Sharpening Mathematical Analysis Skills: Alina Sîntămărian Ovidiu Furdui

Problem Books in Mathematics
Alina Sîntămărian
Ovidiu Furdui
Sharpening
Mathematical
Analysis Skills
Series Editor:
Peter Winkler
Department of Mathematics
Dartmouth College
Hanover, NH
USA
More information about this series at http://www.springer.com/series/714
Alina Sîntămărian • Ovidiu Furdui
Sharpening Mathematical
Analysis Skills
Alina Sîntămărian Ovidiu Furdui
Department of Mathematics Department of Mathematics
Technical University of Cluj-Napoca Technical University of Cluj-Napoca
Cluj-Napoca, Romania Cluj-Napoca, Romania
ISSN 0941-3502 ISSN 2197-8506 (electronic)

ISBN 978-3-030-77138-6 ISBN 978-3-030-77139-3 (eBook)
https://doi.org/10.1007/978-3-030-77139-3
Mathematics Subject Classification: 26-XX, 26Axx, 26Bxx, 26Dxx, 26D15, 40-XX, 40Axx, 40Gxx,
41A58, 11M06, 11M35, 34A05, 45A05
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland
AG 2021
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 translation, 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 information 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
Gold, when multiplied, conspire against his
master; but books, when multiplied, make
great use of those who have them.
—Saint John Chrysostom (347–407)
In the memory of my father, Ioan
and
To my mother, Viorica
Alina
Preface
The library is the friend that all those who

want “to bring something new” must make.
However, it cannot be revealed
—an epsilon no matter how small—
if you do not see and accumulate
what others have done.
Viorel Gh. Vodă
This book is the fruit of our work in the last decade teaching, researching,
and solving and creating problems. Our material offers an unusual collection of
problems of mathematical analysis that someone would study in a standard calculus
course: limits, numerical and power series, derivatives, partial derivatives and their
applications, and implicit functions. Working in a specific field of mathematics, new
ideas come, but in order not to make the book too voluminous, you say “Now, I
really do not add anything more!” many times, and you still add though. . . , because
the ideas, the problems are too beautiful to keep them out.
The book is divided into two parts. Part I contains the problems, arranged by
topics in the first seven chapters of the book, and Chap. 8, which collects two
new proofs of the Sandham–Yeung series. Part II brings solutions to the proposed
exercises in the first seven chapters of the book.
The first chapter of the book collects non-standard problems on limits of special
sequences and integrals. Why limits? First, because in analysis most things reduce
to the calculation of a limit, and second, because limits are the fundamental
problems of analysis. Many problems are original, based on the ideas aroused
during the research work, and not rarely quite unexpectedly. We mention that a
few problems from this chapter, some elementary limits of sequences, have been
taken from Romanian mathematical literature, especially from the famous Gazeta
Matematică, which is the oldest Romanian journal of mathematics that appears
without interruption since 1895.
The second chapter of the book contains series of real numbers. The problems in
this chapter deal with two major topics on series of real numbers: the convergence
of a numerical series and its exact calculation. Regarding the convergence, we have
included in this chapter original problems that can be solved using the classical
way, i.e., the well-known convergence tests for numerical series, or less-known
ix
x Preface
ways . . . The problems about the exact calculation of the value of a series are, most
of them, new in the literature and vary in diversity as well as in difficulty. Their
topics range from the calculation of series involving factorial terms to exotic series
involving the tail of Riemann zeta function values.
The third chapter of the book deals with power series. This chapter includes
the standard formulae on the power series, Taylor series, and Maclaurin series of
the classical elementary functions, and exercises dealing with finding the set of
convergence of a power series and, whenever possible, determining the sum of the
series. The problems in this chapter are as exotic and fascinating as possible. They
cover topics such as the generating functions of special sequences, the calculation
of power series with factorials, series involving polylogarithm functions, as well
as nonlinear logarithmic series. The chapter also includes, as applications, the
calculation of single and multiple series involving Riemann zeta function values.
The last section of this chapter collects several nonlinear Euler series which are new
in the literature (but, clearly, the original problems can be found in other sections,
too).
Chapter 4 is about derivatives and their applications. This chapter contains
problems on the classical topics of mathematical analysis: the calculation of the nth
derivative of a function, Leibniz’s formula, Taylor’s formula, the extremum points
of a function of a single variable, as well as non-standard topics of analysis: the
generalized Leibniz formula, special differential equations, and the computation of
exotic series involving the Maclaurin remainder of special functions.
In Chap. 5, we collect problems on partial derivatives and their applications.
The problems included here deal with the chain rule, Euler’s identity regarding
homogeneous functions, Taylor’s formula, and extrema of functions of several real
variables.
Chapter 6 contains problems on implicit functions. Here is the place where a lot
of curves appear, together with their interesting properties which are pointed out
with the help of the implicit functions theory.
Chapter 7 collects challenging problems on several topics discussed in the
previous chapters as well as gems and mathematical beauties. Special attention is
given to problems concerning evaluation of linear and nonlinear series involving
tails of Riemann zeta function values.
In Chap. 8, we give two new proofs of the famous Sandham–Yeung series.
This volume contains both theory and problems of analysis, many of the
problems being new and original. We do not claim originality of all the problems
included in the book and we are aware that some may be either known or very old.
We have not attempted to document the source of every problem. This would be
a very difficult task: on the one hand, many of this volume’s problems have been
discovered by us over the last decade, some of them have been published in various
journals with a problem column, others see the light of publication for the first time
in this book. Also, there are problems whose history is either lost, with the passing
of time, or the authors have not been aware of it. We have tried to avoid collecting
too many problems that are well-known or published elsewhere, in order to keep
a high level of originality. On the other hand, other problems of this book arose
Preface xi
in a natural way: either as generalizations or motivated by known results that have

long been forgotten (see, for example, in Problem 2.52, the alternating series due to
Hardy). For such problems, when known, the source of the problem is mentioned,
either in the solution or in the statement of the problem.
We have included throughout the book results or quick statements of mathemat-
ical analysis whose proofs have been given in detail. These were called gems of
mathematical analysis, either because they have a very beautiful solution, which
the authors discovered or revived and brought into the light, or because they are
classical results in analysis that the reader should know. The intention of adding
them in the book was to make the material more appealing, and also to inspire the
reader’s creativity for solving the problems.
The book contains theoretical results, added before a problem or a group of
problems, which suggest to the reader the tools for solving the exercises.
The problems have been arranged on sections, each section dealing with a
particular type of topic. Some problems, which are difficult or very challenging,
have been divided into parts, (a), (b), (c), . . . to help the reader find a solution.
Also, there are a couple of open problems, as well as challenge problems, that have
been included in the book. These problems, motivated by knows results, may be
considered as research problems or projects for students with a strong background
in analysis and for the reader who enjoys mathematical research and discovery in
mathematics. The intention of having some difficult problems and open problems,
gathered in the book, is to stimulate creativity and discovery of original methods for
proving known results and establishing new ones.
Many of the problems have been put in gray boxes and some of them have
been colored in blue, either because these are very beautiful results, although
the authors consider all of them to be very beautiful, or they are gems of
mathematical analysis, or have unexpected elegant solutions, or, simply, these
are results that should be studied by the reader who enjoys working on
nonstandard problems of analysis.
This book is addressed to undergraduate students with a strong background in

analysis who prepare for competitions like Putnam and SEEMOUS or other higher-
level mathematical contests. Mathematicians and students interested in problem
solving will find this collection of topics tempting. This volume is also useful
for instructors who are involved in math contests, as well as for individuals who
wish to enrich and test their knowledge by solving problems in analysis. It can
be used by anyone for independent study courses, by students of mathematics,
physics, and engineering, and by anyone who wants to explore the standard topics
of mathematical analysis.
We also address this work to the first- and second-year graduate students who
want to learn more about applications of a certain technique, to do calculations
xii Preface
which happen to have interesting results. Pure and applied mathematicians, who
confront certain difficult computations in their research, might find this book
attractive.
We wish you success in going over this collection of analysis problems, as we
hope that many of them are challenging, worth studying, and splendid. We also
hope you enjoy reading the theory and solving the problems, even if some of them
are quite tough. For questions, generalizations, remarks, or observations regarding
the improvement of this material, please do not hesitate to contact us by regular or
electronic mail.
Cluj-Napoca, Romania Alina Sîntămărian

March 18, 2021 Ovidiu Furdui
Contents
Part I Theory and Problems
1 Sequences of Real Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.1 Limits of Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Applications of Stolz–Cesàro Theorem, the ∞/∞ and the
0/0 Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3 Wolstenholme Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.4 Limits of Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2 Series of Real Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.1 Miscellaneous Series. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.2 Applications of Abel’s Summation Formula . . . . . . . . . . . . . . . . . . . . . . 38
2.3 Series with Positive Terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.4 Alternating Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.5 Series with Harmonic Numbers and Factorials . . . . . . . . . . . . . . . . . . . . 54
2.6 A Mosaic of Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
3 Power Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.1 Convergence and Sum of Power Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.2 Maclaurin Series of Elementary Functions . . . . . . . . . . . . . . . . . . . . . . . . 66
3.3 Gems with Numerical and Power Series . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.4 Single Zeta Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
3.5 Polylogarithm Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.6 Inequalities and Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
3.7 Generating Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
3.8 Series with Harmonic and Skew-Harmonic Numbers. . . . . . . . . . . . . 98
3.9 Remarkable Numerical and Function Series. . . . . . . . . . . . . . . . . . . . . . . 101
3.10 Multiple Series with the Riemann Zeta Function . . . . . . . . . . . . . . . . . 115
3.11 Series Involving Products of Harmonic Numbers . . . . . . . . . . . . . . . . . 119
4 Derivatives and Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
4.1 Apéritif . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
4.2 Integral Equations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
4.3 Differential Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
4.4 Higher Order Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
xv
xvi Contents
4.5 Taylor’s Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136

4.6 Series with the Maclaurin Remainder of a Function f . . . . . . . . . . . 139
4.7 Series with Fractional Part Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
4.8 Extrema of One Variable Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
5 Partial Derivatives and Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
5.1 Partial Derivatives, the Jacobian and the Hessian Matrices,
Differential Operators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
5.2 The Chain Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
5.3 Homogeneous Functions. Euler’s Identity . . . . . . . . . . . . . . . . . . . . . . . . . 165
5.4 Taylor’s Formula for Real Functions of Two Real Variables . . . . . 166
5.5 The Differential of Several Real Variable Functions . . . . . . . . . . . . . . 170
5.6 Extrema of Several Real Variable Functions . . . . . . . . . . . . . . . . . . . . . . 172
6 Implicit Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
6.1 Implicit Functions of One Real Variable Defined by an Equation 177
6.2 Implicit Functions of Two Real Variables Defined by an
Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
6.3 Implicit Functions of One Real Variable Defined by a
System of Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
6.4 Implicit Functions of Two Real Variables Defined by a
7 Challenges, Gems, and Mathematical Beauties . . . . . . . . . . . . . . . . . . . . . . . . . 187
7.3 Convergence and Evaluation of Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
7.4 Harmonic Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
7.5 Series with Factorials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
7.6 Series of Functions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
7.7 Pearls of Series with Tails of Zeta Function Values . . . . . . . . . . . . . . . 213
7.8 Exotic Zeta Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
7.9 Special Differential Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
7.10 Inequalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
7.11 Fabulous Integrals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
8 An Artistry of Quadratic Series: Two New Proofs of
Sandham–Yeung Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
8.1 The First Proof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
8.2 The Second Proof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
Part II Solutions
9 Sequences of Real Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253

0/0 Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
Contents xvii
9.3 Wolstenholme Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270

10 Series of Real Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
10.1 Miscellaneous Series. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
10.2 Applications of Abel’s Summation Formula . . . . . . . . . . . . . . . . . . . . . . 310
10.3 Series with Positive Terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
10.4 Alternating Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
10.5 Series with Harmonic Numbers and Factorials . . . . . . . . . . . . . . . . . . . . 323
10.6 A Mosaic of Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
11 Power Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
11.1 Convergence and Sum of Power Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
11.2 Maclaurin Series of Elementary Functions . . . . . . . . . . . . . . . . . . . . . . . . 340
11.3 Gems with Numerical and Power Series . . . . . . . . . . . . . . . . . . . . . . . . . . . 348
11.4 Single Zeta Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
11.5 Polylogarithm Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
11.6 Inequalities and Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
11.7 Generating Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
11.8 Series with Harmonic and Skew-Harmonic Numbers. . . . . . . . . . . . . 375
11.9 Remarkable Numerical and Function Series. . . . . . . . . . . . . . . . . . . . . . . 387
11.10 Multiple Series with the Riemann Zeta Function . . . . . . . . . . . . . . . . . 398
11.11 Series Involving Products of Harmonic Numbers . . . . . . . . . . . . . . . . . 402
12 Derivatives and Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407
12.1 Apéritif . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407
12.2 Integral Equations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 411
12.3 Differential Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412
12.4 Higher Order Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
12.5 Taylor’s Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 430
12.6 Series with the Maclaurin Remainder of a Function f . . . . . . . . . . . 436
12.7 Series with Fractional Part Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
12.8 Extrema of One Variable Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
13 Partial Derivatives and Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 447
Differential Operators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 447
13.2 The Chain Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 454
13.3 Homogeneous Functions. Euler’s Identity . . . . . . . . . . . . . . . . . . . . . . . . . 462
13.4 Taylor’s Formula for Real Functions of Two Real Variables . . . . . 462
13.5 The Differential of Several Real Variable Functions . . . . . . . . . . . . . . 463
13.6 Extrema of Several Real Variable Functions . . . . . . . . . . . . . . . . . . . . . . 463
14 Implicit Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473
Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474
xviii Contents

15 Challenges, Gems, and Mathematical Beauties . . . . . . . . . . . . . . . . . . . . . . . . . 475
15.3 Convergence and Evaluation of Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 492
15.4 Harmonic Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 498
15.5 Series with Factorials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501
15.6 Series of Functions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 502
15.7 Pearls of Series with Tails of Zeta Function Values . . . . . . . . . . . . . . . 508
15.8 Exotic Zeta Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 519
15.9 Special Differential Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
15.10 Inequalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522
15.11 Fabulous Integrals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 529
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 535
Notations
N The set of natural numbers (N = {1, 2, 3, . . .})

R The set of real numbers
R∗ The set of nonzero real numbers (R∗ = R \ {0})
R R ∪ {±∞}
C The set of complex numbers
(z) The real part of the complex number z
a The floor of the real number a
{a} The fractional part of the real number a, {a} = a − a
n! 0! = 1, n! = 1 · 2 · 3 · · · n, n ∈ N
(2n)!! 0!! = 1, (2n)!! = 2 · 4 · 6 · · · (2n), n ∈ N
(2n − 1)!! (2n − 1)!! = 1 · 3 · 5 · · · (2n − 1), n ∈ N
∼ an ∼ bn , as n → ∞, if lim an /bn exists and is finite
n→∞
Ckn The binomial coefficient of n taken by k, Ckn = k!(n−k)!
n!
e limn→∞ (1 + 1/n) = 2.71828 . . .

n
γ limn→∞ (1 + 1/2 + · · · + 1/n − lnnn) = 0.57721 ...

G Catalan’s constant G = ∞ n=0 (−1) /(2n + 1) 2
Hn The nth harmonic number Hn = 1 + 1/2 + · · · + 1/n, n ∈ N

ψ The Digamma function (the Psi function) ψ(z) = (z)/ (z), for
(z) > 0
ζ The Riemann zeta function ζ (z) = ∞ ∞, for (z) > 1 2
n=1 1/n
z
ζ (2, a) The Hurwitz zeta function ζ (2, a) = n=0 1/(n + a) , a =

0, −1, −2, . . .
β The Dirichlet beta function β(s) = ∞ ∞
n=0 (−1)n /(2n + 1)s , (s) > 0
z function Lin (z) = k=1 z /k , |z| ≤ 1, n ∈ N \
Lin The Polylogarithm k n
{1}, or Lin (z) = 0 Lin−1 (t)/tdt, n ∈ N \ {1, 2}

V (x) The set of all neighborhoods of x, where x ∈ Rn
int(A) The interior of set A ⊆ Rn
f (n) The nth derivative of function f
xix
xx Notations
C k (Ω, R) The set of all functions f : Ω ⊆ Rn → R which have partial

derivatives of order k continuous on Ω, called the set of functions of
class C k ;
when Ω ⊆ R, the set of functions f : Ω → R which have derivative
of order k continuous on Ω
C ∞ (I, R) The set of all functions f : I → R, where I ⊆ R is an open interval,
which are indefinite differentiable on I
Part I
Theory and Problems
Sequences of Real Numbers
1
If we prove that γ is irrational do you give us

a bonus point at the final exam?
A question of a UTCN student for Alina
during an analysis course
This chapter collects problems on sequences of real numbers. The problems

included are challenging and the topics covered vary in diversity: from limits
of sequences involving special numerical terms, applications of Stolz–Cesàro
Theorem, both ∞/∞ and 0/0 cases, Wolstenholme sequences, to the calculation
of spectacular limits of integrals.
1.1 Limits of Sequences
The Constant e The sequence of rational numbers (en )n≥1 defined by

1 n
en = 1 + , n ≥ 1,
n
is strictly increasing and bounded, hence convergent. Its limit is denoted by

e, in honor of the Swiss mathematician Leonhard Euler (1707–1783). It is
known that e = 2.71828 . . ..
Another sequence of rational numbers convergent to e is the sequence
(En )n≥0 defined by
1 1 1
En = 1 + + + ··· + , n ≥ 0.
1! 2! n!
(continued)
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 3

A. Sîntămărian, O. Furdui, Sharpening Mathematical Analysis Skills, Problem Books
in Mathematics, https://doi.org/10.1007/978-3-030-77139-3_1
4 1 Sequences of Real Numbers
Therefore, we have
n
1
e = lim 1+ ;
n→∞ n

1 1 1
e = lim 1+ + + ··· + .
n→∞ 1! 2! n!

n!en
1.1. Prove that the sequence n+ 21
is strictly decreasing.
n n≥1

n
9k 2 +12k+5
1.2. Calculate lim (3k+2)! .
n→∞ k=1

n
1
1.3. Calculate lim .
n→∞ k=1 k(k+1)(k+1)!
1.4. Let (xn )n≥1 be the sequence defined by xn = 1

+ 1
+ 1
+ ··· + 1
n(n+2) ,
n 1·3 2·4 3·5
n ≥ 1. Calculate lim 2xn − 12 .
n→∞
Some Algebraic Identities If a, b ∈ R and n ∈ N, n ≥ 2, then the following

identities hold:
• a n − bn = (a − b)(a n−1 + a n−2 b + · · · + abn−2 + bn−1 );

• a 2n−1 + b2n−1 = (a + b)(a 2n−2 − a 2n−3 b + · · · − ab2n−3 + b2n−2 ).
Particular cases:
• a 2 − b2 = (a − b)(a + b);
• a 3 ± b3 = (a ± b)(a 2 ∓ ab + b2 ).
1.5. Let k, m ∈ N, k, m ≥ 2. Calculate

√ √
n + 3 − nk + 1
k k k
lim √ √ .
n→∞ m nm + 2 − m nm − 5
√
1.6. Determine a, b, c ∈ R such that lim n(an − bn2 + cn − 3) = 1.
n→∞
1.1 Limits of Sequences 5
Three Sums of Natural Numbers If n ≥ 1 is an integer, then:
n(n + 1)
• 1 + 2 + 3 + ··· + n = ;
2
n(n + 1)(2n + 1)
• 12 + 22 + 32 + · · · + n2 = ;
6
2
n(n + 1)
• 13 + 23 + 33 + · · · + n3 = .
2
1.7. Calculate:

1 + 3 + 5 + · · · + (2n − 1) 2n + 1
(a) lim − ;
n→∞ n+1 2

2n
(−1)k (1 + 2 + · · · + k)
k=1
(b) lim ;

n→∞ 2n+1
(−1) (1 + 2 + · · · + k)
k−1
k=1
2n−1
1
(c) lim (−1)k−1 k 2 ;
n→∞ n2
k=1
2n
1
(d) lim 4 (−1)k (13 + 23 + 33 + · · · + k 3 ).
n→∞ n
k=1
1.8. Let {x} be the fractional part of x. Prove that:

√ n √
(a) lim (2 + 3) = 1; (b) lim (3 + 7)n = 1;
n→∞ n→∞
√ 2n √
(c) lim (1 + 2) = 1; (d) lim (1 + 2)2n−1 = 0.
n→∞ n→∞
The Squeeze Theorem Let (an )n≥1 , (bn )n≥1 , (cn )n≥1 be sequences of real
numbers for which there exists n0 ∈ N such that an ≤ bn ≤ cn , for all
n ≥ n0 . If lim an = lim cn = l ∈ R, then lim bn = l.
n→∞ n→∞ n→∞
sin n
1.9. Calculate lim n n .
n→∞
1.10. Calculate:

1 1 1
(a) lim + + ··· + ;
n→∞ (n + 1)2 (n + 2)2 (2n)2

1 1 1
(b) lim √ +√ + ··· + √ ;
n→∞ n2 + 1 n2 + 2 n2 + n
n
5k 3 + 3k 2 + 2k + 1
(c) lim ;
n→∞ n4 + k + 7
k=1
(2n + 1)(2n + 3) · · · (4n + 1)
(d) lim ;
n→∞ (2n)(2n + 2) · · · (4n)
(n + 2)(n + 5) · · · (4n − 1)
(e) lim ;
n→∞ (n + 1)(n + 4) · · · (4n − 2)
C2n
(f) lim n 4nn ;
n→∞ 4 C
2n
33n (Cn2n )2
(g) lim ;
n→∞ Cn C3n
3n 6n
n
1 Ck
(h) lim n √ n .
n→∞ 2
k=0
k+2
Remark 1.1. Regarding parts (d) and (e) of the above problem, more general limits
are given in Problem 1.16 and part (i) of Problem 7.19.
1.11. Let (xn )n∈N be a bounded sequence of real numbers such that lim (2xn +
n→∞
x2n ) = l ∈ R. Prove that lim xn = 3l .
n→∞
Remark 1.2. If k ∈ N, k ≥ 2, and (xn )n∈N is a bounded sequence of real numbers

such that lim (kxn + xkn ) = l ∈ R, then lim xn = k+1
l
.
n→∞ n→∞
1.12. [32] Prove that:

1 3
(a) lim n
x n + (1 − x)n dx = ;
n→∞ 0
4
1 3 π2
(b) lim n 2 n
xn + (1 − x)n dx − = .
n→∞ 0 4 48

π
2 √
sinn x + cosn x dx =
n
(a) lim 2;
n→∞ 0

π
2 √ π2
sinn x + cosn x dx − 2 =
n
(b) lim n 2
√ .
n→∞ 0 12 2
1.14. [A. Sîntămărian, 2016] Calculate

π
2
lim n
sinn x + sinn (2x) + cosn x dx.
n→∞ 0
1.15.
(a) Let (xn )n≥1 be the sequence defined by
1, 2, 2, 3, 3, 3, 4, 4, 4, 4, . . . .

2 terms 3 terms 4 terms
xn
√
Calculate lim n
.
n→∞
(b) A problem of G. Boroica. Let (yn )n≥1 be the sequence defined by
1, 2, 2, 2, 2, 3, 3, . . . , 3, 4, 4, . . . , 4, . . . .

22 terms 32 terms 42 terms
yn
Calculate lim √
3 n.
n→∞
1.16. [128, pp. 82, 83], [130, Remark 4] A Challenging Limit

Let p, q ∈ N, with p ≥ 2. Calculate
(qn + 1)(qn + p + 1) · · · (qn + np + 1)

lim .
n→∞ qn(qn + p) · · · (qn + np)
Remark 1.3. For a generalization of the above limit, see part (i) of Problem 7.19.
1.17. Find the general term of the sequence (an )n≥1 , with a1 = a2 = 1, defined by
the recurrence relation an+2 = an + an+1
1
, n ≥ 1.
√
1.18. Let (xn )n≥1
√ , with x1 = 2, be the sequence defined by the recurrence
relation xn+1 = 2 + xn . Calculate lim 4n (2 − xn ).
n→∞
1.19. Let (xn )n≥0 , with x0 > 0, be the sequence defined by the recurrence relation
xn+1 = 21+x
√ n . Prove that the sequence converges and find its limit.
xn
1.20.
(a) Study the convergence of the sequence (xn )n≥0 , with x0 ≥ 0, defined by the
recurrence relation
√ 1
xn+1 = xn + .
n+1
(b) If k ∈ N, k ≥ 2, study the convergence of the sequence (xn )n≥0 , with x0 ≥ 0,

defined by the recurrence relation
√ 1
xn+1 = k
xn + .
n+1
Cauchy–d’Alembert’s Criterion Let (an )n≥1 be a sequence of positive real

√
numbers such that lim an+1
an = l ∈ R. Then lim
n a = l.
n
n→∞ n→∞
1.21. Calculate:
n
(a) lim √
n
;
n→∞
n!
n (n!)2
(b) lim ;
n→∞ (2n)!8n
√
n
(n + 1)(n + 2) · · · (2n)
(c) lim ;
n→∞ n
(2n)(2n + 3) · · · (8n − 6)
(d) lim n ;
n→∞ (n!)2
√
n
(2n + 1)(2n + 4) · · · (5n − 2)
(e) lim .
n→∞ n
Remark 1.4. For a generalization of parts (d) and (e) in the above problem see parts
(ii) and (iii) of Problem 7.19.
√
1.22. Let a ∈ (0, 1). Calculate lim n
1− n
a.
n→∞
1.23. [103] Traian Lalescu’s Sequence Calculate

√
n
lim n+1
(n + 1)! − n! .
n→∞
The Fibonacci Sequence

One of the most important sequences defined by a recurrence relation with
two terms is the sequence (Fn )n≥0 defined by
F0 = 0, F1 = 1, Fn+1 = Fn + Fn−1 , n ≥ 1,
which bears the name of the Italian mathematician Leonardo Fibonacci (c.
1170–c. 1240). The Fibonacci number Fn+1 is the sum of the other two
previous Fibonacci numbers. It is well-known the following representation
of the Fibonacci numbers:
√ n √ n
1 1+ 5 1− 5
Fn = √ − , n ≥ 0,
5 2 2
√
the irrational number 1+ 5
2 = 1.61803 . . . being called the golden number or
the golden section.
[117] An Equality with the Golden Section

2 arctan 2
arctan √ = .
1+ 5 2
1.24. Two special sums of binomial coefficients. Prove that:
(a) C0n + C2n+1 + C4n+2 + · · · + C2n

2n = F2n+1 , n ≥ 0;
(b) C1n + C3n+1 + C5n+2 + · · · + C2n−1
2n−1 = F2n , n ≥ 1.
1.25. Compositions of homographic functions.
(a) Let f : (0, ∞) → R, f (x) = 3x+2

2x+3 . Calculate lim f ◦ f ◦ · · · ◦ f (x).
n→∞
n times
(b) Let f : (0, ∞) → R, f (x) = 2x+1
x+3 . Calculate lim f ◦ f ◦ · · · ◦ f (x).
n→∞
n times

0/0 Cases
Stolz–Cesàro Theorem, the ∞/∞ Case Let (an )n≥1 and (bn )n≥1 be two
sequences of real numbers such that:
(a) 0 < b1 < b2 < . . . < bn < . . . and lim bn = ∞;

n→∞
an+1 − an
(b) lim = l ∈ R.
n→∞ bn+1 − bn
an
Then lim exists and it is equal to l.
n→∞ bn
This theorem, the discrete version of l’Hôpital’s Theorem, is mentioned in

various papers in the literature, either as a theoretical result or as a problem. In
[10, pp. 413, 414], [26, Appendix B], it appears as a theoretical result regarding the
calculation of limits of expressions xn /yn of the form ∞/∞, and as a problem it is
mentioned, in a slightly modified form, in the famous problem book of Pólya and
Szegö [121, problem 70, pp. 16, 17].
1.26. Calculate:
1+ + · · · + n1
1
2
(a) lim ;
n→∞ ln n
1 + 2 + · · · + np
p p
(b) lim , where p ∈ N;
n→∞ √np+1 √ √
1 + 2 2 + 3 3 + ··· + n n
(c) lim √ ;
n→∞
√ √ n2 n
√
1 + 2 + 3 + ··· + 3 n
3 3
(d) lim √ ;
n→∞
√ n 3 n√
√
1 + 22 2 + 32 3 3 + · · · + n2 n n
(e) lim ;
n→∞ n(n + 1)(n + 2)
√ √ √2
1 + 4 2! + 9 3! + · · · + n n!
(f) lim ;
n→∞
n
n
n
(g) lim ;
n→∞ 1 + 22 + 33 + · · · + nn
a + 2a + · · · + na
(h) lim , where a ∈ R and x is the floor of x ∈ R;
n→∞ n2
a 1! + 2a 2! + · · · + na n!
(i) lim , where a ∈ R and x is the floor of
n→∞ (n + 1)!
x ∈ R.
1.2 Applications of Stolz–Cesàro Theorem, the ∞/∞ and the 0/0 Cases 11
1.27. Let (xn )n≥1 , with x1 > 0, be the sequence defined by the recurrence relation
xn+1 = ln(1 + xn ), n ≥ 1. Calculate lim nxn .
n→∞
1.28. Let (xn )n≥1 , with x1 > 0, be the sequence

√ defined by the recurrence relation
xn+1 = arctan xn , n ≥ 1. Calculate lim nxn .
n→∞
1.29. Let (xn )n≥1 , with x1 ∈ R, be the sequence defined by the recurrence relation
xn+1 = xn + e−xn , n ≥ 1. Calculate lim lnxnn .
n→∞
1.30. Open Problem Let β > 0 and let (xn )n≥1 be the sequence defined by
the recurrence relation
n2β
x1 = a > 0, xn+1 = xn + .
x1 + x2 + · · · + xn

xn ? β +1
Prove that lim β = .
n→∞ n β
Stolz–Cesàro Theorem, the 0/0 Case Let (an )n≥1 and (bn )n≥1 be two
sequences of real numbers such that:
(a) lim an = lim bn = 0;

n→∞ n→∞
(b) (bn )n≥1 is strictly decreasing;
an+1 − an
(c) lim = l ∈ R.
n→∞ bn+1 − bn
an
Then lim exists and it is equal to l.
n→∞ bn
This version of Stolz–Cesàro Theorem is discussed in [94, p. 56], [26, pp. 265,
266], [124, pp. 281–284].
1.31. Euler’s Constant Let (xn )n≥1 be the sequence defined by
1 1 1
xn = 1 + + + · · · + − ln n, n ≥ 1.
2 3 n
(continued)
Prove that (xn )n≥1 converges and, denoting by γ its limit, calculate
lim n(xn − γ ).
n→∞
The constant γ = 0.5772156649 . . . is called Euler’s constant, being

introduced in the literature by Leonhard Euler. It is not known whether γ is a
rational or an irrational number, this being an open problem of great interest
for mathematicians.
1.32.
(a) Calculate
lim n(2Hn − Hn2 − γ ).

n→∞
(b) A generalization. Let k ≥ 2 be an integer. Calculate

lim n kHn − Hnk − (k − 1)γ .
n→∞
1.33. Let Hn be the nth harmonic number and let On = 1 + 13 + · · · + 2n−1

1
, n ≥ 1.
Calculate

1 Hn n 1 On n 1 2On n
lim 1+ , lim √ 1+ and lim 1+ .
n→∞ n n n→∞ n n n→∞ n n
1.34. Calculate
√ 1 + 1 +···+ 1 1 1

lim n 2e 2 4 2n − e1+ 3 +···+ 2n−1 .
n→∞
1.35. Ioachimescu’s Constant Let (xn )n≥1 be the sequence defined by
1 1 1 √
xn = 1 + √ + √ + · · · + √ − 2( n − 1), n ≥ 1.
2 3 n
√ that (xn )n≥1 converges and, denoting by I its limit, calculate

Prove
lim n(xn − I ).
n→∞
The constant I = 0.539645491 . . . is called Ioachimescu’s constant.

Romanian mathematician Andrei G. Ioachimescu (1868–1943), one of the
(continued)
founders of the famous Romanian magazine Gazeta Matematică, has pro-

posed a problem [99] in which he was asking to prove that the sequence
(xn − 2)n≥1 converges to a finite limit between −2 and −1.

π2
n
1.36. Calculate lim n 6 − 1
k2
.
n→∞ k=1

π4
n
1.37. Calculate lim n3 90 − 1
k4
.
n→∞ k=1
1.38. [D. M. Bătineţu-Giurgiu, 1991]

n 4 n3
Let xn = 1
(2k−1) 4 , n ≥ 1. Calculate lim π
96xn .
k=1 n→∞
1.39. Let x ∈ R, x > 1.
(a) Prove that

n

1 1
lim n x−1
ζ (x) − = .
n→∞ kx x−1
k=1
(b) Calculate
n

1 1
lim n nx−1 ζ (x) − − .
n→∞ kx x−1
k=1
∞
The function ζ defined by ζ (z) = n=1 1/nz , for (z) > 1, is called the
Riemann zeta function. The most important open problem in mathematics,
the Riemann hypothesis, is about proving that the nontrivial zeros of the zeta
function are located on the vertical line that passes through 1/2.
1.40. Let k ∈ R, k > 1. Calculate

1 1 1 1
lim 1 + + · · · + ζ (k) − 1 − k − · · · − k .
n→∞ 2 n 2 n
1.41. Limits with the zeta function. Prove that:
(a) lim ζ (x) = 1;

x→∞
ζ (2x) − 1
(b) lim = 0;
ζ (x) − 1
x→∞
2x
(c) lim ζ (x) = e;
x→∞
3 x x

(d) lim ζ (x)2 − e = e.
x→∞ 2
1.42. Two limits involving the constant e. Calculate

n
n

1 1 1
lim (n + 1)! e − and lim (−1)n−1 (n + 1)! − (−1)k .
n→∞ k! n→∞ e k!
k=0 k=0
1.3 Wolstenholme Sequences
1.43. A Wolstenholme Limit
(a) Let s > 0. Prove that

n sn
k es
lim = s .
n→∞ n e −1
k=1
(b) [120] A Limit of Dumitru Popa

Let f : (0, 1] → (0, ∞) be a differentiable function on (0, 1], with
f (1) > 0 and ln f having decreasing derivative, and let (xn )n≥1 be the
n
sequence defined by xn = f n nk .
k=1
Prove that
⎧
⎪
⎪
⎨0 if f (1) < 1
lim xn = 1
if f (1) = 1
n→∞ ⎪
⎪ 1−e−f (1)
⎩∞ if f (1) > 1.
Reference material regarding this problem can be found in [26, pp. 32–33].
More generally, it can be proved that if f is like in part (b) and (yn )n≥1 is the
n
sequence defined by yn = f αn+β nk , α > 0, β ≥ 0, then
k=1
(continued)
1.3 Wolstenholme Sequences 15
⎧
⎪
⎪
⎨0 if f (1) < 1
lim yn = 1
if f (1) = 1
n→∞ ⎪
⎪ 1−e−αf (1)
⎩∞ if f (1) > 1.
1.44. Let a ≥ 1. Prove that

n n n k
k k ae
lim a k−n = lim a k−n = .
n→∞ n n→∞ n ae − 1
k=1 k=1
1.45.
(a) Let s > 0. Prove that

n sk
k es
lim = s .
n→∞ n e −1
k=1
(b) Let a, b ≥ 0, with a + b > 0. Prove that

n ak+bn
k ea+b
lim = a+b .
n→∞ n e −1
k=1
1.46. Let f : (0, 1] → (0, ∞) be a differentiable function on (0, 1], with

f (1) > 0 and ln f having decreasing derivative, and let (xn )n≥1 be the
sequence defined by
n
k
xn = a k−n f αn+β ,
n
k=1
where a ≥ 1, α > 0, and β ≥ 0.

Prove that
(continued)
⎧
⎪
⎪ 0 if f (1) < 1
⎪
⎨
aeαf (1)
lim xn = if f (1) = 1
n→∞ ⎪
⎪ aeαf (1) −1
⎪
⎩
∞ if f (1) > 1.
1.4 Limits of Integrals
1.47. Spectacular Limits of Integrals
(a) Prove that:

1√
n
+1
x
n
2
• lim dx = ;
n→∞ 0 2 3
1√ n
n
x+1 2 2
• lim n dx − = .
n→∞ 0 2 3 27
Remark 1.5. If k > 0, then

1√
n
+k−1
x
n
k
• lim dx = ;
n→∞ 0 k k+1
1√ n
n
x+k−1 k (k − 1)k
• lim n dx − = .
n→∞ 0 k k + 1 (k + 1)3
(b) Prove that:

1 2
n
• lim √ dx = 2;
n→∞ 0 x+1
n
1 n
2
• lim n 2 − √ dx = 2.
n→∞ 0
n
x+1
Remark 1.6. [O. Furdui, A. Sîntămărian, Problem 2, SEEMOUS 2020]

If k > 1, then
1 k
n
k
• lim √ dx = ;
n→∞ 0 n
x+k−1 k−1
(continued)
1.4 Limits of Integrals 17
1 n
k k k
• lim n − √ dx = .
n→∞ k−1 0
n
x+k−1 (k − 1)2
1.48.
(a) Prove that
4n 1 √ n √
lim √ 1− n
x dx = π .
n→∞ n 0
(b) If f : [0, 1] → R is a Riemann integrable function which is continuous at 0,

then
4n 1 √ n √
lim √ 1− n
x f (x)dx = π f (0).
n→∞ n 0
1.49. Let (an )n≥1 be a sequence of nonnegative real numbers such that lim
n→∞
an = a.
(a) Calculate

1
n
lim (1 + an x n )n dx.
n→∞ 0
Application. Let α > 0 be a real number. Calculate

1
n
lim (1 + αx n )n dx.
n→∞ 0
(b) Calculate

1 x + x 3 + · · · + x 2n−1
n
1+
n
lim dx.
n→∞ 0 n
(c) Calculate

1 a1 x + a2 x 2 + · · · + an x n
n
1+
n
lim dx.
n→∞ 0 n
1.50. Let f : [−1, 1] → R be a Riemann integrable function which is continuous

at 0. Prove that
1 hf (x)
lim dx = πf (0).
h→0+ −1 x 2 + h2
1.51. Two identities with the fractional part function.
(a) Let f : [0, 1] → R be a Riemann integrable function. Prove that
1 1
f ({nx})dx = f (x)dx.
0 0
(b) Let f : [0, 1] → R be a Riemann integrable function. Prove that
1 n−1 1
nx f ({nx}) dx = f (x)dx.
0 2 0
1.52. Let f : R → R be a function and let g : [0, 1] → R be a Riemann integrable

function.
(a) Prove that
1 n−1 1
1
f (nx) g ({nx}) dx = f (k) g(x)dx.
0 n 0
k=0
f (x)
(b) If f satisfies lim α = L, where α ≥ 0, then
x→∞ x
1 1 L 1
lim f (nx) g ({nx}) dx = g(x)dx.
n→∞ nα 0 1+α 0
1.53. Let f : [0, 1] → R be a Riemann integrable function. Calculate
1 n
lim f dx,
n→∞ 0 x
where {x} is the fractional part of x.

1.54.
(a) Calculate
!n
1 1
n
lim dx,
n→∞ 0 x
where {x} is the fractional part of x.

(b) Let f : [0, 1] → (0, ∞) be a continuous function. Calculate
!n
1 1
n
lim f (x)dx.
n→∞ 0 x
1.55.
(a) Prove that

1 1 1
lim " #n dx = ,
n→∞ 0 1 2
x
where a is the floor of a.

(b) Let f : [0, 1] → R be a Riemann integrable function which is continuous at 0.
Prove that
⎛ ⎞
1
⎜ 1 ⎟ f (0) + f (1)
lim f ⎝ " #n ⎠ dx = .
n→∞ 0 1 2
x
1.56.
(a) Let k ∈ N. Prove that

1
xk 1 1
lim " #n dx = 1 − k+1 ,
n→∞ 0 1 k+1 2
x

(b) Let f, g : [0, 1] → R be Riemann integrable functions, with f continuous at 0.
Prove that
⎛ ⎞
1
1 1
⎜ 1 ⎟ 2
lim f ⎝ " #n ⎠ g(x)dx = f (0) g(x)dx + f (1) g(x)dx.
n→∞ 0 1 0 1
x 2
1.57. Let f, g : [0, 1] → R be Riemann integrable functions, with f continuous at

0. Prove that

1 1 1
lim f * n + g(x)dx = f (0) g(x)dx.
n→∞ 0 0
x
1.58. Let α > 0 and let f : [−1, 0] → R be a continuous function. Prove

that
0 n f (0)
lim n x + eαx f (x)dx = .
n→∞ −1 1 +α
1.59. Let α > 0. Calculate

0 n
x2
lim n x+ + eαx dx.
n→∞ −1 2
1.60. Let α > 0. Calculate

n
1 √
lim x + α dx
n n
.
n→∞ 0
1.61. Calculate
n dx
lim .
n→∞ 0 1 + n2 cos2 x
1.62. Calculate
n dx
lim , a > 0.
n→∞ 0 1 + n2 sin2 (x + na)
1.63. [57] Calculate
1 n x
lim dx.
n→∞ n 0 1 + n2 cos2 x
More generally (see [55]), it can be proved that if f : [0, 1] → R is a

continuous function, then
(continued)

n f xn 1
lim dx = f (x)dx.
n→∞ 0 1 + n2 cos2 x 0
1.64. Calculate
1 n x
lim √ dx.
n→∞ n n 0 1 + n cos2 x
More generally, it can be proved that if f : [0, 1] → R is a continuous

function, then

1 n f xn 1
lim √ dx = f (x)dx.
n→∞ n 0 1 + n cos2 x 0
1.65.
(a) Calculate
1 1
lim ln 1 + enx dx.
n→∞ n 0
(b) Let k ≥ 1 be an integer. Calculate
1 1
lim lnk 1 + enx dx.
n→∞ nk 0
1.66. [46] Let a, b ∈ R, with 0 < a < b. Calculate

1 b
n
lim lnn (1 + enx ) dx.
n→∞ n a
1.67.
(a) [45] Calculate

1
n
lim (1 + x n )n dx.
n→∞ 0
(b) [35] Let f : [0, 1] → (0, ∞) be a continuous function. Calculate

1
n
lim (1 + x n )n f (x)dx.
n→∞ 0
1.68.
(a) Calculate
1 n 1
lim n x dx.
n→∞ n 1
(b) Let f : [1, ∞) → R be a continuous function such that lim f (x) = L.

x→∞
Calculate
1 n 1
lim n x f (x)dx.
n→∞ n 1

1.69. Let f : 0, π2 → R be a continuous function. Prove that
π π
2 f (0) + f
lim n (cos x − sin x) 2n
f (x)dx = 2
.
n→∞ 0 2

1.70. [37] Let f : 0, π2 → R be a continuous function. Prove that
π 2n π
2 cos x − sin x f (0) + f
lim n f (x)dx = 2
.
n→∞ 0 cos x + sin x 4
1.71. [26, Remark, p. 67] Let f : [a, b] → [0, ∞) be a continuous function.

Calculate
⎛ ⎞
b b
lim n ⎝ f n (x)dx ⎠ .
n+1 n
f n+1 (x)dx −
n→∞ a a
1.72. Let f : [0, ∞) → [0, ∞) be a bounded continuous function. Calculate

∞ ∞
lim n n
f n+1 (x)e−x dx − n
f n (x)e−x dx .
n→∞ 0 0
1.73.
(a) Let a, b ∈ R, a < b, and let f : [a, b] → R be a Riemann integrable function.

Calculate
b f (x)
L = lim dx.
n→∞ a 1 + sin x sin(x + 1) · · · sin(x + n)
√
(b) Let q ∈ [1, 2). Calculate
b
f (x)
lim q n dx − L .
n→∞ a 1 + sin x sin(x + 1) · · · sin(x + n)
1.74.
(a) Let a, b ∈ R. Calculate

b
lim | sinn x| dx.
n→∞ a
(b) Let f : [a, b] → R be a Riemann integrable function. Calculate
b
lim f (x) sinn x dx.
n→∞ a
1.75. Limits with Exotic Integrals
(a) Let f : [0, 1] → R be a continuous function.

Prove that:
1 xn 1
• lim f (x)dx = f (x)dx;
n→∞ 0 x n + (1 − x)n 1
2
1 xn 1
• lim n f (x)dx − f (x)dx = 0.
n→∞ 0 x n + (1 − x)n 1
2
(b) If f ∈ C 1 ([0, 1], R), then

1 xn 1 π2 1
lim n 2
f (x)dx − f (x)dx =− f .
n→∞ 0 x + (1 − x)n
n 1
2
96 2
(continued)
xn
A challenge. Solve the same problem with x n +(1−x)n replaced by
x n −(1−x)n
x n +(1−x)n .
1.76.
(a) Let k ≥ 2 be an integer. Prove that
1 k−1
lim {nx} dx = ,
n→∞ 1
k
2k
where {x} denotes the fractional part of x.

(b) Prove that
1 xn 1
lim {nx} dx = ,
n→∞ 0 x n + (1 − x)n 4
1.77. Let f : R → R be a continuous function such that lim f (x) = f (−∞)

x→−∞
and lim f (x) = f (∞) exist and are finite. Prove that
x→∞
x
lim (f (t + 1) − f (t)) dt = f (∞) − f (−∞).
x→∞ −x
1.78. [66] Let f : [0, 1] → R be a continuous function. Calculate
n 1
lim x n f (x n ) ln(1 − x)dx.
n→∞ ln n 0
1.79. A problem of Radu Gologan. Let k ≥ 1 be an integer and f : [0, 1] → [0, 1]

be a Riemann integrable function. Prove that
1
lim n f n (x)(1 − f k (x))dx = 0.
n→∞ 0
1.80. An Open Problem Solved
(a) [95] Prove that

1
n
lim x(1 − x)x 2 (1 − x 2 ) · · · x n (1 − x n )dx = b(1 − b) = 0.185155 . . . ,
n→∞ 0
where b is the unique solution of the equation
1 x
dx = 1.
0 1 − bx
(b) Prove that

1 1 1 √ √
x 1+ 2 +···+ n (1 − x)(1 −
n
lim x) · · · (1 − n
x)dx = 0.
n→∞ 0
Remark 1.7. Details regarding this problem due to O. Furdui and the motiva-
tion of studying these limits are given in [26, problem 1.79, p. 15].
Series of Real Numbers
2
Read to get wise and teach others when it will be needed.

Saint Basil the Great (329–379)
The problems in this chapter deal with two major topics on series of real numbers:
the convergence of a numerical series and its exact calculation. Regarding the
convergence, the problems can be solved by using the classical way, i.e. the well-
known convergence tests for numerical series, or less known ways. Most of the
problems about the exact calculation of the value of a series are new in the literature
and vary in diversity as well as in difficulty. Their topics range from the calculation
of series involving harmonic numbers and factorials to exotic series involving tails
of Riemann zeta function values.
2.1 Miscellaneous Series
The Geometric Series Let k ≥ 0 be an integer and let q ∈ R. The geometric

∞
series q n converges if and only if |q| < 1, and we have
n=k
∞
qk
qn = .
1−q
n=k
(continued)

28 2 Series of Real Numbers
Remark 2.1. This formula has an equivalent formulation
the first term of the series

the sum of the geometric series = .
1 − the ratio
2.1. Calculate:
∞ ∞ n
1 1 + 3 + · · · + (2n − 1)
(a) (−1)n−1 ; (b) ;
7n (2n + 1) + (2n + 3) + · · · + (4n − 1)
n=1 n=1
∞ ∞
1 1
(c) ; (d) ;
n(n + 3) (2n − 1)(2n + 1)(2n + 3)
n=1 n=1
∞ ∞
1 n
(e) ln 1 − 2 ; (f) , |a| > 1;
n an
n=2 n=1
∞ ∞
n n
(g) ; (h) ;
n4 + n2 + 1 (n − 2)! + (n − 1)! + n!
n=1 n=2
∞ ∞
n(n + 1)(n + 2)(n + 3) + 1 n(n + 3)(n + 6)(n + 9) + 81
(i) ; (j) .
(n − 1)!(n2 + 3n + 1)2 n!(n2 + 9n + 9)
n=1 n=0
2.2. Series with Fractional Part Function
Let {x} be the fractional part of x. Prove that:

∞
√ 2n−1 1
(a) 2)(1 += ;
2
n=1
∞ √
√ n 3−1
(b) 1 − (2 + 3) = ;
2
n=1
∞ √
√ n 7−1
(c) 1 − (3 + 7) = .
3
n=1
We record a theorem, a gem of mathematical analysis, which uses in its proof the
geometric series.
2.1 Miscellaneous Series 29
Theorem 2.1. The set of prime numbers is infinite.
Proof. To prove that the set of prime numbers is infinite we use Euler’s series
∞ 1 π2
n=1 n2 = 6 . We assume, by way of contradiction, that the set of prime
numbers P is finite. Let P = {p1 , p2 , . . . , pk }. Then, for all n ∈ N we have
n = p1α1 p2α2 · · · pkαk , where αi ∈ N ∪ {0}, i = 1, 2, . . . , k. It follows that
∞
1 1
=
n2 2α1 2α2
· · · pk2αk
n=1 α1 ,α2 ,...,αk ≥0 p1 p2
∞ ∞ ∞
1 1 1
= 2α1
···
α1 =0 p1 α2 =0 p22α2 2αk
αk =0 pk
p12 p22 pk2

= · ··· · .
p12 − 1 p22 − 1 pk2 − 1
Therefore,
p12 p22 pk2 π2

· ··· · = ,
p12 −1 p22 −1 pk2 −1 6
which contradicts the irrationality of π 2 .

Remark 2.2. Using the formula ∞ n=1 n3 = ζ (3) and the irrationality of
1
ζ (3), one can prove similarly that the set of prime numbers is infinite.
2.3. Prove that

∞
1 1
= .
(2n + 1)2 − 1 4
n=1
2.4. Calculate:
∞ ∞
1 1
(a) ; (b) .
n2 (n + 1)2 n3 (n + 1)3
n=1 n=1
2.5. Calculate:
∞ ∞
1 1
(a) ; (b) ;
(2n + 1)(2n + 2) (3n + 1)(3n + 2)(3n + 3)
n=0 n=0
∞
1
(c)1 .
(4n + 1)(4n + 2)(4n + 3)(4n + 4)
n=0
Let A be a finite set of natural numbers of the following form 2m 3n 5p . Prove

2.6.
1
that x < 4.
x∈A
2.7. Prove that:

1 1 1 1 1 1 1 1 1 π ln 2
(a) 1 + − − + + − − + + − ··· = + ;
2 3 4 5 6 7 8 9 10 4 2√
1 1 1 1 1 1 1 1 1 1 1 2 3 ln 2
(b) 1 + + − − − + + + − − − +··· = π+ .
2 3 4 5 6 7 8 9 10 11 12 9 3
2.8. [126] Prove that

∞
1 e
= .
n!(n4 + n2 + 1) 2
n=0
2.9.

(a) [97, problem 15, p. 230] Prove that . . 1 − 2
1 . . . 2 = 3
. . . 3.
.
2n digits n digits n digits

(b) Determine the digits a, b, and c such that . . a − b
a . . . b = c .
. . . c.
2n digits n digits n digits
2.10. Prove that:
(a) 0.9999999 . . . = 1;
(b) 0.90909090 . . . = 10
11 ;
(c) 0.0909090909 . . . = 11 1
;
(d) 0.abababababab . . . = 10a+b99 , where a, b ∈ {0, 1, 2, . . . , 9}.
1 Thispart of the problem was given in day 1 of the student contest IMC 2010, Blagoevgrad,
Bulgaria.
2.11. Calculate
∞ ∞
1 (−1)n
and .
(3 + (−1)n )n (3 + (−1)n )n
n=1 n=1
If x, y ≥ 0 are real numbers, then
x−y
arctan x − arctan y = arctan .
1 + xy
2.12. Calculate
∞ ∞
1 1
arctan and arctan .
n +n+1
2 2n2
n=1 n=1
2.13. Let (Fn )n≥0 be the Fibonacci sequence.
(a) Prove that F2n = Fn+1

2 − F 2 , for all n ≥ 1.
n−1
(b) Calculate
∞
F2n
2 F2
.
n=2
Fn−1 n+1
2.14. Calculate the sum of the series

⎛ ⎞
∞
⎝ "√ 2 1 ⎠,
√ # − ln 1 +
n2 − n + 1 + n2 + n + 1 n
n=1
where x denotes the floor of x.
Mathematical Amusements: Dividing Integers!
• 64 : 16 = 4;
• 95 : 19 = 5;
• 98 : 49 = 2.

∞
(−1)n
2.15. Calculate ln 1 + n .
n=2
2.16. Resembling a Series of Ramanujan
(a) Check that
1 1 1 1 2
+ + = + , n ≥ 1.
3n − 1 3n 3n + 1 n (3n) − 3n
3
(b) Prove that

∞
2
= ln 3 − 1.
(3n)3− 3n
n=1
2.17. Two Telescoping Series
(a) Find the polynomial P of degree 2 such that
n2 P (n + 1) P (n)
n
= − n , ∀n ∈ N.
2 2n+1 2

∞
n2
(b) Prove that 2n = 6.
n=1
(c) Find the polynomial Q of degree 3 such that
n3 Q(n + 1) Q(n)
n
= − n , ∀n ∈ N.
2 2n+1 2

∞
n3
(d) Prove that 2n = 26.
n=1
2.18. Let k ∈ N. Prove that

∞
1 1
= 2 (2Hk − H2k ) .
n(n + k)(n + 2k) 2k
n=1
A gem with a numerical series.

∞
1 1
We prove that = .
n(n + 1)(n + 2) 4
n=1
1 1 1 1
Since = − we have that
n(n + 1)(n + 2) 2 n(n + 1) (n + 1)(n + 2)
(continued)
∞
∞ ∞

1 1 1 1
= −
n(n + 1)(n + 2) 2 n(n + 1) (n + 1)(n + 2)
n=1 n=1 n=1
∞
∞
1 1 1 1
= − −
2 n(n + 1) i(i + 1) 2
n=1 i=1
1
= .
4
This is a particular case of Problem 2.18 and also, of part (a) of Problem 2.19.
2.19. Let n ∈ N. Prove that:

∞
1 1
(a) [2] = ;
k(k + 1)(k + 2) · · · (k + n) n · n!
k=1
∞
1 1 1 1
(b) = ζ (2) − 1 − − · · · − .
k 2 (k + 1)(k + 2) · · · (k + n) n! 22 n2
k=1
2.20. [17]
(a) Let n ∈ N, n ≥ 2. Check that

1 n! k! (k + 1)!
= − .
Cnn+k n−1 (n + k − 1)! (n + k)!

∞
1
(b) Calculate Cnn+k
.
k=0
2.21. Prove that

1 !
1 ζ (2)
x dx = 1 − ,
0 x 2
where {a} is the fractional part of a.


⎧
1 dx ⎨ζ (2) − 1 if k = 1
*k+ =
0 ⎩k ζ (2) − 1 − 1
− ··· − 1
−1 if k ≥ 2,
x 22 (k−1)2
2.23. Let n ∈ N. Prove that

1 n
dx
" #n = ζ (n + 1) + (−1)j +n+1 ζ (j ) + (−1)n ,
0 1
x j =2

∞
1
√ < 2.
n(n + 1)
n=1
For the general case, see Problem 7.51.
2.25. [53] [O. Furdui, 2016] Let (an )n≥1 be a strictly increasing sequence of natural
numbers. Prove that the series
∞ √
an
[an , an+1 ]
n=1
converges. Here [a, b] denotes the least common multiple of a and b.
2.26. Prove that

∞
1 1 1 (−1)n−1 π2
1 − + − ··· + = .
n(n + 1) 2 3 n 12
n=1
2.27. A Criterion of Mircea Ivan

Let (xn )n≥1 be a sequence of positive
xn+1 n
real numbers such that lim xn < 1. Prove that lim xn = 0.
n→∞ n→∞
2.28. A series that equals 1. Prove that

∞
1 · 3 · · · (2n − 1) 1
· = 1.
2 · 4 · · · (2n) n+1
n=1
2.29. Let p, q ∈ R, with 0 < q < p.
(a) Prove that
q(q + p)(q + 2p) · · · (q + np)

lim = 0.
n→∞ p(2p) · · · ((n + 1)p)
(b) [130, p. 896] Prove that

∞
q(q + p)(q + 2p) · · · (q + np − p) 1 q
· = .
p(2p) · · · (np) n+1 p−q
n=1
For the above problem, see also Problem 2.64.
2.30. A product with radicals.

(a) Check that n4 + 4 = (n + 1)2 + 1 (n − 1)2 + 1 , ∀n ≥ 1.
,
∞ √
√n +1 = 2.
2
(b) Prove that
n=1
4 n +4
∞
2.31. A zeta series that equals 1. Prove that (ζ (n) − 1) = 1.
n=2
2.32. Prove that

∞ ∞
1 3
(−1)n (ζ (n) − 1) = and (ζ (2n) − 1) = .
2 4
n=2 n=1
∞
2.33. Prove that (−1)n (ζ (n + 1) − ζ (n)) = ζ (2) − 2.
n=2
2.34.
(a) Let p ≥ 3 be a prime number. Prove that

∞
(n, p) 2p − 1
(−1)n−1 = ln 2.
n p
n=1
(b) Prove that

∞ ∞
(n, 4) 11 (n, p) 1 1
= ζ (2) and = 1 + − 2 ζ (2),
n2 8 n2 p p
n=1 n=1
where p ≥ 2 is a prime number and (a, b) is the greatest common divisor of a

and b.
Stirling’s Formula for Approximating the Factorial

If n ∈ N, then for large values of n we have
√ n n
n! ∼ 2π n .
e
We mention that an ∼ bn , as n → ∞, if lim bann exists and is finite. For reference

n→∞
materials about Stirling’s formula the reader is referred to [26, p. 259].
2.35. Prove that

∞
1 1 γ √
n ln 1 + −1+ = 1 + − ln 2π
n 2n 2
n=1
and
∞
1 1 3 γ √
n ln 1 − +1+ = − + + ln 2π .
n 2n 2 2
n=2
2.36. Prove that

∞
1 1
n ln 1 − 2 + = γ − ln 2
n n
n=2
and
∞
n+1
n ln − 2 = 3 − ln(4π ).
n−1
n=2
2.37. An Alternating Logarithmic Series Prove that

∞
1 8
(−1) ln 1 − 2
n
= ln .
n π2
n=2
2.38. Prove that

∞
1 1 ln2 2
ln 1 + ln 1 + = .
2n 2n + 1 2
n=1

∞
1 1 1 ln3 2
ln 1 + ln 1 + ln 1 + = .
n 2n 2n + 1 3
n=1
2.40. Remarkable harmonic series.

Prove that:
∞ ∞
Hn Hn+1
(a) = ζ (2); (b) = 2;
n(n + 1) n(n + 1)
n=1 n=1
∞ ∞
Hn Hn2
(c) = 1; (d) = ζ (2) + 1.
(n + 1)(n + 2) (n + 1)(n + 2)
n=1 n=1
2.41. Two series with Catalan’s number.

Let Cn = 1 n
n+1 C2n be the nth Catalan number. Prove that:
1 n!(n − 1)!
(a) x n (1 − x)n−1 d x = , n ≥ 1;
0 √ (2n)!
∞
1 4 3
(b) =2+ π;
Cn 27
n=0
∞
2n 3
(c) = 5 + π.
Cn 2
n=0
2.42. Let k, p ≥ 1 be integers such that k + 1 < p. Prove that
∞ k
nk
= (−1)k−i Cik ζ (p − i).
(n + 1)p
n=1 i=0
2.2 Applications of Abel’s Summation Formula
Abel’s Summation Formula [26, p. 258]

Let (an )n≥1 and (bn )n≥1 be two sequences of real or complex numbers and
n
let An = ak . Then
k=1
(a) the finite version

n n
ak bk = An bn+1 + Ak (bk − bk+1 );
k=1 k=1
(b) the infinite version

∞ ∞
ak bk = lim An bn+1 + Ak (bk − bk+1 ).
n→∞
k=1 k=1
2.43. Prove that:

cos 1 cos 2 cos n
(a) lim + + ··· + = 0;
n→∞ n + 1 n+2 2n
2
cos 1 cos2 2 cos2 n √
(b) lim + + ··· + = ln 2.
n→∞ n + 1 n+2 2n
Remark 2.3. More generally, it can be proved that if k ≥ 1 is an integer,

then
2k−1
cos 1 cos2k−1 2 cos2k−1 n
• lim + + ··· + = 0;
n→∞ n+1 n+2 2n
2k
cos 1 cos2k 2 cos2k n Ck2k
• lim + + ··· + = 2k ln 2.
n→∞ n+1 n+2 2n 2
2.44. [O. Furdui, Problem 4, SEEMOUS 2017]

Let k ∈ N and let (xn )n≥k be the sequence defined by
(continued)
2.2 Applications of Abel’s Summation Formula 39
n
1 1 1
xn = Cki e − 1 − − − ··· − .
1! 2! i!
i=k
Prove that:
(a) (xn )n≥k is strictly increasing;

(b) (xn )n≥k is bounded;
e
(c) lim xn = .
n→∞ (k + 1)!
2.45. A Bouquet of Exponential Series

Let x ∈ R. Prove that:
∞
x x2 xn
(a) ex − 1 − − − ··· − = xex ;
1! 2! n!
n=0
∞
x x2 xn x2 x
(b) n e −1− −
x
− ··· − = e ;
1! 2! n! 2
n=1
∞ 3
x x2 xn x x2 x
(c) n e −1− −
2 x
− ··· − = + e .
1! 2! n! 3 2
n=1
2.46. Prove that

∞
1 1 π2
(−1)n−1 + + ··· = .
n2 (n + 1)2 8
n=1
2.47. [20 October 2018]

Let k ≥ 0 be an integer. Prove that:
∞
1 1 1 k
(a) ζ (2) − 1 − 2 − · · · − 2 − = Hk+1 + − ζ (2);
2 n n+k k+1
n=1
(continued)
In particular, when k = 0 we have that

∞
1 1 1
ζ (2) − 1 − 2 − · · · − 2 − = 1 − ζ (2);
2 n n
n=1
∞
1 1 1 k
(b) [82] + + · · · − = Hk+1 + ;
n 2 (n + 1)2 n+k k+1
n=1
In particular, when k = 0 we have that
∞
1 1 1
+ + ··· − = 1.
n2 (n + 1)2 n
n=1
2.48.
(a) Let n ∈ N and k > 0 be a real number. Prove that
1 1 1 1 x n−1
+ + + ··· = − ln x dx.
n2 (n + k)2 (n + 2k)2 0 1 − xk
(b) Let k > 0. Prove that

1 1 1 1
lim n + + + ··· = .
n→∞ n2 (n + k)2 (n + 2k)2 k
(c) [67] A series with gaps.

Prove that
∞
1 1 1 1 π2 1
+ + + ··· − = + .
n2 (n + 2)2 (n + 4)2 2n 16 2
n=1
A challenge. Let k ∈ N. Calculate

∞
1 1 1 1
Sk = + + + ··· − .
n2 (n + k)2 (n + 2k)2 kn
n=1
π2 5π 2
We mention that S1 = 1, S2 = 16 + 12 , and S4 = 64 + 1
4 + G
4.
2.49. An alternating series with gaps.

Prove that
∞
1 1 1 π2 ln 2
(−1)n−1 + + + ··· = + .
n2 (n + 2)2 (n + 4)2 16 2
n=1
2.50.
(a) Let n ∈ N and k > 0 be a real number. Prove that
1 1 1 1 1 x n−1 2
+ + + ··· = ln x dx.
n3 (n + k)3 (n + 2k)3 2 0 1 − xk
(b) [90] Let k > 0. Prove that

1 1 1 1
lim n 2
+ + + ··· = .
n→∞ n3 (n + k)3 (n + 2k)3 2k
(c) A series with gaps, ζ (2) and ζ (3).
Prove that
∞
1 1 1 π2 7
+ + + ··· = + ζ (3).
n3 (n + 2)3 (n + 4)3 12 16
n=1
Open problem. Let k > 0. Calculate

∞
1 1 1
+ + + ··· .
n3 (n + k)3 (n + 2k)3
n=1
Wallis Formula
2
2 4 2n 1 π
lim · ··· · = .
n→∞ 1 3 2n − 1 2n + 1 2
(continued)
Remark 2.4. The above formula implies that

∞
n+1 π
(−1)n−1 ln = ln .
n 2
n=1
2.51. A Wallis Product

Let (xn )n≥2 be the sequence defined by
n
(−1)k
xn = (−1)k ln 1 + − ln n.
k
k=2
Prove that:
(a) x2n = x2n+1 , ∀n ≥ 1;

2
(b) lim xn = ln ;
n→∞ π
∞
(−1)k 1 2
(c) (−1) ln 1 +
k
+ ln 1 − = ln .
k k π
k=2
Remark 2.5. Part (c) of the problem has the following equivalent formulation:
∞
- (−1)k
k k π
= .
k + (−1)k k−1 2
k=2
2.52. [10, p. 277] An alternating series of Hardy

Prove that
∞
1 1 1 γ − ln π
(−1)n 1 + + + · · · + − ln n − γ = .
2 3 n 2
n=1
2.53. Prove that:

∞
1 1 1 γ − ln 2
(a) (−1) 1 + + + · · · + − ln n(n + 1) − γ =
n
;
2 3 n 2
n=1
(continued)
∞
1 1 1
(b) (−1) 1 + + + · · · + − ln n(n + 1)(n + 2) − γ
n 3
2 3 n
n=1
γ ln π
= − ;
2 6
∞
1 1 1
(c) [43] 1 + + + · · · + − ln n(n + 1) − γ
2 3 n
n=1
1 √
= − ln 2π + γ .
2
2.54. [50]
(a) Prove that

∞
1 1 1 1 3
1 + + · · · + − ln n + − γ = γ + − ln 2.
2 n 2 2 2
n=1
A challenge. Prove that

∞
1 1 1
n−1
(−1) 1 + + · · · + − ln n + −γ
2 n 2
n=1

γ 5 1
= − + ln 2 + ln π − 2 ln .
2 2 4
(b) Let p ∈ N. Prove that

∞
1 1 1 2−p
1 + + · · · + − ln k + −γ +
2 k p 2pk
k=1

1 1 1 √ 1
=γ + + − ln 2π + ln 1 + ,
2 p 2 p
where denotes the Gamma function.

2.55. Let On = 1 + 1
3 + ··· + 1
2n−1 , n ∈ N. Prove that:
∞
γ ln n 1 2
(a) On − − ln 2 − = γ + ln ;
2 2 4 π
n=1
∞
γ ln n 1 8 π
(b) (−1)n On − − ln 2 − = γ + ln − .
2 2 4 π 2
n=1
2.56. [31] A Harmonic Series with the Tail of ζ (3)
(a) Prove that for any n ∈ N we have
H1 + H2 + · · · + Hn = (n + 1)Hn − n.
(b) Prove that

∞
1 1
Hn ζ (3) − 1 − 3 − · · · − 3 = 2ζ (3) − ζ (2),
2 n
n=1
where Hn = 1 + 1
2 + ··· + 1
n denotes the nth harmonic number.
Remark 2.6. Other spectacular series involving the nth harmonic number Hn
and tails of Riemann zeta function values, of which some are listed below
∞
1 1 5
Hn ζ (4) − 1 − 4 − · · · − 4 = ζ (4) − ζ (3),
2 n 4
n=1
∞
1 1 1 5 5
nHn ζ (4) − 1 − 4 − · · · − 4 = − ζ (2) + ζ (3) − ζ (4),
2 n 4 4 8
n=1
∞
1 1
Hn2 ζ (3) − 1 − 3 − · · · − 3 = 3ζ (4) − 4ζ (3) + 2ζ (2),
2 n
n=1
can be found in [30].

2.3 Series with Positive Terms 45
2.3 Series with Positive Terms
Comparison Criteria for Series with Positive Terms

∞
∞
A. Let an and bn be two series with positive terms with the property
n=1 n=1
that there exist α > 0 and n0 ∈ N such that an ≤ αbn , for all n ∈ N,
n ≥ n0 .

∞
∞
(a) If the series bn converges, then the series an converges.
n=1 n=1
∞
∞
(b) If the series an diverges, then the series bn diverges.
n=1 n=1

∞
∞
B. Let an and bn be two series with positive terms such that the limit
n=1 n=1
exists lim an = l ∈ [0, ∞].
n→∞ bn

∞
∞
(a) If l ∈ (0, ∞), then the series an and bn have the same nature.
n=1 n=1

∞
(b) If l = 0, then the convergence of the series bn implies the
n=1

∞
convergence of the series an .
n=1

∞
(c) If l = ∞, then the convergence of the series an implies the
n=1

∞
convergence of the series bn .
n=1
2.57. Study the nature of the series:
∞ ∞ ∞
1 1 1
(a) √ ; (b) √ ; (c) ;
n +1
3 5
n(n + 2) ln(n + 1)
n=1 n=1 n=1
∞ ∞ ∞
1 1 1 π
(d) ln 1 + 3 ; (e) arctan ; (f) sin ;
n n n 2n
n=1 n=1 n=1
∞
π
∞ ∞
1 n2
(g) √ ; (h) ; (i) 1 − cos .
nn n+1 n(n + 1)5n + 3 n
n=1 n=1 n=1
2.58. Convergence of Series with Norms of Vectors in Rn

Let v = (v1 , v2 , . . . , vn ) ∈ Rn , let ||v||p = p |v1 |p + |v2 |p + · · · + |vp |p
and ||v||∞ = max1≤i≤n |vi |. These are the usual p-norm and ∞-norm on Rn .
(a) Prove that

lim p ||v||p − ||v||∞ = ||v||∞ ln k,
p→∞
where k is the number of entries of v which are equal to either −||v||∞ or

||v||∞ .
(b) [19] If k = 1, then prove that the series
∞

pα ||v||p − ||v||∞
p=1
converges for all α ≥ 0.

Application. Prove that the series
∞ √
nα 1 + 2n + 3n + · · · + 2021n − 2021
n
n=1
converges for all α ≥ 0.
2.59. [25 April 2020] Let k ∈ N and a1 , a2 , . . . , ak > 0. Prove that

√ √
n a + n a + ··· + n a
√ n
1 2 k √
lim n − k a1 a2 · · · ak
n→∞ k
√
k a a ···a
1 2 k √
= ln2 a1 + ln2 a2 + · · · + ln2 ak − ln2 k a1 a2 · · · ak .
2
Application. Study the nature of the series
∞ √ √ √ n
n a1 + n a2 + · · · + n a
k √
− k
a1 a2 · · · ak .
k
n=1
A Consequence of d’Alembert’s Ratio Test for Series with Positive Terms

∞
Let an be a series with positive terms such that the limit exists lim an+1
an =
n=1 n→∞
l ∈ [0, ∞].

∞
(a) If l < 1, then the series an converges.
n=1
∞
(b) If l > 1, then the series an diverges.
n=1
If l = 1, then the test is inconclusive.
∞ ∞
an an
(a) , a > 0, p ∈ R; (b) , a > 0;
np n(7n + 5n )
n=1 n=1
∞ ∞
32n (n!)3 1 · 4 · 7 · · · (3n − 2)
(c) ; (d) .
(3n)! 22n · n!
n=1 n=1
[12] A Convergence Criterion of Chen

Let (an )n≥1 be a strictly decreasing sequence of positive numbers such that
the limit exists
a2n
lim = l ∈ [0, ∞].
n→∞ an

∞
(a) If l < 12 , then the series an converges.
n=1
∞
(b) If l > 12 , then the series an diverges.
n=1
If l = 12 , then the test is inconclusive.

∞ ∞ ∞
1 1 1
(a) √ ; (b) ; (c) .
3 n (ln n)ln n (ln n)ln(ln n)
n=1 n=2 n=2
A Consequence of Cauchy’s Root Test for Series with Positive Terms

∞
Let an be a series with positive terms such that the following limit exists
√
n=1
lim n an = l ∈ [0, ∞].
n→∞

∞
(a) If l < 1, then the series an converges.
n=1
∞
(b) If l > 1, then the series an diverges.
n=1
∞ n ∞ n2
an2 1 n
(a) , a > 0; (b) 1+ ;
n2 − n + 1 3n n2 + 1
n=1 n=1
∞ ∞
1 n 1
(c) n sin ; (d) nn+1 lnn 1 + .
2n 3n
n=1 n=1
A Consequence of Raabe–Duhamel’s Test for Series with Positive Terms

∞
Let an be a series with positive terms such that the following limit exists
n=1
an
lim n an+1 − 1 = l ∈ [0, ∞].
n→∞

∞
(a) If l > 1, then the series an converges.
n=1
∞
(b) If l < 1, then the series an diverges.
n=1

Series for which d’Alembert and Raabe–Duhamel’s tests fail to hold:

∞ 1 ∞
n=2 n lnk n , k ∈ R, k > 1;
1
• convergent n=1 nHnk ,
∞ ∞ x−2
• divergent 1
n=2 n ln n , n=1 n ζ (x) − 1 − 21x − · · · − n1x , x ∈ R,
x ≥ 2.
2.63. Study the nature of the series2 :
∞ ∞
(2n − 1)! 1 · 3 · 5 · · · (2n − 1) 1
(a) ; (b) · ;
22n−1 (n − 1)!(n + 1)! 2 · 4 · 6 · · · (2n) 2n + 1
n=1 n=1
∞ ∞
1 · 5 · 9 · · · (4n − 3)
(c) a ln n , a > 0; (d) , a > 0;
4n · a(a + 1) · · · (a + n − 1)
n=1 n=1
∞ ∞
1 1 1 1
(e) x 1+ 2 +···+ n , x > 0; (f) x sin 1+sin 2 +···+sin n , x > 0.
n=1 n=1
2.64. Let a, b, c > 0, with b > c.
(a) Prove that

b−c
c(a + c) · · · (an + c − a) e− max{a,b} (Hn −ln n)
≤ , ∀ n ≥ 1.
b(a + b) · · · (an + b − a) b−c
n max{a,b}
(b) Let
c(a + c) · · · (an + c − a) 1
an = · , n ≥ 1.
b(a + b) · · · (an + b − a) an + b
Prove that
1
an = [an−1 (an + c − a) − an (an + c)] , ∀ n ≥ 2.
b−c
(continued)
2 Thesum of the series in part (b) is equal to π2 − 1. See the power series expansion of arcsin. The
1
sum of the series in part (d) is equal to 4a−5 , when a > 54 . See Problem 2.64 for the general case.

∞
(c) A Telescoping Series Prove that the series an converges and
n=1
∞
c(a + c) · · · (an + c − a) 1 c
· = .
b(a + b) · · · (an + b − a) an + b b(b − c)
n=1
Applications. Prove that:

∞
1 · 4 · · · (3n − 2) 1 1
(1) · = ;
2 · 5 · · · (3n − 1) 3n + 2 2
n=1
∞
1 · 5 · · · (4n − 3) 1 1
(2) · = ;
2 · 6 · · · (4n − 2) 4n + 2 2
n=1
∞
2 · 6 · · · (4n − 2) 1 2
(3) · = ;
3 · 7 · · · (4n − 1) 4n + 3 3
n=1
∞
3 · 6 · · · 3n 1 3
(4) [81] · = .
7 · 10 · · · (3n + 4) 3n + 7 28
n=1
2.65.
(a) [7, problem 3.35, p. 24] Let a, b, c be positive numbers. Prove that the series
∞
1 1

1 bn + cn
a −
n
2
n=1
converges if and only if a 2 = bc.

(b) [51] Let a, b, c be positive numbers. Prove that the series
∞
1 1

bn + cn
1 a
n· a − n − ln √
2 bc
n=1
converges if and only if 2 ln2 a = ln2 b + ln2 c.
Cauchy’s Integral Criterion Letf : [1, ∞) → (0, ∞) be a monotone

decreasing function. The series ∞ n=1 f (n) converges if and only if the
n
sequence (vn )n≥1 , defined by vn = 1 f (x) dx, converges.
2.66. The Generalized Harmonic Series of Riemann

Let p ∈ R. Prove that the series ∞
n=1 n
1
p converges for p > 1 and diverges
for p ≤ 1.
We record a gem of mathematical analysis regarding the convergence of the

generalized harmonic series of Riemann.

∞
• The generalized harmonic series of Riemann 1
np , p ∈ (1, ∞), con-
n=1
verges.
1 1 1 p p
1+ + p + ··· + p < , ∀ n ≥ 1 ⇒ 1 < ζ (p) ≤
2p 3 n p−1 p−1
.

∞
• The generalized harmonic series of Riemann 1
np , p ∈ (0, 1], diverges.
n=1

∞
1
If the series np would converge, then
n=1
∞ ∞ ∞ ∞
1 1 1 2 1 1
= + > = p−1 ,
np (2n + 1) p (2n + 2)p (2n + 2) p 2 (n + 1)p
n=1 n=0 n=0 n=0
and it follows that 2p−1 > 1, which is a contradiction.
2.67. The Harmonic Series Diverges

(a) Use the inequality ex ≥ 1 + x, ∀x ≥ 0, to prove that

1 1 1 1 1
1+ 1+ ··· 1 + ≤ e1+ 2 +···+ n , ∀ n ≥ 1.
1 2 n

∞
(b) Prove, by using part (a) of the problem, that 1
n = ∞.
n=1
Inequalities and Divergent Series

∞
H1 H2 Hn n Hn
(a) + + ··· + > √n
n
H1 H2 · · · Hn ⇒ diverges.
1 2 n n! n
n=1
(continued)
ln 2 ln 3 ln(n + 1) n
(b) + + ··· + > √ n
ln 2 ln 3 · · · ln(n + 1)
2 3 n+1 n
(n + 1)!
∞
ln n
⇒ diverges.
n
n=2
∞
1 1 1 n2 1
(c) + + ··· + > ⇒ diverges.
ln 2 ln 3 ln(n + 1) ln((n + 1)!) ln n
n=2
(d) p ∈ (0, 1)
∞
1 1 1 n p 1−p 1
p
+ p
+ · · · + p
> √ n ⇒ diverges.
1 2 n n
n! np
n=1
2.68. Let p be a positive number. Study the nature of the series:

∞ 1
∞ 1 ∞ ln n ∞ ln n
(a) ; (b) ; (c) ; (d) .
n=2 n ln n n=2 n lnp n n=2 n n=2 n
p
2.69. Study the nature of the series

∞
ln n
√ √ √ .
n=2
1+ 2 + 3 3 + ··· + n n
2.4 Alternating Series
Leibniz’s Criterion for Alternating Series

If (an )n≥1 is a monotone
decreasing sequence convergent to 0, then the
alternating series ∞n=1 (−1)
n−1 a converges.
n
2.70. Determine which of the following series are semiconvergent or absolutely

convergent3 :
∞ ∞
n=1 |an | converges. A series is
3A series n=1 an is absolutely convergent if the series
semiconvergent if the series converges, but it does not converge absolutely. The sum of the series
2
in part (d) is γ ln 2 − ln2 2 .
2.4 Alternating Series 53
∞ ∞ ∞
1 1 1
(a) (−1)n−1 ; (b) (−1)n ; (c) (−1)n−1 √ ;
n n2 nnn
n=1 n=1 n=1
∞ ∞ ∞
ln n (n + 1)n+1 1
(d) (−1)n ; (e) (−1)n−1 ; (f) (−1)n tan √ .
n nn+2 n n
n=2 n=1 n=1
2.71. Study the nature of the following series:
∞ ∞
(−1)n (−1)n
(a) ; (b) ;
n + sin n n + cos n
n=1 n=1
∞ ∞
(−1)n (−1)n
(c) ; (d) .
n + (−1)n sin n n + (−1)n cos n
n=1 n=1
Remark 2.7. We mention that if (an )n≥1 is a sequence

which converges to 0, but
not monotonically, then the alternating series ∞n=1 (−1) n a may or may not be
n
convergent, as the next problem shows.
. .
1 1
if n is even if n is even
2.72. Let an = n
2
and bn = n2
2
.
n if n is odd n2
if n is odd

∞
∞
5π 2
Prove that (−1)n−1 an = ∞ and (−1)n−1 bn = 24 .
n=1 n=1
.
a
np if n is even
2.73. Let a, b > 0, p ∈ (0, 1] and let an = b
.
np if n is odd

∞
Prove that (−1)n an converges if and only if a = b.
n=1
2.74.
(a) Botez–Catalan identity. Let n ∈ N. Prove that
1 1 1 1 1 1 1 1
1− + − + ··· + − = + + ··· + .
2 3 4 2n − 1 2n n+1 n+2 2n

∞
(b) Calculate (−1)n−1 n1 .
n=1
2.75. Prove that

∞
(−1)n−1 1
lim x = .
x→∞ n+x 2
n=1
2.76. Limits of Special Series

Prove that:
∞
h π
(a) lim = ;
h→0+ 1 + (nh)2 2
n=1
∞
h π
(b) lim = ;
h→0+ 1 + (2n − 1) h
2 2 4
n=1
∞
h π
(c) lim = , where k ∈ (1, ∞);
h→0+ 1 + (nh)k k sin πk
n=1
∞
h π
(d) lim = , where k ∈ (1, ∞).
h→0+ 1 + (2n − 1) h
k k 2k sin πk
n=1
∞ (2016) proved that, if f : [0, ∞) → R is a decreasing

Remark 2.8. M. Ivan
function such that 0 f (x)dx converges, then
∞ ∞ ∞ ∞
1
lim hf (nh)= f (x)dx and lim hf ((2n − 1)h)= f (x)dx.
h→0+ 0 h→0+ 2 0
n=1 n=1
2.5 Series with Harmonic Numbers and Factorials
2.77. Remarkable Series with Harmonic Numbers and Factorials

(a) Check that

(j − 1)! 1 (j − 1)! j!
Hi+j = Hi+j −1 − Hi+j
(i + j )! i (i + j − 1)! (i + j )!

