Using item a) compute the sum of the series Find the power series of x − 1 of f , indicating the
values of x for which the development is valid.
∞
  
π 2n 1 22n
.1 + (−1)n−1 − . 40. Consider the function
n=1
42n n (2n)!  x
arctan(t − 2) − t + 2
.f (x) = dt.
2 (t − 2)2
35. Consider the function f : R → R defined by
Expand f in a power series of x − 2, indicating the
2x values of x for which the development is valid.
.f (x) = sin(2x) + .
(1 − x2 )2 41. Let f be a function of class C ∞ in a neighborhood
of the origin that satisfies the following conditions:

a) Express f in a power series of x indicating the .f (0) = 1 and f  (x) = f (x) + x, ∀x ∈ R.

values of x for which the development is valid.
Find the Maclaurin development of f .
b) Use item a) to calculate the sum of the series
42. Find the Fourier series relative to the periodic func-
∞
   tions of period 2π, defined on ] − π, π[ as follows:
(−1)n n+1
. + 2n .
n=0
(2n + 1)! 2 π, if −π <x≤0
a)
0, if 0<x<π
234 Series of Functions

0, if −π <x≤0 of period 2π defined on the interval [0, 2π] as follows:

b)
ex , if 0<x<π ⎧
⎪ a
⎪
⎪ x, if 0 ≤ x ≤ b
⎪
⎪ b
cos(x), if − π < x ≤ 0 ⎪
⎪
c) ⎪
⎪a, if b < x ≤ π − b
sin(x), if 0 < x < π ⎪
⎨a
.f (x) = (π − x), if π − b < x ≤ π + b
d) x3 ⎪
⎪ b
⎪
⎪ −a, if π + b < x ≤ 2π − b
⎪
⎪
⎪
⎪ a
e) 4 − |x| ⎪
⎩ b (x − 2π),
⎪ if 2π − b < x ≤ 2π.

43. Considering that the following functions, defined on

the indicated intervals, are periodic with period 2π, 46. Let f : [−π, π] → R be a piecewise C 1 function and
find their respective Fourier series and their sums.
∞
a0  
. + an cos(nx) + bn sin(nx)
a) x, 0 < x < 2π 2 n=1

b) x2 , 0 < x < 2π be its Fourier series. Consider the functions

π 3π
c) |x|, − < x ≤ f (x) + f (−x) f (x) − f (−x)
2 2 .g(x) = and h(x) = .
⎧ π π 2 2
⎪
⎨x, if − ≤ x ≤
2 2 Determine the Fourier series of g and h.
d)
⎪
⎩π − x, if π < x ≤ 3π
2 2 47. Show that if f is a periodic function with period 2π
⎧ π π and such that f (x + π) = −f (x), ∀x ∈ R, then the
⎪
⎨x , if − 2 ≤ x ≤ 2
2
Fourier series of f is
e)
⎪ π2
⎩ π 3π
, if <x≤ ∞
     
4 2 2
. a2n−1 cos (2n − 1)x +b2n−1 sin (2n − 1)x .
n=1
44. Consider that the following functions are periodic
and defined on the interval corresponding to one pe-
48. Consider the real-valued function of a real variable,
riod. Determine the Fourier series and the respective
sum for each function.
0, if 0 ≤ x < 2
.f (x) =
a) π sin(πx), 0 < x < 1 x − 2, if 2 ≤ x < 4.

b) 2 − x2 , 0 < x < 1

c) x − 3, 0 < x ≤ 4 a) Find the analytical expression of the odd ex-

⎧ tension, f˜, of f to the interval ] − 4, 4[ and
⎪
⎨0, if − 2 ≤ x ≤ −1
⎪ sketch its graph.
d) x, if − 1 < x < 1 b) Determine the Fourier series of f˜ and sketch,
⎪
⎪
⎩1, if 1 ≤ x < 2
on R, the graph of its sum.
⎧ x
⎪
⎨− , if − 3 < x ≤ 0 49. Obtain the function
e) 3 2
⎪ x
⎩2x − , if 0 < x ≤ 3 1, if 0 ≤ x < π
3 .f (x) =
0, if π < x ≤ 2π
45. Let a ∈ R \ {0} and b ∈ ]0, π2 [. Determine the
Fourier series corresponding to the periodic function as the sum of a Fourier cosine series.
3.6. Solved Exercises 235

50. Consider the function f (x) = π − x. 54. Consider the Fourier series of the periodic function
of period 2π, f˜, defined on the interval [−π, π] by
a) Develop the function into a Fourier sine series f (x) = |x|,
on ]0, π[. Use the result to calculate the sum
∞
of the series π 2  (−1)n − 1
. + cos(nx).
∞
sin(nx) 2 π n=1 n2
. , 0 < x < 2π.
n=1
n
∞

b) Using the result from the previous item, cal- a) Show that the series n an sin(nx), where
culate the sum of the series n=1
an are the Fourier coefficients of the original
∞
cos(nx) series, is convergent on R.
. , 0 < x < 2 π. ∞
n=1
n2 
b) Let g(x) = − n an sin(nx). Sketch the
n=1
c) Develop the function into a Fourier cosine se- graph of g on the interval [−2π, 2π].
ries on ]0, π[.
∞ c) Calculate the sum of the series
 (−1)n
d) Calculate the sum of the series . ∞ ∞
n=1
n2  1  (−1)n
. and .
n=1
(2n − 1)2 n=1
(2n − 1)3
51. Consider the function f (x) = x.

a) Develop the function into a Fourier sine series 55. Determine, using Fourier series, the function f such
on ]0, π[. Use the result to calculate the sum that
of the series
∞
 f  (x) + 3f (x) = 3x, 0<x<1
sin(nx) .
. (−1)n+1 , −π < x < π. f (0) = f (1) = 0.
n=1
n

b) Develop the function into a Fourier cosine se- 56. Consider the periodic function with period 2 defined
ries on ]0, π[. by f (x) = x2 on the interval ]0, 2].

52. Develop into a Fourier series the periodic function of a) Determine the Fourier series of f .
period 2π, defined on ] − π, π[ by f (x) = | sin(x)|, b) Show that the term by term differentiation of
and use the result to show that the series obtained in item a) does not con-
∞
 1 1 ∞
(−1)n 1 π verge to f  .
. = and = − .
n=1
4n2 −1 2 n=1
4n 2−1 2 4 57. Consider the periodic function with period 6 defined
  on ] − 3, 3[ by f (x) = x − x2 .
Hint: sin(a) cos(b) = 1
2
sin(a + b) + sin(a − b) .
a) Find the Fourier series of f .
53. Show that on the interval ]0, π[ we have
b) Let S(x) be the sum of the series
∞
π4 cos(2nx)
.x (π − x) = −3
2 2
a0 x  3   nπx  nπx
, ∞
30 n=1
n4 . + an sin − bn cos ,
2 n=1
nπ 3 3
and from this result, we obtain
∞ ∞
where a0 , an , and bn are the Fourier coeffi-
 1 π4  1 π4
. = and = . cients of f . Determine the expression of S(x),
n 4 90 (2n + 1) 4 96
n=1 n=0 x ∈ ] − 3, 3[.
236 Series of Functions

SOLUTIONS

∞
 sin(n2 x)
1. a) The terms of the series √5
are functions of domain R. The function f is the sum of the series;
n=1 n7 + 3
therefore, its domain is the subset of R where the series converges. The general term satisfies
 
 sin(n2 x) 
.  √ ≤ √ 1 , ∀x ∈ R, ∀n ∈ N, (3.13)
5
n + 3
7 5
n7 + 3
since | sin(x)| ≤ 1, ∀x ∈ R.
∞
 1
We test the convergence of the series of positive terms, √
5
, by comparison with the series
n=1 n7 + 3
∞
 1
, which we know to be convergent. As the limit
n=1
n7/5

1 
√
5
n7 + 3 5 n7
. lim = lim =1
1 n7 + 3
n7/5

is finite and different from zero, we conclude, by Corollary 2 of the General Comparison Test, that the series
∞ ∞  
1  sin(n2 x) 
√ √ 
5
n7 + 3
is convergent. By inequality (3.13), the series  5 n7 + 3  converges, ∀x ∈ R. Then, the
n=1 n=1
∞
 sin(n2 x)
series √5
is convergent, ∀x ∈ R, that is, the domain of f is R.
n=1 n7 + 3
b) To conclude that f is continuous on R, it is sufficient (see Theorem 3.2.3) that the following two conditions
are satisfied:
sin(n2 x)
(i) The functions fn (x) = √ 5
are continuous on R.
n7 + 3
∞
sin(n2 x)
(ii) The series √5
is uniformly convergent to f on R.
n=1 n7 + 3
The first condition is immediately satisfied because the functions hn (x) = sin(n2 x) are continuous on
R. The second is a consequence of the previous item: Inequality (3.13) and the convergence of the se-
∞ ∞
1 sin(n2 x)
ries √
5
are sufficient to conclude the uniform convergence of the series √5
by the
7
n +3 n7 + 3
n=1 n=1
Weierstrass’ test. As the series converges to f , this function is continuous.
∞
 1  1
2. a) The terms of the series n
are functions with domain R, if n is even, and R \ − , if n is
n=0
1 + (3x) 3
 1
odd. The function f is the sum of the series, so its domain is the subset of R \ − where the series
3
converges. For the series to converge, its general term must be a null sequence. But
⎧
⎪
⎪1, if |3x| < 1
⎪
⎨
1 0, if |3x| > 1
. lim =
1 + (3x)n ⎪
⎪1
⎪
⎩ , if 3x = 1;
2
3.6. Solved Exercises 237

1 1
therefore, if |x| ≤ , x = − , the series diverges.
3 3
∞ 


 1 
Let x be such that |3x| > 1. We study the series of positive terms  
 1 + (3x)n  using D’Alembert’s
n=0
test:    
 1   1 1 
     + 
 1 + (3x)n+1   1 + (3x) 
n  (3x)n+1 
. lim   = lim   = lim  3x = 1 .

 1 
 1 + (3x)n+1  
 1 
 |3x|
 1 + (3x)n   (3x)n+1 + 1 

1 1
We conclude that the series converges because < 1, that is, |x| > .
|3x| 3
   
1 1
The domain of f is −∞, − ∪ , +∞ .
3 3
b) To show that f is continuous on [1, +∞[, it is sufficient (see Theorem 3.2.3) that the following two conditions
are met:
1
(i) The functions fn (x) = are continuous on [1, +∞[.
1 + (3x)n
∞
1
(ii) The series is uniformly convergent to f on [1, +∞[.
n=1
1 + (3x)n
The first condition is immediately satisfied. The second condition follows from the inequality
   n
 1 
.  ≤ 1 ≤ 1 , ∀x ∈ [1, +∞[, ∀n ∈ N,
1 + (3x)n  (3x)n 3
∞  n
 1
since the series is convergent, which, by the Weierstrass’ test, is sufficient to conclude the
n=1
3
∞
 1
uniform convergence of the series . As the series converges to f , this function is continuous.
n=1
1 + (3x)n

∞
 x
3. a) Let f (x) = . The terms of the series are functions with domain R. The function f is the sum of
n=1
nx
the series; therefore, its domain is the subset of R where the series converges.

– If x = 0, all terms are zero; hence, the series converges.

∞
 ∞
x 1
– If x = 0, then x
=x x
. This series is a p-series; it converges if and only if x > 1.
n=1
n n=1
n
Therefore, the domain of f is the set ]1, +∞[ ∪ {0}.
∞
 1
b) Let us consider the function g(x) = . To conclude that f is continuous on ]1, +∞[, it is sufficient
n=1
nx
to prove that g is continuous in this set since f (x) = x g(x). If we show that g is continuous on [a, +∞[, for
every a > 1, then g is continuous on ]1, +∞[. For this, it is enough (see Theorem 3.2.3) that the following
two conditions are met:
1
(i) The functions gn (x) = x are continuous on ]a, +∞[.
n
∞
1
(ii) The series x
is uniformly convergent to g on ]a, +∞[.
n=1
n
238 Series of Functions

The functions gn (x) = e−x log(n) are continuous, since the exponential is continuous on R. Let us look at
the second condition. Let a > 1 be a real constant. The inequality
 
 1 
.  ≤ 1 , ∀x ∈ [a, +∞[, ∀n ∈ N,
nx  na

∞
1
combined with the fact that the series is convergent, is sufficient to conclude, by Weierstrass’ test,
n=1
na
∞
1
the uniform convergence of the series x
on [a, +∞[. As the series converges to g, this function is
n=1
n
continuous on [a, +∞[. Since a is an arbitrary constant greater than 1, we also have g continuous on
]1, +∞[.
∞
 1
4. The terms of the series √ are continuous functions on [1, 2]. As
n=1 (nx + 1) n3 + 1
 
 
 1  1 1 1 1 1
. √ = ·√ ≤ ·√ ≤ √ , ∀x ∈ [1, 2], ∀n ∈ N,
 (nx + 1) n3 + 1  |nx + 1| n3 + 1 n+1 n3 + 1 n3

∞
 1 3
and the series of positive terms √ is convergent because it is a p-series with p = , by Weierstrass’ test,
n=1 n3 2
∞
 1
the series √ converges uniformly to f , on [1, 2]. It follows from Theorem 3.2.3 that f is
n=1 (nx + 1) n3 + 1
continuous on [1, 2]; by Theorem 3.2.4, the series is integrable term by term on this interval and
 2  2 ∞
 ∞  2
1 1 1
f (x) dx = √ dx = √ dx
1 1 n=1 (nx + 1) n3 + 1 n=1 n 3 + 1 1 nx + 1

∞
  2 ∞
1 log(nx + 1) log(2n + 1) − log(n + 1)
. = √ = √
n=1 n3 +1 n 1 n=1 n n3 + 1
∞
  
1 2n + 1
= √ log .
n=1 n n3 + 1 n+1

∞
cos(nx)
5. The terms of the series are functions of domain R. The function f is the sum of the series; therefore,
n=1
n2
its domain is the subset of R where the series converges. Since we have
 
 cos(nx) 
.  ≤ 1 , ∀x ∈ R, ∀n ∈ N,
n2  n2

∞
1
because | cos(nx)| ≤ 1, ∀x ∈ R, ∀n ∈ N, and the series of positive terms 2
is convergent as it is a p-
n=1
n
∞
cos(nx)
series with p = 2, by Weierstrass’ test, the series converges uniformly to f , on R. Consequently, by
n=1
n2
cos(nx)
Theorem 3.2.3, given that the functions are continuous on R, f is continuous on R; by Theorem 3.2.4,
n2
3.6. Solved Exercises 239
 π
the series is integrable term by term on every closed bounded interval of R. In particular, f is integrable on 0, ,
2
and
 π/2  π/2 ∞ ∞  π/2
cos(nx) 1
f (x) dx = dx = cos(nx) dx
0 0 n=1
n2 n=1
n2 0
.
∞  π ∞ ∞ ∞
1 sin(nx) 2 sin( nπ
2
) (−1)n−1 (−1)n
= = = = ,
n=1
n 2 n 0 n=1
n 3
n=1
(2n − 1) 3
n=0
(2n + 1)3

because
⎧
 nπ ⎪
⎨0, if n is even
. sin = −1, if n = 3, 7, 11, . . .
2 ⎪
⎩
1, if n = 1, 5, 9, . . .

∞
e−nx
6. The terms of the series f (x) = are functions of domain R. As
n=0
n4 + 5

 −nx 
 e  1 1
.  ≤ ≤ 4, ∀x ∈ [0, 1], ∀n ∈ N0 ,
n + 5
4 n4 + 5 n

∞
1
since |e−nx | = e−nx ≤ e0 = 1, ∀x ∈ [0, 1], ∀n ∈ N0 , and the series of positive terms 4
is convergent
n=1
n
∞
e−nx
as it is a p-series with p = 4. By Weierstrass’ test, the series 4+5
converges uniformly to f , on [0, 1].
n=0
n
e−nx
Therefore, by Theorem 3.2.3 since the functions fn (x) = are continuous on [0, 1], f is continuous on
n4 + 5
[0, 1]. Theorem 3.2.4 shows that the series is integrable term by term on this interval. We have

  ∞
  ∞  1
1 1 1 e−nx 1  1
f (x) dx = + 4
dx = + 4
e−nx dx
0 0 5 n=1 n + 5 5 n=1 n + 5 0
.
∞
  −nx 1 ∞

1 1 e 1 1 − e−n
= + 4
− = + .
5 n=1 n + 5 n 0 5 n=1 n (n4 + 5)

7. a) The series of functions defined, for x ≥ 0, by

∞ 
 n+ x
1 n
. x+
n=1
n

is of positive terms. Let us study this series by applying Cauchy’s Root Test:
 
 n+ x    x
n 1 n n 1 n 1 n
lim x+ = lim x+ x+
n n n
. 
   x    x
1 n 1 n 1 1 n2
= lim x + x+ = lim x + x+ = x.
n n n n
240 Series of Functions

By Cauchy’s Root Test, the series converges if x < 1 and diverges if x > 1. It remains to see what happens
∞   1
1 n+ n
if x = 1. In this case, we obtain the numerical series 1+ , which is divergent because its
n=1
n
general term is not a null sequence. In fact,
  1    1
1 n+ n 1 n 1 n
. lim 1+ = lim 1 + 1+ = e.
n n n

b) Let 0 < α < 1 and x ∈ [0, α].

  x   x   α
 1 n+ n  1 n+ n 1 n+ n

. x + = x+ ≤ α+ , ∀n ∈ N.
 n  n n

∞   α
1 n+ n
As shown in item a), the numerical series α+ is convergent. By Weierstrass’ test, the original
n=1
n
series converges uniformly on [0, α].
∞

8. We will use Weierstrass’ test to prove that the series of functions an fn (x) converges uniformly on R. Since
n=1
1
|fn (x)| ≤ , ∀n ∈ N, ∀ x ∈ R, and (an ) is a sequence of positive terms, we have
n
an
.|an fn (x)| ≤ , ∀n ∈ N, ∀ x ∈ R.
n
∞
an
We prove that the series of positive terms is convergent by applying D’Alembert’s test:
n=1
n

an+1
n+1 n an+1 n an+1 1
. lim a
n
= lim = lim · =
(n + 1) an n+1 an 2
n
an+1 1
because, by hypothesis, lim = . As the limit is less than 1, the series is convergent. By Weierstrass’ test,
an 2
∞

the series of functions an fn (x) converges uniformly on R.
n=1

9. Since the functions fn : [0, 1] → R are continuous, if the convergence was uniform, we would have, by Theo-
rem 3.2.4,
 1
∞ ∞  1
. fn (x) dx = fn (x) dx.
0 n=1 n=1 0

But  
∞
1  1
. fn (x) dx = f (x) dx = 0
0 n=1 0

and
∞ 
 1 ∞ 
 
1 1
. fn (x) dx = − = 1.
n=1 0 n=1
n n+1

Therefore, the series of functions is not uniformly convergent on [0, 1].

3.6. Solved Exercises 241

∞
arccot(nx)
10. a) The terms of the series √ are functions with domain R. The function f is the sum of the
n=1 n5
series; therefore, its domain is the subset of R where the series converges. The terms of the series satisfy
 
 arccot(nx) 
.  √  ≤ √π , ∀x ∈ R, ∀n ∈ N, (3.14)

n5 n5

∞
 1
since |arccot(nx)| ≤ π, ∀x ∈ R. The series of positive terms √ is convergent because it is a
n=1 n5
∞
5 π
p-series with p = . Therefore, the series √ is also convergent. By inequality (3.14), the series
2 n=1 n5
∞   ∞
 arccot(nx)  arccot(nx)
 √  converges, ∀x ∈ R. Then the series √ is convergent, ∀x ∈ R, that is, the
 
n=1 n5 n=1 n5
domain of f is R.

b) To conclude that f is continuous on R, it is sufficient, according to Theorem 3.2.3, that the following two
conditions are satisfied:
arccot(nx)
(i) The functions fn (x) = √ are continuous on R.
n5
∞
arccot(nx)
(ii) The series √ is uniformly convergent to f on R.
n=1 n5

The functions hn (x) = arccot(nx) are continuous on R, which implies the first condition. The second condi-
∞
π
tion is a consequence of the previous item: Inequality (3.14) and the fact that the series √ is conver-
n=1 n5
∞
arccot(nx)
gent are sufficient, by Weierstrass’ test, to conclude the uniform convergence of the series √ .
n=1 n5
As the series converges to f , this function is continuous.

c) To conclude that f is differentiable on R, it is sufficient, by Corollary 2 of Theorem 3.2.4, that the following
conditions are fulfilled:

(i) The functions fn (x) are continuous on R.

∞
arccot(nx)
(ii) The series √ is pointwise convergent to f on R.
n=1 n5
∞ 
  ∞
arccot(nx)  n
(iii) The series √ = − √ is uniformly convergent on R.
n=1 n5 n=1 (1 + n2 x2 ) n5
  n
The first condition is satisfied because the functions arccot(nx) = − are continuous on R. We
1 + n 2 x2
proved the second condition in the previous item. The third is again an application of Weierstrass’ test. We
have the inequality
 
 
 n  n 1
. − √ ≤ √ = √ , ∀x ∈ R, ∀n ∈ N,
 (1 + n2 x2 ) n5  n5 n3

∞
 ∞  
1 3 arccot(nx) 
and the series √ is convergent because it is a p-series with p = , so the series √
n=1 n3 2 n=1 n5
is uniformly convergent. We can conclude that the series converges to f  , that is,
242 Series of Functions

∞   ∞
 arccot(nx)  n
.f (x) = √ = − √ ,
n 5 (1 + n 2 x2 ) n 5
n=1 n=1

which means that f is differentiable on R.

∞

11. If the series an cos(nx) converges uniformly on R, then it converges pointwise on R. In particular, the series
n=0
∞

is convergent at x = 0. At this point, we obtain the numerical series an , which is convergent.
n=0
∞

Conversely, suppose that the series an is convergent. We have
n=0

. |an cos(nx)| ≤ |an | = an , ∀x ∈ R, ∀n ∈ N.

∞

By Weierstrass’ test the series an cos(nx) is uniformly convergent on R.
n=0


+∞ 
n
12. We know that the series (bn − bn+1 ) converges, that is, the sequence Sn = (bk − bk+1 ) = b1 − bn+1
n=1 k=1
converges. Therefore, there exists b = lim bn = b1 − lim Sn , so the sequence (bn ) is bounded, that is, there exists
M > 0 such that |bn | ≤ M, ∀n ∈ N. Based on this fact, we have the following inequalities:

.|bn fn (x)| = |bn | |fn (x)| ≤ |bn | an ≤ M an , ∀x ∈ R, ∀n ∈ N.

∞

Thus, by Weierstrass’ test, the series bn fn (x) is uniformly convergent on R, defining a function f of domain
n=1
R. Since, by hypothesis, the functions fn are continuous on R, ∀n ∈ N, the function f is continuous on its domain.

∞

13. a) Consider the series (−1)n (1 − x) xn , x ∈ [0, 1].
n=0

– If x = 1, all terms are zero. Therefore the series converges.

∞

– If x = 1, as the series (1 − x) (−x)n is geometric with ratio r = −x, it converges because x ∈ [0, 1[.
n=0
In this case,
∞
 1 1−x
. (1 − x) (−x)n = (1 − x) · = .
n=0
1 − (−x) 1+x

∞
 1−x
Then the series (−1)n (1 − x) xn converges pointwise to the function f (x) = on [0, 1]. This
n=0
1+x
convergence will be uniform if
. lim sup |f (x) − Sn (x)| = 0,
n→+∞ x∈[0,1]

where Sn denotes the sequence of partial sums of the original series.

– If x = 1, |f (1) − Sn (1)| = 0.
3.6. Solved Exercises 243

1 − (−x)n+1 1−x
– If x = 1, Sn (x) = (1 − x) · = · (1 − (−x)n+1 ), and therefore,
1 − (−x) 1+x

 
1 − x 1−x  
|f (x) − Sn (x)| =  − · 1 − (−x)n+1 
1+x 1+x
  
 1−x   
=  · 1 − 1 − (−x)n+1 
.
1+x
 
1 − x  1−x
=  · (−x)n+1  = · xn+1 .
1+x 1+x
1−x
Let g(x) = |f (x) − Sn (x)| = · xn+1 .
1+x
The function g is continuous on [0, 1]; therefore, by Weierstrass’ Theorem, it has a maximum and a minimum
on this interval. As g is nonnegative and g(0) = g(1) = 0, the maximum is attained at some point within
the interior of the interval. The function g is differentiable on ]0, 1[, so this point corresponds to a zero of
the first derivative.
 
 1−x xn  
.g (x) = · xn+1 = − 2
· (n + 1)x2 + 2x − (n + 1) ,
1+x (1 + x)
and on the interval ]0, 1[,

 −1 + 1 + (n + 1)2
.g (x) = 0 ⇔ (n + 1)x2 + 2x − (n + 1) = 0 ⇔ x = .
n+1
Then 
   −1 +1 + (n + 1)2
1−
−1 + 1 + (n + 1)2 n+1
. sup |f (x) − Sn (x)| = g ≤  ,
x∈[0,1] n+1 −1 + 1 + (n + 1)2
1+
n+1
and, as 
−1 +1 + (n + 1)2
1−
n+1
. lim  = 0,
n→+∞ −1 + 1 + (n + 1)2
1+
n+1
we can conclude that
. lim sup |f (x) − Sn (x)| = 0,
n→+∞ x∈[0,1]

∞

that is, the series (−1)n (1 − x) xn converges uniformly to the function f , on the interval [0, 1].
n=0
∞

b) Let us consider the series (1 − x) xn , x ∈ [0, 1].
n=0
– If x = 1, all terms are zero. Hence, the series converges.
∞

– If x = 1, as the series (1 − x) xn is geometric with ratio r = x, the series converges because
n=0
x ∈ [0, 1[. In this case,
∞
 1
. (1 − x) xn = (1 − x) · = 1.
n=0
1−x
244 Series of Functions

∞

Then, the series (1 − x) xn , x ∈ [0, 1], converges pointwise to the function
n=0

0, if x = 1
.f (x) =
1, if x ∈ [0, 1[.

Let fn (x) = (1 − x)xn . The functions fn are continuous on [0, 1], ∀n ∈ N0 , because they are polynomial
functions. If the series was uniformly convergent on the interval [0, 1], f would be continuous on this interval.
∞
Since f is not continuous at x = 1 because lim f (x) = 1 and f (1) = 0, the series (1 − x) xn does
x→1−
n=0
not converge uniformly to f on [0, 1].
∞
π n xn πn
14. a) Let us study the power series of x, 2
. Setting an = , the radius of convergence of
n=2
n log (n) n log2 (n)
this series is, by Corollary 1 of Theorem 3.3.1
 
 πn 
   
 an   2  n + 1 log2 (n + 1)
 n log (n)  1
.r = lim   = lim 
  = lim · = ,
an+1  π n+1  π n log 2
(n) π
 
 (n + 1) log2 (n + 1) 
 
1 1
which implies, by the same theorem, that the series is absolutely convergent if x ∈ − , and divergent
    π π
1 1
if x ∈ −∞, − ∪ , +∞ .
π π
1
At x = ± , we have to study the corresponding numerical series.
π
∞  n ∞
1 πn 1 1
– If x = , we obtain the series 2
= 2
. Let us study the convergence of
π n=2
n log (n) π n=2
n log (n)
this series using the Integral Test.
1
Let f (x) = . We have:
x log2 (x)
(i) f (x) > 0, ∀x ≥ 2.
(ii) f is continuous on [2, +∞[.
(iii) f is decreasing on [2, +∞[ because x log2 (x) is positive and increasing.
∞  +∞
1 1
So the numerical series 2
converges if and only if the improper integral 2
dx
n=2
n log (n) 2 x log (x)
converges. We have

   x
+∞ 1 1 x 1 −2
dx = lim dt = lim log(t) dt
. 2 x log2 (x) x→+∞ 2 t log2 (t) x→+∞ 2 t
 x  
1 1 1 1
= lim − = lim − + = ,
x→+∞ log(t) 2 x→+∞ log(x) log(2) log(2)

that is, the improper integral is convergent, so the same happens to the series. As it is a series of
positive terms, it converges absolutely.
3.6. Solved Exercises 245

∞   ∞
1 πn 1 n 1
– If x = − , we obtain the series 2
− = (−1)n 2
, which is alternating.
π n=2
n log (n) π n=2
n log (n)
As we just saw, the series of modules is convergent. Then the series is absolutely convergent.
   