1 (j − 1)! j!
+ 2 − .
i (i + j − 1)! (i + j )!
(b) [148, O. Oloa, 30 December 2005, entry 25]

Prove that
(continued)
2.5 Series with Harmonic Numbers and Factorials 55
∞ ∞
(i − 1)!(j − 1)!
Hi+j = 3ζ (3).
(i + j )!
i=1 j =1
(c) [O. Furdui, 13 September 2019]

Prove that
∞ ∞
(i − 1)!(j − 1)! 11
(−1)i−1 Hi+j = ζ (3).
(i + j )! 8
i=1 j =1
(d) [O. Furdui, 17 September 2019]

An alternating series. Prove that
∞ ∞ (i − 1)!(j − 1)! 11
(−1)i−1 + (−1)j −1 Hi+j = ζ (3).
(i + j )! 4
i=1 j =1
Remark 2.9. One can prove, and this is a challenging problem, that the
alternating version of Oloa’s series
∞ ∞
(i − 1)!(j − 1)!
(−1)i+j Hi+j
(i + j )!
i=1 j =1

is equal to 3ζ (3)+2Li3 − 13 −2Li3 13 +2Li3 − 12 + 23 ln3 2−ln2 2 ln 3+

2 ln 2Li2 − 12 .
A Double Series with Harmonic Numbers and Factorials

We record as a gem another proof of Oloa’s double series formula
∞ ∞
(i − 1)!(j − 1)!
Hi+j = 3ζ (3).
(i + j )!
i=1 j =1
We have
(continued)
∞ ∞ ∞ ∞
(i − 1)!(j − 1)! Hj +i
Hi+j = (i − 1)!
(i + j )! j (j + 1) · · · (j + i)
i=1 j =1 i=1 j =1
∞
(i − 1)! 1
= + Hi
i · i! i
i=1
∞ ∞
1 Hi
= +
i3 i2
i=1 i=1
= 3ζ (3).
We used in the preceding calculations the formula

∞
Hn+k 1 1
= + Hk , k ∈ N,
n(n + 1) · · · (n + k) k · k! k
n=1
whosesolution is given in detail in [26, problem 3.59 (b), p. 149], and Euler’s
series ∞ Hn
n=1 n2 = 2ζ (3).
2.78. Prove that

∞ ∞
(i − 1)!(j − 1)!
Hi+j +1 = 6 − ζ (2) − 3ζ (3).
(i + j + 1)!
i=1 j =1
2.6 A Mosaic of Series

∞
1 1 π2 ln 2 1
(−1) n−1
n ζ (2) − 1 − 2 − · · · − 2 − 1 = − − .
2 n 16 2 2
n=1
Open problem. Let k ≥ 4 be an integer. Calculate

∞
1 1 1
(−1)n−1 nk−1 ζ (k) − 1 − k − · · · − k − .
2 n k−1
n=1
(continued)
2.6 A Mosaic of Series 57
The case when k = 3 is part (d) of Problem 2.82.
2.80. [64] A Nonlinear Harmonic Series
(a) Check that, for n ≥ 2, we have

Hn Hn+1 1 Hn−1 Hn Hn Hn+1 Hn−1 Hn
= − + −
n3 − n 2 (n − 1)n n(n + 1) n−1 n

3 1 1 Hn 1 Hn−1 Hn+1
+ − − 2− −
4 n−1 n 2n 4 n−1 n+1
1
− .
4(n + 1)2
(b) Prove that

∞
Hn Hn+1 5 1
= − ζ (2) − ζ (3).
n −n
3 2 4
n=2
2.81. [27 July 2018]
(a) Prove that
1 1 1 1 xn
− + − ··· = dx, n ∈ N.
n+1 n+2 n+3 0 1+x
(b) Prove that

∞
Hn 1 1 1 1 1 1
− + − ··· = ζ (3)+ ζ (2) ln 2+ ln3 2.
n n+1 n+2 n+3 4 2 6
n=1
(c) Prove that

∞
Hn 1 1 1 1 1 1
(−1)n − + − ··· = ζ (3)− ζ (2) ln 2+ ln3 2.
n n+1 n+2 n+3 4 2 6
n=1
(continued)
(d) Prove that

∞ 2
1 1 1 π2 1
Hn − + − ··· = + ln2 2 − ln 2.
n+1 n+2 n+3 12 2
n=1
A challenge. Calculate
∞ 2
1 1 1
(−1) Hnn
− + − ··· .
n+1 n+2 n+3
n=1
A Gem with an Alternating Quadratic Series

We prove that
∞ 2
1 1 1 π2
(−1) n−1
− + − ··· = .
n n+1 n+2 24
n=1
We have
∞ 2
1 1 1
(−1) n−1
− + − ···
n n+1 n+2
n=1
∞ 2
1 1 1
= ln2 2 + (−1)n−1 − + − ···
n n+1 n+2
n=2
∞ 2
n−1=m 1 1 1
= ln 2 +
2
(−1) m
− + − ···
m+1 m+2 m+3
m=1
∞ 2
1 1 1 1
= ln2 2 − (−1)m−1 − − + − ···
m m m+1 m+2
m=1
∞ ∞
(−1)m (−1)m−1 1 1 1
= ln 2 +
2
+2 − + − ···
m2 m m m+1 m+2
m=1 m=1
∞ 2
1 1 1
− (−1)m−1 − + − ···
m m+1 m+2
m=1
(continued)
and it follows that

∞ 2
1 1 1
2 (−1) n−1
− + − ···
n n+1 n+2
n=1
∞ ∞
(−1)n (−1)n−1 1 1 1
= ln2 2 + +2 − + − ··· .
n2 n n n+1 n+2
n=1 n=1
Since
1 1 1 1 x n−1
− + − ··· = dx,
n n+1 n+2 0 1+x
we obtain that
∞ ∞ 1
(−1)n−1 1 1 1 (−1)n−1 x n−1
− + − ··· = dx
n n n+1 n+2 n 0 1+x
n=1 n=1
1 ∞
1 (−x)n
=− dx
0 x(1 + x) n
n=1
1 ln(1 + x)
= dx
0 x(1 + x)
1 ln(1 + x) 1 ln(1 + x)
= dx − dx
0 x 0 1+x
π2 ln2 2
= − .
12 2
Therefore,
∞ 2
1 1 1 π2
2 (−1) n−1
− + − ··· = .
n n+1 n+2 12
n=1
We mention that this quadratic series was calculated by a different method

in [26, problem 3.45].
2.82. [28 January 2019]
(a) Prove that
−1 + (2n + 1)(−1)n
(−1) · 1 + (−1)2 · 2 + · · · + (−1)n · n = , n ≥ 1.
4
(b) Prove that
∞
1 1 1 1 3
(−1)n n ζ (2) − 1 − 2 − · · · − 2 − = 2 + 2 ln 2 − ζ (2) .
2 n n 4 2
n=1
(c) Prove that
n(n + 1)
(−1) · 12 + (−1)2 · 22 + · · · + (−1)n · n2 = (−1)n , n ≥ 1.
2
(d) Prove that
∞
1 1 1 1 1 ζ (2)
(−1)n n2 ζ (3) − 1 − 3 − · · · − 3 − 2 = ln 2 + − .
2 n 2n 2 2 2
n=1
2.83.
(a) Prove that, if k ∈ N, k ≥ 2, then

∞
1 1 ζ (k)
(−1)n−1 ζ (k) − 1 − k − · · · − k = k .
2 n 2
n=1
(b) Prove that, if k ∈ N, k ≥ 3, then

∞
1 1
(−1) n ζ (k) − 1 − k − · · · − k
n−1
2 n
n=1

1 1 1 1
= ζ (k) − k+1 − ζ (k − 1) − k−1 .
2 2 2 2
2.84. A Bouquet of Factorial Series

Prove that:
∞
1 2 3 3
(a) + + + · · · = e;
n! (n + 1)! (n + 2)! 2
n=1
∞
1 2 3 13
(b) n + + + ··· = e;
n! (n + 1)! (n + 2)! 6
n=1
∞
1 2 3 47
(c) n2 + + + ··· = e.
n! (n + 1)! (n + 2)! 12
n=1
2.85. [16 August 2019]
(a) Let k > 3. Prove that

∞
1 2 3 1
+ + + ··· = (ζ (k − 2) + ζ (k − 1)).
nk (n + 1)k (n + 2)k 2
n=1
(b) Let k > 4. Prove that

∞
1 2 3 1
n + + + ··· = (ζ (k − 3)+3ζ (k − 2) + 2ζ (k − 1)) .
nk (n + 1)k (n + 2)k 6
n=1
2.86. [28 January 2019]
(a) A Quadratic Series with the Tail of ζ (2)

Prove that
∞
1 1 1 2 3 3 1
n ζ (2) − 1 − 2 − · · · − 2 − = ζ (2) − ζ (3) − .
2 n n 2 2 2
n=1
(b) A Quadratic Series with the Tail of ζ (3)

Prove that
∞
1 1 1 2 1 1 3 1
n3 ζ (3) − 1 − 3 − · · · − 3 − 2 = − ζ (2) + ζ (3) + ζ (4) − .
2 n 2n 2 2 8 16
n=1
2.87. An Amazing Series with Double Factorials and the Tail of ζ (2)
Let
π
2
In = x 2 cos2n x dx, n ≥ 0.
0
(a) Prove that
2n − 1 1 1 · 3 · · · (2n − 1) π
In = In−1 − 2 · · , n ≥ 1.
2n n 2 · 4 · · · (2n) 4
(b) Prove that

1 · 3 · · · (2n − 1) 1 1 π
In = · ζ (2) − 1 − 2 − · · · − 2 · , n ≥ 1.
2 · 4 · · · (2n) 2 n 4
(c) Prove that

∞
1 · 3 · · · (2n − 1) 1 1
· ζ (2) − 1 − 2 − · · · − 2 = 4 ln 2 − ζ (2).
2 · 4 · · · (2n) 2 n
n=1
Power Series
3
Learned is a man who is happy to keep learning.

Nicolae Iorga (1871–1940)
This chapter is concentrated on power series, as the title says. It includes the
standard formulae on the power series, Taylor and Maclaurin series of the classical
elementary functions, and exercises dealing with finding the set of convergence
of a power series and, whenever possible, determining the sum of the series. The
problems in this chapter are as exotic and fascinating as possible. They cover topics
such as the generating functions of special sequences, the calculation of power series
with factorials, series involving polylogarithm functions, and nonlinear logarithmic
series. The chapter also includes, as applications, the calculation of single and
multiple series involving Riemann zeta function values. The last section of this
chapter collects several nonlinear Euler series which are new in the literature, but
the original problems can be found in other sections, too.
3.1 Convergence and Sum of Power Series
The Radius and the Convergence set of a Power Series

∞
Let an x n , an ∈ R, be a power series and let R be its radius of convergence.
n=0
Then
1
R= √ ∈ [0, +∞].
lim n
|an |
n→∞
(continued)

64 3 Power Series
If R = 0, then the convergence set of the power series is {0}.

If R ∈ (0, +∞), then (−R, R) is the interval of convergence of the power
series. The nature of the power series has to be studied for x = −R and x =
R, and the convergence set can be (−R, R), [−R, R), (−R, R] or [−R, R].
If R = +∞, / then/ the convergence set of the power series is R.
If limn→∞ / an+1
an /
exists, then the radius of convergence of the power series
can be calculated by the formula
/ /
/ an /
R = lim // /.
/
n→∞ a n+1
3.1. Find the convergence set of the following power series:
∞ ∞ ∞
xn xn (x + 1)n
(a) ; (b) ; (c) (−1)n ;
2n + 5n (n + 1)3n n+7
n=0 n=0 n=1
∞ ∞ ∞
(x − 2)n (x − 1)2n xn
(d) ; (e) ; (f) ;
(3n − 1)5n n4n nn+1
n=1 n=1 n=1
∞ 2 ∞ ∞
1 n −n n n2n n
(g) 1− x ; (h) x ; (i) Hn x n ;
n (n!)3
n=1 n=0 n=1
∞ ∞ n2 ∞ 5
n n xn xn
(j) n x ; (k) 2
; (l) 3
.
n=1 n=1
nn n=1
32n nn
3.2. Find the convergence set of the power series

∞
x n! = 1 + x + x 2 + x 6 + x 24 + x 120 + · · · .
n=0
3.3. Find the convergence set and the sum of the following power series:
∞ ∞
(a) (n + 1)x n+1 ; (b) (n + 1)(n + 2)x n ;
n=0 n=0
∞ ∞
x 3n−1
(c) (−1)n−1 ; (d) (−1)n−1 (5n − 1)x 5n−2 ;
3n − 1
n=1 n=1
∞ ∞
x 4n−1
(e) (−1) (n + 1) x ;
n 3 n
(f) .
4n − 3
n=0 n=1
3.1 Convergence and Sum of Power Series 65
3.4.
A. Find the convergence set of the following power series:

∞
1 1
(a) 1 + + · · · + − ln n − γ x n ;
2 n
n=1
∞
1 1 1
(b) e − 1 − − − ··· − xn;
1! 2! n!
n=1
∞
1 1 1
(c) ζ (k) − k − k − · · · − k x n , k ≥ 2.
1 2 n
n=1
B. Prove that
⎧ x
∞
1 1 1
⎨e − e x + 1 if x = 1
e − 1 − − − ··· − x =
n
x−1
1! 2! n! ⎩
n=1 1 if x = 1.
The polylogarithm function Lik , k ≥ 2, is the special function defined

by
∞
xn
Lik (x) = , x ∈ [−1, 1].
nk
n=1
C. Prove that, for k > 2, we have

⎧
xζ (k) − Lik (x)
1 n ⎨
∞
1 1 if x ∈ [−1, 1)
ζ (k) − k − k − · · · − k x = 1−x
1 2 n ⎩
n=1 ζ (k − 1) − ζ (k) if x = 1
and
∞
1 1 1 xζ (2) − Li2 (x)
ζ (2) − 2 − 2 − · · · − 2 x n = , x ∈ [−1, 1).
1 2 n 1−x
n=1
66 3 Power Series
3.2 Maclaurin Series of Elementary Functions
The Maclaurin Series of Some Elementary Functions

∞
x 2n+1
sin x = (−1)n , x ∈ R;
(2n + 1)!
n=0
∞
x 2n
cos x = (−1)n , x ∈ R;
(2n)!
n=0
∞
(2n − 1)!! x 2n+1
arcsin x = x + · , x ∈ [−1, 1];
(2n)!! 2n + 1
n=1
∞
x 2n+1
arctan x = (−1)n , x ∈ [−1, 1];
2n + 1
n=0
the binomial series
α α(α−1) 2 α(α−1) · · · (α − n+1) n
(1+x)α = 1+ x+ x + ··· + x + ··· ,
1! 2! n!
x ∈ (−1, 1), α ∈ R;
∞ ∞
1 1
= (−1)n x n , x ∈ (−1, 1); = xn, x ∈ (−1, 1);
1+x 1−x
n=0 n=0
∞
1 (α + n) n
= x , α > 0, x ∈ (−1, 1),
(1 − x)α (α)n!
n=0
where is the Gamma function of Euler;
the logarithmic series
∞
xn
ln(1 + x) = (−1)n−1 , x ∈ (−1, 1];
n
n=1
∞
xn
ln(1 − x) = − , x ∈ [−1, 1);
n
n=1
the exponential series
∞ ∞
xn xn
ex = , x ∈ R; e−x = (−1)n , x ∈ R.
n! n!
n=0 n=0
3.2 Maclaurin Series of Elementary Functions 67
Infinite Nested Product Formulae

(a) the geometric series I
1
= 1 + x (1 + x (1 + x (1 + x (1 + · · · )))) , x ∈ (−1, 1);
1−x
(b) the geometric series II
1
= 1 − x (1 − x (1 − x (1 − x (1 − · · · )))) , x ∈ (−1, 1);
1+x
(c) the logarithmic series I

x 2x 3x
− ln(1−x) = x 1 + 1+ 1+ (1 + · · · ) , x ∈ [−1, 1);
2 3 4
(d) the logarithmic series II

x 2x 3x
ln(1+x) = x 1 − 1− 1− (1 − · · · ) , x ∈ (−1, 1];
2 3 4
(e) the binomial series

αx (α − 1)x (α − 2)x
(1 + x) = 1 +
α
1+ 1+ (1 + · · · ) ,
1 2 3
x ∈ (−1, 1), α ∈ R;
(f) the constant e

1 1 1 1 1
e=1+ 1+ 1+ 1+ 1 + (1 + · · · ) ;
1 2 3 4 5
(g) the Maclaurin series of the function ex

x x x x x
ex = 1 + 1+ 1+ 1+ 1 + (1 + · · · ) , x ∈ R;
1 2 3 4 5
(h) the Maclaurin series of the function e−x

x x x x x
e−x = 1 − 1− 1− 1− 1 − (1 − · · · ) , x ∈ R.
1 2 3 4 5
68 3 Power Series
Beautiful Numerical Series

∞
√ n
3− 5
• n = 1;
2
n=1 n
∞ √
2 + k − k 2 + 4k 1
• n = , k > 0;
2 k
n=1
∞ √ n 1
• n 2− 3 = ;
2
n=1
∞ n 1
• n 1 + k − k 2 + 2k = , k > 0.
2k
n=1
3.5. The Power Series Expansion of f k

∞
If f : (−r, r) → R has the power series expansion f (x) = an x n and k ∈ N,
n=0

∞
then f k (x) = cn x n , where cn = ai1 ai2 · · · aik and the sum is over all k-tuples
n=0
of integers with i1 , i2 , . . . , ik ≥ 0 and i1 + i2 + · · · + ik = n.
(i1 , i2 , . . . , ik ),
Application. If k ∈ N, then
∞
1
= Cnn+k−1 x n , |x| < 1.
(1 − x)k
n=0
3.6. Calculate:
∞ ∞
16n2 + 4n − 1 16n2 + 12n + 1
(a) [134] ; (b) .
(4n + 2)! (4n + 3)!
n=0 n=0
3.7. [A. Sîntămărian, 2014] Calculate

∞
1
.
n(2n + 1)(4n + 1)
n=1
Next we record a proof, a gem of mathematical analysis, regarding the divergence

of the harmonic series which uses Abel’s theorem for power series.
The Harmonic Series Diverges

∞
1
If the harmonic series n would converge, then we have, based on Abel’s
n=1
Theorem for power series, that
∞ ∞
1 xn
= lim = − lim ln(1 − x) = ∞.
n x→1− n x→1−
n=1 n=1
3.8. Find the Maclaurin series expansion for the following functions and specify
the convergence set in each case:
(a) f : R → R, f (x) = sinh x; (b) f : R → R, f (x) = cos2 x;

(c) f : R → R, f (x) = sin3 x; (d) f : R → R, f (x) = x 2 e−3x ;
(e) f : R \ {1} → R, f (x) = 3x−1

(x−1)2
;
(f) f : (−3, 3) → R, f (x) = √1 ;
9−x 2
(g) f :
(−1, 1)→ R, f (x) = 1
2 ln 1+x
1−x ;
(h) f : − 15 , 1 → R, f (x) = ln(1 + 4x − 5x 2 );
√
(i) f : R → R, f (x) = ln(x + 1 + x 2 );
(j) f : R → R, f (x) = cos(3x) + x sin(3x);
2
(k) f : R \ {1} → R, f (x) = 1+x
1−x .
3.9. Prove that:

∞
1 1 · 3 · · · (2n − 1) n
(a) √ =1+ x , x ∈ [−1, 1);
1−x 2 · 4 · · · (2n)
n=1
∞
1 · 3 · · · (2n − 1) 1
(b) · = ln 4;
2 · 4 · · · (2n) n
n=1
∞
1 · 3 · · · (2n − 1) 1 2
(c) (−1)n · = 2 ln √ .
2 · 4 · · · (2n) n 1+ 2
n=1
3.10. Prove that

∞
1 2 2π(n + 1) n
=√ sin x , |x| < 1.
x +x+1
2
3 3
n=0
70 3 Power Series

∞
3.11. Let f be a function which has the power series expansion f (x) = an x n ,
n=0
|x| < R, R ∈ (0, ∞]. Prove that
∞
f (x) + f ( x) + f ( 2 x)
a3n x 3n = , |x| < R,
3
n=0
where = 1 is a root of order 3 of unity, i.e. 3 = 1.

Application. Prove that
∞
√
x 3n 1 x − x2 x 3
= e + 2e cos , x ∈ R.
(3n)! 3 2
n=0
3.12.
(a) Prove that

∞ ∞
(n2 + n − 1)3 (n2 + n − 1)3
= 12e and (−1)n = 0.
(n − 2)! + (n + 2)! (n − 2)! + (n + 2)!
n=2 n=2
(b) Prove that

∞
(n2 + n − 1)3
x n−2 = (x 2 + 6x + 5)ex , x ∈ R.
(n − 2)! + (n + 2)!
n=2
3.13. Series with the floor function. Prove that:

∞ * +
log2 n
(a) = 1;
n(n + 1)
n=1
∞
1 ln 2
(b) * += ;
n=2
n(n + 1) log2 n 2
∞
1 π2 ln2 2
(c) * +2 = − ;
n(n + 1) log2 n 24 4
n=2
∞
(−1)n
(d) * +k = 0, k ∈ N.
n=2 log2 n
3.3 Gems with Numerical and Power Series 71
Curiosities with Two Series

∞ * + ∞
log2 n 1
=1 and = 1.
n(n + 1) n(n + 1)
n=1 n=1
For Problems 3.14–3.21 write the power series expansion in terms of x − x0

and determine the convergence set for which the expansion holds.
3.14. f : R \ {−5} → R, f (x) = 1

x+5 , x0 = 2.
3.15. f : (−1, ∞) → R, f (x) = ln(x + 1), x0 = 3.
3.16. f : R \ {1} → R, f (x) = 1

(x−1)2
, x0 = −1.
3.17. f : R \ {−3, −1} → R, f (x) = 1

x 2 +4x+3
, x0 = −5.
3.18. f : R → R, f (x) = 1
x 2 −4x+8
, x0 = 2.
3.19. f : R → R, f (x) = e2x−1 , x0 = −1.
3.20. f : R → R, f (x) = sin(3x + π ), x0 = π .
3.21. f : R → R, f (x) = arctan x − arctan x 2 −x+1

1
, x0 = 1.
3.3 Gems with Numerical and Power Series
3.22. [22 October 2018] Power Series with Factorials
(a) Prove that

∞
xn 1 − ex
= + ex , x ∈ R.
(n − 1)! + n! x
n=1
(continued)
72 3 Power Series
In particular, for x = −1 we obtain the series

∞
(−1)n
= 2e−1 − 1.
(n − 1)! + n!
n=1
(b) Prove that

∞
x n−1 + x n 1 − ex + xex
= (1 + x) , x ∈ R.
(n − 1)! + n! x2
n=1
In particular, for x = 1 we obtain the series

∞
1
= 1.
(n − 1)! + n!
n=1
(c) Prove that

∞
n
x n = (x − 1)ex + 1, x ∈ R.
(n − 2)! + (n − 1)! + n!
n=2

∞
n
= 1.
(n − 2)! + (n − 1)! + n!
n=2
(d) Prove that

∞
xn x 1 − et
= ex − 1 + dt, x ∈ R.
(n − 2)! + (n − 1)! + n! 0 t
n=2

∞
1 1 1 − et
=e−1+ dt.
(n − 2)! + (n − 1)! + n! 0 t
n=2
(e) Prove that

∞
x n−2 + x n−1 + x n
(n − 2)! + (n − 1)! + n!
n=2
(continued)

1 + x + x2 x x 1 − et
= e −1+ dt , x ∈ R.
x2 0 t
3.23.
(a) Prove that

∞
1 1 1 1
(−1)n − + − ··· = .
n+1 n+2 n+3 2
n=0
(b) Prove that

∞
1+x
(−1) 2 x n =
n
, x ∈ (−1, 1),
1 + x2
n=0
where x is the floor of the real number x.
Proof. We give here the solution of this part of the problem. First, we observe that
the power series converges for x ∈ (−1, 1). We calculate the sum of the power
series. We have
∞ ∞
(−1) 2 x n = 1 + x + (−1) 2 x n
n n
n=0 n=2
∞ " #
m+2
=1+x+ (−1) 2
x m+2
m=0
∞
(−1) 2 x m ,
m
= 1 + x − x2
m=0
and the result follows. An alternative solution, based on calculating the 2nth partial
sum of the series, is left to the reader.
(c) Prove that

∞
1 1 1 π
(−1) 2
n
− + − ··· = .
n+1 n+2 n+3 4
n=0
74 3 Power Series
(d) Let k ∈ N. Prove that

∞
1 + x + · · · + x k−1
(−1) k x n =
n
, x ∈ (−1, 1).
1 + xk
n=0
(e) Let k ∈ N. Prove that

∞
1 1 1 1 1 + x + · · · + x k−1
(−1) k
n
− + − ··· = dx.
n+1 n+2 n+3 0 (1 + x)(1 + x k )
n=0
(f) Prove that

∞
x n+1 1
(−1) 2
n
= arctan x + ln(1 + x 2 ), x ∈ [−1, 1].
n+1 2
n=0
3.24. [19 December 2019] Prove that

∞
1 1 1 1 21
− + · · · ζ (2) − 1 − 2 − · · · − 2 = ζ (3)−ζ (2) ln 2.
n+1 n+2 2 n 16
n=1
3.25. [10 January 2020] Prove that:

An alternating series
∞
1 1 1 1 ln 2
(a) (−1)n−1 + − − + ··· = ;
n+1 n+2 n+3 n+4 2
n=1
Quadratic series
∞
1 1 1 1 π ln 2
(b) [91] (−1)n−1 Hn+ − − + ··· = ;
n+1 n+2 n + 3 n + 4 8
n=1
∞ 2
1 1 1 1 π ln 2
(c) + − − + ··· = +G+ ,
n n+1 n+2 n+3 4 2
n=1
where G denotes Catalan’s constant;

∞ 2
1 1 1 1 π2 π ln 2
(d) (−1) n−1
+ − − + ··· = + − .
n n+1 n+2 n+3 16 4 2
n=1
3.4 Single Zeta Series 75
3.26. [12 January 2020]

Prove that
∞
1 1 1 1 1 1 1
+ − − + ··· − + − ···
n n+1 n+2 n+3 n n+1 n+2
n=1
π ln 2
= + G,
8
where G denotes Catalan’s constant.
∞
1 1 1 1
(−1) n−1
+ − − + ···
n n+1 n+2 n+3
n=1

1 1 1
× − + − ···
n n+1 n+2
7π 2 ln2 2
= − .
96 8
3.4 Single Zeta Series
3.27. Prove that

∞ ∞
5
n (ζ (n + 1) − 1) = ζ (2) and (−1)n−1 n (ζ (n + 1) − 1) = ζ (2) − .
4
n=1 n=1
3.28. Prove that

∞ ∞
ζ (n) − 1 ζ (2n) − 1
=1−γ and = ln 2.
n n
n=2 n=1
3.29. Series Mirabili Prove that:

∞
n
(a) [72] 2 (ζ (n) − 1) − 1 = 3;
n=2
(continued)
76 3 Power Series
∞
1 11
(b) 3 ζ (n) − 1 − n − 1 =
n
;
2 2
n=2
∞ 0 1 25
(c) [84] 22n (ζ (2n) − 1) − 1 = ;
12
n=1
∞
1 3
(d) 32n
ζ (2n) − 1 − 2n − 1 = H6 ;
2 2
n=1
∞
1 1
(e) (k + 1) ζ (n) − 1 − n − · · · − n − 1 = (k + 1)Hk+1 , k ∈ N;
n
2 k
n=2
∞
1 1 k+1
(f) (k + 1)2n ζ (2n) − 1 − 2n − · · · − 2n − 1 = H2k+2 ,
2 k 2
n=1
k ∈ N.
3.30. Prove that:

∞ 0 1 11
(a) 22n−1 (ζ (2n − 1) − 1) − 1 = ;
12
n=2
∞ 0 1 337
(b) n 22n−1 (ζ (2n − 1) − 1) − 1 = .
144
n=2
3.31.
(a) Prove that

∞
7
(−1)n 2n (ζ (n) − 1) − 1 = .
6
n=2
(b) Let k ∈ N. Prove that

∞
1 1
(−1) (k + 1) ζ (n) − 1 − n − · · · − n − 1
n n
2 k
n=2
= (k + 1) (H2k+2 − Hk+1 ) .
3.32. [40]
(a) Prove that

∞

(n − 1) 2n (ζ (n) − 1) − 1 = 4ζ (2).
n=2
(b) Let k ∈ N. Prove that

∞
1 1
(n − 1) (k + 1) n
ζ (n) − 1 − n − · · · − n − 1 = (k + 1)2 ζ (2).
2 k
n=2
3.33. Prove that

∞
1 n
2 (ζ (n) − 1) − 1 = 3 − 2γ − ln 2
n
n=2
and
∞
1 n
4 (ζ (2n) − 1) − 1 = ln 6.
n
n=1
∞
3.34. Prove that Hn (ζ (n) − ζ (n + 1)) = ζ (2) − γ .
n=2
3.35. Prove that:

∞
ζ (2) ζ (3) ζ (n)
(a) [39] Hn − γ − − − ··· − = 2γ − 1;
2 3 n
n=2
∞
ζ (2) ζ (3) ζ (n) ζ (2)
(b) n Hn − γ − − − ··· − = − 1 + γ;
2 3 n 2
n=2
∞
ζ (2) ζ (3) ζ (n)
(c) (2n − 1) Hn − γ − − − ··· − = ζ (2) − 1.
2 3 n
n=2
78 3 Power Series
3.36. Prove that

∞ ∞
ζ (2n + 1) − 1 1 1
=− n ln 1 − 2 + = ln 2 − γ .
n+1 n n
n=1 n=2
3.37. Prove that

∞
nζ (2n)
= π 2.
4n−2
n=1
3.38. Series Delights Prove that:

∞
1
(a) = n − ζ (2) − ζ (3) − · · · − ζ (n), n ≥ 2;
k(k + 1)n
k=1
∞
(b) (n + 1 − ζ (2) − ζ (3) − · · · − ζ (n + 1)) = ζ (2) − 1;
n=1
∞
(c) [42] n (n + 1 − ζ (2) − ζ (3) − · · · − ζ (n + 1)) = ζ (3);
n=1
∞
1
(d) (−1)n−1 (n + 1 − ζ (2) − ζ (3) − · · · − ζ (n + 1)) = ;
4
n=1
∞
n
(e) 2 (n − ζ (2) − ζ (3) − · · · − ζ (n)) − 1 = 1;
n=2
∞

(f) n 2n (n − ζ (2) − ζ (3) − · · · − ζ (n)) − 1 = 4ζ (2) − 3.
n=2
The Dirichlet beta function is defined by

∞
(−1)n
β(x) = , x > 0.
(2n + 1)x
n=0
We mention the special values β(1) = π

4 and β(2) = G.
3.39. Series with the Dirichlet Beta Function

Prove that:
(continued)
3.5 Polylogarithm Series 79
∞
ln 2
(a) (β(n) − 1) = − ;
2
n=1
∞
1 − ln 2
(b) (−1)n (β(n) − 1) = ;
2
n=1
∞
ln 2 π 2
(c) n (β(n) − 1) = − − ;
2 48
n=1
∞
1 ln 2 π 2
(d) (−1)n n (β(n) − 1) =
− + ;
4 2 48
n=1
∞
1 5
(e) 5n β(n) − 1 + n − 1 = − ln 2;
3 2
n=1
∞
1 5 47
(f) (−1)n 5n β(n) − 1 + n − 1 = − ln 2 ;
3 2 60
n=1
∞
1 ln 2
(g) (β(2n) − 1) = − .
4 2
n=1
Remark 3.1. We obtain, based on parts (a) and (c), the remarkable series
formula
∞
−8 (n − 1)(β(n) − 1) = ζ (2).
n=2
3.40. Prove that

∞
π ln 2
(ζ (n) − β(n)) = + .
4 2
n=2
3.5 Polylogarithm Series
3.41. A Bouquet of Polylogarithm Series
(a) Prove that, for x ∈ [−1, 1], we have

∞
(Lin (x) − x) = (1 − x) ln(1 − x) + x.
n=2
(continued)
80 3 Power Series
In particular, for x = 1 we obtain the series of Problem 2.31

∞
(ζ (n) − 1) = 1.
n=2
(b) Prove that, for x ∈ [−1, 1], we have

∞
1−x x
(−1)n (Lin (x) − x) = ln(1 − x) − + 1.
x 2
n=2
In particular, for x = 1 we obtain the first series of Problem 2.32

∞
1
(−1)n (ζ (n) − 1) = .
2
n=2
(c) Prove that, for x ∈ [−1, 1], we have

∞
1 − x2 x 1
(Li2n (x) − x) = ln(1 − x) + + .
2x 4 2
n=1
In particular, for x = 1 we obtain the second series of Problem 2.32

∞
3
(ζ (2n) − 1) = .
4
n=1
(d) Prove that, for x ∈ [−1, 1], we have

∞
(1 − x)2 3x 1
(Li2n+1 (x) − x) = − ln(1 − x) + − .
2x 4 2
n=1
In particular, for x = 1 we have

∞
1
(ζ (2n + 1) − 1) = .
4
n=1
(e) Prove that, for x ∈ [−1, 1], we have

∞
1−x x
(Li2n (x) − Li2n+1 (x)) = ln(1 − x) − + 1.
x 2
n=1
(continued)

∞
1
(ζ (2n) − ζ (2n + 1)) = .
2
n=1
(f) Prove that, for x ∈ [−1, 1], we have

∞
1−x
(−1)n (Lin (x) − Lin+1 (x)) = 2 ln(1 − x) − Li2 (x) + 2.
x
n=2
In particular, for x = 1 we obtain the series of Problem 2.33

∞
(−1)n (ζ (n) − ζ (n + 1)) = 2 − ζ (2).
n=1
(g) Prove that, for x ∈ [−1, 1], we have

∞
n (Lin (x) − x) = (1 − x) ln(1 − x) + x Li2 (x) + x.
n=2

∞
n(ζ (n) − 1) = ζ (2) + 1.
n=2
(h) Prove that, for x ∈ [−1, 1], we have

∞
n (Lin (x) − Lin+1 (x)) = (1 − x) ln(1 − x) + Li2 (x).
n=2

∞
n(ζ (n) − ζ (n + 1)) = ζ (2).
n=2
(i) Prove that, for x ∈ [−1, 1], we have

∞ 0 1
x2
2 (Lin (x) − x) − x = 2 (1 − x 2 ) ln(1 − x) + x +
n 2
.
2
n=2
(continued)
82 3 Power Series
In particular, for x = 1 we obtain part (a) of Problem 3.29

∞
n
2 (ζ (n) − 1) − 1 = 3.
n=2
(j) Prove that, for x ∈ [−1, 1], we have

∞ 0 1 1 − x4
1 x x2 x3
4 (Li2n (x) − x) − x =
n 2
ln(1−x)+ 1+ + + .
x2 x 2 3 4
n=1
In particular, for x = 1 we obtain part (c) of Problem 3.29

∞
n 25
4 (ζ (2n) − 1) − 1 = .
12
n=1
(k) Prove that, for x ∈ [−1, 1], we have

∞
x2
3n Lin (x) − x − n − x 3
2
n=2

x2 x3
= 3 (1 − x 3 ) ln(1 − x) + x + + .
2 3
In particular, for x = 1 we obtain part (b) of Problem 3.29

∞
1 11
3 ζ (n) − 1 − n − 1 =
n
.
2 2
n=2
(l) A generalization of parts (i) and (k).

Let k ∈ N. Prove that, for x ∈ [−1, 1], we have
∞
x2 xk
(k + 1) n
Lin (x) − x − n − · · · − n − x k+1
2 k
n=2
x2 x k+1
= (k + 1) 1 − x k+1 ln(1 − x) + x + + ··· + .
2 k+1
In particular, for x = 1 we obtain part (e) of Problem 3.29
(continued)
∞
1 1
(k + 1) n
ζ (n) − 1 − n − · · · − n − 1 = (k + 1)Hk+1 .
2 k
n=2
(m) Let k ∈ N. Prove that, for x ∈ [−1, 1], we have

∞
x2 xk
(n − 1) (k + 1)n Lin (x) − x − n − · · · − n − x k+1
2 k
n=2
= (k + 1)2 x k+1 Li2 (x).

∞
1 1
(n−1) (k + 1)n ζ (n) − 1 − n − · · · − n − 1 = (k+1)2 ζ (2).
2 k
n=2
Remark 3.2. From parts (a) and (g) we have, for x ∈ [−1, 1], that
∞
(n − 1) (Lin (x) − x) = x Li2 (x).
n=2
Multiple Polylogarithm Series
(n) Prove that, for x ∈ [−1, 1], we have

∞ ∞
(Lin+m (x) − x) = x Li2 (x).
n=1 m=1

∞ ∞
(ζ (n + m) − 1) = ζ (2).
n=1 m=1
(o) Prove that, for x ∈ [−1, 1], we have

∞ ∞
n (Lin+m (x) − x) = x (Li2 (x) + Li3 (x)) .
n=1 m=1
(continued)
84 3 Power Series

∞ ∞
n (ζ (n + m) − 1) = ζ (2) + ζ (3).
n=1 m=1
(p) Prove that, for x ∈ [−1, 1], we have

∞ ∞
nm (Lin+m (x) − x) = x (Li2 (x) + 2Li3 (x) + Li4 (x)) .
n=1 m=1

∞ ∞
nm (ζ (n + m) − 1) = ζ (2) + 2ζ (3) + ζ (4).
n=1 m=1
(q) Let k ∈ N, k ≥ 2. Prove that, for x ∈ [−1, 1], we have

∞ ∞

··· Lin1 +···+nk (x) − x = x Lik (x).
n1 =1 nk =1

∞ ∞
··· (ζ (n1 + · · · + nk ) − 1) = ζ (k).
n1 =1 nk =1
(r) Prove that, for x ∈ [−1, 1], we have

∞ ∞
(Li2n+m (x) − Lin+2m (x)) = 0.
n=1 m=1

∞ ∞
(ζ (2n + m) − ζ (n + 2m)) = 0.
n=1 m=1
(continued)
(s) Let k ∈ N and 1 ≤ i ≤ k. Prove that, for x ∈ [−1, 1], we have
∞ ∞ i
j
··· n1 · · · ni Lin1 +···+nk (x) − x = x Ci Lii+k−j (x).
n1 =1 nk =1 j =0
In particular, for x = 1 and i = k we obtain part (b) of Problem 3.100
∞ ∞ k
j
··· n1 · · · nk (ζ (n1 + · · · + nk ) − 1) = Ck ζ (2k − j ).
n1 =1 nk =1 j =0
3.42. Polylogarithm Delights

Let x ∈ [−1, 1]. Prove that:
(a) A remarkable power series

∞
xi
i n (i − 1)
i=2
= nx + (1 − x) ln(1 − x) − Li2 (x) − Li3 (x) − · · · − Lin (x), n ≥ 2;
(b) polylogarithm series I

∞
(Li2 (x) + Li3 (x) + · · · + Lin+1 (x) − (n + 1)x − (1 − x) ln(1 − x))
n=1
= (1 − x) ln(1 − x) − xLi2 (x) + x;
(c) polylogarithm series II

∞
n (Li2 (x) + Li3 (x) + · · · + Lin+1 (x) − (n + 1)x − (1 − x) ln(1 − x))
n=1
= −xLi3 (x);
(continued)
86 3 Power Series
(d) polylogarithm series III

∞
(−1)n−1
n=1
× (Li2 (x) + Li3 (x) + · · · + Lin+1 (x) − (n + 1)x − (1 − x) ln(1 − x))

(1 − x)2 1 3x
= ln(1 − x) + − ;
2x 2 4
(e) polylogarithm series IV
∞ 0 1
2n (Li2 (x) + Li3 (x) + · · · + Lin (x) − nx − (1 − x) ln(1 − x)) + x 2
n=2
= 2(1 − x)2 ln(1 − x) − 3x 2 + 2x;
(f) polylogarithm series V

∞ 0 1
n 2n (Li2 (x) + Li3 (x) + · · · + Lin (x) − nx − (1 − x) ln(1 − x))+ x 2
n=2
= 2(1 − x 2 ) ln(1 − x) − 4x 2 Li2 (x) + x 2 + 2x;
(g) polylogarithm series VI

Let k ∈ N.
∞ ∞
···
n1 =1 nk =1

Li2 (x) + · · · + Lin1 +···+nk (x) − (n1 + · · · + nk )x − (1 − x) ln(1 − x)
= −x Lik+1 (x).
Remark 3.3. The first five parts generalize Problem 3.38 to the case of
polylogarithm functions. When x = 1 we obtain Problem 3.38.
3.43. Alternating Series with Polylogarithm Functions

Let x ∈ [−1, 1] \ {0} and n ∈ N, n ≥ 2. Prove that:
(a) A remarkable power series

∞
xi
+ 1)
i n (i
i=2

1−x
= (−1) Li2 (x) − Li3 (x) + · · · + (−1) Lin (x) −
n n
ln(1 − x) − 1
x
x
− ;
2
(b) A limit with polylogarithm functions

1−x x
lim (−1) Li2 (x)+ · · · +(−1) Lin (x)−
n n
ln(1−x)−1 −
n→∞ x 2
= 0;
In particular, for x = 1 we obtain the limit

1
lim (−1)n ζ (2) − ζ (3) + · · · + (−1)n ζ (n) − 1 − = 0;
n→∞ 2
(c) Alternating series I

∞
1−x x
(−1)n n
Li2 (x) + · · · + (−1) Lin (x) − ln(1 − x) − 1 −
x 2
n=2
(1 − x)2 3x 1
=− ln(1 − x) + − ;
2x 4 2

∞
1 1
(−1)n ζ (2) − ζ (3) + · · · + (−1)n ζ (n) − 1 − = ;
2 4
n=2
(continued)
88 3 Power Series
(d) Alternating series II

∞
1−x x
(−1)n (−1)n Li2 (x) + · · · + (−1)n Lin (x) − ln(1 − x) − 1 −
x 2
n=2
1−x Li2 (x) x

= ln(1 − x) − − + 2;
x x 4

∞
1 7
(−1) (−1)n ζ (2) − ζ (3) + · · · + (−1)n ζ (n) − 1 −
n
= −ζ (2).
2 4
n=2
3.6 Inequalities and Integrals
3.44. [115] Prove that x x ≤ x 2 − x + 1, for all 0 < x ≤ 1.
ex +e−x x2
3.45. Prove that 2 ≤ e 2 , for all x ∈ R.
3.46. Young’s Inequality ((a) and (c))
(a) Let p, q > 1, with 1

p + 1
q = 1, and x, y ≥ 0. Prove that
xp yq
+ ≥ xy.
p q
(b) [111] Let p, q > 1, with 1

p + 1
q = 1, and x, y ∈ (0, 1). Prove that
q p pq
+ ≥ .
1 − xp 1 − yq 1 − xy
(c) Let p, q, r > 1, with 1

p + 1
q + 1
r = 1, and x, y, t ≥ 0. Prove that
xp yq tr
+ + ≥ xyt.
p q r
(d) Let p, q, r > 1, with 1

p + 1
q + 1
r = 1, and x, y, t ∈ (0, 1). Prove that
qr pr pq pqr
+ + ≥ .
1−x p 1−y q 1−t r 1 − xyt
3.6 Inequalities and Integrals 89
3.47. Nesbitt’s Inequality
(a) If a, b, c are positive real numbers, then
a b c 3
+ + ≥ .
b+c a+c a+b 2
(b) If n ∈ N, n ≥ 2, and x1 , x2 , . . . , xn are positive real numbers, then

x1 x2 xn
+ + ··· +
x2 + x3 + · · · + xn x1 + x3 + · · · + xn x1 + x2 + · · · + xn−1
n
≥ .
n−1
3.48. Circular Inequalities
(a) [111] Prove that, for a, b, c ∈ (−1, 1), the following inequality holds:
1 1 1 1 1 1
+ + ≥ + + .
1 − a2 1 − b2 1 − c2 1 − ab 1 − bc 1 − ac
(b) Prove that, for a, b, c, d ∈ (0, 1), the following inequality holds:
1 1 1 1 1 1 1 1
+ + + ≥ + + + .
1 − a 3 1 − b3 1 − c3 1 − d 3 1 − abc 1 − bcd 1 − cda 1 − dab
3.49. Let n ∈ N and

∞ 1
In = dx.
0 (1 + x 2 )n
Prove that:
∞
In
(a) = π;
n
n=1
∞
(2n − 3)!! 1
(b) = .
(2n)!! 2
n=2
90 3 Power Series
3.50. [O. Furdui, Problem 4, SEEMOUS 2016 (parts (a) and (b))]
Let n ∈ N and
∞ arctan x
In = dx.
0 (1 + x 2 )n
Prove that:
∞
In
(a) = ζ (2);
n
n=1
∞
1
(b) arctan x ln 1 + 2 dx = ζ (2);
0 x
∞ 1
(c) arccot x ln 1 + 2 dx = 2ζ (2).
0 x
3.51. Let a > 1. Prove that:

∞
(−1)n π
(a) = π ;
n=−∞
an + 1 a sin a
∞
1 1 1 + x a−2
(b) =− ln x dx.
n=−∞
(an + 1)2 0 1 − xa
When a = 2 one obtains sophomore’s dream for series [9, problem 11, part
a), p. 381]
∞
2 ∞
(−1)n 1
= .
n=−∞
2n + 1 n=−∞
(2n + 1)2
3.52. Sophomore’s dream. Using the power series expansion of the exponential
function, prove that
∞ 1 ∞ 1
1 1 1
= dx and (−1)n−1 = x x dx.
nn 0 xx nn 0
n=1 n=1
Remark 3.4. Formula

∞ 1
1 1
= dx
nn 0 xx
n=1
is known in the mathematical literature as sophomore’s dream because of the

likeness of the terms in the two members of the formula [112, p. 189]. The equality
is too good to be true, however, it is true.
The second integral formula
∞ 1
1
(−1)n−1 = x x dx
nn 0
n=1
was discovered by the Swiss mathematician John Bernoulli (1667–1748) in 1697,

being so fascinated by this result that he called it “series mirabili”, which means
marvelous series [112, p. xvi].
3.53.
(a) An exponential integral with the floor function.

Let a ∈ (0, 1). Calculate
" #
1 1
x
a dx,
0

(b) A particular case. Prove that
" #
1
− x1
2 dx = 1 − ln 2.
0
3.54. Curiosities with integrals. Prove that

∞ ∞
2−x dx = 2 and x 2−x dx = 3,
0 0
3.55. Prove that

∞ 1
x e−x dx = ,
0 e−1

92 3 Power Series
3.56. Prove that

1 ln(1 − x) 1 ln(1 − x) 2
dx = −ζ (2) and dx = 2ζ (2).
0 x 0 x
3.57. Prove that

1 ln(1 + x) π2 1 ln(1 + x) 2 π2
dx = and dx = − 2 ln2 2.
0 x 12 0 x 6
3.58.
(a) [139] Let n ≥ 2 be an integer. Prove that
1 ln(1 + x + x 2 + · · · + x n−1 ) π 2 (n − 1)
dx = .
0 x 6n
(b) Let n ≥ 2 and k ≥ 0 be integers. Prove that

1 lnk x ln(1 + x + x 2 + · · · + x n−1 ) 1
dx = (−1) k! 1 − k+1 ζ (k + 2).
k
0 x n
3.59. Calculate
1 ln(1 − x + x 2 )
dx.
0 x
3.60.
(a) Let n ≥ 3 be an odd integer. Prove that

2
1 ln(1 − x + x 2 − x 3 + · · · + x n−1 ) 1 π
dx = −1 .
0 x n 12
(b) Let n ≥ 3 be an odd integer and k ≥ 0 be an integer. Prove that
lnk x ln(1 − x + x 2 − x 3 + · · · + x n−1 )

1
dx
0 x

1 1
= (−1)k k! 1 − k+1 − 1 ζ (k + 2).
2 nk+1
3.7 Generating Functions 93
3.61. A surprising series with binomial coefficients.
(a) [49] Prove that

1 ln(1 − x + x 2 ) π2
dx = − .
0 x−x 2 9
(b) Deduce the surprising series giving ζ (2)

∞
1
3 = ζ (2).
n2 Cn2n
n=1
3.7 Generating Functions
The Cauchy Product of Two Power Series

∞
∞
The Cauchy product of the power series an x n and bn x n , with an , bn ∈
n=0 n=0
R, is the power series
∞ n
cn x n , where cn = ak bn−k .
n=0 k=0
3.62. Write the first three nonzero terms of the power series expansion in x of the
following functions:
(a) f : R → R, f (x) = e−x sin x; (b) f : R → R, f (x) = e3x cos x;

(c) f : − π2 , π2 → R, f (x) = tan x; (d) f : R → R, f (x) = tanh(2x).
3.63. The generating function of the nth harmonic number Hn .
(a) Prove that

∞
ln(1 − x)
− = Hn x n , x ∈ (−1, 1).
1−x
n=1
(b) Prove, by differentiating the equality in part (a), that

∞
1 − ln(1 − x)
= nHn x n−1 , x ∈ (−1, 1).
(1 − x)2
n=1
94 3 Power Series
(c) Prove, by integrating the equality in part (a), that

∞
Hn n+1
ln2 (1 − x) = 2 x , x ∈ [−1, 1).
n+1
n=1
(d) Prove the identity

n
1 1 Hn−1
· =2 , n ≥ 2.
k−1 n+1−k n
k=2
3.64.

(a) Computing −Li2 1
2 . Prove that
1
2 ln(1 − x) ln2 2 π 2
dx = − .
0 x 2 12
(b) Two series with the harmonic number. Prove that

∞ ∞
Hn Hn
= ζ (2) and = ln2 2.
n2n−1 (n + 1)2n
n=1 n=1
3.65. The generating function of the harmonic number On .
(a) Prove that

∞ 2n
1 2 1+x 1 1 x
ln = 1 + + ··· + , x ∈ (−1, 1).
4 1−x 3 2n − 1 n
n=1
(b) Prove that

∞
x 1+x 1 1
ln = 1 + + ··· + x 2n , x ∈ (−1, 1).
2(1 − x ) 1 − x
2 3 2n − 1
n=1
(c) Prove that

∞
On 7
2
= ζ (3).
n 4
n=1
3.66. The generating function of the skew-harmonic number Hn− .

Let n ∈ N. The skew-harmonic number Hn− is defined by
1 1 (−1)n−1
Hn− = 1 − + − ··· + .
2 3 n
(a) Prove that
1 1 − (−t)n
Hn− = dt, n ≥ 1.
0 1+t
(b) Prove that

∞
ln(1 + x)
Hn− x n = , x ∈ (−1, 1).
1−x
n=1
3.67. The generating function of the sequence (En )n≥0 .

Find the Maclaurin series expansion of the function f : R \ {1} → R,
ex
f (x) = 1−x , and mention the set on which the expansion holds.
3.68. The generating function of Bernoulli numbers.

The Bernoulli numbers, Bn , n ≥ 0, are defined by the equality
∞
x Bn n
= x .
e −1
x n!
n=0
Determine the first five nonzero Bernoulli numbers.
Remark 3.5. The Bernoulli numbers satisfy the recurrence relation
C1n+1 Bn + C2n+1 Bn−1 + · · · + Ckn+1 Bn−k+1 + · · · + Cnn+1 B1 + 1 = 0, n ≥ 1,
which can be written in the following form:
(B + 1)n+1 − B n+1 = 0, n ≥ 1,
where B k is a notation for Bk .

All Bernoulli numbers of odd index, except for B1 , are equal to 0.
3.69. The generating function

√of the
√
Fibonacci
sequence.
Knowing that, for x ∈ 1− 5
2 , 2
5−1
, one has the power series expansion
96 3 Power Series
∞
1
= an x n−1 ,
1 − x − x2
n=1
prove that (an )n≥1 equals the Fibonacci sequence.
The Lucas sequence (Ln )n≥0 is defined by the recurrence relation
L0 = 2, L1 = 1, Ln+2 = Ln+1 + Ln , ∀n ≥ 0.
3.70. The generating function of the Lucas sequence.

√ √
(a) Prove that, for x ∈ 1− 5
2 , 5−1
2 , the following equality holds
∞
x−2
= Ln x n .
x +x−1
2
n=0
√ √
(b) Prove that, for x ∈ 1− 5 5−1
2 , 2 , the following equality holds
∞
1 Ln n
ln = x .
1 − x − x2 n
n=1
3.71.
(a) The generating function of the sequence (sin n)n≥1 .

Prove that, for x ∈ (−1, 1), the following equality holds
∞
x sin 1
(sin n)x n = .
1 − 2x cos 1 + x 2
n=1

sin n
(b) The generating function of the sequence n .
n≥1
Prove that, for x ∈ [−1, 1], the following equality holds
∞
sin n n x − cos 1 π
x = arctan + − 1.
n sin 1 2
n=1
(c) Two special series with sine.

Prove that
∞ ∞
sin n π −1 sin n 1
= and (−1)n−1 = .
n 2 n 2
n=1 n=1
3.72.
(a) The generating function of the sequence (cos n)n≥1 .

Prove that, for x ∈ (−1, 1), the following equality holds
∞
x cos 1 − x 2
(cos n)x n = .
1 − 2x cos 1 + x 2
n=1

(b) The generating function of the sequence cosn n n≥1 .
Prove that, for x ∈ [−1, 1], the following equality holds
∞
cos n n 1
x = − ln(1 − 2x cos 1 + x 2 ).
n 2
n=1
(c) Two special series with cosine.

Prove that
∞ ∞
cos n 1 cos n 1
= − ln 2 sin and (−1)n−1 = ln 2 cos .
n 2 n 2
n=1 n=1
3.73. Let (an )n≥0 be the sequence defined by
∞
1
= an x n , x ∈ (−1, 1).
x − 2x + 1
2
n=0
Prove that the numbers m ∈ N, with the property that there exists n ∈ N ∪ {0}
such that
an an+1 an+2 an+3 = am−2 ,
are perfect squares.
3.74. Let (an )n≥0 and (bn )n≥0 be the sequences defined by
∞ ∞
sinh x = an x n , cosh x = bn x n , x ∈ R.
n=0 n=0
98 3 Power Series
1 ,2n
Eliminate a factor of the form a2k−1 or b12k from the product 1
k=1 a2k−1 b2k such
that the remaining product be a perfect square.
3.8 Series with Harmonic and Skew-Harmonic Numbers
3.75. Skew-harmonic numbers and tails of zeta function.

Prove that:
∞
ln 2
(a) Hn− Hn− − ln 2 = ;
2
n=1
∞
− 1 1 π2 π 2 ln 2 ln2 2
(b) Hn − ln 2 ζ (2) − 1 − 2 − · · · − 2 = − + + ;
2 n 8 6 2
n=1
∞
n − 1 1 π 2 ln 2 21
(c) (−1) Hn ζ (2) − 1 − 2 − · · · − 2 = − ζ (3);
2 n 8 16
n=1
∞ 2
1 1 π ln 2 9
(d) Hn− ζ (3) − 1 − 3 − · · · − 3 = − ζ (3).
2 n 4 8
n=1
3.76. A mosaic of series. Prove that:

∞
Hn− π2 ln2 2
(a) (−1)n =− − ;
n 12 2
n=1
∞
Hn− π2 ζ (3)
(b) 2
= ln 2 − ;
n 4 4
n=1
∞
Hn− 7 π4
(c) 3
= ζ (3) ln 2 − ;
n 4 288
n=1
∞
Hn− Hn+1 5 π2
(d) · = ζ (3) + ;
n n+1 8 12
n=1
∞
Hn− Hn+1 3 π2 2
(e) (−1)n · = ζ (3) − ln 2 − ln2 2 + ln3 2.
n n+1 2 3 3
n=1
3.8 Series with Harmonic and Skew-Harmonic Numbers 99
3.77.
(a) Prove that

∞ −
n−1 Hn Hn+1 1 π2 ln3 2 7
(−1) = 2 Li3 =− ln 2 + + ζ (3).
n+1 2 6 3 4
n=1
(b) Prove that

∞
Hn− Hn ln3 2 π 2 ζ (3)
(−1)n = − ln 2 − .
n 3 6 4
n=1
3.78. Prove that:

∞
Hn − π 2 ln 2 ζ (3) ln3 2
(a) Hn − ln 2 = − − ;
n 12 4 6
n=1
∞
Hn − π 2 ln 2 ζ (3) ln3 2
(b) (−1)n Hn − ln 2 = − − − ;
n 12 4 6
n=1
∞
H2n − ζ (3) ln3 2
(c) H2n − ln 2 = − − ;
n 2 3
n=1
∞
H2n−1 − ζ (2) ln 2
(d) H2n−1 − ln 2 = .
2n − 1 2
n=1
3.79. Euler’s related series. Prove that

∞ ∞
H2n−1 21 H2n 11
= ζ (3) and = ζ (3).
(2n − 1)2 16 n2 4
n=1 n=1
3.80. Gems with skew-harmonic numbers.
(a) Let n ∈ N. Prove that

(continued)
100 3 Power Series
1 1 − (−1)n (−1)n −
x n−1 ln(1 + x)dx = ln 2 + Hn .
0 n n
(b) Prove that
∞ −
H2n−1 3π ln 2
(−1)n−1 = .
2n − 1 8
n=1
(c) Prove that

∞ ∞
Hn− π2 Hn−
= and (−1)n−1 = ln2 2.
n(n + 1) 12 n(n + 1)
n=1 n=1
(d) A nonlinear skew-harmonic sum I.

Prove that
∞ −
Hn− Hn+1 π 2 ln 2 5
· = − ln2 2 − ζ (3).
n n+1 4 8
n=1
(e) A nonlinear skew-harmonic sum II.

Prove that
∞
Hn Hn− 43π 4 π 2 ln2 2 ln4 2 1
= + − − 3Li4 .
n2 1440 8 8 2
n=1
3.81.
(a) Prove that

∞
(−1)n 1 1 π2
ζ (2) − 1 − 2 − · · · − 2 = ζ (3) − ln 2.
n 2 n 4
n=1
(b) Prove that

∞
(−1)n 1 1 19π 4 7
ζ (3) − 1 − 3 − · · · − 3 = − ζ (3) ln 2.
n 2 n 1440 4
n=1
(continued)
3.9 Remarkable Numerical and Function Series 101
(c) Prove that

∞
Hn− 1 1
ζ (2) − 1 − 2
− ··· − 2
n 2 n
n=1

53π 4 π 2 ln2 2 ln4 2 21 ln 2 1
= + − − ζ (3) − 3 Li4 .
1440 4 8 8 2
Open problem. Calculate

∞
Hn− 1 1
ζ (3) − 1 − − ··· − 3 .
n 23 n
n=1
3.9 Remarkable Numerical and Function Series
3.82. Let
x3 x6
u(x) = 1 + + + ··· ,
3! 6!
x4 x7
v(x) = x + + + ··· ,
4! 7!
x2 x5 x8
w(x) = + + + ··· .
2! 5! 8!
Prove that
u2 (x) + v 2 (x) + w 2 (x) − u(x)v(x) − v(x)w(x) − w(x)u(x) = e−x
and
u3 (x) + v 3 (x) + w 3 (x) − 3u(x)v(x)w(x) = 1.
3.83.
(a) Prove that
1 x2 xn x tn
ln −x− − ··· − = dt, x ∈ [−1, 1).
1−x 2 n 0 1−t
102 3 Power Series
(b) Prove that

∞
1 1 x2 xn ln2 (1 − x)
ln −x− − ··· − = , x ∈ [−1, 1).
n 1−x 2 n 2
n=1
3.84. Calculate:
∞
1 x2 xn
(a) ln −x− − ··· − , x ∈ [−1, 1);
1−x 2 n
n=1
∞ 2n+1
x3 n+1 x
(b) arctan x − x + + · · · + (−1) , x ∈ [−1, 1].
3 2n + 1
n=0
3.85. Prove that:
(a) H1 + H2 + · · · + Hn = (n + 1)Hn − n, ∀ n ≥ 1;
(b) harmonic numbers and the tail of − ln(1 − x).
∞
1 x2 xn
Hn ln −x− − ··· −
1−x 2 n
n=1
ln(1 − x) x
=− − , x ∈ [−1, 1);
1−x 1−x
(c) harmonic numbers and the tail of Li2 (x).

∞
x2 xn
Hn Li2 (x) − x − 2 − · · · − 2
2 n
n=1
ln2 (1 − x)
= Li2 (x) + + ln(1 − x), x ∈ [−1, 1);
2
(d) harmonic numbers and the tail of Li3 (x).
∞
x2 xn
Hn Li3 (x) − x − 3 − · · · − 3
2 n
n=1
1 x ln2 (1 − t)
= Li3 (x) − Li2 (x) + dt, x ∈ [−1, 1];
2 0 t
In particular, when x = 1 we obtain the series of Problem 2.56
(continued)
∞
1 1
Hn ζ (3) − 1 − 3 − · · · − 3 = 2ζ (3) − ζ (2).
2 n
n=1
(e) harmonic numbers and the tail of Li4 (x).

∞
x2 xn
Hn Li4 (x) − x − 4 − · · · − 4
2 n
n=1
ln x x ln2 (1 − t)
= Li4 (x) − Li3 (x) + dt
2 0 t
1 x ln t ln (1 − t)
2
− dt, x ∈ [−1, 1];
2 0 t
In particular, when x = 1 we obtain the first series in Remark 2.6

∞
1 1 5
Hn ζ (4) − 1 − 4 − · · · − 4 = ζ (4) − ζ (3).
2 n 4
n=1
3.86.
(a) Prove that, for x ∈ (−1, 1], we have

∞
1 1 1 x
− + − ··· xn = (Li2 (x) − Li2 (−1)) .
n2 (n + 1)2 (n + 2)2 1+x
n=1
For x = −1 we obtain [O. Furdui, A. Sîntămărian, Problem 4, part (b),

SEEMOUS 2019]
∞
1 1 1
(−1)n−1 − + − ··· = ln 2
n2 (n + 1)2 (n + 2)2
n=1
and, for x = 1 we have the series

∞
1 1 1 π2
− + − ··· = .
n2 (n + 1)2 (n + 2)2 8
n=1
(continued)
104 3 Power Series
(b) Prove that

∞
1 1 1 1
− + − ··· xn
n n2 (n + 1)2 (n + 2)2
n=1
x ln(1 − t) ln(1 + t)
= ln(1 + x) (Li2 (x) − Li2 (−1)) + dt, x ∈ [−1, 1].
0 t
In particular, for x = −1 we have

∞
(−1)n−1 1 1 1 5
− + − ··· = ζ (3)
n n2 (n + 1)2 (n + 2)2 8
n=1
and, for x = 1 we obtain the series

∞
1 1 1 1 π 2 ln 2 5
− + − ··· = − ζ (3).
n n2 (n + 1)2 (n + 2)2 4 8
n=1
3.87. [5 April 2019] Remarkable Series of Functions
(a) Prove that

∞
1 x2 xn 1 1
ln −x− − ··· − ζ (2) − 1 − 2 − · · · − 2
1−x 2 n 2 n
n=1
x ln2 (1 − x)
= (ζ (2) − Li2 (x)) + ζ (2) ln(1 − x) + , −1 ≤ x < 1.
1−x 2
In particular, for x = −1 we have

∞
1 (−1)2 (−1)n 1 1
ln − (−1) − − ··· − ζ (2) − 1 − 2 − · · · − 2
2 2 n 2 n
n=1
π2 π2 ln2 2
=− + ln 2 + ,
8 6 2
and, for x = 1
2 we obtain the series
(continued)
∞
1 1 1 1 1
ln 2 − − 2 − · · · − n ζ (2) − 1 − 2 − · · · − 2
2 2 2 2 n 2 n
n=1
π2 π2
= + ln2 2 − ln 2.
12 6
(b) Prove that
∞
x2 xn 1 1
Li2 (x) − x − 2 − · · · − 2 ζ (2) − 1 − 2 − · · · − 2
2 n 2 n
n=1
= − ln(1 − x)(ζ (2) − Li2 (x)) − ζ (2)Li2 (x)

3 x ln2 (1 − t)
+ dt, −1 ≤ x ≤ 1.
2 0 t
In particular, for x = 1 we obtain the remarkable quadratic series [26,

problem 3.22, p. 142]
∞
1 1 2 5
ζ (2) − 1 − 2 − · · · − 2 = 3ζ (3) − ζ (4).
2 n 2
n=1
3.88. [71] A Logarithmic Power Series
(a) Prove that
1 (−1)2 (−1)n 1 xn
ln − (−1) − − ··· − = (−1)n−1 dx.
2 2 n 0 1+x
(b) Prove that the convergence set of the power series
∞
1 (−1)2 (−1)n
ln − (−1) − − ··· − xn
2 2 n
n=1
is (−1, 1].
(continued)
106 3 Power Series
(c) Prove that

∞
1 (−1)2 (−1)n
ln − (−1) − − ··· − xn
2 2 n
n=1
⎧
⎪
⎪ ln 2 − 12 if x = 1
⎨
=
⎪
⎪ ln(1 + x) − x ln 2
⎩ if x ∈ (−1, 1).
1−x

∞
∞
3.89. Let f (x) = an x n , |x| < 1. Prove that if both series an and
n=1 n=1

∞
(f (1) − a1 − a2 − · · · − an ) converge, then
n=1
⎧
∞ ⎨ f (1)x − f (x) if x = 1
(f (1) − a1 − a2 − · · · − an )x n = 1−x
⎩
n=1 f (1) − f (1) if x = 1.
Applications. Problems 3.4 and 3.88.
3.90. Marvelous Numerical and Power Series

(a) Prove that
1 2 2 2 1 1−x
− + − + ··· = x n−1 dx, n ≥ 1.
n n+1 n+2 n+3 0 1+x
(b) [88] Prove that

∞
1 2 2 2
xn − + − + ···
n n+1 n+2 n+3
n=1
⎧
⎪
⎪ 2x ln 2 + (1 − x) ln(1 − x)
⎪
⎨ if x ∈ (−1, 1)
1+x
= ln 2 − 1 x = −1
⎪
⎪
if
⎪
⎩ln 2 if x = 1.
(continued)
We mention the interesting formula

∞
1 2 2 2
− + − + ··· = ln 2.
n n+1 n+2 n+3
n=1
(c) Prove that

∞ 2
1 2 2 2
− + − + ··· = 4 ln 2 − 2 ln2 2 − ζ (2).
n n+1 n+2 n+3
n=1
(d) Prove that

∞ 2
1 2 2 2 π2
(−1)n − + − + ··· = − 2 ln2 2.
n n+1 n+2 n+3 12
n=1
Open problem. Calculate:

∞ 3
1 2 2 2
• − + − + ··· ;
n n+1 n+2 n+3
n=1
∞ 3
1 2 2 2
• (−1)n − + − + ··· .
n n+1 n+2 n+3
n=1
A Jewel with a Quadratic Series

We prove that
∞ 2
1 1 1
− + − ··· = ln 2.
n n+1 n+2
n=1
We apply in our proof Abel’s summation formula with an = 1 and bn =

2
1
n − n+1
1
+ n+21
− · · · . We have

1 1 2 2 2
bn − bn+1 = − + − + ··· .
n n n+1 n+2 n+3
Therefore,
(continued)
108 3 Power Series
∞ 2
1 1 1
− + − ···
n n+1 n+2
n=1
2
1 1 1
= lim n − + − ···
n→∞ n+1 n+2 n+3
∞
1 2 2 2
+ − + − + ···
n n+1 n+2 n+3
n=1
∞
1 2 2 2
= − + − + ···
n n+1 n+2 n+3
n=1
3.90 (b)
= ln 2.
This series was calculated using a different method in [26, problem 3.29].
Another Jewel with a Quadratic Series

We prove that
∞ 2
1 1 1 π 2 ln 2 9
− + − ··· = − ζ (3). (3.1)
n2 (n + 1)2 (n + 2)2 2 4
n=1
We apply in our proof Abel’s summation formula with an = 1 and bn =

2
1
n2
− (n+1)
1
2 + 1
(n+2)2
− · · · . We have

1 1 2 2
bn − bn+1 = 2 − + − ··· .
n n2 (n + 1)2 (n + 2)2
Therefore,
(continued)
∞ 2
1 1 1
− + − ···
n2 (n + 1)2 (n + 2)2
n=1
2
1 1 1
= lim n − + − ···
n→∞ (n + 1)2 (n + 2) 2 (n + 3)2
∞
1 1 2 2
+ − + − · · ·
n n2 (n + 1)2 (n + 2)2
n=1
∞ (3.2)
1 1 2 2
= − + − ···
n n2 (n + 1)2 (n + 2)2
n=1
∞
1 2 2 2 1
= − + − ··· −
n n2 (n + 1)2 (n + 2)2 n2
n=1
∞
1 1 1 1
=2 − + − ··· − ζ (3).
n n2 (n + 1)2 (n + 2)2
n=1
1
Using that 1
k2
=− 0 x k−1 ln x dx, we obtain
1 1 1 1 x n−1
− + − ··· = − ln x dx, n ≥ 1.
n2 (n + 1)2 (n + 2)2 0 1+x
It follows that
∞ ∞ 1
1 1 1 1 1 x n−1
− + − ··· =− ln x dx
n n2 (n + 1)2 (n + 2)2 n 0 1+x
n=1 n=1
1 ∞
ln x x n−1
=− dx
0 1+x n
n=1
1 ln x ln(1 − x)
= dx.
0 x(1 + x)
We calculate the preceding integral and we have
(continued)
110 3 Power Series

1 ln x ln(1 − x) 1 1 1
dx = ln x ln(1 − x) − dx
0 x(1 + x) 0 x 1+x
1 ln x ln(1 − x) ln x ln(1 − x)
1
= dx − dx
0 x 0 1+x

1 1 ln2 x + ln2 (1 − x) − ln2 1−xx
= ζ (3) − dx
2 0 1+x
π 2 ln 2 5
= − ζ (3),
4 8
since
1 ln2 x 3
• dx = ζ (3);
0 1+x 2
1 ln2 (1 − x) 7 π 2 ln 2 ln3 2
• dx = ζ (3) − + ;
0 1+ x 4 6 3
1 ln2 1−xx 1−x
=t ∞ ln2 t π 2 ln 2 ln3 2
• dx x= dt = + .
0 1+x 0 (1 + t)(2 + t) 3 3
It follows that
∞
1 1 1 1 π 2 ln 2 5
− + − ··· = − ζ (3), (3.3)
n n2 (n + 1)2 (n + 2)2 4 8
n=1
and the quadratic series is calculated based on formula (3.2).
A Series with Harmonic Numbers

We prove that
∞
1 1 1 11 π 2 ln 2
Hn − + − ··· = ζ (3) + .
n2 (n + 1)2 (n + 2)2 16 8
n=1
(continued)
We have
∞
1 1 1
S= Hn − + − ···
n2 (n + 1)2 (n + 2)2
n=1
∞
1 1 1 1 1 1
= − 2 + 2 + ··· + Hn − + − ···
12 2 3 n2 (n + 1)2 (n + 2)2
n=2
2 ∞
n−1=m π 1 1 1
= + Hm+1 − + − ···
12 (m + 1) 2 (m + 2) 2 (m + 3)2
m=1
∞
π2 1 1 1 1 1
= + Hm + − − + − ···
12 m+1 m2 m2 (m + 1) 2 (m + 2)2
m=1
∞ ∞
π2 Hm 1 1 1 1
= + −S+ − + − ···
12 m2 m+1 (m + 1) 2 (m + 2) 2 (m + 3)2
m=1 m=1
∞
m+1=n π2 1 1 1 1
= + 2ζ (3) − S + − + − ···
12 n n2 (n + 1)2 (n + 2)2
n=2
(3.3) π 2 ln 2 5
= 2ζ (3) − S + − ζ (3)
4 8
11 π 2 ln 2
= ζ (3) + − S,
8 4
and the result follows.

∞
1 1 1
(−1) Hnn
− + − ···
n2 (n + 1)2 (n + 2)2
n=1
5 π2 ln2 2
= − ζ (3) − + ln 2 + .
8 12 2
112 3 Power Series
Two Nonlinear Series with Gaps

We prove that
∞
1 1 1
+ + + · · ·
n2 (n + 2)2 (n + 4)2
n=1

1 1 1
× + + + ···
(n + 1)2 (n + 3)2 (n + 5)2
21 π 2 ln 2
= ζ (3) −
16 8
and the quadratic series with gaps
∞ 2
1 1 1 3 π4 π 2 ln 2
+ + + · · · = ζ (3) + + .
n2 (n + 2)2 (n + 4)2 16 128 8
n=1
Let
1 1 1
an = + + + ···
n2 (n + 2)2 (n + 4)2
and let
∞ ∞
S1 = an an+1 and S2 = an2 .
n=1 n=1
(a+b)2 −a 2 −b2
On the one hand, using that ab = 2 , we have
∞ 2
1 1 1 1
S1 = + + + ···
2 n2 (n + 1)2 (n + 2)2
n=1
∞ 2
S2 1 1 1
− − + + ···
2 2 (n + 1)2 (n + 3)2
n=1
∞ 2
3
(7.2) S2 1 1 1
= ζ (3) − − + + ···
2 2 2 m2 (m + 2)2
m=2
2
3 S2 1 1 1 1
= ζ (3) − − S2 − + 2 + 2 + ···
2 2 2 12 3 5
3 π4
= ζ (3) − S2 + ,
2 128
(continued)
since 112 + 312 + 1

52
+ · · · = 34 ζ (2).
It follows that
3 π4
S1 + S2 = ζ (3) + .
2 128
a 2 +b2 −(a−b)2
On the other hand, using that ab = 2 , we have
∞ 2
S2 1 1 1
S1 = + + + ···
2 2 (n + 1)2 (n + 3)2
n=1
∞ 2
1 1 1 1 1
− − + − + ···
2 n2 (n + 1)2 (n + 2)2 (n + 3)2
n=1

S2
(3.1) 1 9 2 1 π 2 ln 2 9
= + S2 − ζ (2) − − ζ (3)
2 2 16 2 2 4
π4 π 2 ln 2 9
= S2 − − + ζ (3).
128 4 8
It follows that
π4 π 2 ln 2 9
S1 − S2 = − − + ζ (3).
128 4 8
Solving the system of equations we obtain the values of series S1 and S2 .
3.91. Open Problem. Calculate

∞
1 1 1 1
(−1)n + + ··· + + ···
n2 (n + 2)2 (n + 1)2 (n + 3)2
n=1
and
∞ 2
1 1 1
(−1) n
+ + + ··· .
n2 (n + 2)2 (n + 4)2
n=1
114 3 Power Series
3.92. Prove that

∞
1 1 1 1 13 4
− + − ··· = π .
n2 n2 (n + 1)2 (n + 2)2 1440
n=1
3.93. [15 April 2019]
(a) Prove that

∞
1 x2 xn
ln −x− − ··· −
1−x 2 n
n=1

1 (−1)2 (−1)n
× ln − (−1) − − ··· −
2 2 n
.
ln(1+x)−x ln 2
+ 12 ln 1−x
1+x − ln 2 ln(1 − x) if x ∈ (−1, 1)
= 1−x
ln 2 − ln2 2 if x = −1.
In particular, for x = −1, we obtain the remarkable quadratic series
∞ 2
1 (−1)2 (−1)n
ln − (−1) − − ··· − = ln 2 − ln2 2.
2 2 n
n=1
(b) Prove that

∞
1 x2 xn
(−1)n ln −x− − ··· −
1−x 2 n
n=1

1 (−1)2 (−1)n
× ln − (−1) − − ··· −
2 2 n
⎧
⎨− ln2 (1−x) + 1 Li 1 − 1 Li 1+x − ln 2 ln(1−x) if x ∈ (−1, 1)
4 2 2 2 2 2 2 2
=
⎩ π 2 − ln2 2 if x = −1.
24
(continued)
3.10 Multiple Series with the Riemann Zeta Function 115
In particular, for x = −1, we obtain the remarkable alternating

quadratic series
∞ 2
1 (−1)2 (−1)n π2
(−1)n ln − (−1) − − ··· − = − ln2 2.
2 2 n 24
n=1
3.94. [22 April 2019] A Nonlinear Logarithmic Series

Prove that
∞
1 x2 xn 1 y2 yn
ln −x− − ··· − ln −y− − ··· −
1−x 2 n 1−y 2 n
n=1
x y
= − ln(1 − x) ln(1 − y) − ln(1 − y) − ln(1 − x)
1−x 1−y
1 − xy
+ ln(1 − xy), ∀ x, y ∈ [−1, 1).
(1 − x)(1 − y)
A particular case. When x = y ∈ [−1, 1), we recover the quadratic series

involving the tail of the logarithmic function [26, problem 3.79, p. 152]
∞ 2
1 x2 xn
ln −x− − ··· −
1−x 2 n
n=1
2x 1+x
= − ln2 (1 − x) − ln(1 − x) + ln(1 − x 2 ).
1−x 1−x
3.10 Multiple Series with the Riemann Zeta Function
3.95. An alternating double series.

Prove that
∞ ∞
(−1)n+m ζ (3)
= .
nm(n + m) 4
n=1 m=1
116 3 Power Series
3.96.
(a) Prove that

∞ ∞
(ζ (i + j ) − 1) = ζ (2).
i=1 j =1
(b) Let k ≥ 2 be an integer. Prove that

∞ ∞
··· (ζ (i1 + · · · + ik ) − 1) = ζ (k).
i1 =1 ik =1
3.97. [4 April 2018]
(a) Prove that

∞ ∞
ζ (i + j ) − 1 ζ (2)
i = .
i+j 2
i=1 j =1

∞ ∞
ζ (i1 + · · · + ik ) − 1 ζ (k)
··· i1 = .
i1 + · · · + ik k
i1 =1 ik =1
3.98. Prove that:

∞ ∞
π2 3
(a) (ζ (2i + j ) − 1) = − ;
12 8
i=1 j =1
∞ ∞ 0 1 π2 25
(b) 22i+j (ζ (2i + j ) − 1) − 1 = − .
3 24
i=1 j =1
3.99.
(a) Prove that

∞ ∞
i (ζ (i + j ) − 1) = ζ (2) + ζ (3).
i=1 j =1
(continued)
(b) Let n, k ∈ N be such that n ≥ 2 and 1 ≤ k ≤ n. Prove that
∞ ∞ k
j
··· i1 · · · ik (ζ (i1 + · · · + in ) − 1) = Ck ζ (j + n).
i1 =1 in =1 j =0
3.100.
(a) Prove that

∞ ∞
ij (ζ (i + j ) − 1) = ζ (2) + 2ζ (3) + ζ (4).
i=1 j =1
∞ ∞ k
j
··· i1 · · · ik (ζ (i1 + · · · + ik ) − 1) = Ck ζ (k + j ).
i1 =1 ik =1 j =0
3.101. [5 May 2018]
(a) Prove that

∞ ∞ 0 1
2i+j (ζ (i + j ) − 1) − 1 = 4ζ (2).
i=1 j =1

∞ ∞ 0 1
··· 2i1 +···+ik (ζ (i1 + · · · + ik ) − 1) − 1 = 2k ζ (k).
i1 =1 ik =1
118 3 Power Series
3.102. [6 September 2018]
(a) Prove that

∞ ∞
1
3i+j
ζ (i + j ) − 1 − − 1 = 9ζ (2).
2i+j
i=1 j =1
(b) Prove that

∞ ∞
1
ij 3i+j ζ (i + j ) − 1 − − 1 = 9ζ (2) + 54ζ (3) + 81ζ (4).
2i+j
i=1 j =1
A challenge. Let n ≥ 2 be an integer. Calculate

∞ ∞
1
··· (n + 1)i1 +···+in ζ (i1 + · · · + in ) − 1 − · · · − −1 .
ni1 +···+in
i1 =1 in =1
3.103. Prove that:

∞ ∞
3
(a) (−1)i−1 (ζ (i + j ) − 1) = ;
4
i=1 j =1
∞ ∞ 0 1 25
(b) (−1)i−1 2i+j (ζ (i + j ) − 1) − 1 = .
12
i=1 j =1
3.104.
(a) Prove that

∞ ∞
5
(−1)i+j (ζ (i + j ) − 1) = ζ (2) − .
4
i=1 j =1
(continued)
3.11 Series Involving Products of Harmonic Numbers 119

∞ ∞
1
··· (−1)i1 +···+ik (ζ (i1 + · · · + ik ) − 1) = (−1)k ζ (k) − 1 − k .
2
i1 =1 ik =1
3.11 Series Involving Products of Harmonic Numbers
The next problems, which are new in the literature, are about the calculations
of special nonlinear Euler sums. A series which contains a product of at least
two harmonic numbers is called a nonlinear Euler sum. The investigation of
nonlinear series has been motivated by the discovery of the famous quadratic series
∞ Hn2
n=1 n2 = 4 ζ (4). This series was introduced in the literature by Sandham and
17
was rediscovered 45 years later by Au-Yeung. See Chap. 8 for two new proofs and
for more information about the history of this formula.
3.105. Series with Consecutive Harmonic Numbers

(a) [25] A harmonic sum.
Prove that
∞
Hn Hn+1
· = ζ (2) + 2ζ (3).
n n+1
n=1
(b) An alternating harmonic sum.

Prove that
∞
Hn Hn+1 7 π2 2
(−1)n · = − ζ (3) − − ln3 2 + ln2 2 + ζ (2) ln 2.
n n+1 8 12 3
n=1
Open problem. Calculate

∞ ∞
Hn−1 Hn Hn+1 Hn−1 Hn Hn+1
· · and (−1)n · · .
n−1 n n+1 n−1 n n+1
n=2 n=2
3.106. Prove that

∞
Hn Hn+1 17
· = −2ζ (2) + ζ (4).
n2 n+1 4
n=1
120 3 Power Series
3.107. Telescoping Series

(a) Prove that
Hn2 Hn+1 H3 H3 H2 H2 2
2Hn+1 2Hn+1
· = n − n+1 + n − n+1 + +
n n+1 n n+1 n n + 1 (n + 1)2 (n + 1)2
Hn+1 1
− − , n ≥ 1.
(n + 1)3 (n + 1)3
(b) Prove that

∞
Hn2 Hn+1 29
· = 3ζ (3) + ζ (4).
n n+1 4
n=1
(c) Prove that

2 3 2
Hn Hn+1 Hn3 Hn+1 Hn2 Hn+1 Hn+1
· = − +2 − +3
n n+1 n n+1 n n+1 (n + 1)2
2
Hn+1
1 Hn Hn+1 1
− + − + + , n ≥ 1.
(n + 1)3 n n + 1 (n + 1)2 (n + 1)2
(d) Prove that

∞ 2
Hn Hn+1 17
· = ζ (2) + 5ζ (3) + ζ (4).
n n+1 4
n=1

∞ 2
Hn2 Hn+1 69 33
· = 3ζ (3) + ζ (4) + ζ (5) + 3ζ (2)ζ (3).
n n+1 4 2
n=1
3.108. Prove that

∞
Hn Hn+1
· = ζ (2) + 2ζ (3) − 3ζ (4).
n (n + 1)2
n=1
Derivatives and Applications
4
The teacher you will be studying with,

such knowledge you will be learning.
A Greek Proverb
Chapter 4 is about derivatives and their applications. This chapter contains problems
on the classical topics of mathematical analysis: the calculation of the nth derivative
of a function, Leibniz’s formula, Taylor’s formula, the extremum points of a function
of a single variable, and non-standard topics of analysis such as the generalized
Leibniz formula, special differential equations, and the computation of exotic series
involving the Maclaurin remainder of special functions.
4.1 Apéritif
4.1. Let h : R → R, h(x) = x 3 − 3x. Determine J = h(I ), where I = (1, ∞).

Prove that the function f : I → J , f (x) = h(x) is bijective. Let g be the inverse of
f . Calculate g (2) and g (2).
4.2. Let a ∈ R. Find all differentiable functions f : R → R such that f (x + y) =

f (x) + ay, for all x ∈ R and y ∈ R∗ .
4.3. [113] Find all functions f : R → R, derivable at 0, with f (0) = 1, such that
f (2x) = 2f 2 (x) − 4f (x) + 3, ∀x ∈ R.
4.4. Let f (x) = x(x + 1) · · · (x + 2021) and let g = f ◦ f ◦ f . Calculate g (0).
122 4 Derivatives and Applications
Fig. 4.1 Equal areas
4.5. The areas of two regions determined by a parabola and two tangents.
Let (x0 , y0 ), x0 = 0, be a point on the graph of the parabola y = ax 2 , a > 0.
The tangent to the parabola at (x0 , y0 ) intersects the x axis at (x1 , 0). Prove that the
area of the region bounded by the tangent to the parabola at (x0 , y0 ), parabola, and
the vertical line x = x1 is equal to the area of the region bounded by the parabola,
the x axis, and the vertical line x = x1 . (See Fig. 4.1.)
4.6. Prove that (1 + x)1−x ≤ 1 + x − x 2 , ∀x ∈ [0, 1].
4.7. Let n ∈ N. Calculate:
(a) 1 + x + 2x 2 + 3x 3 + · · · + nx n , x ∈ R;
(b) 1 + 22 x + 32 x 2 + · · · + n2 x n−1 , x ∈ R;
(c) 2 + 3 · 2x + 4 · 3x 2 + · · · + n(n − 1)x n−2 , x ∈ R.
4.8. Two inequalities and a limit with e.
(a) Prove that

e 1 n e
<e− 1+ < , ∀n ≥ 1.
2n + 2 n 2n + 1
(continued)
4.1 Apéritif 123
(b) Calculate

1 n
lim n e − 1 + .
n→∞ n
4.9. Let (xn )n≥1 be the sequence defined by xn = 1 + 12 + · · · + n1 − ln n and

let γ = lim xn . Prove that
n→∞
1 1
< xn − γ < , ∀n ≥ 1.
2n + 1 2n
4.10. [18] Identities with derivatives. Check that:
(f g) f g f g
(a) = + +2 · ;
fg f g f g
(f/g) f g (f/g) g
(b) = − −2 · .
f/g f g f/g g
We record, as gems, the proofs of two amazing formulae of mathematical

analysis.
(a) Euler’s Formula eix = cos x + i sin x, ∀x ∈ R.

Let f : R → C, f (x) = −ix
eix − cos x − i sin x. We have f (x) =
−ix
if (x) ⇒ e f (x) = 0 ⇒ e f (x) = C , C ∈ C. We obtain f (x) =
C eix , C ∈ C. Since f (0) = 0 ⇒ C = 0 ⇒ f (x) = 0, for all x ∈ R.
(b) [127] de Moivre’s Formula If x ∈ R and n ∈ Z, then
(cos x + i sin x)n = cos(nx) + i sin(nx).
Since the formula is true for n = 0, we consider that n = 0. Let

f : R → C, f (x) = cos x+i sin x. We need to prove that f n (x) = f (nx)
or, equivalently, f (nx)f n (−x) = 1. Using that f (x) = if (x), we have

f (nx)f n (−x) = nf n−1 (−x) f (nx)f (−x) − f (nx)f (−x)
= i · n · f n−1 (−x) [f (nx)f (−x) − f (nx)f (−x)]
= 0.
(continued)
It follows that f (nx)f n (−x) = C , C ∈ C, for all x ∈ R. When x = 0 ⇒

C = f n+1 (0) = 1. Therefore, f (nx)f n (−x) = 1, for all x ∈ R.
4.2 Integral Equations
4.11. Find all continuous functions f : R → R which satisfy the equation

x
f (x) = x + e−t f (x − t) dt, ∀x ∈ R.
0
4.12. Let a ∈ R. Find all continuous functions f : R → R which satisfy the

equation:
x
(a) f (x) = a + sin tf (x − t) dt, ∀x ∈ R;
0
x
(b) f (x) = a − sin tf (x − t) dt, ∀x ∈ R.
0
4.13. Find all continuous functions f : R → R which satisfy the equation:

x
(a) f (−x) = 1 + e−t f (x − t) dt, ∀x ∈ R;
0
x
(b) f (−x) = x + e−t f (x − t) dt, ∀x ∈ R;
0
x
(c) f (−x) = sin x + e−t f (x − t) dt, ∀x ∈ R.
0
4.14. Find all continuous functions f : R → R which satisfy the equation:

x
(a) [70] f (−x) = 1 + sin t f (x − t) dt, ∀x ∈ R;
0
x
(b) [78] f (−x) = x + sin t f (x − t) dt, ∀x ∈ R.
0
4.3 Differential Equations 125

equation:
x
(a) f (−x) = a − sin t f (x − t) dt, ∀x ∈ R;
0
x
(b) f (−x) = x − sin t f (x − t) dt, ∀x ∈ R.
0

equation
x
f (x) = a − cos t f (x − t) dt, ∀x ∈ R.
0

equation:
x
(a) f (x) − tf (x − t) dt = a, ∀x ∈ R;
0
x
(b) f (x) + tf (x − t) dt = a, ∀x ∈ R;
0
x
(c) f (−x) + tf (x − t) dt = a, ∀x ∈ R.
0
4.3 Differential Equations
4.18. A problem of Vasile Pop.
(a) Find all differentiable functions f : R → R which satisfy the equation
f (x) − 2xf (−x) = x, for all x ∈ R.
(b) Find all differentiable functions f : R → R which satisfy the equation
f (x) + 2xf (−x) = x, for all x ∈ R.

4.19. [52] Find all differentiable functions f : R → R which satisfy the

equation
xf (x) + f (−x) = x 2 , for all x ∈ R.
4.20. [38] Let k ∈ N. Find all differentiable functions f : R → R which satisfy

the equation
xf (x) + kf (−x) = x 2 , for all x ∈ R.
4.21. A Cauchy problem.

√
(a) Find all differentiable functions f : (0, ∞) → (0, ∞), with f (1) = 2, such
that

1 1
f = , ∀x > 0.
x f (x)
(b) Find all differentiable functions f : (0, ∞) → (0, ∞), with f (1) = 2, such
that

1 1
f = , ∀x > 0.
x f (x)
4.22. Let a, b > 0. Find all differentiable functions f : (0, ∞) → (0, ∞) such
that1
a bx
f = , ∀x > 0.
x f (x)
4.23. [A. Sîntămărian, 2016] Find all differentiable functions f : [−1, 1] →

(0, ∞) such that
sin x
f (sin x) = , ∀x ∈ R.
f (cos x)

4.24. [A. Sîntămărian, 2016] Find all differentiable functions f : 0, π2 → (0, ∞)
such that
1
f (arctan x) = , ∀x > 0.
f arctan x1
1 For b = 1 we obtain problem B–3, Putnam 2005.

4.25. Find all continuous functions f : R → R which satisfy the equation

x+y
f (t)dt = f (x)f (y), ∀x, y ∈ R.
x−y
4.26.
(a) Find all nonconstant differentiable functions f : R → R which satisfy

the equation
f (x + y) − f (x − y) = 2f (x)f (y), ∀x, y ∈ R.
(b) Find all nonconstant differentiable functions f : R → R which satisfy

the equation
f (x + y) − f (x − y) = 2f (x)f (y), ∀x, y ∈ R.
4.27. A Differential Equation Used for the Calculation of a Series
(a) Find all differentiable functions f : R → R which satisfy the equation
f (x) − f (x) = x ex , ∀x ∈ R.
(b) Calculate
∞
x x2 xn
n ex − 1 − − − ··· − .
1! 2! n!
n=1
4.28. Other Differential Equations Used for the Calculation of Some

Series
Let k ≥ 0 be an integer.
(a) Find all differentiable functions f : R → R such that
(continued)
xk x
f (x) − f (x) = e , ∀x ∈ R.
k!
(b) Prove that
∞
x x2 xn x k+1 x
Ckn e −1− −
x
− ··· − = e .
1! 2! n! (k + 1)!
n=k
(c) Find all differentiable functions f : R → R such that
xk
f (x) − f (x) = (x + k) ex , ∀x ∈ R.
k!
(d) Prove that
∞
x x2 xn
nCkn e −1− −
x
− ··· −
1! 2! n!
n=k
⎧ 2
⎪
⎪ x x
⎪
⎪ e if k=0
⎨ 2
=
⎪
⎪
⎪
⎪ x k+1 x k+2
⎩ + ex if k ≥ 1.
(k − 1)!(k + 1) k!(k + 2)
4.29. Let x ∈ R.
(a) Prove that

∞
xn x n+1 x n+2
− + − ··· = sinh x.
n! (n + 1)! (n + 2)!
n=1
In particular, when x = 1 we obtain the series

∞
1 1 1
− + − ··· = sinh 1.
n! (n + 1)! (n + 2)!
n=1
(b) Prove that

∞
xn x n+1 x n+2 1
−2 +3 − ··· = sinh x + xe−x .
n! (n + 1)! (n + 2)! 2
n=1

∞
1 2 3 cosh 1
− + − ··· = .
n! (n + 1)! (n + 2)! 2
n=1
4.30. Let x ∈ R. Prove that

∞
xn x n+1 x n+2 sinh x xex
n − + − ··· = + .
n! (n + 1)! (n + 2)! 2 2
n=1

∞
1 1 1 3e − e−1
n − + − ··· = .
n! (n + 1)! (n + 2)! 4
n=1
4.31. [54] The Behavior of an Exponential Series

Let k ∈ N. Prove that
∞
−x x x2 xn (−1)k
lim e (−1)n Ckn e −1− −
x
− ··· − = k+1 .
x→∞ 1! 2! n! 2
n=k
4.32. Let x ∈ R. Prove that:

∞"n# x x2 xn

1 1
(a) ex − 1 − − − ··· − = sinh x + (x 2 − x)ex ;
2 1! 2! n! 4 4
n=1
∞ "n# x x 2 x n
(b) (−1)n ex − 1 − − − ··· −
2 1! 2! n!
n=1
1 1
= x cosh x − sinh x;
2 2
∞ " # x2 xn

nn x x
(c) (−1) e −1− −
x
− ··· − = − sinh x;
2 1! 2! n! 2
n=1
∞ "n#
x x2 xn
(−1) 2
n
(d) ex − 1 − − − ··· −
2 1! 2! n!
n=1
(continued)
1
= sin x + cos x + x sin x − ex ;
2
∞
x x2 xn
(−1) 2 n ex − 1 − −
n
(e) − ··· −
1! 2! n!
n=1
1 cos x − ex
= x+ sin x + .
2 2
4.4 Higher Order Derivatives
4.33. Higher Order Derivatives Let a, b ∈ R, with a = 0. Prove that:

2 3 a n n!
(a) f : R\ − ab → R, f (x) = ax+b
1
, f (n) (x) = (−1)n (ax+b) n+1 , n ≥ 0;

(b) f : − |a| , ∞ → R, f (x) = ln(|a|x + b),
b
|a|n (n − 1)!
f (n) (x) = (−1)n−1 , n ∈ N;
(|a|x + b)n
(c) f : R → R, f (x) = sin(ax + b),

2 , n ≥ 0;
f (n) (x) = a n sin ax + b + nπ
(d) f : R → R, f (x) = cos(ax + b),
2 , n ≥ 0;
f (n) (x) = a n cos ax + b + nπ
(e) f : R → R, f (x) = eax+b , f (n) (x) = a n eax+b , n ≥ 0.
4.34. Divertimento Prove that:
(a) sin x is not a polynomial function;

(b) ex is not a polynomial function;
(c) ex is not a rational function.
4.35. Let f be a polynomial with real coefficients such that f (x) ≥ 0, for all x ∈ R.
Let g(x) = f (x) + f (x) + f (x) + · · · . Prove that g(x) ≥ 0, for all x ∈ R.
4.4 Higher Order Derivatives 131
An Inequality with Derivatives of a Polynomial Function

Let p be a polynomial with real coefficients and real roots. The following
inequality holds true:
p 2 (x) ≥ p(x)p (x), ∀x ∈ R.
Proof. Let x1 , x2 , . . . , xn be the distinct real roots of p with multiplicities

k1 , k2 , . . . , kn respectively. If x = xi is a root of p, then the inequality is true.
We assume that x is not a root of p.
We have
n
p (x) ki
=
p(x) x − xi
i=1
and it follows by differentiation that

n
p (x)p(x) − p 2 (x) ki
=− ,
p2 (x) (x − xi )2
i=1
and the inequality to prove follows.
4.36. Special Differential Equations
(a) Find all functions f ∈ C ∞ ((−R, R), R), R ∈ (0, ∞], such that
∞
f (x) = f (n) (x), for all x ∈ (−R, R).
n=1
(b) Find all functions f ∈ C ∞ ((−R, R), R), R ∈ (0, ∞], such that
∞
f (x) = (−1)n f (n) (x), for all x ∈ (−R, R).
n=1
(c) Let a ∈ R∗ . Find all functions f ∈ C ∞ ((−R, R), R), R ∈ (0, ∞], such
that
∞
f (x) = a n f (n) (x), for all x ∈ (−R, R).
n=1
(continued)
(d) Find all functions f ∈ C ∞ ((−R, R), R), R ∈ (0, ∞], such that
∞
f (x) = nf (n) (x), for all x ∈ (−R, R).
n=1
(e) Find all functions f ∈ C ∞ ((−R, R), R), R ∈ (0, ∞], such that
∞
f (x) = f (2n−1) (x), for all x ∈ (−R, R).
n=1
(f) Find all functions f ∈ C ∞ ((−R, R), R), R ∈ (0, ∞], such that
∞
f (x) = f (2n) (x), for all x ∈ (−R, R).
n=1
(g) Find all functions f ∈ C ∞ ((−R, R), R), R ∈ (0, ∞], such that
∞
f (−x) = f (2n) (x), for all x ∈ (−R, R).
n=1
4.37. Calculate the derivative of order n of the following functions:
(a) f : R \ {−1} → R, f (x) = 3x

1+x ;
(b) f : R \ −3, 12 → R, f (x) = 1
2x 2 +5x−3
;
(c) f : (0, ∞) → R, f (x) = 1+x
√ ;
x
(d) f : R∗ → R, f (x) = 1+x
√
3x ;
(e) f : R → R, f (x) = sin x;

2
(f) f : R → R, f (x) = cos2 x;

(g) f : R → R, f (x) = sin3 x;
(h) f : R → R, f (x) = cos3 x;
(i) f : R → R, f (x) = sinh x;
(j) f : R → R, f (x) = cosh x.
4.38. Let f : R → R be the function defined by f (x) = sin4 x + cos4 x. Prove that
nπ
f (n) (x) = 4n−1 cos 4x + , n ∈ N.
2
4.39. [M. Ţena, 2011] Let f, g : R → R, f (x) = ex sin x, g(x) = ex cos x. Prove
that, for all n ≥ 0 an integer and x ∈ R, we have
2 2
f (n) (x) + g (n) (x) = 2n e2x .
Remark 4.1. If f (x) = e−x sin x and g(x) = e−x cos x, then for all n ≥ 0 an
integer and x ∈ R, we have
2 2
f (n) (x) + g (n) (x) = 2n e−2x .
4.40. Series with derivatives. Let k ∈ N and f : R → R, f (x) = 1

1+x 2
.
(a) [5] Prove that

∞

f (4n) (1) f (2n) (1)
+ = 1.
(4n)! (2n)!
n=0
(b) [34] Prove that

∞
f (2kn) (1) 2k − cos kπ
2 + sin 2
kπ
= 2k−1 .
n=0
(2kn)! 22k − 2k+1 cos kπ
2 +1
Leibniz’s Formula Let n ≥ 0 be an integer, a, b ∈ R, a < b, and let

f, g : (a, b) → R be n times differentiable functions on (a, b). Then
n
(f · g)(n) (x) = Ckn f (n−k) (x)g (k) (x), x ∈ (a, b).
k=0
4.41. Calculate the derivative of order n of the following functions:
(a) f : R → R, f (x) = (x 2 − 3x − 1)e2x ;

(b) f : R → R, f (x) = x−1
ex ;
(c) f : R → R, f (x) = eax cos(bx + c), a, b, c ∈ R.
4.42. Let n ≥ 0 be an integer and f, g : R → R, f (x) = x sin x, g(x) = x cos x.

Solve the equation
f (2n) (x) + g (2n+1) (x) = (−1)n sin x.

4.43. Let f : (−1, ∞) → R, f (x) = ln2 (x + 1). Prove that, ∀n ∈ N, n ≥ 2, we

have

1 1 1 1
f (n) (x) = 2(−1)n−1 (n−1)! ln(x + 1) − 1 + + + · · · + .
2 3 n−1 (x + 1)n
4.44. Let f : (0, ∞) → R, f (x) = ln x

x . Prove that

n! 1 1 1
f (n)
(x) = (−1) n−1
1 + + + · · · + − ln x , n ∈ N.
x n+1 2 3 n
4.45. Let n ∈ N and f : (0, ∞) → R, f (x) = x n ln x. Prove that

1 1 1
f (n)
(x) = n! 1 + + + · · · + + ln x .
2 3 n
(1+x)2 −ln x
4.46. Let f : (0, ∞) → R, f (x) = x . Prove that

n! 1 1 1
f (n)
(x) = (−1) n
1 + 1 + + + · · · + − ln x , n ≥ 2.
x n+1 2 3 n
1
4.47. Let n ∈ N and f : R∗ → R, f (x) = x n−1 e x . Prove that
f (x)
f (n) (x) = (−1)n .
x 2n
4.48. Let f : [−1, 1] → R, f (x) = arcsin2 x. Calculate f (n) (0).
4.49. Let f, g, h : R → R be the functions
1
f (x) = , g(x) = arctan x, h(x) = g 2 (x).
1 + x2
Calculate f (n) (0), g (n) (0), and h(n) (0).
4.50. Let f : R \ {−1} → R, f (x) = 1

1+x 3
. Calculate f (n) (0).

4.51. Let f : R → R, f (x) = 1
x 2 +x+1
. Calculate f (n) − 12 .
4.52. Let f : (−∞, 1) → R, f (x) = ln(1 − x 3 ). Calculate f (n) (0).

4.53. Limits of nth Order Derivatives Prove that:

ex − ea (n)
• lim n = ex , x, a ∈ R, x = a;
n→∞ x−a

sin x (2n+1) sin x
• lim (−1)n−1 n = , x ∈ R∗ ;
n→∞ x 2

sin x (2n) cos x
• n
lim (−1) n = , x ∈ R∗ ;
n→∞ x 2

1 − cos x (2n) sin x
• n
lim (−1) n = , x ∈ R∗ ;
n→∞ x 2

1 − cos x (2n−1) cos x
• n−1
lim (−1) n = , x ∈ R∗ ;
n→∞ x 2

(1 + x)n ln(1 + x) (n)
• lim (−1)n = 1, x ∈ (−1, 1), x = 0.
n→∞ (n − 1)! x
Remark 4.2. The case when a = 0 in the first limit is problem 4, part (b), section A,
given in 2018 by M. Ivan at Traian Lalescu National Contest for University Students
held every year in Romania [110].
The Generalized Leibniz Formula Let a, b ∈ R, a < b, m ≥ 1 an integer

and let f1 , f2 , . . . , fm : (a, b) → R be n times differentiable functions on
(a, b). Then
n! (k ) (k ) (k )
(f1 f2 · · · fm )(n) (x) = f 1 (x)f2 2 (x) · · · fm m (x),
k1! k2 ! · · · km ! 1
k1 +k2 +···+km =n
where the sum is over all m-tuples (k1 , k2 , . . . , km ) of integers

k1 , k2 , . . . , km ≥ 0 such that k1 + k2 + · · · + km = n.
In particular, when m = 2 we obtain the classical Leibniz formula
n
n! (k ) (k ) (k1 ) (n−k1 )
(f1 · f2 )(n) (x) = f 1 (x)f2 2 (x) = Ckn1 f1 (x)f2 (x).
k1 !k2 ! 1
k1 +k2 =n k1 =0
4.54. [69] Let n ∈ N and fn : R → R, fn (x) = sin x sin(2x) · · · sin(nx). Prove

(n)
that fn (0) = (n!)2 .
4.55. Let n ∈ N and gn : R → R, gn (x) = cos x cos(2x) · · · cos(nx). Calculate

gn(n) (0).
4.56. Let n ∈ N and fn : R → R, fn (x) = sinh x sinh(2x) · · · sinh(nx). Calculate

(n)
fn (0).
4.57. Let n ∈ N and gn : R → R, fn (x) = cosh x cosh(2x) · · · cosh(nx). Calculate

(n)
gn (0).
4.5 Taylor’s Formula
Taylor’s Formula Let I ⊆ R be an open interval and f : I → R be an n + 1

times differentiable function on I . Then, for all x0 , x ∈ I , with x = x0 , there
exists θx0 ,x ∈ (0, 1) such that the Taylor formula or order n corresponding to
the function f and the point x0 holds true
f (x) = (Tn f )(x) + (Rn f )(x),
where
n
f (k) (x0 )
(Tn f )(x) = (x − x0 )k
k!
k=0
is the Taylor polynomial of degree n corresponding to the function f and the

point x0 , and
f (n+1) (x0 + θx0 ,x (x − x0 ))

(Rn f )(x) = (x − x0 )n+1
(n + 1)!
is the remainder of order n corresponding to the function f and the point x0

(the remainder in the form of Lagrange).
We mention three particular cases of Taylor’s formula.
I. If x0 = 0 ∈ I , then we obtain the Maclaurin formula of order n corres-
ponding to the function f
f (x) = (Mn f )(x) + (Rn f )(x),
where
(continued)
4.5 Taylor’s Formula 137
n
f (k) (0) k
(Mn f )(x) = x
k!
k=0
is the Maclaurin polynomial of degree n corresponding to the function f ,

and
f (n+1) (θx x) n+1

(Rn f )(x) = x
(n + 1)!
is the remainder of order n corresponding to the function f .

II. If f : R → R is a polynomial function of degree n and x0 ∈ R, then
f (x) = (Tn f )(x), ∀x ∈ R.
III. If n = 0, then we obtain Lagrange’s Mean Value Theorem
f (x) = f (x0 ) + f (x0 + θx0 ,x (x − x0 ))(x − x0 ).
Taylor’s Formula with the Remainder in Integral Form

Let I be an open interval and let f ∈ C n+1 (I, R). Then, for all x0 , x ∈ I ,
we have
n
f (k) (x0 ) x (x − t)n (n+1)
f (x) = (x − x0 )k + f (t)dt.
k! x0 n!
k=0
4.58. Write the Taylor formula of order n corresponding to the function f and the
point x0 in the following cases:
(a) f : (−3, ∞) → R, f (x) = x+31

, x0 = −1;
(b) f : R → R, f (x) = e 2x−1 , x0 = 2;
(c) f : R → R, f (x) = sin(3x), x0 = π2 ;
(d) f : R → R, f (x) = cosh (x − 1), x0 = 1.
4.59. Write the following functions in terms of powers of x:
(a) f : R → R, f (x) = (x − 1)3 (x + 5)2 ;

(b) f : R → R, f (x) = (x + 1)2 (x − 9)(x + 2),
without expanding the parenthesis.

4.60. Write the function in terms of powers of x − 1
f : R → R, f (x) = x 5 − 5x 4 + 10x 3 − 7x 2 − x + 3.
4.61. Write the function in terms of powers of x + 1
f : R → R, f (x) = x 5 + 5x 4 + 3x 3 − 11x 2 − 16x − 7.
4.62. Write the Maclaurin polynomial of degree n corresponding to the function f
(a) f : (−1, ∞) → R, f (x) = 1

x 2 +4x+3
;
√
(b) f : [−1, ∞) → R, f (x) = x + 1;
(c) f : (−∞, 1) → R, f (x) = ln x−1
x−2 .
4.63. The Maclaurin Formula of Order n for Elementary Functions

Write the Maclaurin formula of order n corresponding to the function f
(a) f : R → R, f (x) = ex ;
(b) f : R → R, f (x) = sin x;
(c) f : R → R, f (x) = cos x;
(d) f : (−1, ∞) → R, f (x) = (1 + x)α , α ∈ R;
(e) f : (−1, ∞) → R, f (x) = ln(1 + x);
(f) f : R → R, f (x) = sinh x;
(g) f : R → R, f (x) = cosh x.
4.64. Let n ∈ N. Prove that
x x2 x 2n
1+ + + ··· + > 0, ∀x ∈ R.
1! 2! (2n)!
√
4.65. Irrational numbers. Prove that e, e−1 , e 2, sin 1, and cos 1 are irrational
numbers.
4.66. (a) Prove that

∞
n3
= 5e.
n!
n=1
(b) Let f be a function which has a Maclaurin series expansion with radius of
convergence R > 1. Prove that
4.6 Series with the Maclaurin Remainder of a Function f 139
∞
n3 (n)
f (0) = f (1) + 3f (1) + f (1).
n!
n=1
4.67. Let f : R → R be a twice derivable function, with f , f , and f continuous

on R. If f (0) = 1, f (0) = 0, and f (0) = p, then prove that
n
x x2
lim f √ = ep 2 .
n→∞ n
4.6 Series with the Maclaurin Remainder of a Function f
4.68. [19 April 2019]
(a) Prove that

∞
xn x n+1 x n+2 x
+ + + ··· = , x ∈ [−1, 1).
n n+1 n+2 1−x
n=1
(b) Let f be a function which has the Maclaurin series expansion f (x) =
∞ (n)
f (0) n
n! x , |x| < R. Prove that
n=0
∞

f (n) (0) n f (n+1) (0) n+1
x + x + ··· = xf (x), |x| < R.
n! (n + 1)!
n=1
(c) Let f be a function which has the Maclaurin series expansion f (x) =
∞ (n)
f (0) n
n! x , |x| < R. Prove that
n=0
∞

f (n) (0) n f (n+1) (0) n+1
an x + x + ···
n! (n + 1)!
n=0
f (x) − af (ax)
= , a = 1, |ax| < R.
1−a

x (x − t)n t x x2 xn
(a) e dt = ex − 1 − − − · · · − , ∀n ≥ 0;
0 n! 1! 2! n!
∞
x x2 xn
(b) ex − 1 − − − ··· − = xex ;
1! 2! n!
n=0
∞
x x2 xn
(c) (−1) e − 1 − −
x n
− ··· − = sinh x.
1! 2! n!
n=0
4.70. Series with the Maclaurin Remainder of a Function f

Let f be a function which has the Maclaurin series expansion with radius
of convergence R ∈ (0, ∞]. Prove that, for |x| < R, we have
∞

f (0) f (0) 2 f (n) (0) n
(a) f (x) − f (0) − x− x − ··· − x
1! 2! n!
n=1
= xf (x) − f (x) + f (0);

∞
f (0) f (0) 2 f (n) (0) n x2
(b) n f (x) − f (0) − 1! x − 2! x − ··· − n! x = 2 f (x);
n=1
∞

f (0) f (0) 2 f (n) (0) n
(c) n2
f (x) − f (0) − x− x − ··· − x
1! 2! n!
n=1
x3 x2
= f (x) + f (x).
3 2
In particular, when f (x) = Lik (x), k ∈ N, k ≥ 2 and x ∈ [−1, 1], we
obtain the polylogarithm series
∞
x2 xn
(a) Lik (x) − x − k − · · · − k = Lik−1 (x) − Lik (x);
2 n
n=1
∞ 2
x xn Lik−2 (x) Lik−1 (x)
(b) n Lik (x) − x − k − · · · − k = − ;
2 n 2 2
n=1
(continued)
∞
x2 xn
(c) n Lik (x) − x − k − · · · − k
2
2 n
n=1
Lik−3 (x) Lik−2 (x) Lik−1 (x)
= − + .
3 2 6
When x = 1 the above formulae become
∞
1 1
(a) ζ (k) − 1 − k − · · · − k = ζ (k − 1) − ζ (k), k > 2;
2 n
n=1
∞
1 1 ζ (k − 2) ζ (k − 1)
(b) n ζ (k) − 1 − k − · · · − k = − , k > 3;
2 n 2 2
n=1
∞
1 1
(c) n ζ (k) − 1 − k − · · · − k
2
2 n
n=1
ζ (k − 3) ζ (k − 2) ζ (k − 1)
= − + , k > 4.
3 2 6
4.71.
(a) Prove that
n(n + 1)
(−1) · 12 + (−1)2 · 22 + · · · + (−1)n · n2 = (−1)n , n ≥ 1.
2
(b) Prove that
∞
x x2 xn x 2 −x
(−1) n e −1− −
n−1 2 x
− ··· − = e , x ∈ R.
1! 2! n! 2
n=1

∞
nπ x x2 xn 1
(a) cos e −1− −
x
− ··· − = sin x − cos x + ex ;
2 1! 2! n! 2
n=0
(continued)
∞
nπ x x2 xn 1
e −1− − − ··· − = − sin x − cos x + ex .
x
(b) sin
2 1! 2! n! 2
n=1
A generalization Let f be a function which has the Maclaurin series expan-

sion with radius of convergence R, and let x ∈ R, |x| < R. Prove that
∞

nπ f (0) f (0) 2 f (n) (0) n
cos f (x) − f (0) − x− x − ··· − x
2 1! 2! n!
n=0
f (x) f (xi) − f (−xi) f (xi) + f (−xi)

= + −
2 4i 4
and
∞

nπ f (0) f (0) 2 f (n) (0) n
sin f (x) − f (0) − x− x − ··· − x
2 1! 2! n!
n=1
f (x) f (xi) − f (−xi) f (xi) + f (−xi)

= − − .
2 4i 4
We record, as a gem, the proof of an amazing series formula involving the

Maclaurin remainder of a function f .
A Series Formula with the Maclaurin Remainder of a Function f

Let f be a function which has the Maclaurin series expansion with radius
of convergence R = ∞. Then
∞

f (0) f (0) 2 f (n) (0) n
2n f (x) − f (0) − x− x − ··· − x = f (2x) − f (x),
1! 2! n!
n=0
for all x ∈ R.
(continued)
Proof. We have
∞

f (0) f (0) 2 f (n) (0) n
S(x) = 2n f (x) − f (0) − x− x − ··· − x
1! 2! n!
n=0
= f (x) − f (0)
∞

f (0) f (0) 2 f (n) (0) n
+ 2n f (x) − f (0) − x− x − ··· − x
1! 2! n!
n=1
= f (x) − f (0)
∞

f (0) f (0) 2 f (m+1) (0) m+1
+ 2m+1 f (x) − f (0) − x− x − ··· − x
1! 2! (m + 1)!
m=0
= f (x) − f (0)
∞

f (0) f (0) 2 f (m) (0) m
+2 2m f (x) − f (0) − x− x − ··· − x
1! 2! m!
m=0
∞
f (m+1) (0)
− (2x)m+1
(m + 1)!
m=0
= f (x) − f (0) + 2S(x) − (f (2x) − f (0)) ,
and it follows that S(x) = f (2x) − f (x).
4.73. Let x ∈ R. Prove that the following series are absolutely convergent
and show that:
∞
1 3 sin nπ
(a) [44] 3 sin x − x + x − · · · −
n
x = sin x cos(2x);
2 n
3! n!
n=0
∞
1 2 cos nπ
(b) [41] 3 cos x − 1 + x − · · · −
n
x = − sin x sin(2x).
2 n
2! n!
n=0
4.74. Let f be a function which has the Maclaurin series expansion with
radius of convergence R and let k ≥ 0. Prove that
∞

x x2 xn 1 x x−t k (k)
Ckn f (n) (0) ex − 1 − − − ··· − = e t f (t)dt, |x| < R.
1! 2! n! k! 0
n=k
4.75. Let f be a function which has the Maclaurin series expansion with radius of
convergence R. Calculate
∞

f (0) f (0) 2 f (n) (0) n
(−1) n
f (x) − f (0) − x− x − ··· − x , |x| < R.
1! 2! n!
n=1

∞
n
x x2 xn
(a) [86] (−1) 2 e − 1 − −
x
− ··· − = 1 − cos x;
1! 2! n!
n=1
∞
n
x x2 x n−1
(b) (−1) 2 e − 1 − −x
− ··· − = sin x.
1! 2! (n − 1)!
n=1
A generalization Let f be a function which has the Maclaurin series expan-

sion with radius of convergence R. Prove that, for |x| < R, we have
∞

n
f (0) f (0) 2 f (n) (0) n
(−1) 2 f (x) − f (0) − x− x − ··· − x
1! 2! n!
n=1
f (ix) + f (−ix)
= f (0) −
2
and
∞

n
f (0) f (0) 2 f (n−1) (0) n−1
(−1) 2 f (x) − f (0) − x− x − ··· − x
1! 2! (n − 1)!
n=1
f (ix) − f (−ix)
= .
2i
4.77. A Spectacular Maclaurin Series
(a) Let k, n ∈ N, n ≥ k. Prove that
Ckk + Ckk+1 + · · · + Ckn = Ck+1

n+1 .
(b) Let k ≥ 0 be an integer and f be a function which has the Maclaurin

series expansion with radius of convergence R ∈ (0, ∞]. Prove that, for
|x| < R, we have
∞

f (0) f (0) 2 f (n) (0) n
Ckn f (x) − f (0) − x− x − ··· − x
1! 2! n!
n=k
f (k+1) (x) k+1

= x .
(k + 1)!

n 2k−1
4.78. Let n ∈ N, Tn (x) = x
(−1)k−1 (2k−1)! be the Maclaurin polynomial of
k=1
degree 2n − 1 corresponding to the sine function and let
∞ Tn (x) − sin x
In = dx.
0 x 2n+1
Prove that:
(a) In = − 2n(2n−1)
1
In−1 , n ≥ 1;
(−1)n−1
(b) In = 2(2n)! π .
4.79. [O. Furdui, Problem 3, Traian Lalescu National Contest for University
Students, 2015]2

n+1
x 2k−2
Let n ≥ 0 be an integer, T2n (x) = (−1)k−1 (2k−2)! be the Maclaurin
k=1
polynomial of degree 2n corresponding to the cosine function and let
∞ T2n (x) − cos x
In = dx.
0 x 2n+2
2 This
problem was given in 2015 at Section B of Traian Lalescu National Contest for University
Students, the 8th edition, Braşov, Romania.
(a) Prove that In = − 2n(2n+1)

1
In−1 , n ≥ 1.
(b) Calculate In .
4.7 Series with Fractional Part Function
4.80. Prove that
lim {n!e} = 0 and lim n {n!e} = 1,

n→∞ n→∞
where {a} is the fractional part of the real number a.
An Invitation to the Calculation of Series with the Floor and the Frac-
tional Part Function
∞ ∞
{n!e} 1 1 1
(a) = e − 1 − − − ··· − = 1;
n! 1! 2! n!
n=1 n=1
∞ ∞
{(2n)!e} 1 1 1 cosh 1
(b) = e − 1 − − − ··· − =1− ;
(2n)! 1! 2! (2n)! 2
n=1 n=1
∞ ∞
{(2n − 1)!e} 1 1 1 cosh 1
(c) = e−1− − − ··· − = ;
(2n − 1)! 1! 2! (2n − 1)! 2
n=1 n=1
∞ ∞
n {n!e} 1 1 1
(d) (−1) = (−1) n
e−1− − − ··· − = 1 − cosh 1;
n! 1! 2! n!
n=1 n=1
∞
{n!e} n ex − ex
(e) x = + 1, for x = 1;
n! x−1
n=1
∞
n!e n ex
(f) x = − 1, for x ∈ (−1, 1);
n! 1−x
n=1
(continued)
4.7 Series with Fractional Part Function 147
∞ !
n! 1 e 3
(g) = − ;
e n! 2 2e
n=1
∞ !
n! xn x ex x + 1 e−x
(h) = + + · , for x = 1;
e n! e(1 − x) 2 x−1 2
n=1
∞ 4 5
n! xn 1 x x + 1 −x
(i) =− e + e , for x ∈ [−1, 1);
e n! 2 x−1
n=1
∞ 1
(j) (−1)n {n!e} = 1 − e + ln 2 + ex ln(2 − x)dx;
n=1 0
∞
{n!e} 1 ex − 1
(k) = dx;
n 0 x
n=1
∞
1 1 ex − 1
(l) {n!e} − =2−e+ dx;
n+1 0 x
n=1
∞
{(2n − 1)! sinh 1} 1
(m) = ;
(2n − 1)! 2e
n=1
∞
{(2n)! cosh 1} sinh 1
(n) = 1 − cosh 1 + ;
(2n)! 2
n=1
∞
{(2n)! cos 1} cosh 1 cos 1 sin 1
(o) = − − ;
(2n)! 2 2 2
n=1
∞
{(2n − 1)! sin 1} cos 1 sinh 1
(p) = + .
(2n − 1)! 2 2
n=1
4.82. [59] Let k ≥ 0 be an integer. Prove that

∞
{(n + k)!e} e 1 1
= − k! e − 1 − − · · · − .
n! k+1 1! k!
n=1
4.8 Extrema of One Variable Functions
The Local Extremum Points of a One Variable Function

Let I ⊆ R be an open interval, n ∈ N, n ≥ 2, f : I → R a function n
times differentiable on I and x0 ∈ I such that
f (x0 ) = f (x0 ) = . . . = f (n−1) (x0 ) = 0, f (n) (x0 ) = 0.
(a) If n is even, then x0 is a point of local extremum of f as follows: a local

minimum point of f if f (n) (x0 ) > 0, and a local maximum point of f if
f (n) (x0 ) < 0.
(b) If n is odd, then x0 is not a local extremum point of f , it is an inflection
point.
4.83. Find:
(a) the minimum and the maximum values of the function
f : R → R, f (x) = 8 sin x + 15 cos x;
(b) the maximum value of the function f : R → R,
1
f (x) = sinh x − cosh2 x − ;
2
(c) the minimum value of the function f : (−∞, 0) → R,
π 1
f (x) = arctan2 x − arctan ;
2 x
(d) the minimum value of the function f : R → R,
f (x) = x(x + 5)(x + 10)(x + 15).
4.84. Find the minimum and the maximum values of the function f : R → R,
f (x) = x 5 − x, on the set C = { x ∈ R | x 4 + 4 ≤ 5x 2 }.
4.8 Extrema of One Variable Functions 149
4.85. Find the local extremum points and the local extremum values of the follo-
wing functions:
(a) f : R → R, f (x) = x
x 2 +1
;
x2
(b) f : R → R, f (x) = x 4 +1
;
(c) f : R → R, f (x) = 3x − 15x 4
5 − 25x 3 + 1;
(d) f : R → R, f (x) = x − 16x − 1;
6 3
: R → R, f (x) = xe−x ;
2
(e) f
: R → R, f (x) = x 2 e−x ;
2
(f) f
(g) f : R → R, f (x) = cosh (3x);
(h) f : (0, ∞) → R, f (x) = x ln2 x;
(i) f : (0, ∞) → R, f (x) = x x ;
(j) f : R → R, f (x) = ex cos x;
(k) f : R → R, f (x) = arctan xex ;
(l) f : R → R, f (x) = 2 sin x + x 2 + π x;
(m) f : R → R, f (x) = 2 sin x − x 2 − π x.
4.86. Find the distance from the point (0, 33) to the parabola x = 4y 2 .
4.87. Let b ∈ R. Find the distance from the point (0, b) to the parabola x 2 = 4y.
4.88. [A. Doboşan, 1978] Let a, b, c ∈ R, with ab < 0. If m and M are the local
minimum and the local maximum values of f : R → R, f (x) = ax 5 + bx 3 + c,
108b5
then M + m = 2c and Mm = 3125a 3 +c .
2
We record, as a gem, the proof of a conditional inequality in two variables.
A Gem with a Conditional Inequality

If x, y ≥ 0, such that x + y = 2, then x 2 + xy + y 2 ≥ 3.
Proof. Let x = 1 + α and y = 1 − α, α ∈ [0, 1]. We have
x 2 + xy + y 2 = (1 + α)2 + 1 − α 2 + (1 − α)2 = 3 + α 2 ≥ 3,
with equality when x = y = 1.

4.89. Let n ∈ N and x, y ≥ 0, with x + y = 2. Prove that x n − xy + y n ≥ 1.
4.90. Open problem. Let a, b ∈ R, a 2 + b2 = 0. Find:
• max {sin(ax) + sin(bx)};

x∈R
• max {sin(ax) + cos(bx)}.
x∈R
An Invitation to Maclaurin Inequalities

I. Let k ∈ N and let f be a function which has the Maclaurin series expansion
with radius of convergence R ∈ {1, ∞}. If f (n) (0) ≥ 0, ∀n ≥ 0, then

x 2k+2 x 2k+1 x 2k
f − 2f +f ≥ 0, ∀x ∈ (−R, R).
2k + 2 2k + 1 2k
Proof. A calculation shows that

2k
x 2k+2 x 2k+1 x
f − 2f +f
2k + 2 2k + 1 2k
∞
f (n) (0) 2kn x 2n xn 1
= x − 2 + .
n! (2k + 2) n (2k + 1)n (2k)n
n=0
2
We denote x n = t and observe that (2k+2)
t
n − (2k+1)n + (2k)n > 0, ∀t ∈ R.
2t 1
This follows since the discriminant of the quadratic function Δ < 0.
The case k = 1. If f is a function which has the Maclaurin series expansion

with radius of convergence R ∈ {1, ∞} and f (n) (0) ≥ 0, ∀n ≥ 0, then

x4 x3 x2
f − 2f +f ≥ 0, ∀x ∈ (−R, R).
4 3 2
Applications. The following inequalities hold:
x4 x3 x2
(a) e −2e 3 + e 2 ≥
4 0, ∀x ∈ R;
x4 x3 x2
(b) cosh − 2 cosh + cosh ≥ 0, ∀x ∈ R;
4 3 2
(continued)
2 3 1
(c) − + ≥ 0, ∀x ∈ [−1, 1];
4−x 4 3−x 3 2 − x2
22n−1 3n 2n−1
(d) − + ≥ 0, ∀x ∈ [−1, 1], n ∈ N;
(4 − x 4 )n (3 − x 3 )n (2 − x 2 )n
2
x4 x2 x3
(e) 1 − 1− ≤ 1− , ∀x ∈ [−1, 1];
4 2 3
2
x 2k+2 x 2k x 2k+1
(f) 1 − 1− ≤ 1− , ∀x ∈ [−1, 1], k ∈ N.
2k + 2 2k 2k + 1
II. Let f be a function which has the Maclaurin series expansion with
radius of convergence R ∈ {1, ∞}. If f (n) (0) ≥ 0, ∀n ≥ 0, then

f x 6 − 4f x 5 + 6f x 4 − 4f x 3 + f x 2 ≥ 0, ∀x ∈ (−R, R).
Proof. A calculation shows that

f x 6 − 4f x 5 + 6f x 4 − 4f x 3 + f x 2
∞
f (n) (0) 2n n 4
= x x − 1 ≥ 0,
n!
n=0
with equality when x = 0 or x = 1.
1
Applications. We consider the particular cases when f is ex , cosh x, 1−x ,
ln(1 − x), and we have that the following inequalities hold:
6 5 4 3 2
(g) [V. Brayman, 2020] ex − 4ex + 6ex − 4ex + ex ≥ 0, ∀x ∈ R;
(h) cosh(x 6 )−4 cosh(x 5 )+6 cosh(x 4 )−4 cosh(x 3 )+cosh(x 2 ) ≥ 0, ∀x ∈ R;
1 4 6 4 1
(i) − + − + ≥ 0, ∀x ∈ (−1, 1);
1−x 6 1−x 5 1−x 4 1−x 3 1 − x2
(j) (1 − x 2 )(1 − x 4 )6 (1 − x 6 ) ≤ (1 − x 3 )4 (1 − x 5 )4 , ∀x ∈ [−1, 1].
A challenge. Prove that the inequality (j) holds for all x ∈ R.
Partial Derivatives and Applications
5
If I were again beginning my studies, I would follow

the advice of Plato and start with mathematics.
Galileo Galilei (1564–1642)
This chapter collects problems on partial derivatives and their applications, the Jaco-
bian and the Hessian matrices, differential operators, the chain rule, homogeneous
functions and Euler’s identity, Taylor’s formula for functions of two variables, as
well as extrema of functions of several variables.

Differential Operators
5.1. Calculate the first order partial derivatives of the following functions:
(a) f (x, y) = sin(cos(xy 2 + 2x − y + 10));

(b) f (x, y) = ln(sin2 (xy) + 1);
(c) f (x, y, z) = ln(x y y z zx ), x, y, z > 0;
(d) f (x, y, z) = exyz cos2 (x − yz);
(e) f (x, y, z) = (xy)z , x, y > 0;
(f) f (x, y, z) = x yz , x > 0;
z
(g) f (x, y, z) = x y , x, y > 0.
5.2. Calculate the first and second order partial derivatives of the following
functions:
(a) f (x, y) = ln(xy + 1), xy + 1 > 0;

(b) f (x, y) = (x − y) cos(xy);
(c) f (x, y) = (x 2 + y 2 ) arctan xy , y = 0;
154 5 Partial Derivatives and Applications
2
(d) f (x, y, z) = ex y sin(x − z);
ye−z
(e) f (x, y, z) = xy−1 , xy − 1 = 0;
x−y
(f) f (x, y, z) = xz−1 , xz − 1 = 0;
(g) f (x, y, z) = yz−1 , yz − 1 = 0;
xz
sin(x−y)
(h) f (x, y, z) = cos(y−z) , y − z = 2
π
+ kπ , k ∈ Z.
5.3. [3, p. 225], [96, p. 98, 99], [143, p. 72] Let f : R2 → R be the function defined
by
.
xy xx 2 −y
2 2
+y 2
if (x, y) = (0, 0)
f (x, y) =
0 if (x, y) = (0, 0).
Prove that:
(a) f is of class C 1 ;
∂2f ∂2f
(b) ∂y∂x (0, 0) = −1, ∂x∂y (0, 0) = 1;
(c) f does not satisfy the conditions of Schwarz’s Theorem.
5.4. [143, p. 73] Let f : R2 → R be the function defined by

.
x 2 arctan yx − y 2 arctan xy if xy = 0
f (x, y) =
0 if xy = 0.
Prove that:
(a) f is of class C 1 ;
∂2f ∂2f
(b) ∂y∂x (0, 0) = −1, ∂x∂y (0, 0) = 1;
(c) f does not satisfy the conditions of Schwarz’s Theorem.
2 3 x
5.5. Let A = (x, y) ∈ R2 | y = 0 and f : A → R, f (x, y) = xy − xe y . Prove
that, for all (x, y) ∈ A, the following equalities hold:
∂f ∂f
(a) x (x, y) + y (x, y) = xy + f (x, y);
∂x ∂y
∂ 2f ∂ 2f ∂ 2f
(b) x 2 2 (x, y) + 2xy (x, y) + y 2 2 (x, y) = 2xy.
∂x ∂y∂x ∂y
2 3 x − yx
5.6. Let A = (x, y) ∈ R2 | xy = 0 and f : A → R, f (x, y) = y3
e .
Calculate L(f ) = x ∂f ∂f
∂x + y ∂y and L(L(f )).
5.1 Partial Derivatives, the Jacobian and the Hessian Matrices, Differential. . . 155
xy−z
5.7. Let A = {(x, y, z) ∈ R3 | x − yz = 0} and f : A → R, f (x, y, z) = x−yz .
Prove that, for all (x, y, z) ∈ A, the following equality holds:
∂ 2f
(x − yz)2 (x, y, z) − 2xf (x, y, z) = 0.
∂z∂y
2 3
5.8. Let A = (x, y, z) ∈ R3 | x 3 + y 3 + z3 − 3xyz > 0 and f, g, h : A → R be
the functions defined by
f (x, y, z) = ln(x 3 + y 3 + z3 − 3xyz),

g(x, y, z) = f 2 (x, y, z),
h(x, y, z) = f 3 (x, y, z).
Prove that, for all (x, y, z) ∈ A, the following equalities hold:
(a) fx (x, y, z) + fy (x, y, z) + fz (x, y, z) = x+y+z 3

;
(b) fx (x, y, z) + fy (x, y, z) + fz (x, y, z) − 3fx (x, y, z)fy (x, y, z)fz (x, y, z)
3 3 3
= x 3 +y 3 +z
27
3 −3xyz ;
(c) fx 3 (x, y, z) + fy 3 (x, y, z) + fz3 (x, y, z) − 3fxyz (x, y, z) = 0;

(d) gx 3 (x, y, z) + gy 3 (x, y, z) + gz3 (x, y, z) − 3gxyz (x, y, z) = 0;
(e) hx 3 (x, y, z) + hy 3 (x, y, z) + hz3 (x, y, z) − 3hxyz (x, y, z) = 162
x 3 +y 3 +z3 −3xyz
.
Remark 5.1. If n ∈ N and ϕ : A → R, ϕ(x, y, z) = f n (x, y, z), then
27n(n − 1)(n − 2)f n−3 (x, y, z)

ϕx 3 (x, y, z)+ϕy 3 (x, y, z)+ϕz3 (x, y, z)−3ϕxyz (x, y, z)= .
x 3 + y 3 + z3 − 3xyz
∂ 2n f
5.9. Let n ≥ 0 and f : R2 → R, f (x, y) = (x n +y n )ex+y . Calculate ∂y n ∂x n (x, y).
xy
5.10. Let n ≥ 0 and f : R2 → R, f (x, y) = (1+x 2 )2 (1+y 2 )2
. Calculate
∂ 2n f
∂y n ∂x n (0, 0).
5.11. Let A = {(x, y) ∈ R2 | xy + 1 = 0} and f : A → R, f (x, y) = 1

xy+1 .
(a) Prove that
∂ 2n f
(0, 0) = (−1)n (n!)2 , n ≥ 0.
∂y n ∂x n
(b) Let A = {(x, y, z) ∈ R2 | xyz + 1 = 0} and f : A → R, f (x, y, z) = 1

xyz+1 .
Prove that
∂ 3n f
(0, 0, 0) = (−1)n (n!)3 , n ≥ 0.
∂zn ∂y n ∂x n
Remark 5.2. Let k ∈ N, k ≥ 2, A = {(x1 , x2 , . . . , xk ) ∈ Rk | x1 x2 · · · xk + 1 = 0},

and
1
f : A → R, f (x1 , x2 , . . . , xk ) = .
x1 x2 · · · xk + 1
The following equality holds
∂ kn f
n · · · ∂x n (0, 0, . . . , 0) = (−1)n (n!)k , n ≥ 0.
∂xkn ∂xk−1 1
5.12. Let f, g : R2 → R, f (x, y) = ex+y sin(x + y), g(x, y) = ex+y cos(x + y).
Prove that, for all n ≥ 0 and (x, y) ∈ R2 , we have:
2 n 2 n 2 n 2
∂nf
(a) ∂x n (x, y)+ ∂∂yfn (x, y) + ∂∂xgn (x, y) + ∂∂ygn (x, y) = 2n+1 e2(x+y) ;
2n 2 2n 2
(b) ∂y∂ n ∂xf n (x, y) + ∂y∂n ∂xg n (x, y) = 22n e2(x+y) .
Let A ⊆ Rn , a ∈ int (A), and f : A → Rm , f = (f1 , f2 , . . . , fm ) be

a function that has first order partial derivatives at a. The following matrix
having m rows and n columns
⎛ ⎞
∂f1 ∂f1 ∂f1
(a) (a) . . . (a)
⎜ ∂x1 ∂x2 ∂xn ⎟
⎜ ∂f ⎟
⎜ 2 (a) ∂f2 ∂f2
(a) ⎟
⎜ (a) . . . ⎟
J (f )(a) = ⎜ ∂x1 ∂x2 ∂xn ⎟
⎜ ⎟
⎜ ... ... ... ... ⎟
⎝ ∂fm ∂fm ∂fm ⎠
(a) (a) . . . (a)
∂x1 ∂x2 ∂xn
is called the Jacobian matrix of f at a.
If m = n, then the determinant of the Jacobian matrix of f is called
functional determinant or the Jacobian
/ /
/ ∂f1 ∂f1 ∂f1 /
/ (a) (a) . . . (a)/
/ ∂x1 ∂x2 ∂xn //
/ ∂f
/ 2 (a) ∂f2 ∂f2
(a)//
D(f1 , f2 , . . . , fn ) / (a) . . .
(a) = / ∂x1 ∂x2 ∂xn / .
D(x1 , x2 , . . . , xn ) / /
/ ... ... ... ... /
/ ∂fn /
/ ∂fn ∂fn
(a)//
/ (a) (a) . . .
∂x1 ∂x2 ∂xn
(continued)
Let A ⊆ Rn , a ∈ int (A), and f : A → R be a function that has second

order partial derivatives at a. The following square matrix of order n
⎛ ⎞
∂ 2f ∂ 2f ∂ 2f
⎜ ∂x 2 (a) ∂x ∂x (a) . . . ∂x ∂x (a)⎟
⎜ 2 1 n 1 ⎟
⎜ ∂ 2 f1 ∂ 2f ∂ 2f ⎟
⎜ ⎟
⎜ (a) (a) . . . (a) ⎟
H (f )(a) = ⎜ ∂x1 ∂x2 ∂x2 2 ∂x ∂x ⎟
⎜ n 2
⎟
⎜ ⎟
⎜ 2. . . ... ... ... ⎟
⎝ ∂ f 2
∂ f 2
∂ f ⎠
(a) (a) . . . (a)
∂x1 ∂xn ∂x2 ∂xn ∂xn2
is called the Hessian matrix of f at a.
5.13. Let f : R2 → R3 , f (x, y) = (x − cos y, arctan(xy), sin(π x − y)).

Determine the matrix J (f )(1, 0).

5.14. Let f : R3 → R2 , f (x, y, z) = ln exy−z + 1 , (x 2 + 1)yz . Determine the
matrix J (f )(−1, 1, −1).
D(x,y)
5.15. Polar coordinates in plane. Calculate D(ρ,θ) if
x = ρ cos θ, y = ρ sin θ.
D(x,y,z)
5.16. Polar coordinates in space. Calculate D(ρ,ϕ,θ) if
x = ρ sin ϕ cos θ, y = ρ sin ϕ sin θ, z = ρ cos ϕ.
5.17. Calculate D(x,y)

D(u,v) if x = √ u
, y=√ v
, u2 + v 2 < 1.
1−u2 −v 2 1−u2 −v 2
D(x,y,z)
5.18. Calculate D(u,v,w) if
x = u2 + v 2 + w 2 , y = u3 + v 3 + w 3 , z = u4 + v 4 + w 4 .
5.19. Let f : (0, ∞) × R → R, f (x, y) = x y . Calculate H (f )(1, −1).

2 3 y−z
5.20. Let A = (x, y, z) ∈ R3 | z − x = 0 and f : A → R, f (x, y, z) = z−x .
Calculate H (f )(1, −1, 0).
Let ∅ = A ⊆ R3 be an open set. We consider the following sets of functions:

F = {f | f : A → R}
F1 = { f | f : A → R having first order partial derivatives on A }
F2 = { f | f : A → R having second order partial derivatives on A }
V = { F | F : A → R3 }
V1 = { F | F = (P , Q, R) : A → R3 , with P , Q, R having first order partial
derivatives on A}.
We define the following differential operators:

∂f ∂f ∂f
the gradient ∇ : F1 → V , ∇f = , , ;
∂x ∂y ∂z
∂P ∂Q ∂R
the divergence div : V1 → F , div F = + + , F = (P , Q, R);
∂x ∂y ∂z

∂R ∂Q ∂P ∂R ∂Q ∂P
the rotation rot : V1 → V , rot F = − , − , − ;
∂y ∂z ∂z ∂x ∂x ∂y
∂ 2f ∂ 2f ∂ 2f
the Laplacian Δ : F2 → F , Δf = + + .
∂x 2 ∂y 2 ∂z2
The rotation is also known as the curl.
Remark 5.3. Let A ⊆ Rn , a ∈ int (A), and f : A → R having first order

partial derivatives at a. The gradient of the function f at a is defined by

∂f ∂f ∂f
∇f (a) = (a), (a), . . . , (a) .
∂x1 ∂x2 ∂xn
If f has second order partial derivatives at a, then
H (f )(a) = J (∇f )(a).
(continued)
5.2 The Chain Rule 159
Also, the Laplacian of f at a is given by
∂ 2f ∂ 2f ∂ 2f
Δf (a) = (a) + (a) + · · · + (a).
∂x12 ∂x22 ∂xn2
5.21. Let f : R3 → R, f (x, y, z) = arctan(xyz). Calculate ∇f and Δf =

div(∇f ).
5.22. Let f : R3 \ {(0, 0, 0)} → R, f (x, y, z) = ln √ 1

. Calculate ∇f and
x 2 +y 2 +z2
Δf = div(∇f ).

5.23. Let f ∈ C 2 R3 , R . Prove that rot(∇f ) = 0.
5.24. Let f : R3 → R be the function defined by
1 3 3
f (x, y, z) = (x + y 3 + z3 ) + (x 2 + y 2 + z2 ) + 9(x + y + z).
3 2
Find all points (x, y, z) ∈ R3 such that ∇f (x, y, z) = (y 3 , z3 , x 3 ).
5.25. Let f : R3 → R, f (x, y, z) = 12 (x +y +z)2 +xyz. Find all points (x, y, z) ∈

(−1, ∞)3 such that ∇f (x, y, z) = (x + 3, y + 8, z + 15).
5.26. Let F : R3 → R3 , F (x, y, z) = (x − y + z, xy − ez , arctan(x − yz)).

Calculate div F and rot F .
5.27. Let F : R3 → R3 , F (x, y, z) = (x cos(yz), yexz , sin(xyz)). Calculate

div F and rot F .
5.2 The Chain Rule
The Chain Rule

I. Let ∅ = A ⊆ Rn be an open set, J ⊆ R an open interval, f ∈ C 1 (A, R)
and ui : J → R, i ∈ {1, . . . , n}, derivable functions on J such that
(u1 (x), . . . , un (x)) ∈ A, for all x ∈ J . Then the function F : J → R,
F (x) = f (u1 (x), . . . , un (x)) is derivable on J and we have
(continued)
n
F (x) = fui (u1 (x), . . . , un (x)) · ui (x), x ∈ J.
i=1
Case n = 2 Let ∅ = A ⊆ R2 be an open set, J ⊆ R an open interval, f ∈

C 1 (A, R) and u, v : J → R derivable functions on J such that (u(x), v(x)) ∈
A, for all x ∈ J . Then the function F : J → R, F (x) = f (u(x), v(x)) is
derivable on J and we have
F (x) = fu (u(x), v(x)) · u (x) + fv (u(x), v(x)) · v (x), x ∈ J.
II. Let I ⊆ R be an open interval, ∅ = B ⊆ Rm an open set, f : I → R a

derivable function on I , and u : B → R a function having first order partial
derivatives on B such that u(x1 , . . . , xm ) ∈ I , for all (x1 , . . . , xm ) ∈ B. Then
the function F : B → R, F (x1 , . . . , xm ) = f (u(x1 , . . . , xm )) has first order
partial derivatives on B and we have, for j ∈ {1, . . . , m},
Fxj (x1 , . . . , xm ) = f (u(x1 , . . . , xm )) · uxj (x1 , . . . , xm ), (x1 , . . . , xm ) ∈ B.
Case m = 2 Let I ⊆ R be an open interval, ∅ = B ⊆ R2 an open set,

f : I → R a derivable function on I , and u : B → R a function having
first order partial derivatives on B such that u(x, y) ∈ I , for all (x, y) ∈ B.
Then the function F : B → R, F (x, y) = f (u(x, y)) has first order partial
derivatives on B and we have
Fx (x, y) = f (u(x, y)) · ux (x, y), (x, y) ∈ B,

Fy (x, y) = f (u(x, y)) · uy (x, y), (x, y) ∈ B.
III. Let ∅ = A ⊆ Rn and ∅ = B ⊆ Rm be open sets, f ∈ C 1 (A, R) and

u1 , . . . , un : B → R functions having first order partial derivatives on B such
that (u1 (x), . . . , un (x)) ∈ A, for all x = (x1 , . . . , xm ) ∈ B. Then the function
F : B → R, F (x) = f (u1 (x), . . . , un (x)) has first order partial derivatives
on B and we have, for j ∈ {1, . . . , m},
n
Fxj (x) = fui (u1 (x), . . . , un (x)) · (ui )xj (x), x = (x1 , . . . , xm ) ∈ B.
i=1
Case m = n = 2 Let ∅ = A, B ⊆ R2 be open sets, f ∈ C 1 (A, R) and

u, v : B → R functions having first order partial derivatives on B such that
(u(x, y), v(x, y)) ∈ A, for all (x, y) ∈ B. Then the function F : B → R,
(continued)
F (x, y) = f (u(x, y), v(x, y)) has first order partial derivatives on B and we
have
Fx (x, y) = fu (u(x, y), v(x, y)) · ux (x, y) + fv (u(x, y), v(x, y)) · vx (x, y),
Fy (x, y) = fu (u(x, y), v(x, y)) · uy (x, y) + fv (u(x, y), v(x, y)) · vy (x, y),
for all (x, y) ∈ B.
Now we give some examples of how to apply the chain rule.
1. Calculate F if F (x) = f (2x − 1, −x), where f = f (u, v) is a function of

class C 3 .
Let u(x) = 2x − 1 and v(x) = −x. Then F (x) = f (u(x), v(x)). We have
F (x) = 2fu (u(x), v(x)) − fv (u(x), v(x)).
We calculate
F (x) = 2[fu2 (u(x), v(x)) · 2 + fuv (u(x), v(x)) · (−1)]

−[fvu (u(x), v(x)) · 2 + fv 2 (u(x), v(x)) · (−1)]
= 4fu2 (u(x), v(x)) − 4fuv (u(x), v(x)) + fv 2 (u(x), v(x)).
We derive F and we obtain
F (x) = 4[fu3 (u(x), v(x)) · 2 + fu2 v (u(x), v(x)) · (−1)]

−4[fuvu (u(x), v(x)) · 2 + fuv 2 (u(x), v(x)) · (−1)]
+[fv 2 u (u(x), v(x)) · 2 + fv 3 (u(x), v(x)) · (−1)]
= 8fu3 (u(x), v(x)) − 12fu2 v (u(x), v(x))
+6fuv 2 (u(x), v(x)) − fv 3 (u(x), v(x)).
2. Calculate Fx 2 y , Fxyz , Fy 2 z , and Fyz2 if F (x, y, z) = f (x − yz), where f = f (u)

is a three times derivable function.
We denote u(x, y, z) = x − yz and we have F (x, y, z) = f (u(x, y, z)).
To determine Fx 2 y , first we calculate Fx (x, y, z) = f (u(x, y, z)) and then
Fx 2 (x, y, z) = f (u(x, y, z)), and taking the derivative of Fx 2 with respect to y
we obtain that
Fx 2 y (x, y, z) = −zf (u(x, y, z)).

For calculating Fxyz we take the derivative of Fx with respect to y and we have
that
Fxy (x, y, z) = −zf (u(x, y, z)),
and by taking the derivative of the preceding equality with respect to z we obtain
that
Fxyz (x, y, z) = −f (u(x, y, z)) − zf (u(x, y, z)) · (−y)

= yzf (u(x, y, z)) − f (u(x, y, z)).
For determining Fy 2 z we calculate one at a time:
Fy (x, y, z) = −zf (u(x, y, z)),
Fy 2 (x, y, z) = (Fy )y (x, y, z) = −zf (u(x, y, z)) · (−z) = z2 f (u(x, y, z)),
Fy 2 z (x, y, z) = (Fy 2 )z (x, y, z) = 2zf (u(x, y, z)) + z2 f (u(x, y, z)) · (−y)
= −yz2 f (u(x, y, z)) + 2zf (u(x, y, z)).
Now we calculate Fyz2 and we have
Fyz (x, y, z) = (Fy )z (x, y, z) = −f (u(x, y, z)) − zf (u(x, y, z)) · (−y)

= yzf (u(x, y, z)) − f (u(x, y, z)),
Fyz2 (x, y, z) = (Fyz )z (x, y, z)
= yf (u(x, y, z)) + yzf (u(x, y, z)) · (−y) − f (u(x, y, z)) · (−y)

= −y 2 zf (u(x, y, z)) + 2yf (u(x, y, z)).
IV , if F (x, y, z, t) = f (x − t, yz, y − t), where f = f (u, v, w) is

3. Calculate Fxyzt
a function of class C 4 .
Let u(x, y, z, t) = x − t, v(x, y, z, t) = yz and w(x, y, z, t) = y − t. Then
we have
F (x, y, z, t) = f (u(x, y, z, t), v(x, y, z, t), w(x, y, z, t)).
First we calculate
Fx (x, y, z, t) = fu (u(x, y, z, t), v(x, y, z, t), w(x, y, z, t)).
Next, for simplifying the writing, we shall drop out the argument of the functions
and we have
Fx = fu ,
Fxy = (Fx )y = fuv · z + fuw = zfuv + fuw ,
Fxyz = (Fxy )z = fuv + zfuv 2 · y + fuwv · y = yzfuv 2 + yfuvw + fuv ,
IV
Fxyzt = (Fxyz )t
= yz[fuv
IV
2 u · (−1) + fuv 2 w · (−1)] + y[fuvwu · (−1) + fuvw 2 · (−1)]
IV IV IV
+[fuvu · (−1) + fuvw · (−1)]

= −yzfuIV2 v 2 − yfuIV2 vw − yzfuv
IV
2 w − yfuvw 2 − fu2 v − fuvw .
IV
5.28. Calculate F and F if:
(a) F (x) = f (sin x, arctan x);

(b) F (x) = f (ln(x 2 + 1), cos x),
where f = f (u, v) is a function of class C 2 .
5.29. Calculate F if:
(a) F (x) = f (−x, 3x);

(b) F (x) = f (x, x 2 ),
5.30. Calculate F and F if:

√
(a) F (x) = f (ex , ln(x + x 2 + 1), x 2 );
2
(b) F (x) = f (2x , arccot x, ex ),
where f = f (u, v, w) is a function of class C 2 .
5.31. Calculate Fx 2 , Fxy , and Fy 2 if:
(a) F (x, y) = f (x 2 − 2xy + 3y);

(b) F (x, y) = f (x sin y − y cos x),
where f = f (u) is a twice derivable function.
5.32. Calculate the derivatives written to the right hand side of the function:
(a) F (x, y, z) = f (x − y − z), Fxyz , Fyz2 ;

(b) F (x, y, z) = f (xyz), Fx 3 , Fx 2 z ;
(c) F (x, y, z) = f (x 2 − xy + z), Fxy 2 , Fxyz ;

(d) F (x, y, z, t) = f (xyt − z + t), IV ,
Fxyt , Fxzt , Fxt 2 , Fyzt , Fxyzt
where f = f (u) is a three times derivable function for parts (a), (b), and (c), and a
four times derivable function for part (d).
5.33. Calculate Fx 2 , Fxy , and Fy 2 if:
(a) F (x, y) = f (xy − x + y, x arctan y);

(b) F (x, y) = f (sin(x − y), xy − ln(y 2 + 1)),
(a) F (x, y, z) = f (x − yz, x − y), Fy 2 , Fyz , Fxyz ;

(b) F (x, y, z) = f (x − z, y − z), Fxz , Fyz , Fxyz , Fxz2 ;
(c) F (x, y, z) = f (xy − z, xz − y), Fxy , Fyz , Fxyz , Fy 2 z ;
(d) F (x, y, z, t) = f (x − t, yz − t), IV ,
Fyt , Fzt , Fxyz , Fxyt , Fxzt , Fxt 2 , Fxyzt
where f = f (u, v) is a function of class C 3 for parts (a), (b), and (c), and of class
C 4 for part (d).
(a) F (x, y, z) = f (x − y, y − z, xy − z), Fxy , Fxz , Fyz , Fxyz , Fyz2 ;

(b) F (x, y, z) = f (y, x − z, x − yz), Fxy , Fxz , Fyz , Fz2 , Fxy 2 , Fxyz , Fy 2 z ;
(c) F (x, y, z) = f (z, x − y − z, x), Fxy , Fxz , Fyz , Fz2 , Fxyz , Fxz2 , Fxy IV ;
3
(d) F (x, y, z, t) = f (xt, y − z, x − t), Fxy , Fyt , Fzt , Ft 2 , Fxyz , Fxyt , Fyzt ,
IV ,
Ft 3 , Fxyzt
where f = f (u, v, w) is a function of class C 3 for parts (a) and (b), and of class
C 4 for parts (c) and (d).
5.36. Let a, b ∈ R, f : R → R be a nonconstant twice derivable function on R

and g : R2 → R be the function defined by g(x, y) = ef (ax+by) . Find the relation
that has to be satisfied by the numbers a and b so that, for all (x, y) ∈ R2 , to have
∂ 2g ∂ 2g ∂ 2g
(x, y) − 2 (x, y) + (x, y) = 0.
∂x 2 ∂y∂x ∂y 2
5.3 Homogeneous Functions. Euler’s Identity 165
5.37. Let a, b, x0 ∈ R, f : R2 → R be a function of class C 2 , and g : R → R

be the function defined by g(x) = ef (ax,bx) . Knowing that the second order partial
derivatives of f at (ax0 , bx0 ) are equal to each other to a nonzero real number, find
the relation that has to be satisfied by the numbers a and b such that g (x0 )g(x0 ) =
g 2 (x0 ).
5.3 Homogeneous Functions. Euler’s Identity
Let α ∈ R and A ⊆ Rn be a cone, i.e. a set with the property that, for all
x ∈ A and t ∈ (0, ∞), we have tx ∈ A. A function f : A → R is called
homogeneous of degree α if
f (tx) = t α f (x), for all x ∈ A and t ∈ (0, ∞).
Euler’s identity. Let α ∈ R, A ⊆ Rn be an open set, which is also a cone, and

let f ∈ C 1 (A, R). The function f is homogeneous of degree α if and only if
n
∂f
xi (x1 , . . . , xn ) = αf (x1 , . . . , xn ), for all (x1 , . . . , xn ) ∈ A.
∂xi
i=1
Euler’s identity can be written using the gradient of the function f as follows:
x · ∇f (x) = αf (x), for all x ∈ A,
where “·” denotes the inner product of vectors in Rn .
5.38. Prove that the following functions are homogeneous in two different ways
(using the definition and Euler’s identity):
(a) f (x, y, z) = √ 1
x+y+z
, x + y + z > 0;
x+y+z
(b) f (x, y, z) = xy+yz+zx , xy + yz + zx = 0;
yz xz xy
(c) f (x, y, z) = xe + ye x2 y2 + ze , x, y, z = 0;
z2
(d) f (x, y, z) = x12 sin yz + 1

y2
sin xz + z12 sin xy , x, y, z = 0;
xy+z2
(e) f (x, y, z) = x−y , x − y = 0.
5.39. Let α ∈ R, A ⊆ R2 be an open set, which is also a cone, and f ∈ C 3 (A, R)

be a homogeneous function of degree α. Prove that:
(a) ∂f ∂f
∂x and ∂y are homogeneous functions of degree α − 1.
(b) The following equality holds:
∂ 2f ∂ 2f ∂ 2f
x2 2
(x, y) + 2xy (x, y) + y 2 2 (x, y) = α(α − 1)f (x, y),
∂x ∂y∂x ∂y
for all (x, y) ∈ A.

(c) The following equality holds:
∂ 3f ∂ 3f 3
2 ∂ f
3
3∂ f
x3 (x, y) + 3x 2
y (x, y) + 3xy (x, y) + y (x, y)
∂x 3 ∂y∂x 2 ∂y 2 ∂x ∂y 3
= α(α − 1)(α − 2)f (x, y),

Remark 5.4. More generally, if n ∈ N and f ∈ C n (A, R) is a homogeneous

function of degree α, then the following equality holds:

∂ ∂ (n)
x +y f (x, y) = α(α − 1) · · · (α − n + 1)f (x, y),
∂x ∂y
for all (x, y) ∈ A, where “(n)” means “symbolic power (symbolic order of
derivability)”, i.e. we have
n
∂ nf
Ckn x n−k y k (x, y) = α(α − 1) · · · (α − n + 1)f (x, y),
∂y k ∂x n−k
k=0
5.4 Taylor’s Formula for Real Functions of Two Real Variables
Taylor’s Formula for Real Functions of Two Real Variables

Let ∅ = A ⊆ R2 be a convex open set and f ∈ C n+1 (A, R). Then
∀(x0 , y0 ), (x, y) ∈ A, ∃(θ1 , θ2 ) ∈ (0, 1)2 such that Taylor’s formula of order
n corresponding to the function f and the point (x0 , y0 ) holds true
(continued)
5.4 Taylor’s Formula for Real Functions of Two Real Variables 167
f (x, y) = (Tn f )(x, y) + (Rn f )(x, y),
where
n
1 ∂ ∂ (k)
(Tn f )(x, y) = f (x0 , y0 ) + (x − x0 ) + (y − y0 ) f (x0 , y0 )
k! ∂x ∂y
k=1
is the Taylor polynomial of degree n corresponding to the function f and the

point (x0 , y0 ), and

1 ∂ ∂ (n+1)
(Rn f )(x, y) = (x − x0 ) + (y − y0 ) f (c1 , c2 ),
(n + 1)! ∂x ∂y
with (c1 , c2 ) = (x0 + θ1 (x − x0 ), y0 + θ2 (y − y0 )), is the remainder of order

n corresponding to the function f and the point (x0 , y0 ) (the remainder in the
Lagrange form).
Here “(k)” means “symbolic power (symbolic order of derivability)”, i.e. we
have

∂ ∂ (k)
(x − x0 ) + (y − y0 ) f (x0 , y0 )
∂x ∂y
k
∂kf
= Cik (x0 , y0 )(x − x0 )k−i (y − y0 )i .
∂y i ∂x k−i
i=0
For example, the Taylor polynomial of degree 3 corresponding to the

function f and the point (x0 , y0 ) is given by
(T3 f )(x, y)

1 ∂f ∂f
= f (x0 , y0 ) + (x0 , y0 )(x − x0 ) + (x0 , y0 )(y − y0 )
1! ∂x ∂y

1 ∂ 2f ∂ 2f
+ (x0 , y0 )(x − x 0 ) 2
+ 2 (x0 , y0 )(x − x0 )(y − y0 )
2! ∂x 2 ∂y∂x
∂ 2f
+ (x0 , y0 )(y − y0 )2
∂y 2

1 ∂ 3f ∂ 3f
+ (x0 , y0 )(x − x0 ) 3
+ 3 (x0 , y0 )(x − x0 )2 (y − y0 )
3! ∂x 3 ∂y∂x 2
(continued)
∂ 3f ∂ 3f
+3 2
(x0 , y0 )(x − x0 )(y − y0 )2 + 3 (x0 , y0 )(y − y0 )3 .
∂y ∂x ∂y
We mention the following three particular cases of Taylor’s formula.
I. When (x0 , y0 ) = (0, 0) ∈ A, then we obtain the Maclaurin formula of

order n corresponding to the function f
f (x, y) = (Mn f )(x, y) + (Rn f )(x, y),
where
n
1 ∂ ∂ (k)
(Mn f )(x, y) = f (0, 0) + x +y f (0, 0)
k! ∂x ∂y
k=1
is the Maclaurin polynomial of degree n corresponding to the function f ,

and

1 ∂ ∂ (n+1)
(Rn f )(x, y) = x +y f (θ1 x, θ2 y)
(n + 1)! ∂x ∂y
is the remainder of order n corresponding to the function f .

II. When f : R2 → R is a polynomial function of degree n and (x0 , y0 ) ∈
R2 , then we have f (x, y) = (Tn f )(x, y), for all (x, y) ∈ R2 .
III. When n = 0, we obtain the Lagrange Mean Value Theorem for a real
function of two real variables
∂f ∂f
f (x, y) = f (x0 , y0 ) + (c1 , c2 )(x − x0 ) + (c1 , c2 )(y − y0 ),
∂x ∂y
where (c1 , c2 ) = (x0 + θ1 (x − x0 ), y0 + θ2 (y − y0 )).
The extension of the previous formulae to functions of more than two variables
is left to the interested reader.
5.40. Write the following functions in terms of powers of x and y, without

expanding the parenthesis:
(a) f : R2 → R, f (x, y) = (x + 2)(y − 1)2 ;

(b) f : R2 → R, f (x, y) = (x − 1)2 (y + 1).
5.4 Taylor’s Formula for Real Functions of Two Real Variables 169
5.41. Write in terms of powers of x − 1 and y + 1 the following function
f : R2 → R, f (x, y) = x 3 + x 2 y + x 2 − 3xy − 3x + 2y + 8.
5.42. Write in terms of powers of x + 1 and y − 2 the following function
f : R2 → R, f (x, y) = x 2 y + 4y 3 − 2x 2 + 4xy − 20y 2 − 8x + 35y − 14.
5.43. Find the Maclaurin polynomial of degree n corresponding to the function f

in the following cases:
(a) f : R2 → R, f (x, y) = sin(xy), n = 2;

(b) f : R2 → R, f (x, y) = ex cos y, n = 3;
(c) f : R2 → R, f (x, y) = exy , n = 4.
5.44. Find the Taylor polynomial of degree 2 corresponding to the function f and
the point (x0 , y0 ), if:
(a) f : R2 → R, f (x, y) = ex +xy , (x0 , y0 ) = (1, −1);

2
(b) f : R2 → R, f (x, y) = cos(xy − 1), (x0 , y0 ) = (1, 1).

√ √
5.45. Using the Taylor polynomial of degree 2 approximate 1, 01 · 3
0, 97.
5.46. Using the Taylor polynomial of degree 2 approximate 0, 983,01 .
5.47. Let n ≥ 0 be an integer. Find the coefficient of x n y n in the Maclaurin

polynomial of degree 2n corresponding to the function
3
f (x, y) = .
(3x + 1)(5y + 3)
5.48. Let n ≥ 0 be an integer. Find the coefficient of (x − 1)n (y − 1)n in the Taylor
2n+1
polynomial of degree 2n corresponding to the function f (x, y) = y n2(xy+1) and the
point (1, 1).
5.49. Find the coefficient of x n y n in the Maclaurin series expansion of the function
1
f (x, y) = .
(1 − x)(1 − x − y)
5.5 The Differential of Several Real Variable Functions
Let A ⊆ Rn , x 0 ∈ int (A), and f : A → R a function k times differentiable

at x 0 .
The differential of order k of the function f at x 0 is defined by
dk f (x 0 ) : Rn → R,

∂ ∂ (k)
dk f (x 0 )(dx) = dx1 + · · · + dxn f (x 0 ), dx = (dx1 , . . . , dxn ) ∈ Rn ,
∂x1 ∂xn
where “(k)” means “symbolic power (symbolic order of derivability).”

We mention the differential expressions of orders one and two of a function
of two and three real variables.
Let A ⊆ R2 , (x0 , y0 ) ∈ int (A), and f : A → R a function twice
differentiable at (x0 , y0 ).
The differential of order one of the function f at (x0 , y0 )
∂f ∂f
d f (x0 , y0 ) : R2 → R, d f (x0 , y0 )(dx, dy) = (x0 , y0 ) dx+ (x0 , y0 ) dy.
∂x ∂y
The differential of order two of the function f at (x0 , y0 )

d2 f (x0 , y0 ) : R2 → R,
∂ 2f ∂ 2f ∂ 2f
d2 f (x0 , y0 )(dx, dy) = (x 0 , y 0 ) dx 2
+2 (x 0 , y 0 ) dx dy+ (x0 , y0 ) dy 2 .
∂x 2 ∂y∂x ∂y 2
Let A ⊆ R3 , (x0 , y0 , z0 ) ∈ int (A), and f : A → R a function twice

differentiable at (x0 , y0 , z0 ).
The differential of order one of the function f at (x0 , y0 , z0 )
d f (x0 , y0 , z0 ) : R3 → R,
∂f ∂f ∂f
d f (x0 , y0 , z0 )(dx, dy, dz) = (x0 , y0 , z0 ) dx+ (x0 , y0 , z0 ) dy+ (x0 , y0 , z0 ) dz.
∂x ∂y ∂z
The differential of order two of the function f at (x0 , y0 , z0 )

d2 f (x0 , y0 , z0 ) : R3 → R,
∂ 2f ∂ 2f
d2 f (x0 , y0 , z0 )(dx, dy, dz) = (x 0 , y 0 , z0 ) dx 2
+ (x0 , y0 , z0 ) dy 2
∂x 2 ∂y 2
(continued)
5.5 The Differential of Several Real Variable Functions 171
∂ 2f ∂ 2f
+ 2
(x0 , y0 , z0 ) dz2 + 2 (x0 , y0 , z0 ) dx dy
∂z ∂y∂x
∂ 2f ∂ 2f
+2 (x0 , y0 , z0 ) dy dz + 2 (x0 , y0 , z0 ) dx dz.
∂z∂y ∂z∂x
2 +y
5.50. Let f : R2 → R, f (x, y) = ex + sin(xy 2 ). Calculate
df (0, 1)(1, 2) and d2 f (0, 1)(1, 2).