1 1
Conclusion: The original series diverges if x ∈ −∞, − ∪ , +∞ and is absolutely convergent if
  π π
1 1
x∈ − , .
π π
 2
log2 (n + 1) log(n + 1)
Remark: To calculate the lim = lim . we proceed as follows: Consider the func-
log2 (n) log(n)
log(x + 1) log(x + 1)
tion g(x) = ; we use L’Hôpital’s Rule applied to the calculation of the limit lim ,
log(x) x→+∞ log(x)
∞
where the indeterminate form of type arises:
∞

  1
log(x + 1) x+1 x
. lim   = lim = lim = 1.
x→+∞ log(x) x→+∞ 1 x→+∞ x + 1

log(x + 1) log2 (n + 1)
Therefore, lim = 1, which implies that lim = 1.
x→+∞ log(x) log2 (n)
∞
2n nn 2n nn
b) Let us study the power series of x, n+1
xn . Setting an = , the radius of conver-
n=1
(n + 1) (n + 1)n+1
gence is by Corollary 1 of Theorem 3.3.1
 
 2n nn 
   
 an    2n nn (n + 2)n+2
   (n + 1) n+1

r = lim   = lim  n+1  = lim
an+1 2 (n + 1)n+1  (n + 1)n+1 2n+1 (n + 1)n+1
 
.  (n + 2)n+2 
 n  
1 n n + 2 n+2 1 1 1
= lim · = · ·e= .
2 n+1 n+1 2 e 2
     
1 1 1 1
Then, the series is absolutely convergent if x ∈ − , and divergent if x ∈ −∞, − ∪ , +∞ .
2 2 2 2
Let us study the convergence of the numerical series corresponding to the endpoints of the interval of
convergence.

∞  n ∞
1 2n nn 1 nn
– If x = , we obtain the series of positive terms n+1
· = .
2 n=1
(n + 1) 2 n=1
(n + 1)n+1
We will study this series by comparing it with the harmonic series, which is divergent. The limit
nn
 n+1  n
(n + 1)n+1 n n n 1
. lim = lim = lim · =
1 n+1 n+1 n+1 e
n
is finite and different from zero; therefore, by Corollary 2 of the General Comparison Test, the series is
divergent.
246 Series of Functions

∞   ∞
1 2n nn 1 n nn
– If x = − , we obtain the series n+1
· − = (−1)n , which is alter-
2 n=1
(n + 1) 2 n=1
(n + 1)n+1
n n
nating, since > 0, ∀n ∈ N. We proved in the previous case that the series of modules is
(n + 1)n+1
divergent. Therefore, we need to verify whether the following conditions hold to apply Leibniz’s test:
 n
nn n 1 1
(i) lim = lim · = × 0 = 0.
(n + 1)n+1 n+1 n+1 e
nn
(ii) > 0, ∀n ≥ 1.
(n + 1)n+1
 n
nn n 1
(iii) The sequence of general term n+1
= · is decreasing, as it is the product
(n + 1) n+1 n+1
of two decreasing positive sequences.
Therefore, the series is convergent. It is conditionally convergent since the series of modules diverges.
 
1 1
Conclusion: The power series of x is absolutely convergent if x ∈ − , , conditionally convergent if
    2 2
1 1 1
x = − , and divergent if x ∈ −∞, − ∪ , +∞ .
2 2 2
∞
 3n log(n)
c) By Corollary 1 of Theorem 3.3.1, the power series of x, an xn , with an = , has a radius of
n=1
7n n2
convergence
 
 3n log(n) 
   
 an    7(n + 1)2 log(n)
 7n n2  7
.r = lim   = lim 
  = lim = .
a  3n+1 log(n + 1)  3n2 log(n + 1) 3
n+1  
 7n+1 (n + 1)2 

     
7 7 7 7
Then, the series converges absolutely if x ∈ − , and diverges if x ∈ −∞, − ∪ , +∞ .
3 3 3 3

At the endpoints of the convergence interval, we obtain two numerical series.

∞   ∞
7 3n log(n) 7 n log(n)
– If x = − , we have the series n 2
· − = (−1)n which is alternating. Let us
3 n=1
7 n 3 n=1
n2
 ∞ ∞
log(n) 1
start by studying the series of modules, 2
, by comparison with the series 3/2
, which is
n=1
n n=1
n
3
convergent because it is a p-series with p = . Since
2

log(n)
. lim
n2 = lim log(n) = 0,
1 n1/2
n3/2

∞
log(n)
by Corollary 3 of the General Comparison Test, the series is convergent. As it is the series
n=1
n2
∞
 log(n)
of modules of (−1)n , this series is absolutely convergent.
n=1
n2
3.6. Solved Exercises 247

∞
 log(n)
7
– If x = , we obtain the series , which we proved in the previous case to be convergent. As
3 n=1
n2
it is of positive terms, it is absolutely convergent.
     
7 7 7 7
Conclusion: The series converges absolutely if x ∈ − , and diverges if x ∈ −∞, − ∪ , +∞ .
3 3 3 3
∞
 2 × 5 × · · · × (3n − 1) n 2 × 5 × · · · × (3n − 1)
d) Let us study the power series of x, x . Setting an = ,
n=1
(n + 2)! (n + 2)!
the radius of convergence is by Corollary 1 of Theorem 3.3.1
 
 2 × 5 × · · · × (3n − 1) 
   
 an   
.r = lim   = lim  (n + 2)!  = lim n + 3 = 1 .
  2 × 5 × · · · × (3n − 1)(3n + 2) 
an+1   3n + 2 3
 
(n + 3)!
     
1 1 1 1
Then, the series converges absolutely if x ∈ − , and diverges if x ∈ −∞, − ∪ , +∞ .
3 3 3 3
Let us study the behavior of the numerical series at the endpoints of the convergence interval.
1
– If x = − , we obtain the alternating series
3

∞   ∞
2 × 5 × · · · × (3n − 1) 1 n 2 × 5 × · · · × (3n − 1)
. · − = (−1)n .
n=1
(n + 2)! 3 n=1
3n (n + 2)!

Let us begin by studying the series of modules

∞ ∞
2 × 5 × · · · × (3n − 1)
. = bn ,
n=1
3n (n + 2)! n=1

using Raabe’s test,

⎛ ⎞
2 × 5 × · · · × (3n − 1)
 
bn ⎜ n
3 (n + 2)! ⎟
lim n − 1 = lim n ⎜ ⎟
⎝ 2 × 5 × · · · × (3n − 1)(3n + 2) − 1⎠
bn+1
.
3n+1 (n + 3)!
 
3n + 9 7n 7
= lim n − 1 = lim = > 1,
3n + 2 3n + 2 3

∞

showing the series is convergent. Then, (−1)n bn is absolutely convergent.
n=1
∞
1 2 × 5 × · · · × (3n − 1)
– If x = , we obtain , which is the series of modules of the previous series
3 n=1
3n (n + 2)!
that we have seen to be convergent. As it is of positive terms, it is absolutely convergent.
     
1 1 1 1
Conclusion: The series converges absolutely if x ∈ − , and diverges if x ∈ −∞, − ∪ , +∞ .
3 3 3 3
248 Series of Functions

∞ ∞
log(n) n log(n) n log(n)
e) Let us study the power series of x, x = x . Setting an = , the radius of
n=1
n n=2
n n
convergence of this series is by Corollary 1 of Theorem 3.3.1
 
 log(n) 
   
 an   
.r = lim   = lim  n  = lim n + 1 · log(n) = 1;
  log(n + 1) 
an+1   n log(n + 1)
 
n+1

therefore, the series converges absolutely if x ∈ ] − 1, 1[ and diverges if x ∈ ] − ∞, −1[ ∪ ]1, +∞[. Regarding
the endpoints of the convergence interval, we need to study the respective numerical series.
∞
 log(n)
– If x = −1, we obtain the alternating series (−1)n . Let us start by studying the series of
n=1
n
∞
log(n)
modules, , using the Integral Test.
n=1
n
log(x)
Let f (x) = , x ≥ 1:
x
(i) If x > 1, then log(x) > 0, therefore, f (x) > 0, ∀x > 1.
(ii) f is continuous, ∀x ≥ 1.
(iii) We can study the monotonicity of f by analyzing its derivative. We have

1
x· − log(x) 1 − log(x)
 x
.f (x) = = ,
x2 x2

and thus, f  (x) < 0, ∀x > e; therefore, f is decreasing on ]e, +∞[.

The conditions of the Integral Test are fulfilled on the interval [3, +∞[. Then the numerical series
∞  +∞
log(n) log(x)
and the improper integral dx are both convergent or both divergent. Since
n=3
n 3 x
 +∞  x  x
log(x) log(t) 1
dx = lim dt = lim log(t) dt
3 x x→+∞ 3 t x→+∞ 3 t
 x
log2 (t)
. = lim
x→+∞ 2 3
 2 
log (x) log2 (3)
= lim − = +∞,
x→+∞ 2 2

the improper integral is divergent; the same happens to the series.

Because the series of modules is divergent, let us apply Leibniz’s test to the alternating series:
log(n) log(x) ∞
(i) lim = 0 (just note that lim is an indeterminate form of type that can be
n x→+∞
 x  ∞
log(x) 1
lifted using L’Hôpital’s Rule: lim = lim = 0).
x→+∞ (x) x→+∞ x
log(n)
(ii) > 0, ∀n > 1.
n  
log(x) log(n)
(iii) We saw that the function f (x) = is decreasing on ]e, +∞[, which implies that
x n
is decreasing for n ≥ 3.
3.6. Solved Exercises 249

Then, the series is convergent. As the series of modules is divergent, the original series is conditionally
convergent.
∞
 log(n)
– If x = 1, we obtain the series which, as we saw before, is divergent.
n=1
n
Conclusion: The power series is absolutely convergent if x ∈ ] − 1, 1[, conditionally convergent if x = −1,
and divergent if x ∈ ] − ∞, −1[ ∪ [1, +∞[.

∞ ∞
log(n) log(n)
Remark: The series and are both divergent because they only differ in a finite
n=1
n n=3
n
number of terms.
∞
 3 n2 3 n2
f) Let us study the power series of x, xn . Setting an = n 3 , the radius of conver-
n=0
4n 3
(n + 2) 4 (n + 2)
gence of this series is, by Corollary 1 of Theorem 3.3.1,
 
 3 n2 
     
 an    3n2 4n+1 (n + 1)3 + 2
 = lim  
n 3
4 (n + 2)
r = lim   = lim
an+1   3 (n + 1)2  3 (n + 1)2 4n (n3 + 2)
 
.  4n+1 (n + 1)3 + 2 
 2
n (n + 1)3 + 2
= lim 4 = 4.
n+1 n3 + 2

The series converges absolutely if x ∈ ] − 4, 4[ and diverges if x ∈ ] − ∞, −4[ ∪ ]4, +∞[.

Regarding the endpoints of the convergence interval, we have to study the respective numerical series.
∞ ∞
3 n2 3 n2
– If x = 4, we obtain the series n 3
4n = . We will study this series by comparing
n=0
4 (n + 2) n=0
n3 + 2
it with the harmonic series, which is divergent. The limit

3 n2
n3 + 2 3 n3
. lim = lim 3 =3
1 n +2
n
is finite and different from zero; therefore, the series is divergent by Corollary 2 of the General Com-
parison Test.
∞
 ∞

3 n2 3 n2
– If x = −4, we obtain the series n (n3 + 2)
(−4) n
= (−1)n 3 , which is alternating since
n=0
4 n=0
n +2
3 n2
> 0, ∀n ∈ N. The series of modules is the series we studied for the case x = 4, which we
n3 + 2
proved to be divergent. Being an alternating series, we apply Leibniz’s test:
3 n2
(i) lim 3 = 0.
n +2
3n 2
(ii) 3 > 0, ∀n ≥ 1.
n +2
 
3 n2
(iii) The sequence is decreasing, because
n3 + 2

3 (n + 1)2 3 n2 −n4 − 2n3 − n2 + 4n + 2

. − 3 =3 < 0, ∀n ≥ 2.
(n + 1)3 + 2 n +2 ((n + 1)3 + 2)(n3 + 2)
250 Series of Functions

Then, the series is convergent. As the series of modules is divergent, the series is conditionally conver-
gent.

Conclusion: The power series is absolutely convergent if x ∈ ] − 4, 4[, conditionally convergent if x = −4,
and divergent if x ∈ ] − ∞, −4[ ∪ [4, +∞[.

∞
1 × 4 × · · · × n2 (n!)2 (n!)2 n
g) Let an = = . By Corollary 1 of Theorem 3.3.1, the power series of x, x ,
(2 n)! (2 n)! n=1
(2 n)!
has radius of convergence
 
 (n!)2 
   
 an   
.r = lim   = lim   (2 n)!  
 = lim
(2n + 2)(2n + 1)
= 4.
a   (n + 1)! 2  (n + 1)2
n+1  
 
(2 n + 2)!

The series converges absolutely if x ∈ ] − 4, 4[ and diverges if x ∈ ] − ∞, −4[ ∪ ]4, +∞[.

Let us study the numerical series corresponding to the endpoints of the interval.
∞ ∞
(n!)2 n
– If x = 4, we obtain the series 4 = bn . Let us study this series using D’Alembert’s test1 :
n=1
(2 n)! n=1

 2
(n + 1)!
4n+1
bn+1 (2 n + 2)! 2n + 2
. lim = lim = lim = 1.
bn (n!)2 n 2n + 1
4
(2 n)!

Since the limit is 1, but for values greater than 1, taking into account the Note after D’Alembert’s
test, we can conclude that the series is divergent.
∞ ∞
(n!)2 (n!)2 n
– If x = −4, we obtain the alternating series (−4)n = (−1)n 4 . The series of
n=1
(2 n)! n=1
(2 n)!
∞
(n!)2 n
modules 4 is divergent. Since we establish the divergence of the series of modules by
n=1
(2 n)!
D’Alembert’s test, we can affirm that the alternating series is divergent (refer to the Note on page 105
for further details).

Conclusion: The power series converges absolutely if x ∈ ] − 4, 4[ and diverges if x ∈ ] − ∞, −4] ∪ [4, +∞[.
∞
  n   n
1 1
h) Let us study the power series of x, nn
sin 2
xn . Let an = nn sin 2
. The radius of
n=1
n n
convergence of this series is, by Theorem 3.3.1,

1 1 1
.r =  =    n =   = +∞.
lim n |an | 1 1
lim n nn sin lim n sin 2
n 2 n

Moreover, the series is absolutely convergent on R.

1 In general, this test is not useful for studying the endpoints of a power series because the limit is always 1.
3.6. Solved Exercises 251

∞
 1 1
i) Let us study the power series of x,   xn . Let an =   . The
2n n arctan(n) − π 2n n arctan(n) − π
n=1 2 2
radius of convergence of this series is, by Corollary 1 of Theorem 3.3.1,
 
 
1
   
   
 an   2n n arctan(n) − π2  2(n + 1) π2 − arctan(n + 1)

.r = lim   = lim   = lim   = 2.
an+1  
 1 
 n π2 − arctan(n)
 2n+1 (n + 1) arctan(n + 1) − π  
2

Then, the series is absolutely convergent if x ∈ ] − 2, 2[ and diverges if x ∈ ] − ∞, −2[ ∪ ]2, +∞[.
At the endpoints of the convergence interval, we obtain two numerical series.
∞ ∞
1 1
– If x = 2, we have the series   2n =   . Let us calculate
2n n arctan(n) − π
n arctan(n) − π2
n=1 2 n=1
the limit of its general term:
1
. lim   = −1.
n arctan(n) − π2
Since the general term of the series is not a null sequence, by Theorem 2.2.1, the series is divergent.
∞
 ∞
(−2)n 1
– If x = −2, we obtain the series   = (−1)n+1  π  , which is
2 n n arctan(n) − π
n − arctan(n)
n=1 2 n=1 2
1
alternating because an =  π  > 0, ∀n ∈ N. The limit of the general term does not
n 2 − arctan(n)
exist, as the subsequence of even order terms has limit −1 and the subsequence of odd order terms
has limit 1. Then, the series is divergent.

Conclusion: The power series converges absolutely if x ∈ ] − 2, 2[ and diverges if x ∈ ] − ∞, −2] ∪ [2, +∞[.
∞
 ∞

3 × 6 × · · · × 3n n 3n n! 3n n!
j) Consider the power series x = xn . If an = , then the radius
n=1
(n + 3)! n=1
(n + 3)! (n + 3)!
of convergence of this series is, by Corollary 1 of Theorem 3.3.1,
3n n!
 
 an  (n + 3)! n+4 1
.r = lim   = lim
 = lim = ;
a
n+1 3n+1 (n + 1)! 3(n + 1) 3
(n + 4)!
     
1 1 1 1
therefore, the series converges absolutely if x ∈ − , and diverges if x ∈ −∞, − ∪ , +∞ .
3 3 3 3
Let us study the numerical series at the endpoints of the convergence interval.
∞ ∞

1 n! 1
– If x = , we obtain the series = . Let us study this series by
3 n=1
(n + 3)! n=1
(n + 3)(n + 2)(n + 1)
∞
1
comparing it with 3
which we know to be convergent (p-series with p = 3):
n=1
n

1
(n + 3)(n + 2)(n + 1) n3
. lim = lim = 1.
1 (n + 3)(n + 2)(n + 1)
n 3

Since this value is finite and different from zero, the series is convergent by Corollary 2 of the General
Comparison Test. As it is a series of positive terms, it is absolutely convergent.
252 Series of Functions

∞
1 n!
– If x = − , we obtain the alternating series (−1)n . As the series of modules is convergent,
3 n=1
(n + 3)!
the alternating series is absolutely convergent.
     
1 1 1 1
Conclusion: The series converges absolutely if x ∈ − , and diverges if x ∈ −∞, − ∪ , +∞ .
3 3 3 3
∞
 n+1 n+1
k) Let us study the power series of x, (−1)n √ xn . Let an = (−1)n √ . The radius of
e 2n + 5 e 2n + 5
n=0
convergence of this series is, by Corollary 1 of Theorem 3.3.1,
 
 n+1  
   (−1)n √  √
 an   e2n + 5  (n + 1) e2n+2 + 5 n+1 e2n+2 + 5
  
.r = lim 
an+1 
= lim 
n + 2  = lim (n + 2)√e2n + 5 = lim n + 2 · e2n + 5
= e;
 (−1) n+1
√ 
 2n+2 
e +5

then, the series is absolutely convergent if x ∈ ] − e, e[ and divergent if x ∈ ] − ∞, −e[ ∪ ]e, +∞[.
At the endpoints of the convergence interval, we obtain two numerical series that we will proceed to study.
∞
 ∞
n+1 (n + 1) en
– If x = −e, we obtain the series (−1)n √ (−e)n = √ . As
n=0 e2n +5 n=0 e2n + 5

(n + 1) en
. lim √ = +∞,
e2n + 5

the general term is not a null sequence; therefore, by Theorem 2.2.1, the series diverges.
∞ ∞
n+1 (n + 1) en
– If x = e, we obtain the series (−1)n √ en = (−1)n √ . As the limit of the
2n
e +5 e2n + 5
n=0 n=0
general term does not exist, the series is divergent.

Conclusion: The power series converges absolutely if x ∈ ] − e, e[ and diverges if x ∈ ] − ∞, −e] ∪ [e, +∞[.

∞
1 × 3 × 5 × · · · × (2n − 1) 1 n
l) Let us study the power series of x, · x . Setting
n=1
2 × 4 × 6 × · · · × 2n n

1 × 3 × 5 × · · · × (2n − 1) 1
.an = · ,
2 × 4 × 6 × · · · × 2n n
the radius of convergence of this series is, by Corollary 1 of Theorem 3.3.1,
 
 1 × 3 × 5 × · · · × (2n − 1) 1 
   · 
 an   × × × · · · × 
.r = lim   = lim  2 4 6 2n n  = lim (n + 1)(2n + 2) = 1.
  1 × 3 × 5 × · · · × (2n − 1)(2n + 1) 1 
an+1  ·  n(2n + 1)
 
2 × 4 × 6 × · · · × 2n(2n + 2) n+1

The series converges absolutely if x ∈ ] − 1, 1[ and diverges if x ∈ ] − ∞, −1[ ∪ ]1, +∞[.

Let us study the numerical series obtained at the endpoints of the convergence interval.
– If x = −1, we obtain the alternating series
∞
 1 × 3 × 5 × · · · × (2n − 1) 1
. (−1)n · .
n=1
2 × 4 × 6 × · · · × 2n n
3.6. Solved Exercises 253

Let us start by studying the series of modules,

∞ ∞
1 × 3 × 5 × · · · × (2n − 1) 1
. · = an ,
n=1
2 × 4 × 6 × · · · × 2n n n=1

using Raabe’s test

   
an (n + 1)(2n + 2) 3n + 2 3
. lim n − 1 = lim n − 1 = lim = > 1,
an+1 n(2n + 1) 2n + 1 2
∞

which shows that the series is convergent. As it is the series of modules of (−1)n an , this series is
n=1
absolutely convergent.
∞

1 × 3 × 5 × · · · × (2n − 1) 1
– If x = 1, we obtain the series · , which we have seen to be convergent.
2 × 4 × 6 × · · · × 2n
n=1
n
As it is of positive terms, it is absolutely convergent.

Conclusion: The power series converges absolutely if x ∈ [−1, 1] and diverges if x ∈ ] − ∞, −1[ ∪ ]1, +∞[.
∞ 
 2n
n
15. a) Let us study the power series of x − 1, (x − 1)n . Let y = x − 1, and consider the
n=1
2n + 1
∞ 
 2n  2n
n n
series y n . Setting an = , the radius of convergence of this series is, by
n=1
2n +1 2n + 1
Theorem 3.3.1,
1 1 1 1
.r =  =  2n =  2 = = 4.
lim n |an | n n n2
lim n lim lim
2n + 1 2n + 1 4n2 + 4n + 1

The series converges absolutely if y ∈ ] − 4, 4[ and diverges if y ∈ ] − ∞, −4[ ∪ ]4, +∞[. As y = x − 1, we

∞  2n
n
can conclude that the series (x − 1)n is absolutely convergent if x ∈ ] − 3, 5[ and divergent
n=1
2n + 1
if x ∈ ] − ∞, −3[ ∪ ]5, +∞[.
At the endpoints of the convergence interval, we have to study the corresponding numerical series.
– If x = −3, we obtain the series
∞ 
 2n ∞
  2n ∞
  2n
n n 2n
. (−4)n = (−1)n 4n = (−1)n .
n=1
2n + 1 n=1
2n + 1 n=1
2n + 1

As  2n
2n 1
. lim = ,
2n + 1 e
the general term does not have limit; therefore, by Theorem 2.2.1, the series is divergent.
– If x = 5, we obtain the series
∞ 
 2n ∞ 
 2n
n 2n
. 4n = ,
n=1
2n + 1 n=1
2n + 1

whose general term does not tend to zero, as we saw earlier; therefore, the series is divergent.
254 Series of Functions

In summary, the power series of x − 1 is divergent if x ∈ ] − ∞, −3] ∪ [5, +∞[ and absolutely convergent if
x ∈ ] − 3, 5[.
∞
 (x − 2)n
b) Let us study the power series of x − 2, (−1)n . Let y = x − 2, and consider the series
n=2
n log(n)
∞

n yn (−1)n
(−1) . Setting an = , the radius of convergence of this series is, by Corollary 1 of
n=2
n log(n) n log(n)
Theorem 3.3.1,
 
 (−1)n 
     
 an   
 n log(n)  n + 1 log(n + 1)
.r = lim   = lim   = lim · = 1;
an+1   (−1)n+1  n log(n)
 
 (n + 1) log(n + 1) 

therefore, the series converges absolutely if y ∈ ]−1, 1[ and diverges if y ∈ ]−∞, −1[ ∪ ]1, +∞[. As y = x−2,
∞
(x − 2)n
the series (−1)n is absolutely convergent if x ∈ ]1, 3[ and divergent if x ∈ ] − ∞, 1[ ∪ ]3, +∞[.
n=2
n log(n)
Let us study the numerical series obtained at the endpoints of the convergence interval.
∞
 ∞
(−1)n 1
– If x = 1, we obtain the series (−1)n = . Let us study it using the Integral
n=2
n log(n) n=2
n log(n)
Test.
1
Let f (x) = :
x log(x)
(i) As x ≥ 2 ⇒ log(x) > 0, we have f (x) > 0, ∀x ≥ 2.
(ii) f is continuous on [2, +∞[.
(iii) f is decreasing on [2, +∞[ because x log(x) is increasing and positive.
∞  +∞
1 1
Then the numerical series and the improper integral dx are both conver-
n=2
n log(n) 2 x log(x)
gent or both divergent. Since

 +∞  x  x 1
1 1 t
dx = lim dt = lim dt
. 2 x log(x) x→+∞ 2 t log(t) x→+∞ 2 log(t)
   x     
= lim log log(t) = lim log log(x) − log log(2) = +∞,
x→+∞ 2 x→+∞

the improper integral is divergent, the same happening to the series.

∞
 1
– If x = 3, we obtain the alternating series (−1)n . We apply Leibniz’s test since we already
n=2
n log(n)
know that the series of modules is divergent:
1
(i) lim = 0.
n log(n)
1
(ii) > 0, ∀n ≥ 2.
n log(n)
1
(iii) We saw earlier that the function f (x) = is decreasing on [2, +∞[, which implies that
  x log(x)
1
is decreasing for n ≥ 2.
n log(n)
3.6. Solved Exercises 255

Then, we conclude that the series is convergent. As the series of modules is divergent, the series is
conditionally convergent.

Conclusion: The power series of x − 2 is absolutely convergent if x ∈ ]1, 3[, conditionally convergent if
x = 3, and divergent if x ∈ ] − ∞, 1] ∪ ]3, +∞[.
∞
 n
c) Let us study the power series of x + 4, n (2n2 − 1)
(x + 4)n . We can rewrite it in the form
n=1
(−3)
∞ ∞
n n
(−1)n n 2 − 1)
(x + 4) n
. Let y = x + 4, and consider the series (−1)n n 2 − 1)
y n . If
n=1
3 (2n n=1
3 (2n
n
an = (−1)n n , the radius of convergence of this series is, by Corollary 1 of Theorem 3.3.1,
3 (2n2 − 1)
 
 n 
   (−1)n n 
 an   2 − 1)  2
.r = lim   = lim  3 (2n  = lim 3n(2n + 4n + 1) = 3.
  n+1  (n + 1)(2n2 − 1)
an+1  (−1)n+1  
 3n+1 2(n + 1)2 − 1 

The series converges absolutely if y ∈ ] − 3, 3[ and diverges if y ∈ ] − ∞, −3[ ∪ ]3, +∞[. As y = x + 4,

∞
n
the series (−1)n n 2 − 1)
(x + 4)n is absolutely convergent if x ∈ ] − 7, −1[ and divergent if
n=1
3 (2n
x ∈ ] − ∞, −7[ ∪ ] − 1, +∞[.

At the endpoints of the convergence interval, we have to study the respective numerical series.
∞
 ∞

n n
– If x = −7, we obtain the series (−1)n (−3)n = .
n=1
−
3n (2n2 1) n=1
2n 2−1

It is of positive terms, and we can compare it with the harmonic series, which is divergent. Since the
limit
n
2n2 − 1 n2 1
. lim = lim =
1 2n2 − 1 2
n
is finite and different from zero, the series is divergent by Corollary 2 of the General Comparison Test.
∞ ∞
n n
– If x = −1, we obtain the alternating series (−1)n n 3n = (−1)n 2 . We apply
n=1
3 (2n − 1)
2
n=1
2n − 1
Leibniz’s test since the series of modules is divergent:
n
(i) lim = 0.
2n2 − 1
n
(ii) > 0, ∀n ≥ 1.
2n2 − 1
n+1 n 2n2 + 2n + 1
(iii) − = − < 0, ∀n ∈ N, which implies that
2(n + 1) − 1 2n2 − 1 (2n2 + 4n + 1)(2n2 − 1)
2
n
is decreasing.
2n2 − 1
We can conclude that the series is convergent. Since the series of modules is divergent, the series is
conditionally convergent.