2 −y 2
5.51. Let f : R2 → R, f (x, y) = arctan(xy) + ex . Calculate
df (1, 0)(2, 1) and d2 f (1, 0)(2, 1).
5.52. Let f : R3 → R, f (x, y, z) = sin(xyz). Calculate
df (π, −1, 1)(−1, 2, 1) and d2 f (π, −1, 1)(−1, 2, 1).
5.53. Let f : R3 → R, f (x, y, z) = cos(xy + yz + xz). Calculate

π π
df , 1, 0 (1, −2, 2) and d2 f , 1, 0 (1, −2, 2).
2 2
5.54. Let f : (0, ∞)3 → R, f (x, y, z) = ln(x x y y zz ). Calculate
d4 f (2, 1, 2)(−1, 1, 1).
Let A ⊆ Rn , x 0 ∈ int (A), and f : A → R be a differentiable

function at x 0 . Then the function f is derivable at x 0 by the direction
v ∈ Rn \ {(0, . . . , 0)} and this derivative is equal to the differential of f
at x 0 evaluated at the unit vector of the direction, i.e.

df 0 v
(x ) = df (x )0
.
dv v
5.55. Let f : R2 → R be a differentiable function and let (x0 , y0 ) ∈ R2 . Knowing

that the derivative of f at (x0 , y0 ) by the direction of a vector parallel to the bisector
of the first quadrant is −5, and the derivative of f at (x0 , y0 ) by the direction of
a vector parallel to the bisector of the second quadrant is 9, calculate ∂f ∂x (x0 , y0 ) ·
∂f
∂y (x0 , y0 ).
5.56. Let f : R3 → R, f (x, y, z) = y 2x+1 + z2y+1 + x 2z+1 . Calculate the derivatives

of f at (1, 1, −1) and (−1, 1, −1) by the direction of a vector parallel to the normal
to the plane x − 2y + 2z + 3 = 0.
5.6 Extrema of Several Real Variable Functions
Local Extremum Points of Several Real Variable Functions

Let A ⊆ Rn be an open set, f ∈ C 2 (A, R), and x 0 ∈ A a critical point of f .
(a) If d2 f (x 0 ) is positive definite, then x 0 is a point of local minimum of f ,

and if d2 f (x 0 ) is negative definite, then x 0 is a point of local maximum of
f.
(b) If d2 f (x 0 ) is indefinite, then x 0 is not a point of local extremum of f , it
is a saddle point.
Remark 5.5. The quadratic form d2 f (x 0 ) : Rn → R is:
(i) positive definite if d2 f (x 0 )(h) > 0, for all h ∈ Rn \ {(0, . . . , 0)};

(ii) negative definite if d2 f (x 0 )(h) < 0, for all h ∈ Rn \ {(0, . . . , 0)};
(iii) indefinite if there exist h1 , h2 ∈ Rn such that
d2 f (x 0 )(h1 ) · d2 f (x 0 )(h2 ) < 0.
Let
∂ 2f 0
d1 = (x )
∂x12
/ /
/ ∂ 2f ∂ 2f /
/ 0 0 /
/ (x ) (x )/
/ ∂x 2 ∂x2 ∂x1 /
d2 = / 2 1 /
/ ∂ f ∂ 2f
0 //
/ (x 0 ) (x )
/ ∂x2 ∂x1 ∂x22 /
/ 2 /
/ ∂ f 0 ∂ 2f ∂ 2f /
/ 0
(x 0 )//
/ ∂x 2 (x ) ∂x2 ∂x1 (x ) ∂x3 ∂x1
/ /
/ ∂ 2 f1 ∂ 2f 0 ∂ 2f /
/ 0 /
d3 = / (x 0 ) (x ) (x )/
/ ∂x2 ∂x1 ∂x22 ∂x3 ∂x2 /
/ 2 /
/ ∂ f ∂ 2f ∂ 2 f 0 //
/ (x 0
) (x 0 ) (x ) /
/ ∂x ∂x ∂x3 ∂x2 ∂x 2
3 1 3
(continued)
5.6 Extrema of Several Real Variable Functions 173
· · ·
dn = det (H (f )(x 0 )).
Sylvester’s Criterion
1. d2 f (x 0 ) is positive definite if and only if dk > 0, for all k ∈ {1, 2, . . . , n};

2. d2 f (x 0 ) is negative definite if and only if (−1)k dk > 0, for all k ∈
{1, 2, . . . , n}.
5.57. A particular case of Huygens’ problem. Prove that the function

xy
f : (0, ∞)2 → R, f (x, y) = ,
(x + 1)(y + 1)(x + y)
has a global maximum and determine this maximum.
5.58. Find the local extremum points and the local extremum values of the
following functions:
(a) f : (0, ∞)2 → R, f (x, y) = x

y + yx ;
(b) f : R2 → R, f (x, y) = (x 2 + 1)(y 2 + 1)(xy + 1);
(c) f : 2R2 → R, f (x, y) = 3xy e−2x−y ;
(d) f : (x, y) ∈ R2 | x = 0 → R, f (x, y) = − yx e−y + ex 1
− x 2 − 4x;
(e) f : (0, π ) → R, f (x, y) = cos x + cos y + cos(x + y);
2
(f) f : (0, ∞)3 → R, f (x, y, z) = xy + yz + xz ;

(g) f : R3 → R, f (x, y, z) = arctan(x 2 + 2x) + y 2 + yz + z2 ;
(h) f : (0, ∞)3 → R, f (x, y, z) = xyz + x1 + y1 + 1z ;
(i) f : R3 → R, f (x, y, z) = (x − z)ex−y −z ;
2
3
(j) f : − π2 , π2 → R, f (x, y, z) = cos x + cos y + cos z − cos(x + y + z).
5.59. Let k, l ∈ R. Find the local extremum points and the local extremum values
of the function f : (0, ∞)2 → R, f (x, y) = xy + xk + yl .
5.60. The Minimum Value of Two Integrals
(a) Calculate
1
min (x 2 − a − bx)2 dx.
a, b ∈ R 0
(continued)
(b) Calculate
1
min (x 3 − a − bx − cx 2 )2 dx.
a, b, c ∈ R −1
5.61. Find the maximum value of the function f : R2 → R, f (x, y) = xy, on the
set C = {(x, y) ∈ (0, ∞)2 | 9x 2 + 23xy + 16y 2 = 141}.
5.62. Find the minimum value of the function f : R2 → R, f (x, y) = x + y 4 , on

the set C = {(x, y) ∈ (0, ∞)2 | xy = 4}.
5.63. Find the local extremum points

2 of the function f : R32 → R, f (x, y) =
x + 2y , relative to the set C = (x, y) ∈ R2 | 2x + 4y = 3 .
2 2
5.64. Find the minimum and maximum2 values of the function f 3 : R2 → R,

f (x, y) = x − y, relative to the set C = (x, y) ∈ R2 | x 2 + y 2 = 2 .
5.65. Find the minimum and maximum 2 values of the function3 f : R2 → R,

f (x, y) = xy, relative to the set C = (x, y) ∈ R2 | x 2 + y 2 = 1 .
3 f : R → R, f (x, y) = xy,
5.66. Find the local extremum points of the function 2
2
relative to the set C = (x, y) ∈ R | x + 2y = 4 .
2
5.67. Prove that, of all rectangles of constant perimeter, the square has the
maximum area.
5.68.
(a) Prove that, of all right trapezoids of constant area, the square has the
minimum perimeter.
(b) Prove that, of all right trapezoids of constant perimeter, the square has the
maximum area.
2 function f : [0,3∞) → R, f (x,3 y, z) =

5.69. Find the maximum value of the 3
sin x + sin y + sin z, on the set C = (x, y, z) ∈ [0, ∞) | x + y + z = π .
5.70. Find the local extremum points

2 of the function f : R3 → R, f3(x, y, z) =
x + y + z , relative to the set C = (x, y, z) ∈ R | x + 2y + 3z = a , a ∈ R.
2 3 3
5.71. Find the local extremum points of2 the function f : (0, ∞)3 → R, 3
f (x, y, z) = xyz2 , relative to the set C = (x, y, z) ∈ (0, ∞)3 | x + y + 2z = a ,
a > 0.
2 the function 3f : R →3R, f (x, y, z) =

5.72. Find the local extremum points of 3
x(y + z) + 3yz, relative to the set C = (x, y, z) ∈ R | xyz = 3 .
5.73. Find the minimum and maximum values of the function f : R3 → R,

f (x, y, z) = x + y + z, on the set
C = (x, y, z) ∈ R3 | x − 2y + z = 2, x 2 + y 2 + z2 = 1 .
5.74. Find the local extremum

2 points of the function f : R3 → R, f (x, y, z)3=
xyz, relative to the set C = (x, y, z) ∈ R | x + y + z = 4, xy + yz + xz = 5 .
3
5.75. Find the minimum and maximum values 2 of the function f : R3 → 3 R,

f (x, y, z) = 1 − x − y − z, on the set C = (x, y, z) ∈ R3 | x 2 + y 2 + z2 ≤ 1 .
5.76. Find the minimum and maximum values 2 of the function f : R3 → 3 R,

f (x, y, z) = 2xy + yz + xz, on the set C = (x, y, z) ∈ R3 | x 2 + y 2 ≤ z ≤ 1 .
5.77. [93] A discrete extremum. The positive integers x, y, z satisfy the equality
xy + z = 160. Find the minimum value of the expression x + yz.
5.78. Prove that

/ /
/ 2/ 1
/(x + y)e−x −y / ≤ √ , for all x, y ∈ R.
2
5.79. The Extremum of Some Linear Functions Over Conics
2 minimum and maximum values of f (x, y) = x +y3 on the ellipse

(a) Find the
E = (x, y) ∈ R2 | 3x 2 − 2xy + 3y 2 + 4x + 4y − 4 = 0 .
2 local extreme values of f (x, y) = 2x + y on the hyperbola
(b) Find the 3
H = (x, y) ∈ R2 | 3x 2 + 10xy + 3y 2 − 16x − 16y − 16 = 0 .
2 local extreme values of f (x, y) = 2x − y on 3the parabola
(c) Find the
P = (x, y) ∈ R2 | 9x 2 + 24xy + 16y 2 − 40x + 30y = 0 .
A gem of geometry, analysis, and linear algebra.

We find the minimum and maximum 2values of the function f (x, y) = 3
x 2 +y 2 −16x−10y relative to the set D = (x, y) ∈ R2 | x 2 + y 2 + 2y ≤ 0 .
Solution. We notice that the set D is the disk with center C(0, −1) and
radius 1. Let M(x, y) ∈ D and let A be the point whose coordinates are
(8, 5). We have that f (x, y) = (x − 8)2 + (y − 5)2 − 89 = AM 2 − 89. Thus,
the minimum and maximum values of f are obtained when AM is minimum
and maximum respectively.
Let C = ∂D and let E and F be thepoints where
theline AC intersects
the
circle C . A calculation shows that E 45 , − 25 and F − 45 , − 85 . It follows
that min f (x, y) = AE 2 − 89 = −8 and max f (x, y) = AF 2 − 89 = 32.
A challenge. Find the minimum and maximum values of the function
f (x, 2y, z) = x 2 + y 2 + z2 − 4x − 6y − 14z relative
3 to the set
B = (x, y, z) ∈ R3 | x 2 + y 2 + z2 − 2z ≤ 0 .
5.80. Let P2 be the set of polynomials of degree at most 2, with real coefficients,
and let J : P2 → R be the function defined by
1
J (f ) = f 2 (x)dx.
0
(a) [15, Problem 2.2.19 (Sp80), p. 35] If Q = {f ∈ P2 | f (1) = 1}, then prove that
J attains its minimum over Q and determine f ∈ Q for which this minimum is
attained.2 3
(b) If Q = f ∈ P2 | f (1) = 1 , then prove that J attains its minimum over Q and
determine2 f ∈ Q for which this
3 minimum is attained.
(c) If Q = f ∈ P2 | f (1) = 1 , then prove that J attains its minimum over Q
and determine f ∈ Q for which this minimum is attained.
Implicit Functions
6
He who neglects learning in his youth

loses the past and is dead for the future.
Euripides (BC. 480–BC. 406)
The problems included in this chapter are about studying the existence and unicity
of implicit functions of one or more variables defined by an equation or by a system
of equations.
6.1 Implicit Functions of One Real Variable Defined by an

Equation
The Theorem of Existence and Unicity of Implicit Functions of One Real

Variable Defined by an Equation
Let ∅ = A, B ⊆ R be open sets, F : A × B → R, and x0 ∈ A, y0 ∈ B. If
(i) F (x0 , y0 ) = 0,
(ii) F ∈ C 1 (A × B, R),
(iii) Fy (x0 , y0 ) = 0,
then there exists U ∈ V (x0 ) open, with U ⊆ A, there exists V ∈ V (y0 ) open,
with V ⊆ B, and there exists a unique function y : U → V such that:
(continued)
178 6 Implicit Functions
1◦ F (x, y(x)) = 0, for all x ∈ U ,

2◦ y(x0 ) = y0 ,
F (x, y(x))
3◦ y ∈ C 1 (U, V ) and y (x) = − x , for all x ∈ U .
Fy (x, y(x))
Remark 6.1. If in condition (ii) we have F ∈ C n (A × B, R), n ≥ 2, then in

conclusion 3◦ we can write y ∈ C n (U, V ).
6.1. Prove that the equation x − y 2 + exy = 0 defines y as an implicit function of

x in the neighborhood of (0, 1) and calculate y (0), y (0) and y (0).
6.2. Prove that the equation x 5 + xy 2 − sin y = 0 defines y as an implicit function

of x in the neighborhood of (0, π ). Calculate y (0) and y (0).
6.3. Prove that the equation x 2 − xy + 2y 2 + x − y + cos(πy) = 0 defines y as an

implicit function of x in the neighborhood of (0, 1) and calculate y (0), y (0) and
y (0).
6.4. Prove that the equation cos(x − y) + x sin y − y cos x + π + 1 = 0 defines y as

an implicit function of x in the neighborhood of (0, π ). Calculate y (0) and y (0).
6.5. Prove that the equation (x 2 + y 2 ) arctan xy + x − y + 1 = 0 defines y as an

implicit function of x in the neighborhood of (0, 1) and calculate y (0) and y (0).
6.6. Prove that the equation x y = y x defines y as an implicit function of x in the

neighborhood of (1, 1) and calculate y (1) and y (1).
6.7. Prove that the sum of the lengths of the segments determined by the origin and
the
√ intersection
√ points with the coordinate axes of a tangent to the curve of equation
√
x + y = c, c > 0, is equal to c. (See Fig. 6.1.)
√ √ √
Remark 6.2. The curve x + y = c, c > 0, is in fact a part of a parabola (prove
α α
it!). In general, the curves defined implicitly by the equation xa α + ybα = 1, where a
and b are positive numbers and α ∈ R, are called Lamé’s curves or superellipses.
6.8. Folium of Descartes. Let us consider the equation x 3 + y 3 − 3axy = 0, a > 0,

and let α = 3a
2 . Prove that the preceding equation defines y as an implicit function
of x in the neighborhood of (α, α) and calculate y (α), y (α), and y (α). (See
Fig. 6.2.)
Fig. 6.1 The tangent to

Lamé’s
√ curve√
√
x + y = c, c > 0
Fig. 6.2 Folium of Descartes

x 3 + y 3 − 3axy = 0, a > 0
6.9. Lemniscate of Bernoulli.

√
Let us consider the equation (x 2 + y 2 )2 − 2a 2 xy = 0,
a > 0, and let α = 2 . Prove that the preceding equation defines y as an implicit
a 2
function of x in the neighborhood of (α, α) and calculate y (α) and y (α). (See
Fig. 6.3.)
6.10. Lemniscate of Bernoulli. Determine the point, in the first quadrant, located on
the lemniscate of the equation (x 2 + y 2 )2 − 2a 2 (x 2 − y 2 ) = 0, a > 0, where the
tangent is parallel to the x axis. (See Fig. 6.4.)
Fig. 6.3 Lemniscate of

Bernoulli
(x 2 + y 2 )2 − 2a 2 xy = 0,
a>0
Fig. 6.4 Lemniscate of

Bernoulli
(x 2 +y 2 )2 −2a 2 (x 2 −y 2 ) = 0,
a>0
2 2 2
6.11. Astroid. Prove that the tangents
√
to the astroid of equation x 3 + y 3 = a 3 ,
a 2
a > 0, at the points of abscissa 4 are perpendicular. (See Fig. 6.5.)
6.12. Cardioid. Prove that the tangents to the cardioid of equation (x 2 +y 2 +ax)2 =
a 2 (x 2 + y 2 ), a > 0, at the points of abscissa 0 and ordinate different from 0 are
perpendicular. (See Fig. 6.6.)
Fig. 6.5 Astroid

2 2 2
x 3 + y 3 = a 3 , a > 0,
and the two tangents
Fig. 6.6 Cardioid

(x 2 +y 2 +ax)2 = a 2 (x 2 +y 2 ),
a > 0, and the two tangents
6.13. Heart. Let us consider the equation (x 2 + y 2 − 1)2 − x 2 y 3 = 0.
(a) Prove that the preceding equation defines y as an implicit function of x in the
neighborhood of (1, 1) and calculate y (1) and y (1).
(b) Prove that the preceding equation defines y as an implicit function of x in the
neighborhood of (−1, 1) and calculate y (−1) and y (−1).
(See Fig. 6.7.)

Fig. 6.7 Heart

(Eugène Beutel, 1909)
(x 2 + y 2 − 1)3 − x 2 y 3 = 0
Fig. 6.8 Four leaf clover

(x 2 + y 2 − 1)3 − x 2 y 2 = 0
6.14. Four leaf clover. Prove that the equation (x 2 + y 2 − 1)3 − x 2 y 2 = 0 defines
y as an implicit function of x in the neighborhood of (1, 1) and calculate y (1) and
y (1). (See Fig. 6.8.)
6.2 Implicit Functions of Two Real Variables Defined by an Equation 183

Equation
The Theorem of Existence and Unicity of Implicit Functions of Two

Variables Defined by an Equation
Let ∅ = A ⊆ R2 be open set, ∅ = B ⊆ R be open set, F : A × B → R, and
(x0 , y0 ) ∈ A, z0 ∈ B. If
(i) F (x0 , y0 , z0 ) = 0,
(ii) F ∈ C 1 (A × B, R),
(iii) Fz (x0 , y0 , z0 ) = 0,
then there exists U ∈ V ((x0 , y0 )) open, with U ⊆ A, there exists V ∈ V (z0 )

open, with V ⊆ B, and there exists a unique function z : U → V such that:
1◦ F (x, y, z(x, y)) = 0, for all (x, y) ∈ U ,

2◦ z(x0 , y0 ) = z0 ,
3◦ z ∈ C 1 (U, V ), and
Fx (x, y, z(x, y)) Fy (x, y, z(x, y))

zx (x, y) = − , zy (x, y) = − ,
Fz (x, y, z(x, y)) Fz (x, y, z(x, y))
for all (x, y) ∈ U .
Remark 6.3. If in condition (ii) we have F ∈ C n (A × B, R), n ≥ 2, then in

conclusion 3◦ we can write z ∈ C n (U, V ).
6.15. Prove that the equation xz−z2 +y−cos z = 0 defines z as an implicit function
of x and y in the neighborhood of (−1, 1, 0) and calculate zx (−1, 1), zy (−1, 1),
zx 2 (−1, 1), zxy (−1, 1), and zy 2 (−1, 1).
6.16. Prove that the equation (2x − y)ez + 3xy − z = 0 defines z as an implicit
function of x and y in the neighborhood of (1, −1, 0) and calculate zx (1, −1),
zy (1, −1), zx 2 (1, −1), zxy (1, −1), and zy 2 (1, −1).
6.17. Prove that the equation xy arctan z + 2x 2 − y + z = 0 defines z as an implicit

function of x and y in the neighborhood of (−1, 2, 0) and calculate zx (−1, 2),
zy (−1, 2), zx 2 (−1, 2), zxy (−1, 2), and zy 2 (−1, 2).

System of Equations
The Theorem of Existence and Unicity of Implicit Functions of One Real

Variable Defined by a System of Equations
Let ∅ = A, B, C ⊆ R be open sets, F, G : A × B × C → R and x0 ∈ A,
y0 ∈ B, z0 ∈ C. If
(i) F (x0 , y0 , z0 ) = 0, G(x0 , y0 , z0 ) = 0,

(ii) /F, G ∈ /C 1 (A × B × C, R),
/F F /
/ /
(iii) / y z / = 0,
/Gy Gz /
(x0 ,y0 ,z0 )
then there exists U ∈ V (x0 ) open, with U ⊆ A, there exists V ∈ V (y0 ) open,
with V ⊆ B, there exists W ∈ V (z0 ) open, with W ⊆ C, and there exist a
unique function y : U → V and a unique function z : U → W such that:
1◦ F (x, y(x), z(x)) = 0, G(x, y(x), z(x)) = 0, for all x ∈ U ,

2◦ y(x0 ) = y0 , z(x0 ) = z0 ,
3◦ y ∈ C 1 (U, V ), z ∈ C 1 (U, W ) and
/ / / /
/F /F Fx //
/ x Fz // / y
/ /
/G Gz /(x,y(x),z(x)) /Gy Gx /
x
y (x) = − / / , z (x) = − / / (x,y(x),z(x)) ,
/F Fz // /F Fz //
/ y / y
/ / / /
/Gy Gz / /Gy Gz /
(x,y(x),z(x)) (x,y(x),z(x))
for all x ∈ U .
Remark 6.4. If in condition (ii) we have F, G ∈ C n (A × B × C, R), n ≥ 2,

then in conclusion 3◦ we can write y ∈ C n (U, V ), z ∈ C n (U, W ).
6.18. Prove that the system of equations

.
z2 − 2x + y + ey = 0
x3 − y2 + z = 0
defines y and z as implicit functions of x in the neighborhood of (1, 0, −1) and

calculate y (1), z (1), y (1), and z (1).
6.4 Implicit Functions of Two Real Variables Defined by a System of Equations 185

.
x 8 − xy + yz = 1
−x 2 z2 + 10x + 5ey = 14
defines y and z as implicit functions of x in the neighborhood of (1, 0, −1) and

calculate y (1), z (1), y (1), and z (1).

.
x sin(yz) − cos z + y = 0
2ex+y + xy − z2 = 1
defines y and z as implicit functions of x in the neighborhood of (−1, 1, 0) and

calculate y (−1), z (−1), y (−1), and z (−1).

System of Equations
The Theorem of Existence and Unicity of Implicit Functions of Two Real

Variables Defined by a System of Equations
Let ∅ = A ⊆ R2 be open set, ∅ = B, C ⊆ R open sets, F, G : A × B × C →
R and (x0 , y0 ) ∈ A, u0 ∈ B, v0 ∈ C. If
(i) F (x0 , y0 , u0 , v0 ) = 0, G(x0 , y0 , u0 , v0 ) = 0,

(ii) /F, G ∈ /C 1 (A × B × C, R),
/F F /
(iii) // u v // = 0,
Gu Gv (x ,y ,u ,v )
0 0 0 0
then there exists W ∈ V ((x0 , y0 )) open, with W ⊆ A, there exists U ∈ V (u0 )

open, with U ⊆ B, there exists V ∈ V (v0 ) open, with V ⊆ C, and there exist
a unique function u : W → U and a unique function v : W → V such that:
1◦ F (x, y, u(x, y), v(x, y)) = 0 and G(x, y, u(x, y), v(x, y)) = 0, for all
(x, y) ∈ W ,
2◦ u(x0 , y0 ) = u0 , v(x0 , y0 ) = v0 ,
3◦ u ∈ C 1 (W, U ), v ∈ C 1 (W, V ), and
(continued)
/ / / /
/ / /F Fv //
/ Fx Fv / / y
/ / / /
/Gx Gv / /Gy Gv /
ux (x, y) = − / /(x,y,u(x,y),v(x,y)) , uy (x, y) = − / /(x,y,u(x,y),v(x,y)) ,
/ / / /
/ Fu Fv / / Fu Fv /
/ / / /
/Gu Gv / /Gu Gv /
(x,y,u(x,y),v(x,y)) (x,y,u(x,y),v(x,y))
/ / / /
/ / /F Fy //
/ Fu Fx / / u
/ / / /
/Gu Gx / /Gu Gy /
vx (x, y) = − / /(x,y,u(x,y),v(x,y)) , vy (x, y) = − / /(x,y,u(x,y),v(x,y)) ,
/ / / /
/ Fu Fv / / Fu Fv /
/ / / /
/Gu Gv / /Gu Gv /
(x,y,u(x,y),v(x,y)) (x,y,u(x,y),v(x,y))
for all (x, y) ∈ W .
Remark 6.5. If in condition (ii) we have F, G ∈ C n (A × B × C, R), n ≥ 2,

then in conclusion 3◦ we can write u ∈ C n (W, U ), v ∈ C n (W, V ).

.
xy + xv − uv + eu = π
xu + x − y − cos v = 3
defines u and v as implicit functions of x and y in the neighborhood of (1, −1, 0, π ).

Calculate ux (1, −1), uy (1, −1), vx (1, −1), and vy (1, −1).

.
x sin u − y cos v − x + y = 1
e2x+y − sin(u − v) = 1
defines u and v as implicit functions of x and y in the neighborhood of (−1, 2, π, 0)

and calculate ux (−1, 2), uy (−1, 2), vx (−1, 2), and vy (−1, 2).
Challenges, Gems, and Mathematical Beauties
7
In mathematics the art of proposing a question

must be held of higher value than solving it.
Georg Ferdinand Cantor (1845–1918)
This chapter contains challenging problems on various topics discussed in the

previous chapters of this book. A special attention is given to the calculation of series
involving harmonic numbers and tails of Riemann zeta function values. The calcu-
lation of these series reduces to the evaluation of linear or nonlinear Euler sums. A
series involving a single harmonic number is called a linear Euler sum, while a series
involving a product of at least two harmonic numbers is called a nonlinear Euler
sum. The study of these series has been initiated by the famous mathematicians
Leonhard Euler and Christian Goldbach in the mid-eighteenth century. Most of the
series in this chapter are new and we have included them here because they are
simply unusual sums containing combinations of harmonic numbers and tails of zeta
function values. These series, linear or quadratic, alternating or not, are evaluated
in terms of zeta function, logarithms, and polylogarithm of special values. Proofs
of special results of analysis, which are called gems, challenges, and several open
problems are scattered throughout the chapter.
7.1. Calculate:
√ n
(a) lim 2 n x − 1 , x > 0;
n→∞ n
√ √
(b) lim n x + n y − 1 , x, y > 0;
n→∞
1
(c) lim (2x x − 1) x .
x→0+
188 7 Challenges, Gems, and Mathematical Beauties
A Gem of Mathematical Analysis

We prove that the sequence 1 + 12 + · · · + 1
n diverges.
n≥1
By way of contradiction, we assume that the sequence converges. This
implies that there exists a ∈ R such that
!
1 1
lim 1 + + ··· + = a.
n→∞ 2 n
Equivalently
4 5
1 1 1 1
lim 1+ + ··· + − 1 + + ··· + = a.
n→∞ 2 n 2 n
We also have
4 5
1 1 1 1
lim 1 + + · · · + − 1 + + ··· + = a.
n→∞ 2 2n 2 2n
Subtracting the previous two limits we obtain that

4 5 4 5
1 1 1 1 1 1
lim + ··· + − 1 + + ··· + + 1 + + ··· + = 0.
n→∞ n+1 2n 2 2n 2 n
It follows that
4 5 4 5
1 1 1 1
lim 1 + + ··· + − 1 + + ··· + = ln 2,
n→∞ 2 2n 2 n
which is a contradiction, since a sequence of integers cannot converge to an

irrational number.
In fact, it can be
proved that, this is left
as a challenge to the interested
reader, the sequence 1 + 12 + · · · + n1 is dense in [0, 1].
n≥1
7.2. Study the convergence of the sequence ({ln n})n≥1 , where {a} denotes the
fractional part of a.
7.3. Let k ∈ N. Calculate

n (kn)!
lim .
n→∞ (n!)k
7.4. Calculate lim (eHn −γ − n).

n→∞
7.5. [A. Boţan, 2016] Prove that:
(a) ln x ≤ x − 1 ≤ ln x + ln2 x, ∀x ∈ [1, 2);

√
n
√
n √
(b) lim 1 + 2 + · · · + n n − n − ln n = −1.
n→∞
γ
7.6. Let a, b ∈ R, with a − b = 2. Calculate
√ 1 1

2e1+ 2 +···+ n −a − n e1+ 3 +···+ 2n−1 −b .
1 1
lim
n→∞
7.7. Calculate

(sin 1)2n + (sin 2)2n + · · · + (sin n)2n .
n
lim
n→∞
7.8. [M. Ivan, 2019]
(a) Calculate

n n sin 1
lim 2 + 2n sin 2 + · · · + 2n sin n .
n→∞
(b) Let (xn )n∈N be a bounded sequence of real numbers. Calculate

√
2n x1 + 2n x2 + · · · + 2n xn .
n
lim
n→∞
7.9. Trigonometric Limits Calculate:

(a) lim n
| sin 1 sin 2 · · · sin n|;
n→∞ 6
n
(b) lim sin2 n sin2 (n + 1) · · · sin2 (3n);
n→∞
(c) lim n (a ± cos 1)(a ± cos 2) · · · (a ± cos n), a ≥ 1;
n→∞
(d) lim (a sin 1 + b cos 1)2 (a sin 2 + b cos 2)2 · · · (a sin n + b cos n)2 ,
n
n→∞
a, b ∈ R.
7.10. Prove that

ln n
1+ 1
3 + ··· + 1
2n−1
lim = 4.
n→∞ 1
2 + 4 + · · · + 2n
1 1
7.11. Let a, b ∈ (0, ∞) and α, β ∈ (0, 1). Prove that

β−1
1 1 1 1
+ + + ··· +
aα (a + 1)α (a + 2)α (a + n − 1)α (1 − β)α−1
lim α−1 = .
n→∞ 1 1 1 1 (1 − α)β−1
+ + + ··· +
b β (b + 1)β (b + 2)β (b + n − 1)β
7.12. A. Exponential Limits

Let x >70. Prove that:
x xn
(a) lim n ex − 1 − − · · · − = ex;
n
n→∞
71! n!

2n x xn x
n · n e − 1 − − ··· − − ex = ex ln √ ;
3 n x
(b) lim n
n→∞ 1! n! 2π

x x 2n−1 ex
(c) lim n sinh x − − · · · − =
n
;
n→∞ 1! (2n − 1)! 2

x2 x 2n ex
(d) lim n cosh x − 1 − − ··· − =
n
;
n→∞ 2! (2n)! 2
(e) A challenge.
⎡
− n+1
1
⎣ x x n+1
lim e − 1 − − ··· −
x
n→∞ 1! (n + 1)!
− 1
x xn n 1
− e − 1 − − ··· −
x
= .
1! n! ex
B. Polylogarithm Limits
Let k ∈N and x ∈ [0, 1). Prove that:
n x2 xn
(a) lim Lik (x) − x − k − · · · − k = x;
n→∞
⎛ 2 n ⎞

x 2 x n x
(b) lim n ⎝ nk · Lik (x) − x − k − · · · − k − x ⎠ = x ln
n n
.
n→∞ 2 n 1−x
We mention that Li1 (x) = − ln(1 − x).
7.13. Prove that

n
1 1
lim + + ··· − ln n = 1 + γ .
n→∞ k2 (k + 1)2
k=1
7.14.
(a) Let n ∈ N. Prove that the equation x n + x n−1 + · · · + x − 1 = 0 has a unique

positive solution xn , with xn ≤ 1.
(b) Prove that the sequence (xn )n≥1 is decreasing.
(c) Prove that lim xn = 12 .
n→∞
(d) Prove that lim 2n xn − 12 − 2n+2 1
= 0.
n→∞
7.15. Prove that:

n−1
n2 π2
(a) [142] lim = ;
n→∞ k 2 (n − k)2 3
k=1
n−1
n3
(b) lim = 2ζ (3).
n→∞ k 3 (n − k)3
k=1
7.16. Taylorian Limits

Let n ≥ 0 be an integer. Calculate:
2 x1
xn
ex − 1 − x
1! − x2! − · · · −
(a) lim (n + 1)! n!
;
x→0 x n+1
x2 xn
x1
1
ln 1−x −x− 2 − ··· − n
(b) lim (n + 1) .
x→0 x n+1
7.17.

(a) Let x ∈ R. Calculate lim n cosn x
n −1 .
n→∞ x
(b) Let f ∈ C 2 (R, R), with f (0) = 1. Prove that lim f n n = ef (0)x and
n→∞
calculate
x
lim n f n − ef (0)x .
n→∞ n
7.18. Let f : [0, 1] → R be a continuous function. Calculate:

∞
∞

1 xk xi
(a) lim n2 f (x)dx;
n→∞ 0 k2 i
k=n i=n
(continued)
∞
2
1 xk
3
(b) lim n f (x)dx.
n→∞ 0 k2
k=n
7.19. [137], [128, problems 59–61, pp. 19, 20], [130, Proposition 2]
Let k ≥ 0 be an integer, p, q, s ∈ N, with p ≥ 2, and a, r ∈ (0, +∞). We
consider the sequence (an )n∈N defined by an = a + (n − 1)r.
Prove that:

aqn+k+1 aqn+k+1+p · · · aqn+k+1+s(n−1)p p ps + q
(i) lim = ;
n→∞ aqn+k aqn+k+p · · · aqn+k+s(n−1)p q
7 q
aqn+k aqn+k+p · · · aqn+k+s(n−1)p p ps + q
(ii) lim n = [(ps+q)r]s ;
n→∞ (n!)s q
√n a
qn+k aqn+k+p · · · aqn+k+s(n−1)p
(iii) lim s
7
n→∞ q n
p ps + q (ps + q)r s
= .
q e
7.20. [131, Problem 3] Evaluate

3
55n Cn2n
lim .
n→∞ C5n n 2n
10n C5n C4n
7.21. [131, Problem 4] Let p, α, β ∈ N, with p ≥ 2 and α < β. We consider

the sequences (xn )n∈N and (yn )n∈N , with x1 = 0 and y1 = 0, defined by the
recurrence relations

1
nxn+1 = n + xn and nyn+1 = (n + 1)yn .
p
yαn (xαn + xαn+1 + · · · + xβn )

Evaluate lim .
n→∞ xαn (yαn + yαn+1 + · · · + yβn )
7.22. [131, Problem 5] Let λ, μ, α, β ∈ N, with λ = μ and α < β. We consider the

sequences (xn )n∈N and (yn )n∈N , with x1 = 0 and y1 = 0, defined by the recurrence
relations

1 1
nxn+1 = n+ xn and nyn+1 = n+ yn .
λ μ
yαn (xαn + xαn+1 + · · · + xβn )

Evaluate lim .
n→∞ xαn (yαn + yαn+1 + · · · + yβn )
7.23. [133] Let p, a, b ∈ N, with a < b. We consider the sequence (xn )n≥1 defined
by the recurrence nxn+1 = (n + 1/p)xn and an initial condition x1 = 0. Evaluate
xan + xan+1 + · · · + xbn

lim .
n→∞ nxan
7.24. [138] Let p ≥ 2 be an integer and let (xn )n∈N be the sequence defined by
-
n−1
n + jp + 1
xn = .
n + jp
j =0
Assuming that lim xn exists, find its value.

n→∞
Remark 7.1. The limit in Problem 7.24 is a particular case of the limit in Problem
1.16 (q = 1) and also a particular case of part (i) in Problem 7.19 (k = 0, q = s =
1). Different solutions are given there, without the assumption that the limit exists,
while here we present a telescoping way for finding the limit (see the solution).
7.25. [135] Let p ≥ 2 be an integer. Determine the limit

∞ √p
n
lim p .
n→∞
k=1 j =1 k (n + k)p−j +1
p j

n
7.26. [136] Let (Sn )n∈N be the sequence defined by Sn = (−1)k−1 1
k − ln k+1
k .
k=1
Calculate:
/ /
/ 2/ 4 //
(a) lim n /Sn − ln /;
n→∞
/ π /
3/
/ (−1)n−1 4 //
(b) lim n /Sn − − ln /.
n→∞ 4n2 π
7.27.
(a) Calculate
2 xn
lim dx.
n→∞ 0 1 + xn
(b) Let f : [0, ∞) → R be a continuous function such that lim f (x) = f (∞)
x→∞
exists and is finite. Prove that
2
lim f (x n )dx = f (0) + f (∞).
n→∞ 0
7.28. Let a ∈ [0, 1) and let f : [a, 1] → R be a continuous function. Prove

that
1
lim x y x f (y)dy = f (1).
x→∞ a
7.29. Calculate
1
lim x (t 2 − t + 1)x dt.
x→∞ 0
7.30. [O. Furdui, 30 December 2016]

Calculate
1
lim x t tx dt.
x→∞ 0
7.31. [89] Let a > 1. Calculate

1
lim x a t (t−1)x dt.
x→∞ 0
7.32. Calculate

∞
lim n
{x}n e−x dx,
n→∞ 0
7.33. Calculate

1 ∞
lim n
xn e−x dx,
n→∞ n 0
7.34. Let a, b > 0. Calculate

1 √
lim n2 xn ax + b − 1 dx.
n
n→∞ 0
7.35. Prove that

1 √ π2
lim n2 1 + x n dx − 1 =
n
.
n→∞ 0 12
7.36. Let a, b ≥ 0.
(a) Calculate
√
π
2
lim n a sin2n x + b cos2n x dx.
n→∞ 0

(b) Let f : 0, π2 → R be a continuous function. Calculate
√
π
2
lim n a sin2n x + b cos2n x f (x) dx.
n→∞ 0
7.37. Let a, b ∈ R, a < b, and let f : [a, b] → R be a Riemann integrable

function. Calculate
b f (x)
lim dx.
n→∞ a 1 + cos x cos(x + 1) · · · cos(x + n)
7.38. Limits of Integrals with Exponential and Trigonometric Functions
(a) [75] Let k > 0 be a real number. Calculate

∞ | sin x|k
lim nk dx.
n→∞ 0 e(n+1)x − enx
(b) [79] Let k > −1 be a real number. Calculate

∞ x k sin x
lim nk+1 dx.
n→∞ 0 e(n+1)x − enx
7.39. Let f : [0, ∞) → R be continuous at 0 and bounded. Prove that

√ n
∞ x2 + 1 − x f (0)
lim n √ f (x) dx = .
n→∞ 0 x2 +1+x 2
7.40.
(a) Calculate
1 n
lim | sin x| dx.
n→∞ n 0
(b) Calculate
1 n
lim | sin x| cos2 x dx.
n→∞ n 0
7.41. Let f : [0, 1] → R be a Riemann integrable function.
(a) Prove that

1 1 π
lim f (| sin nx|)dx = f (sin x)dx.
n→∞ 0 π 0
(b) Prove that

1 1 2π
lim f (sin nx)dx = f (sin x)dx.
n→∞ 0 2π 0
7.3 Convergence and Evaluation of Series 197
7.42. Let k, m ≥ 0. Prove that

1 n
x (1 − x)m dx = e−k−m .
n k
lim
n→∞ 0
An interesting identity
n n n
n(n + 1)(2n + 1)
min {i, j } = i2 = , n ∈ N.
6
i=1 j =1 i=1
7.43. Let k ≥ 2 be an integer. Calculate
k {x n }
lim dx,
n→∞ 1 x
where {a} denotes the fractional part of a.
7.44. Let k ∈ N. Calculate

2
lim nx k dx,
n→∞ 1
where {a} denotes the fractional part of a.
7.3 Convergence and Evaluation of Series

∞
√ α
7.45. Study the convergence of the series n
n − 1 lnβ n, α, β > 0.
n=2
∞
1
7.46. Study the convergence of the series sin n sin .
n
n=1
7.47. A trigonometric series
(a) Prove that | sin x| ≤ |x|, for all x ∈ R.

∞ √
(b) Prove that the series sin π(2 + 3)n converges absolutely.
n=1
∞ sin n

7.48. Study the convergence of the series 1−n n .
n=2
7.49. Let f : [0, 1] → R be a continuous function. Study the convergence of the

series
∞ 1 xn
f (x)dx.
1 + x + x2 + · · · + xn
n=1 0
7.50. Study the convergence of the series:
∞

π sin2 x
n
(a) − dx ;
2
0 x2
n=1
∞ n sin4 x
(b) ln 2 − dx .
0 x3
n=1
7.51. Let p ≥ 2 be an integer. Prove that:
1 p p
(a) √ < √ − √ , ∀n ∈ N;
(n + 1) n
p p
n p
n+1
∞
1
(b) √ < p.
(n + 1) p n
n=1
See also Problem 2.24.
7.52. Calculate
∞ ∞
{x} e−x dx and xe−x dx.
0 0
7.53. Calculate
∞ ∞
1 (−1)n
and .
(n − 1)! + n! (n + (−1)n )!
n=1 n=0
7.54. Calculate
∞ ∞
n n
and .
(n + 1)! (n + (−1)n )!
n=1 n=1

∞
1 1 1 1 1
= 1 + + + ··· + + − e.
(n + k − 1)(n + k)(n + k)! 1! 2! k! k · k!
n=1
Remark 7.2. When k = 1 we obtain Problem 1.3.
7.56. Prove that

∞ ∞
1 1 1
= ln 2 − 1 and = ln 2 + .
(−1)n n + 1 (−1)n n − 1 2
n=2 n=2
7.57. Prove that
(−1) 2 (−1) 3
∞ n ∞ n
π 1 2π 1
= − ln 2 and = √ − ln 2.
n 4 2 n 3 3 3
n=1 n=1
7.58. A Wallis Series Prove that

∞
n
1 4
(−1) 2 ln 1 + = ln ,
n π
n=1
7.59.
(a) Let n, k ∈ N. Prove that

1 n!
(1 − x)n x k−1 dx = .
0 k(k + 1)(k + 2) · · · (k + n)
(b) Prove that

∞
1 1
= .
k(k + 1)(k + 2) · · · (k + n) n · n!
k=1

7.4 Harmonic Series
7.60. A Famous Series of Euler Revisited

Prove, by calculating in two different ways the integral
1 ln x ln(1 − x)
dx,
0 x(1 − x)
that
∞
Hn
= 2ζ (3).
n2
n=1
7.61.
(a) Let k ≥ 1 be an integer. Prove that

∞
1 Hk
= .
n(n + k) k
n=1
(b) Euler’s series. Prove, by using part (a), that

∞
Hk
= 2ζ (3).
k2
k=1
(c) A telescoping term. Prove that
Hn Hn+1 H2 H2 Hn Hn+1 1 Hn+1

= n − n+1 + − + + , n ≥ 1.
n(n + 1) n n+1 n n + 1 (n + 1)2 (n + 1)2
(d) [25] A nonlinear Euler sum. Prove that

∞
Hn Hn+1
= ζ (2) + 2ζ (3).
n(n + 1)
n=1
(e) Prove that
(continued)
7.4 Harmonic Series 201
∞
Hn2
= 3ζ (3).
n(n + 1)
n=1
7.62. [O. Furdui, 16 July 2018]
(a) Prove that

n
Hk2 = (n + 1)Hn2 − (2n + 1)Hn + 2n, n ∈ N.
k=1
(b) [65] Prove that

∞
1 1 1
Hn2 ζ (2) − 1 − 2 − · · · − 2 − = 2 − ζ (2) − 2ζ (3).
2 n n
n=1
(c) Prove that

∞
1 1
Hn2 ζ (3) − 1 − 3 − · · · − 3 = 2ζ (2) − 4ζ (3) + 3ζ (4).
2 n
n=1
7.63. Two series and Apéry’s constant.

Prove that:
∞
1 1 1 1
(a) [26, p. 142] ζ (2) − 2 − 2 − · · · − 2 = ζ (3);
n 1 2 n
n=1
∞
1 1 1 1
(b) + + + · · · = 2ζ (3).
n n2 (n + 1)2 (n + 2)2
n=1
A Famous Series of Goldbach Revisited

We prove that
∞
Hn π4
= .
n3 72
n=1
(continued)
We have
∞ ∞ ∞ ∞ ∞
Hn 1 1 1
S= = = .
n3 n2 k(n + k) n2 k(n + k)
n=1 n=1 k=1 n=1 k=1

∞
∞
It follows, based on symmetry reasons, that S = 1
nk 2 (n+k)
and this
n=1 k=1
shows
∞ ∞ ∞ ∞ ∞ ∞
1 1 1 1 1 π4
2S= + = = = .
n2 k(n+k) nk 2 (n+k) n k2
2 n2 k 2 36
n=1 k=1 n=1 k=1 n=1 k=1
7.64. Prove that

∞
π4 H1 H2 Hn 17
Hn − 3 − 3 − ··· − 3 = ζ (4) − 2ζ (3).
72 1 2 n 4
n=1
7.65. A challenge. Prove that

∞
H1 H2 Hn
(−1)n Hn 2ζ (3) − 2 − 2 − · · · − 2
1 2 n
n=1

ln4 2 π 2 ln2 2 π4 7 1 1
= − − − ζ (3) ln 2 + Li4 .
48 48 1440 8 2 2
This result, which answers a question of O. Furdui and A. Sîntămărian

[60], was communicated to the authors by M. Levy on 24 November, 2019.
The second open problem in [60], of calculating the series
∞
π4 H1 H2 Hn
n
(−1) Hn − 3 − 3 − ··· − 3 ,
72 1 2 n
n=1
was solved by C.I. Vălean and M. Levy on December 2019 in [147].

∞
7.66. Computing Li2 1
and n Hn
2 n=1 (−1) n
(a) Prove the identity

1 Hn
x n−1 ln(1 − x)dx = − , n ≥ 1.
0 n
(b) An alternating harmonic series. Prove that
∞
Hn π2 ln2 2
(−1)n−1 = − .
n 12 2
n=1
We solve this part of the problem here. We have

1 1
1 dx 1
ln = ln 1 − x + x 2 − x 3 + · · · dx
0 1−x 1+x 0 1−x
H2 H3
=1− + − ··· .
2 3
A calculation, based on the substitution x = 1−t
1+t , shows that

1 1 dx π2 ln2 2
ln = − ,
0 1−x 1+x 12 2
and we obtain, probably a new method, that
∞
Hn π2 ln2 2
(−1)n−1 = − .
n 12 2
n=1
(c) Justify the steps in Legendre’s formula

∞ 1
1 1 dx
= ln
n2 2n 0 x 2−x
n=1
1
1
= ln 1 − x + x 2 − · · · dx
0 1−x
H2 H3
=1− + − ···
2 3
π2 ln2 2
= − .
12 2
(continued)
Remark 7.3.
(i) We mention the equality

∞
1 1 π2 ln2 2
Li2 = = − .
2 n2 2n 12 2
n=1

See also Problems 3.64 and 7.83 for the calculation of Li2 12 .
(ii) The calculation of a class of log integrals.
We record the calculation of a class of logarithmic integrals. We have
1 ln(1 − x) 2 1 ln t
2
dx = dt
0 1+x 0 2−t
∞
n+1 1
= t n ln2 tdt
2n+2 0
n=0
∞
1
=
2n+1 (n + 1)2
n=0
7.66 (c) π2 ln2 2

= − .
12 2
More generally, one can prove that, for k ≥ 2, we have
1 ln(1 − x) k k
k−1
1
dx = (−1)i−1 s(k − 1, i)Lik+1−i ,
0 1+x 2k−1 2
i=1
where s(k − 1, i) are the Stirling numbers of the first kind.
7.67. A Gem with Two Series

Prove that
∞ ∞
Hn− − Hn Hn− + Hn
(−1)n−1 = ln2 2 and (−1)n−1 = ζ (2).
n n
n=1 n=1
7.68. Prove that:

∞
1 1 (−1)n−1 1 − ln 2
(a) Hn ln + 1 − + · · · + = ;
2 2 n 2
n=1
∞
Hn 1 1 (−1)n−1 ζ (3) ln3 2
(b) ln + 1 − + · · · + = − .
n+1 2 2 n 8 6
n=1
7.69. Intriguing Series with H2n − Hn − ln 2

Prove that:
∞
1 1 1 1
(a) [68] (−1)n n + + ··· + − ln 2 +
n+1 n+2 2n 4n
n=1
= ln 2−1
8 ;
∞
1 1 1 1 1
(b) n 2
(−1) n + + ··· + − ln 2 + −
n+1 n+2 2n 4n 16n2
n=1
−3
= π32 .
7.70. A Challenging Series with H2n − Hn − ln 2 and the Tail of ζ (2)
(a) Prove that

1 x 2n
dx = Hn − H2n + ln 2, n ≥ 1.
0 1+x
(b) Prove that

∞
1 1
ζ (2) − 1 − 2
− ··· − 2 (H2n − Hn − ln 2)
2 n
n=1
1 7
= − ζ (2) − ζ (3) + ζ (2) ln 2 + ln2 2.
2 8
7.71. The Glaisher–Kinkelin constant A is defined by the formula
n2 n 1 n2 -
n
A = lim n− 2 − 2 − 12 e 4 k k = 1.282427129100622 . . . .
n→∞
k=1
Prove that
∞
1 1 5 γ
n Hn − ln n − γ − + = − ln A + .
2n 12n2 24 12
n=1
7.5 Series with Factorials
7.72. A Mathematical Jewel I
(a) Check the identity
(m − 1)! (m − 1)! m!
= − , m, n ≥ 1.
(n + m)! n(n + m − 1)! n(n + m)!
(b) [149, B. Cloitre, 9 December 2004, entry 8]
Prove that
∞ ∞
(n − 1)!(m − 1)!
= ζ (2).
(n + m)!
n=1 m=1
7.73. [48] A Mathematical Jewel II
(a) Let a ∈ [−1, 1]. Prove that

∞ ∞
(i − 1)!(j − 1)! i+j a a
a = 2Li2 −2Li2 − −2Li2 (a),
(i + j )! 2−a 2−a
i=1 j =1
where Li2 is the Dilogarithm function defined in Problems 3.4 and 7.83.
(continued)
7.5 Series with Factorials 207
(b) A double sum with factorials.

Prove that
∞ ∞
(i − 1)!(j − 1)!
= ζ (2).
(i + j )!
i=1 j =1
A Mathematical Jewel III

We record as a gem the third proof, the first two proofs being given in
Problems 7.72 and 7.73, of the following remarkable formula due to B.
Cloitre, involving a double series with factorials
∞ ∞
(i − 1)!(j − 1)!
= ζ (2).
(i + j )!
i=1 j =1
We have
∞ ∞ ∞ ∞
(i − 1)!(j − 1)! 1
= (i − 1)!
(i + j )! j (j + 1) · · · (j + i)
i=1 j =1 i=1 j =1
∞
(i − 1)!
=
i · i!
i=1
∞
1
=
i2
i=1
= ζ (2).
We used in our calculations the following formula due to M. Andreoli [2]

∞
1 1
= , n ∈ N,
k(k + 1) · · · (k + n) n · n!
k=1
whose solution is given in detail in [26, Problem 3.10, p. 140]. For another
proof of the preceding formula see the solution of Problem 7.59.
Remark 7.4. It is worth mentioning the fabulous series with four factorials
∞ ∞ ∞
(i − 1)!(j − 1)!(k − 1)! 13 π2
= ζ (3) − ln 2
(i + j + k)! 4 2
i=1 j =1 k=1
whose proof, which is challenging, is left to the interested reader.
7.74. Amazing series with three factorials.

Prove that:
∞ ∞
(i − 1)!(j − 1)! ζ (2)
(a) An alternating series (−1)i−1 = ;
(i + j )! 2
i=1 j =1
∞ ∞ (i − 1)!(j − 1)!
(b) A ζ (2) series (−1)i−1 + (−1)j −1 = ζ (2);
(i + j )!
i=1 j =1
∞ ∞
(i − 1)!(j − 1)!
(c) = 2 − ζ (2);
(i + j + 1)!
i=1 j =1
∞ ∞
(i − 1)!(j − 1)! ζ (2) 3
(d) = − .
(i + j + 2)! 2 4
i=1 j =1
7.6 Series of Functions
7.75. A uniformly continuous function.

Prove that the series ∞
n=1
cos(n4 x)
n2
defines a uniformly continuous function on
R.
7.76. Let (an )n≥0 be a sequence of real numbers such that lim an = l. Prove that
n→∞
∞
xn
lim e−x an = l.
x→∞ n!
n=0
7.77.
(a) Let (an )n≥1 be a sequence of positive real numbers. Prove that
7.6 Series of Functions 209
∞
1
x
lim anx = sup an .
x→∞ n≥1
n=1
1
(b) Application. Calculate lim ζ (x) x .
x→∞
7.78. [O. Furdui, M. Ivan, A. Sîntămărian, 28 December 2016]
(a) Prove that

∞ x n 1 1
=x dt, x ∈ R.
n 0 t xt
n=1
(b) [56] Prove that

1
∞ x n x
−1
lim = ee .
x→∞ n
n=1
7.79. [O. Furdui, 29 December 2016]
(a) Prove that

∞ x n 1
(−1)n−1 =x t xt dt, x ∈ R.
n 0
n=1
(b) Prove that

∞ x n
lim (−1)n−1 = 1.
x→∞ n
n=1
7.80. Find the convergence set and the sum of the power series
∞
xn
.
(−1)n n + 1
n=2
7.81. Prove that

∞
1 x
Hn − 1 − x − x2 − · · · − xn =− ln(1−x), x ∈ (−1, 1).
1−x (1 − x)2
n=1
7.82. Prove that

∞ ∞
x x2 x n+m x 2 + 2x x
ex − 1 − − − ··· − = e , x ∈ R.
1! 2! (n + m)! 2
n=0 m=0
7.83. The Dilogarithm function Li2 is the special function defined by

∞
zn z ln(1 − t)
Li2 (z) = =− dt, |z| ≤ 1.
n2 0 t
n=1
2
(a) Prove that Li2 (z) + Li2 (1 − z) = π6 − ln z ln(1 − z).
2 2
In particular, Li2 12 = π12 − ln2 2 (see also Problems 3.64 and 7.66).
(b) Application. Calculate
∞
sin2n θ + cos2n θ
, θ ∈ R.
n2
n=1
7.84. [62] Let k ∈ N. Find the convergence set and the sum of the power series
∞
1 1 1
Ckn e − 1 − − − · · · − xn.
1! 2! n!
n=k
Remark 7.5. More generally, it can be proved that, if f has a Maclaurin series with
radius of convergence R > 1, then
∞ 1
1 1 1 1
Ckn e − 1 − − − ··· − f (n) (0) = e1−t t k f (k) (t)dt.
1! 2! n! k! 0
n=k
7.85. Prove that

∞ ∞
1 1 1 2 1
n! e − 1 − − − · · · − =e .
1! 2! n! n · n!
n=0 n=1
7.86. Harmonic–Trigonometric Series

Prove that
(continued)
∞
Hn n 1
x = Li2 (x) + ln2 (1 − x), −1 ≤ x < 1. (7.1)
n 2
n=1
Remark
7.6. Formula (7.1), which is the generating function for the sequence
Hn
n , is also true for complex numbers z, with |z| ≤ 1, z = 1.
n≥1
πi
Replacing x by e k , k ∈ N, in formula (7.1) one obtains the harmonic–
cosine series formula
π 2 k 2 − 6k + 3 1 2 π
∞
Hn nπ
cos = · + ln 2 sin .
n k 24 k2 2 2k
n=1
In particular, when k equals 1, 2, or 3, we obtain the following formulae:

∞
Hn ln2 2 π 2
• (−1)n = − ;
n 2 12
n=1
∞
H2n ln2 2 5π 2
• (−1)n = − ;
n 4 48
n=1
∞
Hn nπ π2
• cos =− .
n 3 36
n=1
Replacing x by eix and −eix in formula (7.1), we obtain the amazing

Fourier series formulae:
∞
Hn π2 πx x2 1 x
• cos(nx) = − + + ln2 2 sin , x ∈ [0, 2π ];
n 24 4 8 2 2
n=1
∞
Hn x2 π2 1 x
• (−1)n cos(nx) = − + ln2 2 cos , x ∈ [−π, π ].
n 8 12 2 2
n=1

7.87. (a) The Generating Function of 1
n ζ (2) − 1 − 1
22
− ··· − 1
n2
Prove that
∞
1 1 xn
ζ (2) − 1 − 2 − · · · − 2
2 n n
n=1
x ln2 (1 − t)
= ln(1 − x) [Li2 (x) − ζ (2)] − Li3 (x) + dt, x ∈ [−1, 1].
0 t
(continued)
∞
ζ (2) − 1 − 1
22
− ··· − 1
n2
= ζ (3)
n
n=1
and when x = −1 we obtain, see also part (a) of Problem 3.81, that
∞
ζ (2) − 1 − 1
22
− ··· − 1
n2 π2
(−1)n = ζ (3) − ln 2.
n 4
n=1
It follows, from the previous formula, that
∞
1+ 1
22
+ ··· + 1
n2 ln 2
(−1)n = ζ (2) − ζ (3).
n 2
n=1

(b) The Generating Function of 1
n ζ (3) − 1 − 1
23
− ··· − 1
n3
Prove that
∞
1 1 xn
ζ (3) − 1 − 3 − · · · − 3
2 n n
n=1
Li22 (x)
= ln(1 − x) [Li3 (x) − ζ (3)] − Li4 (x) + , x ∈ [−1, 1].
2
∞
ζ (3) − 1 − 1
23
− ··· − 1
n3 ζ (4)
=
n 4
n=1
and when x = −1 we obtain, see also part (b) of Problem 3.81, that
∞
ζ (3) − 1 − 1
23
− ··· − 1
n3 19π 4 7
(−1) n
= − ζ (3) ln 2.
n 1440 4
n=1
It follows, from the previous formula, that
∞
1+ 1
23
+ ··· + 1
n3 3 19
(−1)n = ζ (3) ln 2 − ζ (4).
n 4 16
n=1
7.7 Pearls of Series with Tails of Zeta Function Values 213
7.88. A Gem with a Logarithmic Series

Let x ∈ [−1, 1). Prove that
∞
2
x2 xn x+1
ln (1 − x) − x +
2
+ ··· + = ln(1 − x 2 ) − ln2 (1 − x).
2 n x−1
n=1
In particular, when x = −1 we obtain the amazing logarithmic series
∞
2
(−1)2 (−1)n
(−1) + + ··· + − ln2 2 = ln2 2.
2 n
n=1
7.7 Pearls of Series with Tails of Zeta Function Values
7.89. Two Superb Series with the Tail of ζ (2)

Prove that:
∞
1 1 1 1
(a) [61] n ζ (2) − 1 − 2 − · · · − 2 − 1 + = ;
2 n 2n 4
n=1
∞
1 1 1 1 − ln 4
(b) (−1)n n2 ζ (2) − 1 − 2 − · · · − 2 − n + = .
2 n 2 4
n=1
7.90. Series with ζ (2) − 1 − 1

22
− ··· − 1
n2
− 1
n
(a) Prove that

∞
1 1 1
Hn ζ (2) − 1 − 2 − · · · − 2 − = −1.
2 n n
n=1
(b) Prove that

∞
1 1 1 1 1
1+ 2 + · · · + 2 ζ (2)−1− 2 − · · · − 2 − =ζ (2)−2ζ (3).
2 n 2 n n
n=1
(continued)
(c) Open problem Calculate

∞
1 1 1
H2n ζ (2)−1− 2 − · · · − 2 − .
2 n n
n=1
7.91. Prove that

∞
1 1 1 1 ζ (4)
1 + 2 + ··· + 2 ζ (3) − 1 − 3 − · · · − 3 = .
2 n 2 n 2
n=1
See the connection of this exercise with part (b) of Problem 2.56.
Open Problem Since

1 1 1 1 ζ (2) Hn
Hn 1 + 2 + ··· + 2 ζ (3) − 1 − 3 − · · · − 3 ∼ · 2,
2 n 2 n 2 n
it is natural to consider as an open problem the calculation of the series

∞
1 1 1 1
Hn 1 + 2 + · · · + 2 ζ (3) − 1 − 3 − · · · − 3 .
2 n 2 n
n=1
7.92. Prove that:

∞
1 1 1 1 π2
(a) [85] n ζ (3) − 1 − 3 − · · · − 3 − 2 = − ;
2 n 2n 4 12
n=1
∞
1 1 1 1 ζ (3) ζ (2)
(b) n2 ζ (4) − 1 − 4 − · · · − 4 − 3 = + − ;
2 n 3n 9 6 2
n=1
∞
1 1 1 1 ζ (3) ζ (2)
(c) [76] n ζ (5) − 1 − 5 − · · · − 5 − 4 =
3
+ − .
2 n 4n 16 4 2
n=1
7.93. Exotic Series Prove that:

∞
1 1 1 1 2ζ (2) + 1
(a) n2 ζ (3) − 1 − 3 − · · · − − + = ;
2 n3 2 2n 12
n=1
∞
1 1 1 1 6ζ (2) + 1
(b) n3 ζ (4) − 1 − 4 − · · · − − + = .
2 n4 3 2n 24
n=1
A challenge. If x > 1, one can prove that

n

1 1 1 x
lim n 2
n x−1
ζ (x) − − + = ,
n→∞ kx x − 1 2n 12
k=1

n
which shows that nx−1 ζ (x) − 1
kx − 1
x−1 + 1
2n ∼ 1
n2
.
k=1
It is natural to consider, as a challenge, the calculation of the series
∞
1 1 1 1
nk−1
ζ (k) − 1 − k − · · · − k − + ,
2 n k − 1 2n
n=1
where k ∈ N, k ≥ 5.
7.94. Prove that:

n
n(n + 1) 1
(a) kHk = Hn+1 − , n ≥ 1;
2 2
k=1
∞
1 1 1 1 3
(b) nHn ζ (2) − 1 − 2 − · · · − 2
− + 2
= ;
2 n n 2n 8
n=1
∞
1 1 1 ζ (2) 1
(c) nHn ζ (3) − 1 − 3 − · · · − 3
− 2 = − ζ (3) − ;
2 n 2n 4 8
n=1
∞
1 1 ζ (2) 5 5
(d) nHn ζ (4) − 1 − 4 − · · · − 4
=− + ζ (3) − ζ (4).
2 n 4 4 8
n=1
7.95.
(a) Prove that, for all n ≥ 1, the following identity holds:
n(n + 1)
12 − 22 + 32 − · · · + (−1)n+1 n2 = (−1)n+1 .
2
(b) Prove that
∞
1 1 1 1 ln 2 ζ (2)
(−1) n
n ζ (3) − 1 − 3 − · · · − 3 −
2
= + − .
2 n 2 4 2 4
n=1
(c) Prove that

∞
1 1 ζ (2) 3ζ (3)
(−1)n n2 ζ (4) − 1 − 4 − · · · − 4 = − .
2 n 4 8
n=1
The next problems in this section are about the calculation of exotic series
involving the product of tails of Riemann zeta function values. These exercises have
been motivated by problems 3.22 and 3.45 in [26] and by the results in the papers
[27, 28, 58, 92].
7.96. Fascinating Series Prove that:

∞

1 2 1 1 2
(a) ζ (2) − 1 + 2 + · · · + 2 = 9ζ (5) − 3ζ (2)ζ (3);
n 2 n
n=1
∞ 2
1 1 2ζ (2)
(b) ζ 2 (2) − 1 + 2 + · · · + 2 −
2 n n
n=1
= 2ζ (2) − 3ζ (3) − 52 ζ (4);
(c) Hn enters the scene.
∞

1 1 2 2ζ (2)
Hn ζ (2) − 1 + 2 + · · · + 2 −
2
2 n n
n=1
19
= −2ζ (2) + 3ζ (3) − ζ (4);
4
(continued)
(d) A challenge.
∞

1 1 2
(−1) n
ζ (2) − 1 + 2 + · · · + 2
2
2 n
n=1

38 1 7 ln4 2
=− ζ (4) + 4Li4 + ζ (3) ln 2 − ζ (2) ln2 2 + ;
9 2 2 6
2
(e) Open problem. Since ζ 2 (2) − 1 + 212 + · · · + n12 ∼ 2ζn(2) , it is natural
to consider as an open problem the calculation of the series
∞
2
1 1 2
ζ (2) − 1 + 2 + · · · + 2
2
.
2 n
n=1
7.97. Prove that

∞
2
1 1
ζ 2 (3) − 1 + 3 + · · · + 3 = 10ζ (5) − 4ζ (2)ζ (3) − ζ 2 (3).
2 n
n=1
We mention that, for an integer k ≥ 3, the following formula holds:

∞

1 1 2
ζ (k) − 1 + k + · · · + k
2
2 n
n=1
∞
1 1 1
=2 1 + k + · · · + k − ζ 2 (k) − ζ (2k − 1).
nk−1 2 n
n=1
The last series, which is an Euler series of weight 2k − 1, can be evaluated

in terms of zeta values [22, Theorem 3.1].
7.98. [61] Pearls of Quadratic Series with the Tail of ζ (2)

Prove that:
(a) An alternating sum.

∞

1 1 2 1 ζ (2) 5 π 2 ln 2
(−1) n ζ (2)−1− 2 − · · · − 2 −1 = +
n 2
+ ζ (3)− ;
2 n 2 2 8 4
n=1
(continued)
∞

1 1 2 1 3 ζ (2) 3
(b) n ζ (2) − 1 − 2 − · · · − 2 − = − − ζ (3);
2 n n 2 2 2
n=1
∞ 2
1 1 1 2 ζ (2) ζ (3)
(c) n2 ζ (2) − 1 − 2 − · · · − 2 − 1 + =− + + ;
2 n n 3 2 2
n=1
(d) A quadratic series and Apéry’s constant.
∞

1 1 2 5
(n + n) ζ (2) − 1 − 2 − · · · − 2 − 1 = − ζ (3).
2
2 n 6
n=1
7.99. Prove that:

∞
1 1 1 2
(a) [27] ζ (2) − 1 − 2 − · · · − 2 = 5ζ (2)ζ (3) − 9ζ (5);
n 2 n
n=1
∞
1 1 1 2 5
(b) ζ (2) − 1 − 2 − · · · − 2 = ζ (2)ζ (3) − ζ (4) + ζ (5);
n+1 2 n 2
n=1
∞
1 1 1 2 23
(c) ζ (2) − 1 − − · · · − = ζ 2 (3) − ζ (6);
n2 22 n2 24
n=1
∞
1 1 1 2 101
(d) ζ (2) − 1 − − · · · − = −ζ 2 (3) + ζ (6)
(n + 1) 2 2 2 n2 24
n=1
5
− ζ (4).
2
7.100. Quadratic Series with Tail of ζ (3)

Prove that:
∞
1 1 2 3
(a) n ζ (3) − 1 − 3 − · · · − 3 = ζ (4) − 3ζ (2)ζ (3) + 5ζ (5);
2 n 4
n=1
∞ 2
1 1 2 ζ (3) 3 5
(b) n2 ζ (3) − 1 − 3 − · · · − 3 = ζ (2) − − ζ (4) − ζ (5)
2 n 3 3 4 3
n=1
+ζ (2)ζ (3);
(continued)
(c) A fabulous series.

∞

1 1 2 1 ζ (3) 3 3
n ζ (3)−1− 3 − · · · − 3 −
3
= − ζ (2)+ + ζ (4)+ .
2 n 4n 2 8 16
n=1
7.101. A Quadratic Series with the Tail of ζ (4)

Prove that
∞
1 1 2
n3 ζ (4) − 1 − 4 − · · · − 4
2 n
n=1

1 5
= 2ζ (3) − ζ (4) − 10ζ (5) − ζ (6) + 2ζ 2 (3) + 4ζ (2)ζ (3) .
4 3
7.102.
(a) Prove that

∞
1 1 1 1
n ζ (2) − 1 − 2 − · · · − 2 ζ (3) − 1 − 3 − · · · − 3
2 n 2 n
n=1
1
= (ζ (2) + ζ (3) − 2ζ (4)) .
2
(b) Prove that
∞
1 1 1 1
n ζ (2) − 1 − 2 − · · · − 2 ζ (4) − 1 − 4 − · · · − 4
2 n 2 n
n=1
3 15 3 1
= ζ (2)ζ (3) − ζ (5) + ζ (4) + ζ (3).
2 4 8 2
∞
1 1 1 1 1 1
n + +··· + +··· = ζ (2)+ ζ (3)+ζ (4).
n (n + 1)2
2 n (n + 1)3
3 2 2
n=1
7.103. Open problem. Calculate:

∞
1 1 1 1
(a) (−1) n ζ (2) − 1 − 2 − · · · − 2
n
ζ (3) − 1 − 3 − · · · − 3 ;
2 n 2 n
n=1
∞
1 1 1 1
(b) (−1) n ζ (2) − 1 − 2 − · · · − 2
n
ζ (4) − 1 − 4 − · · · − 4 ;
2 n 2 n
n=1
∞
1 1 1 1
(c) (−1) n ζ (k) − 1 − k − · · · − k
n
ζ (m) − 1 − m − · · · − m ,
2 n 2 n
n=1
where k, m ≥ 2.
7.104. (a) A challenge. Prove that

∞
1 1 1 1
Hn ζ (2) − 1 − 2 − · · · − 2 ζ (3) − 1 − 3 − · · · − 3
2 n 2 n
n=1
3
= −2ζ (4) − ζ (5) + 2ζ (2)ζ (3).
2
Open problems. Calculate:
∞
1 1 1 1
(b) Hn ζ (2) − 1 − 2 − · · · − 2 ζ (4) − 1 − 4 − · · · − 4 ;
2 n 2 n
n=1
∞
1 1 1 1
(c) (−1)n Hn ζ (2) − 1 − 2 − · · · − 2 ζ (3) − 1 − 3 − · · · − 3 ;
2 n 2 n
n=1
∞
1 1 1 1
(d) (−1)n Hn ζ (2) − 1 − 2 − · · · − 2 ζ (4) − 1 − 4 − · · · − 4 .
2 n 2 n
n=1
(e) More generally, calculate in closed form the harmonic series
∞
1 1 1 1
Hn ζ (k) − 1 − k − · · · − k ζ (m) − 1 − m − · · · − m
2 n 2 n
n=1
and
∞
1 1 1 1
(−1)n Hn ζ (k)−1− k − · · · − k ζ (m)−1− m − · · · − m ,
2 n 2 n
n=1
for integers k, m > 1.

A Remarkable Quadratic Series

We prove that
∞ 2
1 1
+ + ··· = 3ζ (3). (7.2)
n2 (n + 1)2
n=1
Solution I. Using Abel’s summation formula, with an = 1 and bn =

2
1
n2
+ (n+1)
1
2 + ··· , we have that
∞ 2 ∞
1 1 1 1 2
+ + ··· = + + ···
n2 (n + 1)2 n n2 (n + 1)2
n=1 n=1
∞
1 1 1
=2 + + ··· − ζ (3)
n n2 (n + 1)2
n=1
∞
(∗) Hn
=2 − ζ (3)
n2
n=1
= 3ζ (3).
We used at step (∗) Abel’s summation formula with an = 1

n and bn =
1
n2
+ (n+1)
1
2 + ···.
Solution II. We have

∞ 2 ∞ 2
1 1 1 1
+ + ··· = ζ (2) + 2
+ + ···
n2 (n + 1)2 n2 (n + 1)2
n=1 n=2
∞ 2
5 1 1
= ζ (4)+ + + ···
2 (m + 1) (m + 2)2
2
m=1
∞ 2
5 1 1
= ζ (4) + ζ (2) − 1 − − ··· − 2
2 22 m
m=1
(∗∗)5 5
= ζ (4) + 3ζ (3) − ζ (4)
2 2
= 3ζ (3),
where the equality (∗∗) follows based on problem 3.22 in [26].

7.105.
(a) Prove that

∞
1 1 1 7
+ + ··· = ζ (4).
n2 n2 (n + 1)2 4
n=1
(b) A generalization. Let k > 1. Prove that

∞
1 1 1 ζ 2 (k)+ζ (2k)
+ +··· = .
nk n (n+1)k
k 2
n=1
Remark 7.7. We mention that, for k > 1, the following equality holds:
∞ ∞
1+ 1
2k
+ ··· + 1
nk 1 1 1 ζ 2 (k) + ζ (2k)
= + + ··· = .
nk nk nk (n + 1)k 2
n=1 n=1
7.106. Prove that

∞
1 1 1 1 7
+ + ··· + + ··· = 3ζ (3)− ζ (4).
n2 (n + 1)2 (n + 1)2 (n + 2)2 4
n=1
7.107. Series with Tails of ζ (k)
(a) Prove that

∞
1 1 1 1 1 9
+ + ··· + + ··· = ζ (6).
n2 n2 (n + 1)2 (n + 1)2 (n + 2)2 8
n=1
(b) A generalization. Let k > 1. Prove that

∞
1 1 1 1 1 ζ 3 (k)−ζ (3k)
+ +··· + +··· = .
nk nk (n + 1)k (n + 1)k (n + 2)k 3
n=1
A Gem with a Cubic Series

We prove that
∞ 3
1 1 17
(2n − 1) + + ··· = 9ζ (3) − ζ (4).
n2 (n + 1)2 4
n=1
Solution. We apply Abel’s summation formula with an = 2n − 1 and bn =

3
1
n 2 + 1
(n+1) 2 + · · · . A calculation shows that
2
1 1 1
bn − bn+1 = 2 + + ···
n n2 (n + 1)2

1 1 1 1
+ + + · · · + + · · ·
n2 (n + 1)2 (n + 1)2 (n + 2)2
2
1 1
+ + + ···
(n + 1)2 (n + 2)2
and it follows
∞ 3
1 1
(2n − 1) + + ···
n2 (n + 1)2
n=1
∞ 2
1 1
= + + ···
n2 (n + 1)2
n=1
∞
1 1 1 1
+ + + ··· + + ···
n2 (n + 1)2 (n + 1)2 (n + 2)2
n=1
∞ 2
1 1
+ + + ···
(n + 1)2 (n + 2)2
n=1
7 5
= 3ζ (3) + 3ζ (3) − ζ (4) + 3ζ (3) − ζ (4)
4 2
17
= 9ζ (3) − ζ (4).
4
(continued)
We used in our calculations formula (7.2), Problem 7.106, and the series
∞ 1 2
n=1 (n+1)2 + 1
(n+2)2
+ · · · = 3ζ (3) − 52 ζ (4), which follows from the
second solution of formula (7.2).
Remark 7.8. We mention that the cubic series

∞ 3
1 1 9 17 25 9
n + + ··· = ζ (3) − ζ (4) − ζ (5) + ζ (2)ζ (3)
n2 (n + 1)2 2 8 4 2
n=1
was calculated in [27].

Subtracting the previous two cubic series one obtains the remarkable
formula
∞ 3
1 1 25
+ + ··· = 9ζ (2)ζ (3) − ζ (5),
n2 (n + 1)2 2
n=1
communicated to the second author, without proof, in 2014 by C.I. Vălean.
7.108. Prove that:

∞
1 1 1 2
(a) + + · · · − = ζ (2) − ζ (3);
n2 (n + 1)2 n
n=1
∞
1 1 1 2 3 1 1
(b) n + + · · · − = ζ (3) − ζ (2) − .
n2 (n + 1)2 n 2 2 2
n=1
Part (b) of the problem is related to part (a) of Problem 2.82.
7.109. Prove that:

∞ 2
1 1 π4
(a) [74] (2n − 1) + + · · · = ;
n3 (n + 1)3 60
n=1
∞
2
1 1 1
(b) [77] (2n − 1) + + · · · − 2 = 3.
n2 (n + 1)2 n
n=1
7.110.
(a) A Quadratic Series, ζ (2) and ζ (3)

Prove that
∞ 2
1 1
(3n2 − 3n + 1) + + ··· = 2ζ (2) − ζ (3).
n3 (n + 1)3
n=1
(b) A Quadratic Series, ζ (3) and ζ (4)

Prove that
∞ 2
1 1
(4n3 − 6n2 + 4n − 1) + + ··· = 2ζ (3) − ζ (4).
n4 (n + 1)4
n=1
7.111. Two Harmonic Series with the Tail of ζ (3)

Prove that:
∞
1 1 5
(a) Hn + + · · · = 2ζ (3) − ζ (2) + ζ (4);
n3 (n + 1)3 4
n=1
∞ 2
1 1 3
(b) n + + · · · = ζ (4) − 5ζ (5) + 3ζ (2)ζ (3).
n3 (n + 1)3 4
n=1
7.112. Prove that

∞
Hn 1 1
+ + ··· = 3ζ (4).
n n2 (n + 1)2
n=1
Remark 7.9. For a similar exercise see problem 3.62 in [26].
7.113. Prove that:

∞
Hn 1 1 ζ (4)
(a) ζ (2) − 1 − 2 − · · · − 2 = ;
n(n + 1) 2 n 2
n=1
(continued)
∞
Hn 1 1 15
(b) ζ (3) − 1 − 3 − · · · − 3 = − ζ (5) + 4ζ (2)ζ (3).
n(n + 1) 2 n 2
n=1
7.114. A Challenging Alternating Zeta Series
(a) Prove that
1 1 1 xn
ζ (2) − 1 − − ··· − 2 = − ln x dx, n ≥ 1.
22 n 0 1−x
(b) Prove that

∞
(−1)n 1 1 π2
ζ (2) − 1 − 2 − · · · − 2 = ζ (3) − ln 2. (7.3)
n 2 n 4
n=1
Remark 7.10. This problem was motivated by the calculation, and this is a
very challenging problem, of the quadratic alternating series
∞ 2
1 1
(−1)n n + + ···
n2 (n + 1)2
n=1

49 5 π2 7 ln4 2 1 ln2 2ζ (2)
=− ζ (4)+ ζ (3)− ln 2+ ln 2ζ (3)+ +2Li4 − .
16 8 4 4 12 2 2
From part (b) of the problem and the previous formula one obtains the
closely related series
∞
1 1 2
(−1) n ζ (2) − 1 − 2 − · · · − 2
n
2 n
n=1

49 7 ln4 2 1 ζ (2) ln2 2 5 π2
= − ζ (4) + ln 2ζ (3) + + 2Li4 − − ζ (3) + ln 2.
16 4 12 2 2 8 4
It is worth mentioning that the non-alternating series (see also part (a) of
Problem 7.63)
∞
1 1 1
ζ (2) − 1 − 2 − · · · − 2 = ζ (3) (7.4)
n 2 n
n=1
(continued)
appears in [26, problem 3.20, p. 142].

Adding and subtracting series (7.3) and (7.4) we obtain the following
interesting formulae:
∞
1 1 1 π2
• ζ (2) − 1 − 2 − · · · − = 2ζ (3) − ln 2;
n 2 (2n)2 4
n=1
∞
1 1 1 π2
• ζ (2) − 1 − 2 − · · · − = ln 2.
2n − 1 2 (2n − 1)2 8
n=1
7.115. Prove that

∞
1 1 7ζ (3) π 2 ln 2
(−1)n Hn ζ (2) − 1 − 2 − · · · − 2 = − .
2 n 16 8
n=1
A challenge. Calculate
∞
1 1
(−1)n Hn2 ζ (2) − 1 − 2 − · · · − 2 .
2 n
n=1
7.116. A Series with the nth Harmonic Number and Tail of ζ (3)
Prove that:
1 1 1 1 xn
(a) ζ (3) − 1 − − · · · − = ln2 x dx, n ≥ 1;
23 n3 2 0 1−x
∞
Hn 1 1 7
(b) [145] ζ (3) − 1 − 3 − · · · − 3 = 2ζ (2)ζ (3) − ζ (5);
n 2 n 2
n=1
(c) Challenges. Calculate
∞
1 1
(−1)n Hn ζ (3) − 1 − 3 − · · · − 3
2 n
n=1
and
∞
1 1
(−1)n Hn2 ζ (3) − 1 − 3 − · · · − 3 .
2 n
n=1
(continued)
Remark 7.11. We mention that the problem involving the calculation of the
series with the nth harmonic number and the tail of ζ (2)
∞
Hn 1 1 7
ζ (2) − 1 − 2 − · · · − 2 = ζ (4)
n 2 n 4
n=1
can be found in [26, problem 3.62, p. 149] and [28].
7.117. Various Series with the Tail of ζ (2)

Open problem. Calculate, in terms of well-known constants, the series:
∞
1 1 2
(a) (−1)n Hn ζ (2) − 1 − 2 − · · · − 2 ;
2 n
n=1
∞ 2
1 1
(b) Hn2 ζ (2) − 1 − 2 − · · · − 2 ;
2 n
n=1
∞
1 1 2
(c) (−1) Hn ζ (2) − 1 − 2 − · · · − 2 .
n 2
2 n
n=1
Remark 7.12. This problem is motivated by the evaluation of the quadratic

harmonic series
∞
1 1 2 19
Hn ζ (2) − 1 − 2 − · · · − 2 = ζ (4) − 3ζ (3),
2 n 4
n=1
which is calculated in [28].
7.118. Alternating Quadratic Zeta Series

Open problem. Let k ∈ N, k ≥ 2, and i = 0, 1, . . . , 2k − 3.
(a) Calculate
∞ 2
1 1
n i
(−1) n + + ··· .
nk (n + 1)k
n=1
(continued)
(b) Calculate
∞
1 1 2
(−1)n ni ζ (k) − 1 − k − · · · − k .
2 n
n=1
The series in part (b), with i = 0 and k = 2, was evaluated by C.I. Vălean.
When i = 1 and k = 2, the series is given in Problem 7.114.
π2
7.119. A Mosaic of Series with 8 −1− 1
32
− ··· − 1
(2n−1)2
(a) Prove that

∞
π2 1 1 1 ln 2 π 2 1
− 1 − 2 − ··· − − = − + .
8 3 (2n − 1)2 4n 2 16 4
n=1
A generalization. Let a > b > 0. Then

∞
1 a−b 1 1 1 1
ζ 2, − − − ··· − −
a2 a (a − b)2 (2a − b)2 (na − b)2 a 2 n
n=1

1 b a−b a−b
= 2 1−γ −ψ 1− − ζ 2, ,
a a a a

where ψ is the Digamma function, ψ(x) = − x1 − γ − ∞ 1
n=1 x+n − 1
n ,
∞
x = 0, −1, −2 . . . and ζ (2, a) = n=0 (n+a)2 , a = 0, −1, −2 . . .,
1
denotes the Hurwitz zeta function.

(b) Prove that
∞
π2 1 1 1 1 π2
n − 1 − 2 − ··· − − = 1− .
8 3 (2n − 1)2 4n 8 8
n=1
(c) Prove that

∞
π2 1 1 1 1 π π2
(−1) n n
− 1 − 2 − ··· − − = + − .
8 3 (2n − 1)2 4n 8 16 32
n=1
(continued)
(d) Prove that

∞
1 π2 1 1 π 2 ln 2 7ζ (3)
− 1 − 2 − ··· − = − .
2n − 1 8 3 (2n − 1)2 8 16
n=1
(e) Prove that

∞
1 π2 1 1 π4
− 1 − 2 − ··· − = .
(2n − 1)2 8 3 (2n − 1)2 384
n=1
This also implies that

∞
1 1 1 5π 4
1 + 2 + ··· + = .
(2n − 1)2 3 (2n − 1)2 384
n=1
(f) Prove that

∞
2
π4 1 1 π2 π2 π4
− 1 + 2 + ··· + − = − .
64 3 (2n − 1)2 16n 16 128
n=1
(g) A Masterpiece of Quadratic Series

Prove that
∞ 2
π2 1 1 π 2 ln 2 π4
− 1 − 2 − ··· − = − .
8 3 (2n − 1)2 8 128
n=1
7.120. Open Problem. Calculate:
∞ 2
π2 1 1
(a) (−1) − 1 − 2 − ··· −
n
;
8 3 (2n − 1)2
n=1
∞ 2 2
π 1 1
(b) n
(−1) n − 1 − 2 − ··· − ;
8 3 (2n − 1)2
n=1
∞ 2
1 π2 1 1
(c) − 1 − 2 − ··· − ;
n 8 3 (2n − 1)2
n=1
(continued)
7.8 Exotic Zeta Series 231
∞ 2
(−1)n π 2 1 1
(d) − 1 − 2 − ··· − ;
n 8 3 (2n − 1)2
n=1
∞ 2 2
π 1 1
(e) Hn − 1 − 2 − ··· − ;
8 3 (2n − 1)2
n=1
∞ 2 2
π 1 1
(f) (−1)n Hn − 1 − 2 − ··· − ;
8 3 (2n − 1)2
n=1
∞ 2 2
π 1 1 π 1 1 1
(g) − 1 − 2 − ··· − − − − · · · − ;
8 3 (2n − 1)2 24 22 42 (2n)2
n=1
(h) Special quadratic sums.
Let k ≥ 2 be an integer. Calculate, in closed form, the series
∞ 2
2k − 1 1 1
ζ (k) k − 1 − k − · · · −
2 3 (2n − 1)k
n=1
and
∞ 2
2k − 1 1 1
(−1) ζ (k) k − 1 − k − · · · −
n
.
2 3 (2n − 1)k
n=1
7.8 Exotic Zeta Series
7.121. Prove that:

∞
1 1 π2
(a) + = ;
(6n + 1)2 (6n + 5)2 9
n=0
∞
1 1 91
(b) + =
ζ (3);
(6n + 1)3 (6n + 5)3
108
n=0
∞
1 1 1 1
(c) + = 1− k 1 − k ζ (k), k > 1;
(6n + 1)k (6n + 5)k 2 3
n=0
∞
1 1 π
(d) (−1)n + = ;
6n + 1 6n + 5 3
n=0
(continued)
∞
1 1 10
(e) (−1) n
+ = G,
(6n + 1)2 (6n + 5)2 9
n=0
where G denotes Catalan’s constant;
∞
1 1 7 3
(f) (−1) n
+ = π ;
(6n + 1)3 (6n + 5) 3 216
n=0
∞
1 1 1
(g) (−1) n
+ = 1 + k β(k), k > 0,
(6n + 1)k (6n + 5)k 3
n=0
where β denotes the Dirichlet beta function.
7.122. Prove that:

∞
n(n + 1)
(a) − 2ζ (3) − 3ζ (4) − · · · − nζ (n + 1) = 2ζ (3) − 1;
2
n=2
∞
n(n + 1) 3
(b) (−1)n − 2ζ (3) − 3ζ (4) − · · · − nζ (n + 1) = ;
2 8
n=2
∞
3 π2 11
(c) + n − ζ (2) − ζ (4) − · · · − ζ (2n) = − ;
4 12 16
n=1
∞
3 π 7
(d) (−1)n + n − ζ (2) − ζ (4) − · · · − ζ (2n) = coth π − ;
4 4 8
n=1
∞
1 1
(e) + n − ζ (3) − ζ (5) − · · · − ζ (2n + 1) = ;
4 16
n=1
∞
ζ (2) ζ (3) ζ (n) π2
(f) ln 2 − 2 − 3 · · · − n = − ln 2;
2 2 2 8
n=2
∞
ζ (2) ζ (3) ζ (n) ln 2
(g) (−1)n Hn − γ − − − ··· − =1−γ − .
2 3 n 2
n=2
7.123. Prove that

∞
1 π2 3
−ζ (2)−ζ (4)− · · · −ζ (2n)+ζ (3)+ζ (5)+ · · · +ζ (2n + 1) = − .
2 12 4
n=1
7.9 Special Differential Equations 233
7.9 Special Differential Equations
7.124.
(a) Let k ∈ R∗ . Find all differentiable functions f : R → R that satisfy the equation
f (x)f (−x) = k, ∀x ∈ R.
(b) Let k be an even positive integer. Find all differentiable functions f : R → R

that satisfy the equation
f (x)f (−x) = x k , ∀x ∈ R.
7.125.
(a) Find all differentiable functions f : R → R that satisfy the equation
f (x) = f (x)f (−x), ∀x ∈ R.
(b) Find all differentiable functions f : R → R that satisfy the equation
f (x) = f 2 (x)f (−x), ∀x ∈ R.
7.126.
(a) Find all differentiable functions f : R → R, with f (0) = 1, which satisfy the
equation
f (x) = f 2 (−x)f (x), ∀x ∈ R.
(b) Find all differentiable functions f : R → R, with f (0) = 1, which satisfy the
equation
f (x) = x 2 f 2 (−x)f (x), ∀x ∈ R.
7.127. Find all differentiable functions f : R → R that satisfy the equation
f (x) = f 3 (−x)f (x), ∀x ∈ R.
7.128. Find all continuous functions f : R → R such that
1 y f (x) + f (y)
f (t)dt = , ∀ x, y ∈ R, x = y.
y−x x 2
7.129. Find all differentiable functions f : R → R such that
f (x) − f (y) f (x) + f (y)

= , ∀ x, y ∈ R, x = y.
x−y 2

f (x) − f (y) x+y
=f , ∀ x, y ∈ R, x = y.
x−y 2

f (x) − f (y) x−y
=f , ∀ x, y ∈ R, x = y.
x−y 2
7.132. [87] Find all continuous functions f : R → R such that
0 x
f (t)dt + tf (x − t)dt = x, ∀x ∈ R.
−x 0
7.10 Inequalities
Two New Proofs of Nesbitt’s Inequality

Method I. Using the AM − GM inequality.
If a, b, c are positive real numbers, then
a b c 3
+ + ≥ .
b+c a+c a+b 2
A new proof of this inequality is based on a method that combines integral

calculus with the AM − GM inequality. The inequality is equivalent to
1 1 1 9
+ + ≥ .
b+c a+c a+b 2(a + b + c)
We have that
(continued)
7.10 Inequalities 235
1 1 1 1
+ + = x b+c−1 + x a+c−1 + x a+b−1 dx
b+c a+c a+b 0
x b+c + x a+c + x a+b

1
= dx
0 x
√
(∗) 1 3 x 2(a+b+c)
≥3 dx
0 x
9
= .
2(a + b + c)
We used at step (∗) the AM − GM inequality.
Generalization. Let n ∈ N, n ≥ 2, and let x1 , x2 , . . . , xn be positive real

numbers. The following inequality holds true:
x1 x2 xn n
+ +· · ·+ ≥ .
x2 +x3 + · · · + xn x1 +x3 + · · · + xn x1 +x2 + · · · + xn−1 n−1
The inequality is equivalent to
1 1 1 n2
+ + ··· + ≥ ,
S − x1 S − x2 S − xn (n − 1)S
where S = x1 + x2 + · · · + xn .
We have that
1 1 1 1
+ + ··· + = t S−x1 −1 + · · · + t S−xn −1 dt
S − x1 S − x2 S − xn 0
(∗) 1 n−1
≥n t n S−1 dt
0
n2
= .
(n − 1)S
We used at step (∗) the AM − GM inequality.
Method II. Using the convexity of the exponential function.

Let S = a + b + c. We have
(continued)
a b c ∞
+ + = ae−(b+c)x + be−(a+c)x + ce−(a+b)x dx
b+c a+c a+b 0
∞ a b −(a+c)x c −(a+b)x

−(b+c)x
=S e + e + e dx
0 S S S
(∗∗) ∞ a(b+c)+b(a+c)+c(a+b)
≥ S e− S x
dx
0
S2
=
2(ab + bc + ac)
3
≥ .
2
We used at step (**) the convexity of the exponential function.
7.133. Prove that if a1 , a2 , . . . , an ∈ R∗ , then

n n
ai aj
≥ 0.
a2
i=1 j =1 i
+ aj2
Remark 7.13. More generally, if m, k ∈ N and a1 , a2 , . . . , an ∈ R∗ , then

n n aim ajm
≥ 0.
i=1 j =1
aik + ajk
7.134. Prove that if a1 , a2 , . . . , an ∈ R, then
n n
n
2
ij
ai aj ≥ ai .
i+j −1
i=1 j =1 i=1
7.11 Fabulous Integrals
7.135. A Fabulous Logarithmic Integral

Prove that
(continued)
7.11 Fabulous Integrals 237

1 ln(x a + (1 − x)a ) π2 1
dx = −a , a > 0.
0 x 12 a
Remark 7.14. The following particular cases are worth mentioning:

√ √
√ 1 + (1 − x) 2−1 )
ln(x 2−1
• a = 2−1 ⇒ dx = ζ (2);
0 √ x √
√ 1 ln(x 37−6 + (1 − x) 37−6 )
• a = 37 − 6 ⇒ dx = π 2 ;
x
√ 0
2
• a = k 2 + 1 − k, k ∈ N, the integral equals kπ6 .
7.136. [36] The Stirling numbers of the first kind s(n, k) are defined by the
generating function
n
z(z + 1)(z + 2) · · · (z + n − 1) = (−1)n+k s(n, k)zk .
k=0
We have that s(0, 0) = 1, s(n, 0) = 0, for all n ∈ N and s(n, n) = 1.

Let n ≥ 1 be an integer. Prove that
1 ln(1 − x) n n−1
dx = n (−1)k−1 s(n − 1, k)ζ (n + 1 − k).
0 x
k=0
7.137. A Harmonic Integral

Prove that
∞ Hx
dx = ζ (2),
1 x2
where Hn denotes the nth harmonic number and x is the floor of x.
7.138. [83] Let m ≥ 0 be an integer. Prove that
(continued)
1 xm ζ (2) + ζ (3) + · · · + ζ (m + 2)
" # dx = − 1,
0 1 m+1
x
∞
7.139. Sophomore’s Dream for Integrals of the Form 1 f (x)dx
Prove that:
∞ 1 1

(a) e − 1 − − ··· − dx = 1;
1 1! x!
∞ 1 1 1
(b) {x} e − 1 − − · · · − dx = ;
1! x! 2
∞
1
1 1 e+1
(c) x e − 1 − − ··· − dx = ;
1 1! x! 2
∞ {e x!}
(d) dx = 1;
1 x!
∞ 1 1 1
(e) sinh 1 − 1 − − · · · − dx = ;
1 3! 2x − 1!
2e
∞ 1 1
x
(f) (−1) e − 1 − − ··· − dx = 1 − cosh 1.
1 1! x!

We record as gems the proofs of two famous harmonic series ∞ Hn
n=1 n2 = 2ζ (3)
∞ Hn 4
and n=1 n3 = π72 . The first series is a result of Leonhard Euler and the second
series is attributed to the German mathematician Christian Goldbach.
A Polylogarithm Integral and a Famous Series of Euler

We prove that
∞
1 ζ (2) − Li2 (x) Hn
2ζ (3) = dx = .
0 1−x n2
n=1
We calculate the preceding integral by two different methods.

We have
(continued)
∞
ζ (2) − Li2 (x) 1 + x + · · · + x n−1
=
1−x n2
n=1
and it follows that

1 ζ (2) − Li2 (x) 1 ∞ 1 + x + · · · + x n−1
dx = dx
0 1−x 0 n=1 n2
∞ 1
1
= (1 + x + x 2 + · · · + x n−1 )dx
n2 0
n=1
∞
Hn
= .
n2
n=1
We calculate the integral by parts and we have
1 ζ (2)−Li2 (x) /1 1 ln2 (1 − x)

dx = −(ζ (2)−Li2 (x)) ln(1 − x)/0 + dx
0 1−x 0 x
1 ln2 x
= dx
0 1−x
∞
1
= 2
ln x x n
dx
0 n=0
∞ 1
= x n ln2 x dx
n=0 0
∞
2
=
(n + 1)3
n=0
= 2ζ (3).
The Euler series formula is proved.
A Polylogarithm Integral and a Famous Series of Goldbach

We prove that
(continued)
∞
π4 1 ζ (3) − Li3 (x) Hn
= dx = .
72 0 1−x n3
n=1
We calculate the preceding integral by two different methods.

We have
∞
ζ (3) − Li3 (x) 1 + x + · · · + x n−1
=
1−x n3
n=1
and it follows that

1 ζ (3) − Li3 (x) 1 ∞ 1 + x + · · · + x n−1
dx = dx
0 1−x 0 n=1 n3
∞ 1
1
= (1 + x + x 2 + · · · + x n−1 )dx
n3 0
n=1
∞
Hn
= .
n3
n=1
We calculate the integral by parts and we have
1 ζ (3) − Li3 (x) /1 1 ln(1 − x)

dx = −(ζ (3) − Li3 (x)) ln(1 − x)/0 − Li2 (x) dx
0 1−x 0 x
1 ln(1 − x)
=− Li2 (x) dx
0 x
/
Li22 (x) //1
= /
2 0
ζ 2 (2)
=
2
π4
= .
72
The Goldbach series formula is proved.
7.140. Euler and Goldbach Related Series
(a) An Euler related series.

Prove that
∞
H2n 1 ζ (2) − Li2 (x 2 ) 11
= dx = ζ (3).
n2 0 1−x 4
n=1

More generally, one can prove that if k ∈ N, then
∞
Hkn 1 ln(1 − x) ln(1 − x k )
=k dx
n2 0 x
n=1
k 1 ln2 (1 + x + · · · + x k−1 )
= (k + 1)ζ (3) − dx.
2 0 x
We leave, as a challenge, the calculation of the previous logarithmic

integrals that lead to the evaluation of the series ∞ Hkn
n=1 n2 .
(b) Prove, using the identity On = H2n − 12 Hn , that
∞
On 7
2
= ζ (3).
n 4
n=1
See also part (c) of Problem 3.65.

(c) A Goldbach related series.

∞
H2n 1 1
= −6ζ (4) + 7ζ (3) ln 2 − 2ζ (2) ln2 2 + ln4 2 + 8Li4 .
n3 3 2
n=1
7.141. A Series with Double Factorials and H2n − Hn − ln 2
(a) Prove that

π
2 π (2n − 1)!!
sin2n x ln sin x dx = · (H2n − Hn − ln 2) , n ≥ 1.
0 2 (2n)!!
(continued)
(b) Prove that

∞
(2n − 1)!!
(H2n − Hn − ln 2) = ln 2 − 1.
(2n)!!
n=1
For other series involving the term H2n −Hn −ln 2, see also Problems 7.69
and 7.70.
7.142. A Series with Double Factorials and H2n+1 − Hn − ln 2
(a) Prove that

π
2 (2n)!!
sin2n+1 x ln sin x dx = (ln 2 + Hn − H2n+1 ) , n ≥ 0.
0 (2n + 1)!!
(b) Prove that

∞
(2n)!!
(H2n+1 − Hn − ln 2) = ln 2,
(2n + 1)!!
n=0
with the convention H0 = 0.
7.143. [O. Furdui, Problem 4, SEEMOUS 2018, Iaşi, Romania]
(a) Let f : R → R be a polynomial function. Prove that

∞
e−x f (x)dx = f (0) + f (0) + f (0) + · · · .
0
(b) Let f be a function that has a Maclaurin series expansion with radius of
∞
convergence R = ∞. Prove that if f (n) (0) converges absolutely, then
∞ −x n=0
0 e f (x)dx converges and
∞ ∞
f (n) (0) = e−x f (x)dx.
n=0 0
7.144. [O. Furdui, 23 May 2018]
(a) Let f be a function that has a Maclaurin series expansion with radius of

∞
convergence R = ∞. Prove that if f (2n) (0) converges absolutely, then
∞ −x n=0
0 e (f (x) + f (−x))dx converges and
∞ ∞
1
f (2n) (0) = e−x (f (x) + f (−x))dx.
2 0
n=0
(b) Let f be a function that has a Maclaurin series expansion with radius of

∞
convergence R = ∞. Prove that if f (2n−1) (0) converges absolutely,
∞ n=1
then 0 e−x (f (x) − f (−x))dx converges and
∞ ∞
1
f (2n−1) (0) = e−x (f (x) − f (−x))dx.
2 0
n=1
An Artistry of Quadratic Series: Two New
Proofs of Sandham–Yeung Series 8
Truth is ever to be found in simplicity, and not in the

multiplicity and confusion of things.
Sir Isaac Newton (1642–1727)
In this chapter we give two new proofs of the following remarkable series formula:
∞
Hn2 17
2
= ζ (4), (8.1)
n 4
n=1
where Hn = 1 + 12 + · · · + n1 denotes the nth harmonic number.

This formula has an interesting history. It was the first quadratic series introduced
in the literature by H.F. Sandham, in 1948, as a problem in the American Mathemat-
ical Monthly [125]. Apparently, the series went unnoticed. D. Castellanos recorded
it in his survey article [11, p. 86], attributed it rightly to Sandham, but with a
wrong entry in the bibliography. P.J. De Doelder [14] evaluated the associated series
∞ Hn2 11π 4
n=1 (n+1)2 = 360 without any reference to Sandham’s series. In April 1993, the
series was discovered numerically by Enrico Au-Yeung, an undergraduate student
in the Faculty of Mathematics in Waterloo, and proved rigorously by D. Borwein
and J. Borwein in [8], who used Parseval’s theorem to prove it. Formula (8.1) was
rediscovered by P. Freitas as Proposition A.1 in the appendix section of [23]. Freitas
proved it by calculating a double integral involving a logarithmic function. This
formula was revived and brought into light by C.I. Vălean and O. Furdui [146], who
proved it by calculating a special integral involving a quadratic logarithmic function.
The series has also appeared as a problem in [26, Problem 3.70, p. 150] and [122,
Problem 2.6.1, p. 110]. It is clear that this remarkable quadratic harmonic series has
attracted lots of attention lately and has become a classic in the theory of nonlinear
harmonic series.
246 8 An Artistry of Quadratic Series: Two New Proofs of Sandham–Yeung Series
8.1 The First Proof
The first proof of formula (8.1) is based on calculating the series

∞
1 H1 H2 Hn π4
2ζ (3) − 2 − 2 − · · · − 2 =
n 1 2 n 30
n=1
in two different ways

[60].Hn
We have, since ∞ n=1 n2 = 2ζ (3) (see part (b) of Problem 7.61), that
∞ ∞ ∞
1 H1 H2 Hn Hn+m
2ζ (3) − 2 − 2 − · · · − 2 = .
n 1 2 n n(n + m)2
n=1 n=1 m=1
It follows, based on symmetry reasons, that

∞ ∞ ∞ ∞
Hn+m Hn+m
= ,
n(n + m)2 m(n + m)2
n=1 m=1 n=1 m=1
and this implies

∞ ∞ ∞ ∞ ∞ ∞
Hn+m 1 Hn+m Hn+m 1 Hn+m
2
= + = .
n(n+m) 2 m(n+m) n(n+m)2
2 2 nm(n+m)
n=1 m=1 n=1 m=1 n=1 m=1
Therefore,
∞ ∞ ∞
1 H1 H2 Hn 1 Hn+m
2ζ (3) − 2 − 2 − · · · − 2 = . (8.2)
n 1 2 n 2 nm(n + m)
n=1 n=1 m=1
∞ Hn
Using (8.2) and the identity 1
k=1 k(k+n) = n (see part (a) of Problem 7.61), we
have
8.1 The First Proof 247
∞ ∞ ∞ ∞
1 H1 H2 Hn 1 1
2ζ (3) − 2 − 2 − · · · − 2 =
n 1 2 n 2 nmk(k + n + m)
n=1 n=1 m=1 k=1
∞ ∞ ∞ 1
1 1
= x n+m+k−1 dx
2 nmk 0
n=1 m=1 k=1
∞ ∞ ∞

1 1 x k−1 xn xm
= dx
2 0 k n m
k=1 n=1 m=1
1 1 ln3 (1 − x)
=− dx
2 0 x
1 1 ln3 x
=− dx
2 0 1−x
∞
1 1
=− 3
ln x x i
dx
2 0 i=0
∞ 1
1
=− x i ln3 x dx
2 0
i=0
∞
1
=3
(i + 1)4
i=0
= 3ζ (4).
Now we calculate the series by using Abel’s summation formula with
1 H1 H2 Hn
an = and bn = 2ζ (3) − 2
− 2 − ··· − 2 ,
n 1 2 n
and we have that
248 8 An Artistry of Quadratic Series: Two New Proofs of Sandham–Yeung Series
∞
1 H1 H2 Hn H1 Hn+1
2ζ (3)− 2 − 2 − · · · − 2 = lim Hn 2ζ (3)− 2 − · · · −
n 1 2 n n→∞ 1 (n+1)2
n=1
∞
Hn+1
+ Hn
(n+1)2
n=1

∞ Hn+1 − 1
n+1 Hn+1
=
(n+1)2
n=1
∞ 2 ∞
Hn+1 Hn+1
= −
n+1 (n+1)3
n=1 n=1
∞ 2 ∞
Hn Hn
= −
n n3
n=1 n=1
∞ 2
Hn π4
= − ,
n 72
n=1
since

H1 H2 Hn+1
lim Hn 2ζ (3) − 2 − 2 − · · · − =0
n→∞ 1 2 (n + 1)2

and ∞ Hn
n=1 n3 =
π4
72 (see A Famous Series of Goldbach Revisited, after Problem
7.63).
It follows that
∞
Hn2 5 17
2
= 3ζ (4) + ζ (4) = ζ (4).
n 4 4
n=1
8.2 The Second Proof
The second proof of formula (8.1) is based on calculating the series

∞
Hn 1 1
+ + ··· = 3ζ (4) (8.3)
n n2 (n + 1)2
n=1
in two different ways.

One method for calculating the series (8.3) is given in the solution of Problem
7.112. Before we calculate the series (8.3) by another method we record a formula
we need in our analysis.
8.2 The Second Proof 249
Let n ≥ 1 be an integer. The following formula holds true:

n
Hk 1 1
2 = Hn2 + 1 + 2 + · · · + 2 . (8.4)
k 2 n
k=1
Formula (8.4) can be proved by mathematical induction or by an application of

Abel’s summation formula with ak = k1 and bk = Hk .
Now we calculate the series (8.3) by Abel’s summation formula with
Hn 1 1
an = and bn = + + ···
n n2 (n + 1)2
and we have that

∞
1
Hn 1
+ + · · ·
n2
n (n + 1)2
n=1

H1 H2 Hn 1 1
= lim + + ··· + + + ···
n→∞ 1 2 n (n + 1)2 (n + 2)2
∞
H1 H2 Hn 1
+ + + ··· +
1 2 n n2
n=1
∞
H1 H2 Hn 1
= + + ··· +
1 2 n n2
n=1
∞ ∞
(8.4) 1 Hn2 1 1+ 1
22
+ ··· + 1
n2
= +
2 n2 2 n2
n=1 n=1
∞
1 Hn2 7
= 2
+ ζ (4),
2 n 8
n=1
since (see Remark 7.7 with k = 2, after Problem 7.105)
∞
1+ 1
22
+ ··· + 1
n2 7
= ζ (4).
n2 4
n=1
∞ Hn2 ∞ Hn2
It follows that 3ζ (4) = 1
2 n=1 n2 + 78 ζ (4) ⇒ n=1 n2 = 17
4 ζ (4).
Part II
Solutions
Sequences of Real Numbers
9

n!en
1.1. Let xn = n+ 21
, n ≥ 1. We have
n
xn+1 e
= n+ 1 , n ≥ 1.
xn 2
1+ n
1
x+ 1
2
One can prove that the function f (x) = 1 + x1 is strictly decreasing on
[1, ∞). Since lim f (x) = e, we get that f (x) > e, for all x ∈ [1, ∞). Therefore,
x→∞
xn+1
xn < 1, for all n ≥ 1, and this implies that the sequence (xn )n≥1 is strictly
decreasing.

n
9k 2 +12k+5
1.2. Let xn = (3k+2)! , n ≥ 1. We have
k=1
n 3n+2
1 1 1 1 1 1
xn = + + = −1− − , n ≥ 1.
(3k)! (3k + 1)! (3k + 2)! k! 1! 2!
k=1 k=0
Therefore, lim xn = e − 52 .
n→∞
1.3. We have
n n+1
1 1 1
xn = =3− − , n ≥ 1.
k(k + 1)(k + 1)! k! (n + 1)(n + 1)!
k=1 k=0
Therefore, lim xn = 3 − e.
n→∞
0 1
1.4. We have xn = 1
2
3
2 − 2n+3
(n+1)(n+2) , n ≥ 1. It follows that

1 n 2n + 3 n
lim 2xn − = lim 1 − = e−2 .
n→∞ 2 n→∞ (n + 1)(n + 2)
1.5. We have
⎧
√
√ ⎪
nk + 3 − nk + 1
k
k ⎨∞ if k < m
⎪
lim √ √ = 27 if k = m
n→∞ m nm + 2 − m nm − 5 ⎪
⎪
⎩0 if k > m.
1.6. a = 32 , b = 94 , and c = 0.
1.7.
(a) The limit equals − 32 .

2n
(b) Let Sn = 1 + 2 + · · · + n, n ≥ 1. We have (−1)k Sk = 2Sn , n ≥ 1. Thus,
k=1

2n
(−1)k Sk
k=1 2Sn
lim = lim

n→∞ 2n+1 n→∞ −2Sn + S2n+1
(−1)k−1 Sk
k=1
n(n + 1) n
= lim = lim = 1.
n→∞ −n(n + 1) + (2n + 1)(n + 1) n→∞ n+1
(c) The limit equals 2.

(d) The limit equals 2.
1.8.
√ √
(a) Prove (see the solution of Problem 7.47) that (2 + 3)n = 1 − (2 − 3)n .
The other parts of the problem can be solved similarly.
√ −1 sin n √
1.9. The limit equals 1. Observe that n
n ≤ n n ≤ n n, ∀n ≥ 1.
1.10.
(a) Let xn = 1
(n+1)2
+ 1
(n+2)2
+ ··· + 1
(2n)2
, n ≥ 1. We have
n n
≤ xn ≤ , n ≥ 1.
(2n)2 (n + 1)2
It follows, based on the Squeeze Theorem, that lim xn = 0.

n→∞
(b) Let xn = √ 12 + √ 12 + · · · + √ 12 , n ≥ 1. We have
n +1 n +2 n +n
n n
√ ≤ xn ≤ √ , n ≥ 1.
n +n
2 n +1
2
It follows, based on the Squeeze Theorem, that lim xn = 1.

n→∞
n
5k 3 +3k 2 +2k+1
(c) Let xn = n4 +k+7
, n ≥ 1. We have
k=1
n n
5k 3 + 3k 2 + 2k + 1 5k 3 + 3k 2 + 2k + 1
≤ xn ≤ , n ≥ 1.
n4 + n + 7 n4 + 8
k=1 k=1
It follows, based on the Squeeze Theorem, that lim xn = 54 .

n→∞
(d) We mention that this limit is a particular case of a problem proposed by L.
Pârşan in Gazeta Matematică [114].
(2n + 1)(2n + 3) · · · (4n + 1)
Let xn = , n ≥ 1. Since n(n + 2) < (n + 1)2 ,
(2n)(2n + 2) · · · (4n)
we have
(4n + 1)2 (2n + 1)2 (2n + 3)2 · · · (4n + 1)2

= < xn2
(2n)(4n) (2n)(2n + 1)2 (2n + 3)2 · · · (4n − 1)2 (4n)
(2n + 1)(2n + 2)2 (2n + 4)2 · · · (4n)2 (4n + 1)
<
(2n)2 (2n + 2)2 · · · (4n)2
(2n + 1)(4n + 1)
= , n ≥ 1.
(2n)2
√
It follows, based on the Squeeze Theorem, that lim xn2 = 2, so lim xn = 2.
n→∞ n→∞
We notice that if we take p = q = 2 in Problem 1.16 or k = 0, a = r = 1,
p = q = 2, and s = 1 in part (i) of Problem 7.19, then we get the limit in (d).
(n + 2)(n + 5) · · · (4n − 1)
(e) Let xn = , n ≥ 1. Since n(n + 3)2 < (n + 2)3 and
(n + 1)(n + 4) · · · (4n − 2)
n2 (n + 3) < (n + 1)3 , we have that
(4n − 1)3 (n + 2)3 (n + 5)3 · · · (4n − 4)3 (4n − 1)3

= < xn3
(n + 1)(4n − 2)2 (n + 1)(n + 2)3 (n + 5)3 · · · (4n − 4)3 (4n − 2)2
(n + 2)2 (n + 4)3 (n + 7)3 · · · (4n − 2)3 (4n − 1)
<
(n + 1)3 (n + 4)3 · · · (4n − 2)3
(n + 2)2 (4n − 1)
= , n ≥ 1.
(n + 1)3
It follows, based on the Squeeze Theorem, that lim xn3 = 4. Thus, lim xn =
√3
n→∞ n→∞
4.
We remark that if we take k = 1, a = r = 1, p = 3, and q = s = 1 in part (i)
of Problem 7.19, then we get the limit in (e).
(f) [128, problem 63, p. 20] We have
C2n
4n 1 (4n)! (n!)2
= · ·
n
4n C2n 4n [(2n)!]2 (2n)!
1 (2n + 1)(2n + 2) · · · (4n)
= ·
4n [(n + 1)(n + 2) · · · (2n)]2
(2n + 1)(2n + 2) · · · (4n)
=
[(2n + 2)(2n + 4) · · · (4n)]2
(2n + 1)(2n + 3) · · · (4n − 1)
= , n ≥ 1.
(2n + 2)(2n + 4) · · · (4n)
√
C2n
4n 2
Using part (d) of the problem, we get that lim n n = .
n→∞ 4 C 2
2n
(g) [128, problem 64, p. 20] We have, for n ≥ 1, that
33n (Cn2n )2 [(2n)!]2 n!(2n)! [(3n)!]2

= 33n · · ·
Cn3n C3n
6n
(n!)4 (3n)! (6n)!
[(n + 1)(n + 2) · · · (2n)]3

= 33n ·
(3n + 1)(3n + 2) · · · (6n)
[(3n + 3)(3n + 6) · · · (6n)]3
=
(3n + 1)(3n + 2) · · · (6n)
(3n + 3)(3n + 6) · · · (6n) (3n + 3)(3n + 6) · · · (6n)
= · .
(3n + 1)(3n + 4) · · · (6n − 2) (3n + 2)(3n + 5) · · · (6n − 1)
Using the idea in the solution of part (e), we get that
(3n + 3)(3n + 6) · · · (6n) √

3
lim = 4
n→∞ (3n + 1)(3n + 4) · · · (6n − 2)
(3n + 3)(3n + 6) · · · (6n) √

3
lim = 2.
n→∞ (3n + 2)(3n + 5) · · · (6n − 1)
Therefore,
33n (Cn2n )2 √
3
√
3
lim = 4 · 2 = 2.
n→∞ Cn3n C3n
6n
(h) The limit equals 0. We have

n n √
1 Ckn 1 k+2
√ = n Ckn
2n k+2 2 k+2
k=0 k=0
n √
1 (n + 1)! k+2 k+1
= · ·
2n (k + 1)!(n − k)! n + 1 k + 2
k=0
n √
1 k+2 k+1
= n Ck+1 ·
2 n+1
n+1 k+2
k=0
√ n
n+2 1
< · Ck+1
n+1 2n n+1
k=0
√
n+2 2n+1 − 1
= · .
n+1 2n
Remark. More generally, if α ≥ 0, then

n
1 Ck
lim √ n = 0.
n→∞ 2n
k=1
k+α
The case α = 0 is a problem given at the entrance examination test of Technical

University of Cluj-Napoca on July 2012.
1.11. For > 0 there exists n0 ∈ N such that l − < 2xn +x2n < l + , ∀n ≥ n0 . Let
k ≥ 1 be an integer. Replacing n by 2n, 22 n, . . . , 2k−1 n in the previous inequalities,
we obtain that
k−1 1 i
1 k−1
k−1 1 i
l − (−1)i − < 2xn + − x2k n < l + (−1)i − ,
2 2 2
i=0 i=0
∀n ≥ n0 . This implies that

2l 1 k 1 1 k−1
1− − −2 1− k < 2xn + − x2k n
3 2 2 2

2l 1 k 1
< 1− − +2 1− k , ∀n ≥ n0 .
3 2 2
Passing to the limit as k → ∞ in the previous inequalities, we have that 2l3 − 2 ≤

2xn ≤ 2l3 + 2 , ∀n ≥ n0 . This implies that 3l − ≤ xn ≤ 3l + , ∀n ≥ n0 ⇒
lim xn = 3l .
n→∞
1.12.
1
(a) Let In = n
x n + (1 − x)n dx. We have
0
1
2 1 √ 3√n
In < n
(1 − x)n + (1 − x)n dx + x + x n dx =
n n
2
0 1
2
4
and
1
2 1 3
In > (1 − x)dx + xdx = .
0 1
2
4
1.13. For part (a) see the solution of Problem 1.12.

1.14. The limit equals 32 .
1.15.
√
(a) The limit equals 2. Observe that xn = k, for 1 + 2 + · · · + (k − 1) + 1 ≤ n ≤
1 + 2 + · · · + k ⇒ xn = k, for (k−1)k
2 + 1 ≤ n ≤ k(k+1) 2 . If n ≥ 1, then there
exists a unique k ≥ 1 such that 1 + 2 + · · · + (k − 1) + 1 ≤ n ≤ 1 + 2 + · · · + k
or 1 + 2 + · · · + (k − 1) < n ≤ 1 + 2 + · · · + k. It follows that √xnn = √kn . We
have

(k − 1)k k(k + 1) 2 1 2
<n≤ ⇒ ≤√ <
2 2 k(k + 1) n (k − 1)k
and it follows that

2 k 2
k ≤ √ <k .
k(k + 1) n (k − 1)k
xn
√
Thus, lim√ = lim √k = 2.
n→∞ n n→∞ n √
3
(b) Similarly, one can prove that the limit equals 3.
1.16. Solution I. The first solution, based on inequalities, is due to the problem
author, Alina Sîntămărian.
(continued)
One can prove that the following inequalities hold:
n(n + p)p−1 < (n + p − 1)p
and
np−1 (n + p) < (n + 1)p ,
for all n ∈ N and p ∈ N, p ≥ 2.

Let
(qn + 1)(qn + p + 1) · · · (qn + (n − 1)p + 1)
xn = , n ≥ 1.
qn(qn + p) · · · (qn + (n − 1)p)
We have
n−1
-
p−1
(qn+1)p−1 (qn+(k−1)p+1)(qn+kp+1) (qn+(n−1)p+1)
p k=1
xn =
(qn)p (qn+p)p · · · (qn+(n−1)p)p
(qn+1)p−1 (qn+p)p · · · (qn+(n−1)p)p (qn+(n−1)p+1)
<
(qn)p (qn+p)p · · · (qn+(n−1)p)p
(qn+1)p−1 (qn+(n−1)p+1)
= ,
(qn)p
for all n ∈ N, n ≥ 2.
Also,
p (qn+1)p (qn+p+1)p · · · (qn+(n−1)p+1)p

xn = n−1
-
p−1
(qn) (qn+(k−1)p) (qn+kp) (qn+(n−1)p)p−1
k=1
(qn+1)p (qn+p+1)p · · · (qn+(n−2)p+1)p (qn+(n−1)p+1)p

>
(qn)(qn+1)p (qn+p+1)p · · · (qn+(n−2)p+1)p (qn+(n−1)p)p−1
(qn+(n−1)p+1)p
= ,
(qn)(qn+(n−1)p)p−1
Therefore, we write
(continued)
(qn + (n − 1)p + 1)p p (qn + 1)p−1 (qn + (n − 1)p + 1)

< xn < ,
(qn)(qn + (n − 1)p)p−1 (qn)p
p p+q
It follows that lim xn = q , since
n→∞
(qn + (n − 1)p + 1)p p+q

lim =
n→∞ (qn)(qn + (n − 1)p)p−1 q
and
(qn + 1)p−1 (qn + (n − 1)p + 1) p+q
lim p
= .
n→∞ (qn) q
Thus,

qn + 1 qn + p + 1 qn + np + 1 p p+q
lim · ··· = .
n→∞ qn qn + p qn + np q
Solution II. The second solution, based on calculus, is due to Dorian Popa.
This solution holds for p, q > 0.
qn + 1 qn + p + 1 qn + np + 1
Let xn = · ··· , n ≥ 1.
qn qn + p qn + np
We have

n
qn + kp + 1
n
1 ln 1 + qn+kp1
ln xn = ln = · ,
qn + kp qn + kp 1
qn+kp
k=0 k=0
for all n ≥ 1.
ln(1+x)
Let f : (0, +∞) → R, f (x) = x . Since f is strictly decreasing on
(0, +∞), we have that

ln 1 + 1
qn ln 1 + 1
qn+kp ln 1 + 1
qn+np
1
≤ 1
≤ 1
,
qn qn+kp qn+np
for all k ∈ {0, 1, . . . , n} and n ≥ 1.

Therefore, we have that

ln 1 + qn 1 n
1 ln 1 + 1
qn+np
n
1
≤ ln xn ≤ ,
1
qn
qn + kp 1
qn+np
qn + kp
k=0 k=0
(continued)
for all n ≥ 1.
It follows that
1 p+q
lim ln xn = ln ,
n→∞ p q
since

ln 1 + 1
qn ln 1 + 1
qn+np
lim 1
= 1, lim 1
=1
n→∞ n→∞
qn qn+np
and
n n
1 1 1 1 1 1 q +p
lim = lim = dx = ln .
n→∞ qn + kp n→∞ n q + p nk 0 q + px p q
k=0 k=0
1
p+q p
Thus, lim xn = q .
n→∞
(2k−1)!!
1.17. Observe that an+2 an+1 = an+1 an + 1, ∀n ≥ 1. It follows that a2k = (2k−2)!! ,
(2k)!!
k ≥ 2, and a2k+1 = (2k−1)!! , k ≥ 1.
1.18. Prove by mathematical induction that xn = 2 cos 2n+1
π
, ∀n ≥ 1. It follows that
π2
lim 4n (2 − xn ) = 4 .
n→∞
1.19. We consider the case when x0 = 1. We have, based on the AM − GM
√
inequality, that 1 + xn > 2 xn and it follows xn+1 > 1, ∀n ≥ 0. Also,
1 < xn+1 = 21+x√ n < 1+xn < xn , ∀n ≥ 0. The sequence is strictly decreasing and
xn 2
bounded, since 1 < xn+1 < xn < x0 , ∀n ≥ 0. Thus, the sequence converges. Let
l = lim xn . Since 1 < xn+1 < 1+x
2 , ∀n ≥ 0, we have that 1 ≤ l ≤ 2 ⇒ l = 1.
n 1+l
n→∞
1.20.
(a) By induction one can prove that xn ≥ 1, ∀n ≥ 1. Using the AM−GM inequality
√ √
we have that xn + n+1 1
+ 1 = xn+1 + 1 ≥ 2 xn+1 , ∀n ≥ 0. It follows that
√ √ n+1 √x n √x , ∀n ≥ 0. This
2 xn+1 ≤ n+2 n+1 + xn ⇒ 2 n+1 ≤ 2 · n+1 + 2
n n+2
n
implies that
n √ 2
1 kk+2 x0
1 ≤ xn+1 ≤ 2 + n+1 , n ≥ 1.
2n+1 k+1 2
k=0
Passing to the limit, as n → ∞, in the previous inequalities, we have that

lim xn = 1.
n→∞
(b) Similarly, one can prove that lim xn = 1.
n→∞
1.21.
1
(a) e; (b) 32 ; (c) 4e ;
(d) Let xn = (2n)(2n+3)···(8n−6)
(n!)2
, n ≥ 1. We have
xn+1 (2n+2)(2n+5) · · · (8n−4)(8n−1)(8n+2) (n!)2

= ·
xn [(n+1)!]2 (2n)(2n+3) · · · (8n−6)
(2n+2)(2n+5) · · · (8n−4) (8n−1)(8n+2)
= · , n ≥ 1.
(2n)(2n+3) · · · (8n−6) (n+1)2
Using the idea from the solution of part (e) of Problem 1.10, we get that
(2n + 2)(2n + 5) · · · (8n − 4) √

3
lim = 2 2.
n→∞ (2n)(2n + 3) · · · (8n − 6)
An application of Cauchy–d’Alembert’s criterion shows that

√ xn+1 √
3
√
3
limn
xn = lim = 2 2 · 64 = 128 2.
n→∞ n→∞ xn
(2n + 1)(2n + 4) · · · (5n − 2)

(e) Let xn = , n ≥ 1. We have
nn
xn+1 (2n + 3)(2n + 6) · · · (5n)(5n + 3) nn

= ·
xn (n + 1) n+1 (2n + 1)(2n + 4) · · · (5n − 2)
(2n + 3)(2n + 6) · · · (5n) 5n + 3 1
= · · n , n ≥ 1.
(2n + 1)(2n + 4) · · · (5n − 2) n + 1 1 + n1
Using the idea from the solution of part (e) of Problem 1.10, we get that
7
(2n + 3)(2n + 6) · · · (5n) 3 25
lim = .
n→∞ (2n + 1)(2n + 4) · · · (5n − 2) 4
It follows, based on Cauchy–d’Alembert’s criterion, that

7 7
√ xn+1 3 25 1 5 3 25
lim n xn = lim = ·5· = .
n→∞ n→∞ xn 4 e e 4
1.22. The limit equals 1. Use Cauchy–d’Alembert’s criterion.

1.23. Solution due
to D.M. Bătineţu-Giurgiu.
√
n
Let Ln = n+1 (n + 1)! − n!, n ≥ 2. We consider the sequence (xn )n≥2 defined
by
√
n+1
(n + 1)!
xn = √n
− 1, n ≥ 2.
n!
We have
√ √
n+1
n+1
(n + 1)! (n + 1)! n n+1 1
lim √ = lim ·√ · = · e · 1 = 1,
n→∞ n
n! n→∞ n+1 n
n! n e
and this implies lim xn = 0.

n→∞
It follows that

√
n xn √n
lim Ln = lim xn n! = lim · n! · ln(1 + xn )
n→∞ n→∞ n→∞ ln(1 + xn )
√
xn √
n
n+1
(n + 1)!
= lim · n! · ln √
n→∞ ln(1 + xn ) n
n!
√
xn n
n! n+1
= lim · · ln n+1
√
n→∞ ln(1 + xn ) n (n + 1)!
1 1
=1· · ln e = .
e e
Remark. We mention that a generalization of Traian Lalescu’s sequence involving
the Gamma function was studied in [24].
1.24. Let (an )n≥0 and (bn )n≥0 be the sequences defined by

n
an = C0n + C2n+1 + C4n+2 + · · · + C2n
2n = n+k , n ≥ 0;
C2k
k=0

n
bn = C1n + C3n+1 + C5n+2 + · · · + C2n−1
2n−1 = C2k−1
n+k−1 , n ≥ 1.
k=1
Using the recurrence formula Ckn = Ckn−1 + Ck−1

n−1 , 0 < k < n, we have
n+1 n n
2k−1
an+1 = n+1+k = 2 +
C2k n+1+k = 2 +
C2k n+k + Cn+k
C2k
k=0 k=1 k=1
n n n
=1+ n+k + 1 +
C2k C2k−1
n+k = an + 1 + C2k−1 2k−2
n+k−1 + Cn+k−1
k=1 k=1 k=1
n n n−1
2p
= an + 1 + C2k−1
n+k−1 + C2k−2
n+k−1 = an + bn + 1 + Cn+p
k=1 k=1 p=0
= 2an + bn ,
and
n+1 n n
bn+1 = C2k−1
n+k = C2k−1
n+k + 1 = C2k−1 2k−2
n+k−1 + Cn+k−1 + 1
k=1 k=1 k=1
n n−1
2p
= C2k−1
n+k−1 + Cn+p + 1
k=1 p=0
= bn + an .
It follows that an+2 − 3an+1 + an = 0, with a0 = 1 and a1 = 2. We determine

the sequence (an )n≥0 using the method for solving sequences defined by recurrence
relations of order two and we get that an = F2n+1 . This implies bn = an+1 − 2an =
F2n+3 − 2F2n+1 = F2n .
1.25.
(a) The limit equals 1. Let fn (x) = f ◦ f ◦ · · · ◦ f (x). We have that fn (x) =

n times
an x+bn
cn x+dn , where an , bn , cn , and dn are sequences of real numbers defined by
n
an bn 3 2 1 5n + 1 5n − 1
= = .
cn dn 2 3 2 5n − 1 5n + 1
(5n +1)x+5n −1
It follows that fn (x) = (5n −1)x+5n +1 ⇒ lim fn (x) = 1.
√ n→∞
5−1
(b) The limit equals 2 . See the solution of part (a).
For information regarding sequences defined by homographic recurrence rela-

tions the reader is referred to [119, section 3.3, p. 129].

0/0 Cases
1
1.26. (a) 1; (b) p+1 ; (c) 25 ; (d) 34 ; (e) 13 ; (f) 1; (g) 1; (h) a2 ; (i) a.
1.27. Since x1 > 0, we get that xn > 0, for all n ≥ 1. The inequality ln(1 + x) < x,
for x > 0, implies xn+1 = ln(1+xn ) < xn , for all n ≥ 1. It follows that the sequence
(xn )n≥1 is strictly decreasing and bounded, hence convergent. Let l = lim xn .
n→∞
Passing to the limit in the recurrence relation we obtain l = ln(1 + l), and this
implies l = 0.
Since (xn )n≥1 is strictly decreasing and lim xn = 0, we have that the sequence
n→∞
(1/xn )n≥1 is strictly increasing and lim x1n = ∞. An application of Stolz–Cesàro’s
n→∞
theorem shows that
n+1−n xn xn+1
lim nxn = lim = lim
n→∞ n→∞ 1
xn+1 − 1
xn
n→∞ xn − xn+1
xn ln(1 + xn ) ln(1 + xn ) xn2

= lim = lim · = 2.
n→∞ xn − ln(1 + xn ) n→∞ xn xn − ln(1 + xn )

1.28. The sequence (xn )n≥1 is bounded because arctan x ∈ − π2 , π2 , for all x ∈ R.
Since x1 > 0, we have xn > 0, for all n ≥ 1. The inequality arctan x < x, for
x > 0, implies xn+1 = arctan xn < xn , for all n ≥ 1. It follows that the sequence
(xn )n≥1 is strictly decreasing. The sequence (xn )n≥1 converges, since it is bounded
and decreases. Let l = lim xn . Passing to the limit in the recurrence relation we
n→∞
obtain l = arctan l, and√it follows l = 0.
To calculate lim nxn we consider the sequence (nxn2 )n≥1 . Since (xn )n≥1 is
n→∞
strictly decreasing and lim xn = 0, we get that the sequence x12 is strictly
n→∞ n n≥1
increasing and lim 1
= ∞. We apply Stolz–Cesàro’s theorem and we have
n→∞ xn2
2
xn2 xn+1
n+1−n
lim nxn2 = lim = lim
n→∞ n→∞
2
1
− 1
xn2
n→∞ xn2 − xn+1
2
xn+1
2
x 2 arctan2 xn arctan xn xn4 3
= lim 2 n = lim · 2 = .
n→∞ x − arctan2 xn n→∞ xn xn − arctan2 xn 2
n
√ 6
Thus, lim nxn = 3
2.
n→∞
1.29. We have xn < xn + e−xn = xn+1 , for all n ≥ 1, and this shows that the
sequence (xn )n≥1 is strictly increasing. Also, we observe that the sequence (xn )n≥1
is unbounded. If (xn )n≥1 would be bounded, then the sequence would converge.
Denoting by l its limit we get, from the recurrence relation, that l = l + e−l , which
is impossible. Since the sequence (xn )n≥1 is strictly increasing and unbounded, it
follows that lim xn = ∞. We apply Stolz–Cesàro’s theorem and we have
n→∞
xn xn+1 − xn 1
lim = lim = lim n = 1,
n→∞ ln n n→∞ ln(n + 1) − ln n n→∞ exn
ln 1 + n1
n
since, based on the same theorem of Stolz–Cesàro, one has
e xn exn+1 − exn −xn

e exn − 1
1
lim = lim = lim exn +e − exn = lim = 1.

n→∞ n n→∞ n + 1 − n n→∞ n→∞ 1
exn
1.31. Let (yn )n≥1 be the sequence defined by

1 1 1
yn = 1 + + + · · · + − ln(n + 1), n ≥ 1.
2 3 n
A calculation shows that:

1 1
xn+1 − xn = + ln 1 − < 0, n ≥ 1;
n+1 n+1

1 1
yn+1 − yn = − ln 1 + > 0, n ≥ 1;
n+1 n+1
n+1
xn − yn = ln > 0, n ≥ 1.
n
It follows 0 < 1 − ln 2 = y1 < y2 < . . . < yn < xn < xn−1 < . . . < x1 = 1,
which implies that (xn )n≥1 is bounded and being strictly decreasing it converges.
We denote its limit by γ .
To calculate lim n(xn − γ ) we apply Stolz–Cesàro’s theorem, the 0/0 case, and
n→∞
we have
n+1 − ln(n+1)+ ln n 1
1
xn −γ xn+1 −xn
lim n(xn −γ )= lim = lim = lim = .
n+1 − n n+1 − n
n→∞ n→∞ 1 n→∞ 1 1 n→∞ 1 1 2
n
Remark. We mention that interesting properties of Euler’s constant are studied in

[129].
1.32.
(a) The limit equals 1. We use the following limit (see Problem 1.31):

1 1 1 1
lim n 1 + + + · · · + − ln n − γ = lim n(Hn − ln n − γ ) = .
n→∞ 2 3 n n→∞ 2
We have
0 1
lim n 2Hn − Hn2 − γ = lim n 2(Hn − ln n − γ ) − (Hn2 − ln(n2 ) − γ )
n→∞ n→∞
= 2 lim n(Hn − ln n−γ )− lim n(Hn2 − ln(n2 )−γ )

n→∞ n→∞
= 1.
(b) The limit equals k2 . See the solution of part (a) of the problem.
1.33. Use that x = eln x , ∀x > 0.

1 Hn n γ 1 On n γ 1 2On n
lim 1+ =e , lim √ 1+ =2e 2 , lim 1+ =4eγ .
n→∞ n n n→∞ n n n→∞ n n
γ
1.34. 12 e 2 .
1.35. Let (yn )n≥1 be the sequence defined by
1 1 1 √
yn = 1 + √ + √ + · · · + √ − 2( n + 1 − 1), n ≥ 1.
2 3 n
A calculation shows that:

1 √ √
xn+1 − xn = √ − 2( n + 1 − n) < 0, n ≥ 1;
n+1
1 √ √
yn+1 − yn = √ − 2( n + 2 − n + 1) > 0, n ≥ 1;
n+1
√ √
xn − yn = 2( n + 1 − n) > 0, n ≥ 1.
√
It follows 0 < 3 − 2 2 = y1 < y2 < . . . < yn < xn < xn−1 < . . . < x1 = 1,
which implies that the sequence (xn )n≥1 is bounded and being strictly decreasing it
converges. We denote√its limit by I .
To calculate lim n(xn − I ) we apply Stolz–Cesàro’s theorem, the 0/0 case,
n→∞
and we have
√ xn − I xn+1 − xn
lim n(xn − I ) = lim = lim
n→∞ n→∞ √1
n
n→∞ √ 1
n+1
− √1n
√ √
√1
n+1
− 2( n + 1 − n) 1
= lim = .
n→∞ √1
n+1
− n
√1 2
1.36. 1.
1.37. 13 .
2
1.38. e π 4 .
1.39. (b) − 12 .
1.40. The limit equals 0. We have, based on part (a) of Problem 1.39, that

1 1 1 1
lim 1 + + · · · + ζ (k) − 1 − k − · · · − k
n→∞ 2 n 2 n

1 + 12 + · · · + n1 1 1
= lim · lim n k−1
ζ (k) − 1 − k − · · · − k
n→∞ nk−1 n→∞ 2 n
= 0.
1.41. Let x > 1 and k ≥ 1. We have
∞ ∞ i+1 ∞
1 1 1 1
= dt = dt >
(x − 1)k x−1 k tx i tx (i + 1)x
i=k i=k
and it follows that
1 1 1
ζ (x) − 1 − − ··· − x < . (9.1)
2x k (x − 1)k x−1
(a) We have, based on inequality (9.1) with k = 1, that 1 < ζ (x) < x
x−1 , ∀x > 1.
This implies that lim ζ (x) = 1.
x→∞
(b) Using inequality (9.1) with k = 2, we have 1
2x < ζ (x) − 1 < 1
2x + 1
(x−1)2x−1
⇒
1 < 2x (ζ (x) − 1) < x+1
x−1 , ∀x > 1. This implies lim 2x (ζ (x) − 1) = 1. It
x→∞
follows that
ζ (2x) − 1 22x (ζ (2x) − 1) 1

lim = lim · lim x = 0.
x→∞ ζ (x) − 1 x→∞ 2x (ζ (x) − 1) x→∞ 2
(c) We have
x x lim 2x (ζ (x)−1)
lim ζ (x)2 = lim (1 + ζ (x) − 1)2 = ex→∞ = e.
x→∞ x→∞
(d) We have, based on Lagrange Mean Value Theorem, that

x 3 x x 3 x
3 x
ζ (x)2 − e = e2 ln ζ (x) − e = 2x ln ζ (x) − 1 · eθ(x) ,
2 2 2
where θ (x) is between 1 and 2x ln ζ (x). Observe that lim θ (x) = 1.

x→∞
We are able to write

x x
3 x 3 x
2 ln ζ (x) − 1 = 2 (ln ζ (x) − ln 1) − 1
2 2
x
3 1
= 2x (ζ (x) − 1) · −1 ,
2 θ1 (x)
where 1 < θ1 (x) < ζ (x). Observe that lim θ1 (x) = 1.

x→∞
It follows that
x x !
3 x 1 x 3 x 3
2 ln ζ (x) − 1 = 2 (ζ (x) − 1)−1 + (1 − θ1 (x)) .
2 θ1 (x) 2 2
A calculation, based on inequality (9.1) with k = 3, shows that

x 3 x 1
lim 2 (ζ (x) − 1) − 1 = lim ζ (x) − 1 − x 3x = 1.
x→∞ 2 x→∞ 2
The inequalities
x x x
3 3 3
0 < (θ1 (x) − 1) < (ζ (x) − 1) = 2x (ζ (x) − 1)
2 2 4
x
imply that lim (θ1 (x) − 1) 32 = 0.
x→∞ x x x
Thus, lim 32 (2x ln ζ (x) − 1) = 1 ⇒ lim 32 ζ (x)2 − e = e.
x→∞ x→∞
1.42. Both limits are equal to 1. The first limit can be calculated by applying the
Stolz–Cesàro theorem, the 0/0 case.
For the calculation of the second limit we consider separately the cases when
n = 2m and n = 2m − 1.
The case n = 2m. Let
2m
1 1
xm = (−1)k − .
k! e
k=0
We have
1
xm+1 − xm = − ,
(2m)!(2m + 2)
which shows the sequence (xm )m≥1 decreases and converges to 0. We apply Stolz–
Cesàro’s theorem, the 0/0 case, and we have
2m
1 1 xm+1 − xm
lim (2m + 1)! (−1) −
k
= lim
(2m+3)! − (2m+1)!
m→∞ k! e m→∞ 1 1
k=0
− (2m)!(2m+2)
1
= lim 2 +10m+5
m→∞ − 4m(2m+3)!
= 1.
The case n = 2m − 1. Let
2m−1
1 1
ym = − (−1)k .
e k!
k=0
We have
1
ym+1 − ym = − ,
(2m − 1)!(2m + 1)
which shows the sequence (ym )m≥1 decreases to 0. We apply Stolz–Cesàro’s

theorem, the 0/0 case, and we have
2m−1

1 1 ym+1 − ym
lim (2m)! − (−1) k
= lim
m→∞ e
k=0
k! m→∞ 1
(2m+2)! − 1
(2m)!
− (2m−1)!(2m+1)
1
= lim
+6m+1
2
m→∞ − 4m(2m+2)!
= 1.
9.3 Wolstenholme Sequences
1.43.
(a) This part follows from part (b) with f (x) = x s . Two solutions of this
problem, one elementary and the other one, which uses Lebesgue Dominated
Convergence Theorem, are given in [98].
(b) If f (1) > 1, then xn > f n (1) ⇒ lim xn = ∞.

n→∞
Let f (1) ≤ 1. We have
(continued)
n−1 n−1
k n ln f 1− nk
xn = f 1−
n
= e .
n
k=0 k=0
We apply Lagrange Mean Value

Theorem to the function ln f and we have
that there exists θ ∈ 1 − nk , 1 such that

k k f (θ ) k f (1)
ln f (1) − ln f 1− = · ≥ · ,
n n f (θ ) n f (1)
where the last inequality holds since the function (ln f ) = ff decreases.

f (1) n ln f 1− nk
It follows that ln f 1 − n ≤ ln f (1) − n · f (1) ⇒ e
k k
≤
−k ff (1)
(1)
f n (1)e , ∀ 1 ≤ k ≤ n. Therefore,
n−1 −n ff (1)
(1)
−k ff (1)
(1) 1−e
0 < xn < f (1) n
e = f (1)
n
. (9.2)
− ff (1)
(1)
k=0 1−e
If f (1) < 1, then we have, based on inequality (9.2), that lim xn = 0.

n→∞
If f (1) = 1, then inequality (9.2) implies xn < 1
, ∀n ≥ 1, and this
1−e−f (1)
shows lim sup xn ≤ 1
.
1−e−f (1)
Let i ≥ 1 be a fixed integer. Then, for all n ≥ i, we have
i−1
n ln f 1− nk
xn ≥ e . (9.3)
k=0
If k is fixed, then

k ln f 1 − nk f 1 − nk − 1
lim n ln f 1 − = −k lim · = −kf (1).
n→∞ n n→∞ f 1 − k − 1 − nk
n
Passing to the limit as n → ∞ in (9.3), we get
i−1
1 − e−if (1)
lim inf xn ≥ e−kf (1)
= , ∀i ≥ 1.
1 − e−f (1)
k=0
Passing to the limit as i → ∞ in the previous inequality, we have that

lim inf xn ≥ 1
−f (1)
.
1−e
(continued)
For studying the sequence (yn )n≥1 see the solution of Problem 1.46, with
a = 1.
1.44. The first limit is a particular case of Problem 1.46, with f (x) = x,
α = 1, β = 0. We calculate the second limit.
n k
Let xn = a k−n nk .
k=1
We have
√
n− 3 n−1 k n k
k k
xn = a k−n
+ a k−n
= S1 (n) + S2 (n).
n √ n
k=1 k=n− 3 n
Observe that
√
n− 3 n−1 k n− √
3n k n− √
3 n √
n− 3 n k
k k−n k k k 1 k
S1 (n)= a k−n
< a < = + .
n n n n n
k=1 k=1 k=1 k=2
√ x x
Let f : [2, n − 3 n] → R, f (x) = n , and g(x) = ln f (x) =

x ln x − x ln n. A calculation shows that g decreases on the interval 2, ne
n √
and increases on the interval e , n − 3 n , so it follows that
k . 3 n:
√ n− √
k 4 n− 3 n
max √ ≤ max , .
2≤k≤n− 3 n n n2 n
This implies that

. 3 n:
√ n− √
1 4 n− 3 n √
S1 (n) < + max 2
, (n − 3 n − 1).
n n n
√3
√
3
Since ln(1 − x) ≤ −x, ∀x ∈ [0, 1), we have that ln 1 − nn ≤ − nn ⇒
√ √3
√ √ n− √
3n √ 1
− 3 n+ √
n− 3 n
(n − 3 n) ln 1 − nn ≤ − 3 n + √ 1
3n ⇒ n ≤ e 3n
.
This implies
√ n− √
3n √
n− 3 n √ 1
− 3 n+ √ √
0 ≤ lim (n− 3 n−1) ≤ lim e 3n
(n− 3 n−1) = 0
n→∞ n n→∞
(continued)
√
√ n− 3 n √
n− 3 n
and it follows that lim n (n − 3
n − 1) = 0.
n→∞
Thus,
. 3 n:
√ n− √
4 n− 3 n √
lim max , (n − 3 n − 1) = 0
n→∞ n2 n
and this shows that lim S1 (n) = 0.

n→∞
We can write
√
n k 3n
k−n k 1 i n−i
S2 (n) = a = 1− .
√ n ai n
k=n− 3 n i=0
i2
n−i
We have (n − i) ln 1 − ni ≤ − ni (n − i) = −i + n ⇒ 1 − ni ≤
1
i2 √
e−i+ n ≤ e−i · e 3 n . It follows that
√
3
1
n 1 ∞ 1
√ 1 √ 1 √ ae
S2 (n) ≤ e 3n
< e 3n
= e 3n
.
(ae)i (ae)i ae − 1
i=0 i=0
The previous inequality implies that lim sup S2 (n) ≤ ae−1

ae
.
√
Let k ≥ 1 be a fixed integer. Then, for n > k, we have
3
k
1 i n−i
S2 (n) > 1− .
ai n
i=0
It follows
k+1
k
1 −i 1− 1
ae
lim inf S2 (n) ≥ e = .
i=0
ai 1− 1
ae
Passing to the limit, as k → ∞, in the previous inequality, we have that

lim inf S2 (n) ≥ ae−1
ae
.
1.45.
(a) Solution I. Use a similar technique to the one given in the solution of
Problem 1.44.
Solution II. We use Tannery’s Theorem for Series.
Tannery’s
n Theorem for Series [106, p. 216] For each natural number
n, let m k=1 ak (n) be a finite sum such that limn→∞ mn = ∞. If, for
each k, limn→∞ ak (n) exists, and there is a convergent series ∞ k=1 Mk
of nonnegative real numbers such that |ak (n)| ≤ Mk , for all n ∈ N and
1 ≤ k ≤ mn , then
mn ∞
lim ak (n) = lim ak (n),
n→∞ n→∞
k=1 k=1
that is, both sides are well defined (the limits and the sums converge) and
are equal.
" #
Let k0 ∈ N be such that sk0 > 1, i.e. k0 = 1s + 1. We have
n sk k0 sk n sk

k k k
= + .
n n n
k=1 k=1 k=k0 +1
k0 sk

It is easy to see that lim k
= 0.
n→∞ k=1 n
We are able to write
n sk n−k0 −1 n−k0 sk0
k i s(n−i) i s(n−i) k0
= 1− = 1− − .
n n n n
k=k0 +1 i=0 i=0
Now we observe that the function f : [i, ∞) → R, f (x) =

s(x−i)
1 − xi is strictly decreasing on [i, ∞). It follows that, for n ≥
k0 + i, we have
s(n−i) sk0
i k0
1− ≤ .
n i + k0
(continued)

∞ s(n−i)
Since the series 1
(i+k0 )sk0
converges and lim 1 − ni = e−si ,
i=0 n→∞
we have, based on Tannery’s Theorem for Series, that
n−k0 −1 s(n−i) ∞
i es
lim 1− = e−si = .
n→∞ n es − 1
i=0 i=0
(b) Observe that

n (a+b)n n ak+bn n (a+b)k
k k k
≤ ≤ .
n n n
k=1 k=1 k=1
1.46. If f (1) > 1, then xn > f αn+β (1) ⇒ lim xn = ∞.

n→∞
Let f (1) ≤ 1. We have
n−1 n−1
1 αn+β k 1 (αn+β) ln f 1− nk
xn = f 1− = e .
ak n ak
k=0 k=0
We apply Lagrange Mean Value

Theorem to the function ln f and we have
that there exists θ ∈ 1 − nk , 1 such that

k k f (θ ) k f (1)
ln f (1) − ln f 1 − = · ≥ · ,
n n f (θ ) n f (1)
where the last inequality holds since the function (ln f ) = ff decreases. It

follows that ln f 1 − nk ≤ ln f (1) − nk · ff (1)
(1)
⇒ (αn + β) ln f 1 − nk ≤
αn+β f (1)
(αn + β) ln f (1) − k · · ≤ (αn + β) ln f (1) − k · α · ff (1)
(1)
. This
n f (1)
k f (1)
(αn+β) ln f 1− n −kα f (1)
implies that e ≤ f αn+β (1)e , ∀ 1 ≤ k ≤ n. Therefore,
f (1)
n
−α
⎛ ⎞k 1− e f (1)
n−1 −α ff (1)(1) a
0 < xn < f αn+β
(1) ⎝e ⎠ =f αn+β
(1) .
f (1)
a −α
f (1)
k=0
1− e
a
(9.4)
If f (1) < 1, then we have, based on inequality (9.4), that lim xn = 0.
n→∞
(continued)
If f (1) = 1, then inequality (9.4) implies that xn < 1

−αf (1)
, ∀n ≥ 1,
1− e a
αf (1)
and it follows lim sup xn ≤ aeαf (1) .
ae −1
Let i ≥ 1 be a fixed integer. Then, for all n ≥ i, we have
i−1
1 (αn+β) ln f 1− nk
xn ≥ e . (9.5)
ak
k=0
If k is fixed, then

k ln f 1 − nk f 1 − nk − 1 αn + β
(αn + β) ln f 1 − = −k · · ·
n f 1 − nk − 1 − nk n

and we have that lim (αn + β) ln f 1 − nk = −kαf (1).
n→∞
Passing to the limit as n → ∞ in (9.5), we get that
k i
e−αf (1)
i−1
e−αf (1) 1− a
lim inf xn ≥ = , ∀ i ≥ 1.
a e−αf (1)
k=0 1− a
aeαf (1)
We let i → ∞ in the previous inequality and we obtain lim inf xn ≥ .
aeαf (1) −1
1.47.
(a) The case when k = 1 is trivial, so we consider only the case when k = 1.
Using the substitution x = t n , we have
1√
n
x +k−1
n
n 1
In = dx = [t (t + k − 1)]n−1 (t + k − 1)dt.
0 k kn 0
√
(k−1)2 +4y−k+1
We make the substitution t (t + k − 1) = y ⇒ t = 2 and we
get, after some calculations, that

n k k−1
In = n y n−1
1+ dy
2k 0 (k − 1)2 + 4y
1 n(k − 1) k y n−1
= + dy.
2 2k n 0 (k − 1)2 + 4y
We integrate by parts and we obtain that
k k−1 k yn
In = + n 3 dy.
k+1 k 0
(k − 1)2 + 4y
It follows lim In = k
k+1 , since
n→∞
1 k yn k
0< 3 dy < .
kn (n + 1)|k − 1|3
0
(k − 1)2 + 4y
Now we prove that the second limit holds. We have
k k−1 k yn
In − = 3 dy.
k+1 kn 0
(k − 1)2 + 4y
We integrate by parts and we get
k k(k − 1) k−1 6 k y n+1

In − = + n · 5 dy.
k+1 (n + 1)(k + 1)3 k n+1 0
(k − 1)2 + 4y
This implies that

k
lim n In −
n→∞ k+1
⎡ ⎤
⎢ k(k − 1)n k−1 6n k y n+1 ⎥
= lim ⎣ + n · 5 dy ⎦
n→∞ (n + 1)(k + 1)3 k n+1 0
(k − 1)2 + 4y
k(k − 1)
= ,
(k + 1)3
since
1 k y n+1 k2
0< 5 dy < .
kn (n + 2)|k − 1|5
0
(k − 1)2 + 4y
(b) Using the substitution x = y n , we have
1 k
n 1 y
n−1
dy
In = √ dx = nk n
.
0
n
x+k−1 0 y+k−1 y+k−1
y (k−1)t
We make the substitution y+k−1 = t ⇒y = 1−t and we obtain, after
some calculations, that
1
k t n−1
In = nk n
dt.
0 1−t
We integrate by parts and we get

1
k k tn
In = − kn dt.
k−1 0 (1 − t)2
It follows lim In = k
k−1 , since
n→∞
1 1
k tn k n+2 k k
0 < kn dt < t n dt = .
0 (1 − t)2 (k − 1)2 0 (n + 1)(k − 1)2
Now we prove that the second limit holds. We have

1
k k tn
− In = k n dt.
k−1 0 (1 − t)2
We integrate by parts and we obtain

1
k 1 k 2k n k t n+1
− In = · − dt.
k−1 n + 1 (k − 1) 2 n+1 0 (1 − t)3
This implies that

1 n
k k
lim n − √ dx
n→∞ k−1 0
n
x+k−1
1

n k 2k n n k t n+1
= lim · − dt
n→∞ n + 1 (k − 1)2 n + 1 0 (1 − t)3
k
= ,
(k − 1)2
since
1 1
k t n+1 k n+3 k k
0<k n
dt < t n+1 dt = .
0 (1 − t)3 (k − 1)3 0 (k − 1)3 (n + 2)
1.48.
(a) Using the substitution x = t n , we have that
4n 1 √ n √ 1 √ n!(n − 1)!
√ 1− n
x dx = 4n n (1 − t)n t n−1 dt = 4n n ·
n 0 0 (2n)!
and the limit follows based on Stirling’s formula.

(b) Let > 0. Since f is continuous at 0, there exists δ > 0 such that |f (x) −
f (0)| < , for |x| < δ. We have
4n 1 √ n 4n δ √ n
√ 1− n
x f (x)dx = √ 1 − n x (f (x) − f (0))dx
n 0 n 0
4n 1 √ n
+√ 1− n
x (f (x) − f (0))dx
n δ
4n 1 √ n
+ f (0) √ 1− n
x dx.
n 0
(9.6)
We can write
/ n /
/ 4 δ √ n / 4n δ √ n
/√ 1 − n x (f (x) − f (0))dx /≤ √ 1 − n x dx
/ n / n
0 0
4n 1 √ n
< √ 1− n
x dx
n 0
and it follows, based on part (a), that

/ n /
/ 4 δ √ n / √
lim / √ 1 − n x (f (x) − f (0))dx /≤ π.
n→∞ / n 0
/
This implies, since > 0 was arbitrary taken, that
4n δ √ n
lim √ 1 − n x (f (x) − f (0))dx = 0. (9.7)
n→∞ n 0
Since f is Riemann integrable, we have that f is bounded, i.e. there exists M > 0
such that |f (x)| ≤ M, ∀x ∈ [0, 1]. It follows
/ n /
/ 4 1 √ n / 4n √
/√ 1 − n x (f (x) − f (0))dx /≤ 2M √ n
(1 − δ)(1 − δ)n .
/ n / n
δ
n √
Passing to the limit in the previous inequality we have, since lim √4 n (1− n δ)n = 0,
n→∞
that
4n 1 √ n
lim √ 1− n
x (f (x) − f (0))dx = 0. (9.8)
n→∞ n δ
Combining (9.6), (9.7), (9.8), and part (a) of the problem, we obtain
4n 1 √ n √
lim √ 1− n
x f (x)dx = π f (0).
n→∞ n 0
1.49. Solution I.
6
1
(a) The limit equals 1 + a. Let yn = n
0 (1 + an x ) dx.
n n We have yn ≤ 1 + an ,
∀n ≥ 1. Let 0 < < 1. We have

6
1 1
n √
yn ≥ n
√ (1 + an x n )n dx ≥ n
√ (1 + an )n dx = (1 + an ) 1 − n .
n n
Thus,
6
n √
(1 + an ) 1 − n ≤ yn ≤ 1 + an , ∀n ≥ 1.
√
Passing to the limit in the previous inequalities we get, since lim n
1− n
=
n→∞
1, that
1 + a ≤ lim yn ≤ 1 + a.
n→∞
Since ∈ (0, 1) was arbitrary taken, we obtain, by letting → 1, that lim yn =

n→∞
1 + a.
Solution II.
(a) Using integration by parts we have
1 n n /1 1
1+an x n dx = x 1+an x n /0 −n2 an x n (1+an x n )n−1 dx
0 0
1 n 1 n−1
= (1+an )n −n2 1+an x n dx+n2 1+an x n dx
0 0
1 n
≥ (1+an )n −n2 1+an x n dx.
0
6
1
This implies n
0 (1 + an x ) dx
n n ≥ √1+an
n 2 , ∀n ≥ 1.
n +1
Since

1 + an n
1
√ ≤ (1 + an x n )n dx ≤ 1 + an , ∀n ≥ 1,
n +1
n 2
0
we have, by passing to the limit as n → ∞, that lim yn = 1 + a.

n→∞
(b) The limit equals 1 + a. We have

1 a1 x+a3 x 3 + · · · +a2n−1 x 2n−1
n
a1 + a3 + · · · + a2n−1
dx ≤ 1+
n
1+ .
0 n n
The AM − GM inequality implies that

7 6
1 a1 x+a3 x 3 +···+a2n−1 x 2n−1 n √
1 n n dx.
0 1+ ≥ 0 1 + a1 a3 · · · a2n−1 x
n n
n dx n
Thus,

1 n 1 a1 x + a3 x 3 + · · · + a2n−1 x 2n−1
n
n √
1 + n a1 a3 · · · a2n−1 x n dx ≤ 1+
n
dx
0 0 n
a1 + a3 + · · · + a2n−1
≤1+ .
n
√
Since lim n a1 a3 · · · a2n−1 = a = lim a1 +a3 +···+a
n
2n−1
, we have, by part
n→∞ n→∞
(a) and by passing to the limit as n → ∞ in the previous inequalities, that the
limit equals 1 + a.
(c) The limit equals 1 + a. See the solution of part (b).
1.51.
(a) Using the substitution nx = y, we have

1 1 n
f ({nx})dx = f ({y}) dy
0 n 0
n−1 k+1
1
= f (y − k) dy
n
k=0 k
n−1 1
y−k=t 1
= f (t)dt
n
k=0 0
1
= f (t)dt.
0
(b) See the solution of part (a) of Problem 1.52.
1.52.
(a) Using the substitution nx = y, we have

1 1 n
f (nx) g ({nx}) dx = f (y) g ({y}) dy
0 n 0
n−1 k+1
1
= f (k)g(y − k)dy
n
k=0 k
n−1 1
y−k=t 1
= f (k) g(t)dt.
n 0
k=0
(b) We apply Stolz–Cesàro’s theorem, the ∞/∞ case, and we have, based on part
(a) of the problem, that
1 1
lim f (nx) g ({nx}) dx
n→∞ nα 0
f (0) + f (1) + · · · + f (n − 1) 1
= lim g(x)dx
n→∞ n1+α 0
f (n) 1
= lim g(x)dx
n→∞ (1 + n)1+α − n1+α 0
f (n) nα 1
= lim · lim g(x)dx
n→∞ nα n→∞ (1 + n)1+α − n1+α 0
L 1
= g(x)dx.
1+α 0
1
1.53. The limit equals 0 f (x)dx. Since f = f + − f − , where f + and f − are
the positive and negative parts of f , without losing the generality, we consider that
f ≥ 0. Using the substitution xn = t, we have
1 n ∞ f ({t})
f dx = n dt
0 x n t2
∞ k+1 f ({t})
=n dt
t2
k=n k
∞ k+1 f (t − k)
=n dt
t2
k=n k
∞ 1
t−k=y f (y)
= n dy
(k + y)2
k=n 0
∞
1 n
= f (y) dy.
0 (k + y)2
k=n
Since
∞ ∞ ∞
n n n
< < ,
(k + 1)2 (k + y)2 k2
k=n k=n k=n
it follows that
∞
n 1 1 n ∞
n 1
f (y)dy < f dx < f (y)dy.
(k + 1)2 0 0 x k2 0
k=n k=n
Passing to the limit in the preceding inequalities and using the limit
∞
1
lim n = 1, p > 0,
n→∞ (k + p)2
k=n
1 2 3 1
we obtain lim 0 f xn dx = 0 f (x)dx.
n→∞
A more general problem can be found in [26, Problem 1.70, p. 13].
1.54.
(a) The limit equals 1. Using the substitution 1

x = t, we have
!n ∞
1 1 {t}n 2 {t}n 2 (t − 1)n 1
dx = dt > dt = dt >
0 x 1 t2 1 t2 1 t2 4(n + 1)
and this implies that

!n
1 1 1
n
√ < dx < 1.
n
4(n + 1) 0 x
(b) The limit equals 1. Since f is continuous on [0, 1], there exist m, M > 0 such
that 0 = m ≤ f (x) ≤ M, ∀x ∈ [0, 1]. It follows that
!n !n !n
1 1 1 1 1 1
dx ≤ f (x)dx ≤
n n n
m M dx,
0 x 0 x 0 x
and the desired limit holds, based on part (a) of the problem.
1.55.
(a) Using the substitution 1

x = t, we have
1 ∞ ∞ i+1 ∞
1 1 1 1 1 1 1
" #n dx = dt = dt = −
0 1 1 t tn
2 in i t2 in i i+1
x i=1 i=1
∞ ∞
1 1 1
= ζ (n + 1) − = ζ (n + 1) − − .
i (i + 1)
n 2 i n (i + 1)
i=1 i=2
It follows
1 ∞
1 1 1 1
lim " #n dx = lim ζ (n + 1) − − lim = .
n→∞ 0 1 n→∞ 2 n→∞ i n (i + 1) 2
x i=2
We used that lim ζ (n + 1) = 1 (see part (a) of Problem 1.41) and

n→∞
∞ ∞
1 1
< = ζ (n) − 1,
i (i + 1)
n in
i=2 i=2

∞
which shows lim n
1
= 0.
n→∞ i=2 i (i+1)
(b) Using the substitution x1 = t, we have
⎛ ⎞
1 ∞ 1
1 ∞ f i+1 f
⎜ 1 ⎟ tn in
f ⎝ " #n ⎠ dx = dt = dt
0 1 1 t2 i t2
x i=1
∞
1 1 1
= f −
in i i+1
i=1
∞
f (1) 1 1 1
= + f − .
2 in i i+1
i=2
Let 4 at50, there exists δ > 0 such that |f (x) −

> 0. Since f is continuous
ln 1δ
f (0)| < , for |x| < δ. Let n0 = ln 2 + 1.
We write
∞ n0
1 1 1 1 1 1
f − = f n − f (0) −
in i i+1 i i i+1
i=2 i=2
∞
1 1 1 f (0)
+ f n −f (0) − + .
i i i+1 2
i=n0 +1
/ /
For n ≥ n0 +1 and i ≥ 2, we have 1
in < 1
2n < 1
2n0 < δ ⇒ /f i1n −f (0)/ <
. It follows that
/ ∞ / ∞
/ 1 1 1 / 1 1
/ f n − f (0) − /≤ − < ,
/ i i i+1 / i i+1
i=n0 +1 i=n0 +1
∞ 0
1
which implies lim f i1n − f (0) 1i − 1
i+1 = 0.
n→∞ i=n +1
0
The continuity of f at 0 shows that
n0
1 1 1
lim f n − f (0) − = 0.
n→∞ i i i+1
i=2
Putting all these together the desired limit holds and the problem is solved.
1.56 and 1.57. See the solution of Problem 1.55.

1.58. Let β ∈ (−1, 0) be the unique solution of the equation x + eαx = 0, x ∈
[−1, 0].
We have
0 n β n 0 n
n x + eαx f (x)dx = n x + eαx f (x)dx + n x + eαx f (x)dx.
−1 −1 β
A calculation shows
/ /
/ β / β
/n x+eαx n
f (x)dx //≤ n||f || |x + eαx |n dx ≤ n||f ||(1 − e−α )n (1 + β),
/
−1 −1
and it follows, by passing to the limit as n → ∞ in the previous inequalities, that
β n
lim n x + eαx f (x)dx = 0.
n→∞ −1
The substitution h(x) = x + eαx = y shows that x = h−1 (y) ⇒ dx =

1
h (h−1 (y))
dy = 1
αh−1 (y)
dy and we have
1+αe
0 n 1 f (h−1 (y))
n x + eαx f (x)dx = n yn −1 (y)
dy.
β 0 1 + αeαh
It follows, based on Problem 7.28, that

0 n 1 f (h−1 (y))
lim n x + eαx f (x)dx = lim n yn −1 (y)
dy
n→∞ β n→∞ 0 1 + αeαh
f (h−1 (1))
= −1 (1)
1 + αeαh
f (0)
= .
1+α
1
1.59. The limit equals 1+α . See the technique used in the solution of Problem 1.58.
√ √
1.60. The limit equals α. Since n α ≤ n x n + α, x ∈ [0, 1], we have
n
1 √
α≤ x + αdx
n n
.
0
We apply Hölder’s inequality for the integral1 and we obtain that
1 Hölder’s inequality for the integral. If p > 1 and f : [0, 1] → R is a Riemann integrable
1 p 1
function, then |f (x)| dx ≤ |f (x)|p dx.
0 0
n
1 √ 1 1
n
x n + αdx ≤ (x n + α)dx = + α.
0 0 n+1
Thus,
n
1 √ 1
α≤ n
x n + αdx ≤ + α.
0 n+1
1.61. The limit equals 1. The problem, in a general form, is discussed in [55].
1.62. The limit equals 1. Observe that
n dx n(a+1) dy
=
0 1 + n2 sin2 (x + na) na 1 + n2 sin2 y
n(a+1) dy na dy
= − .
0 1 + n2 sin y 2
0 1 + n2 sin2 y
Now prove that if b > 0, then
nb dx
lim = b.
n→∞ 0 1 + n2 sin2 x
1.63. The limit equals 12 . First we show that if a ∈ R, then

π dx π
=√ . (9.9)
0 1 + a cos x
2 2
1 + a2
We have
π dx π/2 dx
=2
0 1 + a 2 cos2 x 0 1 + a 2 cos2 x
∞ dt
tan x=t
= 2
0 t2 + 1 + a2
/∞
2 t /
=√ arctan √ /
/
1 + a2 1 + a2 0
π
=√ .
1 + a2
*n+
Let k = π ∈ N and we observe that kπ ≤ n < (k + 1)π . Let
n x
In = dx.
0 1 + n2 cos2 x
Since kπ ≤ n < (k + 1)π , it follows

kπ x (k+1)π x
dx ≤ In < dx. (9.10)
0 1 + n2 cos2 x 0 1 + n2 cos2 x
mπ
Let m ∈ N and let Jm = 0 x
1+n2 cos2 x
dx. We calculate Jm using the
substitution mπ − x = t and we have
mπ mπ − t
Jm = dt
0 1 + n2 cos2 t
mπ 1
= mπ dt − Jm
0 1 + n2 cos2 t
π 1
= m2 π dt − Jm
0 1 + n2 cos2 t
π 2 m2
(9.9)
= √ − Jm .
1 + n2
Therefore,
π 2 m2
Jm = √ . (9.11)
2 1 + n2
Using relations (9.10) and (9.11), one has
π 2k2 In π 2 (k + 1)2 "n#

√ ≤ < √ , where k= .
2n 1 + n2 n 2n 1 + n2 π
Passing to the limit in the preceding inequalities, we obtain that
1 n x 1
lim dx = .
n→∞ n 0 1 + n2 cos2 x 2
Remark. Using the same method one can also prove that (see [55])
1 n x 1
lim dx = .
n→∞ n 0 1 + n2 sin2 x 2
1.65.
(a) The limit equals 12 . Integrate by parts.

(b) The limit equals k+11
. The substitution nx = y shows that
1 1 1 n
lnk 1 + enx dx = lnk 1 + ey dy.
nk 0 nk+1 0
Using Stolz–Cesàro Theorem, the ∞/∞ case, we have that

n+1
1 n lnk (1 + ey ) dy
lim lnk 1 + ey dy = lim n
n→∞ nk+1 0 n→∞ (n + 1)k+1 − nk+1

lnk 1 + eθn
= lim
n→∞ (n + 1)k+1 − nk+1
1
= ,
k+1
where θn ∈ (n, n + 1).

ln 1+eθn
We used, in the previous calculations, the limit lim n = 1, which follows
n→∞
from the inequalities

ln (1 + en ) ln 1 + eθn ln 1 + en+1
< < .
n n n
1.66. The limit equals b. We have

n bn+1 − a n+1 1 n
b
= lnn (enx )dx
n+1 n a

1 b
n
≤ lnn (1 + enx )dx
n a
ln(1 + enb ) √
≤ · b − a.
n
n
Also, we have
ln(1 + enb ) √
· b − a = b,
n
lim
n→∞ n
7
bn+1 − a n+1 n+1
a n+1 1
n n
lim = lim b n · lim 1− · lim √ = b,
n→∞ n+1 n→∞ n→∞ b n→∞ n
n+1
and the result follows based on the Squeeze Theorem.

1.67.
(b) The limit equals 2. Since f is continuous, there exist m and M such that 0 <
m ≤ f (x) ≤ M, ∀x ∈ [0, 1]. We have

n
1 √
n
(1 + x n )n f (x)dx ≤ 2 M.
0
n
The inequality 1 + x n ≥ 2x 2 implies that

1 1 n2
n n
(1 + x n )n f (x)dx ≥ 2 x 2 f (x)dx.
0 0
We have, since f (x) ≥ m, that

7
2m n
1 √
n
2n 2 ≤ (1 + x n )n f (x)dx ≤ 2 M,
n +2 0
and the desired limit follows based on the Squeeze Theorem.
1.68. The solution is given in [29].

1.69. We have
π
2
In = f (x) (cos x − sin x)2n dx
0
π
2
= f (x) (1 − sin(2x))n dx
0
π
2x=t 1 t
= f (1 − sin t)n dt
2 0 2
π
1 2 π −t
= f (t) + f (1 − sin t)n dt.
2 0 2
We use the substitution 1 − sin t = y and we have that t = arcsin(1 − y), which
implies dt = − √ 1 2 dy. It follows that
2y−y
1
1 y n− 2 1 1
In = F (y) √ dy = g(y)y n− 2 dy,
0 2−y 0
where F and g are the continuous functions

1 arcsin(1−y) π − arcsin(1−y) F (y)
F (y)= f +f and g(y)= √ .
2 2 2 2−y
We prove that if h : [0, 1] → R is continuous, then
1 1
lim n h(x)x n− 2 dx = h(1),
n→∞ 0
see also Problem 7.28, and this implies that

π
f (0) + f
lim nIn = g(1) = F (1) = 2
.
n→∞ 2
Let > 0. Since h is continuous at 1, there exists δ > 0 such that |h(x)−h(1)| <
, for 1 − δ < x < 1. Therefore,
1 1 1−δ 1 1 1 1 1
h(x)x n− 2 dx= h(x)x n− 2 dx+ (h(x)−h(1))x n− 2 dx+h(1) x n− 2 dx.
0 0 1−δ 1−δ
(9.12)
On the one hand
/ / 1
/ 1−δ / n(1 − δ)n+ 2
/n h(x)x n− 12 /
dx /≤ ||h|| ,
/ n + 12
0
1−δ 1
which implies that lim n h(x)x n− 2 dx = 0. Also,
n→∞ 0
1 1 n 1

lim n x n− 2 dx = lim 1 − (1 − δ)n+ 2 = 1.
n→∞ 1−δ n→∞ n+ 1
2
On the other hand
n 1
1 1
− 1 − (1 − δ)n+ 2 < n (h(x) − h(1))x n− 2 dx
n+ 1
2 1−δ
n 1

< 1 − (1 − δ)n+ 2 ,
n+ 1
2
1 n− 12
and we have − ≤ lim n 1−δ (h(x) − h(1))x
dx ≤ .
n→∞
1 1
Since was arbitrary taken, we obtain that lim n 1−δ (h(x)−h(1))x n− 2 dx = 0.
n→∞
1 1
Equality (9.12) implies lim n 0 h(x)x n− 2 dx = h(1).
n→∞
1.71. The limit equals 0. We have

⎛ ⎞
b b 1 b n+1 b n
n⎝ f n (x)dx ⎠ =n e n+1 ln a f (x)dx −e n ln a f (x)dx
n+1 n 1
f n+1 (x)dx−
a a
b b
1 1
=n ln f n+1 (x)dx − ln f n (x)dx eθn
n+1 a n a
b b
n
= ln f n+1 (x)dx − ln f n (x)dx eθn
n+1 a a
⎛ ⎞
b n+1 b
f (x)dx
= ⎝ln a b f n+1 (x)dx ⎠ eθn ,
n+1
− ln
n
a f (x)dx a
6 6
b b
where θn is between ln n+1 a f n+1 (x)dx and ln n a f n (x)dx.
This shows that lim θn = ln ||f ||. Passing to the limit in the previous equality,
n→∞
b
f n+1 (x)dx
the result follows since lim a b n = ||f || (see [26, problem 1.47, p. 9]) and
n→∞ a f (x)dx
6
b
lim n+1 a f n+1 (x)dx = ||f ||.
n→∞
1.72. The limit equals ||f ||∞ ln ||f ||∞ , where ||f ||∞ = sup {f (x) : x ∈ [0, ∞)}.
First we need to prove that the following limit holds:

∞
lim n
f n (x)e−x dx = ||f ||∞ . (9.13)
n→∞ 0
It is easy to see that

∞
lim n
f n (x)e−x dx ≤ ||f ||∞ . (9.14)
0
Let b > a ≥ 0 be fixed real numbers. Then

∞ b b
f n (x)e−x dx ≥ f n (x)e−x dx ≥ e−b f n (x)dx,
0 a a
and it follows that

∞ b
− nb
f n (x)e−x dx ≥ e
n n
f n (x)dx.
0 a
Thus,

∞
lim n
f n (x)e−x dx ≥ ||f ||[a,b] = sup {f (x) : x ∈ [a, b]} .
0
Letting a → 0 and b → ∞ in the preceding inequality, we get that

∞
lim n
f n (x)e−x dx ≥ ||f ||∞ . (9.15)
0
Combining (9.14) and (9.15), we have that (9.13) is proved.