Conclusion: The power series of x + 4 is absolutely convergent if x ∈ ] − 7, −1[, conditionally convergent if

x = −1 and divergent if x ∈ ] − ∞, −7] ∪ ] − 1, +∞[.
256 Series of Functions

∞
 en
d) Let us study the power series of x − e, (x − e)n . Let y = x − e, and consider the series
n=2
log(n)
∞
 en en
y n . If an = , the radius of convergence of this series is, by Corollary 1 of Theorem 3.3.1,
n=2
log(n) log(n)

en
 
 an  log(n) 1 log(n + 1) 1
.r = lim   = lim
 = lim · = ,
a
n+1 en+1 e log(n) e
log(n + 1)

     
1 1 1 1
so the series is absolutely convergent if y ∈ − , and divergent if y ∈ −∞, − ∪ , +∞ . As
e e e e
∞
  
en 1 1
y = x − e, the series (x − e)n is absolutely convergent if x ∈ e − , e + and divergent if
log(n) e e
  n=2
 
1 1
x ∈ −∞, e − ∪ e + , +∞ .
e e

Let us study the numerical series corresponding to the endpoints of the interval.
∞   ∞
1 en 1 n (−1)n
– If x = e − , we get the series − = . It is an alternating series because
e n=2
log(n) e n=2
log(n)
∞

1 1
an = > 0, ∀n > 1. Let us consider the series of modules, , and study it by
log(n) n=2
log(n)
comparing it with the harmonic series. As the limit

1
log(n) n
. lim = lim = +∞
1 log(n)
n
∞
 1
by Corollary 4 of the General Comparison Test, the series is divergent. Let us apply Leibniz’s
n=2
log(n)
test:
1
(i) lim = 0.
log(n)
1
(ii) > 0, ∀n ≥ 2.
log(n)
 
  1
(iii) The sequence log(n) is increasing; therefore, is a decreasing sequence for n ≥ 2.
log(n)
Then, the series is convergent. Since the series of modules is divergent, the alternating series is
conditionally convergent.
∞
  n ∞
1 en 1 1
– If x = e + , we get the series = which, as we have seen, is divergent.
e n=2
log(n) e n=2
log(n)
 
1 1
Conclusion: The power series of x − e is absolutely convergent if x ∈ e − , e + , conditionally conver-
    e e
1 1 1
gent if x = e − and divergent if x ∈ −∞, e − ∪ e + , +∞ .
e e e
3.6. Solved Exercises 257

e) Let us consider the series

 π  π  π
∞
 cos ∞ cos  n ∞ cos  n
2n 2n 1 2n 1
. = = ,
n=1
(2n − 1)2 (x − 1)n n=1
(2n − 1)2 x − 1 n=2
(2n − 1)2 x − 1

∞ π 
1  cos 2n
and let y = . We study the power series of y, y n . Note that if n > 1, then
x−1 (2n − 1)2

 n=2
π π  π π
cos 2n
0< < ; therefore, cos > 0. Let an = . The radius of convergence of this series
2n 2 2n (2n − 1)2
is, by Corollary 1 of Theorem 3.3.1,
 
   
 cos π 
   2n  π 
 an   − 1)2  cos 2n (2n + 1)2
.r = lim   = lim  (2n 

 = lim  = 1.
an+1   cos

π  π
(2n − 1)2
 cos 2(n+1)
 2(n+1)

 (2n + 1)2 

Therefore, the series converges absolutely if y ∈ ] − 1, 1[ and diverges if y ∈ ] − ∞, −1[ ∪ ]1, +∞[. If
y = ±1, we obtain two numerical series we will study.
∞ π 
 cos 2n
– If y = 1, we obtain the series , which is of positive terms. We are going to compare it
n=2
(2n − 1)2
∞
1
with 2
, which we know to be convergent, as it is a p-series with p = 2. The limit
n=1
n
 π

cos 2n π 
(2n − 1)2 n2 cos 2n 1
. lim = lim =
1 (2n − 1)2 4
n2

is finite and different from zero; therefore, by Corollary 2 of the General Comparison Test, the series is
convergent. As it is of positive terms, it is absolutely convergent.
∞ π 
 cos 2n
– If y = −1, we obtain the alternating series (−1)n , whose series of modules is the one
n=2
(2n − 1)2
we studied for the case y = 1, which we proved to be convergent. Therefore, the series is absolutely
convergent.

Conclusion: The power series of y is divergent if y ∈ ] − ∞, −1[ ∪ ]1, +∞[ and absolutely convergent if
y ∈ [−1, 1].
1
Concerning the original series, we have, because y = ,
x−1
 
 1 
.|y| ≤ 1 ⇔  ≤ 1 ⇔ |x − 1| ≥ 1 ⇔ x ≥ 2 ∨ x ≤ 0,
x − 1

which implies that the series converges absolutely if x ∈ ] − ∞, 0] ∪ [2, +∞[ and diverges if x ∈ ]0, 2[ \{1},
1
because 1 does not belong to the domain of .
x−1
258 Series of Functions

∞
(2x2 − 5)n
f) Let us study the power series of 2x2 − 5, . Let y = 2x2 − 5 and consider the series
n=1
n 3n+1
∞
 1
yn .
n=1
n 3n+1
1
Setting an = , the radius of convergence of this series is, by Corollary 1 of Theorem 3.3.1,
n 3n+1
 
 1 
   
 an   

.r = lim   
= lim  n 3n+1
 = lim 3(n + 1) = 3;
 1 
an+1   n
 (n + 1) 3n+2 

therefore, the series converges absolutely if y ∈ ] − 3, 3[ and diverges if y ∈ ] − ∞, −3[ ∪ ]3, +∞[.
At the endpoints of the interval of convergence, it is necessary to study the respective numerical series.
∞ ∞ ∞
1 (−1)n 1  (−1)n
– If y = −3, we obtain the series (−3) n
= = , which is the
n=1
n 3n+1 n=1
3n 3 n=1 n
alternating harmonic series that we know to be conditionally convergent.
∞ ∞
1 1  1
– If y = 3, we obtain the series = , which is divergent.
n=1
3n 3 n=1 n
But
|y| < 3 ⇔ |2x2 − 5| < 3 ⇔ −3 < 2x2 − 5 < 3 ⇔ 1 < x2 < 4
.  
⇔ x ∈ ] − 2, 2[ ∩ ] − ∞, −1[ ∪ ]1, +∞[ ⇔ x ∈ ] − 2, −1[ ∪ ]1, 2[.
In addition,
.2x
2
− 5 = 3 ⇔ x = −2 ∨ x = 2
and
.2x
2
− 5 = −3 ⇔ x = −1 ∨ x = 1.

Conclusion: The original series is absolutely convergent if x ∈ ] − 2, −1[ ∪ ]1, 2[, conditionally convergent
if x = −1 or x = 1 and divergent if x ∈ ] − ∞, −2] ∪ ] − 1, 1[ ∪ [2, +∞[.
√ ∞ √

n+1− n
g) Let us study the power series of x + 2, √ √ (x + 2)n . We can rewrite the series in the form
n=1
n+1+ n
∞ ∞
1 1
 (x + 2)n . Let y = x + 2 and consider the series  yn .
n=1 2n + 1 + 2 n(n + 1) n=1 2n + 1 + 2 n(n + 1)
1
Let an =  . The radius of convergence of this series is, by Corollary 1 of Theorem 3.3.1,
2n + 1 + 2 n(n + 1)
1
   
 an  2n + 1 + 2 n(n + 1) 2n + 3 + 2 (n + 1)(n + 2)
.r = lim   = lim
 = lim  = 1;
a 1 2n + 1 + 2 n(n + 1)
n+1 
2n + 3 + 2 (n + 1)(n + 2)

therefore, the series absolutely converges if y ∈ ] − 1, 1[ and diverges if y ∈ ] − ∞, −1[ ∪ ]1, +∞[. As
∞
1
y = x + 2, we can conclude that the series  (x + 2)n is absolutely convergent if
n=1 2n + 1 + 2 n(n + 1)
x ∈ ] − 3, −1[ and divergent if x ∈ ] − ∞, −3[ ∪ ] − 1, +∞[.

Let us study the numerical series at the endpoints of the interval of convergence.
3.6. Solved Exercises 259

∞
 1
– If x = −1, we obtain the series  . As it is of positive terms and the limit
n=1 2n + 1 + 2 n(n + 1)

1

2n + 1 + 2 n(n + 1) n 1
. lim = lim  =
1 2n + 1 + 2 n(n + 1) 4
n

is finite and different from zero, the series is divergent because the harmonic series is divergent (see
Corollary 2 of the General Comparison Test).
∞
 (−1)n
– If x = −3, we obtain the series  . It is an alternating series, and we apply
n=1 2n + 1 + 2 n(n + 1)
Leibniz’s test since the series of modules is divergent:
1
(i) lim  = 0.
2n + 1 + 2 n(n + 1)
1
(ii)  > 0, ∀n ≥ 1.
2n + 1 + 2 n(n + 1)
1 1
(iii)  −  < 0, which implies that the sequence is
2n + 3 + 2 (n + 1)(n + 2) 2n + 1 + 2 n(n + 1)
decreasing.
Then, the alternating series is convergent. It is conditionally convergent because the series of modules
is divergent.

Conclusion: The power series of x + 2 is absolutely convergent if x ∈ ] − 3, −1[, conditionally convergent if

x = −3 and divergent if x ∈ ] − ∞, −3[ ∪ [−1, +∞[.

∞  n
log2 (n) 2x 2x
h) Let us consider the series 2
· 2
and set y = 2 . We are going to study the power
n=1
n x + 1 x +1
∞
log2 (n) n
series of y, y . The radius of convergence of this series is, by Corollary 1 of Theorem 3.3.1,
n=1
n2
log2 (n)
with an = ,
n2
 
 log2 (n) 
   
 an    (n + 1)2 log2 (n)
 n2 
.r = lim   = lim 
  = lim 2 = 1.
a  log2 (n + 1)  n log2 (n + 1)
n+1  
 (n + 1)2 

The series is absolutely convergent if y ∈ ] − 1, 1[ and diverges if y ∈ ] − ∞, −1[ ∪ ]1, +∞[.

About the original series, we have
 
 2x  2x
.|y| < 1 ⇔  < 1 ⇔ −1 < 2 < 1 ⇔ x = −1 ∧ x = 1.
x2 + 1  x +1

Let us study the numerical series at x = −1 and x = 1.

260 Series of Functions

∞ ∞
log2 (n) log2 (n)
– If x = 1, we obtain the series 2
= , which is of positive terms, and we will study
n=1
n n=2
n2
∞
1
it by comparing it with the convergent series . The limit
n=1
n3/2

log2 (n)
n2 log2 (n)
. lim = lim √ =0
1 n
n3/2
allows us to conclude, by Corollary 3 of the General Comparison Test, that the series is convergent. As
it is of positive terms, it is absolutely convergent.
∞
log2 (n)
– If x = −1, we obtain the series (−1)n , which is alternating. As we have seen, the series of
n=2
n2
modules is convergent, and the alternating series is absolutely convergent.

Conclusion: The original power series is absolutely convergent on R.

∞
1  1 1
i) Let us study the power series of , √ · . It is important to note that the set where the
x n=1 2n n + 1 xn
series converges does not include the point x = 0 since the functions are not defined at that point. Rather,
∞
1 1 1
we can set y = , and consider the series n
√ y n . By setting an = n √ , we can apply
x n=1
2 n + 1 2 n+1
Corollary 1 of Theorem 3.3.1 to determine the radius of convergence of this series:
1
  √ √
 an  2n n+1 2 n+2
.r = lim   = lim
 = lim √ = 2.
a 1 n+1
n+1
n+1
√
2 n+2
1
The series converges absolutely if y ∈ ] − 2, 2[ and diverges if y ∈ ] − ∞, −2[ ∪ ]2, +∞[. Since y = ,
x
∞
 1 1 1
we conclude that the series √ · n converges absolutely if ∈ ] − 2, 2[ and diverges if
2n n + 1 x x
n=1    
1 1 1
∈ ] − ∞, −2[ ∪ ]2, +∞[, that is, it is absolutely convergent if x ∈ −∞, − ∪ , +∞ and divergent
x   2 2
1 1
if x ∈ − , \ {0}.
2 2
1 1
Let us study the numerical series obtained at x = − and x = .
2 2
∞
 ∞

1 1 1
– If x = , we obtain the divergent series √ = √ .
2 n=1
n + 1 n=2
n
∞
1 1
– If x = − , we obtain the alternating series (−1)n √ . We will apply Leibniz’s test since the
2 n=1
n+1
series of modules is divergent:
1
(i) lim √ = 0.
n+1
1
(ii) √ > 0, ∀n ≥ 1.
n+1
3.6. Solved Exercises 261
√ √  
1 1 n− n+1 1
(iii) √ −√ = √ √ < 0, ∀n ∈ N, which implies that √ is decreasing.
n+1 n n n+1 n
It follows that the series is convergent. As the series of the modules is divergent, the series is condi-
tionally convergent.
   
1 1
Conclusion: The power series is absolutely convergent if x ∈ −∞, − ∪ , +∞ , conditionally
  2 2
1 1 1
convergent if x = − , and divergent if x ∈ − , \ {0}.
2 2 2
∞ 
 n
2 1
j) Let us study the power series + (2x − 3)n . Let y = 2x − 3, and consider the series
n=1
3 2n
∞ 
 n  n
2 1 2 1
+ y n . If an = + , the radius of convergence of this series is, by Theorem 3.3.1,
n=1
3 2n 3 2n

1 1 1 3
.r =  =  n =   = .
lim n |an | 2 1 2 1 2
lim n + lim +
3 2n 3 2n

     
3 3 3 3
The series converges absolutely if y ∈ − , and diverges if y ∈ −∞, − ∪ , +∞ . However,
2 2 2 2
∞    
2 1 n n 3 9
y = 2x − 3, so the series + (2x − 3) is absolutely convergent if x ∈ , and divergent
3 2n 4 4
   n=1 
3 9
if x ∈ −∞, ∪ , +∞ .
4 4
Regarding the endpoints of the interval of convergence, we have to study the respective numerical series.

∞     ∞  
9 2 1 n 3 n 3 n
– If x = , we obtain the series + = 1+ . As
4 n=1
3 2n 2 n=1
4n

    1
3 n 3 4n 4 3
. lim 1 + = lim 1+ = e4 ,
4n 4n

the general term is not a null sequence; therefore, by Theorem 2.2.1, the series diverges.
∞     ∞
  
3 2 1 n 3 n 3 n
– If x = , we obtain the series + − = (−1)n 1 + . As the limit of the
4 n=1
3 2n 2 n=1
4n
general term does not exist, the series is divergent.
   
3 9
Conclusion: The power series of x is divergent if x ∈ −∞, ∪ , +∞ and absolutely convergent if
  4 4
3 9
x∈ , .
4 4
∞  2
 23n (n − 1)!
k) Let us study the power series of x2 , x2n . Let y = x2 . The radius of convergence of
n=2
(2n)!
262 Series of Functions

∞  2  2
 23n (n − 1)! 23n (n − 1)!
the series y n is, by Corollary 1 of Theorem 3.3.1, and setting an = ,
n=2
(2n)! (2n)!
 
 23n (n − 1)!2 
 
   
 an   (2n)!  (2n + 2)(2n + 1) 1
.r = lim   = lim   = lim = .
an+1   23n+3 (n!)2 
  23 n2 2
 (2n + 2)! 
     
1 1 1 1
The series is absolutely convergent if y ∈ − , and diverges if y ∈ −∞, − ∪ , +∞ . Since
2 2 2 2
∞  2  
 23n (n − 1)! 1 1
2
y = x , the series x is absolutely convergent if x ∈ − √ , √
2n
and divergent if
(2n)! 2 2
 n=2 
1 1
x ∈ −∞, − √ ∪ √ , +∞ .
2 2
At the endpoints of the interval of convergence, we obtain two numerical series.
∞  2 ∞
1  23n (n − 1)! 1 
– If x = − √ , we obtain the series · n = bn . Using Raabe’s test,
2 n=1
(2n)! 2 n=1
⎛ 2n  2 ⎞
2 (n − 1)!
  ⎜ ⎟  
bn ⎜ (2n)! ⎟ (2n + 2)(2n + 1) 3
. lim n − 1 = lim n ⎜ − 1⎟ = lim n − 1 = > 1,
bn+1 ⎝ 22n+2 (n!)2 ⎠ 4n2 2
(2n + 2)!

so the series is convergent. As it is of positive terms, it is absolutely convergent.

∞  2
1  22n (n − 1)!
– If x = √ , we again obtain the series , which we have seen to be absolutely
2 n=1
(2n)!
convergent.
   
1 1
Conclusion: The series is divergent if x ∈ −∞, − √ ∪ √ , +∞ and is absolutely convergent if
  2 2
1 1
x ∈ −√ , √ .
2 2
∞ 
 
2n − 1 2n
l) Let us study the power series of x − 3, (x − 3)n . Let y = x − 3, and consider the series
n=0
3n + 1
∞ 
   2n
2n − 1 2n 2n − 1
y n . If an = , the radius of convergence of this series is, by Theorem 3.3.1,
n=0
3n + 1 3n + 1
1 1 1 9
.r =  =  2n = 2 = . 
lim n |an |
2n − 1 2n − 1 4
lim n lim
3n + 1 3n + 1
     
9 9 9 9
The series is absolutely convergent if y ∈ − , and diverges if y ∈ −∞, − ∪ , +∞ . Since
4 4 4 4
∞    
2n − 1 2n 3 21
y = x − 3, the series (x − 3)n is absolutely convergent if x ∈ , and divergent if
3n + 1 4 4
   n=0 
3 21
x ∈ −∞, ∪ , +∞ .
4 4
3.6. Solved Exercises 263

At the endpoints of the interval of convergence, we obtain two numerical series.

∞     ∞  
3 2n − 1 2n 9 n 6n − 3 2n
– If x = , we have the series − = (−1)n . As
4 n=0
3n + 1 4 n=1
6n + 2

 2n
6n − 3 5
. lim = e− 3 ,
6n + 2

the limit of the general term does not exist, and, by Theorem 2.2.1, the series diverges.
∞     ∞  
21 2n − 1 2n 9 n 6n − 3 2n
– If x = , we obtain the series = . The general term is not
4 n=0
3n + 1 4 n=1
6n + 2
a null sequence, so the series is divergent.
   
3 21
Conclusion: The power series of x − 3 is divergent if x ∈ −∞, ∪ , +∞ and absolutely convergent
  4 4
3 21
if x ∈ , .
4 4
∞ π ∞ π
 sin n  sin n
2n
m) Let us study the power series of x−1, (x−1) = (x−1)2n . Let y = (x−1)2 ,
4n (n + 1) 4n (n + 1)
∞  
n=1 n=2
 sin n π
and consider the series y n . The radius of convergence of this series is, by Corollary 1 of
4n (n + 1)
n=2 π
sin n
Theorem 3.3.1 and setting an = n ,
4 (n + 1)
 
 sin π  
 n 
   n   π  n+1
 an   4 (n + 1)  sin n 4 (n + 2)
.r = lim   = lim
   π   = lim  π  = 4;
a  sin  sin 4n (n + 1)
n+1  n+1  n+1
 n+1 
4 (n + 2) 

the series is absolutely convergent if y ∈ ] − 4, 4[ and divergent if y ∈ ] − ∞, −4[ ∪ ]4, +∞[.

∞ π
 sin n
As y = (x − 1)2 , the series (x − 1)2n is absolutely convergent if x ∈ ] − 1, 3[ and divergent
n=2
4n (n + 1)
if x ∈ ] − ∞, −1[ ∪ ]3, +∞[.
At the endpoints of the interval of convergence, we obtain two numerical series.
∞ π ∞ π
 sin n  sin n
– If x = −1, we get the series (−2) 2n
= . Let us study it by comparison with
n=2
4n (n + 1) n=2
n+1
∞
1
2
, which is convergent (p-series with p = 2):
n=1
n
π
sinn π
n+1 n sin n
. lim = lim · = π.
1 n+1 1
n
n2

Since this limit is finite and different from zero, the series is convergent by Corollary 2 of the General
Comparison Test. Being of positive terms, it is absolutely convergent.
264 Series of Functions

∞ π ∞ π
 sin  sin
– If x = 3, we again obtain the series n
22n = n
, which we have seen to be
n=2
4n (n + 1) n=2
n+1
absolutely convergent.

Conclusion: The power series of x − 1 is divergent if x ∈ ] − ∞, −1[ ∪ ]3, +∞[ and absolutely convergent if
x ∈ [−1, 3].

∞
  
x2n x2
16. The series √ log 1 + √ is of positive terms. For it to be convergent, its general term must be a null
n=1
n n
sequence (see Theorem 2.2.1). But
 
x2n x2 0, if |x2 | ≤ 1
. lim √ log 1 + √ =
n n +∞, if |x2 | > 1;

therefore, the series can only converge if |x| ≤ 1.

We have    
x2n x2 x2n x2 x2n x2n+2
.0 ≤ √ log 1 + √ ≤ √ 1+ √ = √ + , ∀x ∈ R;
n n n n n n
∞ ∞
x2n x2n+2
let us study the power series √ and .
n=1
n n=1
n
1
Let an = √ . The radius of convergence of the first series is, by Corollary 1 of Theorem 3.3.1,
n
1
  √ √
 an  n n+1
. lim   = lim
 = lim √ = 1;
a 1 n
n+1 √
n+1
therefore, the series is absolutely convergent if x ∈ ] − 1, 1[.
1
Let bn = . The radius of convergence of the second series is, by Corollary 1 of Theorem 3.3.1,
n
1
 
 bn  n n+1
. lim   = lim
 = lim = 1;
b
n+1 1 n
n+1
therefore, the series is absolutely convergent if x ∈ ] − 1, 1[.
∞  
x2n x2
We can conclude that the series √ 1+ √ is convergent if x ∈ ] − 1, 1[, which implies, by the General
n=1
n n
Comparison Test, that the original series is convergent if x ∈ ] − 1, 1[.
∞  
1 1
For x2 = 1, that is, x = 1 or x = −1, we obtain the series √ log 1 + √ . Comparing it with the
n=1
n n
harmonic series, which is divergent,
 
1 1
√ log 1 + √    √n
n n √ 1 1
. lim = lim n log 1 + √ = lim log 1 + √ = 1,
1 n n
n
3.6. Solved Exercises 265

we conclude, by Corollary 2 of the General Comparison Test, that the series is divergent. The original series
converges if x ∈ ] − 1, 1[ and diverges if x ∈ ] − ∞, −1] ∪ [1, ∞[.

∞ 
 π n  π n
17. Let y = x2 . The series cos + 2 − (−1)n y n setting an = cos n + 2 − (−1)n has radius of
n=1
n
convergence

1 1 1 1
.  = ! π =  π = .
lim n |an | n
lim cos + 2 − (−1) n 4
lim n cos + 2 − (−1)n n
n

     
1 1 1 1
The series is absolutely convergent if y ∈ − , and diverges if y ∈
∪ , +∞ .−∞, −
4 4 4 4
∞ 
 π    
n 1 1 1 1
The series cos + 2 − (−1)n (x2 )n is absolutely convergent if x2 ∈ − , , that is, x ∈ − , ,
n=1
n 4 4 2 2
and this is the largest open interval on which the series converges.

∞
 4n 5
18. Consider the power series of x, n + 5n
xn . If the radius of convergence of the series is , then, by Theo-
2 4
n=1      
5 5 5 5
rem 3.3.1, the series is absolutely convergent on the interval − , and diverges on −∞, − ∪ , +∞ .
  4 4 4 4
5 5
But 1 ∈ − , , therefore, the series is absolutely convergent at this point. The series is divergent at x = 2
4 4   
5 5
because 2 ∈ −∞, − ∪ , +∞ .
4 4
5
At x = , the series may be convergent or divergent as it is one of the endpoints of the interval of convergence.
4  n
∞ ∞
4n 5 5n
Let us study the series n + 5n
= n + 5n
. Its general term is not a null sequence. In fact,
n=1
2 4 n=1
2

5n 1
. lim = lim  n = 1;
2n + 5n 2
+1
5

therefore, by Theorem 2.2.1, the series is divergent.

∞

19. The series an xn is a power series of x. We have
n=0

√ 1, if n is even
.n an = √n
2, if n is odd,
√ √
n
which implies that lim n an = 1, because lim 2 = 1. Then the radius of convergence of this series is, by
Theorem 3.3.1,
1
.r = √ = 1;
lim n an

therefore, the series is absolutely convergent if x ∈ ] − 1, 1[ and divergent if x ∈ ] − ∞, −1[ ∪ ]1, +∞[.
266 Series of Functions

∞

If x = −1, we obtain the series (−1)n an , whose general term does not have a limit; therefore, by Theo-
n=1
rem 2.2.1, the series is divergent.
∞
If x = 1, we obtain the series an . As lim an does not exist, the series is divergent.
n=1

Conclusion: The power series is absolutely convergent if x ∈ ] − 1, 1[ and diverges if x ∈ ] − ∞, −1] ∪ [1, +∞[.
∞
 ∞

20. Let us consider the power series an xn . If, by hypothesis, the series an 5n converges, then either the power
n=1 n=1
series converges absolutely on R or there exists r ∈ R+ such that the series converges absolutely if x ∈ ] − r, r[
∞

and diverges if x ∈ ] − ∞, −r[ ∪ ]r, +∞[. In the first case, it is evident that the series an (−2)n is absolutely
n=1
convergent. In the second case, if the series converges at x = 5, then r ≥ 5. As −2 ∈ ] − 5, 5[ ⊂ ] − r, r[ , the
∞

series an (−2)n is absolutely convergent.
n=1

21. Suppose the radius of convergence of the power series of x − 3 is r > 0. In that case, the series is absolutely
convergent on the interval ]3 − r, 3 + r[, which is the largest open interval where the series is convergent. The
∞

series may also be convergent at the endpoints of the interval. If the series an (x − 3)n converges at x = 0,
n=0
then 0 ∈ [3 − r, 3 + r], which implies that r ≥ 3. But in that case, 1 ∈ ]3 − r, 3 + r[, so the series cannot diverge
at x = 1.
If the radius of convergence of the series is +∞, then the series converges absolutely at all points of R, with no
points where it can be divergent.
∞

22. If the power series of x, bn xn , has a radius of convergence R > 1, then it is absolutely convergent if x = 1,
n=0
∞

that is, the series bn is absolutely convergent. If, for each n ∈ N, |fn (x)| ≤ |bn |, ∀x ∈ R, Weierstrass’ test
n=0
∞

allows us to conclude that the series of functions fn (x) is uniformly convergent on R.
n=0
∞

23. a) Let us consider a power series an (x − x0 )n with a radius of convergence r > 0. By Theorem 3.3.1, the
n=1
series is absolutely convergent if x ∈ ]x0 − r, x0 + r[ and divergent if x ∈ ] − ∞, x0 − r[ ∪ ]x0 + r, +∞[,
which implies that it can only be conditionally convergent at the endpoints of the interval of convergence,
that is, at x = x0 − r or x = x0 + r.
∞
If the series an (x − 1)n is conditionally convergent at x = −1, then x0 − r = −1. Since x0 = 1, we
n=1
have r = 2.
b) From the previous item, we know that the interval of convergence of the series is ] − 1, 3[, so the series
∞

can only be conditionally convergent at x = −1 or x = 3. If x = 3, we get the series an 2n . As,
n=1
by hypothesis, (an ) is a sequence of positive terms, this is a series of positive terms that, if convergent,
∞
will be absolutely convergent. Therefore, the only point where the series an (x − 1)n is conditionally
n=1
convergent is x = −1.
3.6. Solved Exercises 267

∞
 ∞
1 n 1
24. a) Let us study the power series of x−2, n
(x−2) . Let y = x−2, and consider the series yn .
n=1
n 5 n=1
n 5n
1
The radius of convergence of this series is, by Corollary 1 of Theorem 3.3.1 and setting an = ,
n 5n
 
 1 
   
 an   n 5n  (n + 1) 5n+1
. lim   = lim 
  = lim = 5;
a  1  n 5n
n+1  
 (n + 1) 5n+1 

the series is absolutely convergent if y ∈ ] − 5, 5[ and diverges if y ∈ ] − ∞, −5[ ∪ ]5, +∞[. As y = x − 2, the
∞
1
series (x − 2)n is absolutely convergent if x ∈ ] − 3, 7[ and divergent if x ∈ ] − ∞, −3[ ∪ ]7, +∞[.
n=1
n 5n

∞
 ∞
1 1
If x = 7, we get the series n
5n = , which is divergent.
n=1
n5 n=1
n
∞
 ∞
1 (−1)n
If x = −3, we get the series n
(−5)n = , which is conditionally convergent.
n=1
n5 n=1
n

Conclusion: The power series of x − 2 is absolutely convergent if x ∈ ] − 3, 7[, conditionally convergent if

x = −3 and divergent if x ∈ ] − ∞, −3[ ∪ [7, +∞[.
∞
 1
b) Let f (x) = (x − 2)n , ∀x ∈ ] − 3, 7[. By Theorem 3.3.3, we have
n=1
n 5n

 ∞
 ∞   ∞
 1  1  1
f  (x) = (x − 2)n = (x − 2) n
= n (x − 2)n−1
n=1
n 5n n=1
n 5 n
n=1
n 5 n
.
∞ ∞   ∞  
1 1 x−2 n 1  x−2 n
= (x − 2) n−1
= = , ∀x ∈ ] − 3, 7[.
n=1
5n n=0
5 5 5 n=0 5

x−2
This series is geometric with r = . Then
5

∞ 
 
x−2 n 1 5
. = =
5 x−2 7 − x
n=0 1−
5

and, finally,
 1 5 1
.f (x) = · = , ∀x ∈ ] − 3, 7[.
5 7−x 7−x

∞
(−1)n  n+1
25. Let us consider the series log(x) .
n=0
n + 1
268 Series of Functions

 ∞
(−1)n n+1
a) Let y = log(x), and let us study the power series y . The radius of convergence of this series
n=0
n+1
(−1) n
is, by Corollary 1 of Theorem 3.3.1 and setting an = ,
n+1
 
 (−1)n 
   
 an   n+1  n+2

. lim   = lim   = lim = 1;
an+1   (−1)n+1 
  n+1
 n+2 

so the series is absolutely convergent if y ∈ ] − 1, 1[ and diverges if y ∈ ] − ∞, −1[ ∪ ]1, +∞[. Since
∞
(−1)n  n+1
y = log(x), we conclude that the series log(x) is absolutely convergent if log(x) ∈ ] − 1, 1[
n=0
n + 1
and divergent if log(x) ∈ ] − ∞, −1[ ∪ ]1, +∞[, that is, it is absolutely convergent if x ∈ ]e−1 , e[ and
divergent if x ∈ ]0, e−1 [ ∪ ]e, +∞[, because the domain of log(x) is R+ .
∞ ∞ ∞ ∞
(−1)n (−1)2n+1 1 1
– If x = e−1 , we obtain the series (−1)n+1 = = − = − ,
n=0
n+1 n=0
n+1 n=0
n+1 n=1
n
which is divergent.
∞ ∞ ∞
(−1)n (−1)n+1 (−1)n
– If x = e, we obtain the series =− =− , which is the alternating
n=0
n + 1 n=0
n + 1 n=1
n
harmonic series, therefore conditionally convergent.

Conclusion: The original series is absolutely convergent if x ∈ ]e−1 , e[, conditionally convergent if x = e
and divergent if x ∈ ]0, e−1 ] ∪ ]e, +∞[.
∞
(−1)n  n+1
b) The function log(x) is the sum of a power series with interval of convergence ]e−1 , e[, so
n=0
n + 1
it is differentiable term by term on this interval (see Theorem 3.3.3), that is, if x ∈ ]e−1 , e[, we have
 ∞  ∞   ∞
 (−1)n  n+1  (−1)n  n+1  1   n
. log(x) = log(x) = (−1)n log(x) .
n=0
n + 1 n=0
n + 1 x n=0

∞
  n
But the series (−1)n log(x) is geometric with ratio r = − log(x); therefore,
n=0

∞
1   n 1 1
. (−1)n log(x) = · ,
x n=0 x 1 + log(x)
so  
∞
(−1)n  n+1 1 1
. log(x) = · .
n=0
n + 1 x 1 + log(x)
Then 
∞
(−1)n  n+1 1 1  
. log(x) = · dx = log 1 + log(x) + C.
n=0
n + 1 x 1 + log(x)

Let us determine the value of the constant C. For x = 1 (note that this value belongs to the interval of
convergence), we obtain
∞
(−1)n  n+1  
.0 = log(1) = log 1 + log(1) + C ⇔ C = 0;
n=0
n + 1
3.6. Solved Exercises 269

thus,

∞
(−1)n  n+1  
. log(x) = log 1 + log(x) .
n=0
n + 1

26. a) The domain of f is the subset of R where the series is convergent. Let us consider the power series of x2 ,
∞ ∞
(−1)n (−1)n
2n 2
x2n . Let y = x2 . The radius of convergence of the series 2n 2
y n is, by Corollary 1
n=0
2 (n!) n=0
2 (n!)
(−1)n
of Theorem 3.3.1 and setting an = 2n ,
2 (n!)2

 
 (−1)n 
   
 an   22n (n!)2 
. lim   = lim 
  = lim 22 (n + 1)2 = +∞.
a  (−1)n+1 
n+1  
   
 22n+2 (n + 1)! 2 

∞
 (−1)n
Then, the series converges absolutely for all y ∈ R. As y = x2 , the series x2n is absolutely
n=0
22n(n!)2
convergent for all x ∈ R; therefore the domain of f is R.

b) As the power series is differentiable term by term on its interval of convergence (see Theorem 3.3.3), then,
for all x ∈ R,
 ∞  ∞
 (−1)n  (−1)n 2n 2n−1
 2n
.f (x) = x = x
2n
2 (n!) 2 22n (n!)2
n=0 n=1

and
 ∞
 ∞
 (−1)n 2n 2n−1  (−1)n 2n(2n − 1) 2n−2

.f (x) = x = x .
n=1
22n (n!)2 n=1
22n (n!)2

Then
∞ ∞
2 (−1)n (−1)n
.x f (x) = x2 2n (n!)2
x 2n
= 2n (n!)2
x2n+2
n=0
2 n=0
2

∞ ∞
 (−1)n 2n 2n−1 (−1)n 2n 2n
.x f (x) = x x = x
22n (n!)2 22n (n!)2
n=1 n=1

and
∞ ∞
(−1)n 2n(2n − 1) 2n−2 (−1)n 2n(2n − 1) 2n
.x
2
f  (x) = x2 2n 2
x = x .
n=1
2 (n!) n=1
22n (n!)2
270 Series of Functions

Therefore,

x2 f  (x) + xf  (x) + x2 f (x) =

∞ ∞ ∞
(−1)n 2n(2n − 1) 2n  (−1)n 2n 2n  (−1)n
= x + x + x2n+2
n=1
22n (n!)2 n=1
22n (n!)2 n=0
22n (n!)2
∞   ∞
 (−1)n 2n(2n − 1) + 2n  (−1)n
= x2n + x2n+2
n=1
22n (n!)2 n=0
2 (n!)2
2n

∞ ∞
.
(−1)n 4n2 2n  (−1)n−1 2n
= 2n 2
x +  2 x
n=1
2 (n!) n=1 22n−2 (n − 1)!
∞ ∞
(−1)n 4n2 2n  (−1)n 4
= 2n (n!)2
x −  2 x
2n

n=1
2 n=1 2
2n (n − 1)!
∞
 ∞
(−1)n 4 (−1)n 4
2 x −
2n 2n
=    x = 0.
22n (n − 1)! 2n (n − 1)! 2
n=1 n=1 2

∞

27. Let (an ) be a sequence of real numbers such that the series an 2n is conditionally convergent.
n=1
∞
 ∞

a) If the series an 2n is conditionally convergent, then the radius of convergence of the series an xn
n=1 n=1
is 2, which implies that the power series of x is absolutely convergent if x ∈ ] − 2, 2[ and divergent if
∞

x ∈ ] − ∞, −2[ ∪ ]2, +∞[. It follows that the series an is absolutely convergent (x = 1) and the series
n=1
∞

n
an 3 is divergent (x = 3).
n=1
∞
 sin(an )
b) The series an is convergent; therefore, by Theorem 2.2.1, lim an = 0. Then lim = 1.
n=1
an

∞

28. Let f (x) = an xn .
n=0

a) We know, by Theorem 3.4.3, that the power series of x is unique and is equal to the Maclaurin series of f ,
∞
f (n) (0) n f (n) (0)
that is, f (x) = x , ∀n ∈ N0 . Therefore, an = . We will prove, by induction, that
n=0
n! n!

.f
(n)
(x) = f (x), ∀n ∈ N, ∀x ∈ R.

By hypothesis, f  (x) = f (x), ∀x ∈ R, which proves the equality for n = 1.

Assume that f (n) (x) = f (x), ∀x ∈ R. We prove that f (n+1) (x) = f (x), ∀x ∈ R.
   
.f
(n+1)
(x) = f (n) (x) = f (x) = f (x), ∀x ∈ R.

f (n) (0) 1
Then f (n) (0) = f (0) = f  (0) = 1 and an = = , ∀n ∈ N.
n! n!
3.6. Solved Exercises 271

∞
xn
b) Taking into account part a), we have f (x) = , ∀x ∈ R. However, this series is the Maclaurin series
n=0
n!
of ex , that is, f (x) = ex .

5x − 1
29. a) Let f (x) = . As x2 − x − 2 = (x + 1)(x − 2), we have
x2 − x − 2
5x − 1 A B A(x + 1) + B(x − 2) (A + B)x + A − 2B
.f (x) = = + = = .
x2 − x − 2 x−2 x+1 (x − 2)(x + 1) (x − 2)(x + 1)
Then
A+B =5 A=3
. ⇔
A − 2B = −1 B = 2;
therefore,
3 2 3 2 3 1 2
.f (x) = + =− + =− · + .
x−2 x+1 2−x 1 − (−x) 2 1− x 1 − (−x)
2
1 1
Since is the sum of a geometric series with ratio r = −x and the first term equal to 1 and x
1 − (−x) 1−
2
x
is the sum of a geometric series with ratio r = and the first term equal to 1, we have
2
∞ ∞
1
. = (−x)n = (−1)n xn
1 − (−x) n=0 n=0

and
∞ 
 x n 1
x = 2
..
1− n=0
2
x
 
These equalities are valid if | − x| < 1 and   < 1, respectively, that is, |x| < 1 and |x| < 2. We can write
2
the Maclaurin series expansion of f :
∞  
3  x n
∞ ∞  3
.f (x) = − +2 (−1)n xn = 2 (−1)n − n+1 xn ,
2 n=0 2 n=0 n=0
2

which is valid if |x| < 1.

 x
2 2
b) Let f (x) = et dt. The function g(t) = et is continuous on R, so the domain of f is R. We know,
0
∞
xn
from Example 3.4.5, that ex = , ∀x ∈ R; therefore,
n=0
n!

∞ ∞ 2n
t2 (t2 )n t
.e = = , ∀t ∈ R.
n=0
n! n=0
n!

As it is a power series, by Corollary 1 of Theorem 3.3.2, it is integrable term by term on every closed bounded
interval of R. Therefore,
 x  x ∞ 2n ∞  x 2n ∞  x
2 t t 1
f (x) = et dt = dt = dt = t2n dt
0 0 n=0 n! n=0 0
n! n=0
n! 0
.
∞  x ∞ ∞
1 t2n+1 1 x2n+1 x2n+1
= = · = ,
n=0
n! 2n + 1 0 n=0
n! 2n + 1 n=0
n!(2n + 1)

with the expansion valid on R.

272 Series of Functions

c) We know, from Example 3.4.5, that

∞
xn
.e
x
= , ∀x ∈ R.
n=0
n!

Therefore, for all x ∈ R and all t ∈ R,

∞ ∞
−tx (−tx)n (−1)n tn xn
.e = = .
n=0
n! n=0
n!

Then
∞ ∞
(−1)n tn xn (−1)n tn xn
  1−  −
1 1 − e−tx 1
n=0
n! 1
n=1
n!
f (x) = dt = dt = dt
. 0 t 0 t 0 t
 ∞
1   ∞
1 
(−1)n tn−1 xn (−1)n+1 xn n−1
= − dt = t dt.
0 n=1 n! 0 n=1 n!

For each x ∈ R \ {0}, the power series of t has an infinite radius of convergence:
 
 (−1)n xn  |x|n
 
 n!  n+1
. lim   = lim n! = lim = +∞,
 (−1)n+1 xn+1  |x|n+1 |x|
 
 (n + 1)!  (n + 1)!

and if x = 0, all terms of the series are zero.

The interval of convergence of the series is R; therefore, by Corollary 1 of Theorem 3.3.2, the series is
integrable term by term on every closed interval of R. In particular, we have on [0, 1]
 ∞
1  ∞  1
(−1)n+1 xn n−1 (−1)n+1 xn n−1
f (x) = t dt = t dt
0 n=1 n! n=1 0
n!
.
∞  ∞   ∞
(−1)n+1 xn 1 n−1 (−1)n+1 xn tn 1 (−1)n+1 n
= t dt = = x .
n=1
n! 0 n=1
n! n 0 n=1
n · n!

This development is valid on R.

 x
d) Let f (x) = arctan(t) dt. By the Fundamental Theorem of Integral Calculus, f  (x) = arctan(x).
0
 1 
1
Considering that arctan(x)and is the sum of a geometric series with ratio r = −x2
=
1 + x2 1 + x2
and the first term equal to 1, we have

∞ ∞
1  n
.
2
= − x2 = (−1)n x2n
1+x n=0 n=0

if, and only if, |x2 | < 1, that is, |x| < 1. Then
 x  ∞
x  ∞ 
 x 
1
arctan(x) = dt = (−1)n t2n dt = (−1)n t2n dt
0 1 + t2 0 n=0 n=0 0
.
∞
  x ∞
t2n+1 x2n+1
= (−1)n = (−1)n
n=0
2n + 1 0 n=0
2n + 1
3.6. Solved Exercises 273

because power series are integrable term by term on every interval contained on its interval of convergence
(see Corollary 1 of Theorem 3.3.2).
By the same theorem, we have
 x  x  ∞
 ∞  x
 
t2n+1 t2n+1
arctan(t) dt = (−1)n dt = (−1)n dt
0 0 n=0
2n + 1 n=0 0 2n + 1
∞
 (−1) n  2n+2  x ∞
 (−1) n 2n+2
t x
. = = ·
n=0
2n + 1 2n + 2 0 n=0
2n + 1 2n +2
∞
 (−1)n
= x2n+2 .
n=0
(2n + 1)(2n + 2)

Therefore,
∞
 (−1)n
.f (x) = x2n+2 ;
n=0
(2n + 1)(2n + 2)

this development is valid on ] − 1, 1[.

 
1 2
e) Taking into account that =− , we can write
2x + 5 (2x + 5)2
 
1 1 1
.f (x) = =− .
(2x + 5)2 2 2x + 5

We know that
1
∞  n ∞
1 5 1  2  2n
. = = − x = (−1)n n+1 xn ,
2x + 5 2 5 n=0 5 5
1+ x n=0
5
   
 2  5 5
and this development is valid if − x < 1, that is, x ∈ − , . It is a power series, so it is differentiable
5 2 2
term by term on its interval of convergence (see Theorem 3.3.3), that is,
   ∞
 ∞  ∞  
1  2n  2n   n2n n−1 5 5
. = (−1)n xn = (−1)n xn = (−1)n x , ∀x ∈ − , .
2x + 5 n=0
5n+1 n=0
5n+1 n=1
5n+1 2 2

Then
∞ ∞
1 1  n2n  n2n−1
f (x) = 2
=− (−1)n n+1 xn−1 = (−1)n+1 n+1 xn−1
(2x + 5) 2 n=1 5 n=1
5
.
∞ n  
2 (n + 1) n 5 5
= (−1)n n+2
x , ∀x ∈ − , .
n=0
5 2 2

2
f) Let f (x) = + e2x . We know that
3 + 2x
∞   ∞
2 2 1 2  2x n  2n+1 xn
= · . = − = (−1)n n+1 ,
3 + 2x 3 1 + 2x 3 n=0 3 n=0
3
3
   
 2x  3 3
and this development is valid if −  < 1, that is, x ∈ − , .
3 2 2
274 Series of Functions

We also know that

∞ ∞
(2x)n 2n xn
.e
2x
= = , ∀x ∈ R.
n=0
n! n=0
n!

Then

∞ ∞ ∞  
2 2n+1 xn 2n xn (−1)n 2 1
.f (x) = + e2x = (−1)n n+1 + = 2n n+1
+ xn ,
3 + 2x n=0
3 n=0
n! n=0
3 n!

 
3 3
development valid on − , .
2 2

g) By Example 3.4.4, we have the Maclaurin series representation of the function sin(x):

∞
 x2n+1
. sin(x) = (−1)n , ∀x ∈ R.
n=0
(2n + 1)!

Then, by Theorem 3.3.3,

∞  
  x2n+1
. cos(x) = sin(x) = (−1)n , ∀x ∈ R,
n=0
(2n + 1)!

that is,
∞
 ∞
(2n + 1) x2n x2n
. cos(x) = (−1)n = (−1)n , ∀x ∈ R.
n=0
(2n + 1)! n=0
(2n)!

Therefore, for every t ∈ R,

∞  2 2n ∞
 2   t4n
n t
. cos t = (−1) = (−1)n ,
n=0
(2n)! n=0
(2n)!

and
 x  ∞
x  t4n
.f (x) = cos(t2 ) dt = (−1)n dt.
0 0 n=0
(2n)!

The interval of convergence of the series is R; therefore, by Corollary 1 of Theorem 3.3.2, the series is
integrable term by term on every closed interval of R. We have
 ∞
x  ∞   
t4n (−1)n x 4n
f (x) = (−1)n dt = t dt
0 n=0
(2n)! n=0
(2n)! 0
.
∞
  x ∞

(−1)n t4n+1 (−1)n
= = x4n+1 .
n=0
(2n)! 4n + 1 0 n=0
(2n)!(4n + 1)

Thus, the power series development of x is

∞
 (−1)n
.f (x) = x4n+1 , ∀x ∈ R.
n=0
(2n)!(4n + 1)
3.6. Solved Exercises 275
 
1 1
h) Let us consider the following derivative: = . Based on this, we can express f (x) as
1−x (1 − x)2
follows:  
x + x2 1
.f (x) = = (x + x2 ) .
(1 − x)2 1−x

We know that
∞
1
. = xn .
1−x n=0

Moreover, this development is valid on the interval ] − 1, 1[. According to Theorem 3.3.3, this power series
is differentiable term by term on its interval of convergence, that is,

   ∞
 ∞ ∞
1   
. = xn = (xn ) = n xn−1 , ∀x ∈ ] − 1, 1[.
1−x n=0 n=0 n=1

Then
  ∞
 ∞
 ∞

1
f (x) = (x + x2 ) = (x + x2 ) n xn−1 = x n xn−1 + x2 n xn−1
1−x n=1 n=1 n=1
∞
 ∞
 ∞
 ∞

. = n xn + n xn+1 = (n + 1) xn+1 + n xn+1
n=1 n=1 n=0 n=1
∞
 ∞

= x+ (2n + 1) xn+1 = (2n + 1) xn+1 , ∀x ∈ ] − 1, 1[.
n=1 n=0

  1
i) Let f (x) = x arctan(x). We know that arctan(x) = , and this function is the sum of a geometric
1 + x2
series with ratio r = −x2 and the first term equal to 1. Therefore,

∞ ∞
1
.
2
= (−x2 )n = (−1)n x2n
1+x n=0 n=0

if and only if | − x2 | < 1, that is, if and only if |x| < 1. Then, if x ∈ ] − 1, 1[,

 x  ∞
x  ∞ 
 x 
1
arctan(x) = dt = (−1)n t2n dt = (−1)n t2n dt
0 1 + t2 0 n=0 n=0 0
.
∞
  x ∞
t2n+1 x2n+1
= (−1)n = (−1)n
n=0
2n + 1 0 n=0
2n + 1

because power series, by Corollary 1 of Theorem 3.3.2, are integrable term by term on every closed interval
contained in their interval of convergence. Therefore,

∞
 ∞
x2n+1 x2n+2
.f (x) =x (−1)n = (−1)n ,
n=0
2n + 1 n=0
2n +1

and this development is valid on ] − 1, 1[.

276 Series of Functions
 
1 2x
j) Taking into account that =− , we can write
1 + x2 (1 + x2 )2
 
2x 1
.f (x) =
2 2
=− 2
.
(1 + x ) 1+x
1
We know that is the sum of a geometric series with ratio r = −x2 and the first term equal to 1.
1 + x2
Thus,
∞
1  n
.
2
= − x2 ;
1+x n=0

this development is valid if | − x2 | < 1, that is, if x ∈ ] − 1, 1[. This is a power series that, by Theorem 3.3.3,
is differentiable term by term on its interval of convergence, that is,
    ∞

∞  ∞
1 
.
2
= (−1) xn 2n
= (−1)n x2n = (−1)n 2n x2n−1 , ∀x ∈ ] − 1, 1[.
1+x n=0 n=0 n=1

Then
∞ ∞
2x
.f (x) = 2 2
=− (−1)n 2n x2n−1 = (−1)n+1 2n x2n−1 , ∀x ∈ ] − 1, 1[.
(1 + x ) n=1 n=1

     3x2 
k) Let f (x) = log 2 + x3 . We have f  (x) = log 2 + x3 and =
2 + x3
∞  n ∞
1 1 1 1  x3  (−1)n 3n
.
3
= ·  3  = − = x ;
2+x 2 x 2 n=0 2 2n+1
1− − n=0
2
 
 x3  √
this development is valid if, and only if, −  < 1, that is, if and only if |x| < 3 2. Thus,
2
∞ ∞
 (−1)n 3n 3 (−1)n 3n+2
.f (x) = 3x2 n+1
x = x ,
n=0
2 n=0
2n+1
√ √
if and only if x ∈ ] − 3 2, 3 2[.
 x
Since f  (t) dt = f (x) − f (0) = f (x) − log(2), we have
0
 x  ∞
x  3 (−1)n 3n+2
f (x) = f  (t) dt + log(2) = log(2) + t dt
0 0 n=0
2n+1
.
∞
   x 
3 (−1)n
= log(2) + t3n+2 dt
n=0
2n+1 0

because, by Corollary 1 of Theorem 3.3.2, a power series is integrable term by term on its interval of
convergence. Then
∞
  x ∞
3 (−1)n t3n+3 3 (−1)n x3n+3
f (x) = log(2) + n+1
= log(2) +
n=0
2 3n + 3 0 n=0
2n+1 3n + 3
.
∞
 (−1)n
= log(2) + x3n+3 ,
n=0
2n+1 (n + 1)
√ √
and the development is valid on ] − 3
2, 3
2[.
3.6. Solved Exercises 277
! 
1+x
l) Let f (x) = log , and calculate its derivative:
1−x
 
1+x 
!  1−x
!
 !  1+x 1+x 2
2
 1+x 1−x 1−x (1 − x)2 1
.f (x) = log = ! = ! = = .
1−x 1+x 1+x 2
1+x 1 − x2
1−x 1−x 1−x

1
Knowing that is the sum of a geometric series with ratio r = x2 and the first term equal to 1, we
1 − x2
have
∞
1
. = x2n , |x2 | < 1,
1−x 2
n=0

and we can write the power series expansion of x of f  :

∞


.f (x) = x2n ;
n=0

the development is valid on ] − 1, 1[. As, by Corollary 1 of Theorem 3.3.2, a power series is term by term
integrable on its interval of convergence, we have
 x ∞ ∞  x ∞  2n+1 x ∞
t x2n+1
.f (x) − f (0) = f (x) = t2n dt = t2n dt = = .
0 n=0 n=0 0 n=0
2n + 1 0 n=0
2n + 1

The development of f in a power series of x is

∞
x2n+1
.f (x) = ,
n=0
2n +1

which is valid on the interval ] − 1, 1[.

  2x x 1 1
m) Considering that log(4 − x2 ) = − =− · and that the function g(x) = is
4 − x2 2 1 − ( x2 )2 1 − ( x2 )2
 2
x
the sum of a geometric series with ratio r = and the first term equal to 1, we can write
2
∞   ∞
 x  x 2n  x2n+1
.f (x) = − · =− ;
2 n=0 2 n=0
22n+1
 
 x 2
this development is valid if   < 1, that is, x ∈ ] − 2, 2[. This is a power series that, by Corollary 1 of
2 
Theorem 3.3.2, is term by term integrable on its interval of convergence, that is,
 ∞
x  ∞  x 2n+1 ∞
t2n+1 t 1
.f (x)−log(4) =− dt = − dt = − x2n+2 , ∀x ∈ ]−2, 2[.
0 n=0
22n+1 n=0 0
22n+1 n=0
22n+1 (2n + 2)

Then
∞
 1
.f (x) = log(4) − x2n+2 ;
n=0
22n+1 (2n + 2)
this development is valid on ] − 2, 2[.
278 Series of Functions
 
1 1
30. a) Since = , we can write
1−x (1 − x)2
 
x 1
.f (x) = =x .
(1 − x)2 1−x

1
We know that is the sum of a geometric series with ratio r = x and the first term equal to 1.
1−x
Therefore,
∞
1
. = xn ;
1−x n=0

this development is valid on ]−1, 1[. It is a power series that, by Theorem 3.3.3, is term by term differentiable
on its interval of convergence, that is,

   ∞
 ∞ ∞
1   
. = xn = (xn ) = n xn−1 , ∀x ∈ ] − 1, 1[.
1−x n=0 n=0 n=1

Then
∞ ∞
x
.f (x) = =x n xn−1 = n xn , ∀x ∈ ] − 1, 1[.
(1 − x) 2
n=1 n=1

b) Rewriting the general term of the numerical series, we get

∞ ∞  n
n 1
.
n
= n .
n=0
3 n=1
3

1
By item a), with x = ,
3
 n   1
∞
 1 1 3 3
. n =f =  2 = .
3 3 1 4
n=1 1−
3

⎧
⎨ sin(x)
, if x = 0
31. Let f (x) = x
⎩1, if x = 0.

a) We know that
∞
 x2n+1
. sin(x) = (−1)n , ∀x ∈ R;
n=0
(2n + 1)!

therefore,
∞
sin(x) x2n
. = (−1)n , ∀x ∈ R \ {0}.
x n=0
(2n + 1)!

Observing that for x = 0, the previous series is convergent and has sum 1, we conclude that

∞
 x2n
.f (x) = (−1)n , ∀x ∈ R.
n=0
(2n + 1)!
3.6. Solved Exercises 279

b) Given the uniqueness of the power series development (see Theorem 3.4.3), the series obtained in item a)
f (n) (0)
is the Maclaurin series of f , so, for each n ∈ N, the coefficient an of xn in the previous series is ,
n!
that is, if ⎧
⎪ k
⎨ (−1) , if n = 2k, k ∈ N
0
.an = (2k + 1)!
⎪
⎩0, if n = 2k + 1, k ∈ N0 ,

then ⎧ k
⎨ (−1)
(n) , if n = 2k, k ∈ N
.f (0) = n! an = 2k + 1
⎩
0, if n = 2k + 1, k ∈ N0 .

2
32. a) Let f (x) = ex + cos(2x). Taking into account that
∞
 x2n
. cos(x) = (−1)n , ∀x ∈ R,
n=0
(2n)!

we can write
∞
 (2x)2n
. cos(2x) = (−1)n , ∀x ∈ R.
n=0
(2n)!

Knowing that
∞
xn
.e
x
= , ∀x ∈ R,
n=0
n!

we have
∞
x2 (x2 )n
.e = , ∀x ∈ R.
n=0
n!

Therefore, the power series development of f is

∞ ∞ ∞  
(x2 )n (2x)2n 1 22n
.f (x) = + (−1)n = + (−1)n x2n ,
n=0
n! n=0
(2n)! n=0
n! (2n)!

which is valid on R.
b) By Theorem 3.4.3, the power series of x of a function f is unique. Therefore, the series we obtained in the
previous item is the Maclaurin series of f , that is,

∞ ∞  
f (n) (0) n 1 22n
. x = + (−1)n x2n .
n=0
n! n=0
n! (2n)!

Considering that in this series, all terms of odd order are zero, all derivatives of odd order of f at x = 0 are
zero. In particular, f (17) (0) = 0.
∞
 ∞
x2n+1 1
33. a) We know that sin(x) = (−1)n , ∀x ∈ R. If x = 1, we obtain (−1)n = sin(1).
n=0
(2n + 1)! n=0
(2n + 1)!
∞ ∞
xn 2n
b) Upon substituting x = 2 in the equality ex = , which holds true on R, we obtain e2 = .
n=0
n! n=0
n!
280 Series of Functions

∞
1
c) We have the following equality, valid on ] − 1, 1[ : = xn . By Corollary 1 of Theorem 3.3.2, we
1−x n=0
can integrate this series term by term obtaining

∞ ∞
xn+1 xn
. − log(1 − x) + log(1) = = .
n=0
n+1 n=1
n

1
If x = − , we arrive at the following result:
2

  ∞ ∞
3 (− 12 )n 1
. − log = = (−1)n n .
2 n=1
n n=1
2 n

34. a) Let f (x) = log(x2 + 1) + cos(2x), and calculate its derivative:

 2x
.f (x) = − 2 sin(2x).
x2 + 1

Taking into account that

∞
 x2n+1
. sin(x) = (−1)n , ∀x ∈ R,
n=0
(2n + 1)!

we can write
∞
 (2x)2n+1
. sin(2x) = (−1)n , ∀x ∈ R.
n=0
(2n + 1)!

1
Knowing that is the sum of a geometric series with ratio r = −x2 and the first term equal to 1, we
1 + x2
have
∞
1
.
2
= (−1)n x2n , | − x2 | < 1.
1+x n=0

Then, the power series expansion of x of f  is

∞
 ∞
 (2x)2n+1
f  (x) = 2x (−1)n x2n − 2 (−1)n
n=0 n=0
(2n + 1)!
∞
 ∞
 22n+2 x2n+1
. = (−1)n 2 x2n+1 − (−1)n
n=0 n=0
(2n + 1)!
∞  
22n+2
= (−1) 2− n
x2n+1 ,
n=0
(2n + 1)!

which is valid on ] − 1, 1[.

3.6. Solved Exercises 281

By Corollary 1 of Theorem 3.3.2, we have

 x ∞
  
22n+2
f (x) − f (0) = f (x) − 1 = (−1)n 2− t2n+1 dt
0 n=0 (2n + 1)!
∞   x
22n+2
= (−1)n 2 − t2n+1 dt
n=0
(2n + 1)! 0
.
∞  
2 22n+2
= (−1)n − x2n+2
n=0
2n + 2 (2n + 2)!
∞  
1 22n+2
= (−1)n − x2n+2 .
n=0
n+1 (2n + 2)!