Now we prove that
∞
f n+1 (x)e−x dx
lim ∞
0
= ||f ||∞ . (9.16)
n→∞ n −x
0 f (x)e dx
∞ ∞
Let M = ||f ||∞ . We have 0 f n+1 (x)e−x dx ≤ M 0 f n (x)e−x dx, and hence
∞
f n+1 (x)e−x dx
0
∞ ≤ M.
n −x
0 f (x)e dx
An application of Hölder’s inequality, with p = n+1

n and q = n + 1, shows that
∞ ∞ n ∞ 1
n+1 n+1
f n (x)e−x dx ≤ f n+1 (x)e−x dx e−x dx ,
0 0 0
and it follows
∞
∞ f n+1 (x)e−x dx
n
f n (x)e−x dx ≤ 0 ∞ n −x
. (9.17)
0 0 f (x)e dx
Letting n → ∞ in (9.17) and using (9.13), we get that

∞ ∞
f n+1 (x)e−x dx f n+1 (x)e−x dx
M ≤ lim ∞ 0
and this implies that lim ∞
0
=M.
0 f n (x)e−x dx n→∞ n −x
0 f (x)e dx
Now we are ready to solve the problem. We have, based on Lagrange Mean Value
Theorem, that

∞ ∞
xn = n n
f n+1 (x)e−x dx − n
f n (x)e−x dx
0 0
∞ ∞
1 1
= n exp f n+1 (x)e−x dx − exp
ln ln f n (x)e−x dx
0 n n 0
∞
n+1 (x)e−x dx
0 f
= ln ∞ n −x
exp(θn ),
0 f (x)e dx
6 6
∞ ∞
where θn is between ln n 0 f n+1 (x)e−x dx and ln n 0 f n (x)e−x dx. This
implies that lim θn = ln ||f ||∞ . Letting n → ∞ in the preceding equality
n→∞
and using (9.16), we get that
lim xn = ||f ||∞ ln ||f ||∞ .

n→∞
If f is not bounded, then the limit might not be finite, an example being f (x) =
x, in which case the limit equals ∞.
1.73.
b
(a) L = a f (x)dx. Use the technique given in the solution of Problem 7.37 and
prove that
/ /
/ b b f (x) /
/ f (x)dx − dx //
/
a a 1 + sin x sin(x + 1) · · · sin(x + n)
n+1
2
1
2 + 1
2(n+1) sin 1
≤ M(b − a) n+1 ,
2
1 − 2 + 2(n+1) sin 1
1 1
where M = sup {|f (x)| : x ∈ [a, b]}.

(b) The limit equals 0. We have
/ /
/ b b f (x) /
q n // f (x)dx − dx //
a a 1 + sin x sin(x + 1) · · · sin(x + n)
n+1
2
1
2 + 1
2(n+1) sin 1
≤ M(b − a)q n
n+1 ,
2
1 − 2 + 2(n+1) sin 1
1 1
and the result follows since

n+1
1 1 2
lim q n
+
n→∞ 2 2(n + 1) sin 1
n
q n 1 2 1 1
= lim √ 1+ · + = 0.
n→∞ 2 (n + 1) sin 1 2 2(n + 1) sin 1
1.74.
(a) The limit equals 0. It suffices to consider only the case when n is even. Also,
since the function x → | sin x| is a periodic function of period π , it is enough to
consider the case when a = 0 and b = π .
We have

π π
2 √ n + 12
sin2n x dx = 2 sin2n x dx = π
0 0 (n + 1)
and it follows, based on Stirling’s formula, that

π
lim sin2n x dx = 0.
n→∞ 0
(b) The limit equals 0. We have

/ /
/ b / b
/ f (x) sinn x dx //≤ M | sinn x| dx,
/
a a
where M = {|f (x)| : x ∈ [a, b]} and the desired limit follows based on part
(a) of the problem.
1.77. We have
x 1+x x
(f (t + 1) − f (t)) dt = f (y)dy − f (t)dt
−x 1−x −x
1+x 1−x
= f (y)dy − f (t)dt
x −x
= f (θ1 ) − f (θ2 ),
where θ1 ∈ (x, 1 + x) and θ2 ∈ (−x, 1 − x).

It follows that
x
lim (f (t + 1) − f (t)) dt = f (∞) − f (−∞),
x→∞ −x
since f (θ1 ) → f (∞) and f (θ2 ) → f (−∞), when x → ∞.

1
1.78. The limit equals − 0 f (x)dx. Approximate f by a polynomial function.
1.79. Since 1 − f ≤ 1 − f k = (1 − f )(1 + f + · · · + f k−1 ) ≤ k(1 − f ),
it suffices to solve the problem for the case when k = 1. Let A = f −1 (1) =
{x ∈ [0, 1] : f (x) = 1}.
We have
1
n f n (x)(1−f (x))dx=n f n (x)(1−f (x))dx+n f n (x)(1−f (x))dx
0 A [0,1]\A
=n f n (x)(1−f (x))dx.
[0,1]\A
Let fn : [0, 1] \ A → R, fn (x) = nf n (x)(1 − f (x)). Since
fn+1 (x) n+1

lim = lim f (x) = f (x) < 1,
n→∞ fn (x) n→∞ n
we get that lim fn (x) = 0.

n→∞
A calculation shows
1 − f n (x)
nf n (x) = f n (x) + · · · + f n (x) < 1 + f (x) + f 2 (x) + · · · + f n−1 (x) =
1 − f (x)
and this implies nf n (x)(1 − f (x)) < 1 − f n (x) < 1, ∀x ∈ [0, 1] \ A. It follows,
based on Lebesgue Dominated Convergence Theorem, that
1
lim n f n (x)(1−f (x))dx= lim fn (x)dx= lim fn (x) dx=0.
n→∞ 0 n→∞ [0,1]\A [0,1]\A n→∞
1.80.
(a) This part of the problem is solved in [95].

(b) Solution due to students Kim Il Jin and Mun Chung Jin. We prove that

1 1 1 √ √ 1
x 1+ 2 +···+ n (1 − x)(1 −
n
x) · · · (1 − n x)dx < √ .
0 (Hn + 1) n Hn + 1
The preceding inequality is in fact based on the exact calculation of the integral
under the nth root. We need the following lemma.
Lemma 9.1. Let {i1 , i2 , . . . , ik } be of {1, 2, .. . , n} with distinct
a subset integers
1 1 1
i1 , i2 , . . . , ik and let F (x) = x n 1 − x 1
H i 1 − x 2 · · · 1 − x k . Then
i i
1 1 1

1 1
i1 i2 ik
F (x)dx = M F (x) 1
+ 1
+ ··· + 1
dx,
0 0 1 − x i1 1 − x i2 1 − x ik
where M = 1
.
Hn +1+ i1 + i1 +···+ i1
1 2 k
Proof. We integrate by parts and we have that
1 1
F (x)dx = − xF (x)dx.
0 0
A calculation shows that

⎡ ⎤
k k 1
1 ij
xF (x) = F (x) ⎣Hn + − 1
⎦.
ij
j =1 j =1 1−x ij
It follows that
⎛ ⎞
1 k 1
1⎠
F (x)dx = − ⎝Hn + F (x)dx
0 ij 0
j =1
1 1 1

1
i1 i2 ik
+ F (x) 1
+ 1
+ ··· + 1
dx,
0 1 − x i1 1 − x i2 1 − x ik
and the lemma is proved.
1 √ √
We apply the lemma to the integral 0 x Hn (1 − x)(1 − x) · · · (1 − n
x)dx and
1
we note the lemma allows us to remove the factors 1 − x ij , j = 1, . . . , n, one by
one to obtain n integrals. We apply the lemma one more time to each of these n
integrals and we obtain n(n − 1) integrals. We continue this process until only the
term x Hn remains under the integral sign, and we see that there are n! integrals of the
1
form ai1 ,i2 ,...,in 0 x Hn dx, where the coefficient ai1 ,i2 ,...,in is obtained by eliminating
the product

1 1 1
1 − x i1 1 − x i2 · · · 1 − x in ,
and the distinct integers i1 , i2 , . . . , in are such that {i1 , i2 , . . . , in } = {1, 2, . . . , n}.
The coefficient, obtained by eliminating the factor

1 1 1
1−x i 1 1−x i 2 ··· 1 − x ik
,
is given by
1 1 1
i1 i2 ik
· ··· .
Hn + 1 + 1
i1 + ··· + 1
ik Hn + 1 + 1
i2 + ··· + 1
ik Hn + 1 + 1
ik
It follows that
1 1 1
i1 i2 in
ai1 ,i2 ,...,in = · ···
Hn + 1 + 1
i1 + ··· + 1
in Hn + 1 + 1
i2 + ··· + 1
in Hn + 1 + 1
in
1 1 1 1
= · · ··· ,
n! Hn + 1 + i1 + · · · + 1
in Hn + 1 + 1
i2 + ··· + 1
in Hn + 1 + 1
in
1
,
n
since ij = n!. Therefore, we obtain the following formula
i=1
1 √ √
x Hn (1−x)(1− x) · · · (1− n x)dx
0

1 1 1 1 1
= · ··· x Hn dx
0 n! Hn +1+ i11 + · · · + i1n Hn +1+ i12 + · · · + i1n Hn +1+ i1n
1 1 1 1
= · ··· ,
n!(Hn +1) Hn +1+ i11 + · · · + i1n Hn +1+ i12 + · · · + i1n Hn +1+ i1n
where the summation is done by the number of all n-tuples (i1 , i2 , . . . , in ) of distinct
integers i1 , i2 , . . . , in , with {i1 , i2 , . . . , in } = {1, 2, . . . , n}.
It follows, since the previous sum has n! terms and each of the term under the
1
summation sign is strictly less than (Hn +1) n , that
1 √ √ 1 1 n! 1
x Hn (1 − x)(1 − x) · · · (1 − n x)dx < · · = .
0 Hn + 1 n! (Hn + 1)n (Hn + 1)n+1
As an example, we calculate based on the given technique the previous integral

for the cases when n = 2 and n = 3.
The case n = 2. We have
1 √ 1 1 1
x H2 (1 − x) 1 − x dx = · ,
0 2!(H2 + 1) H2 + 1 + 1
i + 1
j H2 + 1 + 1
j
where the summation is done by the number of all pairs (i, j ) of distinct integers
with {i, j } = {1, 2}.
We have (i, j ) ∈ {(1, 2), (2, 1)} and it follows that
1 √ 1 1 1
x H2 (1 − x) 1 − x dx = · ·
0 2!(H2 + 1) H2 + 1 + 1
1 + 1
2 H2 + 1 + 1
2
1 1 1
+ · ·
2!(H2 + 1) H2 + 1 + 1
2 + 1
1 H2 + 1 + 1
1
13
= .
420
The case n = 3. We have
1 √ √
x H3 (1 − x) 1 − x 1 − 3 x dx
0
1 1 1 1
= · · ,
3!(H3 + 1) H3 + 1 + 1
i + 1
j + 1
k H3 + 1 + 1
j + 1
k H3 + 1 + 1
k
where the summation is done by the number of all 3-tuples (i, j, k) of distinct
integers with {i, j, k} = {1, 2, 3}.
We have (i, j, k) ∈ {(1, 2, 3), (1, 3, 2), (2, 3, 1), (2, 1, 3), (3, 1, 2), (3, 2, 1)} and
it follows that
1 √ √
x H3 (1 − x) 1 − x 1 − 3 x dx
0
1 1 1 1
= · · ·
3!(H3 + 1) H3 + 1 + 11 + 1
2 + 1
3 H3 + 1 + 1
2 + 1
3 H3 + 1 + 1
3
1 1 1 1
+ · · ·
3!(H3 + 1) H3 + 1 + 11 + 1
3 + 1
2 H3 + 1 + 1
3 + 1
2 H3 + 1 + 1
2
1 1 1 1
+ · · ·
3!(H3 + 1) H3 + 1 + 1
2 + 1
3 + 1
1 H3 + 1 + 1
3 + 1
1 H3 + 1 + 1
1
1 1 1 1
+ · · ·
3!(H3 + 1) H3 + 1 + 1
2 + 1
1 + 1
3 H3 + 1 + 1
1 + 1
3 H3 + 1 + 1
3
1 1 1 1
+ · · ·
3!(H3 + 1) H3 + 1 + 1
3 + 1
1 + 1
2 H3 + 1 + 1
1 + 1
2 H3 + 1 + 1
2
1 1 1 1
+ · · ·
3!(H3 + 1) H3 + 1 + 1
3 + 1
2 + 1
1 H3 + 1 + 1
2 + 1
1 H3 + 1 + 1
1
1021194
= .
185910725
We mention that another solution of part (b) of the problem, based on an −δ
argument, was given by I.J. Kim.
Series of Real Numbers
10
10.1 Miscellaneous Series
2.1.
1
(a) 8; (b) 12 ; (c) 11
18 ; (d) 1
12 ; (e) using the identity

1
ln 1 − 2 = ln(n + 1) + ln(n − 1) − 2 ln n
n
it follows that the series equals − ln 2; (f) a

(a−1)2
; (g) using the identity

n 1 1 1
= −
n4 + n2 + 1 2 n2 − n + 1 n2 + n + 1
we have that the series equals 12 ;

n
(h) (n−2)!+(n−1)!+n! = (n−1)!
1
− n!
1
and the series equals 1;
(i) e; (j) 20 e.
2.2.
(a) We have, see the solution of Problem 7.47, that the following equality holds
√ √
(1 + 2)2n−1 = ( 2 − 1)2n−1 , ∀n ≥ 1. It follows that
∞ √ ∞ √ 2n−1 1
(1 + 2)2n−1 = 2−1 = .
2
n=1 n=1

2.3. Use that 1
(2n+1)2 −1
= 1
4
1
n − 1
n+1 , ∀n ≥ 1.
π2
2.4. (a) − 3; (b) 10 − π 2 .
3 √
2.5. (a) ln 2; (b) π123 − 14 ln 3; (c) 14 ln 2 − 24
π
.
2.6. Let N = max {m, n, p : 2m 3n 5p ∈ A}. We have
N N N
1 1
≤
x 2m 3n 5p
x∈A m=0 n=0 p=0
N N N
1 1 1
=
2m 3n 5p
m=0 n=0 p=0

1 3 1 5 1
=2 1− 1 − N +1 1 − N +1
2N +1 2 3 4 5
15
< .
4
2.7.
(a) We calculate the 4nth partial sum of the series and we have that
1 1 1 1 1 1 1
S4n = 1 + − − + ··· + + − −
2 3 4 4n − 3 4n − 2 4n − 1 4n
n n
1 1
= H4n − 2 −2
4i − 1 4i
i=1 i=1
4n 1 n 1 n 1
= x i−1 dx − 2 x 4i−2 dx − 2 x 4i−1 dx
i=1 0 i=1 0 i=1 0
1 − x 4n
1 1 1 − x 4n 1 1 − x 4n
= dx − 2 x2 dx − 2 x3 dx
0 1−x 0 1 − x4 0 1 − x4
1 1 − x 4n 1 − x 4n
= − 2x 2 (1 + x) dx
0 1−x 1 − x4
1 1 − x 4n 2x 2

= 1− dx
0 1−x 1 + x2
1 (1 − x 4n )(1 + x)
= dx.
0 1 + x2
It follows that
1 1+x π ln 2
lim S4n = dx = + .
n→∞ 0 1+x 2 4 2
(b) We have
1 1 1 1 1 1 1 1 1 1 1
1+ + − − − + + + − − − +···
2 3 4 5 6 7 8 9 10 11 12
1
= (1+x+x 2 − x 3 − x 4 − x 5 +x 6 +x 7 +x 8 − x 9 − x 10 − x 11 + · · · )dx
0
1
= (1+x+x 2 )(1 − x 3 +x 6 − x 9 + · · · )dx
0
1 1+x+x 2
= dx
0 1+x 3
√
2 3 ln 2
= π+ .
9 3
2.8. Use the identity

1 1 n n−1 1
= − + , n ≥ 1.
n!(n4 +n2 +1) 2 (n+1)!(n2 +n+1) n!(n2 −n+1) (n+1)!
2.9. (b) (a, b, c) ∈ {(1, 2, 3), (4, 8, 6)}.

2.10.
(a) We have
∞ ∞
9 9 1
0.9999999 . . . = n
= = 1.
10 10 10n−1
n=1 n=1
2.11. We have
∞ ∞
1 11 (−1)n 3
= and =− .
(3 + (−1)n )n 15 (3 + (−1)n )n 5
n=1 n=1
Calculate the 2nth partial sum of the series.

2.12. Both series are equal to π4 .
2.13.
√ √
(a) Let α = 1+ 5
2 and β = 1− 5
2 . We have
2 2
α n+1 − β n+1 α n−1 − β n−1
2
Fn+1 − Fn−1
2
= √ − √
5 5
−2 −2 − β 2
2n α − α
1 2
2n β
= √ α √ −β √
5 5 5
α 2n − β 2n
= √
5
= F2n .
(b) The series equals 2. Use that
F2n 1 1
2 2
= 2
− 2
, n ≥ 2.
Fn−1 Fn+1 Fn−1 Fn+1
2.14. The series equals γ .

2n
(−1)n
2.15. The series equals 0. We have S2n = ln 1 + n = ln 2n+1
2n , n ≥ 1.
k=2
2.17. (a) P (x) = −2x 2 − 4x − 6; (b) Q(x) = −2x 3− 6x 2 − 18x − 26.
1
2.18. Use that n(n+k)(n+2k) = 2k12 n1 − n+k
2
+ n+2k
1
and calculate the nth partial
sum of the series.
2.19.
(a) Let
∞
1
xn = .
k(k + 1)(k + 2) · · · (k + n)
k=1
We have
1
k(k + 1)(k + 2) · · · (k + n)

1 1 1
= −
n k(k + 1) · · · (k + n − 1) (k + 1)(k + 2) · · · (k + n)
and it follows that

∞

1 1
xn = xn−1 −
n (k + 1)(k + 2) · · · (k + n)
k=1
∞
k+1=m 1 1 1
= xn−1 − −
n m(m + 1) · · · (m + n − 1) n!
m=1

1 1
= xn−1 − xn−1 +
n n!
1
= .
n · n!
2.21. We have
! ∞
1 1 {y}
x dx = dy
0 x 1 y3
∞ k+1 y−k
= dy
y3
k=1 k
∞ /k+1
1 k /
= − + 2 //
y 2y k
k=1
∞
1 1 1 1
= − −
2 k k + 1 (k + 1)2
k=1
ζ (2)
=1− .
2
1
2.24. Use that √n(n+1) < √2
n
− √2 ,
n+1
∀n ≥ 1.
2.26. We have
∞
1 1 (−1)n−1
1 − + ··· +
n(n + 1) 2 n
n=1
∞
(−1)n−1 (−1)n

1− 1
2 + ··· + 1− 1
2 + ··· + n+1 (−1)n
= n
− +
n n+1 (n + 1)2
n=1
∞
(−1)n−1 (−1)n
∞
1− 1
2 + ··· + 1− 1
2 + ··· + n+1 (−1)n
= n
− +
n n+1 (n + 1)2
n=1 n=1
∞
(−1)n
=1+
(n + 1)2
n=1
π2
= .
12
n
2.27. Let l = lim xn+1 x n
< 1. For ∈ (0, 1 − l), ∃n0 ( ) ∈ N such that 0 <
n→∞
xn+1 √ 1
+ 1 +···+ n−1
1
xn <
n
l + , ∀n ≥ n0 . It follows that 0 < xn < xn0 (l + ) n0 n0 +1 ,
∀n ≥ n0 . Passing to the limit, as n → ∞, in the previous inequalities, we get that
lim xn = 0.
n→∞
(2n−1)!!
2.28. Let an = (2n)!!(n+1) . Observe that
an = 2(n + 1)an − 2(n + 2)an+1 , ∀n ≥ 1.

2.29.
(a) Use Problem 2.27.

(b) Let an = q(q+p)(q+2p)···(q+np−p)
p(2p)···(np) · n+1
1
. Calculate the nth partial sum of the
p
series and use that an = p−q [(n + 1)an − (n + 2)an+1 ], n ≥ 1.
2.31. We have
∞ ∞ ∞ ∞ ∞ ∞
1 1 1
(ζ (n) − 1) = = = = 1.
kn kn k2 −k
n=2 n=2 k=2 k=2 n=2 k=2
2.32. We have
∞ ∞ ∞ ∞ ∞ ∞
1 1 n 1 1
(−1) (ζ (n) − 1) =
n
(−1) n
= − = =
kn k k2 +k 2
n=2 n=2 k=2 k=2 n=2 k=2
and
∞ ∞ ∞ ∞ ∞ n ∞
1 1 1 3
(ζ (2n) − 1) = = = = .
k 2n k2 k2 − 1 4
n=1 n=1 k=2 k=2 n=1 k=2
2.33. We have
∞ ∞ ∞
1 1
(−1) (ζ (n + 1) − ζ (n)) =
n
(−1) n
−1
in i
n=2 n=2 i=2
∞ ∞
1 1 n
= −1 −
i i
i=2 n=2
∞
1 1
= −1 · 2
i i +i
i=2
∞
1 1 1
= −2 −
i2 i i+1
i=2
= ζ (2) − 2.
2.34.
(a) We have
.
p if n = pi
(n, p) =
1 if n = pi + j, j = 1, 2, . . . , p − 1
and it follows that

∞
(n, p) p 1
(−1)n−1 = (−1)pi−1 + (−1)pi+j −1
n pi pi + j
n=1 n=pi n=pi+j
j =1,...,p−1
∞
1 1 1
= (−1)i−1 + (−1)pi+j −1 − (−1)pi−1
i pi + j pi
i=1 n=pi+j n=pi
j =0,...,p−1
∞ ∞
1 1 1
= ln 2 + (−1)n−1 − (−1)i−1
n p i
n=1 i=1
1
= ln 2 + ln 2 − ln 2
p
2p − 1
= ln 2.
p
(b) We have
⎧
⎪
⎪ n = 4k + 1, n = 4k + 3
⎨1 if
(n, 4) = 2 if n = 4k + 2
⎪
⎪
⎩4 if n = 4k
and it follows that

∞ ∞ ∞ ∞ ∞
(n, 4) 1 1 1 1
= + +2 +4
n2 (4k + 1)2 (4k + 3)2 (4k + 2)2 (4k)2
n=1 k=0 k=0 k=0 k=1
∞
1 1 1 1
= + + +
(4k + 1)2 (4k + 2)2 (4k + 3)2 (4k + 4)2
k=0
∞ ∞
1 3 1
+ +
(4k + 2)2 16 k2
k=0 k=1
∞
1 1 3
= ζ (2) + + ζ (2)
4 (2k + 1)2 16
k=0
11
= ζ (2).
8
The second series of this part of the problem can be calculated similarly.
2.35. Let Sn be the nth partial sum of the series. A calculation shows that
n
1 1 1
Sn = k ln 1 + −1+ = n ln(n + 1) − ln n! − n + Hn ,
k 2k 2
k=1
where Hn = 1 + 12 + · · · + 1
n denotes the nth harmonic number.
Using Stirling’s formula
1 1
ln n! ∼ ln(2π ) + ln n + n ln n − n,
2 2
we have that
n+1 1 1
Sn ∼ n ln − ln(2π ) + (Hn − ln n) ,
n 2 2
which implies lim Sn = 1 + γ2 − 12 ln(2π ).

n→∞
The second series can be calculated similarly.
2.36. Add and subtract the series from Problem 2.35. Another method would be to
calculate the nth partial sum of the series.
2.37. We calculate S2n , the 2nth partial sum of the series, and we have

2n + 1 (1 · 3 · 5 · · · (2n − 1))4
S2n = ln ·
n (2 · 4 · · · 2n)2 · (2 · 4 · · · (2n − 2))2
2n + 1 ((2n − 1)!)2
= ln + 2 ln 4n−3 .
n 2 · n! · ((n − 1)!)3
It follows, based on Stirling’s formula, that

2n + 1 2 2n − 1 4n−1 n−1 n−1 n
S2n ∼ ln + 2 ln + 2 ln + ln + 2 ln
n πe 2n − 2 n n
and lim S2n = ln π82 .

n→∞
2.38. Use the formula ln 1 + 2n
1
+ ln 1 + 2n+1
1
= ln 1 + n1 , ∀n ≥ 1. It
follows that

1 1
ln 1 + ln 1 +
2n 2n + 1

1 1 1 1
= ln2 1 + − ln2 1 + − ln2 1 + ,
2 n 2n 2n + 1
and calculate, based onthe preceding

formula,
the nth partial
sum
of the series.
2.39. Using formula ln 1 + 2n 1
+ ln 1 + 2n+1
1
= ln 1 + n1 , ∀n ≥ 1, we have
that

1 1 1
ln 1 + ln 1 + ln 1 +
n 2n 2n + 1

1 1 1 1
= ln 1 +
3
− ln 1 +
3
− ln 1 +
3
.
3 n 2n 2n + 1
We calculate, based on the preceding formula, the nth partial sum of the series
and we have
2n+1
ln3 2 1 1
Sn = − ln 1 +
3
.
3 3 k
k=n+1
Since
2n+1
1 1 1
(n + 1) ln3 1 + < ln3 1 + < (n + 1) ln3 1 + ,
2n + 1 k n+1
k=n+1
ln3 2
we obtain that lim Sn = 3 .
n→∞
2.40.
Hn Hn Hn+1
(a) Since n(n+1) = n − n+1 + 1
(n+1)2
, ∀n ≥ 1, we have that
∞ ∞ ∞
Hn Hn Hn+1 1
= − + = 1 + ζ (2) − 1 = ζ (2).
n(n + 1) n n+1 (n + 1)2
n=1 n=1 n=1
The other series can be calculated similarly.

2.41.
1
(a) Let In,n−1 = 0 x n (1 − x)n−1 d x. We integrate by parts and we obtain the
recurrence formula In,n−1 = n−1
n+1 In+1,n−2 .
(b) We have, based on part (a), that
∞ ∞
1 (n + 1)(n!)2
=
Cn (2n)!
n=0 n=0
∞ 1
=1+ (n + 1)n x n (1 − x)n−1 d x
n=1 0
1 ∞
=1+ x n(n + 1)(x(1 − x))n−1 d x
0 n=1
1 2x
=1+ dx
0 (1 − x + x 2 )3
2x − 1
1 1 1
=1+ dx + dx
0 (1 − x + x 2 ) 3
0 (1 − x + x 2 )3
/1
1 / 1 1
=1− / + dx
2(x − x + 1) 0
2 2 / ((x − 1/2) 2 + 3/4)3
0
1/2 1
x−1/2=u
= 1+2 du
0 (u2 + 3/4)3
√ √ /
16u 8u 4 3 2 3u //1/2
=1+2 + + arctan /
3(4u2 + 3)2 3(4u2 + 3) 9 3 0
√
4 3
=2+ π.
27
(c) This part of the problem can be solved in the same way as part (b).
2.42. We have
∞ ∞ ∞ k
nk (n + 1 − 1)k 1
= = Cik (n + 1)i (−1)k−i
(n + 1)p (n + 1)p (n + 1)p
n=1 n=1 n=1 i=0
k ∞ k
1
= Cik (−1)k−i = (−1)k−i Cik ζ (p − i).
(n + 1)p−i
i=0 n=1 i=0
10.2 Applications of Abel’s Summation Formula
2.43.
(a) We use Abel’s summation formula with ak = cos k and bk = 1

n+k , and we have
that
cos 1 cos 2 cos n cos 1 + cos 2 + · · · + cos n

+ + ··· + =
n+1 n+2 2n 2n + 1
n
cos 1 + cos 2 + · · · + cos k
+
(n + k)(n + k + 1)
k=1
n
sin n2 cos n+1 sin k2 cos k+1
= 2
+ 2
.
(2n + 1) sin 12 k=1
(n + k)(n + k + 1) sin 12
n
sin cos n+1
We used that cos 1 + cos 2 + · · · + cos n = 2 2
, ∀n ≥ 1.
sin 12
It follows that

cos 1 cos 2 cos n
lim + + ··· + = 0,
n→∞ n+1 n+2 2n
since
/ /
/ n
sin k2 cos k+1 / n
/ 2 /≤ 1 1
=
1
.
/ (n + k)(n + k + 1) sin 12 / sin 1 n2 n sin 12
k=1 2 k=1
(b) We have

cos2 1 cos2 2 cos2 n 1 1 1 1
+ + ··· + = + + ··· +
n+1 n+2 2n 2 n+1 n+2 2n

1 cos 2 cos 4 cos(2n)
+ + + ··· + .
2 n+1 n+2 2n
Exactly as in the solution of part (a), it can be proved that

cos 2 cos 4 cos(2n)
lim + + ··· + = 0,
n→∞ n+1 n+2 2n

and the result follows since lim 1
n+1 + 1
n+2 + ··· + 1
2n = ln 2.
n→∞
The limits in Remark 2.3 can be proved by using the formulae [100, p. 130]
k−1
1
• cos2k−1 x = Ci2k−1 cos(2k − 2i − 1)x;
22k−2
k−1i=0
1
• cos2k x = 2k 2Ci2k cos 2(k − i)x + Ck2k .
2
i=0

2.44. (a) xn+1 − xn = Ckn+1 e − 1 − 1
1! − 1
2! − ··· − 1
(n+1)! > 0, ∀n ≥ k.
(b) and (c) We have
∞
1 1 1
xn < Cki e − 1 − − − · · · − ,
1! 2! i!
i=k
and we prove that the series converges by calculating its value.

One can check, using integration by parts, that the following formula holds:
1 1 1 1 1
e−1− − − ··· − = (1 − t)i et dt, i ≥ 0.
1! 2! i! i! 0
It follows that
∞
1 1 1
lim xn = Cki e − 1 − − − · · · −
n→∞ 1! 2! i!
i=k
∞
Cki 1
= (1 − t)i et dt
i! 0
i=k
∞

1 1 (1 − t)i−k
= (1 − t) e k t
dt
k! 0 (i − k)!
i=k
e 1
= (1 − t)k dt
k! 0
e
= .
(k + 1)!
2.45. Solution I. We use Abel’s summation formula, with
x x2 xn
an = 1 and bn = ex − 1 − − − ··· −
1! 2! n!
for the first series, and
x x2 xn
an = n and bn = ex − 1 − − − ··· −
1! 2! n!
for the second series, and
x x2 xn
an = n2 and bn = ex − 1 − − − ··· −
1! 2! n!
for the third series.
Solution II. We calculate the first series. Let y(x) be the sum of the series. We
have
∞
x x2 xn
y(x) = e −1− −
x
− ··· −
1! 2! n!
n=0
∞
x x2 xn
= ex − 1 + ex − 1 − − − ··· − .
1! 2! n!
n=1
It follows that
∞
x x2 x n−1
y (x) = e + x
e −1− −
x
− ··· −
1! 2! (n − 1)!
n=1
∞ ∞
x x2 xn xn
= ex + − − ··· −
ex − 1 − +
1! 2! n! n!
n=1 n=1

= ex + y(x) − ex + 1 + ex − 1
= ex + y(x).

Therefore, y (x) − y(x) = ex ⇔ y (x)e−x − y(x)e−x = 1 ⇔ y(x)e−x =
1 ⇒ y(x)e−x = x + C , C ∈ R. We obtain that y(x) = (x + C )ex . Since
y(0) = 0 we have that y(x) = xex .
The other two series can be calculated similarly.
2.46. Observe that the 2nth partial sum of the series is S2n = 112 + 312 +· · ·+ (2n−1)
1
2,
∀n ≥ 1. An alternative solution is based on the use of Abel’s summation formula.
2.47. Use Abel’s summation formula with:
(a) an = 1 and bn = ζ (2) − 1 − 212 − · · · − n12 − n+k

1
;

(b) an = 1 and bn = n2 + (n+1)2 + · · · − n+k .
1 1 1
Alternatively, the series in part (b) follows directly from the series in part (a).
2.48. Parts (a) and (b) follow by direct computations (see the solution of parts (a)
and (b) of Problem 2.50).
(c) Solution I. Use part (a) of the problem.
Solution II. Let

∞
1 1 1 1
A= + + + ··· −
n2 (n + 2)2 (n + 4)2 2n
n=1
∞
1 1 1 1
B= + + + · · · − .
(n + 1)2 (n + 3)2 (n + 5)2 2n
n=1
We have, based on Problem 2.47, that

∞
1 1 1
A+B = + + ··· − = 1.
n2 (n + 1)2 n
n=1
On the other hand,

∞
1 1 1 1
B= + + + ··· −
(n + 1)2 (n + 3)2 (n + 5)2 2n
n=1
∞
n+1=i 1 1 1 1
= + + + ··· −
i 2 (i + 2) 2 (i + 4)2 2(i − 1)
i=2
∞ ∞
1 1 1 1 1 1
= + + + ··· − + −
i2 (i + 2)2 (i + 4)2 2i 2i 2(i − 1)
i=2 i=2
∞
1 1 1 1 1
= + + + ··· − −
i 2 (i + 2) 2 (i + 4)2 2i 2
i=2
∞
1 1 1 1
= + + + ··· −
i2 (i + 2)2 (i + 4)2 2i
i=1

1 1 1 1 1
− + + + · · · − −
12 32 52 2 2
π2
=A− .
8
It follows that
∞
1 1 1 1 1 π2
A= + + + ··· = + −
n2 (n + 2)2 (n + 4)2 2n 2 16
n=1
∞
1 1 1 1 1 π2
B= + + + · · · − = − .
(n + 1)2 (n + 3)2 (n + 5)2 2n 2 16
n=1
2.50.
(a) We have
∞
1 1 1 1
+ + + ··· =
n3 (n + k)3 (n + 2k)3 (n + ki)3
i=0
∞ 1
1
= x n+ki−1 ln2 x dx
2 0
i=0
1 1 ∞
(∗)
= x n+ki−1 ln2 x dx
2 0 i=0
1 1 x n−1 2
= ln x dx.
2 0 1 − xk
We used at step (*) Tonelli’s theorem for nonnegative functions.

1 1 1 n2 1 x n−1 2
n2 + + + · · · = ln x dx
n 3 (n + k)3 (n + 2k)3 2 0 1 − xk
x n =y 1 1 ln2 y
= k
dy
2n 0 1−yn
1 k
1
= n
k
ln2 y dy.
2k 0 1−y n
k
Let fn (y) = n
k ln2 y, y ∈ (0, 1). We have lim fn (y) = − ln y.
1−y n n→∞
k
ln2 y
On the other hand, for n > k, we have fn (y) = k
n
ln2 y < 1−y , since
1−y n
k
n
k ≤ 1−y , ∀n
1
≥ k. The last inequality follows from the fact that the function
1−y n
g : [0, 1] → R, g(x) = x
1−y x is an increasing function when y ∈ (0, 1). The
ln2 y
function y → 1−y is integrable over the interval [0, 1] and we have, based on
Lebesgue Dominated Convergence Theorem, that
1 k
1 1 1 1
lim n 2
+ + + ··· = lim n
ln2 ydy
n→∞ n3 (n + k)3 (n + 2k)3 n→∞ 2k 0 1−yn
k
1 1
= (− ln x)dx
2k 0
1
= .
2k
(c) Solution I. Use part (a) of the problem.
Solution II. Let

∞
1 1 1
A= + + + ···
n3 (n + 2)3 (n + 4)3
n=1
∞
1 1 1
B= + + + ··· .
(n + 1)3 (n + 3)3 (n + 5)3
n=1
We have, based on Abel’s summation formula, that

∞ ∞
1 1 1 1
A+B = + + + ··· = = ζ (2).
n3 (n + 1)3 (n + 2)3 n2
n=1 n=1
On the other hand,

∞
1 1 1
B= + + + ···
(n + 1)3 (n + 3)3 (n + 5)3
n=1
∞
n+1=i 1 1 1
= + + + ···
i3 (i + 2)3 (i + 4)3
i=2
∞
1 1 1
= + + + ···
i 3 (i + 2) 3 (i + 4)3
i=1

1 1 1
− + 3 + 3 + ···
13 3 5
7
= A − ζ (3).
8
It follows that
∞
1 1 1 1 7
A= + + + ··· ζ (2) + ζ (3)=
n3 (n + 2)3 (n + 4)3 2 16
n=1
∞
1 1 1 1 7
B= + + + · · · = ζ (2) − ζ (3).
(n + 1)3 (n + 3)3 (n + 5)3 2 16
n=1
2.52. Use Abel’s summation formula with

1 1 1
an = (−1)n and bn = 1 + + + · · · + − ln n − γ .
2 3 n
Alternatively, calculate the 2nth partial sum of the series and use Wallis formula.
√
(a) an = (−1)n and bn = 1 + 12 + 13 + · · · + n1 − ln n(n + 1) − γ ;
√
(b) an = (−1)n and bn = 1 + 12 + 13 + · · · + n1 − ln 3 n(n + 1)(n + 2) − γ ;
√
(c) an = 1 and bn = 1 + 12 + 13 + · · · + n1 − ln n(n + 1) − γ .
2.54.
(a) We calculate the nth partial sum of the series,

usingAbel’s summation formula
with ak = 1 and bk = 1 + 2 + · · · + k − ln k + 12 − γ , and we have that
1 1

n
1 1 1
Sn = 1 + + · · · + − ln k + −γ
2 k 2
k=1

1 1 3
= n 1 + + ··· + − ln n + −γ
2 n+1 2
n
1
+ k − + ln(2k + 3) − ln(2k + 1)
k+1
k=1

3
= n Hn+1 − ln n + −γ
2
− (n + 1) + Hn+1 + n ln(2n + 3) − ln(2n + 1)! + n ln 2 + ln n!,
where Hk = 1 + 12 + · · · + k1 denotes the kth harmonic number.

Using Stirling’s formula,
1 2n + 1
ln n! ∼ ln(2π ) + ln n − n,
2 2
we obtain

3
Sn ∼ n Hn+1 − ln n + − γ + Hn+1 − ln(n + 1)
2
4n2 + 6n 1 n n+1
+ n ln + ln + ln ,
4n + 4n + 1 2 2n + 1
2 2n + 1
and it follows that

1 1 1 1 1 3
lim Sn = γ + + ln + ln = γ + − ln 2.
n→∞ 2 2 2 2 2 2
(b) First we note that, if p ∈ N, then one has

∞
kp 1 1 γ
ln + = ln 1 + + . (10.1)
kp + 1 kp p p
k=1
Formula (10.1) follows from the Weierstrass product formula for the Gamma
function [150, p. 236]
∞
1 - z −z
= zeγ z 1+ e n , z = 0, −1, −2, . . . ,
(z) n
n=1
by setting z = p1 and by taking logarithms of both sides.

Now we are ready to calculate the series for any p ≥ 1. We have

1 1 1 2−p
1 + + · · · + − ln k + −γ +
2 k p 2pk

1 1 1 kp 1 2k 1
= 1+ + · · · + − ln k + −γ + ln + − ln + ,
2 k 2 kp + 1 kp 2k + 1 2k
and this implies, based on part (a) and formula (10.1), that
∞
1 1 1 2−p
1+ + · · · + − ln k + −γ +
2 k p 2pk
k=1
∞
1 1 1
= 1 + + · · · + − ln k + −γ
2 k 2
k=1
∞ ∞
kp 1 2k 1
+ ln+ − ln +
kp + 1 kp 2k + 1 2k
k=1 k=1

1 3 1 γ 1 γ
= γ + − ln 2 + ln 1 + + − ln 1 + −
2 2 p p 2 2

1 1 1 √ 1
=γ + + − ln 2π + ln 1 + .
2 p 2 p
It is worth mentioning that, by letting p → ∞ in the previous series formula, we

obtain
∞
1 1 1 γ 1 √
1 + + · · · + − ln k − γ − = + − ln 2π ,
2 k 2k 2 2
k=1
which is problem 3.42 in [26].

(a) an = 1 and bn = On − γ2 − ln 2 − ln2n ;

(b) an = (−1)n and bn = On − γ2 − ln 2 − ln n
2 .
2.56.
(a) Use mathematical induction.

(b) Use Abel’s summation formula with
1 1
an = Hn and bn = ζ (3) − 1 − 3
− ··· − 3.
2 n
10.3 Series with Positive Terms
2.57. (a) Converges; (b) diverges; (c) diverges; (d) converges; (e) diverges; (f)
converges; (g) diverges; (h) converges; (i) converges.
2.60. (a) Converges for a ∈ (0, 1) and diverges for a > 1. When a = 1 we obtain
the generalized harmonic series, which converges for p > 1 and diverges for p ≤ 1;
(b) converges for a ∈ (0, 7) and diverges for a ≥ 7; (c) converges; (d) converges.
2.61. (a) Converges; (b) converges; (c) diverges. Parts (b) and (c) can also be solved
ln n = nln ln n < n2
1 1 1
using the comparison criterion. For n large enough, we have (ln n)
and (ln n)1ln ln n = (ln ln1 n)2 > eln1 n = n1 .
e
2.62. (a) Converges for a ∈ (0, 1) and diverges for a ≥ 1; (b) converges; (c)
converges; (d) converges.
2.63. (a) Converges; (b) converges; (c) converges for a ∈ 0, 1e and diverges for
1
a ≥ 1e ; (d) diverges for a ∈ 0, 54 and converges for a > 54 ; (e) converges for

x ∈ 0, 1e and diverges for x ≥ 1e ; (f) converges for x ∈ 0, 1e and diverges for
x ≥ 1e .
c(a+c)···(an+c−a)
2.64. (a) Let t = max {a, b} and let xn = b(a+b)···(an+b−a) .
We have
n n n
ak + c − a b−c 1
ln xn = ln = ln 1 − ≤ −(b−c) ,
ak + b − a ak + b − a ak + b − a
k=1 k=1 k=1
where the last inequality follows since ln(1 − x) ≤ −x, ∀x ∈ [0, 1).
Thus,

1 1 1
ln xn ≤ −(b − c) + + ··· +
b a+b (n − 1)a + b

1 1 1
≤ −(b − c) + + ··· +
t 2t nt
b−c
=− Hn
t
b−c b−c
=− (Hn − ln n) − ln n.
t t
This implies that
b−c
e− max{a,b} (Hn −ln n)
xn ≤ b−c
, ∀n ≥ 1.
n max{a,b}
Part (a) of the problem is used
for proving that the series in part (c) converges.
2.66. When p ≤ 0 the series ∞ 1
n=1 np diverges since:
∞
1
p<0 un = = n−p −→ ∞ ⇒ un diverges;
np
n=1
∞
p=0 un = 1 0 ⇒ un diverges.
n=1
When p > 0 the function f (x) = 1

xp is decreasing on [1, ∞) and we have
n 1 n−p+1 1
p=1 vn = dx = − ;
1 x p −p + 1 −p + 1
n 1
p=1 vn = dx = ln n.
1 x
It follows, based on Cauchy’s integral criterion, that

∞
n−p+1 1 1
0 < p < 1 lim vn = lim − = ∞ ⇒ diverges;
n→∞ n→∞ −p + 1 −p + 1 np
n=1
∞
1
p = 1 lim vn = lim ln n = ∞ ⇒ diverges;
n→∞ n→∞ n
n=1
10.4 Alternating Series 321
∞
n−p+1 1 1 1
p > 1 lim vn = lim − = ⇒ converges.
n→∞ n→∞ −p + 1 −p + 1 p−1 np
n=1
2.68. (a) Diverges; (b) converges for p > 1 and diverges for p ≤ 1; (c) diverges;
(d) converges for p > 1 and diverges for p ≤ 1.

∞
ln n
2.69. The series diverges. Compare the series with n .
n=2
10.4 Alternating Series
2.70. (a) Semiconvergent; (b) absolutely convergent; (c) semiconvergent; (d)

semiconvergent; (e) semiconvergent; (f) absolutely convergent.
2.71. (a) The series converges. The series alternates since n + 1 + sin(n + 1) >
n + sin n, ∀n ≥ 1; (b) The series converges. We have an alternating series since
/ +
n 1 + cos(n + 1) > n + cos / n,∞∀n ≥ 1; (c) The series converges. Observe that
/ ∞ ∞ n / 1
/ (−1) n (−1) /≤
/ n+(−1)n sin n − n / (n−1)n = 1; (d) The series converges. Observe
n=2/ n=2 /
n=2
/ ∞ ∞ / ∞
that // (−1)n
n+(−1)n cos n −
(−1)n /
n /≤ (n−1)n = 1.
1
n=2 n=2 n=2
2.74.
(a) We have
1 1 1 1 1 1 1 1 1 1
1− + − + ··· + − = 1 + + + + ··· + +
2 3 4 2n − 1 2n 2 3 4 2n − 1 2n

1 1 1
−2 + + ··· +
2 4 2n
1 1 1 1 1
=1+ + + + ··· + +
2 3 4 2n − 1 2n

1 1 1
− 1 + + + ··· +
2 3 n
1 1 1
= + + ··· + .
n+1 n+2 2n
(b) Let S2n be the 2nth partial sum of the series. We have, based on part (a), that
1 1 1 1 1
S2n = 1 − + − + ··· + −
2 3 4 2n − 1 2n

1 1 1 1 1
= 1 + + + + ··· + + − ln(2n)
2 3 4 2n − 1 2n

1 1 1
− 1 + + + · · · + − ln n + ln 2,
2 3 n
and it follows that lim S2n = ln 2.

n→∞
Remark. From the solution of part (b) of Problem 2.74 it follows that

1 1 1
lim + + ··· + = ln 2.
n→∞ n+1 n+2 2n
More generally, when k ≥ 2 is an integer, it can be proved that

1 1 1
lim + + ··· + = ln k.
n→∞ n+1 n+2 kn

∞
(−1)n−1
2.75. By calculating the 2nth partial sum of the series, we have that n+x =
n=1

∞
1 1 1 1
(x+2n)(x+2n−1) . Since (x+2n)2
< (x+2n)(x+2n−1) < (x+2n−1)2
, we get that
n=1
∞ ∞ ∞
1 1 1
x <x <x .
(x + 2n)2 (x + 2n)(x + 2n − 1) (x + 2n − 1)2
n=1 n=1 n=1

∞
Now the problem is solved if we prove that lim x 1
(x+n)2
= 1.
x→∞ n=1
We have
∞ ∞ n+1 ∞
1 1 1 1
= dt = dt < .
x+1 1 (x + t)2 (x + t)2 (x + n)2
n=1 n n=1
10.5 Series with Harmonic Numbers and Factorials 323

∞
Similarly, one can prove that 1
(x+n)2
< 1
x+1 + 1
(x+1)2
. It follows that
n=1
∞
x 1 x x
<x < + , x > 0.
x+1 (x + n) 2 x + 1 (x + 1)2
n=1
2.76. We prove the first limit in Remark 2.8. Let h > 0.

Since f decreases, we have that
n−1 n n−1 k+1 n−1

f ((k + 1)h) ≤ f (xh)dx = f (xh)dx ≤ f (kh)
k=0 0 k=0 k k=0
and it follows that

n
f (h) + f (2h) + · · · + f (nh) ≤ f (xh)dx ≤ f (0) + f (h) + · · · + f ((n − 1)h).
0
n
1 nh
This implies, since 0 f (xh)dx = h 0 f (t)dt, that
nh
h (f (h)+f (2h)+ · · · +f (nh)) ≤ f (t)dt ≤ h (f (0)+f (h)+ · · · +f ((n−1)h)) .
0
Passing to the limit, as n → ∞, in the previous inequalities we get that

∞ ∞ ∞
h f (nf ) ≤ f (t)dt ≤ hf (0) + h f (nh).
n=1 0 n=1

∞ ∞
This implies that lim h f (nh) = 0 f (t)dt. The second limit can be proved
h→0+ n=1
similarly.
10.5 Series with Harmonic Numbers and Factorials
2.77.

∞ ∞ ∞ ∞
(i − 1)!(j − 1)! (j − 1)!
Hi+j = (i − 1)! Hi+j
(i + j )! (i + j )!
i=1 j =1 i=1 j =1
∞
Hi 1
= (i − 1)! + 2
i · i! i · i!
i=1
∞ ∞
Hi 1
= +
i2 i3
i=1 i=1
= 3ζ (3).

∞
Hi
We used that i2
= 2ζ (3) (see Problems 7.60 and 7.61).
i=1
(c) We have, based on part (a), that
∞ ∞ ∞ ∞
(i − 1)!(j − 1)! (j − 1)!
(−1)i−1 Hi+j = (−1)i−1 (i − 1)! Hi+j
(i + j )! (i + j )!
i=1 j =1 i=1 j =1
∞
Hi 1
= (−1) i−1
(i − 1)! + 2
i · i! i · i!
i=1
∞ ∞
Hi 1
= (−1)i−1 + (−1)i−1
i2 i3
i=1 i=1
11
= ζ (3).
8
∞
We used that (−1)i−1 Hi 2i = 58 ζ (3). This series can be calculated by
1i=1n
using the formula 0 x ln(1 − x)dx = − Hn+1 n+1
, n ≥ 0, and by observing that
∞
Hi 1 ln(1 − x) ln(1 + x)
(−1)i−1 =− dx,
i2 0 x
i=1
where the integral is calculated in the solution of Problem 2.81.

(d) Use symmetry combined to part (c) of the problem.
2.78. We have, based on part (a) of Problem 2.77, with i replaced by i + 1, that
∞ ∞ ∞ ∞
(i − 1)!(j − 1)! (j − 1)!
Hi+j +1 = (i − 1)! Hi+j +1
(i + j + 1)! (i + j + 1)!
i=1 j =1 i=1 j =1
∞
Hi+1 1
= (i − 1)! +
(i + 1) · (i + 1)! (i + 1) · (i + 1)!
2
i=1
∞ ∞
Hi+1 1
= +
i(i + 1)2 i(i + 1)3
i=1 i=1
∞ ∞
Hi Hi+1 1 Hi+1 1
= − + − +
i i + 1 i(i + 1) (i + 1)2 i(i + 1)3
i=1 i=1
= 1 + 1 − 2ζ (3) + 1 + 3 − ζ (2) − ζ (3)

= 6 − ζ (2) − 3ζ (3).
10.6 A Mosaic of Series
2.79. Solution I (A jewel in the theory of series).

We have
∞
1 1
S= (−1) n−1
n ζ (2)−1− 2 − · · · − 2 −1
2 n
n=1
∞
1 1
= ζ (2)−2+ (−1)n−1 n ζ (2)−1− 2 − · · · − 2 −1
2 n
n=2
∞
n−1=m 1 1
= ζ (2)−2+ (−1) (m+1) ζ (2)−1− 2 − · · · −
m
−1
2 (m+1)2
m=1
∞
1 1
= ζ (2)−2− (−1)m−1 m ζ (2)−1− 2 − · · · − 2 −1
2 m
m=1
∞ ∞
1 1 (−1)m−1
− (−1) m−1
ζ (2)−1− 2 − · · · − 2 +
2 m m+1
m=1 m=1
∞ ∞
1 1 (−1)j −2
= ζ (2)−2−S− (−1)m−1 + +··· +
(m+1) (m+2)2
2 j
m=1 j =2
(continued)
∞
1 1
= ζ (2)−2−S+ (−1) k−1
+ +··· − ln 2+1
k 2 (k+1)2
k=2
(∗) π2
= ζ (2)−1−S+ −ζ (2)
8
π2
= −1− ln 2−S.
8
We used at step (*) Problem 2.46.
Solution II. First we observe that

∞ ∞
1 1 1 1 1
ζ (2) − 1 − 2 − · · · − 2 = > =
2 n (n + k)2 (n + k)(n + k + 1) n+1
k=1 k=1
and
∞ ∞
1 1 1 1 1
ζ (2) − 1 − − ··· − 2 = < = .
22 n (n + k)2 (n + k − 1)(n + k) n
k=1 k=1
1
These imply that n+1 < ζ (2) − 1 − 212 − · · · − n12 < n1 , ∀n ≥ 1.

Let xn = n ζ (2) − 1 − 212 − · · · − n12 − 1. A calculation shows that
1 1 1
xn+1 − xn = ζ (2) − 1 − − ··· − 2 − > 0.
22 n n+1
Also, from the previous inequalities, we have that

1 1
lim n ζ (2) − 1 − 2 − · · · − 2 = 1,
n→∞ 2 n

∞
which implies that lim xn = 0. Thus, the series (−1)n−1 (−xn ) is a Leibniz
n→∞ n=1
series; hence, it converges.
Now we calculate the series. We apply Abel’s summation formula with

1 1
an = (−1)n−1 and bn = n ζ (2) − 1 − 2 − · · · − 2 − 1
2 n
and we have, since lim (a1 + a2 + · · · + an )bn+1 = 0, that

n→∞
∞
1 1
(−1)n−1 n ζ (2) − 1 − 2 − · · · − 2 − 1
2 n
n=1
∞ 1

1 1

= (−1) 1−1
+ · · · + (−1) n−1
− ζ (2) − 1 − 2 − · · · − 2
n+1 2 n
n=1
∞
1 1 1
= − ζ (2) − 1 − 2 − · · · − .
2k 2 (2k − 1)2
k=1
We apply one more time Abel’s summation

formula, with ak = 1 and bk =
1
2k − ζ (2) − 1 − 212 − · · · − (2k−1)
1
2 , and we have that
∞
1 1 1
− ζ (2) − 1 − 2 − · · · −
2k 2 (2k − 1)2
k=1

1 1 1
= lim k − ζ (2) − 1 − 2 − · · · −
k→∞ 2k + 2 2 (2k + 1)2
∞
1 1 1 1
+ k − − −
2k 2k + 2 (2k)2 (2k + 1)2
k=1
∞
1 1 1 1
= − − +
2(k + 1) 4k 2(2k + 1) 2(2k + 1)2
k=1
∞ ∞
1 1 1 1 1
= − − +
2(k + 1) 4k 2(2k + 1) 2 (2k + 1)2
k=1 k=1
ln 2 π2
1
=− + − .
2 16 2
2.80. Part (a) follows by direct computations and the series in part (b) telescopes.
2.81.
(a) Let k ≥ 1 be an integer. We have
1 1 1 (−1)k−1
− + + ··· +
n+1 n+2 n+3 n+k
1
= x n − x n+1 + · · · + (−1)k−1 x n+k−1 dx
0
1 1 − (−x)k
= xn dx
0 1+x
and it follows that

1
1 1 1 (−1)k−1 xn
lim − + + ··· + = dx.
k→∞ n + 1 n+2 n+3 n+k 0 1+x
(b) First we prove that the following logarithmic integral formula holds
1 ln(1 − x) ln(1 + x) 5
dx = − ζ (3). (10.2)
0 x 8
We have, since ab = 12 (a 2 + b2 − (a − b)2 ), that
1 ln(1 − x) ln(1 + x) 1 ln2 (1 − x)

1 1 1 ln2 (1 + x)
dx = dx + dx
0 x 2 0 x 2 0 x

2 1−x
1 1 ln 1+x
− dx
2 0 x

1−x
ζ (3) 1 1 ln2 1+x
= ζ (3) + − dx
8 2 0 x
1−x
1+x =t 9 1 ln2 t
= ζ (3) − dt
8 0 1 − t2
9 1 1 ln2 t 1 1 ln2 t
= ζ (3) − dt − dt
8 2 0 1−t 2 0 1+t
5
= − ζ (3),
8
1 2
since 0 ln (1+x)
x dx = ζ (3)
4 (see [6, pp. 291–292]).
Now we are ready to solve part (b) of the problem. We have, based on part (a),
that
∞ ∞ 1
Hn 1 1 1 Hn xn
− + −··· = dx
n n+1 n+2 n+3 n 0 1+x
n=1 n=1
∞

1 1 Hn n
= x dx
0 1+x n
n=1
1
7.86 1 1
= Li2 (x)+ ln2 (1−x) dx
0 1+x 2
1 Li2 (x) 1 1 ln2 (1−x)
= dx+ dx.
0 1+x 2 0 1+x
We calculate the first integral by parts and we have, based on (10.2), that
1 Li2 (x) /1 1 ln(1 − x) ln(1 + x)

dx = Li2 (x) ln(1 + x)/0 + dx
0 1+x 0 x
5
= ζ (2) ln 2 − ζ (3).
8
We calculate the second integral and we have
1 ln2 (1 − x) 1 ln2 t
dx = dt
0 1+x 0 2−t
1 1 ln2 t
= dt
2 0 1 − 2t
1 ∞ n
1 t
= ln2 t dt
2 0 2
n=0
∞ 1
1 1
= t n ln2 t dt
2 2n 0
n=0
∞
1
=2
2n+1 (n + 1)3
n=0

1
= 2Li3 .
2
Putting all these together and using that (see [21, p. 44])

1 1
Li3 = −2π 2 ln 2 + 4 ln3 2 + 21ζ (3) ,
2 24
part (b) of the problem is solved.

∞
Hn 1 1 1
(−1)n − + − ···
n n+1 n+2 n+3
n=1
∞
Hn 1 x n
= (−1)n dx
n 0 1+x
n=1
∞
1 1 Hn
= n
(−x) dx
0 1+x n
n=1
1
7.86 1 1
= Li2 (−x) + ln2 (1 + x) dx
0 1+x 2
1 Li2 (−x) 1 1 ln2 (1 + x)
= dx + dx
0 1+x 2 0 1+x
1 Li2 (−x) 1
= dx + ln3 2.
0 1+x 6
We calculate the preceding integral by parts and we have that
1 Li2 (−x) /1 1 ln2 (1+x) π2 ζ (3)

dx=Li2 (−x) ln(1+x)/0 + dx=− ln 2+ ,
0 1+x 0 x 12 4
and part (c) of the problem is solved.

(d) We have, based on part (a), that
∞ 2 ∞ 1 1
1 1 1 xnyn
Hn − + − ··· = dxdy
n+1 n+2 n+3 (1 + x)(1 + y)
n=1 n=1 0 0
1 1 ∞
1
= Hn (xy)n dxdy
0 0 (1 + x)(1 + y)
n=1
3.63 (a)
1 1 ln(1 − xy)
= − dxdy
0 0 (1 + x)(1 + y)(1 − xy)
1
1 1 ln(1 − xy)
=− dy dx.
0 1+x 0 (1 + y)(1 − xy)
We calculate the inner integral and we have that
1 ln(1−xy) 1 1 ln(1−xy) 1 1 x ln(1−xy)

dy = dy+ dy
0 (1+y)(1−xy) 1+x 0 1+y 1+x 0 1−xy
1 2
1 ln(1−xy) ln (1−x)
= dy− .
1+x 0 1+y 2(1+x)
We calculate the preceding logarithmic integral and we have
1 ln(1 − xy) xy=t x ln(1 − t)

dy = dt
0 1+y 0 x+t
x+t=u
2x ln(1 + x − u)
= du
x u
2x
u=(1+x)v 1+x ln(1 − v)
= ln 2 ln(1 + x) + dv.
x
1+x
v
Putting all these together, we have that
1 1
1 ln(1−xy)
dy dx
0 1+x 0 (1+y)(1−xy)
2x

1 1 ln(1−v) ln2 (1−x)
1+x
= ln 2 ln(1+x)+ dv− dx
0 (1+x)2 x
1+x
v 2
2x
1 ln(1+x) 1 1 1+x ln(1−v) 1 1 ln2 (1−x)
= ln 2 2
dx+ 2
dv dx− dx
0 (1+x) 0 (1+x) x
1+x
v 2 0 (1+x)2
2x
1− ln 2 1 1 1+x ln(1−v) 1 1
= ln 2 + 2
dv dx− Li2 ,
2 0 (1+x) x
1+x
v 2 2
where the last equality follows based on part (ii) of Remark 7.3 after Problem 7.66.
We calculate the logarithmic integral by parts, with
2x
1+x ln(1 − v) ln(1 − x)
f (x) = dv and f (x) = ,
x
1+x
v x(1 + x)
g (x) = 1
(1+x)2
and g(x) = − 1+x
1
, and we have that
2x

1 1 1+x ln(1 − v)
dv dx
0 (1 + x)2 x
1+x
v
/
ln(1 − v) //1
2x
1 1+x 1ln(1 − x)
=− dv / + dx
1+x x
1+x
v 0 0 x(1 + x)
2

1 1 ln(1 − v) 1 1 1 1
=− dv + ln(1 − x) − − dx
2 1
2
v 0 x 1+x (1 + x)2
1 1 ln(1 − v) 1 ln(1 − x) 1 ln(1 − x) 1 ln(1 − x)

=− dv + dx − dx − dx
2 12 v 0 x 0 1+x 0 (1 + x)2

1 1 1 ln 2
= Li2 (1) − Li2 − Li2 (1) + Li2 +
2 2 2 2

1 1 1 ln 2
= − Li2 (1) + Li2 + .
2 2 2 2
1
We used in the preceding calculations the integrals 0 ln(1−x)
1+x dx = −Li2 2
1
1 ln(1−x)
(see Problem 7.66) and 0 (1+x)2 dx = − ln22 , which can be calculated by the
method given in part (ii) of Remark 7.3 after Problem 7.66.
It follows that
1
1 1 ln(1 − xy) 1 π2
dy dx = ln 2 − ln2 2 − ,
0 1+x 0 (1 + y)(1 − xy) 2 12
and the problem is solved.

2.82.
(a) Prove the formula by mathematical induction.

an = (−1)n n and bn = ζ (2) − 1 − 212 − · · · − n12 − n1 .
(c) Prove the formula by mathematical induction.
(d) Use Abel’s summation formula with
an = (−1)n n2 and bn = ζ (3) − 1 − 213 − · · · − n13 − 2n1 2 .
(a) an = (−1)n−1 and bn = ζ (k) − 1 − 21k − · · · − n1k ;

(b) an = (−1)n−1 n and bn = ζ (k) − 1 − 21k − · · · − n1k .
2.84.
(a) We use Abel’s summation formula with an = 1 and bn = 1

n! + (n+1)!
2
+ (n+2)!
3
+
· · · and we have, since bn − bn+1 = 1
n! + 1
(n+1)! + · · · , that
∞
1 2 3
+ + + ···
n! (n + 1)! (n + 2)!
n=1

1 2 3
= lim n + + + ···
n→∞ (n + 1)! (n + 2)! (n + 3)!
∞
1 1
+ n + + ···
n! (n + 1)!
n=1
∞
1 1
= n + + ···
n! (n + 1)!
n=1
∞
(∗) n(n + 1) 1
= ·
2 n!
n=1
∞
1 n+1
=
2 (n − 1)!
n=1
3
= e.
2
We used at step (*) Abel’s summation formula with an = n and bn = 1

n! +
(n+1)! + · · · .
1
Parts (b) and (c) of the problem can be solved similarly.
2.85.
(a) We use Abel’s summation formula with an = 1 and bn = 1

nk
+ (n+1)
2
k + (n+2)k +
3
· · · , and we have that

∞ ∞
1 2 3 1 1 1
+ + + · · · = n + + +···
nk (n+1)k (n+2)k nk (n+1)k (n+2)k
n=1 n=1
∞
(∗) n(n+1) 1
= · k
2 n
n=1
1
= (ζ (k−2)+ζ (k−1)).
2
We used at step (*) Abel’s summation formula with

an = n and bn = n1k + (n+1)
1
k + ···.
Part (b) of the problem can be solved similarly.
2.86.
(a) We use Abel’s summation formula with

1 1 1 2
an = n and bn = ζ (2) − 1 − 2 − · · · − 2 − ,
2 n n
and we have, since
bn −bn+1

1 1 1 1 1 1
=− 2
2 ζ (2)−1− 2
− · · · − 2
+ 2
− − ,
n(n+1) 2 (n+1) (n+1) n+1 n
that
∞
1 1 1 2
n ζ (2)−1− 2 − · · · − 2 −
2 n n
n=1
∞
1 1 1 1 1 1 1
=− 2 ζ (2)−1− 2 − · · · − + − −
2 n+1 2 (n+1)2 (n+1)2 n+1 n
n=1
∞ ∞
1 1 1 1 1
=− ζ (2)−1− 2 − · · · − 2
−
n+1 2 (n+1) 2 (n+1)3
n=1 n=1
∞ ∞
1 1 1 1
+ +
2 (n+1)2 2 n(n+1)
n=1 n=1
∞
1 1 1 1
=− ζ (2)−1− 2 − · · · − 2 +ζ (2)−1− (ζ (3)−1)
m 2 m 2
m=1
1 1
+ (ζ (2)−1)+
2 2
3 3 1
= ζ (2)− ζ (3)− .
2 2 2

∞
We used the formula 1
m ζ (2) − 1 − 1
22
− ··· − 1
m2
= ζ (3), which can
m=1
be proved using Abel’s summation formula with
am = m1 and bm = ζ (2) − 1 − 212 − · · · − m12 .
(b) We use Abel’s summation formula with

1 1 1 2
an = n3 and bn = ζ (3) − 1 − 3 − · · · − 3 − 2 ,
2 n 2n
and we have, since

3n + 1 1 1
bn − bn+1 =− 2 2 ζ (3) − 1 − 3 − · · · −
2n (n + 1)3 2 (n + 1)3
1 1 1
+ − − 2 ,
(n + 1)3 2(n + 1)2 2n
that
∞ ∞
1 1 1 2
n3 ζ (3) − 1 − 3 − · · · − 3 − 2 = An (bn − bn+1 ),
2 n 2n
n=1 n=1
where

3n + 1 1 1
An (bn − bn+1 ) = − 2 ζ (3) − 1 − 3 − · · · −
8(n + 1) 2 (n + 1)3
1 1 1
+ − − 2
(n + 1)3 2(n + 1)2 2n

1 2 1 1
=− 3− 2 ζ (3) − 1 − 3 − · · · −
8 n+1 2 (n + 1)3
1 1 1
+ − − 2 ,
(n + 1)3 2(n + 1)2 2n
and it follows, after some calculations, that

∞
1 1 1 2
n3 ζ (3)−1− 3 − · · · − 3 − 2
2 n 2n
n=1
∞ ∞ ∞
3 1 1 3 1 3 1
=− ζ (3)−1− 3 − · · · − 3
− 3
+
4 2 (n+1) 8 (n+1) 16 (n+1)2
n=1 n=1 n=1
∞ ∞
3 1 1 1 1 1
+ + ζ (3)−1− 3 − · · · −
16 n2 2 n+1 2 (n+1)3
n=1 n=1
∞ ∞ ∞
1 1 1 1 1 1
+ − −
4 (n+1)4 8 (n+1)3 8 n2 (n+1)
n=1 n=1 n=1
1 1 3 1
= − ζ (2)+ ζ (3)+ ζ (4)− .
2 2 8 16
We used in the preceding computations the formulae:
∞
1 1
• ζ (3) − 1 − 3 − · · · − = ζ (2) − 2ζ (3) + 1;
2 (n + 1)3
n=1
∞
1 1 1 ζ (4)
• ζ (3) − 1 − 3 − · · · − = − ζ (3) + 1.
n+1 2 (n + 1)3 4
n=1
Power Series
11
11.1 Convergence and Sum of Power Series
3.1. (a) (−5, 5); (b) [−3, 3); (c) (−2, 0]; (d) [−3, 7); (e) (−1, 3); (f) R; (g)
(−e, e); (h) R; (i) (−1, 1); (j) {0};
. (k) [−1, 1]; (l) [−1, 1].
1 if k = n!
3.2. (−1, 1). Observe that ak = ⇒ R = 1.
0 if k = n!
3.3. (a) We have
∞ ∞
x
(n + 1)x n+1 = mx m = , x ∈ (−1, 1).
(1 − x)2
n=0 m=1
(b) We have
∞ ∞
(n + 1)(n + 2)x n = m(m + 1)x m−1 , x ∈ (−1, 1).
n=0 m=1
For the calculation of the preceding series, we use the power series
∞
x2
mx m+1 = , x ∈ (−1, 1),
(1 − x)2
m=1
which can be obtained by multiplying the derivative of the geometric series by x 2 .
338 11 Power Series
We differentiate the preceding series, and we divide it by x = 0 to obtain

∞
2
m(m + 1)x m−1 = , x ∈ (−1, 1).
(1 − x)3
m=1
(c) Let
∞
x 3n−1
f (x) = (−1)n−1 , x ∈ (−1, 1].
3n − 1
n=1
We have
∞
x
f (x) = x (−x 3 )n−1 = , x ∈ (−1, 1).
1 + x3
n=1
For x ∈ (−1, 1), it follows that

x t 1 x 2 −x+1 1 2x − 1 π
f (x)−f (0)=f (x)= dt= ln + √ arctan √ + √ .
0 1+t 3 6 (x + 1) 2
3 3 6 3
When x = 1 we have
π ln 2
f (1) = lim f (x) = √ − .
x→1 3 3 3
x<1
(d) Let
∞
f (x) = (−1)n−1 (5n − 1)x 5n−2 , x ∈ (−1, 1).
n=1
We have
∞

x x
f (t)dt = (−1) n−1
(5n − 1)t 5n−2
dt
0 0 n=1
∞ x
= (−1)n−1 (5n − 1) t 5n−2 dt
n=1 0
∞
= x4 (−x 5 )n−1
n=1
x4
= .
1 + x5
11.1 Convergence and Sum of Power Series 339
It follows that

x4 4x 3 − x 8
f (x) = = , x ∈ (−1, 1).
1 + x5 (1 + x 5 )2
(e) We observe that

∞ ∞
(−1) (n + 1) x =
n 3 n
(−1)m−1 m3 x m−1 .
n=0 m=1
We differentiate the geometric series and we multiply it by x to obtain

∞
x
nx n = , x ∈ (−1, 1).
(1 − x)2
n=1
We differentiate the preceding power series, we multiply it by x, and we have

∞
x + x2
n2 x n = , x ∈ (−1, 1).
(1 − x)3
n=1
We differentiate the above series and replace x by −x to obtain

∞
x 2 − 4x + 1
(−1)n−1 n3 x n−1 = , x ∈ (−1, 1).
(1 + x)4
n=1
(f) We have, for x ∈ (−1, 1), that

∞ ∞
∞

x 4n−1 x x
= x2 t 4n−4
dt = x 2
t 4n−4
dt
4n − 3
n=1 n=1 0 0 n=1
x 1 x2 x2 1 + x
= x2 dt = arctan x + ln .
0 1−t 4 2 4 1−x
3.4. A. (a) [−1, 1); (b) R; (c) [−1, 1), for k = 2 and [−1, 1], for k ≥ 3.
B. Let x = 1. We have
340 11 Power Series
∞
1 1 1
S= e − 1 − − − ··· − xn
1! 2! n!
n=1
∞
1 1 1
= (e − 2)x + e − 1 − − − ··· − xn
1! 2! n!
n=2
∞
n−1=m 1 1 1
= (e − 2)x + e − 1 − − − ··· − x m+1
1! 2! (m + 1)!
m=1
∞ ∞
1 1 1 x m+1
= (e − 2)x + e − 1 − − − ··· − x m+1 −
1! 2! m! (m + 1)!
m=1 m=1

= (e − 2)x + xS − ex − 1 − x ,
−e x x
and it follows that S = e x−1 + 1.
When x = 1, the series can be calculated by Abel’s summation formula.
Part C can be solved similarly.
11.2 Maclaurin Series of Elementary Functions
3.5. Solve the problem by mathematical induction.

3.6. (a) The series equals 1e − sin 1. We have
16n2 + 4n − 1 (4n + 1)(4n + 2) − 8n − 3

un = =
(4n + 2)! (4n + 2)!
1 8n + 3 1 2(4n + 2) − 1
= − = −
(4n)! (4n + 2)! (4n)! (4n + 2)!
1 2 1
= − +
(4n)! (4n + 1)! (4n + 2)!

1 1 1 1 1 1
= − + − − − .
(4n)! (4n + 1)! (4n + 2)! (4n + 3)! (4n + 1)! (4n + 3)!
Therefore,
n n
1 1 1 1
uk = − + −
(4k)! (4k + 1)! (4k + 2)! (4k + 3)!
k=0 k=0
n
1 1
− −
(4k + 1)! (4k + 3)!
k=0
4n+3 2n+1
1 1
= (−1)k − (−1)k .
k! (2k + 1)!
k=0 k=0
It follows that
∞ n
16n2 + 4n − 1 1
= lim uk = − sin 1.
(4n + 2)! n→∞ e
n=0 k=0
(b) The series equals cos 1 − 1e . We have
16n2 + 12n + 1 (4n + 2)(4n + 3) − 8n − 5

vn = =
(4n + 3)! (4n + 3)!
1 8n + 5 1 2(4n + 3) − 1
= − = −
(4n + 1)! (4n + 3)! (4n + 1)! (4n + 3)!
1 2 1
= − +
(4n + 1)! (4n + 2)! (4n + 3)!

1 1 1 1 1 1
= − − − + − .
(4n)! (4n + 2)! (4n)! (4n + 1)! (4n + 2)! (4n + 3)!
Therefore,
n n
1 1
vk = −
(4k)! (4k + 2)!
k=0 k=0
n
1 1 1 1
− − + −
(4k)! (4k + 1)! (4k + 2)! (4k + 3)!
k=0
2n+1 4n+3
1 1
= (−1)k − (−1)k .
(2k)! k!
k=0 k=0
It follows that
∞ n
16n2 + 12n + 1 1
= lim vk = cos 1 − .
(4n + 3)! n→∞ e
n=0 k=0
3.7. We have
1 1 2 8
un = = + −
n(2n + 1)(4n + 1) n 2n + 1 4n + 1
342 11 Power Series

1 1 1 1 1 1
=4 − + − −4 − .
4n 4n + 1 4n + 2 4n + 3 4n + 1 4n + 3
Therefore,
n n n
1 1 1 1 1 1
uk = 4 − + − −4 −
4k 4k + 1 4k + 2 4k + 3 4k + 1 4k + 3
k=1 k=1 k=1
4n+3 2n+1
k−1 1 1 1 1 1
= −4 (−1) +4 1− + −4 (−1) k
+4 1−
k 2 3 2k + 1 3
k=1 k=0
4n+3 2n+1
1 1
= 6−4 (−1)k−1 −4 (−1)k .
k 2k + 1
k=1 k=0
It follows that
∞ n
1
= lim uk = 6 − 4 ln 2 − 4 arctan 1 = 6 − 4 ln 2 − π.
n(2n + 1)(4n + 1) n→∞
n=1 k=1
3.8.
(a) We have
∞ ∞
ex − e−x 1 xn 1 (−x)n
sinh x = = −
2 2 n! 2 n!
n=0 n=0
∞ ∞
1 1 − (−1)n n n=2m−1 x 2m−1
= x = , x ∈ R.
2 n! (2m − 1)!
n=0 m=1
(b) We have
∞
1 + cos(2x) 22n−1 2n
cos2 x = =1+ (−1)n x , x ∈ R.
2 (2n)!
n=1
(c) We use the identity sin(3x) = 3 sin x − 4 sin3 x, x ∈ R, and we have that
∞ ∞

3 sin x − sin(3x) 1 x 2n+1 (3x)2n+1
sin x =
3
= 3 (−1) n
− (−1)n
4 4 (2n + 1)! (2n + 1)!
n=0 n=0
∞
1 3 − 32n+1
= (−1)n x 2n+1 , x ∈ R.
4 (2n + 1)!
n=0
(d) We have
∞ ∞
(−3x)n 3n n+2
x 2 e−3x = x 2 = (−1)n x , x ∈ R.
n! n!
n=0 n=0
(e) We have
∞ ∞ ∞
3x − 1 −3 2
= + = −3 x n +2 (n+1)x n = (2n−1)x n ,
(x − 1) 2 1 − x (1 − x)2
n=0 n=0 n=0
for x ∈ (−1, 1).

2
(f) We use the binomial series with α = − 12 and x → − x9 and we have that
⎡ ⎤
− 2
1
∞ − 12 − 32 · · · −n+ 12 x 2 n
1 1 x2 1
√ = 1− = ⎣1+ − ⎦
9−x 2 3 9 3 n! 9
n=1
∞
1 1 (2n−1)!! 2n
= + x , x ∈ (−3, 3).
3 3 (2n)!! · 32n
n=1
(g) We have
∞ ∞

1 1+x 1 x n+1 x n+1
ln = (−1) n
+
2 1−x 2 n+1 n+1
n=0 n=0
∞
1 (−1)n + 1 n+1
= x
2 n+1
n=0
∞
x 2n+1
= , x ∈ (−1, 1).
2n + 1
n=0
(h) We have
∞ ∞
xn (5x)n
f (x) = ln(1 − x) + ln(1 + 5x) = − + (−1)n−1
n n
n=1 n=1
∞
(−1)n−1 5n − 1 n 1 1
= x , x∈ − , .
n 5 5
n=1
344 11 Power Series
(i) We use the binomial series and we have

∞
1 1 (2n − 1)!! 2n
f (x) = √ = (1+x 2 )− 2 = 1+ (−1)n x , x ∈ (−1, 1).
1 + x2 n=1
(2n)!!
We integrate the preceding relation and we obtain

∞
(2n − 1)!!
f (x) = x + (−1)n x 2n+1 , x ∈ [−1, 1].
(2n)!!(2n + 1)
n=1
(j) We have
∞ 2n ∞
n (3x) (3x)2n−1
cos(3x) + x sin(3x) = (−1) +x (−1)n−1
(2n)! (2n − 1)!
n=0 n=1
∞ ∞
(3x)2n (3x)2n−1
=1+ (−1)n +x (−1)n−1
(2n)! (2n − 1)!
n=1 n=1
∞
32n−1 (2n − 3) 2n
=1+ (−1)n−1 x , x ∈ R.
(2n)!
n=1
(k) We have
∞
1 + x2 1
= 1 + x + 2x 2 · =1+x+2 xn, x ∈ (−1, 1).
1−x 1−x
n=2
3.9. A method for calculating series of this form is given in [130].

3.10. Use that

1 1 1 1
= − ,
x2 + x + 1 α − β β(1 − β 2 x) α(1 − α 2 x)
√ √
where α = − 12 + 23 i and β = − 12 − 23 i.
3.12. Observe that (n − 2)! + (n + 2)! = (n − 2)!(n2 + n − 1)2 , ∀n ≥ 2. The second
series of part (a) is due to T. Andreescu [1].
3.13. (a) We have
* + ⎛ * +⎞
∞ ∞
log2 n log2 n
= ⎝ ⎠
n(n + 1) n(n + 1)
n=1 k=0 2k ≤n<2k+1
∞
1 1
= k −
n n+1
k=0 2k ≤n<2k+1
∞
k
=
2k+1
k=0
= 1.
(b) We have
⎛ ⎞
∞ ∞
1 ⎝ 1
* += * +⎠
n(n + 1) log2 n n(n + 1) log2 n
n=2 k=1 2k ≤n<2k+1
∞
1 1 1
= −
k n n+1
k=1 2k ≤n<2k+1
∞
1
=
k2k+1
k=1
ln 2
= .
2
3.14. We use the geometric series and we have
∞ ∞
1 1 1 1 1 x−2 n (x − 2)n
= = · = − = (−1)n ,
x+5 x−2+7 7 1+ 7
x−2 7 7 7n+1
n=0 n=0
for x ∈ (−5, 9).
Remark. Another method for solving the problem is based on the calculation of the
derivative of order n of the function f (x) = x+5
1
and its evaluation at 2, i.e.
(n)
1
(2).
x+5
346 11 Power Series
3.15. We use the logarithmic series and we have

∞
x−3 x−3 (x − 3)n
ln(1+x) = ln 4 1 + = ln 4+ln 1 + = ln 4+ (−1)n−1 ,
4 4 4n n
n=1
for x ∈ (−1, 7].
derivative of order n of the function f (x) = ln(1 + x) and its evaluation at 3, i.e.
(ln(1 + x))(n) (3).
3.16. We use the power series expansion

∞
1
= nx n−1 , x ∈ (−1, 1),
(1 − x)2
n=1
and we have that

∞ n−1 ∞
1 1 1 x+1 n
= 2 = n = (x + 1)n−1 ,
(x − 1)2 4 2 2n+1
4 1− x+1
2
n=1 n=1
for x ∈ (−3, 1).
derivative of order n of the function f (x) = (x−1)
1
2 and its evaluation at −1, i.e.
(n)
1
(−1).
(x − 1)2
3.17. We have

1 1 1 1 1 1
= − = −
x + 4x + 3
2 2 x+1 x+3 4 1− x+5
8 1− x+5
2 4
∞ ∞
1 (x + 5)n 1 (x + 5)n
= −
4 2n 8 4n
n=0 n=0
∞
1 1
= − 2n+3 (x + 5)n ,
2n+2 2
n=0
for x ∈ (−7, −3).

3.18. We use the geometric series and we have
1 1 1 1
= = · 2
x2 − 4x + 8 (x − 2) + 4
2 4
1+ x−2
2
∞ 2n ∞
1 x−2 (x − 2)2n
= (−1) n
= (−1)n ,
4 2 22n+2
n=0 n=0
for x ∈ (0, 4).

3.19. We use the power series for the exponential function and we have
∞ ∞
1 2(x+1) 1 (2(x + 1))n 2n
e2x−1 = e2(x+1)−3 = ·e = 3 = (x + 1)n ,
e3 e n! e3 n!
n=0 n=0
for x ∈ R.
derivative of order n of the function f (x) = e2x−1 and its evaluation at −1, i.e.
(e2x−1 )(n) (−1).
3.20. We use the power series expansion for the sine function and we have
∞
32n+1
sin(3x+π ) = sin(3(x−π )+4π ) = sin(3(x−π )) = (−1)n (x−π )2n+1 ,
(2n + 1)!
n=0
for x ∈ R.
derivative of order n of the function f (x) = sin(3x + π ) and its evaluation at π , i.e.
(sin(3x + π ))(n) (π ).
3.21. We use the trigonometric formula
a+b
arctan a + arctan b = arctan , ab < 1,
1 − ab
and we have that

348 11 Power Series

1 1
arctan x − arctan 2 = − arctan (−x) + arctan 2
x −x+1 x −x+1
−x + 1
x 2 −x+1
= − arctan
1+ x
x 2 −x+1
= − arctan (−(x − 1))
= arctan (x − 1), ∀x ∈ R.
Therefore, f (x) = arctan (x −1), for x ∈ R. Now, using the power series expansion
of the arctan function we write
∞
1 (x − 1)2n+1
arctan x − arctan = arctan (x − 1) = (−1)n ,
x2 − x + 1 2n + 1
n=0
for x ∈ [0, 2].
11.3 Gems with Numerical and Power Series
3.22. (c) We have

∞ ∞
n xn
xn =
(n − 2)! + (n − 1)! + n! (n − 2)!n
n=2 n=2
∞
n−1 n
= x
n!
n=2
∞ ∞
xn xn
= −
(n − 1)! n!
n=2 n=2

= x ex − 1 − ex − 1 − x
= (x − 1)ex + 1.