The power series expansion of f in terms of x is

∞
   ∞
  
1 22n+2 1 22n
.1 + (−1)n − x2n+2 = 1 + (−1)n−1 − x2n ,
n=0
n+1 (2n + 2)! n=1
n (2n)!

which is valid on the interval ] − 1, 1[.

π
b) Let x = . Substituting this value into the expansion obtained in item a), we get the series
4

∞
  
π 2n 1 22n
.1 + (−1)n−1 − .
n=1
42n n (2n)!

π    π  
π 2 π 2
This series has sum f = log + 1 + cos 2 = log +1 .
4 4 4 4

35. a) We can write

 
2x 1
.f (x) = sin (2x) + = sin (2x) + .
(1 − x2 )2 1 − x2

Considering that
∞
 x2n+1
. sin(x) = (−1)n , ∀x ∈ R,
n=0
(2n + 1)!

we get
∞
 (2x)2n+1
. sin(2x) = (−1)n , ∀x ∈ R.
n=0
(2n + 1)!

1
Since is the sum of a geometric series with ratio r = x2 and the first term equal to 1, we have
1 − x2

∞
1
. = x2n , |x2 | < 1.
1−x 2
n=0
282 Series of Functions

As, by Theorem 3.3.3, power series are differentiable term by term on their interval of convergence,
∞
 ∞ 
 (2x)2n+1 
f (x) = (−1)n + x2n
n=0
(2n + 1)! n=0
∞
 ∞

(2x)2n+1
= (−1)n + 2nx2n−1
n=0
(2n + 1)! n=1
.
∞ ∞

(2x)2n+1
= (−1)n + 2(n + 1)x2n+1
n=0
(2n + 1)! n=0
∞  
22n+1
= (−1)n + 2(n + 1) x2n+1 ,
n=0
(2n + 1)!

development valid on ] − 1, 1[.

1
b) Let x = . Substituting this value into the development obtained in item a), we get
2
∞    2n+1 ∞  
22n+1 1 (−1)n n+1
. (−1)n + 2(n + 1) = + 2n ;
n=0
(2n + 1)! 2 n=0
(2n + 1)! 2
 
1 1 16
therefore, the sum of this series is f = sin(1) +  2 = sin(1) + .
2 1 9
1−
4
 x
1 1
36. Let f (x) = + log(4 − x). Taking into account that log(4 − x) = − dt, we can write
4−x 3 4−t
 x
1 1
.f (x) = − dt.
4−x 3 4 − t
We know that
∞
1 1
. = = (x − 3)n ;
4−x 1 − (x − 3) n=0

this development is valid if |x − 3| < 1, that is, x ∈ ]2, 4[. It is a power series so, by Corollary 1 of Theorem 3.3.2,
it is integrable term by term on its interval of convergence, that is,
 x  x  ∞

∞  x ∞
1 (x − 3)n+1
. log(4 − x) = − dt = − (t − 3) dt = − (t − 3)n dt = −
n
.
3 4 − t 3 n=0 n=0 3 n=0
n+1

Then
∞ ∞
1 (x − 3)n+1
f (x) = + log(4 − x) = (x − 3)n −
4−x n=0 n=0
n+1
.
∞
 ∞ ∞
(x − 3)n n−1
= 1+ (x − 3)n − =1+ (x − 3)n , ∀x ∈ ]2, 4[.
n=1 n=1
n n=1
n
 
1 1 1
37. Let f (x) = . Considering that = − , we can write
x2 x2 x
     
1 1 1 1
.f (x) = − =− =− .
x 3+x−3 3 1 + x−3
3
3.6. Solved Exercises 283

Since
∞ 
 n ∞

1 x−3 (x − 3)n
.
x−3
= − = (−1)n ,
1+ 3 n=0
3 n=0
3n

we also have    
∞
 ∞

1 (x − 3)n (x − 3)n
.f (x) =− (−1)n = (−1)n+1 ;
3 n=0
3n n=0
3n+1

this development is valid if |x − 3| < 3, that is, x ∈ ]0, 6[. It is a power series that, by Theorem 3.3.3, is
differentiable term by term on its interval of convergence, that is,

∞ 
  ∞

(x − 3)n n (x − 3)n−1
.f (x) = (−1)n+1 = (−1)n+1 .
n=0
3n+1 n=1
3n+1

The development is valid on the interval ]0, 6[.

1
38. Let f (x) = . As x2 + x − 6 = (x − 2)(x + 3), we get
x2 + x − 6

1 A B A(x + 3) + B(x − 2) (A + B)x + 3A − 2B

.f (x) = = + = = ,
x2 + x − 6 x−2 x+3 (x − 2)(x + 3) (x − 2)(x + 3)

so ⎧
⎪ 1
A+B =0 ⎨A =
. ⇔ 5
3A − 2B = 1 ⎪ 1
⎩B = − ;
5
therefore,
1 1 1 1 1 1 1 1
f (x) = · − · =− · − ·
5 x−2 5 x+3 5 2−x 5 3+x
1 1 1 1
= − · − ·
. 5 3 − (x + 1) 5 2 + (x + 1)
1 1 1 1
= − · − ·  .
15 x+1 10 x+1
1− 1− −
3 2

1 x+1
Knowing that is the sum of a geometric series with ratio r = and the first term equal to 1 and
x+1 3
1−
3
1 x+1
  is the sum of a geometric series with ratio r = − and the first term equal to 1, we have
x+1 2
1− −
2

∞  
1 x+1 n
. =
x+1 3
1− n=0
3

and
∞   ∞  
1 x+1 n x+1 n
.   = − = (−1)n ,
x+1 2 2
1− − n=0 n=0
2
284 Series of Functions
   
x + 1  
equalities valid if   < 1 and − x + 1  < 1, respectively. If x ∈ ] − 3, 1[, we can write the development of f
3   2 
in powers of x + 1:

∞   ∞   ∞  
1  x+1 n 1  x+1 n  1 (−1)n+1 1
.f (x) =− − (−1)n = n+1
− n+1 (x + 1)n .
15 n=0 3 10 n=0 2 n=0
5 2 3

 1 1  1
39. Let g(t) = log(t). Since log(t)
and = = is the sum of a geometric series with ratio
t t 1 + (t − 1)
r = −(t − 1) and the first term equal to 1, we have

∞ ∞
1  n
. = − (t − 1) = (−1)n (t − 1)n
1 + (t − 1) n=0 n=0

if, and only if, |t − 1| < 1, that is, t ∈ ]0, 2[. Then
 ∞
t  ∞ 
 t 
log(t) − log(1) = log(t) = (−1)n (x − 1)n dx = (−1)n (x − 1)n dx
1 n=0 n=0 1
.
∞
  t ∞

(x −
1)n+1 (t − 1)n+1
= (−1)n = (−1)n ,
n=0
n+1 1 n=0
n+1

because, by Corollary 1 of Theorem 3.3.2, power series are integrable term by term on every interval contained on
their interval of convergence. Again, by Corollary 1 of Theorem 3.3.2, we have

∞
 ∞

(t − 1)n+1 (t − 1)n+1
 (−1)n − (t − 1)  (−1)n
x
n=0
n+1 x
n=1
n+1
f (x) − f (1) = f (x) = dt = dt
1 t−1 t−1 1
 x∞ ∞  
(t − 1)n x (t − 1)n
= (−1)n dt = (−1)n dt
. 1 n=1 n+1 n=1 1 n+1
∞ 
 (−1)n (t − 1)n+1  x ∞
 (−1)n (x − 1)n+1
= = ·
n=1
n + 1 n + 1 1 n=1
n+1 n+1
∞
(−1)n
= (x − 1)n+1 .
n=1
(n + 1)2

Therefore,
∞
(−1)n
.f (x) = (x − 1)n+1 .
n=1
(n + 1)2

Moreover, this development is valid on ]0, 2[.

  1 1
40. Let g(t) = arctan(t − 2). Taking into account that arctan(t − 2) = and is the sum
1 + (t − 2)2 1 + (t − 2)2
of a geometric series with ratio r = −(t − 2) and the first term equal to 1, we have
2

∞ ∞
1  n
. = − (t − 2)2 = (−1)n (t − 2)2n
1 + (t − 2) 2
n=0 n=0
3.6. Solved Exercises 285

if, and only if, |(t − 2)2 | < 1, that is, t ∈ ]1, 3[. Then, by Corollary 1 of Theorem 3.3.2,

 ∞
t  ∞ 
 t 
arctan(t − 2) − arctan(0) = arctan(t − 2) = (−1)n (x − 2)2n dx = (−1)n (x − 2)2n dx
2 n=0 n=0 2
.
∞
  t ∞

(x − 2)2n+1 (t − 2)2n+1
= (−1)n = (−1)n .
n=0
2n + 1 2 n=0
2n + 1

By the corollary mentioned above, we have

∞
 ∞

(t − 2)2n+1 (t − 2)2n+1
 (−1)n − (t − 2)  (−1)n
x
n=0
2n + 1 x
n=1
2n + 1
f (x) = dt = dt
2 (t − 2)2 2 (t − 2)2
 ∞
x  ∞  x 
(t − 2)2n−1 (t − 2)2n−1
= (−1)ndt = (−1)n dt
. 2 n=1 2n + 1 n=1 2 2n + 1
∞  x ∞
(−1) n (t − 2) 2n (−1) n (x − 2) 2n
= = ·
n=1
2n + 1 2n 2 n=1
2n + 1 2n
∞
 (−1)n
= (x − 2)2n .
n=1
2n (2n + 1)

Therefore,
∞
 (−1)n
.f (x) = (x − 2)2n .
n=1
2n (2n + 1)

Moreover, this development is valid on ]1, 3[.

∞
f (n) (0) n
41. We know that the Maclaurin series is x . Therefore, we need to calculate f (n) (0), ∀n ∈ N. We will
n=0
n!
prove, by induction, that f (n) (x) = f  (x), ∀n > 2, ∀x ∈ R.
By hypothesis, f  (x) = f (x) + x, ∀x ∈ R, which implies that f  (x) = f  (x) + 1 and f  (x) = f  (x), ∀x ∈ R
which proves the equality for n = 3.
Suppose that f (n) (x) = f  (x), ∀x ∈ R. We prove that f (n+1) (x) = f  (x), ∀x ∈ R.
   
.f
(n+1)
(x) = f (n) (x) = f  (x) = f  (x) = f  (x).

Then, f  (0) = f (0) + 0 = 1 and f (n) (0) = f  (0) = f  (0) + 1 = 2, so the Maclaurin series of f is

∞ ∞
f (n) (0) n 2 n
. x =1+x+ x .
n=0
n! n=2
n!

π, if − π < x ≤ 0
42. a) Let f (x) =
0, if 0 < x < π.
286 Series of Functions

(a) The function f (b) The periodic extension, f˜, of the function f

Figure 3.29: Graphs of exercise 42 a)

The graph of f can be seen in Fig. 3.29. By Definition 3.5.3, the Fourier coefficients are
 
1 π 1 0 " #0
a0 = f (x) dx = π dx = x −π = π;
π −π π −π
 0  0
1 sin(nx)
an = π cos(nx) dx = = 0;
. π −π n −π
⎧
  0 ⎨0, if n is even
1 0 cos(nx) 1 cos(−n π) (−1)n − 1
bn = π sin(nx) dx = − =− + = =
π −π n −π n n n ⎩− 2 , if n is odd.
n

Thus, the Fourier series of f is

∞
π 2  
. − sin (2n − 1)x .
2 n=1
2n − 1

Since f is a piecewise C 1 function, by Theorem 3.5.1 the series converges to f˜(x) if x = kπ, k ∈ Z, and
π
takes the value , if x = kπ, k ∈ Z. In Fig. 3.30, we see an illustration of the sum of the series.
2

Figure 3.30: The sum of the series of exercise 42 a)

0, if −π <x≤0
b) Let us consider the function f (x) =
ex , if 0 < x < π.

(a) The function f (b) The periodic extension, f˜, of f

Figure 3.31: Graphs of exercise 42 b)

3.6. Solved Exercises 287

The graph of f can be seen in Fig. 3.31. By Definition 3.5.3, the Fourier coefficients are
 
1 π 1 π x 1  x π 1
a0 = f (x) dx = e dx = e = (eπ − 1);
π −π π 0 π 0 π

1 π x
an = e cos(nx) dx;
π 0
.   π  π 
1 π x 1  x
bn = e sin(nx) dx = e sin(nx) − n ex cos(nx) dx
π 0 π 0 0

n π x
= − e cos(nx) dx.
π 0
We have
 π  π  π  π
ex cos(nx) dx = ex cos(nx) + n ex sin(nx) dx = eπ cos(nπ) − e0 + n ex sin(nx) dx
0 0 0 0
 π  π 
. = (−1)n eπ − 1 + n ex sin(nx) − n ex cos(nx) dx
0 0
 π
= (−1)n eπ − 1 − n2 ex cos(nx) dx.
0

From this equality, we obtain

 π
.(1 + n2 ) ex cos(nx) dx = (−1)n eπ − 1;
0

therefore,  π (−1)n eπ − 1
. ex cos(nx) dx = .
0 n2 + 1
Using this equality, the coefficients are

1 π (−1)n eπ − 1
an = ex cos(nx) dx = ;
π 0 π(n2 + 1)
.
n (−1)n eπ − 1
bn = − · .
π n2 + 1
The Fourier series of f is
∞  
eπ − 1 (−1)n eπ − 1 n (−1)n eπ − 1
+ cos(nx) − · sin(nx)
2π n=1
π(n2 + 1) π n2 + 1
.
∞
eπ − 1 1  (−1)n eπ − 1  
= + cos(nx) − n sin(nx) .
2π π n=1 n2 + 1

Since f is a piecewise C 1 function, by Theorem 3.5.1 the series converges pointwise to the function
⎧˜
⎪
⎪ f (x), if x = kπ, k ∈ Z
⎪
⎪
⎪
⎨1
.S(x) = , if x = 2kπ, k ∈ Z
⎪
⎪ 2
⎪
⎪
⎪ eπ
⎩ , if x = (2k + 1)π, k ∈ Z.
2
Figure 3.32 displays the sum of the series.
288 Series of Functions

Figure 3.32: The sum of the series of exercise 42 b)

cos(x), if − π < x ≤ 0
c) Let f be the function defined by f (x) =
sin(x), if 0 < x < π.

(a) The function f (b) The periodic extension, f˜, of f

Figure 3.33: Graphs of exercise 42 c)

The graph of f can be seen in Fig. 3.33. Using equalities (3.7), (3.8), and (3.11), we can compute the
Fourier coefficients:

 π  0  π 
1 1
. a0 = f (x) dx = cos(x) dx + sin(x) dx
π −π π −π 0
 π 
1  0  2
= sin(x) + − cos(x) = ;
π −π 0 π
 0  π 
1
an = cos(x) cos(nx) dx + sin(x) cos(nx) dx
π −π 0
     
1 1 0     1 π    
= cos (n + 1)x + cos (n − 1)x dx + sin (n + 1)x + sin (1 − n)x dx
π 2 −π 2 0
⎧ ⎛$     %0 $     %π ⎞
⎪
⎪ 1 ⎝ sin (n + 1)x sin (1 − n)x cos (n + 1)x cos (1 − n)x
⎪
⎪ + − − ⎠, if n = 1
⎪
⎨ 2π
+
n+1 1−n n+1 1−n
= −π 0
⎪  0   
⎪
⎪ 1 sin(2x) cos(2x) π
⎪
⎪ −
⎩ + x + , if n = 1
2π 2 −π 2 0
⎧      
⎪
⎪ 1 cos (n + 1)π cos (1 − n)π 1 1
⎪
⎨ − − + + , if n = 1
2π n+1 1−n n+1 1−n
=
⎪
⎪
⎪
⎩1, if n = 1
2
3.6. Solved Exercises 289
⎧  
⎪
⎪ 1 (−1)n + 1 (−1)n + 1 1 (−1)n + 1
⎪
⎨ 2π − =− · , if n = 1
n+1 n−1 π n2 − 1
=
⎪
⎪
⎪ 1
⎩ , if n = 1;
2
 0  π 
1
bn = cos(x) sin(nx) dx + sin(x) sin(nx) dx
π −π 0
  0   
1 1     1 π    
= sin (n + 1)x + sin (n − 1)x dx + cos (n − 1)x − cos (n + 1)x dx
π 2 −π 2 0
⎧ ⎛$     %0 $     %π ⎞
⎪
⎪ 1 ⎝ cos (n + 1)x cos (n − 1)x sin (n − 1)x sin (n + 1)x
⎪
⎪ − − − ⎠, if n = 1
⎪
⎨ 2π
+
n+1 n−1 n−1 n+1
= −π 0
⎪     
⎪
⎪ 1 cos(2x) 0 sin(2x) π
⎪
⎪ − −
⎩ + x , if n = 1.
2π 2 −π 2 0

⎧     
⎪
⎪ 1 1 1 cos (n + 1)(−π) cos (n − 1)(−π)
⎪
⎨ − − + + , if n = 1
2π n+1 n−1 n+1 n−1
=
⎪
⎪
⎪
⎩1, if n = 1
2
⎧    
⎪
⎪ 1 (−1)n+1 − 1 (−1)n−1 − 1 1 n 1 + (−1)n
⎪
⎨ 2π + = − · , if n = 1
n+1 n−1 π n2 − 1
=
⎪
⎪1
⎪
⎩ , if n = 1.
2

The Fourier series of f is

∞
1 1 1 1  (−1)n + 1  
. + cos(x) + sin(x) − cos(nx) + n sin(nx)
π 2 2 π n=2 n2 − 1

∞
1 1  2  1  
. = + cos(x) + sin(x) − cos(2nx) + 2n sin(2nx) .
π 2 π n=1 4n2 − 1

Since f is a piecewise C 1 function, by Theorem 3.5.1 the series converges pointwise to the function

⎧
⎪
⎪ f˜(x), if x = kπ, k ∈ Z
⎪
⎪
⎨1
.S(x) = , if x = 2kπ, k ∈ Z
⎪
⎪ 2
⎪
⎪
⎩− 1 , if x = (2k + 1)π, k ∈ Z.
2

The sum of the series is illustrated in Fig. 3.34.

290 Series of Functions

Figure 3.34: The sum of the series of exercise 42 c)

d) The function f (x) = x3 defined on ] − π, π[ is odd (see Fig. 3.35); therefore, an = 0, ∀n ∈ N0 . Let us
calculate the remaining Fourier coefficients.

(a) The function f (b) The periodic extension, f˜, of f

Figure 3.35: Graphs of exercise 42 d)

    π 
2 π 2 cos(nx) π cos(nx)
bn = x3 sin(nx) dx = −x3 + 3x2 dx
π 0 π n 0 0 n
 3  π  π 
2 π cos(nπ) 3 sin(nx) sin(nx)
= − + x2 − 2x dx
π n n n 0 0 n
 π    π 
2
2π (−1) n 12 2
2π (−1) n 12 cos(nx) π cos(nx)
. = − − 2 x sin(nx) dx = − − 2 −x + dx
n n π 0 n n π n 0 0 n
   
2π 2 (−1)n 12 cos(nπ) sin(nx) π 2π 2 (−1)n 12(−1)n
= − − 2 −π + =− +
n n π n n2 0 n n3
 2 
6 π
= 2 (−1)n − .
n3 n

The Fourier series of f is

∞
  
6 π2
.2 (−1)n − sin(nx).
n=1
n3 n

Since f is a piecewise C 1 function, by Theorem 3.5.1 the series converges pointwise to the function

f˜(x), if x = (2k + 1)π, k ∈ Z

.S(x) =
0, if x = (2k + 1)π, k ∈ Z.
3.6. Solved Exercises 291

The sum of the series is illustrated in Fig. 3.36.

Figure 3.36: The sum of the series of exercise 42 d)

e) The function f (x) = 4 − |x| defined on ] − π, π[ is even (see Fig. 3.37); therefore, bn = 0, ∀n ∈ N. The
Fourier coefficients are

(a) The function f (b) The periodic extension, f˜, of f

Figure 3.37: Graphs of exercise 42 e)

 π  π  π
2 2 2 x2
a0 = f (x) dx = 4x −
(4 − x) dx = = 8 − π;
π0 0 π π 2 0
 π  π  π 
2 2 4 sin(nx)
. an = (4 − x) cos(nx) dx = − x cos(nx) dx
π 0 π n 0 0
   π   
2 x sin(nx) π sin(nx) 2 cos(nx) π 2(1 − (−1)n )
= − − dx = − 2
= .
π n 0 0 n π n 0 n2 π

The Fourier series of f is

∞ ∞
π 2  1 − (−1)n π 4  1  
.4 − + cos(nx) = 4 − + cos (2n − 1)x .
2 π n=1 n2 2 π n=1 (2n − 1)2

As f˜ is a piecewise C 1 function and extendable by continuity to R, by Theorem 3.5.1 the series converges
pointwise to the extension by continuity of f˜ on R.

43. a) Let us consider the periodic function of period 2π defined on the interval ]0, 2π[ by f (x) = x (see Fig. 3.38).
292 Series of Functions

Figure 3.38: The function of exercise 43 a)

By Proposition 3.5.3, the coefficients of the Fourier series can be calculated by the formulas

 2π  2π
1 1 x2
a0 = x dx = = 2π;
π 0 π 2 0
  2π  
1 2π 1 sin(nx) 2π sin(nx)
an = x cos(nx) dx = x − dx = 0;
π 0 π n 0 0 n
.
  2π  
1 2π 1 cos(nx) 2π cos(nx)
bn = x sin(nx) dx = −x + dx
π 0 π n 0 0 n
  2π 
1 2π sin(nx) 2
= − + =− .
π n n2 0 n

The Fourier series of f is

∞
1
.π −2 sin(nx).
n=1
n

By Theorem 3.5.2, the series converges pointwise to the function, S, periodic of period 2π, illustrated in
Fig. 3.39 and whose analytical expression is

x − 2kπ, if 2kπ < x < (2k + 2)π, k ∈ Z

.S(x) =
π, if x = 2kπ, k ∈ Z.

Figure 3.39: The sum of the series of exercise 43 a)

b) Let us consider the periodic function of period 2π defined on ]0, 2π[ by f (x) = x2 (refer to Fig. 3.40).
3.6. Solved Exercises 293

Figure 3.40: The function of exercise 43 b)

Using the results from the previous item and Proposition 3.5.3, the coefficients of the Fourier series are

 2π  2π
1 1 x3
8π 2
a0 = x2 dx = ; =
π 0 0 π 33
 2π   2π  2π 
1 1 2 sin(nx) sin(nx) 4
. an = 2
x cos(nx) dx = x − 2x dx = 2 ;
π 0 π n 0 0 n n
 2π  2π  2π 
1 1 cos(nx) cos(nx) 4π
bn = 2
x sin(nx) dx = −x 2
+ 2x dx = − .
π 0 π n 0 0 n n

The Fourier series of f is

∞  
4π 2 1 π
. +4 cos(nx) − sin(nx) .
3 n=1
n2 n

By Theorem 3.5.2, the series converges pointwise to the function, S, periodic of period 2π, illustrated in
Fig. 3.41 and whose analytical expression is

(x − 2kπ)2 , if 2kπ < x < (2k + 2)π, k ∈ Z

.S(x) =
2π 2 , if x = 2kπ, k ∈ Z.

Figure 3.41: The sum of the series of exercise 43 b)

 
π 3π
c) Let us consider the periodic function of period 2π defined on − , by f (x) = |x| (see Fig. 3.42).
2 2
294 Series of Functions

Figure 3.42: The function of exercise 43 c)

By Proposition 3.5.3, the coefficients of the Fourier series are

 3π/2  π/2  3π/2
1 2 x2 1 x2 5π
a0 = |x| dx = + = ;
π −π/2 π 2 0 π 2 π/2 4
 3π/2  π/2  3π/2
1 2 1
an = |x| cos(nx) dx = x cos(nx) dx + x cos(nx) dx
π −π/2 π 0 π π/2
   π/2     3π/2 
2 sin(nx) π/2 sin(nx) 1 sin(nx) 3π/2 sin(nx)
= x − dx + x − dx
.
π n 0 0 n π n π/2 π/2 n
1  nπ 2  nπ 2
= − sin + cos − ;
n 2 π n2 2 π n2
     3π/2 
1 3π/2 1 3π/2 1 cos(nx) 3π/2 cos(nx)
bn = |x| sin(nx) dx = x sin(nx) dx = −x + dx
π −π/2 π π/2 π n π/2 π/2 n
2  nπ 1  nπ
= − sin − cos .
π n2 2 n 2

The Fourier series of f is

∞   nπ  nπ    nπ  nπ 
5π  1 2 2 2 1
. − sin − 2
cos + cos(nx)+ sin + cos sin(nx).
8 n=1 n 2 πn 2 π n2 πn 2 2 n 2

By Theorem 3.5.1, the series converges pointwise to the function, S, periodic with period 2π, illustrated in
Fig. 3.43 and whose analytical expression is
⎧  
⎪ 1 3
⎨|x − 2kπ|, if 2k − π < x < 2k + π, k ∈ Z
.S(x) =  2 3 2
⎪
⎩π, if x = 2k + π, k ∈ Z.
2
 
π 3π
d) Let f be the periodic function with period 2π defined on − , by
2 2
⎧ π π
⎪
⎨x, if − ≤x≤
2 2
.
⎪
⎩π − x, π 3π
if <x≤ .
2 2
3.6. Solved Exercises 295

Figure 3.43: The sum of the series of exercise 43 c)

Figure 3.44: The function of exercise 43 d)

It is easy to verify that it is an odd function (see Fig. 3.44). As it is periodic with period 2π, it follows from
Propositions 3.5.3 and 3.5.6 that an = 0, ∀n ∈ N0 . Let us calculate the remaining coefficients.
  
2 π/2 π
bn = x sin(nx) dx + (π − x) sin(nx) dx
π 0 π/2
 π/2     π 
2 cos(nx) cos(nx) π/2 cos(nx) π
= −x − dx + π −
− − x sin(nx) dx
π n
0 0 n n π/2 π/2
  π/2  π 
2 −π  nπ sin(nx) π   nπ
= cos + − cos(nπ) − cos − x sin(nx) dx
π 2n 2 n2 0 n 2 π/2
    π 
. 2 π  nπ 1  nπ π cos(nx) π cos(nx)
= cos + 2 sin − (−1) − −xn
+ − dx
π 2n 2 n 2 n n π/2 π/2 n
   
2 π  nπ 1  nπ π π π  nπ sin(nx) π
= cos + 2 sin − (−1) + n
cos(nπ) − cos −
π 2n 2 n 2 n n 2n 2 n2 π/2
⎧
 nπ ⎪
⎨ 0, if n is even
4
= sin =
πn 2 2 ⎪
⎩ 4 (−1)
(n−1)/2
, if n is odd.
n2 π

The Fourier series of f is

∞
4  (−1)n+1
. sin((2n − 1)x).
π n=1 (2n − 1)2
296 Series of Functions

Since f is a continuous function and piecewise C 1 , Theorem 3.5.1 allows us to conclude that the series
converges pointwise to f on R.
 
π 3π
e) Let f be the periodic function of period 2π defined on − , by
2 2
⎧ π π
⎪ 2
⎨x , if − ≤x≤
2 2
.
⎪ 2
⎩π , π 3π
if <x≤ .
4 2 2

Figure 3.45: The function of exercise 43 e)

It is easy to verify that it is an even function (see Fig. 3.45). As it is periodic with period 2π, it follows from
Propositions 3.5.3 and 3.5.6 that bn = 0, ∀n ∈ N. Let us calculate the remaining coefficients.
    π/2
1 3π/2 2 π/2 1 3π/2 π2 2 x3 π  3π/2 π2
a0 = f (x) dx = x2 dx + dx = + x = ;
π −π/2 π 0 π π/2 4 π 3 0 4 π/2 3
 π/2 
2 1 3π/2 π 2
an = x2 cos(nx) dx + cos(nx) dx
π 0 π π/2 4
 π/2  π/2   
2 2 sin(nx) sin(nx) π sin(nx) 3π/2
= x − 2x dx +
π n 0 0 n 4 n π/2
  π/2    
2 π 2  nπ 2 cos(nx) 2
 π/2
cos(nx) π 3nπ  nπ 
. = sin + x − dx + sin − sin
π 4n 2 n n 0 n 0 n 4n 2 2
   
2 π2  nπ π  nπ 2 sin(nx) π/2
π  nπ
= sin + 2 cos − 2 − sin
π 4n 2 n 2 n n 0 2n 2
 2     
2 π nπ π nπ 2 nπ π nπ
= sin + 2 cos − 3 sin − sin
π 4n 2 n 2 n 2 2n 2
2  nπ 4  nπ
= cos − 3 sin .
n2 2 n π 2

The Fourier series of f is

∞   nπ  nπ 
π2 2 4
. + 2
cos − sin cos(nx).
6 n=1
n 2 n3 π 2

Since f is a continuous function and piecewise C 1 , by Theorem 3.5.1, the series converges pointwise to f
on R.
3.6. Solved Exercises 297

Figure 3.46: The function of exercise 44 a)

44. a) Let f be the periodic function of period 1 defined on the interval ]0, 1[ by f (x) = π sin(π x). It is easy to
verify that it is an even function (see Fig. 3.46) and, therefore, bn = 0, ∀n ∈ N. As it is periodic with period
2l = 1, it follows from Proposition 3.5.9 that the coefficients of its development in Fourier series are given
by
  1/2
1 l " #1/2
a0 = f (x) dx = 4 π sin(πx) dx = 4 − cos(πx) 0 = 4;
l −l 0
.
  nπx  1/2
1 l
an = f (x) cos dx = 4 π sin(πx) cos(2nπx) dx.
l −l l 0

Using equality (3.8), the previous integral is equal to

 $     %1/2
1/2      cos (2n + 1)π x cos (2n − 1)π x
.2 π sin (2n + 1)π x − sin (2n − 1)π x dx = 2 − +
0 2n + 1 2n − 1
0

and so
4
.an =− .
(2n − 1)(2n + 1)

The Fourier series of f is

∞
 1
.2 −4 cos(2nπx).
n=1
(2n − 1)(2n + 1)

Since f is a continuous function and piecewise C 1 , Theorem 3.5.2 allows us to conclude that the series
converges pointwise to f on R.
b) Let f be the periodic function of period 1 defined on the interval ]0, 1[ by f (x) = 2 − x2 (see Fig. 3.47).

Figure 3.47: The function of exercise 44 b)

298 Series of Functions

As it is periodic with period 2l = 1, it follows from Propositions 3.5.9 and 3.5.3 that the coefficients of its
Fourier series development are given by
 l  1/2  1  1
1 x3 10
a0 = f (x) dx = 2 f (x) dx = 2 (2 − x2 ) dx = 2 2x − = ;
l −l −1/2 0 3 0 3
 l  nπx  1
1
an = f (x) cos dx = 2 (2 − x2 ) cos(2nπx) dx
l −l l 0

 1   1  
sin(2nπx) 1 sin(2nπx) 1 sin(2nπx)
= 4 −2 x2 cos(2nπx) dx = −2 x2 − 2x dx
2nπ 0 0 2nπ 0 0 2nπ
  1  
2 1 2 cos(2nπx) 1 cos(2nπx)
= x sin(2nπx) dx = −x + dx
nπ 0 nπ 2nπ 0 0 2nπ
  1 
. 2 cos(2nπ) sin(2nπx) 1
= − + =− ;
nπ 2nπ (2nπ)2 0 (nπ)2
 l  nπx  1
1
bn = f (x) sin dx = 2 (2 − x2 ) sin(2nπx) dx
l −l l 0
   1    1 
cos(2nπx) 1 cos(2nπx) 1 cos(2nπx)
= 4 − −2 x2 sin(2nπx) dx = −2 −x2 + 2x dx
2nπ 0 0 2nπ 0 0 2nπ
  1  1 
1 1 sin(2nπx) 1 sin(2nπx)
= −2 − + x − dx
2nπ nπ 2nπ 0 nπ 0 2nπ
 
1 1 cos(2nπ) 1 1
= + − = .
nπ (nπ)2 2nπ 0 nπ

The Fourier series of f is

∞ 
 
5 1 1
. + − cos(2nπx) + sin(2nπx) .
3 n=1 (nπ)2 nπ

Since f is piecewise C 1 , by Theorem 3.5.2 the series converges pointwise to the function, S, periodic with
period 1, illustrated in Fig. 3.48 and analytically defined by
⎧
⎪
⎨f (x), if x ∈/Z
.S(x) =
⎪
⎩ ,
3
if x ∈ Z.
2

c) Let f be the periodic function with period 4 defined on the interval ]0, 4] by f (x) = x − 3 (see Fig. 3.49).
As it is periodic with period 2l = 4, it follows from Proposition 3.5.9 that the coefficients of its development
in Fourier series are given by
 l  2  4  4
1 1 1 1 x2
a0 = f (x) dx = f (x) dx = (x − 3) dx = − 3x = −2;
l −l 2 −2 2 0 2 2 0
.  l  nπx  4  nπx
1 1
an = f (x) cos dx = (x − 3) cos dx
l −l l 2 0 2
3.6. Solved Exercises 299

Figure 3.48: The sum of the series of exercise 44 b)

Figure 3.49: The function of exercise 44 c)

  nπx    nπx 4  4 
1 4 6   nπx 4 1   nπx
= x cos dx − sin = x sin − sin dx
2 0 2 nπ 2 0 nπ 2 0 0 2

2   nπx 4
= cos = 0;
(nπ)2 2 0
  nπx   nπx
1 l 1 4
. bn = f (x) sin dx = (x − 3) sin dx
l −l l 2 0 2
  nπx  nπx 4    nπx 4  4 
1 4 6  1   nπx
= x sin cos dx + = −x cos + cos dx
2 0 2 nπ 2 0 nπ 2 0 0 2
    
1 2 nπx 4 4
= −4 + sin =− .
nπ nπ 2 0 nπ

The Fourier series of f is

∞
1  nπx
. −1−4 sin .
n=1
nπ 2

Since f is piecewise C 1 , Theorem 3.5.2 allows us to conclude that the series converges pointwise to the
function, S, periodic with period 4, illustrated in Fig. 3.50.
300 Series of Functions

Figure 3.50: The sum of the series of exercise 44 c)

The analytical expression of the sum of the series is

f (x), if x = 4k, k ∈ Z
.S(x) =
−1, if x = 4k, k ∈ Z.

d) Let us consider the function f , periodic with period 2l = 4, defined on [−2, 2[ by

⎧
⎪
⎪ 0, if − 2 ≤ x ≤ −1
⎨
.f (x) = x, if − 1 < x < 1
⎪
⎪
⎩
1, if 1 ≤ x < 2

Figure 3.51: The function of exercise 44 d)

(see Fig. 3.51). As it is periodic with period 2l = 4, it follows from Proposition 3.5.9 that the coefficients
of its Fourier series development are given by
 l  2  1  2 
1 1 1 1
.a0 = f (x) dx = f (x) dx = x dx + 1 dx = ;
l −l 2 −2 2 −1 1 2

 l  nπx  1  nπx  2  nπx 

1 1
an = f (x) cos dx = x cos dx + cos dx
l −l l 2 −1 2 1 2

1   nπx 2 1  nπ
= sin =− sin ;
nπ 2 1 nπ 2
3.6. Solved Exercises 301
 l  nπx  1  nπx  2  nπx 
1 1
bn = f (x) sin dx = x sin dx + sin dx
l −l l 2 −1 2 1 2

  nπx 1  1  nπx   nπx 2 

1
= −x cos + cos dx − cos
nπ 2 −1 −1 2 2 1

  nπ  −nπ  nπ 
1 2   nπx 1
= − cos − cos + sin − cos(nπ) + cos
nπ 2 2 nπ 2 −1 2

  nπ  nπ 
1 4
= − cos + sin − (−1)n .
nπ 2 nπ 2

The Fourier series of f is

∞ 
  nπ  nπx   nπ  nπ   nπx 
1 1 1 4
. + − sin cos + − cos + sin − (−1)n sin .
4 n=1 nπ 2 2 nπ 2 nπ 2 2

As f is piecewise C 1 , by Theorem 3.5.2 the series converges pointwise to the function, S, periodic with
period 4, illustrated in Fig. 3.52 and whose analytical expression is
⎧
⎪
⎪f (x), if x = 4k + 2 ∧ x = 4k + 3, k ∈ Z
⎪
⎪
⎪
⎨1
.S(x) =
, if x = 4k + 2, k ∈ Z
⎪ 2
⎪
⎪
⎪
⎪
1
⎩− , if x = 4k + 3, k ∈ Z.
2

Figure 3.52: The sum of the series of exercise 44 d)

e) Let us consider the function f , periodic with period 6, defined on the interval ]−3, 3] by

⎧ x
⎪
⎨− , if − 3 < x ≤ 0
.f (x) =
3
⎪ x2
⎩2x − , if 0 < x ≤ 3
3
302 Series of Functions

Figure 3.53: The function of exercise 44 e)

As it is periodic with period 2l = 6 (see Fig. 3.53), it follows from Proposition 3.5.9 that the coefficients of
its Fourier series development are given by

 l   0  3 
1 1 3 1 x x2
a0 = f (x) dx = f (x) dx = − dx + 2x − dx
l −l 3 −3 3 −3 3 0 3
  0   3

1 x2 x3 5
= − + x2 − = ;
3 6 −3 9 0 2
 l   0  nπx  3  nπx 
1 nπx 1 x x2
an = f (x) cos dx = − cos dx + 2x − cos dx
l −l l 3 −3 3 3 0 3 3
    0    3 2 
1 x nπx 0 1 nπx 1 x nπx
= − sin + sin dx + 2x − cos dx
nπ 3 3 −3 3 −3 3 3 0 3 3
 
1   nπx 0 1  x2  nπx 3  3  2x  nπx
= − cos + 2x − sin − 2 − sin dx
(nπ)2 3 −3 nπ 3 3 0 0 3 3
 3  3 
1 1  2x  nπx  nπx
= (−1 + cos(nπ)) − −3 2 − cos − 2 cos dx
(nπ)2 (nπ)2 3 3 0 0 3
.
   nπx 3  n−7
1 1 6 (−1)
= (−1 + (−1)n ) − 6− sin = ;
(nπ)2 (nπ)2 nπ 3 0 (nπ)2
  nπx  0  nπx  3  nπx 
1 l 1 x x2
bn = f (x) sin dx = − sin dx + 2x − sin dx
l −l l 3 −3 3 3 0 3 3
    0    3 2 
1 x nπx 0 1 nπx 1 x nπx
= cos − cos dx + 2x − sin dx
nπ 3 3 −3 3 −3 3 3 0 3 3
   
1 1   nπx 0 x2  nπx 3  3  2x  nπx
= (−1)n − sin + − 2x − cos + 2− cos dx
nπ nπ 3 −3 3 3 0 0 3 3
  3  3 
1 3 2x  nπx 3 2  nπx
= (−1)n − 3 cos(nπ) + 2− sin + sin dx
nπ nπ 3 3 0 nπ 0 3 3
  nπx 3 
1 6  6(1 − (−1)n ) 2(−1)n
= −2(−1)n + − cos = − .
nπ (nπ)2 3 0 (nπ)3 nπ

The Fourier series of f is

∞ 
  nπx    nπx 
5 (−1)n − 7 6(1 − (−1)n ) 2(−1)n
. + cos + − sin .
4 n=1 (nπ)2 3 (nπ)3 nπ 3
3.6. Solved Exercises 303

Since f is piecewise C 1 , Theorem 3.5.2 allows us to conclude that the series converges pointwise to the
function, S, periodic of period 4, illustrated in Fig. 3.54 and whose analytical expression is

f (x), if x = 6k + 3, k ∈ Z
.S(x) =
2, if x = 6k + 3, k ∈ Z.

Figure 3.54: The sum of the series of exercise 44 e)

45. In Fig. 3.55 we can see the graph of the periodic function of period 2π defined on the interval [0, 2π] in the
following way: ⎧a
⎪
⎪ x, if 0 ≤ x ≤ b
⎪
⎪
⎪
⎪
b
⎪
⎪
⎪
⎪ a, if b < x ≤ π − b
⎪
⎪
⎨a
.f (x) = (π − x), if π − b < x ≤ π + b
⎪
⎪ b
⎪
⎪
⎪
⎪ −a, if π + b < x ≤ 2π − b
⎪
⎪
⎪
⎪
⎪
⎪ a
⎩ (x − 2π), if 2π − b < x ≤ 2π
b
for the case a = b = 1.

Figure 3.55: The function of exercise 45 if a = b = 1

It is easy to see that f is odd; therefore, an = 0, ∀n ∈ N0 . By Proposition 3.5.3 the coefficients bn can be
calculated by the formulas:
 2π  π  π
1 1 2
.bn = f (x) sin(nx) dx = f (x) sin(nx) dx = f (x) sin(nx) dx
π 0 π −π π 0

 b  π−b  π 
2 a a
= x sin(nx) dx + a sin(nx) dx + (π − x) sin(nx) dx
π 0 b b π−b b
304 Series of Functions
       
2 a cos(nx) b a b cos(nx) cos(nx) π−b a π
= −x + dx − a + (π − x) sin(nx) dx
π b n 0 b 0 n n b b π−b

  b  π−b    
2 a a sin(nx) cos(nx) a cos(nx) π a π cos(nx)
= − cos(nb) + −a − (π − x) − dx
π n b n2 0 n b b n π−b b π−b n

  π−b  
2 a a cos(nx) a a π cos(nx)
= − cos(nb) + 2 sin(nb) − a + cos(nπ − nb) − dx
π n n b n b n b π−b n

  π 
2 a a a a a sin(nx)
= − cos(nb) + 2 sin(nb) − (cos(nπ − nb) − cos(nb)) + cos(nπ − nb) −
π n n b n n b n2 π−b

2 a a 2a 2a
= sin(nb) + 2 sin(nπ − nb) = (sin(nb) + sin(nπ − nb)) = (1 − (−1)n ) sin(nb).
π n2 b n b b n2 π b n2 π

The Fourier series of f is

∞ ∞  
 2a 4a  sin (2n − 1)b  
. (1 − (−1) n
) sin(nb) sin(nx) = sin (2n − 1)x .
n=1
b n2 π b π n=1 (2n − 1)2

Since f is a continuous function and piecewise of class C 1 , Theorem 3.5.1 allows us to conclude that the series
converges pointwise to f .

46. Let f : [−π, π] → R be piecewise C 1 and

∞
a0
. + (an cos(nx) + bn sin(nx))
2 n=1

its Fourier series.

It is easy to prove by change of variables that
 π  π
f (−x) dx = f (x) dx;
−π −π
 π  π
. f (−x) cos(nx) dx = f (x) cos(nx) dx;
−π −π
 π  π
f (−x) sin(nx) dx = − f (x) sin(nx) dx.
−π −π

f (x) + f (−x)
Let us calculate the Fourier coefficients of g(x) = .
2
 π  π  π  π 
1 1 f (x) + f (−x) 1
.c0 = g(x) dx = dx = f (x) dx + f (−x) dx
π −π π −π 2 2π −π −π

 π
1
= f (x) dx = a0 ;
π −π
3.6. Solved Exercises 305
 π  π
1 1 f (x) + f (−x)
cn = g(x) cos(nx) dx = cos(nx) dx
π −π π −π 2

 π  π 
1
= f (x) cos(nx) dx + f (−x) cos(nx) dx
2π −π −π

 π
1
= f (x) cos(nx) dx = an ;
π −π

 π  π
1 1 f (x) + f (−x)
dn = g(x) sin(nx) dx = sin(nx) dx
π −π π −π 2

 π  π 
1
= f (x) sin(nx) dx + f (−x) sin(nx) dx = 0.
2π −π −π

The Fourier series of g is

∞
a0
. + an cos(nx).
2 n=1

f (x) − f (−x)
The coefficients of h(x) = are
2
   π  π 
1 π 1 π f (x) − f (−x) 1
c∗0 = h(x) dx = dx = f (x) dx − f (−x) dx = 0;
π −π π −π 2 2π −π −π
 
1 π 1 π f (x) − f (−x)
c∗n = h(x) cos(nx) dx = cos(nx) dx
π −π π −π 2
 π  π 
1
= f (x) cos(nx) dx − f (−x) cos(nx) dx = 0;
2π −π −π
.  
1 π 1 π f (x) − f (−x)
d∗n = h(x) sin(nx) dx = sin(nx) dx
π −π π −π 2
 π  π 
1
= f (x) sin(nx) dx − f (−x) sin(nx) dx
2π −π −π

1 π
= f (x) sin(nx) dx = bn .
π −π

The Fourier series of h is

∞

. bn sin(nx).
n=1

47. Let a ∈ R. From f (x + π) = −f (x), ∀x ∈ R, and the fact that f is periodic with period 2π, we get
 a+π  a+2π
. f (x) cos(2nx) dx = − f (x) cos(2nx) dx (3.15)
a a+π
 a+π  a+2π
f (x) sin(2nx) dx = − f (x) sin(2nx) dx. (3.16)
a a+π
306 Series of Functions

Indeed, by the variable change t = x − π, we obtain

 a+2π  a+π 
  a+π
. f (x) cos(2nx) dx = f (t + π) cos 2n(t + π) dt = − f (t) cos(2nt + 2nπ) dt
a+π a a
  
a+2π a+π   a+π
f (x) sin(2nx) dx = f (t + π) sin 2n(t + π) dt = − f (t) sin(2nt + 2nπ) dt.
a+π a a

Moreover, the result follows from the periodicity of the two trigonometric functions. Similarly, we can show that
 a+π  a+2π
   
. f (x) cos (2n − 1)x dx = f (x) cos (2n − 1)x dx (3.17)
a a+π
 
a+π   a+2π  
f (x) sin (2n − 1)x dx = f (x) sin (2n − 1)x dx. (3.18)
a a+π

Let us calculate the Fourier coefficients of the function f using equalities (3.15), (3.16), (3.17), and (3.18).
  0  π    π  π 
1 π 1 1
a0 = f (x) dx = f (x) dx + f (x) dx = − f (x) dx + f (x) dx = 0;
π −π π −π 0 π 0 0
  0  π 
1 π 1
an = f (x) cos(nx) dx = f (x) cos(nx) dx + f (x) cos(nx) dx
π −π π −π 0
⎧   π  π 
⎪ 1
⎪
⎪ − f (x) cos(nx) dx + f (x) cos(nx) dx = 0, if n is even
⎨π 0 0
=  π  π  
⎪
⎪ 1 2 π
⎪
⎩ f (x) cos(nx) dx + f (x) cos(nx) dx = f (x) cos(nx) dx, if n is odd;
.
π 0 0 π 0
  0  π 
1 π 1
bn = f (x) sin(nx) dx = f (x) sin(nx) dx + f (x) sin(nx) dx
π −π π −π 0
⎧   π  π 
⎪ 1
⎪
⎪ − f (x) sin(nx) dx + f (x) sin(nx) dx = 0, if n is even
⎨π 0 0
=  π  π  
⎪
⎪ 1 2 π
⎪
⎩ f (x) sin(nx) dx + f (x) sin(nx) dx = f (x) sin(nx) dx, if n is odd.
π 0 0 π 0

The Fourier series of f is

∞
     
. a2n−1 cos (2n − 1)x + b2n−1 sin (2n − 1)x .
n=1

48. a) The graph of the function

0, if 0 ≤ x < 2
.f (x) =
x − 2, if 2 ≤ x < 4
and its odd extension can be seen in Fig. 3.56.
The analytical expression of the odd extension to the interval ] − 4, 4[ is
⎧
⎪
⎪ x + 2, if − 4 < x ≤ −2
⎨
.f˜(x) = 0, if − 2 < x < 2
⎪
⎪
⎩
x − 2, if 2 ≤ x < 4.
3.6. Solved Exercises 307

(a) The function f (b) The odd extension of f

Figure 3.56: Graphs of exercise 48

b) Since f˜ is an odd function, the Fourier series expansion is, in fact, a Fourier sine series. We can consider f˜
periodic with period 2l = 8. The coefficients of the Fourier sine series are
 4  nπx  4  nπx
2 1
bn = f (x) sin dx = (x − 2) sin dx
4 0 4 2 2 4
 4  
1 4(x − 2)  nπx 4  nπx 4
= − cos + cos dx
2 nπ 42 2 nπ 4
  2   
1 8 4 nπx 4
. = − cos(nπ) + sin
2 nπ nπ 4 2
   
1 8(−1)n 4 2  nπ
= − − sin
2 nπ nπ 2

4(−1)n 8  nπ
= − − 2
sin ,
nπ (nπ) 2

and the series is

∞   nπ   nπx
4  (−1)n 2
. − − 2 sin sin .
π n=1 n n π 2 4

By Theorem 3.5.2, the series converges pointwise to the periodic function with period 8, illustrated in
Fig. 3.57.

Figure 3.57: The sum of the series of exercise 48

308 Series of Functions

49. If we want the Fourier cosine series of the function

1, if 0 ≤ x < π
.f (x) =
0, if π < x ≤ 2π,

we must consider its extension as an even function, as illustrated in Fig. 3.58. By Proposition 3.5.7, the Fourier
series of f is
a0 ∞  nx
. + an cos ,
2 n=1
2

Figure 3.58: The even extension of the function f of exercise 49

 π  nx
2
where an = f (x) cos dx, n ∈ N0 , that is,
2π 0 2
 2π  
2 1 2π 1 π 1  π
a0 = f (x) dx = f (x) dx = 1 dx = x = 1;
2π 0 π 0 π 0 π 0
.   nx   nx   nx π  nπ
1 2π 1 π 1 2 2
an = f (x) cos dx = cos dx = sin = sin .
π 0 2 π 0 2 π n 2 0 nπ 2

The Fourier series of f is


∞ nπ  nx ∞  
1 2  sin 2 1 2  (−1)n+1 (2n − 1)x
. + cos = + cos .
2 π n=1 n 2 2 π n=1 2n − 1 2

50. Let us consider the function f (x) = π − x.

a) If we want the Fourier sine series of the function, we must consider its extension as an odd function, as
illustrated in Fig. 3.59a.
∞ 
2 π
By Proposition 3.5.8, the Fourier series of f is bn sin(nx), where bn = f (x) sin(nx) dx, n ∈ N,
n=1
π 0
that is,  
2 π 2 π
bn = f (x) sin(nx) dx = (π − x) sin(nx) dx
π 0 π 0
   π 
2 cos(nx) π cos(nx)
. = −(π − x) − dx
π n 0 0 n
  π 
2 π sin(nx) 2
= − = .
π n n2 0 n
3.6. Solved Exercises 309

(a) (b)
Figure 3.59: (a) The odd extension of f (x) = π − x. (b) The even extension of f (x) = π − x

The Fourier series of f is

∞
2
. sin(nx).
n=1
n

By Theorem 3.5.1, the series converges pointwise, on the interval ]0, 2π[, to the function f˜(x) = π − x.
From this fact, it follows that
∞
sin(nx) π−x
. = , 0 < x < 2π.
n=1
n 2

π−x
b) The periodic function of period 2π defined on ]0, 2π[ by g(x) = is piecewise continuous. By Theo-
2
rem 3.5.7 we have
  ∞
x  ∞  x
x π−t sin(nt) sin(nt)
. dt = dt = dt, 0 < x < 2 π,
0 2 0 n=1
n n=1 0
n

or,
∞   ∞ ∞
πx x2 cos(nx) 1 cos(nx) 1
. − = − 2
+ 2
=− 2
+ .
2 4 n=1
n n n=1
n n=1
n2

Taking x = 0 in exercise 43 b), we obtain

∞
2 4π 2 1
.2π = +4 ;
3 n=1
n2

therefore,
∞
1 π2
.
2
= .
n=1
n 6
Then
∞
cos(nx) π2 πx x2 3x2 − 6πx + 2π 2
.
2
= − + = .
n=1
n 6 2 4 12
310 Series of Functions

c) If we want the Fourier cosine series of the function, we must consider its extension as an even function, as
illustrated in Fig. 3.59b. By Proposition 3.5.7,
 π  π  π  π
1 2 2 2 x2
a0 = f (x) dx = f (x) dx = (π − x) dx = πx − = π;
π −π π 0 π 0 π 2 0
 π 
2 2 π
an = f (x) cos(nx) dx = (π − x) cos(nx) dx
π 0 π 0
.  π  π 
2 (π − x) 1
= sin(nx) + sin(nx) dx
π n 0 0 n
 π
2 cos(nx) 2
= − = 2 (1 − (−1)n ) .
nπ n 0 n π

The Fourier series of f is

∞ ∞
π 2  1 − (−1)n π 4  1  
. + cos(nx) = + cos (2n − 1)x .
2 π n=1 n2 2 π n=1 (2n − 1)2

d) As the even extension of the function f , f˜(x) = π − |x|, is a continuous function and piecewise C 1 on
] − π, π[, we have
∞
π 2  1 − (−1)n
.π − |x| = + cos(nx), −π < x < π.
2 π n=1 n2

In particular, if x = 0,
∞
π 2  1 − (−1)n
.π = + .
2 π n=1 n2

∞ ∞
1 (−1)n
Since the series 2
and are convergent, Theorem 2.2.2 allows us to write
n=1
n n=1
n2

∞ ∞
π 2  1 2  (−1)n
.π = + − .
2 π n=1 n2 π n=1 n2

By item b),
∞
π 2 π2 2  (−1)n
.π = + · − .
2 π 6 π n=1 n2

Finally,
∞
(−1)n π2
.
2
=− .
n=1
n 12

51. Let us consider the function f (x) = x.

a) If we want the Fourier sine series of the function, we must consider its extension as an odd function, as illus-
∞
trated in Fig. 3.60a. By Proposition 3.5.8, the Fourier series of f is bn sin(nx), where
n=1
3.6. Solved Exercises 311
 π
2
bn = f (x) sin(nx) dx, n ∈ N, that is,
π 0

 π  π
2 2
bn = f (x) sin(nx) dx = x sin(nx) dx
π 0 π 0
 π  π 
2 cos(nx) cos(nx)
. = −x + dx
π n 0 0 n
   
2 π(−1)n sin(nx) π 2(−1)n+1
= − + = .
π n n2 0 n

(a) (b)
Figure 3.60: (a) The odd extension of f (x) = x. (b) The even extension of f (x) = x

The Fourier series of f is

∞
2(−1)n+1
. sin(nx).
n=1
n

By Theorem 3.5.1, the series converges pointwise, on the interval ] − π, π[, to the function f˜(x) = x. From
this fact, it follows that
∞
sin(nx) x
. (−1)n+1 = , −π < x < π.
n=1
n 2

b) If we want the Fourier cosine series of the function, we must consider its extension as an even function, as
illustrated in Fig. 3.60b. By Proposition 3.5.7,
 π  π  π  π
1 2 2 2 x2
a0 = f (x) dx = f (x) dx = x dx = = π;
π −π π 0 π 0 π 2 0
 π  π
2 2
an = f (x) cos(nx) dx = x cos(nx) dx
π 0 π 0
. 
π  π 1 
2x
= sin(nx) − sin(nx) dx
πn 0 0 n
 π
2 cos(nx) 2  
= = 2 (−1)n − 1 .
nπ n 0 n π
312 Series of Functions

The Fourier series of f is

∞ ∞
π 2  (−1)n − 1 π 4  1  
. + cos(nx) = − cos (2n − 1)x .
2 π n=1 n2 2 π n=1 (2n − 1)2

52. The function f (x) = | sin(x)| is even (see Fig. 3.61); therefore, bn = 0, ∀n ∈ N. In addition, f is periodic with
period π, so it also has period 2π. Let us calculate the Fourier coefficients.

Figure 3.61: The function f (x) = | sin(x)|

 
2 π 2 π 2" #π 4
a0 = f (x) dx = sin(x) dx = − cos(x) 0 = ;
π 0 π 0 π π
 
2 π 1  2  π 
an = sin(x) cos(nx) dx = sin (n + 1)x + sin (1 − n)x dx
π 0 0 2 π
⎧ $     %π
⎪
⎪ 1 cos (n + 1)x cos (1 − n)x
⎪
⎪ − − , if n = 1
.
⎨ π n+1 1−n
= 0
⎪
⎪  
⎪
⎪ 1 cos(2x) π
⎩ − , if n = 1
π 2 0
⎧  
⎪ 1 (−1)n + 1 (−1)n + 1 2 (−1)n + 1
⎨ − =− · , if n = 1
= π n+1 n−1 π n2 − 1
⎪
⎩
0, if n = 1.

The Fourier series of f is

∞ ∞
2 2  (−1)n + 1 2 4  1
. − cos(nx) = − cos(2nx).
π π n=2 n − 1
2 π π n=1 4n2 − 1

Since f is a continuous function and piecewise C 1 , Theorem 3.5.1 allows us to conclude that the series converges
pointwise to f on R.
Let x = 0. Then
∞
2 4  1
.f (0) = − .
π π n=1 4n2 − 1

Since f (0) = 0, we have

∞
 1 1
. = .
n=1
4n2 − 1 2

π
Let x = . Then
2
π 2 4 
∞
1 2
∞
4  (−1)n
.f = − cos(nπ) = − .
2 π π n=1 4n − 1
2 π π n=1 4n2 − 1
3.6. Solved Exercises 313
π
Since f = 1, we have
2
∞
2 4  (−1)n
. −1= ,
π π n=1 4n2 − 1

that is,
∞
(−1)n 1 π
.
2−1
= − .
n=1
4n 2 4

53. Since the function f (x) = x2 (π − x)2 is defined on ]0, π[ and we want to obtain a Fourier cosine series, we consider
an even extension of f to the interval ] − π, π[ (see Fig. 3.62). Since the function is even, the coefficients bn are
zero. Let us calculate a0 and an .