3.23. (a) We have, based on part (a) of Problem 2.81, that
∞ ∞ 1
1 1 1 xn
(−1)n − + − ··· = (−1)n dx
n+1 n+2 n+3 0 1+x
n=0 n=0
1 ∞
1
= (−x)n dx
0 1+x
n=0
1 1
= dx
0 (1 + x)2
1
= .
2
(c) Observe that S4n−1 = 1 − 1

3 + 1
5 − ··· − 1
4n−1 , n ≥ 1.
(d) We have
∞
(−1) k x n = 1 + x + x 2 + · · · + x k−1 − (x k + x k+1 + · · · + x 2k−1 )
n
n=0
+ (x 2k + x 2k+1 + · · · + x 3k−1 ) − · · ·
= (1 + x + x 2 + · · · + x k−1 )(1 − x k + x 2k − x 3k + · · · )
1 + x + x 2 + · · · + x k−1
= .
1 + xk
(e) We have
∞ ∞ 1
1 1 1 xn
(−1) k (−1) k
n 2.81 (a) n
− + − ··· = dx
n+1 n+2 n+3 0 1+x
n=0 n=0
∞

1 1
(−1) k x n dx
n
=
0 1+x
n=0
(d)
1 1 + x + · · · + x k−1
= dx.
0 (1 + x)(1 + x k )
(f) We have
∞ ∞ x
x n+1
(−1) 2 (−1) 2
n n
= t n dt
n+1 0
n=0 n=0
∞

x
= n
n
(−1) 2 t dt
0 n=0
(b)
x 1+t
= dt
0 1 + t2
1
= arctan x + ln(1 + x 2 ).
2
350 11 Power Series
3.24. We have, based on Problem 2.81, that

∞
1 1 1 1
− + ··· ζ (2) − 1 − 2 − · · · − 2
n+1 n+2 2 n
n=1
∞ 1
xn 1 1
= dx ζ (2) − 1 − 2 − · · · − 2
1+x 2 n
n=1 0
1 ∞
1 1 1
= ζ (2) − 1 − 2 − · · · − 2 x n dx
0 1+x 2 n
n=1
3.4 C
1 xζ (2) − Li2 (x)
= dx
0 1 − x2
ζ (2) 1 x 1 1 Li2 (x) 1 1 xζ (2) − Li2 (x)
= − dx + dx.
2 0 1+x 2 0 1+x 2 0 1−x
We calculate the second integral. We have
1 Li2 (x) /1 1 ln(1 − x) ln(1 + x)

dx = Li2 (x) ln(1 + x)/0 + dx
0 1+x 0 x
5
= ζ (2) ln 2 − ζ (3).
8
The last equality holds based on formula (10.2).
On the other hand,
1 xζ (2) − Li2 (x) /1
dx = − (xζ (2) − Li2 (x)) ln(1 − x) /0
0 1−x
1 1 ln2 (1 − x)
+ ζ (2) ln(1 − x)dx + dx
0 0 x
1 1 ln2 (1 − x)
= ζ (2) ln(1 − x)dx + dx
0 0 x
= −ζ (2) + 2ζ (3).
Putting all these together, we have that
∞
1 1 1 1
− + ··· ζ (2) − 1 − 2 − · · · − 2
n+1 n+2 2 n
n=1
1 xζ (2) − Li2 (x) 21
= dx = ζ (3) − ζ (2) ln 2.
0 1 − x2 16
3.25. First we prove the following formula:
1 1 1 1 1 1+x
+ − − + ··· = xn dx. (11.1)
n+1 n+2 n+3 n+4 0 1 + x2
We have
1 1 1 1 1
+ − − +··· = x n +x n+1 −x n+2 −x n+3 + · · · dx
n+1 n+2 n+3 n+4 0
1
= x n (1+x−x 2 −x 3 +x 4 +x 5 − · · · )dx
0
1
= x n (1+x)(1−x 2 +x 4 −x 6 + · · · )dx
0
1 1+x
= xn dx.
0 1+x 2
(a) We have, based on formula (11.1), that

∞ ∞ 1
1 1 1 1 1+x
(−1)n−1 + − − +··· = (−1)n−1 xn dx
n+1 n+2 n+3 n+4 0 1+x 2
n=1 n=1
1 ∞
1+x
=− (−x)n dx
0 1+x 2
n=1
1 x
= dx
0 1+x 2
ln 2
= .
2
(b) We have, based on formula (11.1), that
352 11 Power Series
∞
1 1 1 1
(−1)n−1 Hn + − − + ···
n+1 n+2 n+3 n+4
n=1
∞ 1 1+x
= (−1)n−1 Hn xn dx
0 1 + x2
n=1
∞
1 1+x
=− Hn (−x)n dx
0 1 + x2
n=1
3.63 (a)
1 ln(1 + x)
= dx.
0 1 + x2
We calculate the preceding integral. Using the substitution x = 1−t

1+t , we have that
1 ln(1 + x) 1 1
I= dx = ln 2 dt − I,
0 1 + x2 0 1 + t2
and it follows that I = π ln

8 .
2
(c) We have, based on formula (11.1), that

∞ 2
1 1 1 1
S= + − − + ···
n n+1 n+2 n+3
n=1
∞ 1 1+x 1 1+y
= x n−1 dx y n−1 dy
1 + x2 1 + y2
n=1 0 0
∞
1 1 (1 + x)(1 + y)
= (xy)n−1 dxdy
0 0 (1 + x 2 )(1 + y 2 )
n=1
(1 + x)(1 + y)
1 1
= dxdy
0 0 (1 + x )(1 + y )(1 − xy)
2 2
1 1+x 1 1+y

= dy dx.
0 1+x 0 (1 + y )(1 − xy)
2 2
We calculate the inner integral and we have that

⎛ ⎞
x 2 +x
1 1+y 1 1−x
+
1+x
y
dy = ⎝ 1+x 2 1+x 2
+ 1+x 2 ⎠ dy
0 (1 + y 2 )(1 − xy) 0 1+y 2 1 − xy
π 1−x ln 2 1 + x 1+x
= · + · − ln(1 − x).
4 1+x 2 2 1+x 2 1 + x2
It follows that
1 1 + x 2
π1 − x2
1 ln 2 1 1 + x 2
S= dx + dx − ln(1 − x)dx
0 (1 + x ) 2 0 1 + x2 1 + x2
4 2 2
0
1 1 + x 2
π ln 2 π 1
= + + − ln(1 − x)dx
8 2 4 2 0 1 + x2
π ln 2
= +G+ ,
4 2
since
1 1+x
2
π ln 2 π ln 2
ln(1 − x)dx = −G− − .
0 1 + x2 8 8 4
To calculate the previous integral we integrate by parts, with f (x) = ln(1 − x),
2
f (x) = − 1−x
1
, g (x) = 1+x
1+x
2 , g(x) = arctan x − 1+x
1
2 + 2 − 4 , and we get that
1 π
1 2 /1
1+x 1 1 π /
ln(1 − x)dx = ln(1 − x) arctan x − + − /
1 + x2 1+x 2 2 4 /
0 0
1 arctan x − arctan 1 + 1
2 − 1
1+x 2
+ dx
0 1−x
1 arctan x − arctan 1 1 1+x 1
= dx − dx
0 1−x 2 0 1 + x2

1 arctan x − arctan 1 1 π ln 2
= dx − +
0 1−x 2 4 2
π ln 2 π ln 2
= −G− − ,
8 8 4
since
1 arctan x − arctan 1 π ln 2
A Catalan integral dx = − G.
0 1−x 8
We prove the previous equality using integration by parts and we have that
354 11 Power Series
1 arctan x − arctan 1 /1
dx = − ln(1 − x) (arctan x − arctan 1) /0
0 1−x
1 ln(1 − x)
+ dx
0 1 + x2
1 ln(1 − x)
= dx
0 1 + x2
(11.2)
x= 1−t 1 ln 2 + ln t − ln(1 + t)
=
1+t
dt
0 1 + t2
π ln 2 1 ln t 1 ln(1 + t)
= + dt − dt
4 0 1 + t2 0 1 + t2
π ln 2
= − G,
8
since
1 ln x 1 ln(1 + x) π ln 2
dx = −G and dx = .
0 1 + x2 0 1 + x2 8
The last integral was calculated in the solution of part (b).
Part (d) can be solved similarly to part (c) of the problem.
3.26. We have, based on part (a) of Problem 2.81 and formula (11.1), that
∞
1 1 1 1 1 1 1
+ − − + ··· − + − ···
n n+1 n+2 n+3 n n+1 n+2
n=1
∞ 1 1+x 1 y n−1
= x n−1 dx dy
1 + x2 1+y
n=1 0 0
∞
1 1 1+x
= (xy)n−1 dxdy
0 0 (1 + x 2 )(1 + y)
n=1
1 1 1+x
= dxdy
0 0 (1 + x 2 )(1 + y)(1 − xy)

1 1+x 1 1
= dy dx
0 1 + x2 0 (1 + y)(1 − xy)
1 ln 2 − ln(1 − x)
= dx
0 1 + x2
π ln 2 1 ln(1 − x)
= − dx
4 0 1 + x2
π ln 2
= + G.
8
The last integral follows based on formula (11.2).
11.4 Single Zeta Series
3.27. We have
∞ ∞ ∞ ∞ ∞ ∞
1 1 n 1
n (ζ (n + 1) − 1) = n = = = ζ (2).
i n+1 i in (i − 1)2
n=1 n=1 i=2 i=2 n=1 i=2
3.28. We have
n
∞ ∞ ∞ ∞ ∞ 1 ∞
ζ (n) − 1 1 1 k 1 1
= = =− ln 1 − + = 1−γ
n n kn n k k
n=2 n=2 k=2 k=2 n=2 k=2
and
n
∞ ∞ ∞ ∞ ∞ 1 ∞
ζ (2n) − 1 1 1 k2 1
= = =− ln 1 − 2 = ln 2.
n n k 2n n k
n=1 n=1 k=2 k=2 n=1 k=2
3.29. (e) We have

∞ ∞
∞

1 1 1
(k + 1)n ζ (n) − 1 − n − · · · − n − 1 = (k + 1) n
−1
2 k in
n=2 n=2 i=k+1
∞ ∞ n
k+1
=
i
n=2 i=k+2
∞ ∞ n
k+1
=
i
i=k+2 n=2
∞
(k + 1)2
=
i(i − k − 1)
i=k+2
= (k + 1)Hk+1 .
The other parts of the problem are solved similarly.

356 11 Power Series
3.30. (a) We have

∞ 0 1 ∞ ∞
1
22n−1 (ζ (2n − 1) − 1) − 1 = 22n−1
−1
i 2n−1
n=2 n=2 i=2
∞ ∞ 2n−1
2
=
i
n=2 i=3
∞ ∞ 2n
i 2
=
2 i
i=3 n=2
∞
1
=8
i(i 2 − 4)
i=3
11
= .
12
Part (b) is solved similarly.
3.31. (b) We have
∞
1 1
(−1)n (k + 1)n ζ (n) − 1 − n − · · · − n − 1
2 k
n=2
∞
∞

1
= (−1) (k + 1)
n n
−1
in
n=2 i=k+1
∞ ∞ n
k+1
= (−1)n
i
n=2 i=k+2
∞ ∞ n
k+1
= −
i
i=k+2 n=2
∞
(k + 1)2
=
i(i + k + 1)
i=k+2
= (k + 1)(H2k+2 − Hk+1 ).
Part (a) follows from part (b) with k = 1.

3.32. (b) We have
∞
1 1
(n − 1) (k + 1)n ζ (n) − 1 − n − · · · − n − 1
2 k
n=2
∞
∞

1
= (n − 1) (k + 1)n
−1
in
n=2 i=k+1
∞ ∞ n
k+1
= (n − 1)
i
n=2 i=k+2
∞ ∞ n
k+1
= (n − 1)
i
i=k+2 n=2
∞
(k + 1)2
=
(i − k − 1)2
i=k+2
= (k + 1)2 ζ (2).
3.33. We have
∞ ∞
∞

1 n 1 1
4 (ζ (2n) − 1) − 1 = 4n −1
n n i 2n
n=1 n=1 i=2
∞ ∞ n
1 4
=
n i2
n=1 i=3
∞ ∞ n
1 4
=
n i2
i=3 n=1
∞
4
=− ln 1 − 2
i
i=3
= ln 6.
The first series is calculated similarly.

3.34. Use the identity
ζ (n + 1) − 1
Hn (ζ (n) − ζ (n + 1)) = Hn (ζ (n)−1)−Hn+1 (ζ (n+1)−1)+ , n ≥ 2,
n+1
and the first series of Problem 3.28.

3.35. (a) See [16]. A different solution, for both parts (a) and (b), is based on Abel’s
summation formula. Part (c) follows from parts (a) and (b) by subtraction.
358 11 Power Series
3.36. To prove the first equality use the definition of the zeta function and change
the order of summation of terms, which is possible since the terms of the series are
positive, and to prove the second equality calculate the nth partial sum of the series.
3.37. We have
∞ ∞ ∞ ∞ ∞ ∞
nζ (2n) n 1 n i2
= = 16 = 64 = π 2.
4n−2 4n−2 i 2n (4i 2 )n (4i 2 − 1)2
n=1 n=1 i=1 i=1 n=1 i=1
3.38. (a) This is problem 3.8 in [26, p. 140].

(e) We have, based on part (a), that
∞ ∞
∞

n 1
2 (n − ζ (2) − ζ (3) − · · · − ζ (n)) − 1 = n
2 −1
k(k + 1)n
n=2 n=2 k=1
∞ ∞
2n
=
k(k + 1)n
n=2 k=2
∞ ∞ n
1 2
=
k k+1
k=2 n=2
∞
4
=
(k − 1)k(k + 1)
k=2
= 1.

3.39. (a) We have
∞ ∞ ∞ ∞ ∞
(−1)i π 1
(β(n) − 1) = β(1) − 1 + = −1+ (−1)i
(2i + 1)n 4 (2i + 1)n
n=1 n=2 i=1 i=1 n=2
∞
π (−1)i ln 2
= −1+ =− .
4 2i(2i + 1) 2
i=1

3.40. We have, based on Problem 2.31 and part (a) of Problem 3.39, that
∞ ∞ ∞
ln 2 π ln 2
(ζ (n) − β(n)) = (ζ (n) − 1) − (β(n) − 1) = β(1) + = + .
2 4 2
n=2 n=2 n=2
11.5 Polylogarithm Series
3.41. (m) We have

∞
x2 xk
(n − 1) (k + 1)n Lin (x) − x − n − · · · − n − x k+1
2 k
n=2
∞
∞

xi
= (n − 1) (k + 1)n
− x k+1
in
n=2 i=k+1
∞ ∞ n
k+1
= (n − 1) xi
i
n=2 i=k+2
∞ 2 ∞ n−2
k+1 k+1
= x i
(n − 1)
i i
i=k+2 n=2
∞
xi
= (k + 1)2
(i − k − 1)2
i=k+2
∞
x k+1+j
= (k + 1)2
j2
j =1
= (k + 1)2 x k+1 Li2 (x).

3.42. (a) We have
∞ ∞
xi xi 1 1
Sn = = − = Sn−1 − (Lin (x) − x) .
i (i − 1)
n i n−1 i−1 i
i=2 i=2
It follows that
Sn = S1 − Li2 (x) − Li3 (x) − · · · − Lin (x) + (n − 1)x

= nx + (1 − x) ln(1 − x) − Li2 (x) − Li3 (x) − · · · − Lin (x),
since S1 = (1 − x) ln(1 − x) + x.
360 11 Power Series
∞
(Li2 (x) + Li3 (x) + · · · + Lin+1 (x) − (n + 1)x − (1 − x) ln(1 − x))
n=1
∞ ∞ ∞ ∞ ∞
xi xi 1 xi
=− =− =−
i n+1 (i − 1) i−1 i n+1 (i − 1)(i 2 − i)
n=1 i=2 i=2 n=1 i=2
∞ i
xi xi x
=− − + = (1 − x) ln(1 − x) − xLi2 (x) + x.
(i − 1) 2 i−1 i
i=2

3.43. (a) We have
∞ ∞
xi xi 1 1
Sn = = − = (Lin (x) − x) − Sn−1 .
i n (i + 1) i n−1 i i+1
i=2 i=2
It follows, since S1 = 1−x

x ln(1 − x) − x
2 + 1, that
n
1−x x
(−1) Sn =
n
(−1) (Lik (x) − x) −
k
ln(1 − x) − + 1
x 2
k=2
n
x 1−x x
= (−1)k Lik (x) − (1 − (−1)n−1 ) − ln(1 − x) + − 1
2 x 2
k=2
n
1−x x
= (−1)k Lik (x) − ln(1 − x) − 1 + (−1)n−1 .
x 2
k=2

∞
1−x x
(−1)n Li2 (x) + · · · + (−1)n Lin (x) − ln(1 − x) − 1 −
x 2
n=2
∞ ∞ ∞ ∞ ∞
xi xi 1 1 1 1
= = = x i
− +
i n (i + 1) i+1 in 2(i − 1) i 2(i + 1)
n=2 i=2 i=2 n=2 i=2
(1 − x)2 3x 1
=− ln(1 − x) + − .
2x 4 2
Part (d) is solved similarly.
11.6 Inequalities and Integrals
3.44. It suffices to prove that x ln x ≤ ln(x 2 − x + 1), for x ∈ (0, 1]. We change
variables x = 1 − y and we obtain the inequality (1 − y) ln(1 − y) ≤ ln(1 − y + y 2 ),
for y ∈ [0, 1). Now we use the power series expansion of the logarithmic function
and we have that
∞ ∞
(y − y 2 )n yn
ln(1 − (y − y )) − (1 − y) ln(1 − y) = −
2
+ (1 − y)
n n
n=1 n=1
∞
yn
= (1 − y) 1 − (1 − y)n−1
n
n=1
≥ 0.
Remark. From the solution of Problem 3.44 we also have that the following
inequality holds true (1 − x)1−x ≤ x 2 − x + 1, for x ∈ [0, 1). For another
solution of Problem 3.44, which uses concavity, see the remark after the solution
of Problem 4.6.
3.45. Since (2n)! ≥ 2n n!, for all n ≥ 0, we have that

∞ ∞
ex + e−x x 2n x 2n x2
= ≤ n
=e2 .
2 (2n)! 2 n!
n=0 n=0
3.46. (a) Let f : [0, ∞) → R, f (x) = qx p + py q − pqxy. A calculation

1
shows that f (x) = pq(x p−1 − y) and f (x) = 0 ⇒ x = y p−1 . We have that
1 1
f decreases on [0, y p−1 ) and increases on (y p−1 , ∞), and it follows, since
1
f (y p−1 ) = 0, that f (x) ≥ 0, for all x ≥ 0.
(b) Solution due to C. Mortici [111]. The inequality is equivalent to
1 1 1
+ ≥ , x, y ∈ (0, 1).
p(1 − x ) q(1 − y )
p q 1 − xy
We have
∞ ∞
1 1 (x n )p (y n )q (a) 1
+ = + ≥ xnyn = .
p(1 − x ) q(1 − y q )
p p q 1 − xy
n=0 n=0
362 11 Power Series
3.47. (a) Solution due to C. Mortici [111]. Let S = a + b + c and x = a

S, y = b
S
and z = Sc . The inequality is equivalent to
x y z 3
+ + ≥ .
1−x 1−y 1−z 2
Jensen’s inequality for the convex function f (x) = x n , n ≥ 2, states that

n
x n + y n + zn x+y+z
≥ , x, y, z ≥ 0.
3 3
We have, since x + y + z = 1, that
x y z
∞
+ + = x n+1 + y n+1 + zn+1
1−x 1−y 1−z
n=0
∞ n+1
x+y+z
≥ 3
3
n=0
∞
1
=
3n
n=0
3
= .
2
Two new proofs of Nesbitt’s inequality are given in Chap. 7, after Problem 7.132.
3.48. (a) Solution due to C. Mortici [111]. First we note that, if x, y, and z are
nonnegative real numbers, then x 2 + y 2 + z2 ≥ xy + yz + xz. We have
1 1 1
∞
+ + = a 2n + b2n + c2n
1 − a2 1 − b2 1 − c2
n=0
∞

≥ a n bn + bn cn + a n cn
n=0
1 1 1
= + + .
1 − ab 1 − bc 1 − ac
(b) We observe that if a, b, c, and d are nonnegative real numbers, then
(continued)
a 3 + b3 + c3 + d 3 ≥ abc + bcd + cda + dab.
We prove the previous inequality. We have, based on the AM − GM

inequality, that

3(a 3 + b3 + c3 + d 3 ) = a 3 + b3 + c3 + b3 + c3 + d 3

+ c3 + d 3 + a 3 + d 3 + a 3 + b3
≥ 3abc + 3bcd + 3cda + 3dab.
Now, based on the previous inequality, we have that
1 1 1 1
∞
+ + + = a 3n +b3n +c3n +d 3n
1−a 1−b 1−c 1−d 3
3 3 3
n=0
∞

≥ (abc)n +(bcd)n +(cda)n +(dab)n
n=0
1 1 1 1
= + + + .
1−abc 1−bcd 1−cda 1−dab
3.49. (a) Use the recurrence relation Inn = 2(In − In+1 ), ∀ n ≥ 1.

(b) A direct calculation shows that
(2n − 3)!! π
In = · , ∀ n ≥ 2,
(2n − 2)!! 2
and use part (a) of the problem.

3.50. (a) We prove that In = − 2n1
+ 2n (In − In+1 ), n ≥ 1. We calculate In using
integration by parts, with
arctan x 1 2nx
f (x) = , f (x) = − arctan x,
(1 + x 2 )n (1 + x 2 )n+1 (1 + x 2 )n+1
g (x) = 1, g(x) = x, and we have that

/ ∞
x arctan x //∞ x x2
In = − − 2n arctan x dx
(1 + x 2 )n /0 0 (1 + x 2 )n+1 (1 + x 2 )n+1
∞ x ∞ x2
=− dx + 2n arctan x dx
0 (1 + x 2 )n+1 0 (1 + x 2 )n+1
364 11 Power Series
/∞ ∞
1 / arctan x arctan x
= / +2n − dx
2n(1 + x ) /0
2 n
0 (1 + x )
2 n (1 + x 2 )n+1
1
=− + 2n(In − In+1 ).
2n
In
It follows that n = − 2n1 2 + 2(In − In+1 ) and this implies
∞ ∞ ∞
In 1 1 1 π2
=− +2 (In − In+1 ) = − ζ (2) + 2I1 = ,
n 2 n2 2 6
n=1 n=1 n=1
∞ /
arctan2 x /∞ π2
since I1 = 0 arctan
1+x 2
x
dx = 2 0
= 8 .
(b) We have
∞ ∞
1 1
arctan x ln 1 + 2 dx = − arctan x ln 1 − dx
0 x 0 1 + x2
⎛ n ⎞
∞ 1
∞
⎜ 1+x 2 ⎟
= arctan x ⎝ ⎠ dx
0 n
n=1
∞
In
=
n
n=1
= ζ (2).
(c) Use the identity arctan x + arccot x = π2 , ∀ x > 0, and part (b) of the
problem.
3.52. Use that x1x = e−x ln x and x x = ex ln x .
3.53. (a) The integral equals 1 + 1−aa ln(1 − a). We have
" # ∞
1 1 x= y1 ∞ a y k+1 ak
a x
dx = dy = dy
0 1 y2 y2
k=1 k
∞
1 1
= ak −
k k+1
k=1
∞ ∞
ak ak
= −
k k+1
k=1 k=1
1
= − ln(1 − a) − (− ln(1 − a) − a)
a
1−a
= 1+ ln(1 − a).
a
(b) This part follows from part (a) with a = 12 .

3.54. We have
∞ ∞ k+1 ∞
1
2−x dx = 2−x dx = = 2.
0 2k
k=0 k k=0
The second integral can be calculated similarly.

3.55. We have
∞ ∞ k+1 ∞
k 1
x e−x dx = x e−x dx = 1 − e−1 = .
0 ek e−1
k=0 k k=0
3.56. We have
∞
∞ ∞
1 ln(1 − x) 1 x n−1 1 1 1
dx= − dx=− x n−1 dx= − = − ζ (2)
0 x 0 n n 0 n2
n=1 n=1 n=1
and
1 ln(1 − x) 2 1 ln2 x
dx = dx
0 x 0 (1 − x)2
∞
1
= 2
ln x (n + 1)x n
dx
0 n=0
∞ 1
= (n + 1) x n ln2 x dx
n=0 0
∞
1
=2
(n + 1)2
n=0
= 2ζ (2).
Another method for calculating the second integral is based on the identity of
part (c) of Problem 3.63. We have
366 11 Power Series
1 ln(1 − x) 2 1 ∞ Hn n−1
dx = 2 x dx
0 x 0 n=1 n+1
∞ 1
Hn
=2 x n−1 dx
n+1 0
n=1
∞
Hn
=2
n(n + 1)
n=1
∞
Hn Hn
=2 −
n n+1
n=1
∞ ∞
Hn Hn+1 1
=2 − +2
n n+1 (n + 1)2
n=1 n=1
= 2 + 2ζ (2) − 2
= 2ζ (2).
3.57. We have
∞

1 ln(1 + x) 1
n−1 x
n−1
dx = (−1) dx
0 x 0 n
n=1
∞ 1
(−1)n−1
= x n−1 dx
n 0
n=1
∞
1
= (−1)n−1
n2
n=1
π2
= .
12
We calculate the second integral by parts and then we use the value of the first
integral. So, we have
1 ln(1 + x) 2
/
ln2 (1 + x) //1 1 ln(1 + x)
dx = − / +2 dx
0 x x 0 0 x(1 + x)
1 ln(1 + x) ln(1 + x)

= − ln2 2 + 2 − dx
0 x 1+x
1 ln(1 + x)
= −2 ln2 2 + 2 dx
0 x
π2
= − 2 ln2 2.
6
3.58. (a) We have
1 ln(1 + x + x 2 + · · · + x n−1 ) 1 ln(1 − x n ) 1 ln(1 − x)
dx = dx − dx
0 x 0 x 0 x
x n =y ln(1 − y)
1 1 1 ln(1 − x)
= dy − dx
0 n
y 0 x
1
1 ln(1 − x)
= −1 dx
n 0 x
3.56 π 2 (n − 1)
= .
6n
2
3.59. The integral equals − π18 . We have
1 ln(1 − x + x 2 ) 1 ln(1 + x 3 ) 1 ln(1 + x)

dx = dx − dx
0 x 0 x 0 x
x 3 =y 1 1 ln(1 + y) 1 ln(1 + x)
= dy − dx
3 0 y 0 x
2 1 ln(1 + x)
=− dx
3 0 x
3.57 π2
= − .
18
368 11 Power Series
3.60. (b) We have
1 lnk x ln(1 − x + x 2 − x 3 + · · · + x n−1 )

dx
0 x
1 lnk x (ln(1 + x n ) − ln(1 + x))
= dx
0 x
lnk x ln(1 + x n )
1 1 lnk x ln(1 + x)
= dx − dx
0 x 0 x
1 k
x n =t 1 ln x ln(1 + x)
= − 1 dx
nk+1 0 x
1 ∞
1 n−1
n−1 x
= −1 k
ln x (−1) dx
nk+1 0 n
n=1
∞ 1
1 1
= −1 (−1)n−1 x n−1 lnk x dx
nk+1 n 0
n=1
∞
(−1)n−1
1
= (−1)k k! −1
nk+1 nk+2
n=1

1 1
= (−1)k k! 1 − k+1 − 1 ζ (k + 2).
2 nk+1
Part (a) is solved similarly.

3.61. (a) We have

1 ln(1 − x + x 2 ) 1 1 1
dx = ln(1 − x + x 2 ) + dx
0 x − x2 0 1−x x
1 ln(1 − x + x 2 ) 1 ln(1 − x + x 2 )
= dx + dx
0 1−x 0 x
1 ln(1 − x + x2)
=2 dx,
0 x
and the result follows based on Problem 3.59.

(b) As A. Stadler has observed [141], part (a) of the problem can be used for
proving the surprising series with binomial coefficient giving ζ (2)
∞
1 π2
3 = . (11.3)
n2 Cn2n 6
n=1
We have
∞
π2 1 ln(1 − x + x 2 ) 1 1
=− dx = (x − x 2 )k−1 dx
9 0 x − x2 k 0
k=1
∞ 1
1
= x k−1 (1 − x)k−1 dx
k 0
k=1
∞
(k)(k)
=
k (2k)
k=1
∞
1
=2 .
k=1
k 2 Ck2k
Formula (11.3) has an interesting history: it was mentioned by R. Apéry in his

notorious lecture at Marseille, where he announced the irrationality of ζ (3) (see
[118]). It seems that the formula is not due to R. Apéry, since it was recorded in
[13], together with other series of this form, at the end of problem 36 on page 89.
11.7 Generating Functions
3.62. (a) We have

∞
∞
∞
−x nx
n x 2n+1
e sin x = (−1) (−1)n
= cn x n , x ∈ R,
n! (2n + 1)!
n=0 n=0 n=0

n
where cn = ak bn−k , an = (−1)n n!
1
and b2n = 0, b2n+1 = (−1)n (2n+1)!
1
.
k=0
c0 = a0 b0 = 0;
c1 = a0 b1 + a1 b0 = 1 · 1 + −1
1! · 0 = 1;
c2 = a0 b2 + a1 b1 + a2 b0 = 1 · 0 + −1
1! · 1! + 2! · 0 = −1;
1 1
c3 = a0 b3 + a1 b2 + a2 b1 + a3 b0 = 1 · −1 −1
3! + 1! · 0 + 2! · 1 +
1 −1
3! · 0 = 13 .
It follows that
x3
e−x sin x = x − x 2 + + ··· , x ∈ R.
3
370 11 Power Series
(b) We have
∞
∞
∞
(3x)n x 2n
e3x
cos x = (−1) n
= cn x n , x ∈ R,
n! (2n)!
n=0 n=0 n=0

n
3n
where cn = ak bn−k , an = n! and b2n = (−1)n (2n)!
1
, b2n+1 = 0.
k=0
c0 = a0 b0 = 1 · 1 = 1;
c1 = a0 b1 + a1 b0 = 1 · 0 + 3
1! · 1 = 3;
−1 32
c2 = a0 b2 + a1 b1 + a2 b0 = 1 · 2! + 3
1! ·0+ 2! · 1 = 4.
It follows that
e3x cos x = 1 + 3x + 4x 2 + · · · , x ∈ R.
(c) We have

∞
x 2n+1
(−1)n (2n+1)! ∞ π π
sin x n=0
tan x = = = cn x n , x∈ − , .
cos x ∞
x 2n 2 2
(−1)n (2n)! n=0
n=0
It follows that
∞
∞
∞
x 2n x 2n+1
(−1) n
· cn x n
= (−1)n .
(2n)! (2n + 1)!
n=0 n=0 n=0
We calculate the coefficients cn by solving the system of equations
0 = 1 · c0 ⇒ c0 = 0;
1 = 1 · c1 + 0 · c0 ⇒ c1 = 1;
1
0 = 1 · c2 + 0 · c1 − · c0 ⇒ c2 = 0;
2!
1 1 1
− = 1 · c3 + 0 · c2 − · c1 + 0 · c0 ⇒ c3 = ;
3! 2! 3
1 1
0 = 1 · c4 + 0 · c3 − · c2 + 0 · c1 + · c0 ⇒ c4 = 0;
2! 4!
1 1 1 2
= 1 · c5 + 0 · c4 − · c3 + 0 · c2 + · c1 + 0 · c0 ⇒ c5 = ,
5! 2! 4! 15
and we have that
1 2 π π
tan x = x + x 3 + x 5 + · · · , x∈ − , .
3 15 2 2
(d) We have

∞
22n+1 2n+1
(2n+1)! x ∞
sinh (2x) n=0
tanh (2x) = = = cn x n , x ∈ R.
cosh (2x) ∞
22n 2n
(2n)! x
n=0
n=0
It follows that
∞
∞
∞
22n 2n 22n+1 2n+1
x · cn x n
= x .
(2n)! (2n + 1)!
n=0 n=0 n=0
We calculate the coefficients cn by solving the system of equations
0 = 1 · c0 ⇒ c0 = 0;
2
= 1 · c1 + 0 · c0 ⇒ c1 = 2;
1!
22
0 = 1 · c2 + 0 · c1 + · c0 ⇒ c2 = 0;
2!
23 22 8
= 1 · c3 + 0 · c2 + · c1 + 0 · c0 ⇒ c3 = − ;
3! 2! 3
22 24
0 = 1 · c4 + 0 · c3 + · c2 + 0 · c1 + · c0 ⇒ c4 = 0;
2! 4!
25 22 24 64
= 1 · c5 + 0 · c4 + · c3 + 0 · c2 + · c1 + 0 · c0 ⇒ c5 = ,
5! 2! 4! 15
and we have that
8 64
tanh (2x) = 2x − x 3 + x 5 + · · · , x ∈ R.
3 15
372 11 Power Series
3.63. (a) Use the Cauchy product to multiply the power series of ln(1 − x) and 1−x
1
.
(d) One way to prove part (d) is by direct computation. Another method is based
on part (c) and the Cauchy product for calculating the power series of ln2 (1 − x).
3.64. (a) We have
/1
1 1
2 ln(1 − x) 2 ln x
dx = ln(1 − x) ln x /02 + dx
0 x 0 1−x
1 ln x 1 ln x
= ln2 2 + dx − dx
0 1−x 1
2
1−x
π2 1 ln x
= ln2 2 − − dx
6 1
2
1−x
1
1−x=t π2 2 ln(1 − t)
= ln2 2 − − dt,
6 0 t

(b) We have
∞ ∞ 1 1 ∞ 1
Hn 2 2 1 ln(1 − x)
3.63 (a) 2
=2 Hn x n−1 dx=2 Hn x n dx dx= −2
n=1
n2n−1
n=1 0 0 x
n=1 0 x(1 − x)
1 1

2 ln(1 − x) 2 ln(1 − x) (a) ln2 2 π 2
=−2 dx−2 dx = −2 − + ln2 2
0 x 0 1−x 2 12
π2
= .
6
On the other hand,
1 1 ∞ ∞ 1 ∞
2 ln(1 − x) 2 2 Hn
ln2 2=−2 dx=2 Hn x n dx=2 Hn x n dx= .
0 1−x 0 0 (n + 1)2n
n=1 n=1 n=1
3.66. Part (a) follows by direct computation. Part (b) can be solved either by using
the Cauchy product to multiply the power series of the functions f (x) = ln(1 + x)
and g(x) = 1−x 1
or by using part (a) of the problem.
3.67. We have
∞ ∞ ∞
ex xn
= · x =
n
cn x n , x ∈ (−1, 1).
1−x n!
n=0 n=0 n=0

n
We calculate cn = ak bn−k , with an = 1
n! and bn = 1, and we have that cn =
k=0

n
1
k! . Therefore,
k=0
∞
ex 1 1 1
= 1 + + + ··· + xn, x ∈ (−1, 1).
1−x 1! 2! n!
n=0
x
Remark. The function f (x) = 1−x e
is known as the generating function of the
sequence (En )n≥0 defined by En = 1 + 1!1 + 2!1 + · · · + n!
1
.
3.68. We have
∞

Bn n
x = (e − 1)
x
x ,
n!
n=0
which means
∞
∞

xn Bn n
x= x
n! n!
n=1 n=0
and, therefore,
1 = B0 ,
1 B1 1 1
0= · + B0 ⇒ B1 = − ,
1! 1! 2! 2
1 B2 1 B1 1 1
0= · + · + B0 ⇒ B2 = ,
1! 2! 2! 1! 3! 6
1 B3 1 B2 1 B1 1
0= · + · + · + B0 ⇒ B3 = 0,
1! 3! 2! 2! 3! 1! 4!
1
... ... B4 = − ,
30
... ... B5 = 0,
1
... ... B6 = .
42
3.69. Solution I. We have
∞
1 = (1 − x − x 2 ) an x n−1
n=1
374 11 Power Series
and it follows that
1 = a1 ,
0 = a2 − a1 ⇒ a2 = a1 ,
0 = a3 − a2 − a1 ⇒ a3 = a2 + a1 ,
0 = a4 − a3 − a2 ⇒ a4 = a3 + a2 ,
..
.
0 = an+1 − an − an−1 ⇒ an+1 = an + an−1 .
√ √
Solution II. For x ∈ 1− 5 5−1
2 , 2 we have

1 1 1 1
= √ √ − √
1 − x − x2 5 x+ 1+ 5
x+ 1− 5
2 2
⎛ ⎞
1 ⎝ 2 1 2 1 ⎠
= √ √ · − √ ·
5 1 + 5 1 + 2x√ 1 − 5 1 + 2x√
1+ 5 1− 5
∞ n
1 2 2x
= √ √ · (−1)n √
5 1 + 5 n=0 1+ 5
∞ n
2 2x
− √ · (−1) n
√
1 − 5 n=0 1− 5
∞
2n 1 1
= (−1) √ n−1
√ − √ x n−1
n=1
5 (1 + 5) n (1 − 5) n
∞
√ n √ n
1 1+ 5 1− 5
= √ − x n−1 .
5 2 2
n=1
3.70. The problem is solved similarly to Problem 3.69.

−in
3.71. (a) Use that sin n = e −e
in
2i .
(b) Divide by x the series in part (a) and integrate term by term.
−in
3.72. (a) Use that cos n = e +e
in
2 .
(b) Divide by x the series in part (a) and integrate term by term.
3.73. Observe that an = n + 1, ∀n ≥ 0. It follows that m = (n2 + 5n + 5)2 .
3.74. Eliminate b12n = (2n)!.
11.8 Series with Harmonic and Skew-Harmonic Numbers
3.75. Part (a) of Problem 3.66 implies that
1 (−x)n
Hn− − ln 2 = − dx.
0 1+x
(a) We have
∞ ∞
∞

− 1 (−x)n 1 1
Hn− Hn − ln 2 = − Hn− dx = − Hn− (−x)n dx
0 1+x 0 1+x
n=1 n=1 n=1
3.66 (b)
1 ln(1 − x) ln 2
= − dx = .
0 (1 + x) 2 2
(b) We have
∞
− 1 1
Hn − ln 2 ζ (2) − 1 − 2 − · · · − 2
2 n
n=1
∞ 1
(−x)n 1 1
=− dx ζ (2) − 1 − 2 − · · · − 2
1+x 2 n
n=1 0
1 ∞
1 1 1
=− ζ (2) − 1 − − ··· − 2 (−x)n dx
0 1+x 22 n
n=1
3.4 C
1 xζ (2) + Li2 (−x)
= dx.
0 (1 + x)2
Integrating by parts, we have that

/
1 xζ (2) + Li2 (−x) xζ (2) + Li2 (−x) //1 1 ζ (2) − ln(1+x)
dx = − / +
x
dx
0 (1 + x)2 1+x 0 0 1+x
ζ (2) + Li2 (−1) 1 ln(1 + x) ln2 2
=− + ζ (2) ln 2 − dx +
2 0 x 2
2
π2 π 2 ln 2 ln 2
=− + + .
8 6 2
376 11 Power Series
(c) We have, based on part (a) of Problem 3.66, that

∞
1 1
(−1) n
Hn− ζ (2) − 1 − 2 − · · · − 2
2 n
n=1
∞
1 1 − (−x)n 1 1
= (−1)n dx ζ (2) − 1 − 2 − · · · − 2
0 1+x 2 n
n=1
1 ∞
1 1 1
= (−1) ζ (2) − 1 − 2 − · · · − 2 dx
n
0 1+x 2 n
n=1
1 ∞
1 1 1
− x n ζ (2) − 1 − 2 − · · · − 2 dx
0 1+x 2 n
n=1
3.4 C −ζ (2) − Li2 (−1) xζ (2) − Li2 (x) 1

= ln 2 − dx
2 0 (1 + x)(1 − x)

−ζ (2) − Li2 (−1) 1 1 1 1
= ln 2 − (xζ (2) − Li2 (x)) + dx.
2 2 0 1+x 1−x
We calculate the preceding integral using integration by parts, with f (x) =

xζ (2) − Li2 (x), f (x) = ζ (2) + ln(1−x)
x , g (x) = 1+x 1
+ 1−x1
and g(x) =
ln(1 + x) − ln(1 − x), and we obtain that
1
1 1
(xζ (2) − Li2 (x)) + dx
0 1+x 1−x
/1 1
= (xζ (2)−Li2 (x)) (ln(1 + x) − ln(1 − x)) /0 − ζ (2) (ln(1 + x) − ln(1 − x)) dx
0
1 ln(1 − x) ln(1 + x) 1 ln2 (1 − x)
− dx + dx
0 x 0 x
21
= −2ζ (2) ln 2 + ζ (3).
8
It follows that
∞
1 1
(−1) n
Hn− ζ (2) − 1 − 2 − · · · − 2
2 n
n=1
−ζ (2) − Li2 (−1) 21 π 2 ln 2 21

= ln 2 + ζ (2) ln 2 − ζ (3) = − ζ (3).
2 16 8 16
(d) We have, based on part (a) of Problem 3.66, that

∞
1 1
Hn− ζ (3) − 1 − 3 − · · · − 3
2 n
n=1
∞
1 1 − (−x)n 1 1
= dx ζ (3) − 1 − 3 − · · · − 3
1+x 2 n
n=1 0
1 ∞
1 1 1
= ζ (3) − 1 − 3 − · · · − 3 dx
0 1+x 2 n
n=1
1 ∞
1 1 1
− (−x)n ζ (3) − 1 − 3 − · · · − 3 dx
0 1+x 2 n
n=1

3.4 C
1 1 −xζ (3) − Li3 (−x)
= ζ (2) − ζ (3) − dx
0 1+x 1+x
1 ζ (2) − ζ (3) + xζ (2) + Li3 (−x)
= dx.
0 (1 + x)2
We calculate the preceding integral by parts with f (x) = ζ (2) − ζ (3) + xζ (2) +
Li3 (−x), f (x) = ζ (2) + Li2 (−x)
x , g (x) = (1+x)
1
2 and g(x) = − 1+x , and we have
1
that
ζ (2) − ζ (3) + xζ (2) + Li3 (−x)

1
dx
0 (1 + x)2
/ 1
ζ (2) − ζ (3) + xζ (2) + Li3 (−x) //1 ζ (2) Li2 (−x)
=− / + + dx
1+x 0 0 1+x x(1 + x)
ζ (3) 1 Li2 (−x) 1 Li2 (−x)
=− + ζ (2) ln 2 + dx − dx.
8 0 x 0 1+x

1 Li2 (−x) 1 ∞ x n−1
∞
(−1)n 3
dx = (−1)n dx = = Li3 (−1) = − ζ (3)
0 x 0 n=1 n2 n3 4
n=1
and
1 Li2 (−x) /1 1 ln2 (1 + x)
dx = Li2 (−x) ln(1 + x)/0 + dx
0 1+x 0 x
ζ (3)
= Li2 (−1) ln 2 +
4
ζ (2) ln 2 ζ (3)
=− + .
2 4
378 11 Power Series
Putting all these together, we get that

∞
1 1 π 2 ln 2 9
Hn− ζ (3) − 1 − 3 − · · · − 3 = − ζ (3).
2 n 4 8
n=1

∞ ∞
Hn− (−1)n 1 1 − (−x)n
(−1)n = dx
n n 0 1+x
n=1 n=1
1 ∞
1 (−1)n xn
= − dx
0 1+x n n
n=1
1 − ln 2 + ln(1 − x)
= dx
0 1+x
1 ln(1 − x)
= − ln2 2 + dx
0 1+x
2
π2 ln 2
=− − ,
12 2
where the last equality follows based on part (b) of Problem 7.66.
(b) We have, based on part (a) of Problem 3.66, that
∞ ∞
Hn− 1 1 1 − (−x)n
= dx
n2 n2 0 1+x
n=1 n=1
1 ∞
1 1 (−x)n
= − dx
0 1+x n2 n2
n=1
1 ζ (2) − Li2 (−x)
= dx
0 1+x
/1 1 ln2 (1 + x)
= (ζ (2) − Li2 (−x)) ln(1 + x)/0 − dx
0 x
π2 ζ (3)
= ln 2 − ,
4 4
1 ln2 (1+x)
since Li2 (−1) = − ζ (2)
2 and 0 x dx = ζ (3)
4 (see [6, pp. 291–292]).
(c) We have, based on part (a) of Problem 3.66, that
∞ ∞ ∞
Hn− 1 1 1 − (−x)n 1 1 1 (−x)n
= dx = − dx
n3 n3 0 1+x 0 1+x n3 n3
n=1 n=1 n=1
1 ζ (3) − Li3 (−x)
= dx
0 1+x
/1 1 ln(1 + x)
= (ζ (3) − Li3 (−x)) ln(1 + x)/0 + Li2 (−x)dx
0 x
/
Li22 (−x) //1
= (ζ (3) − Li3 (−1)) ln 2 − /
2 0
7 π4
= ζ (3) ln 2 − .
4 288
1
(d) We have, based on the formula 0 x n−1 ln(1 − x)dx = − Hnn , n ≥ 1, (see [26,
entry (3.17), p. 206]), that
∞ ∞
∞

Hn− Hn+1 Hn− 1 1 Hn− n
· =− x ln(1 − x)dx = −
n
ln(1 − x) x dx
n n+1 n 0 0 n
n=1 n=1 n=1

3.66 (b)
1ln(1 + t) x
= − ln(1 − x)
dt dx
0 0 t (1 − t)
/1
x ln(1 + t) /
= (1 − x) [ln(1 − x) − 1] dt //
0 t (1 − t) 0
1 ln(1 + x)
− (ln(1 − x) − 1) dx
0 x
1 ln(1 − x) ln(1 + x) 1 ln(1 + x)
=− dx + dx
0 x 0 x
5 π2
= ζ (3) + .
8 12
The last equality follows based on formula (10.2) and Problem 3.57.
(e) Using the same idea as in the solution of part (d) we have, after some
calculations, that
380 11 Power Series
1−x 2
∞
Hn− Hn+1 1
(−1)n · =− ln (1 − x) − ln(1 − x) dx
n n+1 0 x(1 + x)
n=1
ln(1 − x) 1 1 ln2 (1 − x)
= −ζ (2) − 2ζ (3) − 2 dx + 4 dx
0 1+x 0 1+x

1 1
= −ζ (2) − 2ζ (3) + 2Li2 + 4Li3
2 2
3 π2 2
= ζ (3) − ln 2 − ln2 2 + ln3 2.
2 3 3
1
3.77. (a) We have, based on the formula 0 x n−1 ln(1 − x)dx = − Hnn , n ≥ 1, (see
[26, entry (3.17), p. 206]), that
∞ ∞
Hn− Hn+1 1
(−1)n−1 = (−1)n Hn− x n ln(1 − x)dx
n+1 0
n=1 n=1
∞

1
= ln(1 − x) Hn− (−x)n dx
0 n=1
3.66 (b) ln2 (1 − x) 1

= dx
0 1+x

1
= 2 Li3 .
2
The preceding integral is calculated in the solution of part (b) of Problem 2.81.
(b) We have, based on part (a) of Problem 3.66, that
∞ ∞
Hn− Hn Hn 1 1 − (−x)n
(−1)n = (−1)n dx
n n 0 1+x
n=1 n=1
∞ ∞

1 1 n Hn Hn n
= (−1) − x dx
0 1+x n n
n=1 n=1
1
(11.9) 1 1 2
= S − Li2 (x) − ln (1 − x) dx,
0 1+x 2
(11.9)
∞
ln2 2 π2
where S = (−1)n Hnn = Li2 (−1) + 1
2 ln2 2 = 2 − 12 .
n=1
We calculate the integral by parts, with f (x) = S − Li2 (x) − 12 ln2 (1 − x),
f (x) = ln(1−x)
x + ln(1−x)
1−x , g (x) = 1+x and g(x) = ln(1 + x) − ln 2, and we have
1
that
1
1 1
S − Li2 (x) − ln2 (1 − x) dx
0 1+x 2

1 2 /1
= S − Li2 (x) − ln (1 − x) (ln(1 + x) − ln 2) /0
2

1 ln(1 − x) ln(1 − x)
− (ln(1 + x) − ln 2) + dx
0 x 1−x
1 ln(1 − x) ln(1 + x) 1 ln(1 − x)
= S ln 2 − dx + ln 2 dx
0 x 0 x
1 ln(1 − x)
− (ln(1 + x) − ln 2) dx
0 1−x
5 1 ln(1 − x)
= S ln 2 + ζ (3) − ζ (2) ln 2 − (ln(1 + x) − ln 2) dx.
8 0 1−x
We calculate
1 ln(1 − x) ln y
1 y
(ln(1 + x) − ln 2) dx = ln 1 − dy
0 1−x 0 y 2
∞
1 y n−1
=− ln y dy
0 n2n
n=1
∞ 1
1
=− y n−1 ln y dy
n2n 0
n=1
∞
1
=
n3 2n
n=1

1
= Li3 .
2
It follows that
∞ −
n Hn Hn 5 1
(−1) = S ln 2 + ζ (3) − ζ (2) ln 2 − Li3
n 8 2
n=1
ln3 2 π 2 ζ (3)
= − ln 2 − ,
3 6 4

since Li3 1
2 = 1
24 −2π 2 ln 2 + 4 ln3 2 + 21ζ (3) (see [21, p. 44]).
382 11 Power Series
1 n
3.78. (a) and (b) Use that Hn− − ln 2 = − 0 (−x) 1+x dx, n ≥ 1 (part (a) of
Problem 3.66).
(c) and (d) Add and subtract the series in parts (a) and (b).
1
3.79. We have, based on the formula 0 x n−1 ln(1 − x)dx = − Hnn , n ≥ 1, (see [26,
entry (3.17), p. 206]), that
∞ ∞ 1
H2n−1 1
=− x 2n−2 ln(1 − x)dx
(2n − 1)2 2n − 1 0
n=1 n=1
∞

1 ln(1 − x) x 2n−1
=− dx
0 x 2n − 1
n=1

1 1 ln(1 − x) 1+x
=− ln dx
2 0 x 1−x
1 1 ln(1 − x) ln(1 + x) 1 1 ln2 (1 − x)
=− dx + dx
2 0 x 2 0 x
(10.2) 21
= ζ (3).
16
The second part of the problem follows from the formula
∞ ∞ ∞
7.60 Hn H2n−1 1 H2n
2ζ (3) = = + .
n2 (2n − 1)2 4 n2
n=1 n=1 n=1
3.80. (a) We have

∞

1 1 i
i−1 x
x n−1
ln(1 + x)dx = x n−1
(−1) dx
0 0 i
i=1
∞
(−1)i−1
=
i(i + n)
i=1
∞ ∞

1 (−1)i−1 (−1)i−1
= −
n i n+i
i=1 i=1
⎛ ⎞
∞
1 (−1)j −1 ⎠
= ⎝ln 2 − (−1)n
n j
j =n+1
1 − (−1)n (−1)n −
= ln 2 + Hn .
n n

− 1
H2n−1 2 ln 2
= − x 2n−2 ln(1 + x)dx
2n − 1 2n − 1 0
and it follows that

∞ − ∞ 1
H2n−1 2 ln 2
(−1) n−1
= (−1) n−1
− x 2n−2
ln(1 + x)dx
2n − 1 2n − 1 0
n=1 n=1
∞
∞

(−1)n−1 1
= 2 ln 2 − ln(1 + x) 2 n−1
(−x ) dx
2n − 1 0
n=1 n=1
π ln 2 1 ln(1 + x)
= − dx
4 0 1 + x2
3π ln 2
= .
8
1
The integral 0 ln(1+x)
1+x 2
dx = π ln 2
8 was calculated in the solution of part (b) of
Problem 3.25.
(c) Observe that
Hn− H− H− (−1)n
= n − n+1 + , n ≥ 1. (11.4)
n(n + 1) n n + 1 (n + 1)2
It follows that
∞ ∞
∞ ∞
Hn− Hn− H− (−1)n (−1)n π2
= − n+1 + =1+ = .
n(n + 1) n n+1 (n + 1)2 (n + 1)2 12
n=1 n=1 n=1 n=1
The second series can be calculated based on formula (11.4) and part (a) of
Problem 3.76.
−
Hn+1 1 1 − (−1)n+1
= (−1)n+1 x n ln(1 − x)dx + ln 2.
n+1 0 n+1
It follows that
384 11 Power Series
∞ − ∞
Hn− Hn+1 Hn− 1 1 − (−1)n+1
· = (−1)n+1 x n ln(1 − x)dx + ln 2
n n+1 n 0 n+1
n=1 n=1
∞
1 Hn−
=− ln(1 + x) n
(−x) dx
0 n
n=1
∞ ∞

Hn− Hn−
+ ln 2 − (−1) n−1
n(n + 1) n(n + 1)
n=1 n=1
−x 2
3.66 (b)
1 ln(1 + t) π
= − ln(1 + x) dt dx + ln 2 − ln2 2 .
0 0 t (1 − t) 12
We calculate the preceding integral by parts, and we have

−x
1 ln(1 + t)
ln(1 + x) dt dx
0 0 t (1 − t)
/
−x ln(1 + t) //1 1 ln(1 − x)
= (1 + x) (ln(1 + x) − 1) dt − (ln(1 + x) − 1)dx
0 t (1 − t) /0 0 x
−1 ln(1 + t) 1 ln(1 − x) ln(1 + x) 1 ln(1 − x)
= 2(ln 2 − 1) dt − dx + dx
0 t (1 − t) 0 x 0 x
t=−y
1 ln(1 − y) 5 π2
= 2(ln 2 − 1) dy + ζ (3) −
0 y(1 + y) 8 6

ln(1 − y) 1 ln(1 − y)
1 5 π2
= 2(ln 2 − 1) dy − dy + ζ (3) −
0 y 0 1+y 8 6
2 2
7.66 π 1 5 π
= 2(ln 2 − 1) − + Li2 + ζ (3) −
6 2 8 6
π 2 ln 2 5
=− − ln3 2 + ln2 2 + ζ (3).
6 8
Putting all these together, we have that part (d) of the problem is solved.
H− 1
(e) We have, based on part (a), that nn = (−1)n 0 x n−1 ln(1 + x)dx +
1−(−1)n
n ln 2, and it follows that
∞ ∞
Hn Hn− Hn 1 1 − (−1)n
= (−1)n x n−1 ln(1 + x)dx + ln 2
n2 n 0 n
n=1 n=1
∞ ∞
1 ln(1 + x) Hn Hn
= (−x) dx + ln 2
n
(1 − (−1)n )
0 x n n2
n=1 n=1
∞
1 ln(1 + x) 1 Hn
= Li2 (−x) + ln2 (1 + x) dx + ln 2 (1 − (−1)n ).
0 x 2 n2
n=1
On the other hand,

∞ ∞
Hn H2n−1 3.79 21
(1 − (−1)n ) = 2 = ζ (3).
n2 (2n − 1)2 8
n=1 n=1
It follows that
∞
Hn Hn− 1 ln(1 + x) 1 1 ln3 (1 + x) 21
= Li2 (−x)dx + dx + ζ (3) ln 2
n2 0 x 2 0 x 8
n=1
/
Li22 (−x) //1 1 ln3 (1 + x)
1 21
=− / +2 dx + ζ (3) ln 2
2 0 0 x 8

43π 4 π 2 ln2 2 ln4 2 1
= + − − 3Li4 .
1440 8 8 2
We used a result of V.I. Levin involving the tetralogarithm Li4 ([21, p. 47], [104])
1 ln3 (1 + x)
π 4 π 2 ln2 2 ln4 2 21 ln 2 1
dx = + − − ζ (3) − 6 Li4 . (11.5)
0 x 15 4 4 4 2
(−1)n
3.81. (a) We apply Abel’s summation formula, with an = n and bn = ζ (2) −
1 − 212 − · · · − n12 , and we have that
∞
(−1)n 1 1
ζ (2) − 1 − 2 − · · · − 2
n 2 n
n=1

1 1
= − lim Hn− ζ (2) − 1 − 2 − · · · −
n→∞ 2 (n + 1)2
∞
−1 (−1)2 (−1)n 1
+ + + ··· +
1 2 n (n + 1)2
n=1
∞
−1 (−1)2 (−1)n (−1)n+1 (−1)n+1 1
= + + ··· + + −
1 2 n n+1 n+1 (n + 1)2
n=1

∞ −
Hn+1 (−1)n+1
∞
(−1)j −1 Hj−
= − − = − 2
(n + 1)2 (n + 1)3 j3 j
n=1 j =2

∞
(−1)j −1 Hj−
= −
j3 j2
j =1
∞
(−1)j −1
∞ Hj−
= −
j3 j2
j =1 j =1
3.76 (b) π2
= ζ (3) − ln 2.
4
386 11 Power Series
n
(b) Use Abel’s summation formula, with an = (−1) n and bn = ζ (3) − 1 − 23 −
1
· · · − n13 , combined with part (c) of Problem 3.76.

H− 1 n
(c) Since nn = (−1)n 0 x n−1 ln(1 + x)dx + 1−(−1) n ln 2 (see part (a) of
Problem 3.80), we have that
∞
Hn− 1 1
ζ (2)−1− 2
−···− 2
n 2 n
n=1
∞ 1
1−(−1)n 1 1
= (−1)n x n−1
ln(1+x)dx+ ln 2 ζ (2)−1− 2 − · · · − 2
0 n 2 n
n=1
∞
1 ln(1+x) 1 1
= ζ (2)−1− 2 − · · · − 2 (−x)n dx
0 x 2 n
n=1
∞
1−(−1)n 1 1
+ ln 2 ζ (2)−1− 2 − · · · − 2
n 2 n
n=1
3.4 C ln(1+x) −xζ (2)−Li2 (−x)

1
= · dx
0 x 1+x
∞
1−(−1)n 1 1
+ ln 2 ζ (2)−1− 2 − · · · − 2 .
n 2 n
n=1
(11.6)
Using that
∞
1 1 1
ζ (2) − 1 − 2 − · · · − 2 = ζ (3)
n 2 n
n=1
(see [26, problem 3.20, p. 142]) and part (a) of the problem, we have
∞
1 − (−1)n 1 1 π2
ζ (2) − 1 − 2 − · · · − 2 = ln 2. (11.7)
n 2 n 4
n=1
We calculate
1 ln(1 + x) −xζ (2) − Li2 (−x)

I= · dx
0 x 1+x
ln2 2ln(1 + x)
1 1 ln(1 + x)
= −ζ (2) − Li2 (−x)dx + Li2 (−x)dx
02 x 0 1+x
/
π 2 ln2 2 Li22 (−x) //1 1 ln(1 + x)
=− + / + Li2 (−x)dx
12 2 0 0 1+x
π 2 ln2 2 π4 1 ln(1 + x)
=− + + Li2 (−x)dx.
12 288 0 1+x
Integrating by parts, with f (x) = Li2 (−x), f (x) = − ln(1+x)

x , g (x) = ln(1+x)
1+x
ln2 (1+x)
and g(x) = 2 , we have that
/
1 ln(1 + x) Li2 (−x) ln2 (1 + x) //1 1 1 ln3 (1 + x)
Li2 (−x)dx = / +2 dx
0 1+x 2 0 0 x
Li2 (−1) ln 2 12 1 ln (1 + x)
3
= + dx
2 2 0 x

π 2 ln2 2 π4 4
ln 2 21 ln 2 1
= + − − ζ (3) − 3 Li4 ,
12 30 8 8 2
where the last equality follows based on formula (11.5).

It follows that

53π 4 ln4 2 21 ln 2 1
I= − − ζ (3) − 3 Li4 . (11.8)
1440 8 8 2
Combining (11.6), (11.7), and (11.8), we have that part (c) of the problem is solved.
11.9 Remarkable Numerical and Function Series
3.82. We notice that u (x) = w(x), v (x) = u(x), and w (x) = v(x). Let
f (x) = u2 (x) + v 2 (x) + w 2 (x) − u(x)v(x) − v(x)w(x) − w(x)u(x).
Then
f (x) = 2u(x)u (x) + 2v(x)v (x) + 2w(x)w (x)

−u (x)v(x)−u(x)v (x)−v (x)w(x)−v(x)w (x)−w (x)u(x)−w(x)u (x)
= −f (x).
388 11 Power Series
It follows that f (x) = C e−x and, since f (0) = 1, we obtain that f (x) = e−x .
To solve the second part of the problem we observe that
u(x) + v(x) + w(x) = ex
and we use the identity
x 3 + y 3 + z3 − 3xyz = (x + y + z)(x 2 + y 2 + z2 − xy − yz − zx).
We have
(u3 (x) + v 3 (x) + w 3 (x) − 3u(x)v(x)w(x))

= 3u2 (x)u (x) + 3v 2 (x)v (x) + 3w 2 (x)w (x)
− 3(u (x)v(x)w(x) + u(x)v (x)w(x) + u(x)v(x)w (x))
= 3u2 (x)w(x) + 3v 2 (x)u(x) + 3w 2 (x)v(x)
− 3(v(x)w2 (x) + u2 (x)w(x) + u(x)v 2 (x))
= 0,
and it follows that u3 (x) + v 3 (x) + w 3 (x) − 3u(x)v(x)w(x) = C . When x = 0 we

have u(0) = 1, v(0) = 0, w(0) = 0, and we obtain that
u3 (x) + v 3 (x) + w 3 (x) − 3u(x)v(x)w(x) = 1.
3.83. Part (a) can be solved by direct computation.

∞ ∞
1 1 x2 xn 1 tn x
ln −x− − ··· − = dt
n 1−x 2 n n 0 1−t
n=1 n=1
∞
x 1 tn
= dt
0 1−t n
n=1
x ln(1 − t)
=− dt
0 1−t
ln (1 − x)
2
= .
2
3.84. (a) ln(1 − x) + x

1−x ; (b) x
2(1+x 2 )
− arctan x
2 .
3.85. (a) Use mathematical induction.

(b) Solution I. Apply Abel’s summation formula with an = Hn and bn =
2 n
1
ln 1−x − x − x2 − · · · − xn .
Solution II. We have, based on part (a) of Problem 3.83, that
∞ ∞
1 x2 xn x
tn
Hn ln −x− − ··· − = Hn dt
1−x 2 n 0 1−t
n=1 n=1
∞
x 1
= Hn t n dt
0 1−t
n=1
3.63 (a)
x ln(1 − t)
= − dt
0 (1 − t)2
ln(1 − x) x
=− − .
1−x 1−x
(c) Divide by x = 0 the series in part (b) and integrate from 0 to x. Another
method is based on the application of Abel’s summation formula with an = Hn and
2 n
bn = Li2 (x) − x − x22 − · · · − xn2 . Parts (d) and (e) are solved similarly.
∞
1 x2 xn 1 1
ln −x− − ··· − ζ (2) − 1 − 2 − · · · − 2
1−x 2 n 2 n
n=1
∞ x
tn 1 1
= dt ζ (2) − 1 − 2 − · · · − 2
1−t 2 n
n=1 0
x ∞
1 1 1
= ζ (2) − 1 − 2 − · · · − 2 t n dt
0 1−t 2 n
n=1
3.4 C tζ (2) − Li2 (t)

x
= dt
0 (1 − t)2
/ x ζ (2)
tζ (2) − Li2 (t) //x ln(1 − t) ln(1 − t)
= / − + + dt
1−t 0 0 1−t t 1−t
x ln2 (1 − x)
= (ζ (2) − Li2 (x)) + ζ (2) ln(1 − x) + .
1−x 2
(b) Let
∞
x2 xn 1 1
f (x) = Li2 (x) − x − 2 − · · · − 2 ζ (2) − 1 − 2 − · · · − 2 .
2 n 2 n
n=1
We have, based on part (a), that

390 11 Power Series
∞
1 1 x2 xn 1 1
f (x) = ln −x− − ··· − ζ (2) − 1 − 2 − · · · − 2
x 1−x 2 n 2 n
n=1
ζ (2) − Li2 (x) ln(1 − x) ln2 (1 − x)

= + ζ (2) + .
1−x x 2x
It follows that
x ζ (2) − Li2 (t) x ln(1 − t) 1 x ln2 (1 − t)
f (x) = dt + ζ (2) dt + dt
0 1−t 0 t 2 0 t
/x x ln (1 − t)
2
= −(ζ (2) − Li2 (t)) ln(1 − t)/0 + dt
0 t
1 x ln (1 − t)
2
− ζ (2)Li2 (x) + dt
2 0 t
3 x ln2 (1 − t)
= − ln(1 − x)(ζ (2) − Li2 (x)) − ζ (2)Li2 (x) + dt.
2 0 t
3.88. See [4].

3.89. Let x ∈ (−1, 1). We apply Abel’s summation formula with an = x n and
n+1
bn = f (1) − a1 − a2 − · · · − an and we have, since An = x−x
1−x , that
∞
x−x n+1
(f (1)−a1 − a2 − · · · − an )x n = lim (f (1)−a1 −a2 − · · · − an+1 )
n→∞ 1−x
n=1
∞
x
+ (1 − x n )an+1
1−x
n=1
∞ ∞
x 1
= an+1 − an+1 x n+1
1−x 1−x
n=1 n=1
x f (x) − a1 x
= (f (1) − a1 ) −
1−x 1−x
f (1)x − f (x)
= .
1−x

∞
Now we consider the case x = 1. Since the series (f (1) − a1 − a2 − · · · − an )
n=1
converges, we have, based on Abel’s theorem for power series, that
∞ ∞
(f (1) − a1 − a2 − · · · − an ) = lim (f (1) − a1 − a2 − · · · − an )x n
x→1−
n=1 n=1
f (1)x − f (x)
= lim
x→1− 1−x
= f (1) − f (1).

∞
We mention that, when x = 1, the power series (f (1) − a1 − a2 − · · · − an )x n
n=1
can be calculated by shifting the index of summation. We have
∞
S= (f (1) − a1 − a2 − · · · − an )x n
n=1
∞
= (f (1) − a1 )x + (f (1) − a1 − a2 − · · · − an )x n
n=2
∞
n−1=i
= (f (1) − a1 )x + (f (1) − a1 − a2 − · · · − ai+1 )x i+1
i=1
∞ ∞
= (f (1) − a1 )x + x (f (1) − a1 − a2 − · · · − ai )x i − ai+1 x i+1
i=1 i=1
= f (1)x + xS − f (x).
3.90. (a) Let k ∈ N. We have
k k k
(−1)i−1 1 1 1 1−(−x)k
= (−1)i−1 x n+i−1 dx= xn (−x)i−1 dx= xn dx.
n+i 0 0 0 1+x
i=1 i=1 i=1

∞
(−1)i−1
k
(−1)i−1 1 xn
It follows that n+i = lim n+i = 0 1+x dx, since
i=1 k→∞ i=1
/ /
/ 1 x n+k / 1 1
0 <// (−1)k
dx /<
/ x n+k dx = .
0 1+x 0 n+k+1
Therefore,
1 2 2 2 1 1 2x n 1 x n−1 (1 − x)
− + − + ··· = x n−1 dx − dx = dx.
n n+1 n+2 n+3 0 0 1+x 0 1+x
392 11 Power Series
(b) We calculate the radius of convergence of the power series by the formula
R = lim sup1 √
n |a | . We have
n

1 n
1 x n−1 − x n n
1 x n−1 (1 − x) n
1
√ = dx < dx < (x n−1 − x n )dx
n
2n(n + 1) 0 2 0 1+x 0
1
= √
n
n(n + 1)
6
1 n−1 (1−x)
and this implies that lim sup n 0 x 1+x dx = 1. Therefore, R = 1.
Let x ∈ R, |x| < 1, and let N ∈ N. We have
N N
1 2 2 2 1 1−y
xn − + − + ··· = xn y n−1 dy
n n+1 n+2 n+3 0 1+y
n=1 n=1
1 (1 − y)x 1 − (xy)N 1 (1 − y)x 1 (1 − y)x(xy)N
= · dy = dy − dy.
0 1+y 1 − xy 0 (1 + y)(1 − xy) 0 (1 + y)(1 − xy)
Passing to the limit, when N → ∞, we get that

∞
1 2 2 2 1 (1 − y)x
xn − + − + ··· = dy,
n n+1 n+2 n+3 0 (1 + y)(1 − xy)
n=1
since
/ /
/ 1 (1 − y)x(xy)N / 1 1 1 1
0 <// dy //≤ y N dy = · → 0 (N → ∞).
0 (1 + y)(1 − xy) 1 − |x| 0 N + 1 1 − |x|
On the other hand,

1 (1 − y)x x 2 1 1 x

dy = − + dy
0 (1 + y)(1 − xy) 1+x 0 1+y 1 − xy 1 − xy

x ln(1 − x)
= 2 ln 2 + − ln(1 − x)
1+x x
2x ln 2 + (1 − x) ln(1 − x)
= .
1+x
Case x = −1. Let an = n1 − n+1 2

+ n+2
2
− n+32
+ · · · . We have, based on
1 y n−2 (1−y) 1 y n−1 (1−y)
part (a), that an−1 = 0 1+y dy > 0 1+y dy = an , which implies that

∞
the series (−1)n an is a Leibniz series, hence it converges. It follows, based on
n=1
Abel’s theorem for power series, that

∞ ∞
1 2 2 2
(−1) an = lim
n
x n
− + − + ···
x→−1 n n+1 n+2 n+3
n=1 n=1
2x ln 2 + (1 − x) ln(1 − x)
= lim
x→−1 1+x
= ln 2 − 1.
Case x = 1. The series formula can be proved either by using part (a) or by
observing that S2n = 1 − 12 + 13 − · · · − 2n1
, n ≥ 1.
(c) This part is solved similarly to part (d).
∞ 2
1 2 2 2
S= (−1)n − + − + ···
n n+1 n+2 n+3
n=1
∞ 1 1−x 1 1−y
= (−1)n x n−1 dx y n−1 dy
0 1+x 0 1+y
n=1
∞
1 1 (1 − x)(1 − y)
=− (−xy)n−1 dxdy
0 0 (1 + x)(1 + y)
n=1
1 1 (1 − x)(1 − y)
=− dxdy
0 0 (1 + x)(1 + y)(1 + xy)
1 1−x 1 1−y

=− dy dx
0 1+x 0 (1 + y)(1 + xy)
1 1−x 1 1−y 1
x
=− − dy dx
0 1+x 0 1−x 1+y 1 + xy
1
1 1 1−y 1 1−y
=− dy − x dy dx
0 1+x 0 1+y 0 1 + xy
1 1 1−y
1
=− 2 ln 2 − 1 − x dy dx.
0 1+x 0 1 + xy

1 1−y 1+x
x dy = ln(1 + x) − 1.
0 1 + xy x
394 11 Power Series
It follows that

1 1 1+x
S=− 2 ln 2 − ln(1 + x) dx
0 1+x x
1 ln(1 + x)
= −2 ln2 2 + dx
0 x
π2
= − 2 ln2 2.
12
3.92. We have
∞ 2
1 1 1
S= − + − ···
n2 (n + 1)2 (n + 2)2
n=1
2
1 1 1
= − 2 + 2 − ···
12 2 3
∞ 2
1 1 1
+ − + − ···
n2 (n + 1)2 (n + 2)2
n=2
∞ 2
n−1=i π4 1 1 1
= + − + − ···
144 (i + 1)2 (i + 2)2 (i + 3)2
i=1
∞ 2
π4 1 1 1 1 1
= + − − + − + ···
144 i2 i 2 (i + 1) 2 (i + 2) 2 (i + 3)2
i=1
∞
π4 π4 1 1 1 1
= + −2 − + − ··· + S,
144 90 i2 i2 (i + 1)2 (i + 2)2
i=1
and it follows that

∞
1 1 1 1 1 π4 π4 13 4
− + − ··· = + = π .
i2 i 2 (i + 1) 2 (i + 2)2 2 144 90 1440
i=1
3.93. (a) Let x ∈ (−1, 1) and let

∞
1 x2 xn 1 (−1)2 (−1)n
f (x)= ln −x− − · · · − ln −(−1)− −···− .
1−x 2 n 2 2 n
n=1
We have
∞
1 1 (−1)2 (−1)n
f (x)= −1−x− · · · −x n−1 ln −(−1)− −···−
1−x 2 2 n
n=1
∞
xn 1 (−1)2 (−1)n
= ln −(−1)− −···−
1−x 2 2 n
n=1
3.88 (c) ln(1+x)−x ln 2

= .
(1−x)2
It follows that
ln(1 + t) − t ln 2
x
f (x) = dt
0 (1 − t)2
/
ln(1 + t) − t ln 2 //x x 1 ln 2
= / − − dt
1−t 0 0 1−t 2 1−t
ln(1 + x) − x ln 2 1 1 − x
= + ln − ln 2 ln(1 − x).
1−x 2 1+x
Letting x → −1, we get that
∞ 2
1 (−1)2 (−1)n
ln − (−1) − − ··· −
2 2 n
n=1

ln(1 + x) − x ln 2 1 1 − x
= lim + ln − ln 2 ln(1 − x)
x→−1 1−x 2 1+x
= ln 2 − ln2 2.
The case when x = −1 also follows from Problem 3.94.

(b) Let x ∈ (−1, 1) and let S(x) be the function
∞
1 x2 xn 1 (−1)2 (−1)n
n
(−1) ln −x− − · · · − ln −(−1)− −···− .
1−x 2 n 2 2 n
n=1
396 11 Power Series
We have that S (x) is equal to

∞
1 1 (−1)2 (−1)n
(−1)n −1−x− · · · −x n−1 ln −(−1)− −···−
1−x 2 2 n
n=1
∞
xn 1 (−1)2 (−1)n
= (−1) n
ln −(−1)− −···−
1−x 2 2 n
n=1
3.88 (c) ln(1−x)+x ln 2

= .
1−x 2
It follows that
x ln(1 − t) + t ln 2
S(x) = dt
0 1 − t2
x
1 1 1
= (ln(1 − t) + t ln 2) + dt
2 0 1−t 1+t

1 1 x ln(1 − t)
= − ln2 (1 − x) + dt − ln 2 ln(1 − x 2 ) .
2 2 0 1+t
We calculate the preceding integral. We have
x ln(1 − t) 1+t=y
1+x ln(2 − y)
dt = dy
0 1+t 1 y

1+x ln 2 + ln 1 − y2
= dy
1 y
1+x
y=2u ln(1 − u) 2
= ln 2 ln(1 + x) + du
u 1
2

1 1+x
= ln 2 ln(1 + x) + Li2 − Li2 .
2 2
Putting all these together, we get that

ln2 (1 − x) 1 1 1 1+x ln 2 ln(1 − x)
S(x) = − + Li2 − Li2 − .
4 2 2 2 2 2
Letting x → −1, we get that
∞ 2
1 (−1)2 (−1)n
(−1)n ln − (−1) − − ··· −
2 2 n
n=1

ln2 (1 − x) 1 1 1 1+x ln 2 ln(1 − x)
= lim − + Li2 − Li2 −
x→−1 4 2 2 2 2 2
π2
= − ln2 2.
24
3.94. We have, based on part (a) of Problem 3.83, that
∞
1 x2 xn 1 y2 yn
S= ln −x− − ··· − ln −y− − ··· −
1−x 2 n 1−y 2 n
n=1
∞ x y
tn un
= dt du
1−t 1−u
n=1 0 0
x y ∞
1
= (ut)n dtdu
0 0 (1 − t)(1 − u)
n=1
x y ut
= dtdu
0 0 (1 − t)(1 − u)(1 − ut)
x y
t u
= du dt.
0 1−t 0 (1 − u)(1 − tu)

y y
u 1 u ut
du = − du
0 (1 − u)(1 − tu) 0 1−t 1 − u 1 − ut
y
1 1 1
= − du
1−t 0 1 − u 1 − ut

1 1
= − ln(1 − y) + ln(1 − ty) .
1−t t
398 11 Power Series
It follows that
x
t 1
S= − ln(1 − y) + ln(1 − ty) dt
0 (1 − t)2 t
x t x ln(1 − ty)
= − ln(1 − y) dt + dt
0 (1 − t)2 0 (1 − t)2
x ln(1 − ty)
x
= − ln(1 − y) + ln(1 − x) + dt
1−x 0 (1 − t)2
x
= − ln(1 − x) ln(1 − y) − ln(1 − y)
1−x
y 1 − xy
− ln(1 − x) + ln(1 − xy).
1−y (1 − x)(1 − y)
The last integral has been calculated by parts as follows:

/
x ln(1 − ty) ln(1 − yt) //x 1 x
dt = / +y dt
0 (1 − t)2 1−t 0 0 (1 − ty)(1 − t)
ln(1 − xy) y x 1 y

= + − dt
1−x 1−y 0 1−t 1 − ty
1 − xy y
= ln(1 − xy) − ln(1 − x).
(1 − x)(1 − y) 1−y
11.10 Multiple Series with the Riemann Zeta Function
3.95. We have
∞ ∞ ∞ ∞ 1
(−1)n+m (−1)n+m
= x n+m−1 dx
nm(n + m) nm 0
n=1 m=1 n=1 m=1
1 ∞ ∞
1 (−x)n (−x)m
= dx
0 x n m
n=1 m=1
1 ln (1 + x)
2
= dx
0 x
ζ (3)
= .
4
The last integral is calculated in [6, pp. 291–292].

3.96. (a) We have
∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞
1 1 1 1
(ζ (i+j )−1)= = = =ζ (2).
k i+j ki kj (k−1)2
i=1 j =1 i=1 j =1 k=2 k=2 i=1 j =1 k=2

3.97. (a) We have, based on symmetry reasons, that
∞ ∞ ∞ ∞
ζ (i + j ) − 1 ζ (i + j ) − 1
i = j
i+j i+j
i=1 j =1 i=1 j =1
and it follows, based on part (a) of Problem 3.96, that

∞ ∞ ∞ ∞
ζ (i + j ) − 1 1 ζ (2)
i = (ζ (i + j ) − 1) = .
i+j 2 2
i=1 j =1 i=1 j =1

3.98. (b) We have

∞ ∞ 0 1 ∞ ∞ ∞
1
22i+j (ζ (2i + j ) − 1) − 1 = 2 2i+j
−1
n2i+j
i=1 j =1 i=1 j =1 n=2
∞ ∞ ∞ 2i+j
2
=
n
i=1 j =1 n=3
∞ ∞ 2i ∞ j
2 2
=
n n
n=3 i=1 j =1
∞
1
=8
(n2 − 4)(n − 2)
n=3
π2 25
= − .
3 24
400 11 Power Series

3.99. Part (a) follows from part (b), with k = 1 and n = 2.
(b) We have
∞ ∞
··· i1 · · · ik (ζ (i1 + · · · + in ) − 1)
i1 =1 in =1
∞ ∞ ∞
1
= ··· i1 · · · ik
mi1 +···+in
i1 =1 in =1 m=2
∞ ∞ ∞ ∞ ∞
i1 ik 1 1
= ··· ···
mi1 mik mik+1 min
m=2 i1 =1 ik =1 ik+1 =1 in =1
⎡ ⎤k
∞
n−k
1 1
⎢ ⎥
= ⎣
m
2 ⎦
m
m=2 1− m1 1− 1
m
∞
mk
=
(m − 1)n+k
m=2
∞ k
1 j
= Ck (m − 1)k−j
(m − 1)n+k
m=2 j =0
k ∞
j 1
= Ck
(m − 1)j +n
j =0 m=2
k
j
= Ck ζ (j + n).
j =0
3.100. See the solution of Problem 3.99.

3.101. (b) We have

∞ ∞ 0 1
··· 2i1 +···+ik (ζ (i1 + · · · + ik ) − 1) − 1
i1 =1 ik =1
∞ ∞
∞

1
= ··· 2i1 +···+ik −1
ni1 +···+ik
i1 =1 ik =1 n=2
∞ ∞ ∞ i1 +···+ik
2
= ···
n
i1 =1 ik =1 n=3
∞ ∞ i1 ∞ ik
2 2
= ···
n n
n=3 i1 =1 ik =1
∞
k
2
= n
n=3
1− 2
n
∞
2k
=
(n − 2)k
n=3
= 2k ζ (k).
3.102. (a) We have

∞ ∞ ∞ ∞
∞

1 1
3i+j ζ (i + j ) − 1 − −1 = i+j
3 −1
2i+j ni+j
i=1 j =1 i=1 j =1 n=3
∞ ∞ ∞ i+j
3
=
n
i=1 j =1 n=4
∞ ∞ i ∞ j
3 3
=
n n
n=4 i=1 j =1
∞
2
3
= n
n=4
1− 3
n
∞
9
=
(n − 3)2
n=4
= 9ζ (2).

The challenge. The sum of the series is (n + 1)n ζ (n).
402 11 Power Series
3.103. (b) We have

∞ ∞ 0 1 ∞ ∞ ∞
1
(−1) i−1
2i+j (ζ (i + j ) − 1) − 1 = (−1) i−1 i+j
2 −1
ni+j
i=1 j =1 i=1 j =1 n=2
∞ ∞ ∞ i+j
2
= (−1) i−1
n
i=1 j =1 n=3
∞ ∞ ∞ j
2 i 2
=− −
n n
n=3 i=1 j =1
∞
4
=
n2 −4
n=3
25
= .
12
3.104. (b) We have
∞ ∞
··· (−1)i1 +···+ik (ζ (i1 + · · · + ik ) − 1)
i1 =1 ik =1
∞ ∞ ∞
1
= ··· (−1)i1 +···+ik
ni1 +···+ik
i1 =1 ik =1 n=2
∞ ∞ ∞
1 i1 1 ik
= − ··· −
n n
n=2 i1 =1 ik =1
∞
k
− n1
=
n=2
1+ 1
n
∞
(−1)k
=
(n + 1)k
n=2

1
= (−1)k ζ (k) − 1 − k .
2
11.11 Series Involving Products of Harmonic Numbers
3.105. Solution I. (a) See Problem 7.61.