Figure 3.62: The even extension of the function of exercise 53

 π  π  π
2 2 2
a0 = f (x) dx = x2 (π − x)2 dx = (x4 − 2π x3 + π 2 x2 ) dx
π 0 π 0 π 0
 π
2 x5 π x4 π 2 x3 π4
= − + = ;
π 5 2 3 0 15
 π  π
2 2
an = x2 (π − x)2 cos(nx) dx = (x4 − 2π x3 + π 2 x2 ) cos(nx) dx
π 0 π 0
 π  π 
2 sin(nx) sin(nx)
= (x4 − 2π x3 + π 2 x2 ) − (4x3 − 6π x2 + 2π 2 x) dx
π n 0 0 n
.    π 
2 cos(nx) π cos(nx)
= (4x3 − 6π x2 + 2π 2 x) − (12x2 − 12π x + 2π 2 ) dx
πn n 0 0 n
  π  π 
2 2 sin(nx) sin(nx)
= − (12x 2
− 12π x + 2π ) + (24x − 12π) dx
π n2 n 0 0 n
    π 
2 cos(nx) π cos(nx)
= − (24x − 12π) + 24 dx
π n3 n 0 0 n
  π 
2 sin(nx) 24
= −12π cos(nπ) − 12π + 24 = − 4 ((−1)n + 1)
π n4 n 0 n
314 Series of Functions

The Fourier series of f is

∞ ∞
π4 (−1)n + 1 π4 1
. − 24 4
cos(nx) = −3 4
cos(2nx).
30 n=2
n 30 n=1
n

The extension, f˜, of f is a continuous function on [−π, π], and Theorem 3.5.1 allows us to conclude that the
series converges pointwise to f˜ on [−π, π], that is,

∞
π4 1
.f˜(x) = −3 cos(2nx), ∀x ∈ [−π, π].
30 n 4
n=1

The analytical expression of f˜ is

x2 (π + x)2 , if − π ≤ x ≤ 0
.f˜(x) =
x2 (π − x)2 , if 0 ≤ x ≤ π.

Let x = 0. Then
∞
π4 1
.f˜(0) = −3 .
30 n 4
n=1

Since f˜(0) = 0, we have

∞
1 π4
.
4
= .
n=1
n 90

π
Let x = . Then
2
π π4 ∞
1 π4 ∞
(−1)n
.f˜ = −3 cos(nπ) = −3 .
2 30 n 4 30 n4
n=1 n=1

π π4
Since f˜ = , we have
2 16

π π4 π4 ∞
(−1)n π4 ∞
1 ∞
1 − (−1)n
.f˜ − f˜(0) = = −3 − + 3 = 3 ,
2 16 30 n 4 30 n 4 n4
n=1 n=1 n=1

∞ ∞
π4 1 π4 1
that is, =6 , which is equivalent to = .
16 n=1
(2n − 1) 4 96 n=0
(2n + 1)4

∞

54. a) The series n an sin(nx) results from term by term differentiation of the original series. In fact, according
n=1
to Theorem 3.5.8, by term by term differentiation of the series of f˜, we obtain the Fourier series of f˜ , that
is,
 ∞
 ∞   ∞
π 2  (−1)n − 1 2  (−1)n − 1 2  (−1)n − 1
. + cos(nx) = cos(nx) = − n sin(nx).
2 π n=1 n2 π n=1 n 2 π n=1 n2
3.6. Solved Exercises 315

By the same theorem, this series converges to f˜ at the points of continuity of f˜ . At the points of
discontinuity the sum of the series is equal to the arithmetic mean of the left- and right-hand derivatives of
the function f˜ at these points. Thus,

∞
2  (−1)n − 1 f˜ (x), if x = kπ, k ∈ Z
. − n sin(nx) =
π n=1 n2 0, if x = kπ, k ∈ Z.

∞

Therefore, the series n an sin(nx) converges pointwise on R.
n=1
∞

b) Let g(x) = − n an sin(nx). By the previous item,
n=1

f˜ (x), if x = kπ, k ∈ Z

g(x) =
0, if x = kπ, k ∈ Z

. ⎧
⎪
⎪ 1, if x ∈ ]2kπ, (2k + 1)π[, k ∈ Z
⎨
= −1, if x ∈ ](2k + 1)π, (2k + 2)π[, k ∈ Z
⎪
⎪
⎩
0, if x = kπ, k ∈ Z.

In Fig. 3.63 is illustrated the graph of g on the interval [−2π, 2π] .

Figure 3.63: The graph of the function g of exercise 54 b)

c) The function f˜ is continuous and piecewise C 1 on R; therefore,

∞ ∞
π 2  (−1)n − 1 π 4  1  
.f˜(x) = + cos(nx) = − cos (2n − 1)x , ∀x ∈ R.
2 π n=1 n 2 2 π n=1 (2n − 1)2

Let x = 0. Then, as f (0) = 0,

∞
π 4  1
.0 = − ,
2 π n=1 (2n − 1)2

which implies that

∞
 1 π2
. = .
n=1
(2n − 1) 2 8
316 Series of Functions

By Theorem 3.5.7, we have

  ∞  x
x x π 4  1  
. f (t) dt = dt − cos (2n − 1)t dt,
0 0 2 π n=1 (2n − 1) 2
0

that is, if 0 < x < π,

∞
x2 π 4  1  
. = x− sin (2n − 1)x .
2 2 π n=1 (2n − 1)3
π
For x = , we obtain
2
∞
π2 π2 4  (−1)n
. = + ,
8 4 π n=1 (2n − 1)3

that is,
∞
 (−1)n π3
. =− .
n=1
(2n − 1) 3 32

∞

55. Suppose that the function f has a Fourier sine series development, that is, f (x) = Bn sin(nπx) (note that,
n=1
in this case, the two conditions f (0) = f (1) = 0 are satisfied).
Assuming that f and its derivatives meet the conditions of Theorem 3.5.8, we have

∞ 
 ∞
 ∞
 ∞



.f (x) + 3f (x) = Bn sin(nπx) +3 Bn sin(nπx) = − n2 π 2 Bn sin(nπx) + 3 Bn sin(nπx).
n=1 n=1 n=1 n=1

Let g(x) = 3x, 0 < x < 1. By extending g to an odd function, we obtain the Fourier sine series of g on the
interval ] − 1, 1[
∞

.g(x) ∼ bn sin(nπx),
n=1

where
  1  
1 3x 1 3 6 (−1)n
.bn =2 3x sin(nπx) dx = 2 − cos(nπx) + cos(nπx) dx =− .
0 nπ 0 0 nπ nπ

Since g(x) = 3x is a C 1 function on ] − 1, 1[, we have the equality

∞
6 (−1)n+1
.3x = sin(nπx), ∀x ∈ ] − 1, 1[ .
n=1
nπ

Thus,
∞
 ∞
 ∞
6 (−1)n+1
f  (x) + 3f (x) = 3x ⇔ − n2 π 2 Bn sin(nπx) + 3 Bn sin(nπx) = sin(nπx)
n=1 n=1 n=1
nπ
.
∞
 ∞
  6 (−1)n+1
⇔ −n2 π 2 Bn + 3Bn sin(nπx) = sin(nπx).
n=1 n=1
nπ
3.6. Solved Exercises 317

From the uniqueness of the Fourier series, it follows that

6 (−1)n+1
. − n2 π 2 Bn + 3Bn = , ∀n ∈ N,
nπ

that is,
6 (−1)n+1
.Bn = , ∀n ∈ N.
nπ(3 − n2 π 2 )

∞
 6 (−1)n+1
The desired function is f (x) = sin(nπx).
n=1
nπ(3 − n2 π 2 )

56. a) The Fourier coefficients of f are

 2  2  2
x3 8
a0 = f (x) dx = x2 dx = = ;
0 0 3 0 3
 2   2 2
x2 2x
an = x2 cos(nπx) dx = sin(nπx) − sin(nπx) dx
0 nπ 0 0 nπ
 2  2  2 cos(nπx) 
2 2  x
= − x sin(nπx) dx = − − cos(nπx) + dx
nπ 0 nπ nπ 0 0 nπ
  2   
2 2 1 4 2 sin(nπx) 2 4
. = − − + cos(nπx) dx = − = ;
nπ nπ nπ 0 (nπ)2 (nπ)2 nπ 0 (nπ) 2

 2  2  2
x2 2x
bn = x2 sin(nπx) dx = − cos(nπx) − − cos(nπx) dx
0 nπ 0 0 nπ
 2  2  2 sin(nπx) 
4 2 4 2  x
= − + x cos(nπx) dx = − + sin(nπx) − dx
nπ nπ 0 nπ nπ nπ 0 0 nπ
 
4 2 cos(nπx) 2 4
= − + =− .
nπ (nπ)2 nπ 0 nπ

The Fourier series of f is

∞ 
 
4 4 4
. + cos(nπx) − sin(nπx)
3 n=1 (nπ)2 nπ

and converges pointwise to the function

(x − 2k)2 , if 2k < x < 2k + 2, k ∈ Z

.S(x) =
2, if x = 2k, k ∈ Z.

b) Term by term differentiation of the series obtained in item a) results in the series

∞ 
  ∞  
4 4 4
. − n π sin(nπx) − n π cos(nπx) = − sin(nπx) − 4 cos(nπx) ,
n=1
(nπ)2 nπ n=1
nπ

which is divergent because if x ∈ ]0, 2[, the general term of the series does not have a limit. Therefore, the
sum of the series does not exist and cannot define the function f  .
318 Series of Functions

Note that the function f does not verify the conditions of Theorem 3.5.8. In fact, the periodic extension of
f to the interval ] − 1, 1[
(x + 2)2 , if − 1 < x ≤ 0,
.g(x) =
x2 , if 0 < x < 1

is not continuous at 0.

57. a) Let us calculate the Fourier coefficients of f .

 3  3  3
1 1 1 x2 x3
a0 = f (x) dx = (x − x2 ) dx = − = −6;
3 −3 3 −3 3 2 3 −3
 
1
 3  nπx
3(x − x2 )  nπx 3 1
 3
3(1 − 2x)  nπx
an = (x − x2 ) cos sin dx = − sin dx
3 −3 3 nπ 3 3 −3 −3 nπ 3
 3    
1  nπx 1 3(2x − 1)  nπx 3 3 6  nπx
= (2x − 1) sin dx = − cos + cos dx
nπ −3 3 nπ nπ 3 −3 −3 nπ 3
 n     n
1 36(−1) 18 nπx 3 36(−1)
. = − + sin =− ;
nπ nπ (nπ)2 3 −3 (nπ)2
 

1 3  nπx 1 3(x2 − x)  nπx 3  3
3(1 − 2x)  nπx
bn = (x − x2 ) sin dx = cos + cos dx
3 −3 3 3 nπ 3 −3 −3 nπ 3
    
1 3(1 − 2x)  nπx 3 3 6  nπx
= − 6(−1)n − sin − sin dx
nπ nπ 3 −3 −3 nπ 3
  nπx 3 
1 18  6(−1)n
= − 6(−1)n + cos =− .
nπ (nπ)2 3 −3 nπ

The Fourier series of f is

∞   nπx  nπx 
36(−1)n 6(−1)n
. −3+ − 2
cos − sin
n=1
(nπ) 3 nπ 3

and, on the interval ] − 3, 3[, its sum is equal to f because f is continuous and of class C 1 .
b) Since
     
x nπt 3 nπt x 3   nπx
. sin dt = − cos = 1 − cos
0 3 nπ 3 0 nπ 3

and     
  nπx
x nπt 3 nπt x 3
. cos dt = sin = sin ,
0 3 nπ 3 0 nπ 3

the series
3   nπx  nπx
∞ ∞
a0 x 3
. + bn + an sin − bn cos ,
2 n=1
nπ n=1
nπ 3 3
3.6. Solved Exercises 319

where a0 , an , and bn are the Fourier coefficients of f , results from the term by term integration between
0 and x of the series of f . By Theorem 3.5.7, with a = 0 and b = x, the sum of the series is equal to
 x
f (t) dt, that is,
0

∞  x  x  2 x
3 t t3 x2 x3
.S(x) + bn = f (t) dt = (t − t2 ) dt = − = − , x ∈ ] − 3, 3[.
n=1
nπ 0 0 2 3 0 2 3

Therefore, using the result of exercise 50d),

∞
x2 x3 18(−1)n x2 x3 3
.S(x) = − + 2 2
= − − , x ∈ ] − 3, 3[.
2 3 n=1
n π 2 3 2
320 Series of Functions

3.7 Proposed Exercises

∞
1. For each value of x, test the convergence of the se- sin(nx)
∞ 9. Consider the series of functions .
 x 3 n2
ries e 2 +nx and, in case of convergence, find n=1

n=1 a) Show that the series converges uniformly on

its sum.
R, but that there are points in R where the
2. For each value of x, test the convergence of the se- series of derivatives diverges.
∞
 ∞
 sin(nx)
ries (xn + n−3x ). b) Prove that f (x) = is integrable
n=1 n=1
n2
 1
∞ on [0,1] and express f (x)dx as the sum of
xn
3. Considering that ex = , show that 0
n=0
n! a series.
∞
 n
ex − 1 10. Consider the series √ .
. lim = 1. n5 x + 1
x→0 x n=1

a) Study the uniform convergence of the series

4. For each value of x = 0, examine the convergence
∞ on the interval [2, 4].
nα
of the series , α ∈ R, and show that it is b) If f is the sum of the series, compute
n=1
xn  4
uniformly convergent on the interval [2,3]. f (x) dx.
2
∞ 
 1 1
5. Show that the series of functions x n −x n−1 11. Consider f , a real-valued function of a real variable,
∞

n=2 x
converges pointwise but not uniformly on [0,1]. defined by f (x) = , x ∈ R+0 .
n=1
n(x + n)
nn+1 n −nx
6. For each n ∈ N, let fn (x) = x e . a) Prove that f is continuous on [0,1].
n!  1
b) Calculate f (x)dx.
a) Study the pointwise convergence of the series 0
∞
fn on the interval ]1, 2]. ∞
 cos(nx)
n=1 12. Show that the series of functions con-
2n
∞
 n=0
b) Show that the series fn is uniformly con- verges uniformly on R and that the sum is a differ-
n=1 entiable function.
vergent on [2, +∞[. What can be said about
the continuity of the sum of the series on this 13. Consider f , a real-valued function of a real variable,
∞ arctan(nx) sin 1

interval? n2
defined by f (x) = .
∞ n
1 n=1
7. Consider the function f (x) = x
.
n=1
n a) Find the domain of f . Give a clear explanation
of the answer.
a) Determine the domain of f .
b) Prove that f is differentiable on R and write
b) Prove that f is continuous on its domain.
its derivative as the sum of a series.
∞
 enx 14. Indicate the real values of x for which the following
8. Consider the function f (x) = 1 + . Show
n=1
n3 +1 series are absolutely convergent, conditionally con-
 ∞
0 1 − e−n vergent, and divergent. Explain the findings for each
that f (x) dx = 1 + .
−1 n=1
n (n3 + 1) question.
3.7. Proposed Exercises 321

∞
 ∞
 (−1)n
a) n n xn r) √
3
(x − 2)n
n=1 n=0 4n n2 + 1
∞
    n ∞
 3n
1 2n x s) (x − 2)n
b) 1+ · 1 + 9n
n=1
n x+1 n=0
∞

∞   (−1)n
(−1)n 1 − x n t) (x − 2)n
c) 2n (n2
+ 3)
n=1
2n − 1 1 + x n=0

∞ 15. Let a, b ∈ R+ . Determine the values of x for which

 (−1)n n
d) xn the following series are absolutely convergent:
n=1
n
2 (n2 + 7)
∞
 2
∞
(x + 3)n a) an xn , a < 1
e) n=0
(n + 1)2n
n=0 ∞
 xn
∞ √ b)
 (−1)n n + 3 an + bn
f) n n4
(x − 3)n n=0
n=1
5
16. Expand the following functions into a Maclaurin se-
∞
(1 − x2 )n ries and indicate the interval on which the develop-
g) √ ment is valid. Give a justification for the answers
n=0
n+2
provided.
∞  x 3t
 21/n π n e −1
h) (x − 2)n a) ax , a > 0 j) dt
n(n + 1)(n + 2) b) x e x 2
0 t
n=1
∞ 
 n 1 2
(−1)n c) , a=0 k)
i) 1+ (2x + 1)n a 2 + x2 x2 − 4x + 3
n=0
2
d) arctan(x) x+1
∞ l)
 (n + 1)n e) sinh(x) (1 − x)2
j) (x + 1)n  x
n=1
(2 n2 )n/2 f) cosh(x) 1 − cos(t)
1 m) dt
∞
 g) 2x + 0 t2
n 2 + x
k) (x − 1)n  x 1
2n (3 n − 1) 2 n)
n=1 h) e−t dt x2 − 3 x + 2
∞ 0
(−1)n 1 3 x
l) √ (x − 4)n i) o) +
n=1
3n 3n
x2 + x − 2 x−5 (x − 5)2
∞
  n
2n 1 17. Expand the function
m)
n=1
n+2 x−1
∞ .f (x) = x log(1 + x3 )
n3 + 3n
n) (x − 1)n
2n
n=0 into a Maclaurin series and use the development to
∞ prove that the function has a minimum at x = 0.
n+2
o) 2+1
(x + 1)n
n=0
n
18. Consider f , a real-valued function of a real variable,
∞ √  
2n n x + 2 n defined by f (x) = x ex .
p)
n=0
n2 + 1 4
a) Develop f into a Maclaurin series and indicate
∞  
1 1 for which values of x the expansion is valid.
q) n
− n+1 (x − e)n
e e Explain the response given.
n=0
322 Series of Functions

b) Using the series of derivatives from the result 23. Obtain the Maclaurin series of the function
of the previous item, calculate the sum of the
series .f (x) = (1 + x)−2
∞
n+1
. .
n=0
n! by two different processes. What is the radius of
convergence of the series?
19. Consider the function f : R → R defined by
24. Determine two power series that represent the func-
.f (x) = x sin(2x). tion
1
.f (x) =
  2 − x
1 3
a) Write Maclaurin series of f and indicate on the interval , . Justify the answer.
2 2
for which values of x the development
is valid. 25. Develop the function
b) Using the result of the previous item and by
term by term differentiation, compute the sum .f (x) = log(x)
of the series
∞ into a power series of x − 2 and indicate an open in-
 (2n + 2)π 2n+1
. (−1)n . terval on which the function coincides with the sum
n=0
(2n + 1)! of the series.
Justify the answer.
26. Expand the function
20. Consider the function f : R → R defined by
.f (x) = e−x+4 + 3x
.f (x) = sin2 (x) + log(1 − x2 ).
into a power series of x − 1 and indicate the largest
open interval on which the development represents
a) Using term by term integration, develop f
the function.
into a Maclaurin series and indicate the
real values of x for which the development
27. Let f be the function defined by f (x) = x2 log(x2 )
is valid.
on R \ {0}. Develop f into a power series of x − 1
Note: 2 sin(x) cos(x) = sin(2x).
and indicate the largest open interval where this ex-
b) Using the result from the previous item, indi- pansion represents the function.
cate the value of f (15) (0) and f (16) (0).
28. Determine whether the following functions are even
21. Develop the function
or odd and sketch their graphs:
2
.f (x) = a) x2 − x g) x|x| + 3
4x + 5
into a power series of x + 3 and determine the radius b) x2 sin(3x) h) sin2 (x)
of convergence of the series.
i) e−|x|
c) tan(6x)
22. Develop the function
2
x3 e−x
1 d) x4 + 20 cos(5x) j)
.f (x) = cos(x)
x2 − 6x + 5
e) x5 − 2x k) x + 5 cos(x)
into a power series of x−3 and determine the interval
of convergence of the series.
f) sin(x) + cos(2x) l) log(x4 )
3.7. Proposed Exercises 323
⎧ π
29. Sketch the graph of the following functions, check if ⎨0, <x≤0
if −
d) 2
they are periodic, and, if so, indicate a period:
x ⎩π, if 0 < x < 3π
2
a) sin(7x) e) sin + cos(2x) ⎧ 3π
3 ⎨0, if − <x≤0
b) sinh(2x) 2
f) sin (x) e) 2
x ⎩sin(x), if 0 < x < π
c) tan(π x) g) sin cos(x) 2
π x π
d) sin h) x + cos(8x) 33. The following functions are periodic and defined on
4
the interval corresponding to one period. Determine
0, if 2n − 1 ≤ x < 2n, n ∈ Z
i) the Fourier series and the respective sum for each
1, if 2n ≤ x < 2n + 1, n ∈ Z
function.
30. Determine the Fourier series related to the func- a) x, 0 < x < 3
tions periodic with period 2π, defined on the interval
] − π, π[ as follows: b) 2 − x, 0 < x < 2
⎧
⎧ ⎪
⎨−1, if −π <x≤0 ⎨1, if − 2 ≤ x ≤ −1
⎪
a) c) 0, if − 1 < x < 1
⎩1, ⎪
⎪
if 0 < x < π ⎩1, if 1 ≤ x ≤ 2
2
1, if − π < x ≤ 0 x + 1, if − 1 ≤ x ≤ 0
b) d)
3, if 0 < x < π 1 − x, if 0 < x ≤ 1
x2 , if − π < x ≤ 0
c) 0, if − 3 ≤ x ≤ 0
−x2 , if 0 < x < π e)
x2 (3 − x), if 0 < x ≤ 3
d) x + |x|
34. Let L > 0. Develop the following functions into
e) e−|x| Fourier series on the indicated intervals:
f) cos2 (2x) a) |x|, ] − L, L[

31. Show that, on the interval ]0, π[, we have b) ex , ] − L, L[

c) cosh(|x| − πL), ] − 2πL, 2πL[
a) (π − x) sin(x) =
∞ 35. Determine the Fourier sine series and the respective
1  1
=1+ cos(x) − 2 2−1
cos(nx) sum for each function.
2 n=2
n
∞
a) 1, 0 < x < π
 1   ⎧
b) π x (π −x) = 8 sin (2n −1)x ⎪
(2n − 1) 3 ⎪
⎨0, if 0 < x < π
n=1
b) 1, if π < x < 2π
⎪
⎪
32. Considering that the following functions, defined on ⎩2, if 2π < x < 3π
the indicated intervals, are periodic with period 2π,
determine their respective Fourier series and the sum c) 2 − x2 , 0 < x < 2
of the obtained series.
x, if 0 < x ≤ 1
d)
a) | cos(x)|, 0 < x < 2π 2 − x, if 1 < x < 2

b) 3x(π 2 − x2 ), −2π < x < 0 e) x3 , 0<x<1

−2x, if 0 < x ≤ π 36. Determine the Fourier cosine series and its respective
c)
2x, if π < x < 2π sum for each function.
324 Series of Functions

a) x cos(x), 0 < x < π 42. Let f : R → R be a periodic function with period

2π such that
b) x2 − 2x, 0 < x < 4
π, if − π < x < 0
.f (x) =
x, if 0 < x ≤ 1 −x + π, if 0 ≤ x < π.
c)
2 − x, if 1 < x < 2
a) Represent f as the sum of a Fourier series on
37. Let h ∈ ]0, π[. Determine the Fourier cosine series the interval ] − π, π[ and sketch, on R, the
of the function defined by graph of its sum.
∞
1 π2
1, if 0 < x < h b) Show that = .
.f (x) =
n=1
(2n − 1) 2 8
0, if h < x < π.
43. Let f : R → R be a periodic function with period 4
 π such that
38. Let h ∈ 0, . Determine the Fourier cosine series ⎧
2 ⎪
of the function defined by ⎪
⎪ −x − 2, if − 2 < x ≤ −1
⎨
|x|
⎧ .f (x) = , if |x| < 1 and x = 0
⎪ ⎪
⎪
⎪
⎪ 1, if x = 0 ⎪ x
⎩
⎨ −x + 2, if 1 ≤ x < 2
2h − x
.f (x) = , if 0 < x < 2h
⎪
⎪ 2h
⎪
⎩0, and S(x) be the Fourier series associated with f .
if 2h < x < π,
a) Without calculating S(x), indicate the value
where f is a periodic function with period 2π. of S(612345 ) and S(65). Provide a clear ex-
π x planation of the answer.
39. Let f (x) = − , x ∈ ]0, π[.
4 2 b) Determine S(x).
a) Develop f into a Fourier sine series. 44. Let f be the function defined by
b) Develop f into a Fourier cosine series. ⎧π
⎪
⎨ , if − π < x < 0
.f (x) =
4
40. Let f : R → R be a function such that ⎪ π
⎩ − x, if 0 ≤ x < π.
⎧ 4
⎪
⎨1, if 0 ≤ x ≤
1
a) Determine the Fourier series of f .
.f (x) = 2
⎪ 1 b) Let
⎩−1, if < x ≤ 1.
2 ∞
a0 x  1  
.S(x) = + an sin(nx)−bn cos (nx) ,
a) Determine the Fourier cosine series of the 2 n=1 n
function f on [0, 1] and indicate the values where a0 , an , and bn are the Fourier coeffi-
for which the series converges. Explain the cients of f . Determine the expression of S(x),
reasoning behind the answer. x ∈ ] − π, π[.
b) Sketch the graph of the sum of the Fourier
45. Use the Fourier series of
cosine series of f . ⎧
⎨−1, if − 3 ≤ x < 0
41. a) Determine the Fourier cosine series of the .f (x) =
⎩1, if 0 ≤ x < 3
function f (x) = sin(x), 0 ≤ x < π.
b) Determine the Fourier sine series of the func- to determine the Fourier series of
⎧
tion g(x) = cos(x), 0 < x < π. ⎪
⎨−x − 3, if − 3 ≤ x < 0
.F (x) =
c) Can the function f be developed into a Fourier ⎪
⎩x − 3, if 0 ≤ x < 3.
cosine series on −π ≤ x ≤ π?
3.7. Proposed Exercises 325

46. Using the equality 47. Determine, using Fourier series, the function f such
that
∞
(−1)n+1 ⎧
.x =2 sin(nx), −π < x < π, ⎪
n ⎨f  (x) + 10f (x) = 5x, −2 < x < 2
n=1
.
⎪
⎩f (−2) = f (2) = 0.
obtain the Fourier series development of f (x) = x4 .
 Answers to
Proposed Exercises

Chapter 1
1
3. a) .− 15 c) 1 e) .− 16. a) True b) False c) False
4
b) .+∞ d) 0
17. a) .lim un = +∞; .lim un = 0
4. a) 0 c) 0 e) 1
b) .lim un = 1; .lim un = −1
1
b) .−1 d) . c) .lim un = +∞; .lim un = −1
2
√
5. c) .nn , .n!, .en , .2n , .n3 , .2 n, . 10 n, .log(n) d) .lim un = 1; .lim un = −1
1 1 1 1 1 1 1 1 e) .lim un = +∞; .lim un = 0
6. . ,. ,. ,. ,. ,. , .√ ,.
nn n! en 2n n3 2 n 10 n log(n)
f) .lim un = 1; .lim un = 0
7. a) .e4 c) .+∞ g) .lim un = +∞; .lim un = −∞
 
b) 0 d) 1 π 3π
h) If .a ∈ + 2kπ, + 2kπ , .k ∈ Z,
1 4 4 4
8. a) . b) . .lim un = sin(a); .lim un = − sin(a)
2e e  
5π 7π
If .a ∈ + 2kπ, + 2kπ , .k ∈ Z,
1 4 4
9. .p = .lim un = − sin(a); .lim un = sin(a)
3e  π π 
If .a ∈ − + 2kπ, + 2kπ , .k ∈ Z,
10. a) 1 if .x = 2kπ, .k ∈ Z 4 4
limit does not exist if .x = (2k + 1)π, .k ∈ Z .lim un = cos(a); .lim un = − cos(a)
 
0 if .x ∈ R \ {kπ, k ∈ Z} 3π 5π
If .a ∈ + 2kπ, + 2kπ , .k ∈ Z,
4 4
b) 0 e) 2 h) .+∞
.lim un = − cos(a); .lim un = cos(a)
c) .+∞ 1
f) . i) 0
2 e i) .lim un = +∞; .lim un = 0
d) . g) .+∞ j) 1
e j) .lim un = +∞; .lim un = −∞
13. .−1
 n
√ 8
15. c) . 2 22. a) .
9

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 327
A. Alves de Sá, B. Louro, Sequences and Series,
https://doi.org/10.1007/978-3-031-67202-6
328 Answers to Proposed Exercises

Chapter 2
1 1 3
1. a) . = ;. h) Convergent or divergent, depending on the se-
(n + 1)2 − 1 n(n + 2) 4
ries
1 1
b) . ;.
n(n + 1)(n + 2) 4 9. a) Convergent e) Divergent
1 1
c) . ;. b) Divergent f) Convergent
n(n + 1)(n + 2)(n + 3) 18

2. a) .−1 c) Convergent g) Divergent

1 d) Convergent h) Divergent
b) .− cot(a) +
a
c) 1 10. a) Convergent d) Convergent
17
d) .− b) Convergent e) Divergent
4
e) .−e−1 c) Convergent f) Convergent
6
f) . 11. a) Convergent e) Divergent
7
1
g) . b) Divergent f) Convergent
arctan(2)
c) Convergent g) Convergent
(x + 1)3
4. a) .x ∈ ] − ∞, −3 [ ∪ ]1, +∞ [ ; .
(x + 1)3 − 8 d) Convergent h) Convergent
|x| − 1
b) .x ∈ ] − 2, 0[ ∪ ]0, 2 [ ; . 12. a) Convergent e) Convergent
2 − |x|
x b) Divergent
c) .x ∈ ] − 1, 1 [ ; .
1 + x2 f) Convergent
c) Convergent
6. a) Conditionally convergent
d) Divergent g) Convergent
b) Absolutely convergent
c) Divergent 13. a) Convergent c) Divergent

98
(−1)n b) Convergent d) Divergent
7. b) .S98 =
n=1
(n + 1)2 14. a) Divergent

8. a) Convergent b) Absolutely convergent

b) Convergent or divergent, depending on the se- c) Divergent

ries d) Absolutely convergent
c) Convergent e) Absolutely convergent

d) Divergent f) Absolutely convergent

e) Convergent or divergent, depending on the se- g) Absolutely convergent

ries h) Conditionally convergent
f) Divergent i) Divergent

g) Convergent or divergent, depending on the se- j) Divergent

ries k) Absolutely convergent
Answers to Proposed Exercises 329

l) Absolutely convergent r) Divergent

m) Divergent s) Absolutely convergent
n) Divergent t) Absolutely convergent
o) Absolutely convergent u) Absolutely convergent
p) Absolutely convergent
16. a) Absolutely convergent if
q) Divergent .(2k + 1)π < α < (2k + 2)π, .k ∈ Z
r) Absolutely convergent diverges if .2kπ ≤ α ≤ (2k + 1)π, .k ∈ Z
s) Conditionally convergent b) Absolutely convergent if .α > 1
conditionally convergent if .0 < α ≤ 1
15. a) Absolutely convergent divergent if .α ≤ 0
b) Absolutely convergent
17. a) Absolutely convergent
c) Divergent
b) 0
d) Conditionally convergent
e) Conditionally convergent 19. Convergent

f) Absolutely convergent 20. a) Divergent

g) Absolutely convergent b) Convergent
h) Absolutely convergent
23. a) Absolutely convergent
i) Absolutely convergent
b) i. Absolutely convergent if .β − α > 1
j) Absolutely convergent
divergent if .β − α ≤ 1
k) Conditionally convergent ii. It is a sum with a finite number of terms.
3
l) Absolutely convergent if .a > − ; 
2 n
(−1)k 1
2 25. a) . · √
3 k! (n − k + 1) n − k + 1
divergent if .a ≤ − n=0 k=0
2 1 1
.= √ +
m) Absolutely convergent 3 3 2
n) Divergent b) Convergent
o) Divergent 8
26. .
p) Absolutely convergent 7
q) Absolutely convergent 27. 32
330 Answers to Proposed Exercises

Chapter 3
x +x3 −1
1. Convergent on .] − ∞, 0[, .S = e 2 1 − ex
3
, e) Absolutely convergent on .] − 5, −1[
divergent on .[0, +∞[ Conditionally convergent at .x = −5
Divergent on .] − ∞, −5[ ∪ [−1, +∞[
 
1
2. Convergent on . , 1 , f) Absolutely convergent on .[−2, 8]
 3  Divergent on .] − ∞, −2[ ∪ ]8, +∞[
1 √ √
divergent on . − ∞, ∪ [1, +∞[ g) Absolutely convergent on .] − 2, 2[\{0}
3 √
Conditionally convergent at .x = − 2 and at
√
4. Absolutely convergent on .] − ∞, −1[ ∪ ]1, +∞[ and .x = 2
√ √
divergent on .] − 1, 1[, .∀α ∈ R Divergent on .] − ∞, − 2[ ∪ ] 2, +∞[ ∪ {0}
if .x = 1: absolutely convergent if .α < −1 and diver-  
1 1
gent if .α ≥ −1 h) Absolutely convergent on . 2− , 2+
π π
if .x = −1: absolutely convergent if .α < −1, condi-    
1 1
tionally convergent if .−1 ≤ α < 0 and divergent if Divergent on . −∞, 2 − ∪ 2 + , +∞
π π
.α ≥ 0  
5 1
i) Absolutely convergent on . − , −
6. a) Convergent b) Is continuous   6 6 
5 1
Divergent on . −∞, − ∪ − , +∞
6 6
7. a) .]1, +∞[ √ √
j) Absolutely convergent on .] − 2 − 1, 2 − 1[
√ √
∞ Divergent on .] − ∞, − 2 − 1] ∪ [ 2 − 1, +∞[
1 − cos(n)
9. b) .
n=1
n3 k) Absolutely convergent on .] − 1, 3[
Divergent on .] − ∞, −1] ∪ [3, +∞[
10. a) Converges uniformly
l) Absolutely convergent on .]1, 7[
∞ √ √ 
 2 4n5 + 1 − 2n5 + 1 Conditionally convergent at .x = 7
b) .
n4 Divergent on .] − ∞, 1] ∪ ]7, +∞[
n=1
m) Absolutely convergent on
∞ 
  
1 n+1 .] − ∞, −1[ ∪ ]3, +∞[
11. b) . − log
n=1
n n Conditionally convergent at .x = −1
Divergent on .] − 1, 3] \ {1}
1  
∞ sin
 1 5
13. a) .R b) .
n2 n) Absolutely convergent on . ,
1 + (nx)2 3 3
n=1    
1 5
Divergent on . −∞, ∪ , +∞
14. a) Divergent on .R \ {0} 3 3
  o) Absolutely convergent on .] − 2, 0[
1
b) Absolutely convergent on . − , +∞ Conditionally convergent at .x = −2
  2
1 Divergent on .] − ∞, −2[ ∪ [0, +∞[
Divergent on . −∞, − \ {−1}
2 p) Absolutely convergent on .[−4, 0]
c) Absolutely convergent on .]0, +∞[ Divergent on .] − ∞, −4[ ∪ ]0, +∞[
Conditionally convergent at .x = 0 q) Absolutely convergent on .]0, 2e[
Divergent on .] − ∞, 0[ \{−1} Divergent on .] − ∞, 0] ∪ [2e, +∞[
d) Absolutely convergent on .] − 2, 2[ r) Absolutely convergent on .] − 2, 6[
Conditionally convergent at .x = 2 Conditionally convergent at .x = 6
Divergent on .] − ∞, −2] ∪ ]2, +∞[ Divergent on .] − ∞, −2] ∪ ]6, +∞[
Answers to Proposed Exercises 331

∞
s) Absolutely convergent on .] − 1, 5[ n−3 n
o) . x , .|x| < 5
Divergent on .] − ∞, −1] ∪ [5, +∞[ n=0
5n+1
t) Absolutely convergent on .[0, 4]
∞
Divergent on .] − ∞, 0[ ∪ ]4, +∞[ (−1)n 3n+4
17. . x , .|x| < 1
n=0
n+1
15. a) Absolutely convergent on .R
b) Absolutely convergent on ∞
xn+1
.] − a, a[, if .a ≥ b
18. a) . , .∀x ∈ R
n=0
n!
.] − b, b[, if .b > a
b) .2 e
∞ n
 log(a) ∞
16. a) . xn , for all .x ∈ R  (−1)n 22n+1 2n+2
n! 19. a) . x , .∀x ∈ R
n=0 (2n + 1)!
n=0
∞
x2n+1
b) . , for all .x ∈ R b) .−π
n=0
n!
∞ ∞ 
 
(−1)n 2n (−1)n 22n+1 1
c) . x , .|x| < |a| 20. a) . − x2n+2 , .|x| < 1
n=0
a2n+2 n=0
(2n + 2)! n+1
∞
(−1)n 2n+1 b) .f (15) (0) = 0; .f (16) (0) = −215 − 2 × 15!
d) . x , .|x| < 1
n=0
2n +1
∞
∞
 −22n+1
1 21. . (x + 3)n , .− 19 < x < − 45 , .r = 7
e) . x2n+1 , for all .x ∈ R 7n+1 4 4
n=0
(2n + 1)! n=0

∞
1 ∞

f) x2n , for all .x ∈ R −1
.
(2n)! 22. . (x − 3)2n , .1 < x < 5
n=0
n=0
22n+2
∞ n
 log(2) (−1)n
g) . + n+1 xn , .|x| < 2 ∞

n! 2
n=0 23. . (−1)n (n + 1) xn , .|x| < 1, .r = 1
∞
(−1)n n=0
h) . x2n+1 , for all .x ∈ R
n!(2n + 1) ∞
n=0 
∞
   24. . (x − 1)n , .0 < x < 2
1 (−1)n+1
i) . − 1 xn , .|x| < 1 n=0
n=0
3 2n+1
∞

∞ 1
 3n n . xn , .|x| < 2
j) . x , for all .x ∈ R n=0
2n+1
n=1
n! n
∞ 
  ∞
1  (−1)n
k) . 1 − n+1 xn , .|x| < 1 25. .log(2) + (x − 2)n+1 , .0 < x < 4
n=0
3
n=0
(n + 1)2n+1
∞

l) . (2n + 1) x , .|x| < 1
n
∞
e3 (−1)n + 3 logn (3)
n=0 26. . (x − 1)n , .∀x ∈ R
∞
n!
(−1)n+1
n=0
m) . x2n−1 , .∀x ∈ R
(2n)!(2n − 1) ∞

n=1 (−1)n (x − 1)n+3
∞   27. .2(x − 1) + 3(x − 1)2 + 4 ,
 1 (n + 1)(n + 2)(n + 3)
n) . 1 − n+1 xn , .|x| < 1 n=0
2 .0 <x<2
n=0
332 Answers to Proposed Exercises

28. a) Neither even nor odd f) Neither even nor odd

b) Odd

g) Neither even nor odd

c) Odd

h) Even

d) Even

i) Even

e) Odd
Answers to Proposed Exercises 333

j) Odd c) Periodic with period 1

k) Neither even nor odd

d) Periodic with period 8

l) Even
e) Periodic with period .6π

2π f) Periodic with period .π

29. a) Periodic with period .
7

g) Not periodic
b) Not periodic
334 Answers to Proposed Exercises

∞  
h) Not periodic  cos (2n − 1)x sin (2n − 1)x
c) .π + 8 −
n=1
(2n − 1)2 π 2n − 1
⎧
⎪
⎪ −2(x − 2kπ),if x ∈ ]2kπ, (2k + 1)π[
⎪
⎪
⎨0, if x = (2k + 1)π
.
⎪2(x − 2kπ), if
⎪ x ∈ ](2k + 1)π, (2k + 2)π[
⎪
⎪
⎩
2π, if x = 2kπ

.k ∈Z
∞ 
3π  sin 3nπ
2
i) Periodic with period 2 d) . + cos(nx)
4 n=1
n
3nπ

2 sin2 4
.+ sin(nx)
n
⎧
⎪
⎪ if x ∈ ](4k − 1) π2 , 2kπ[
⎨0,
. π, if x = (4k − 1) π2 ∨ x = 2kπ
⎪⎪2
⎩π, if x ∈ ]2kπ, (4k + 3) π2 [
∞
1 3  1 
30. a) .− + sin (2n − 1)x .k∈Z
4 π n=1 2n − 1
1 1 1
∞ e) . + cos(x) + sin(x)
4  1  2π 2π 4
b) .2 + sin (2n − 1)x
π n=1 2n − 1 ∞ 
1  n sin 2 − 1
nπ
∞ .+ cos(nx)
2  2 + (−1)n (π 2 n2 − 2) π n=2 n2 − 1
c) . sin(nx) 
π n=1 n3
n cos nπ
.−
2
∞  sin(nx)
π  (−1)n − 1 (−1)n n2 − 1
d) . +2 2
cos(nx)− sin(nx) ⎧
2 πn n
⎪
⎨sin(x), if x ∈ [2kπ, (4k + 1) 2 [
n=1 π
⎪
∞
1 − e−π 2  1 − (−1)n e−π . 1, if x = (4k + 1) 2π
e) . + cos(nx) ⎪
⎪ 2
π π n=1 n2 + 1 ⎩0, if x ∈ ](4k + 1) π , (2k + 2)π[
2
1 1 .k ∈Z
f) . + cos(4x)
2 2
∞
3 3  1 2nπx
33. a) . − sin
∞ 2 π n=1 n 3
2 4  (−1)n
32. a) . − cos(2nx) 
π π n=1 4n2 − 1 x − 3k, if x ∈]3k, 3k + 3[
.
3
.| cos(x)|, .∀x
∈R 2
, if x = 3k
∞ 
 2π 
n2 π 2 − 2 .k ∈Z
b) .3π 3 +18 cos(nx)+ sin(nx) ∞
n=1
n2 n3 2  1
b) .1 + sin(nπx)
⎧  π n=1 n
⎪
⎪3(x − 2kπ) π 2 − (x − 2kπ)2 ,
⎨ 
. if x ∈ ](2k − 2)π, 2kπ[ 2 − (x − 2k), if x ∈ ]2k, 2k + 2[
⎪
⎪
⎩9π 3 , if x = 2kπ
.
1, if x = 2k

.k ∈Z .k ∈Z
Answers to Proposed Exercises 335

∞   ⎧
2  (−1)n (2n − 1)πx ⎪
⎪−2, if x ∈ ] − 3π + 6kπ, −2π + 6kπ[
1 ⎪
c) . + cos ⎪
2 π n=1 2n − 1 2 ⎪
⎪
⎪
⎪ −1, if x ∈ ] − 2π + 6kπ, −π + 6kπ[
⎧ ⎪
⎪
⎪
⎪ if x ∈ ] − π + 6kπ, π + 6kπ[
⎪ ⎪
⎪ 0,
⎨1, if x ∈ ]1 + 4k, 3 + 4k[
⎪ ⎪
⎪
⎪
⎪
⎪ 1, if x ∈ ]π + 6kπ, 2π + 6kπ[
. 0, if x ∈ ]3 + 4k, 5 + 4k[ ⎪
⎪
⎪
⎪ ⎨2,
⎩ 1 , if x = 2k + 1 if x ∈ ]2π + 6kπ, 3π + 6kπ[
2 .
⎪1,
⎪ if x = π + 6kπ
⎪
⎪ 2
.k ∈Z ⎪
⎪− 1 ,
⎪
⎪ if x = −π + 6kπ
∞ ⎪
⎪ 2
1 4  1  ⎪3
⎪
d) . + 2 cos (2n − 1)πx ⎪
⎪ , if x = 2π + 6kπ
2 π n=1 (2n − 1) 2 ⎪
⎪
2
⎪
⎪ − 23 , if x = −2π + 6kπ
⎪
⎪
 ⎪
⎩
x − 2k + 1, if x ∈ [2k − 1, 2k] 0, if x = 3π + 6kπ
.
−x + 2k + 1, if x ∈ [2k, 2k + 1] .k ∈Z
∞  
.k ∈Z 4  (−1)n + 1 4((−1)n −1) nπx
c) . − sin
∞  π n=1 n n3 π 2 2
9 27  6 + (−1)n (n2 π 2 − 6) nπx
e) .− 4 cos ⎧
8 π n=1 n4 3 ⎪
⎪2 − (x − 4k)2 , if x ∈ ]4k, 4k + 2[
⎨

2 π(1 + 2(−1)n ) nπx . (x − 4k)2 − 2, if x ∈ ]4k − 2, 4k[
.+ sin ⎪
⎪
n3 3 ⎩0, if x = 2k

(x − 6k)2 (3 + 6k − x),if x ∈ [6k, 3 + 6k] .k ∈Z
.
∞  
0, if x ∈ [3 + 6k, 6 + 6k] 8  (−1)n+1 (2n − 1)πx
d) . 2 sin
.k ∈Z π n=1 (2n − 1)2 2

∞   x − 4k, if x ∈ [−1 + 4k, 1 + 4k]
L 4L  1 (2n − 1)πx
34. a) . − 2 cos .
2 π n=1 (2n − 1)2 L 2 − (x − 4k), if x ∈ [1 + 4k, 3 + 4k]
∞
 .k ∈Z
1 L(−1)n nπx
∞
b) .sinh(L)
L
+2
L 2 + n2 π 2
cos
L 2  6 − n2 π 2
n=1 e) . 3 (−1)n sin(nπx)
n  π n=1 n3
nπ(−1) nπx
.− sin 
L2 + n2 π 2 L (x − 2k)3 , if x ∈ ]2k − 1, 2k + 1[
.
∞

1 2L 1 nx 0, if x = 2k + 1
c) .sinh(πL) + cos
πL π L2 + n2 L
n=1 .k ∈Z
∞
4 
∞
1  2 π 4  4n2 + 1
35. a) sin (2n − 1)x 36. a) .− + cos(x) − cos(2nx)
.
π n=1 2n − 1 π 2 π n=1 (4n2 − 1)2
⎧ 
⎪ if x ∈ ]2kπ, (2k + 1)π[ −(x − 2kπ) cos(x), if x ∈ [(2k − 1)π, 2kπ]
⎪
⎨1, .
. −1, if x ∈ ](2k + 1)π, (2k + 2)π[ (x − 2kπ) cos(x), if x ∈ [2kπ, (2k + 1)π]
⎪
⎪
⎩0, if x = kπ .k ∈Z
∞
.k ∈Z 4 16  1 + 3(−1)n nπx
b) . + 2 cos
∞   3 π n=1 n2 4
4  cos − (−1)n
nπ nπ
2
cos 6 nx
b) . sin
π n=1 n 3 .(x − 8k)2 − 2|x − 8k|, if .x ∈ [8k − 4, 8k + 4]
.k ∈ Z
336 Answers to Proposed Exercises

∞  ∞
1 4  1 + (−1)n − 2 cos 4 
nπ
nπx 2 1
c) . − 2 2
cos 41. a) . + cos(2nx)
2 π n=1 n2 2 π π n=1 1 − 4n2
∞
1
∞
4  1  8  n
.= − 2 cos (2n − 1)πx b) . sin(2nx)
2 π n=1 (2n − 1)2 π n=1 4n2 − 1
 c) No
x − 2k, if x ∈ [2k, 2k + 1]
.
2 − x + 2k, if x ∈ [2k + 1, 2k + 2] ∞  
3π  1 − (−1)n (−1)n
42. a) . + 2
cos(nx)+ sin(nx)
.k ∈Z 4 n=1 πn n
∞
h 2  sin(nh)
37. . + cos(nx)
π π n=1 n
∞
h 2  sin2 (nh)
38. . + cos(nx)
π πh n=1 n2
43. a) .S(612345 ) = 0, .S(65) = 1
∞
 sin(2nx) ∞  
39. a) . 2  1 2 nπ nπx
2n b) . + 2 sin sin
n=1 π n=1 n n π 2 2
∞ 
2  cos (2n − 1)x
b) . ∞ 
 
π n=1 (2n − 1)2 1 − (−1)n (−1)n
44. a) . cos(nx)+ sin(nx)
∞ πn2 n
4  (−1)n+1  n=1
40. a) . cos (2n − 1)πx
π n=1 2n − 1 ⎧
⎪
⎪ π π2
⎧ ⎨ x+ , if − π < x < 0
⎪
⎪ 1, if x ∈ ] − 1
+ 2k, 1
+ 2k[ b) .S(x) = 4 12
⎪
⎨ 2 2 ⎪
⎪ 2 π2
⎩πx − x
+ , if 0 ≤ x < π
. −1, if x ∈ ] 12 + 2k, 3
+ 2k[ 4 2 12
⎪
⎪ 2
⎪
⎩
0, if x = 2k+1 ∞
  
2 4 (2n − 1)πx
45. .f ∼ sin
.k ∈Z n=1
(2n − 1)π 3
∞
  
3 12 (2n − 1)πx
.F (x) =− − cos
b) 2 n=1 (2n − 1) π
2 2 3

∞
π4 (−1)n 2 2
46. .x4 = +8 (π n − 6) cos(nx)
5 n=1
n4

∞
80  (−1)n nπx
47. .  sin
π n=1 n (nπ)2 − 40 2
 Bibliography

1. Anton, H., Bivens, I., Davis, S.: Calculus, 12th edn. Wiley, Hoboken, New Jersey (2022)

2. Apostol, T.: Calculus, 2nd edn. Wiley, New Delhi (2003)

3. Bourchtein, L., Bourchtein, A.: Theory of Infinite Sequences and Series. Birkhäuser, Cham, Switzer-
land (2022)

4. Ellis, R., Gulick, D.: Calculus with Analytic Geometry, 5th edn. Saunders College Publishing,
Harcourt Brace College Publishers, Fort Worth (1994)

5. Hunt, R.A.: Calculus, 2nd edn. HarperCollins College Publishers, New York (1994)

6. Kreyszig, E.: Advanced Engineering Mathematics, 10th edn. Wiley, New York (2011)

7. Laczkovich, M., Sós, V.T.: Real Analysis, Series, Functions of Several Variables, and Applications.
Springer, New York (2017)

8. Larson, R., Hostetler, R., Edwards, B.H.: Calculus with Analytic Geometry, 7th edn. Houghton
Mifflin, Boston (2001)

9. Little, C.H.C., Teo, K.L., Van Brunt, B.: Real Analysis via Sequences and Series. Springer, New
York (2015)

10. Pinkus, A., Zafrany, S.: Fourier Series and Integral Transforms. Cambridge University Press, Cam-
bridge (1997)

11. Mcquarrie, D.A.: Mathematical Methods for Scientists and Engineers, University Science Books,
Sausalito, California (2003)

12. Salas, S., Hille, E., Etgen, G.: Calculus, One and Several Variables, 10th edn. Wiley, Hoboken,
New Jersey (2007)

13. Spivak, M.: Calculus, 4th edn. Publish or Perish, Inc, Houston, Texas (2008)

14. Stewart, J.: Calculus, 9th edn. Cengage Learning, Boston, MA (2021)

15. Stewart, J.: Calculus, Concepts and Contexts, 5th edn. Cengage Learning, Boston, MA (2023)

16. Taylor, A., Mann, R.: Advanced Calculus, 3rd edn. Wiley, New York (1983)

17. Tolstov, G.P.: Fourier Series. Dover Publications, New York (1962)

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 337
A. Alves de Sá, B. Louro, Sequences and Series,
https://doi.org/10.1007/978-3-031-67202-6
 Index

Symbols E
.Int(x), integer part of x, viii Euler’s constant, 30
.R, extended real line, 15 Euler, Leonhard, 29
extended real line, 15
A
Alternating series, 82 F
arithmetic progression, see arithmetic sequence Fibonacci
arithmetic sequence, 6 sequence, 2
common difference, 6 Fibonacci, Leonardo, 2
definition, 6 Fourier series
general term, 7 coefficients, 205, 223
sum of the first n terms, 7 convergence, 225
cosine series, 216
definition, 205
B
differentiation term by term, 226
bounded, viii
even function, ix
bounded from above, viii
integration term by term, 226
bounded from below, viii
odd function, ix
period, 201
C periodic extension, 206
Cauchy periodic function, 201
Cauchy product, 113 piecewise continuous, 211
Cauchy’s Criterion, 78 piecewise .C 1 , 212
Cauchy’s Root Test, 106 pointwise convergence, 212
Condensation Test, 90 sine series, 217
sequence, 35 trigonometric polynomial, 202
Cauchy, Augustin Louis, 36 trigonometric series, 202
coefficients, 202
D uniform convergence, 225
D’Alembert’s Test, 102 Fourier, Joseph, 205
D’Alembert, Jean Le Rond, 102 function
difference of sequences, 4 even, ix, 214
Dirichlet’s Test, 83 odd, ix, 214
Dirichlet, Peter Gustave Lejeune, 83 of class .C 1 , ix

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 339
A. Alves de Sá, B. Louro, Sequences and Series,
https://doi.org/10.1007/978-3-031-67202-6
340 Index

of class .C p , ix Method of Mathematical Induction, viii

piecewise continuous, 211 minimum, viii
piecewise .C 1 , 212

G
N
Gauss, Carl Friedrich, 7
neighborhood
General Comparison Test, 95
of a, vii, 15
general term of a sequence, 1
of .+∞, 15
geometric progression, see geometric sequence
geometric sequence, 8 of .−∞, 15
common ratio, 8 Neper’s number, 30
definition, 8 Neper, Jhone, 30
general term, 9 Newton Binomial, viii
sum of the first n terms, 9 Numerical series
absolutely convergent, 87
I alternating harmonic, 86
indeterminate forms, ix alternating series, 82
infimum, viii associative property, 80
integer part, viii
Cauchy Condensation Test, 90
Integral Test, 93
Cauchy product, 113
interval, vii
Cauchy’s Criterion, 78
K Cauchy’s Root Test, 106
Kummer’s Test, 109 conditionally convergent, 87
Kummer, Ernst Eduard, 109 convergent, 69
definition, 69
L divergent, 69
L’Hôpital’s Rule, ix General Comparison Test, 95
Leibniz’s Test, 82 general term, 69
Leibniz, Gottfried Wilhelm, 82 geometric, 70
lower bound, viii
harmonic, 71
Integral Test, 93
M
Maclaurin Kummer’s Test, 109
binomial series, 194 p-series, 91
Maclaurin series of .sin(x), 196 partial sums, 69
Maclaurin series of .ex , 197 product of series, 113
Maclaurin, Collin, 194 Ratio Test, 101
maximum, viii rearrangement, 87
Index 341

remainder of order p, 79 pointwise convergence, 167

Root Test, 106 uniform convergence, 169
sum, 69 Sequence of real numbers, 1
telescopic, 74 arithmetic progression, see arithmetic se-
terms of the series, 69 quence
Theorem of Mertens, 115 arithmetic sequence, 6
common difference, 6
P definition, 6
period, 201 general term, 7
periodic extension, 206 sum of the first n terms, 7
periodic function, 201 bounded, 2
piecewise continuous, 211 bounded above, 2
piecewise .C 1 , 212 bounded below, 2
Power series Cauchy sequence, 35
absolute convergence, 186 convergent, 13
convergence at endpoints, 189 decreasing, 4
definition, 186 definition, 1
divergence, 186 divergent, 13
interval of convergence, 187 Fibonacci, 2
Maclaurin series, 193 general infinite limit, 11
radius of convergence, 187 general term, 1
Taylor series, 193 geometric progression, see geometric se-
term by term differentiation, 191 quence
term by term integration, 190 geometric sequence, 8
uniform convergence, 190 common ratio, 8
product of sequences, 4 definition, 8
general term, 9
Q
sum of the first n terms, 9
quotient of sequences, 4
graph, 1
R increasing, 4
Raabe’s Test, 111 indeterminate form
Raabe, Joseph L., 111 division by the highest power, 41
Ratio Test, 101 inferior limit, 35
Root Test, 106 infinite limit, 11
limit, 13
S lower limit, 35
Sequence of functions, 167 monotonic, 4
convergence at a point, 167 null sequence, 17
limit function, 167 operations with limits, 16
342 Index

range, 1 divergence, 186

recursive relation, 2 interval of convergence, 187
Squeeze Theorem, 19 radius of convergence, 187
strictly decreasing, 4 term by term differentiation, 191
strictly increasing, 4 term by term integration, 190
strictly monotonic, 4 uniform convergence, 190
sublimit, 33 sum function of the series, 172
subsequence, 31 Taylor series, 193
superior limit, 35 uniform convergence, 174
tends to, 11, 13 Weierstrass’ Test, 175
terms, 1 sum of sequences, 4
uniqueness of the limit, 16 supremum, viii
upper limit, 35
Sequences of real numbers T
difference, 4 Taylor series, 193
product, 4 Maclaurin series, 193
quotient, 4 uniqueness, 197
sum, 4 Taylor’s formula, 193
Series of functions Lagrange remainder, 193
convergence at a point, 172 Taylor, Brook, 193
differentiable term by term, 183 triangle inequality, vii
Fourier series, 201 trigonometric polynomial, 202
coefficients, 205 trigonometric series, 202
definition, 205 coefficients, 202
periodic extension, 206
U
general term, 172
upper bound, viii
integrable term by term, 180, 183
Maclaurin series, 193 W
partial sum, 172 Weierstrass’ Test, 175
pointwise convergence, 172 Weierstrass, Karl Theodor Wilhelm, 175
power series, 186
absolute convergence, 186 Z
convergence at endpoints, 189 Zeno, 68
definition, 186 Zeno’s Paradoxes, 68