Solution II. We use the formula (see [26, entry (3.17), p. 206])
1 Hn
x n−1 ln(1 − x)dx = − , n ≥ 1,
0 n
and the power series

∞
Hn n 1
x = Li2 (x) + ln2 (1 − x), x ∈ [−1, 1). (11.9)
n 2
n=1
The power series formula can be proved by dividing by x the power series of part
(a) of Problem 3.63 and then integrating the series from 0 to x.
(a) We have
∞ ∞ 1
Hn Hn+1 Hn
· =− x n ln(1 − x)dx
n n+1 n 0
n=1 n=1
∞

1 Hn n
=− ln(1 − x) x dx
0 n
n=1
1
1 2
=− ln(1 − x) Li2 (x) + ln (1 − x) dx
0 2
1
=− ln(1 − x)Li2 (x)dx + 3.
0
We calculate the preceding integral by parts, and we have that
1 /1
ln(1 − x)Li2 (x)dx = −(1 − x) [ln(1 − x) − 1] Li2 (x)/0
0
1−x 0 2 1 1
− ln (1 − x) − ln(1 − x) dx
0 x
1 1−x 0 1
=− ln2 (1 − x) − ln(1 − x) dx
0 x
1 ln2 (1 − x) 1
=− dx + ln2 (1 − x)dx
0 x 0
1 ln(1 − x) 1
+ dx − ln(1 − x)dx
0 x 0
= −2ζ (3) − ζ (2) + 3.
Putting all these together, we get that part (a) of the problem is solved.
404 11 Power Series
We mention that the series in part (a) of Problem 3.105, as well as a generaliza-
tion, are calculated by a different method in [25].
(b) We have
∞ ∞ 1
Hn Hn+1 Hn
(−1)n · =− (−1)n x n ln(1 − x)dx
n n+1 n 0
n=1 n=1
∞

1 Hn
=− ln(1 − x) n
(−x) dx
0 n
n=1
1
1 2
=− ln(1 − x) Li2 (−x) + ln (1 + x) dx
0 2
1 1 1
=− ln(1 − x)Li2 (−x)dx − ln(1 − x) ln2 (1 + x)dx.
0 2 0
(11.10)
We calculate the first integral by parts and we have
1 /1
ln(1 − x)Li2 (−x)dx = −(1 − x) [ln(1 − x) − 1] Li2 (−x)/0
0
1 1−x
− ln(1 + x) [ln(1 − x) − 1] dx
0 x
1 1−x
=− ln(1 + x) [ln(1 − x) − 1] dx
0 x
1 ln(1 − x) ln(1 + x) 1 ln(1 + x)
=− dx + dx
0 x 0 x
1 1
+ ln(1 − x) ln(1 + x)dx − ln(1 + x)dx
0 0
5 π2
= ζ (3) − + ln2 2 − 4 ln 2 + 3.
8 12
(11.11)
We calculate the second integral as follows:

1 1 1 1−x 1 1
ln(1 − x) ln2 (1 + x)dx = ln3 dx − ln3 (1 − x)dx
0 3 0 1+x 3 0
1 1 1
+ ln2 (1 − x) ln(1 + x)dx + ln3 (1 + x)dx
0 3 0
ζ (3) 4 π 2 ln 2 π2
= + 8 ln 2 − 4 ln2 2 + ln3 2 − −6+ .
2 3 3 3
(11.12)
We used that

1 1−x x= 1−t 1 ln3 t
= dt = −9ζ (3)
1+t
ln3 dx 2
0 1+x 0 (1 + t)2
and
1 π2 7 π 2 ln 2 2 ln3 2
ln2 (1−x) ln(1+x)dx = −6+ +4 ln 2−2 ln2 2+ ζ (3)− + .
0 3 2 3 3
We sketch below the calculation of the preceding integral. We have
1 1
ln2 (1 − x) ln(1 + x)dx = ln2 t ln(2 − t)dt
0 0
1
t
= ln2 t ln 2 + ln 1 − dt
0 2
∞ 1
1
= 2 ln 2 − n
t n ln2 tdt
2 n 0
n=1
∞
1
= 2 ln 2 − 2 .
2n n(n + 1)3
n=1

The calculation of the last series, which can be expressed in terms of Li2 12 and

Li3 12 , is left as a challenge to the interested reader.
Combining (11.10), (11.11), and (11.12), part (b) of the problem is solved.
3.106. Use that
Hn Hn+1 H2 Hn Hn Hn+1 1 Hn Hn+1

· = 2n + 2 − + − − .
n2 n+1 n n n n + 1 (n + 1)2 n(n + 1)
3.108. Subtract the series in parts (d) and (b) of Problem 3.107.
Derivatives and Applications
12
12.1 Apéritif
4.1. J = (−2, ∞). Since f is strictly increasing we get that f is injective. We also
have that f is surjective and it follows that f is bijective. Since f (2) = 2 we have
that g (2) = f 1(2) = 19 . From f (x) = x 3 −3x, x ∈ I , we can write g 3 (y)−3g(y) =
y, y ∈ J . Taking successive derivatives, one has that 3g 2 (y)g (y) − 3g (y) = 1,
6g(y)g 2 (y) + 3g 2 (y) · g (y) − 3g (y) = 0, y ∈ J . When y = 2 we obtain that
6 · 2 · 81
1
+ 3 · 4 · g (2) − 3g (2) = 0. Therefore, g (2) = − 243 4
.
4.2. We have
f (x + y) − f (x) f (x + y) − f (x)
= a ⇒ lim = a,
y y→0 y
and it follows that f (x) = a, which implies f (x) = ax + b, b ∈ R.

4.3. f (x) = 12 e2x + 1.
4.4. g (0) = (2021!)3 .
4.5. The equation of the tangent to the parabola at the point (x0 , y0 ) is given by
y − ax02 = 2ax0 (x − x0 ), and it follows that x1 = x20 . Let A1 be the area of the
domain bounded by the tangent to the parabola at (x0 , y0 ), parabola and the vertical
line x = x20 , and let A2 be the area of the domain bounded by the parabola, the x
axis and the vertical line x = x20 . We have
x0 ax 3
A1 = ax 2 − 2ax0 (x − x0 ) − ax02 d x = 0
x0
2
24
and
x0
2 ax03
A2 = ax 2 d x = .
0 24
Remark. More generally, if f is a twice differentiable function such that f is

continuous and positive for all x, then the graph of f is a parabola if and only if
the area of the domains bounded by the graph of f , the tangents to the graph of
f at two distinct points on the graph, and the vertical line that passes through the
intersection point of the two tangents are equal [152].
4.6. It suffices to prove that ln(1 + x − x 2 ) ≥ (1 − x) ln(1 + x), for all x ∈ [0, 1]. Let
f : [0, 1] → R, f (y) = ln(1 + y). We have f (y) = − (1+y) 1
2 < 0, and it follows
that f is concave down. We have, based on the definition of the concavity of f , that
f (α · x + (1 − α) · y) ≥ αf (x) + (1 − α)f (y), for all x, y, α ∈ [0, 1]. Therefore,
ln(1 + x(1 − x)) = ln(1 + (1 − x)x + x · 0) ≥ (1 − x) ln(1 + x) + x ln(1 + 0) =
(1 − x) ln(1 + x).
Remark. The same idea can be used for solving Problem 3.44. It suffices to prove
that ln(x 2 − x + 1) ≥ x ln x, for all x ∈ (0, 1]. Let f : (0, 1] → R, f (x) = ln x.
We have f (x) = − x12 < 0, and it follows that f is concave down. This implies,
based on the definition of the concavity of f , that f (α · x + (1 − α) · y) ≥ αf (x) +
(1 − α)f (y), for all x, y ∈ (0, 1] and α ∈ [0, 1]. Therefore, ln(x 2 − x + 1) =
ln(x · x + (1 − x) · 1) ≥ x ln x + (1 − x) ln 1 = x ln x.
4.7.
(a) Let S1 (x) = 1 + x + 2x 2 + 3x 3 + · · · + nx n , x ∈ R. Using the formula
1 − x n+1
1 + x + x2 + · · · + xn = , x = 1,
1−x
we obtain by differentiation that

1 − x n+1
1 + 2x + 3x + · · · + nx
2 n−1
= .
1−x
It follows that

1 − x n+1
S1 (x) = 1 + x(1 + 2x + 3x + · · · + nx
2 n−1
)=1+x
1−x
nx n+1 − (n + 1)x n + 1
=1+x· , x = 1.
(1 − x)2
n2 +n+2
When x = 1 we have S1 (1) = 1 + 1 + 2 + · · · + n = 1 + n(n+1) 2 = 2 .
(b) Let S2 (x) = 1 + 22 x + 32 x 2 + · · · + n2 x n−1 , x ∈ R. We have
12.1 Apéritif 409

nx n+1 − (n + 1)x n + 1
S2 (x) = S1 (x) = 1 + x ·
(1 − x)2
−n2 x n+2 + (2n2 + 2n − 1)x n+1 − (n + 1)2 x n + x + 1
= , x = 1.
(1 − x)3
When x = 1 we obtain that S2 (1) = 1 + 22 + 32 + · · · + n2 = n(n+1)(2n+1)

6 .
(c) Let S3 (x) = 2 + 3 · 2x + 4 · 3x + · · · + n(n − 1)x , x ∈ R. We have that
2 n−2

nx n+1 − (n + 1)x n + 1
S3 (x) = (1 + 2x + 3x 2 + · · · + nx n−1 ) =
(1 − x)2
−n(n − 1)x n+1 + 2(n2 − 1)x n − n(n + 1)x n−1 + 2
= , x = 1.
(1 − x)3
When x = 1 we have
n n n
S3 (1) = 2 + 2 · 3 + 3 · 4 + · · · + (n − 1) · n = (k − 1)k = k2 − k
k=2 k=2 k=2

n(n + 1)(2n + 1) n(n + 1)
= −1− −1
6 2
(n − 1)n(n + 1)
= .
3
4.8.
(a) The first inequality is equivalent to

e 1 n 1 1 n
<e− 1+ ⇔ 1+ 1+ < e.
2n + 2 n 2n + 1 n
Let f : [1, ∞) → R be the function defined by

x
1 1
f (x) = 1 + 1+ .
2x + 1 x
We have

1 x 2x + 2 2 x+1
f (x) = 1 + · · − + ln .
x 2x + 1 2x + 1 x
Let g : [1, ∞) → R, g(x) = − 2x+1

2
+ ln x+1
x . A calculation shows that
g (x) = − x(x+1)(2x+1)2 < 0, ∀x ≥ 1. It follows that g is strictly decreasing
1
on [1, ∞) and, since lim g(x) = 0, we get that g(x) > 0, for all x ≥ 1. This
x→∞
implies that f (x) > 0, for x ≥ 1, hence f is strictly increasing on [1, ∞) and,
since lim f (x) = e, we obtain that f (x) < e.
x→∞
The second inequality is equivalent to

1 n e 1 1 n
e− 1+ < ⇔e< 1+ 1+ .
n 2n + 1 2n n
Let u : [1, ∞) → R be the function defined by

1 1 x
u(x) = 1 + 1+ .
2x x
We have

1 x 2x + 1 1 2 x+1 1
u (x) = 1 + · · − + + ln − .
x 2x x 2x + 1 x x+1
Let v : [1, ∞) → R, v(x) = − x1 + 2

2x+1 + ln x+1
x −
1
x+1 . A calculation
5x 2 +5x+1
shows that v (x) = x 2 (x+1)2 (2x+1)2 > 0, x ∈ [1, ∞). It follows that v is strictly
increasing on [1, ∞) and, since lim v(x) = 0, we obtain that v(x) < 0, for
x→∞
x ∈ [1, ∞). This implies u (x) < 0, x ∈ [1, ∞), hence u is strictly decreasing
on [1, ∞) and, since lim u(x) = e, we get that u(x) > e, for x ∈ [1, ∞).
x→∞
n
1 e
lim n e − 1 + = .
n→∞ n 2
The limit can also be calculated directly using l’Hôpital’s rule.
4.9. The first inequality is equivalent to γ < xn − 1

2n+1 . Let
1 1 1
un = 1 + + · · · + − ln n − , n ≥ 1.
2 n 2n + 1
We have un+1 − un = n+1 1

− ln n+1
n − 2n+3 + 2n+1 . Let u : [1, ∞) → R be the
1 1
function defined by u(x) = x+1

1
− ln x+1
x − 2x+3 + 2x+1 . We have
1 1
16x 3 + 40x 2 + 32x + 9

u (x) = > 0, x ≥ 1,
x(x + 1)2 (2x + 1)2 (2x + 3)2
12.2 Integral Equations 411
from which it follows that u is strictly increasing on [1, ∞) and, since lim u(x) =
x→∞
0, we obtain that u(x) < 0, for all x ≥ 1. This implies un+1 < un , n ≥ 1 and, since
lim un = γ , we have that un > γ .
n→∞
The second inequality is equivalent to xn − 1
2n < γ . Let
1 1 1
vn = 1 + + · · · + − ln n − , n ≥ 1.
2 n 2n
We have vn+1 − vn = 2(n+1) + 2n − ln n . Let v : [1, ∞) → R

1 1 n+1
be the function
defined by v(x) = 2(x+1) + 2x − ln x . A calculation shows that
1 1 x+1
1
v (x) = − < 0, x ∈ [1, ∞).
2x 2 (x + 1)2
It follows that v is strictly decreasing and, since lim v(x) = 0, we have that v(x) >
x→∞
0, for all x ∈ [1, ∞). This implies that vn+1 > vn , n ≥ 1, which shows that the
sequence (vn )n≥1 is strictly increasing and, since lim vn = γ , we get that vn < γ ,
n→∞
for n ≥ 1.
Remark. We recover the inequalities in Problem 4.9 by taking a = 1 in [129,

Theorem 1.7.2, p. 29] and [129, Theorem 1.9.1, p. 35], where different proofs
are given. See [129] for other inequalities and interesting results regarding Euler’s
constant γ .
12.2 Integral Equations
4.11. We change variables according to x − t = u and we have that

x
f (x) = x + e−x eu f (u) du.
0
x
Since f is continuous, we have that the function x → 0 eu f (u) du is differen-
tiable, therefore f is differentiable. We multiply both sides of the preceding equation
by ex and we have
x
ex f (x) = x ex + eu f (u) du.
0
Let g(x) = ex f (x). A calculation shows that
g (x) = ex + xex + g(x).

This implies (g(x) e−x ) = 1 + x. It follows that

x2
g(x) = e x +x
+C ,
2
and
x2
f (x) = x + + C, C ∈ R.
2
2
Since f (0) = 0, we obtain that f (x) = x + x2 .
√
4.12. (a) f (x) = a2 x 2 + a, x ∈ R; (b) f (x) = a
2 cos( 2x) + a2 , x ∈ R.
4.13.
(a) f (x) = −x 2 − x + 1, x ∈ R;
2
(b) f (x) = −x − x2 , x ∈ R;
(c) f (x) = cos x − sin x − 1, x ∈ R.
4.14.
x2
(a) f (x) = + 1, x ∈ R;
2 √
(b) f (x) = − sin(√ 2x) − x2 , x ∈ R.
2 2
√ x3
4.15. (a) f (x) = a
+ a
2 cos( 2x), x ∈ R; (b) f (x) = −x − 6 , x ∈ R.
2
x
√
−
4.16. f (x) = a − 2a
√ e 2 sin 3
2 x , x ∈ R.
3
4.17.
(a) f (x) = a cosh x, x ∈ R;

(b) f (x) = a cos x, x ∈ R;
(c) f (x) = a cos x, x ∈ R.
12.3 Differential Equations
4.18.
(a) We replace x by −x and we have
f (x) − 2xf (−x) = x, ∀ x ∈ R;

f (−x) + 2xf (x) = −x, ∀ x ∈ R.
We add the preceding equations and we obtain that

f (x) + f (−x) + 2x(f (x) − f (−x)) = 0, ∀x ∈ R. (12.1)
Let g : R → R, g(x) = f (x) − f (−x). We have, based on equation (12.1),

that
g (x) + 2xg(x) = 0, ∀x ∈ R.
2
We multiply the preceding equation by ex and we obtain that

g(x) = C e−x ,
2 2
g(x)ex = 0, ∀x ∈ R ⇒ ∀x ∈ R,
where C is a real arbitrary constant. Therefore,
f (x) − f (−x) = C e−x ,

2
∀x ∈ R. (12.2)
When x = 0 we obtain, based on (12.2), that C = 0. Therefore, f (−x) = f (x),

∀x ∈ R. The initial differential equation implies that
f (x) − 2xf (x) = x, ∀x ∈ R.
It follows

−x 2 −x 2 1 −x 2
f (x)e = xe = − e , ∀x ∈ R,
2
2
and we have f (x) = − 12 + C ex , x ∈ R.
(b) The solution of this part of the problem is similar to the solution of part (a). We
obtain that
1
+ C e−x ,
2
f (x) = x ∈ R.
2
2
4.19. f (x) = x3 + C x, x ∈ R.
4.20. We have
⎧
⎪
⎪ x2
⎨ if k is even
f (x) = k + 2
⎪
⎪ x2
⎩ C xk + , C ∈R if k is odd.
k+2
4.21.
√
(a) We prove that f (x) = x 2 + 1. We have, by multiplying by − x12 , that

1 1 1
− 2f f (x) = − 2 , x > 0. (12.3)
x x x

We replace x by 1
x in f 1
x = 1
f (x) and we obtain that

1
f (x)f = 1, x > 0. (12.4)
x
We add equations (12.3) and (12.4) and we have that

1 1 1 1
− f f (x) + f (x)f = 1 − 2, x > 0.
x2 x x x
It follows that

1 1
f f (x) = 1 − 2 , x > 0.
x x
Thus,

1 1
f f (x) = x + + C .
x x
√
Since f (1) = 2 we have that C = 0 ⇒ f x1 f (x) = x + x1 , ∀x > 0. Using
equation (12.4) we obtain that
f (x) x
= , x > 0.
f (x) 1 + x2
√
Therefore,
√ (ln f (x)) − (ln 1 + x 2 ) = 0, ∀x > 0. It follows
√ ln f (x) −
ln 1 +√ x 2 = ln C1 , where C1 > 0. We obtain
√ f (x) = C1 1 + x . Since
2
f (1) = 2 we have that C1 = 1 and f (x) = 1 + x .
2
1−x
(b) f (x) = (1 + x)e 2(1+x) .
4.22. The functions are of the following form:
√ ab
x C
f (x) = C √ , C > 0.
a

√ 1
x 2 − 12
f (x) = C e 2C , C > 0.

√ 1 x− π
f (x) = C e C ( 4 ), C > 0.
4.25. f (x) = 0, f (x) = 2x, f (x) = 2

b sin(bx), b ∈ R∗ , f (x) = 2
a sinh(ax),
a ∈ R∗ .
4.26.
(a) We prove that these functions are:
f (x) = x;
f (x) = sin(αx) ∗
α ,α ∈R ;
f (x) = sinh(βx)
β , β ∈ R∗ .
We let y = 0 in
f (x + y) − f (x − y) = 2f (x)f (y), x, y ∈ R (12.5)
and we get that 2f (x)f (0) = 0, for all x ∈ R, and since f is not constant we get
that f (0) = 0. We differentiate (12.5) with respect to x and then with respect to y
and we get that
.
f (x + y) − f (x − y) = 2f (x)f (y), x, y ∈ R
(12.6)
f (x + y) + f (x − y) = 2f (x)f (y), x, y ∈ R.
We let y = 0 in the second equation of (12.6) and we get that 2f (x) =

2f (x)f (0). It follows that f (x)(1 − f (0)) = 0, for all x ∈ R. This implies,
since f is not constant, that f (0) = 1.
Now we differentiate the equations in (12.6), the first one with respect to x and
the second one with respect to y and we get that
.
f (x + y) − f (x − y) = 2f (x)f (y), x, y ∈ R
f (x + y) − f (x − y) = 2f (x)f (y), x, y ∈ R.
We subtract the preceding equations and we get that f (x)f (y) − f (x)f (y) = 0,
for all x, y ∈ R. Since f is not constant, there is x0 ∈ R such that f (x0 ) = 0. The
preceding equation implies
f (x0 )
f (y) − f (y) = 0, ∀y ∈ R.
f (x0 )
f (x0 )
Let k = f (x0 ) . Now we consider the following three cases:
Case k = 0. This implies that f (y) = 0, for all y ∈ R, which implies that f (y) =
ay + b. Since f (0) = 0 and f (0) = 1, we get that f (y) = y.
Case k < 0. Let k = −α 2 , with α = 0. We have that f (y) + α 2 f (y) = 0 and this
implies f (y) = C1 cos(αy) + C2 sin(αy). Since f (0) = 0 and f (0) = 1, we get
that f (y) = sin(αy)
α .
Case k > 0. Let k = β 2 , with β = 0. We have that f (y) − β 2 f (y) = 0 and this
implies that f (y) = C3 eβy + C4 e−βy . Since f (0) = 0 and f (0) = 1, we get
that f (y) = sinh(βy)
β .
(b) These functions are:

2
f (x) = x2 + C , C ∈ R;
f (x) = C − cos(αx)
α2
, C ∈ R, α ∈ R∗ ;
f (x) = C + cosh(βx)
β2
, C ∈ R, β ∈ R∗ .
Use a similar technique to that given in part (a) of the problem.

4.27.
2
(a) f (x) = C ex + x2 ex , C ∈ R, x ∈ R.
∞ 2

xn
(b) Let y(x) = n ex − 1 − 1! x
− x2! − · · · − n! , x ∈ R. A calculation shows
n=1
that y (x) − y(x) = xex , with y(0) = 0. It follows, based on part (a), that
2
y(x) = x2 ex .
4.29.
(a) Calculate the 2nth partial sum of the series.

∞
xn x n+1 x n+2
S= −2 +3 −···
n! (n+1)! (n+2)!
n=1
∞
x x2 x3 x4 xn x n+1 x n+2
= −2 +3 −4 + · · · + −2 +3 −···
1! 2! 3! 4! n! (n+1)! (n+2)!
n=2
∞
n−1=m −x x m+1 x m+2 x m+3
= xe + −2 +3 +···
(m+1)! (m+2)! (m+3)!
m=1
−x
=xe
∞
xm x m+1 x m+2 xm x m+1 x m+2
− −2 +3 +··· − − + +···
m! (m+1)! (m+2)! m! (m+1)! (m+2)!
m=1
(a)
=xe−x −S+ sinh x.
4.30. Let
∞
xn x n+1 x n+2
y(x) = n − + − ··· , x ∈ R,
n! (n + 1)! (n + 2)!
n=1
and observe that y satisfies the differential equation y (x) + y(x) = (x + 1)ex ,
x ∈ R.
4.31. Let
∞
x2 xn
f (x) = (−1)n Ckn ex − 1 − x − − ··· − .
2! n!
n=k
Then
∞
x2 x n−1
f (x) = (−1)n Ckn ex − 1 − x − − ··· −
2! (n − 1)!
n=k
∞
x2 x n−1 xn
= (−1)n Ckn e −1−x−
x
− ··· − −
2! (n − 1)! n!
n=k
∞
xn
+ (−1)n Ckn
n!
n=k
∞
(−x)k (−x)n−k
= f (x) +
k! (n − k)!
n=k
(−x)k
= f (x) + e−x .
k!
This implies that
x k −2x (−1)k x
(f (x)e−x ) = (−1)k e ⇒ f (x)e−x = t k e−2t dt+C , C ∈ R.
k! k! 0
It follows, since f (0) = 0, that C = 0 and this implies that
(−1)k x
f (x)e−x = t k e−2t dt.
k! 0
Therefore,
(−1)k ∞
lim e−x f (x) = t k e−2t dt.
x→∞ k! 0
The substitution 2t = y shows that

∞ 1 ∞ (k + 1) k!
t k e−2t dt = y k e−y dy = = k+1 .
0 2k+1 0 2k+1 2
k
It follows that lim e−x f (x) = (−1)
2k+1
x→∞
∞ * +

4.32. (a) Let y(x) = n
e x − 1 − x − x 2 − · · · − x n , x ∈ R.
2 1! 2! n!
n=1
We have
∞ "n# x x2 x n−1

y (x) = ex − 1 − − − ··· −
2 1! 2! (n − 1)!
n=1
∞ "n# x x2 xn
∞ " n # xn
= e −1− −
x
− ··· − +
2 1! 2! n! 2 n!
n=1 n=1
∞ ∞
x 2k x 2k−1
= y(x) + k + (k − 1)
(2k)! (2k − 1)!
k=1 k=1
∞ ∞
x−1 x 2k−1 x x 2k−2
= y(x) + +
2 (2k − 1)! 2 (2k − 2)!
k=1 k=1
x−1 x
= y(x) + sinh x + cosh x.
2 2
We obtain the linear differential equation y (x) = y(x) + x−1

2 sinh x + 2 cosh x,
x
x ∈ R, with initial condition y(0) = 0. A calculation shows that y(x) = 14 sinh x +

4 (x − x)e .
1 2 x
12.4 Higher Order Derivatives
4.33. Prove the formulae by mathematical induction.

4.34.
(a) We assume that sin x is a polynomial function, i.e. sin x = P (x), ∀x ∈ R, where
P is a polynomial of degree n. We differentiate the preceding equality n + 1
times and we get, based on part (c) of Problem 4.33, that sin x + n+1 2 π = 0,
∀x ∈ R, which is a contradiction.
(b) We assume that ex is a polynomial function, i.e. ex = P (x), ∀x ∈ R. Let
n = degree P . We differentiate the preceding equality n + 1 times and we get
that ex = 0, ∀x ∈ R, which contradicts ex > 0, ∀x ∈ R.
P (x)
(c) We assume that ex is a rational function. We have ex = Q(x) , ∀x ∈ R, where
P and Q are polynomials. It follows that Q(x)ex = P (x), ∀x ∈ R. Let
n = degree P . We differentiate n times the preceding equality and we have

n
that Ckn Q(n−k) (x)ex = n!an , where an = 0 is the leading coefficient of P .
k=0

n
It follows that ex = 1
n!an Ckn Q(n−k) (−x), ∀x ∈ R, which contradicts the fact
k=0
that ex is not a polynomial function.
4.35. We have g (x) − g(x) = −f (x) ≤ 0, ∀x ∈ R. It follows that (e−x g(x)) ≤ 0,

∀x ∈ R. This implies that the function x → e−x g(x) decreases on R and it follows
that e−x g(x) ≥ lim e−x g(x) = 0 ⇒ g(x) ≥ 0, ∀x ∈ R.
x→∞
x x x
4.36. (a) f (x) = C e 2 , C ∈ R;√ (b) f (x) = C e− 2 , C ∈ R; (c)√ f (x) = C e 2a ,
3− 5 5−1
C ∈ R; (d) f (x) = C e 2 x , C ∈ R; (e) f (x) = C e 2 x , C ∈ R;
− √x √x
(f) f (x) = C1 e 2 + C2 e 2 , C1 , C2 ∈ R; (g) f (x) = C cosh √x , C ∈ R.
2
4.37.
(n)
3x 3 · n!
(a) = (−1)n+1 , n ≥ 1;
1+x (x + 1)n+1
(n)
1 n n! 2n+1 1
(b) = (−1) − , n ≥ 0;
2x 2 + 5x − 3 7 (2x − 1)n+1 (x + 3)n+1
(n)
1+x 1 · 3 · 5 · · · (2n − 3) − 2n+1
(c) √ = (−1)n−1 x 2 (x − 2n + 1), n ≥ 2;
x 2n

1 + x (n) 1 · 4 · 7 · · · (3n − 5) − 3n+1
(d) √ = (−1)n−1 x 3 (2x − 3n + 2), n ≥ 2;
3
x 3n
nπ
(e) (sin2 x)(n) = −2n−1 cos 2x + , n ∈ N;
nπ
2
(f) (cos2 x)(n) = 2n−1 cos 2x + , n ∈ N;
2
(g) use the formula sin(3x) = 3 sin x − 4 sin3 x and it follows that
3 nπ 3n nπ
(sin3 x)(n) = sin x + − sin 3x + , n ≥ 0;
4 2 4 2
(h) use the formula cos(3x) = 4 cos3 x − 3 cos x and it follows that
3 nπ 3n nπ
(cos3 x)(n) = cos x + + cos 3x + , n ≥ 0;
x 4 2 4 2
e − e−x 1 x
(n)
(i) = e − (−1)n e−x , n ≥ 0;
2 2
x (n)
e + e−x 1 x
(j) = e + (−1)n e−x , n ≥ 0.
2 2
4.38. We have
1 3 cos(4x)
sin4 x+cos4 x = (sin2 x+cos2 x)2 −2 sin2 x cos2 x = 1− sin2 (2x) = + ,
2 4 4

and it follows that f (n) (x) = 4n−1 cos 4x + nπ
2 .
4.39. Solution due to Andreea Ilieş, computer science student 2017/2018.

We have
f (x) = ex (cos x + sin x) = f (x) + g(x),
g (x) = ex (cos x − sin x) = g(x) − f (x).
It follows that
2 2 0 (n−1) 12 0 (n−1) 12
f (n) (x) + g (n) (x) = f (x) + g (x)
0 12 0 12
= (f (x)+g(x))(n−1) + (g(x)−f (x))(n−1)
0 12 0 12
= f (n−1) (x)+g (n−1) (x) + g (n−1) (x)−f (n−1) (x)
2 2
=2 f (n−1) (x) + g (n−1) (x)
2 2
=2 f (n−2) (x) + g (n−2) (x)
2
=···
0 2 2 1
=2n−1 f (x) + g (x)
0 1
=2n f 2 (x)+g 2 (x)
=2n e2x .
4.40.

(a) Since 1
1+x 2
= 1
2i
1
x−i − 1
x+i , we have

n n! 1 1 n! (x+i)n+1 −(x−i)n+1
f (n)
(x)=(−1) − =(−1)n · ,
2i (x−i)n+1 (x+i)n+1 2i (x 2 +1)n+1
and it follows that

2nk+1 2nk+1
f (2nk) (1) 1 1+i 1−i
= − .
(2nk)! 2i 2 2
√ iπ √ iπ
Because 1 + i = 2e 4 and 1 − i = 2e− 4 , we have that
∞
∞ ∞
f (2nk) (1) 1 1+i 1 + i 2nk 1 − i 1 − i 2nk
= −
(2nk)! 2i 2 2 2 2
n=0 n=0 n=0

1 1+i 22k1−i 22k
= · 2k − ·
2i 2 2 − (1 + i)2k 2 22k − (1 − i)2k
2k − cos kπ
2 + sin 2
kπ
= 2k−1 .
22k − 2k+1 cos kπ
2 +1
(b) When k = 1 and k = 2, we obtain the formulae

∞ ∞
f (2n) (1) 3 f (4n) (1) 2
= and = .
(2n)! 5 (4n)! 5
n=0 n=0
4.41.
(a) f : R → R, f (x) = (x 2 − 3x − 1)e2x . Using Leibniz’s formula, we have
f (n) (x) = C0n (x 2 − 3x − 1) · 2n · e2x + C1n (2x − 3) · 2n−1 e2x + C2n · 2 · 2n−2 e2x

= 2n−2 e2x 4x 2 + 4(n − 3)x + n2 − 7n − 4 .
(b) f : R → R, f (x) = x−1

ex . We use Leibniz’s formula and we have
x−n−1
f (n) (x) = C0n (x − 1)(−1)n e−x + C1n (−1)n−1 e−x = (−1)n .
ex
(c) f : R → R, f (x) = eax cos(bx + c), a, b, c ∈ R. We have, based on Leibniz’s

formula, that
n
kπ
f (n) (x) = Ckn a n−k bk eax cos bx + c +
2
k=0
n
kπ
= eax Ckn a n−k bk cos bx + c + .
2
k=0
4.42. x = π4 + kπ , k ∈ Z.
4.43. We use that
2 (k − 1)!
f (x) = ln(x + 1) and (ln(1 + x))(k) = (−1)k−1 , k ≥ 1.
x+1 (x + 1)k
We write f (x) = ln2 (x + 1) = ln(x + 1) · ln(x + 1), we apply Leibniz’s formula

and we have that
(n − 1)!
f (n) (x) = C0n (−1)n−1 · ln(x + 1)
(x + 1)n
n−1
(n − k − 1)! (k − 1)!
+ Ckn (−1)n−k−1 · (−1)k−1
(x + 1)n−k (x + 1)k
k=1
(n − 1)!
+ Cnn ln(x + 1) · (−1)n−1
(x + 1)n
1
= 2(−1)n−1 (n − 1)! ln(x + 1)
(x + 1)n
n−1
n! 1
+ (−1)n−2 · (k − 1)!(n − k − 1)! ·
k!(n − k)! (x + 1)n
k=1
1
= 2(−1)n−1 (n − 1)! ln(x + 1)
(x + 1)n
n−1
n 1
+ (−1)n−2 (n − 1)! ·
k(n − k) (x + 1)n
k=1
1
= 2(−1)n−1 (n − 1)! ln(x + 1)
(x + 1)n
n−1
1 1 1
+ (−1) (n − 1)!
n−2
+
k n − k (x + 1)n
k=1
1
= 2(−1)n−1 (n − 1)! ln(x + 1)
(x + 1)n
n−1
1 1
− 2(−1)n−1 (n − 1)!
(x + 1)n k
k=1
n−1

1 1
= 2(−1) n−1
(n − 1)! ln(x + 1) − , n ≥ 2.
k (x + 1)n
k=1
4.44. We use that

(k)
1 k! (k − 1)!
= (−1)k k+1 , k≥0 and (ln x)(k) = (−1)k−1 , k ∈ N.
x x xk
We have
n
n! (n − k)! (k − 1)!
f (n) (x) = C0n (−1)n ln x + · (−1)k−1
Ckn (−1)n−k
x n+1 x n−k+1 xk
k=1

n−1 n! 1 1 1
= (−1) 1 + + + · · · + − ln x , n ∈ N.
x n+1 2 3 n
4.45. Solution I. Mathematical induction. Let n ≥ 1 and let P (n) be the statement
(x n ln x)(n) = n!(Hn +ln x). One can check that (x ln x) = 1+ln x = 1!(H1 +ln x),
so P (1) is true. We assume that P (n − 1) is true and we prove that P (n) holds true.
Since f (x) = x n ln x, we have that f (x) = nx n−1 ln x+x n−1 and by differentiation
n − 1 times we obtain that
f (n) (x) = n(x n−1 ln x)(n−1) + (x n−1 )(n−1) (P (n − 1) holds true)

= n(n − 1)!(Hn−1 + ln x) + (n − 1)!

1
= n! Hn−1 + ln x +
n
= n!(Hn + ln x).
Solution II. Leibniz’s formula. We have

n
(x n ln x)(n) = (x n )(n) ln x + Ckn (x n )(n−k) (ln x)(k)
k=1
n
n! k (k − 1)!
= n! ln x + Ckn · · x · (−1)k−1
k! xk
k=1
n
1
= n! ln x + n! (−1)k−1 Ckn · .
k
k=1
Now we prove that the following binomial identity holds true.
A Binomial Identity Let n ∈ N. Then

n
1 1 1 1
(−1)k−1 Ckn = 1 + + + · · · + .
k 2 3 n
k=1
Proof. We have
n n 1
1
(−1)k−1 Ckn = Ckn (−x)k−1 d x
k 0
k=1 k=1
1 n
1
=− Ckn (−x)k d x
0 x
k=1
1 1 − (1 − x)n
= dx (1 − x = t)
0 x
1 1 − tn
= dt
0 1−t
1
= (1 + t + t 2 + · · · + t n−1 )d t
0
1 1 1
=1+ + + ··· + ,
2 3 n
and the identity is proved.
4.46. Use Problem 4.44.

1
4.47. Let f (x) = x n−1 e x . We prove that
f (x)
1 (n) 1 1
f (n) (x) = (−1)n , i.e. x n−1 e x = (−1)n n+1 e x .
x 2n x
We solve the problem by mathematical induction. Let n ≥ 1 and let P (n) be the
statement

1 (n) 1 1
x n−1 e x = (−1)n n+1 e x .
x
1 1
Since e x = − x12 e x , we have that P (1) holds true. We assume that P (k) is
true for all k < n and we prove that P (n) holds true.
We have
1 1
f (x) = (n − 1)x n−2 e x − x n−3 e x .
We differentiate the preceding equality n − 1 times and we have that

f (n) (x) = (f (x))(n−1) (x)

1 (n−1)

1 (n−1)
= (n − 1) x n−2 e x − x n−3 e x (P (n − 1) holds true)

1 1 1 (n−2)
= (n − 1) (−1) n−1
ex − x n−3 e x (P (n − 2) holds true)
xn

1 1 n−2 1 1
= (n − 1)(−1) e − (−1)
n−1x e x
xn x n−1

1 1 n−1 1 1 1
= (n − 1)(−1)n−1 n e x + (−1)n−1 − n e x − n+1 e x
x x x
1 1
= (−1)n ex .
x n+1
4.48. We have
2 arcsin x
f (x) = √
1 − x2
2 xf (x)
f (x) = + .
1−x 2 1 − x2
Therefore,
(1 − x 2 )f (x) − xf (x) − 2 = 0.
We differentiate n − 1 times the preceding formula, using Leibniz’s formula, and

we have that
(1 − x 2 )f (n+1) (x) − (2n − 1)xf (n) (x) − (n − 1)2 f (n−1) (x) = 0, n ≥ 2.
When x = 0 we obtain the recurrence formula
f (n+1) (0) = (n − 1)2 f (n−1) (0), n ≥ 2.
When n = 2p, p ∈ N, we have that
f (2p+1) (0) = (2p − 1)2 (2p − 3)2 · · · 12 · f (0).
Since f (0) = 0, we obtain that f (2p+1) (0) = 0, for all p ∈ N.

When n = 2p + 1, p ∈ N, we obtain from the recurrence formula that
f (2p+2) (0) = (2p)2 (2p − 2)2 · · · 22 · f (0).
Since f (0) = 2, we have that f (2p+2) (0) = 22p+1 (p!)2 , for all p ∈ N.
4.49. We have (1 + x 2 )f (x) = 1. We differentiate n times the preceding equality,

using Leibniz’s formula, and we have
(x 2 + 1)f (n) (x) + 2nxf (n−1) (x) + n(n − 1)f (n−2) (x) = 0, n ≥ 2.
When x = 0 we obtain the recurrence formula
f (n) (0) = −n(n − 1)f (n−2) (0), n ≥ 2.
When n = 2p, p ∈ N, it follows that
f (2p) (0) = (−1)p (2p)(2p − 1) · · · 2 · 1 · f (0) (0).
Since f (0) (0) = f (0) = 1, we obtain that f (2p) (0) = (−1)p (2p)!, for all p ∈ N.
When n = 2p + 1, p ∈ N, we obtain from the recurrence formula that
f (2p+1) (0) = (−1)p (2p + 1)(2p) · · · 3 · 2 · f (0).
Since f (0) = 0, we have that f (2p+1) (0) = 0, for all p ∈ N.

For the function g we have
1
g (x) = = f (x).
1 + x2
Therefore, g (2n) (0) = f (2n−1) (0) = 0, n ∈ N. Since g (0) (0) = g(0) = 0, it follows
that g (2n) (0) = 0, ∀n ≥ 0. Also, g (2n+1) (0) = f (2n) (0) = (−1)n (2n)!, ∀n ≥ 0.
For the function h, using Leibniz’s formula and the derivatives (already obtained)
of the function g, we have that
n
h(n) (0) = Ckn g (n−k) (0)g (k) (0), n ≥ 0.
k=0
Therefore,
2n
h(2n) (0) = Ck2n g (2n−k) (0)g (k) (0)
k=0
n
= C2k−1
2n g
(2n−2k+1)
(0)g (2k−1) (0)
k=1
n
= C2k−1
2n (−1)
n−k
(2n − 2k)!(−1)k−1 (2k − 2)!
k=1
k
1
= (−1)n−1 (2n)!
(2n − 2k + 1)(2k − 1)
k=1
n
(2n)! 1 1
= (−1)n−1 +
2n 2n − 2k + 1 2k − 1
k=1
n
1
= (−1) n−1
2(2n − 1)! , ∀n ∈ N.
2k − 1
k=1
When n = 0 we have h(0) (0) = h(0) = 0.

On the other hand,
2n+1
h(2n+1) (0) = Ck2n+1 g (2n−k+1) (0)g (k) (0) = 0, ∀n ≥ 0.
k=0
4.50. We have
f (3n) (0) = (−1)n (3n)!, n≥0

f (3n+1) (0) = 0, n≥0
f (3n+2)
(0) = 0, n ≥ 0.
4.51. We have
n+1
1 4
f (2n) − = (−1)n (2n)!, n≥0
2 3

1
f (2n+1) − = 0, n ≥ 0.
2
4.52. We have
f (3n+1) (0) = 0, n≥0

f (3n+2) (0) = 0, n≥0
f (3n+3)
(0) = −3(3n + 2)!, n ≥ 0.
4.53.
• We have
1 ex − ea
ea e(x−a)t dt = , x = a.
0 x−a
We obtain, by differentiating n times the preceding formula, that

(n)
1 ex − ea
e a (x−a)t n
e t dt = .
0 x−a
It follows that
(n)
ex − ea 1 1 √
n y√
n = nea e(x−a)t t n dt = ea e(x−a) n
y dy.
x−a 0 0
Thus,
(n)
ex − ea 1 √
n y√
1
lim n = lim ea e(x−a) n
y dy = ea ex−a dy = ex .
n→∞ x−a n→∞ 0 0
• We have
sin x 1
= cos(xt)dt, x ∈ R∗ .
x 0
We differentiate 2n + 1 times the preceding equality and we have that

sin x (2n+1) 1 2n+1 1
= t 2n+1 cos xt+ π dt=(−1)n−1 t 2n+1 sin (xt) dt.
x 0 2 0
It follows that
(2n+1) 1
sin x
(−1) n−1
n =n t 2n+1 sin (xt) dt
x 0
n 1
√ √
= 2n+1
y sin x 2n+1
y dy
2n+1 0
and
(2n+1)
sin x n 1
√ √
lim (−1) n−1
n = lim 2n+1
y sin x 2n+1
y dy
n→∞ x n→∞ 2n + 1 0
1 1
= sin x dy
2 0
sin x
= .
2
• We have
ln(1 + y) ∞ dx
= , y ∈ (−1, 1), y = 0
y 1 x(1 + (1 + y)(x − 1))
and it follows that

(n) ∞
ln(1 + y) (−1)n n!(x − 1)n dx
= ·
y 1 [1 + (1 + y)(x − 1)]n+1 x
∞ tn dt
x−1=t
= (−1)n n! ·
0 [1 + (1 + y)t] n+1 1+t
u 1
t= 1−u(1+y) 1+y un
= n
(−1) n! du
0 1 − uy
v
u= 1+y (−1)n n!vn 1
= vy dv
(1 + y)n+1 0 1 − 1+y
√
v n =t (−1) (n − 1)!
n 1 n
t
= √
n dt.
(1 + y)n+1 0 1− y t
1+y
This implies that

(n) √
(1 + y)n ln(1 + y) 1 1 n
t
lim (−1)n = lim √ dt
n→∞ (n − 1)! y 1 + y n→∞ 0 1− ynt
1+y
1 1 1
= y dt
1+y 0 1 − 1+y
= 1.
(n)
4.54. Solution I due to H. Ricardo [123]. The crucial observation is that fn (0) is
n! times the coefficient of x n in the Taylor expansion of fn (x) about the origin. We
have sin(kx) = kx + O(x 3 ), which gives
-
n
fn (x) = (kx + O(x 3 )) = n!x n + O x n+2 .
k=1
(n)
Thus, fn (0) = n!2 .
(k)
Solution II. We apply the
generalized Leibniz formula. We have, since sin (ax+
b) = a sin ax + b + 2 , k ≥ 0, a, b ∈ R, that
k kπ

(n) n! k π kn π
fn (x) = 1k1 2k2 · · · nkn sin x + 1 · · · sin nx +
k1 !k2 ! · · · kn ! 2 2
k1 +k2 +···+kn =n
and when x = 0 one has that


n! k1 π kn π
fn(n) (0) = 1 2 · · · n sin
k1 k2 kn
· · · sin ,
k1 !k2 ! · · · kn ! 2 2
k1 +k2 +···+kn =n
(12.7)
where ki ≥ 0 and k1 + k2 + · · · + kn = n.
If ki is an even integer, i.e. ki = 2mi , mi ≥ 0, one has that sin ki2π = sin mi π =
0. If ki are odd integers, i.e. ki = 2mi − 1, mi ≥ 1, then sin ki2π = sin (2mi2−1)π =
(−1)mi −1 . The equation k1 + k2 + · · · + kn = n implies that 2m1 − 1 + 2m2 − 1 +
· · · + 2mn − 1 = n ⇒
m1 + m2 + · · · + mn = n. (12.8)
However, mi ≥ 1, ∀i = 1, 2, . . . , n and this implies that m1 +m2 +· · ·+mn ≥ n.

Thus, the only case when one has equality in (12.8) is when m1 = m2 = · · · =
mn = 1 ⇒ k1 = k2 = · · · = kn = 1. It follows, from formula (12.7), that
n!
fn(n) (0) = · 1 · 2 · 3 · · · n = n!2 .
1! · 1! · · · 1!
4.55. If n is odd, then gn(n) (0) = 0. If n is even, then we have
(2n) (2n)!3 1
g2n (0) = (−1)n .
2n i 2 i 2 · · · in2
1≤i1 <i2 <···<in ≤2n 1 2
The preceding formula is obtained by calculating the coefficient of x n in the

Taylor expansion of gn (x) about the origin. See the first solution of Problem 4.54.
(n)
4.56. fn (0) = n!2 . See the first solution of Problem 4.54.
(n)
4.57. If n is odd, then gn (0) = 0. If n is even, then we have
(2n) (2n)!3 1
g2n (0) = 2i2 · · · i2
.
2n i
1≤i1 <i2 <···<in ≤2n 1 2 n
The preceding formula is obtained by calculating the coefficient of x n in the

Taylor expansion of gn (x) about the origin. See the first solution of Problem 4.54.
12.5 Taylor’s Formula
4.58.
(a) ∀ x ∈ (−3, ∞), ∃ θ−1,x ∈ (0, 1) such that

n
1 (x + 1)k (x + 1)n+1
= (−1)k + (−1)n+1 ;
x+3 2k+1 [2 + θ−1,x (x + 1)]n+2
k=0
(b) ∀ x ∈ R, ∃ θ2,x ∈ (0, 1) such that

n

2k 2n+1 2θ2,x (x−2)
e 2x−1
=e 3
(x − 2) +
k
e (x − 2) n+1
;
k! (n + 1)!
k=0
(c) when n = 2m, m ≥ 0, ∀ x ∈ R, ∃ θ π2 ,x ∈ (0, 1) such that
32k π 2k
m
sin(3x) = (−1)k+1 x−
(2k)! 2
k=0
32m+1 π π 2m+1
+ (−1)m+2 sin 3θ π2 ,x x − x− ,
(2m + 1)! 2 2
and, when n = 2m + 1, m ≥ 0, ∀ x ∈ R, ∃ θ π2 ,x ∈ (0, 1) such that
32k π 2k
m
sin(3x) = (−1)k+1 x−
(2k)! 2
k=0
32m+2 π π 2m+2
+ (−1)m+2 cos 3θ π2 ,x x − x− ;
(2m + 2)! 2 2
(d) when n = 2m, m ≥ 0, ∀ x ∈ R, ∃ θ1,x ∈ (0, 1) such that
(x − 1)2 (x − 1)4 (x − 1)2m

cosh(x − 1) = 1 + + + ··· +
2! 4! (2m)!
(x − 1)2m+1
+ sinh(θ1,x (x − 1)),
(2m + 1)!
and, when n = 2m + 1, m ≥ 0, ∀ x ∈ R, ∃ θ1,x ∈ (0, 1) such that
(x − 1)2 (x − 1)4 (x − 1)2m

cosh(x − 1) = 1 + + + ··· +
2! 4! (2m)!
(x − 1)2m+2
+ cosh(θ1,x (x − 1)).
(2m + 2)!
4.59.
(a) f (x) = x 5 + 7x 4 − 2x 3 − 46x 2 + 65x − 25;

(b) f (x) = x 4 − 5x 3 − 31x 2 − 43x − 18.
4.60. f (x) = (x − 1)5 + 3(x − 1)2 + 1.

4.61. f (x) = (x + 1)5 − 7(x + 1)3 − 1.
4.62.

(a) (Mn f )(x) = 12 nk=0 (−1)k 1 − 3k+1
1
xk ;

(b) (Mn f )(x) = 1 + 12 x + nk=2 (−1)k−1 (2k−3)!! xk ;
n (2k)!!
k
(c) (Mn f )(x) = − ln 2 − k=1 1 − 21k xk .
4.63.
(a) ∀ x ∈ R, ∃ θx ∈ (0, 1) such that
x x2 x3 xn x n+1 θx x
ex = 1 + + + + ··· + + e ;
1! 2! 3! n! (n + 1)!
(b) when n = 2m, m ≥ 0, ∀ x ∈ R, ∃ θx ∈ (0, 1) such that
x3 x5 x 2m−1 x 2m+1
sin x = x− + −· · ·+(−1)m−1 +(−1)m cos(θx x),
3! 5! (2m − 1)! (2m + 1)!
and, when n = 2m − 1, m ∈ N, ∀ x ∈ R, ∃ θx ∈ (0, 1) such that
x3 x5 x 2m−1 x 2m
sin x = x − + −· · ·+(−1)m−1 +(−1)m sin(θx x);
3! 5! (2m − 1)! (2m)!
(c) when n = 2m, m ≥ 0, ∀ x ∈ R, ∃ θx ∈ (0, 1) such that
x2 x4 x 2m x 2m+1
cos x = 1− + −· · ·+(−1)m +(−1)m+1 sin(θx x),
2! 4! (2m)! (2m + 1)!
and, when n = 2m + 1, m ≥ 0, ∀ x ∈ R, ∃ θx ∈ (0, 1) such that
x2 x4 x 2m x 2m+2
cos x = 1− + −· · ·+(−1)m +(−1)m+1 cos(θx x);
2! 4! (2m)! (2m + 2)!
(d) ∀ x ∈ (−1, ∞), ∃ θx ∈ (0, 1) such that
α α(α − 1) 2 α(α − 1)(α − 2) 3

(1 + x)α = 1 + x+ x + x + ···
1! 2! 3!
α(α − 1)(α − 2) · · · (α − n + 1) n
+ x
n!
(continued)
α(α − 1)(α − 2) · · · (α − n)
+ (1 + θx x)α−n−1 x n+1 ;
(n + 1)!
(e) ∀ x ∈ (−1, ∞), ∃ θx ∈ (0, 1) such that
x2 x3 n−1 x n n+1
ln(1 + x) =x − 2 + 3 − · · · + (−1) n + (−1)n xn+1 1
(1+θx x)n+1
;
(f) when n = 2m, m ≥ 0, ∀ x ∈ R, ∃ θx ∈ (0, 1) such that
x3 x5 x 2m−1 x 2m+1
sinh x = x + + + ··· + + cosh(θx x),
3! 5! (2m − 1)! (2m + 1)!
and, when n = 2m − 1, m ∈ N, ∀ x ∈ R, ∃ θx ∈ (0, 1) such that
x3 x5 x 2m−1 x 2m
sinh x = x + + + ··· + + sinh(θx x);
3! 5! (2m − 1)! (2m)!
(g) when n = 2m, m ≥ 0, ∀ x ∈ R, ∃ θx ∈ (0, 1) such that
x2 x4 x 2m x 2m+1
cosh x = 1 + + + ··· + + sinh(θx x),
2! 4! (2m)! (2m + 1)!
and, when n = 2m + 1, m ≥ 0, ∀ x ∈ R, ∃ θx ∈ (0, 1) such that
x2 x4 x 2m x 2m+2
cosh x = 1 + + + ··· + + cosh(θx x).
2! 4! (2m)! (2m + 2)!
4.64. It suffices to consider only the case when x < 0. We have, based on Maclaurin
formula of order 2n for the exponential function that, ∀x ∈ R, ∃ θ ∈ (0, 1) such that
x x2 x 2n x 2n+1 θx
ex = 1 + + + ··· + + e .
1! 2! (2n)! (2n + 1)!
x2 x 2n x 2n+1 θx
When x < 0 ⇒ x 2n+1 < 0 ⇒ 1 + x
1! + 2! + ··· + (2n)! = ex − (2n+1)! e > 0.
4.65. e is not a rational number

We assume that e is a rational number, i.e. ∃ a, b ∈ N such that e = ab . We
have, based on Maclaurin’s formula of order n for the exponential function,
that ∀x ∈ R, ∃ θ ∈ (0, 1) such that
x xn x n+1 θx
ex = 1 + + ··· + + e .
1! n! (n + 1)!
When x = 1 we have that
a 1 1 eθ
= e = 1 + + ··· + + .
b 1! n! (n + 1)!
The preceding equality implies that

a 1 1 eθ
n! − 1 − − ··· − = . (12.9)
b 1! n! n+1
θ
We note that 0 < n+1e
< 1, for n ≥ 2. Let n > max {2, b}. Equality (12.9) is
a contradiction since the left hand side of (12.9) is an integer, while the right
hand√side of (12.9) is a real number between 0 and 1.
e 2 is not a rational
√ number √
We assume that e 2 is a rational number. − 2 is also a
√ This implies that e
√ it follows that cosh 2 is a rational number, i.e. ∃ a, b ∈
rational number and
N such that cosh 2 = ab . We have, based on Maclaurin’s formula for cosh,
that ∀x ∈ R, ∃ θ ∈ (0, 1) such that
x2 x 2n x 2n+2
cosh x = 1 + + ··· + + cosh(θ x).
2! (2n)! (2n + 2)!
√
When x = 2 we have that
a √ 2 2n 2n+1 √
= cosh 2 = 1 + + · · · + + cosh θ 2 .
b 2! (2n)! (2n + 2)!
2k
Since (2k)! = 1
(2k−1)!!k! , k ≥ 1, the preceding equality becomes
√
a √ 1 1 cosh θ 2
= cosh 2 = 1 + + ··· + + .
b 1!!1! (2n − 1)!!n! (2n + 1)!!(n + 1)!
(continued)
It follows that
√
cosh θ 2
a 1 1
(2n−1)!!n! −1− − ··· − = .
b 1!!1! (2n − 1)!!(n − 1)! (2n + 1)(n + 1)
For n large enough, the preceding equality contradicts the fact that the
left hand side of the equality is an integer, while the right hand side is a real
number between 0 and 1.
To prove that sin 1 and cos 1 are irrational numbers, one can apply the same
technique as above.
4.66.
(a) We have
∞ ∞ ∞ ∞ ∞
n3 n2 k 2 + 2k + 1 k2 k
= = = +2 +e
n! (n − 1)! k! k! k!
n=1 n=1 k=0 k=1 k=1
∞ ∞
k 1
= +2 +e
(k − 1)! (k − 1)!
k=1 k=1
∞ ∞
1 1
= + + 3e
(k − 2)! (k − 1)!
k=2 k=1
= 5e.
(b) We have
∞
f (n) (0) n
f (x) = x , |x| < R.
n!
n=0
It follows that
∞
n3 (n)
x xf (x) = f (0)x n−1 .
n!
n=1
This implies that
∞
n3 (n)
x xf (x) (1) = f (0).
n!
n=1

A calculation shows that x xf (x) (1) = f (1) + 3f (1) + f (1).
4.67. We have, based on Maclaurin’s formula of order 1, that ∀x ∈ R there exists

θ ∈ (0, 1) such that
1
f (x) = 1 + f (θ x)x 2
2
and it follows that

x 1 x x2
f √ =1+ f θ√ .
n 2 n n
This implies that

n 2 n
x 1 x x
lim f √ = lim 1 + f θ √
n→∞ n n→∞ 2 n n
2

lim x f θ √xn
=e n→∞ 2
x2
= ep 2 .
12.6 Series with the Maclaurin Remainder of a Function f
4.68.
(a) Use that
xn x n+1 x n+2 x t n−1

+ + + ··· = dt, x ∈ [−1, 1).
n n+1 n+2 0 1−t
(b) Use Taylor’s formula with the remainder in integral form or Abel’s summation
formula.
(c) We have
∞

f (n) (0) n f (n+1) (0) n+1
S= an x + x +···
n! (n+1)!
n=0
∞

f (0) f (0) 2 f (n) (0) n f (n+1) (0) n+1
=f (0)+ x+ x +···+ an x + x +···
1! 2! n! (n+1)!
n=1
∞

f (n) (0) n f (n+1) (0) n+1
=f (x)+ an x + x +···
n! (n+1)!
n=1
∞

n−1=m f (m+1) (0) m+1 f (m+2) (0) m+2
= f (x)+ a m+1 x + x +···
(m+1)! (m+2)!
m=0
∞

f (m) (0) m f (m+1) (0) m+1 f (m) (0) m
=f (x)+a am x + x +··· − x
m! (m+1)! m!
m=0
∞
f (m) (0)
=f (x)+aS−a (ax)m
m!
m=0
=f (x)+aS−f (ax).
4.69.
(a) Integrate by parts.

(b) We have
∞ ∞
x x2 xn x (x − t)n t
e −1− −
x
− ··· − = e dt
1! 2! n! n!
n=0 n=0 0
∞

x (x − t)n
= e t
dt
0 n!
n=0
x
= et ex−t dt
0
x
= ex dt
0
= xex .
(c) This part of the problem is solved similarly to part (b).

4.70. Use Taylor’s formula with the remainder in integral form or apply Abel’s
summation formula.
4.71.
(a) Use mathematical induction.

(b) Apply Abel’s summation formula with an = (−1)n−1 n2 and bn = ex − 1 − 1!
x
−
x2 xn
2! − ··· − n! .
4.72. We have
∞

nπ f (0) f (0) 2 f (n) (0) n
S= cos f (x)−f (0)− x− x −···− x
2 1! 2! n!
n=0
∞

f (0) f (0) 2 f (2k) (0) 2k
= (−1) k
f (x)−f (0)− x− x −···− x
1! 2! (2k)!
k=0
∞

f (0) f (2k) (0) 2k
=f (x)−f (0)+ (−1) k
f (x)−f (0)− x− · · · − x
1! (2k)!
k=1
∞

k−1=i f (0) f (2i+2) (0) 2i+2
= f (x)−f (0)+ (−1) i+1
f (x)−f (0)− x− · · · − x
1! (2i+2)!
i=0
∞

f (0) f (2i) (0) 2i
=f (x)−f (0)− (−1) i
f (x)−f (0)− x− · · · − x
1! (2i)!
i=0
∞ ∞
f (2i+1) (0) 2i+1 f (2i+2) (0) 2i+2
+ (−1)i x + (−1)i x
(2i+1)! (2i+2)!
i=0 i=0
f (xi)−f (−xi) f (xi)+f (−xi)

=f (x)−f (0)−S+ +f (0)− .
2i 2
We used in the previous calculations the following formulae:
∞
f (2i+2) (0) 2i+2 f (xi) + f (−xi)
• (−1)i x = f (0) − , |x| < R;
(2i + 2)! 2
i=0
∞
f (2i+1) (0) 2i+1 f (xi) − f (−xi)
• (−1)i x = , |x| < R.
(2i + 1)! 2i
i=0
The second series formula can be proved similarly.

4.73.
(a) First, we prove that the series converges absolutely. We apply the Maclaurin
formula of order n to the function f (x) = sin x and we have, since f (x) =
Mn (x) + Rn (x), that
1 3 sin nπ
x − ··· −
sin x − x + x = Rn (x).
2 n
3! n!

Since sin(k) (ax) = a k sin ax + kπ2 , k ≥ 0, a, x ∈ R, we have, by using the
formula for the remainder Rn , that

(n+1)π
sin(n+1)
(θx x) n+1 sin θx x + 2
Rn (x) = x = x n+1 ,
(n + 1)! (n + 1)!
for some θx ∈ (0, 1). It follows that

∞ / / ∞ / sin θ x+ (n+1)π /
/ n sin nπ / / n x /
/ 3 sin x−x+ 1 x 3 − · · · − 2 n /
x /= /3 2
x n+1 /
/ 3! n! / (n+1)! /
n=0 n=0
∞
1 (3|x|)n+1
≤
3 (n+1)!
n=0
1
= e3|x| −1 ,
3
which shows that the series converges absolutely.
Now, we calculate its sum. We have
∞
1 3 sin nπ
f (x) = 3 sin x − x + x − · · · −
n 2 n
x
3! n!
n=0
∞
1 sin nπ
= sin x + 3n sin x − x + x 3 − · · · − 2 n
x
3! n!
n=1
∞
(m+1)π

n−1=m 1 sin
= sin x + 3m+1 sin x − x + x 3 − · · · − 2
x m+1
3! (m + 1)!
m=0
∞
1 sin mπ
= sin x + 3 3m sin x − x + x 3 − · · · − 2
xm
3! m!
m=0
∞
sin (m+1)π
− 3m+1 2
x m+1
(m + 1)!
m=0
∞
sin(m+1) (0)
= sin x + 3f (x) − (3x)m+1
(m + 1)!
m=0
= sin x + 3f (x) − sin(3x),
and it follows that 2f (x) = sin(3x) − sin x ⇒ f (x) = sin x cos(2x).

(b) See [144].

∞ ∞
x x2 xn x (x − t)n t
Ckn f (n) (0) e −1− −
x
− ··· − = Ckn f (n) (0) e dt
1! 2! n! 0 n!
n=k n=k
x ∞ x ∞
1 f (n) (0) n−k=i 1 f (k+i) (0)
= et (x − t)n dt = et (x − t)k+i dt
k! 0 (n − k)! k! 0 i!
n=k i=0
x ∞ (k) (i) x
1 f (0) 1
= et (x − t)k (x − t)i dt = et (x − t)k f (k) (x − t)dt
k! 0 i! k! 0
i=0
1 x
x−t=y
= ex−y y k f (k) (y)dy.
k! 0
4.75. Apply Abel’s summation formula with an = (−1)n and bn = f (x) −

(Tn f )(x). The series equals f (0) − f (x)+f
2
(−x)
.
4.76.
(a) Solution I. Calculate the 4nth partial sum of the series.

Solution II. We have
∞
n
x x2 xn
y(x) = (−1) 2 e − 1 − −
x
− ··· −
1! 2! n!
n=1
∞
x x x2 xn
(−1) 2 ex − 1 − −
n
= ex − 1 − + − ··· −
1! 1! 2! n!
n=2
and it follows that

∞
x x2 x n−2
(−1) 2 ex − 1 − −
n
y (x) = ex + − ··· −
1! 2! (n − 2)!
n=2
∞ " #
n−2=m x m+2 x x2 xm
= e + (−1) 2
ex − 1 − − − ··· −
1! 2! m!
m=0
∞
m
x x2 xm
=e −x
(−1) 2 e −1− −
x
− ··· −
1! 2! m!
m=0
= ex − (ex − 1) − y(x)
= 1 − y(x).
We obtained the differential equation with constant coefficients y + y = 1.

The solution of this equation is y(x) = C1 cos x + C2 sin x + 1, where C1 , C2 ∈
R. Since y(0) = 0 and y (0) = 0, we get that C1 = −1 and C2 = 0 and these
imply that y(x) = 1 − cos x.
(b) Calculate the 4nth partial sum of the series or differentiate the series in part (a).
The general case " #

4n k (k)

Let S4n = (−1) 2 f (x) − f (0) − f 1!(0) x − f 2!(0) x 2 − · · · − f k!(0) x k
k=1
be the 4nth partial sum of the first series.
f (0) 2 f (4) (0) 4 f (4n−2) (0) 4n−2 f (4n) (0) 4n

S4n = x − x + ··· + x − x
2! 4! (4n − 2)! (4n)!
and it follows that

f (ix) + f (−ix)
lim S4n = f (0) − .
n→∞ 2
The second series can be calculated similarly.
4.77. Solution I.

(b) We apply Taylor’s formula with the remainder in integral form and we have
∞

f (0) f (0) 2 f (n) (0) n
Ckn f (x) − f (0) − x− x − ··· − x
1! 2! n!
n=k
∞ x (x − t)n (n+1)
= Ckn f (t)dt
0 n!
n=k
1 x ∞ (x − t)n (n+1)
= f (t)dt
k! 0 n=k (n − k)!
(x − t)i (k+1) (i)

x ∞
n−k=i 1
= (x − t)k f (t)dt
k! 0 i!
i=0
y i (k+1) (i)
x ∞
x−t=y 1
= yk f (x − y)dy
k! 0 i!
i=0
1 x
= y k f (k+1) (x)dy
k! 0
f (k+1) (x) k+1

= x .
(k + 1)!
Solution II. Use Abel’s summation formula combined to part (a) of the problem.
The challenging part of this solution would be to prove that

f (0) f (0) 2 f (n+1) (0) n+1
lim Ck+1 f (x) − f (0) − x− x − ··· − x = 0.
n→∞ n+1 1! 2! (n + 1)!
This limit can be proved using an idea of R. Mabry, which appears in the solution
given by him [107] for a problem proposed by O. Furdui [33].
4.78. We integrate by parts twice and we have
/
∞
Tn (x) − sin x Tn (x) − sin x //∞ 1 ∞ T (x) − cos x
n
In = dx = − / + dx
0 x 2n+1 2nx 2n 0 2n 0 x 2n
/
1 ∞ T (x) − cos x
n Tn (x) − cos x //∞
= dx = −
2n 0 x 2n 2n(2n − 1)x 2n−1 /0
1 ∞ Tn (x) + sin x
+ dx
2n(2n − 1) 0 x 2n−1
1 ∞ Tn−1 (x) − sin x
=− dx
2n(2n − 1) 0 x 2n−1
1
=− In−1 ,
2n(2n − 1)
since Tn (x) = −Tn−1 (x). It follows that
1 (−1)n−1
In = − In−1 and In = I1 .
2n(2n − 1) 2n(2n − 1) · · · 4 · 3
We calculate I1 integrating by parts twice and we have

/
∞x − sin x x − sin x //∞ 1 ∞ 1 − cos x
I1 = dx = − / +2 dx
0 x3 2x 2 0 0 x2
/
1 ∞ 1 − cos x 1 − cos x //∞ 1 ∞ sin x
= dx = − / +2 dx
2 0 x2 2x 0 0 x
1 ∞ sin x
= dx
2 0 x
π
= .
4
12.7 Series with Fractional Part Function
4.80, 4.81, and 4.82. See [59].
12.8 Extrema of One Variable Functions

√
4.83. (a) min f (R) = −17, max f (R) = 17; (b) max f (R) = f ln 5+1
2 = − 54 ;
2
16 ; (d) min f (R) = −625.
(c) min f ((−∞, 0)) = f (−1) = 3π
4.84. min f (C) = f (−2) = −30, max f (C) = f (2) = 30.
4.85.
(a) x = −1 is a global minimum point and f (−1) = − 12 is the global minimum

value, x = 1 is a global maximum point, and f (1) = 12 is the global maximum
value;
(b) x = −1 is a global maximum point and f (−1) = 12 is the global maximum
value, x = 0 is a global minimum point and f (0) = 0 is the global minimum
value, x = 1 is a global maximum point, and f (1) = 12 is the global maximum
value;
(c) x = −1 is a local maximum point and f (−1) = 8 is the local maximum value,
x = 0 is not a local extremum point of f (it is an inflection point), x = 5 is a
local minimum point, and f (5) = −3124 is the local minimum value;
(d) x = 0 is not a local extremum point of f (it is an inflection point), x = 2 is a
global minimum point, and f (2) = −65 is the global minimum value of f ;
√ √ √ 1
(e) x = − 2
2
is a global minimum point of f and f − 22 = − 22 e− 2 is the
√
global minimum value of f , x = 2
is a global maximum point of f , and
√ √ 2
2 − 12
f ( 22 )= 2 e
is the global maximum value of f ;
(f) x = −1 is a global maximum point of f and f (−1) = 1e is the global maximum
value of f , x = 0 is a global minimum point of f and f (0) = 0 is the global
minimum value of f , x = 1 is a global maximum point of f , and f (1) = 1e is
the global maximum value of f ;
(g) x = 0 is a global minimum point of f and f (0) = 1 is the global minimum
value of f ;
(h) x = e12 is a local maximum point of f and f e12 = e42 is the local maximum
value of f , x = 1 is a global minimum point of f , and f (1) = 0 is the global
minimum value of f ;
1
e
(i) x = 1e is a global minimum point of f and f 1e = 1e is the global
(j) x = π4 + 2kπ , k ∈ Z, are local maximum points of f and f π4 + 2kπ =
√
2 π4 +2kπ
2 e k ∈ Z, are local maximum values of f , x = π4 + (2k + 1)π , k ∈ Z,
,
√ π
are local minimum points of f , and f π4 + (2k + 1)π = − 22 e 4 +(2k+1)π ,
k ∈ Z, are local minimum values of f ;
(k) x = −1 is a global minimum point of f and f (−1) = − arctan 1e is the global
2
(l) x = − π2 is a global minimum point of f and f − π2 = − π4 − 2 is the global
2
(m) x = − π2 is a global maximum point of f and f − π2 = π4 − 2 is the global
maximum value of f .
We have f (x) = 2 cos x − 2x − π . Therefore, f (x) = 0 if and only if cos x =

x + π2 . Since y = x + π2 is the equation of the tangent line to the graph of the
cosine function at − π2 , it follows that the only critical point of f is − π2 . Since

f − π2 = f − π2 = 0, f IV − π2 = −2 < 0, we have that x = − π2 is a local
2
maximum point of f and f − π2 = π4 − 2 is the local maximum value of f .
Now we prove that x = − π2 is in fact a global maximum point of f , since
π2 π2
f (x) ≤ − 2 ⇐⇒ 2 sin x − x 2 − π x ≤ −2
4 4
π 2
⇐⇒ 2(1 + sin x) ≤ x +
2
y=x+ π2
⇐⇒ 2(1 − cos y) ≤ y 2
y y 2
⇐⇒ sin2 ≤ .
2 2
on the graph of the parabola x = 4y .

4.86. Let M(0, 33) and let P (x, y) be a point 2
The distance from M to P is d(M, P ) = x 2 + (y − 33)2 . Let
f (x, y) = x 2 +(y−33)2 and g(y) = f (4y 2 , y) = 16y 4 +y 2 −66y+1089, y ∈ R.
Solution I. We observe that the function g can be written as
g(y) = 16(y 2 − 1)2 + 33(y − 1)2 + 1040, y ∈ R.
It follows that the minimum value of g is attained at y = 1, which shows that the
distance from M to the graph of the parabola√x = 4y 2 is obtained at the point
P0 (4, 1) and the distance equals d(M, P0 ) = 4 65.
Solution II. The only critical point of g is y = 1 and, since g (1) = 194 > 0,
it follows that this is a local minimum point of g. In fact, this point is a global
minimum point of g, since g(y) − g(1) = 16(y 2 − 1)2 + 33(y − 1)2 ≥ 0, for all
y ∈ R. Therefore, the distance from M to the graph of √ the parabola x = 4y 2 is
attained at P0 (4, 1) and the distance equals d(M, P0 ) = 4 65.
on the graph of the parabola x = 4y. The
4.87. Let M(0, b) and P (x, y) be a point 2
distance from M to P is d(M, P ) = x + (y − b) . Let

2 2
√
f (x, y) = x 2 +(y −b)2 and g(y) = f (±2 y, y) = y 2 +2(2−b)y +b2 , y ≥ 0.
Solution I. We write the function g as follows:
g(y) = (y − b + 2)2 + 4b − 4, y ≥ 0.
When b ≥ 2, the minimum value of g is attained at y = b − 2, which implies that

√ from M to the graph of the parabola x = 4y is obtained for
the distance 2
√ the points
P1,2 (±2 b − 2, b − 2) and the distance is d(M, P1 ) = d(M, P2 ) = 2 b − 1. We
mention that, when b = 2 the points P1 and P2 are identical to O(0, 0).
When b < 2, the minimum value of g is attained at y = 0, which implies that
the distance from M to the graph of the parabola x 2 = 4y is obtained for the point
P0 (0, 0) = O(0, 0) and the distance is d(M, O) = |b|.
Solution II. We distinguish several cases.
Case b ≥ 2. The only critical point of g is y = b−2 and since g (b−2) = 2 > 0,
it follows that this is a local minimum point of g. In fact, this point is a global
minimum point of g, since g(y) − g(b − 2) = (y − b + 2)2 ≥ 0, for all y ≥ 0.
Therefore, √ the distance from M to the graph of the parabola x 2 = 4y is √attained
at P1,2 (±2 b − 2, b −2) and the distance is d(M, P1 ) = d(M, P2 ) = 2 b − 1.
We mention that, when b = 2 points P1 and P2 are identical to O(0, 0).
Case 0 < b < 2. In triangle OM P , where O(0, 0), M (0, 2), and P (x, y) is a
point on the parabola, with y > 0, we have OM = 2 and M P > 2. From
here, it follows that M OP > M P O. Therefore, in triangle OMP we have

MOP > MP O, which implies OM < MP . Thus, the distance from M to
the graph of the parabola is attained at P0 (0, 0) = O(0, 0) and the distance is
d(M, O) = b.
Case b = 0. The point M(0, 0) = O(0, 0) is on the parabola and the distance is
d(O, O) = 0.
Case b < 0. The distance is d(M, O) = −b.
√
Therefore, the distance from M to the graph of the parabola x 2 = 4y is 2 b − 1,
for b ≥ 2, and |b|, for b < 2. 6
4.88. The critical points of f are 0 and ± − 5a 3b
. Since f (0) = 0 and f (0) =
6b = 0, it follows
6 that 0 is not a local extremum point of f .
Let α = − 5a 3b
. Since f (∓α) = ±6bα = 0, it follows that points −α and α
are local extremum points of f . We have
M + m = f (−α) + f (α) = −aα 5 − bα 3 + c + aα 5 + bα 3 + c = 2c.
On the other hand,
Mm = f (−α)f (α)
= (−aα 5 − bα 3 + c)(aα 5 + bα 3 + c)
= [−α 3 (aα 2 + b) + c][α 3 (aα 2 + b) + c]

2b 2b
= −α 3 +c α3 +c
5 5
4b2
= −α 6 + c2
25
108b5
= + c2 .
3125a 3
4.89. First, we consider the case n = 1. We have x − xy + y − 1 = (x − 1)(1 − y) =

(x − 1)2 ≥ 0.
Now, we consider the case n ≥ 2. Let f : [0, 2] → R, f (x) = x n − 2x + x 2 +
(2 − x)n . A calculation shows that
f (x) = nx n−1 − 2 + 2x − n(2 − x)n−1

f (x) = n(n − 1)x n−2 + 2 + n(n − 1)(2 − x)n−2 .
Since f (x) > 0, ∀x ∈ [0, 2], we have that f is a strictly increasing function and
it follows, since f (1) = 0, that f (x) ≤ 0, ∀x ∈ [0, 1] and f (x) ≥ 0, ∀x ∈ [1, 2].
This shows that f (x) ≥ f (1) = 1, ∀x ∈ [0, 2].
Partial Derivatives and Applications
13

Differential Operators
5.1.
∂f
(a) (x, y) = −(y 2 + 2) cos(cos(xy 2 + 2x − y + 10)) sin(xy 2 + 2x − y + 10),
∂x
∂f
(x, y) = −(2xy − 1) cos(cos(xy 2 + 2x − y + 10)) sin(xy 2 + 2x − y + 10);
∂y
∂f 2y sin(xy) cos(xy) ∂f 2x sin(xy) cos(xy)
(b) (x, y) = , (x, y) = ;
∂x sin (xy) + 1
2 ∂y sin2 (xy) + 1
∂f y ∂f z ∂f x
(c) (x, y) = + ln z, (x, y) = + ln x, (x, y) = + ln y;
∂x x ∂y y ∂z z
∂f
(d) (x, y, z) = e cos(x − yz) [yz cos(x − yz) − 2 sin(x − yz)],
xyz
∂x
∂f
(x, y, z) = exyz cos(x − yz) [xz cos(x − yz) + 2z sin(x − yz)],
∂y
∂f
(x, y, z) = exyz cos(x − yz) [xy cos(x − yz) + 2y sin(x − yz)];
∂z
∂f ∂f ∂f
(e) (x, y, z) = zx z−1 y z , (x, y, z) = zx z y z−1 , (x, y, z) = (xy)z ln(xy);
∂x ∂y ∂z
∂f ∂f ∂f
(f) (x, y, z) = yz · x yz−1 , (x, y, z) = x yz · z ln x, (x, y, z) = x yz · y ln x;
∂x ∂y ∂z
∂f ∂f
(x, y, z) = y z · x y −1 ,
z z
(g) (x, y, z) = z · y z−1 · x y ln x,
∂x ∂y
∂f z
(x, y, z) = y z · x y ln x ln y.
∂z
5.2.
∂f y ∂f x ∂ 2f y2
(a) (x, y) = , (x, y) = , (x, y) = − ,
∂x xy + 1 ∂y xy + 1 ∂x 2 (xy + 1)2
∂ 2f 1 ∂ 2f x2
(x, y) = , (x, y) = − ;
∂y∂x (xy + 1)2 ∂y 2 (xy + 1)2
∂f
(b) (x, y) = cos(xy) − y(x − y) sin(xy),
∂x
∂f
(x, y) = − cos(xy) − x(x − y) sin(xy),
∂y
∂ 2f
(x, y) = −2y sin(xy) − y 2 (x − y) cos(xy),
∂x 2
∂ 2f
(x, y) = −2(x − y) sin(xy) − xy(x − y) cos(xy),
∂y∂x
∂ 2f
(x, y) = 2x sin(xy) − x 2 (x − y) cos(xy);
∂y 2
∂f x ∂f x
(c) (x, y) = 2x arctan + y, (x, y) = 2y arctan − x,
∂x y ∂y y
∂ 2f x 2xy ∂ 2f −x 2 + y 2
(x, y) = 2 arctan + 2 , (x, y) = 2 ,
∂x 2 y x + y2 ∂y∂x x + y2
∂ 2f x 2xy
(x, y) = 2 arctan − 2 ;
∂y 2 y x + y2
∂f 2
(d) (x, y, z) = ex y [2xy sin(x − z) + cos(x − z)],
∂x
∂f 2 ∂f 2
(x, y, z) = x 2 ex y sin(x − z), (x, y, z) = −ex y cos(x − z),
∂y ∂z
∂ 2f x2y 2
(x, y, z) = (4x y + 2y − 1)e sin(x − z) + 4xyex y cos(x − z),
2 2
∂x 2
∂ 2f 2 2
(x, y, z) = 2x(x 2 y + 1)ex y sin(x − z) + x 2 ex y cos(x − z),
∂y∂x
∂ 2f 2 2
(x, y, z) = ex y sin(x − z) − 2xyex y cos(x − z),
∂z∂x
∂ 2f 2
2
(x, y, z) = x 4 ex y sin(x − z),
∂y
∂ 2f 2
(x, y, z) = −x 2 ex y cos(x − z),
∂z∂y
∂ 2f 2
2
(x, y, z) = −ex y sin(x − z);
∂z
∂f y 2 e−z ∂f e−z
(e) (x, y, z) = − , (x, y, z) = − ,
∂x (xy − 1)2 ∂y (xy − 1)2
∂f ye−z
(x, y, z) = − ,
∂z xy − 1
∂ 2f 2y 3 e−z ∂ 2f 2ye−z
(x, y, z) = , (x, y, z) = ,
∂x 2 (xy − 1)3 ∂y∂x (xy − 1)3
∂ 2f y 2 e−z ∂ 2f 2xe−z
(x, y, z) = , (x, y, z) = ,
∂z∂x (xy − 1)2 ∂y 2 (xy − 1)3
∂ 2f e−z ∂ 2f ye−z
(x, y, z) = , (x, y, z) = ;
∂z∂y (xy − 1)2 ∂z2 xy − 1
∂f yz − 1 ∂f 1
(f) (x, y, z) = , (x, y, z) = − ,
∂x (xz − 1)2 ∂y xz − 1
∂f x(x − y)
(x, y, z) = − ,
∂z (xz − 1)2
∂ 2f 2z(yz − 1) ∂ 2f z
(x, y, z) = − , (x, y, z) = ,
∂x 2 (xz − 1) 3 ∂y∂x (xz − 1)2
∂ 2f 2x − y − xyz ∂ 2 f
(x, y, z) = , (x, y, z) = 0,
∂z∂x (xz − 1)3 ∂y 2
∂ 2f x ∂ 2f 2x 2 (x − y)
(x, y, z) = , (x, y, z) = ;
∂z∂y (xz − 1)2 ∂z2 (xz − 1)3
∂f z ∂f xz2
(g) (x, y, z) = , (x, y, z) = − ,
∂x yz − 1 ∂y (yz − 1)2
∂f x
(x, y, z) = − ,
∂z (yz − 1)2
∂ 2f ∂ 2f z2
(x, y, z) = 0, (x, y, z) = − ,
∂x 2 ∂y∂x (yz − 1)2
2
∂ f 1 2
∂ f 2xz3
(x, y, z) = − , (x, y, z) = ,
∂z∂x (yz − 1)2 ∂y 2 (yz − 1)3
∂ 2f 2xz ∂ 2f 2xy
(x, y, z) = , (x, y, z) = ;
∂z∂y (yz − 1)3 ∂z2 (yz − 1)3
∂f cos(x − y) ∂f cos(x − z)
(h) (x, y, z) = , (x, y, z) = − 2 ,
∂x cos(y − z) ∂y cos (y − z)
∂f sin(x − y) sin(y − z)
(x, y, z) = − ,
∂z cos2 (y − z)
2
∂ f sin(x − y) 2
∂ f sin(x − z)
(x, y, z) = − , (x, y, z) = ,
∂x 2 cos(y − z) ∂y∂x cos2 (y − z)
∂ 2f cos(x − y) sin(y − z)
(x, y, z) = − ,
∂z∂x cos2 (y − z)
∂ 2f 2 cos(x − z) sin(y − z)
(x, y, z) = − ,
∂y 2 cos3 (y − z)
∂ 2f cos(x − z) sin(y − z) − sin(x − y)
(x, y, z) = ,
∂z∂y cos3 (y − z)
∂ 2f 1 + sin2 (y − z)
(x, y, z) = sin(x − y) · .
∂z 2 cos3 (y − z)
5.3.
(a) We have
. x 4 y+4x 2 y 3 −y 5
∂f (x 2 +y 2 )2
if (x, y) = (0, 0)
(x, y) =
∂x 0 if (x, y) = (0, 0)
and
. x 5 −4x 3 y 2 −xy 4
∂f (x 2 +y 2 )2
if (x, y) = (0, 0)
(x, y) =
∂y 0 if (x, y) = (0, 0).
∂f
Evidently, the function ∂x is continuous on R2 \ {(0, 0)}. Since
/ /
/ ∂f /
/≤ |y| x + 4x y + y ≤ 2|y|,
4 2 2 4
/ (x, y)
/ ∂x / (x 2 + y 2 )2
∂f ∂f ∂f
for all (x, y) = (0, 0), it follows that lim (x, y) =0= ∂x (0, 0), so
(x,y)→(0,0) ∂x ∂x
is continuous at (0, 0). We proved that the partial derivative ∂f

∂x is continuous on
∂f
R . Similarly, one can prove that the partial derivative ∂y is continuous on R2 ,
2
so f is of class C 1 .
(b) We have
∂f ∂f
∂ 2f ∂ ∂f ∂x (0, y) − ∂x (0, 0) −y
(0, 0) = (0, 0) = lim = lim = −1
∂y∂x ∂y ∂x y→0 y−0 y→0 y
and
∂f ∂f
∂ 2f ∂ ∂f ∂y (x, 0) − ∂y (0, 0) x
(0, 0) = (0, 0) = lim = lim = 1.
∂x∂y ∂x ∂y x→0 x−0 x→0 x
∂2f ∂2f
Therefore, ∂y∂x (0, 0) = ∂x∂y (0, 0).
(c) We prove that f does not satisfy the conditions of Schwarz’s theorem. We have
. x 6 +9x 4 y 2 −9x 2 y 4 −y 6
∂ 2f (x 2 +y 2 )3
if (x, y) = (0, 0)
(x, y) =
∂y∂x −1 if (x, y) = (0, 0).

1
The sequence k,0 converges to (0, 0), but
k∈N

∂ 2f 1 ∂ 2f
lim ,0 = 1 = (0, 0).
k→∞ ∂y∂x k ∂y∂x
∂2f
Therefore, the function ∂y∂x is not continuous at (0, 0). Similarly, one can
∂2f
prove that the other partial derivative ∂x∂y is not continuous at (0, 0).
5.4.
(a) For all (x, y) ∈ R2 , with xy = 0, we have
∂f y ∂f x
(x, y) = 2x arctan − y and (x, y) = −2y arctan + x.
∂x x ∂y y
Let a ∈ R be arbitrary. We have
f (x, 0) − f (a, 0) 0
lim = lim = 0,
x→a x−a x→a x − a
∂f
so, f is partially derivable with respect to the variable x at (a, 0) and ∂x (a, 0) =
0. Also, if a = 0, then
f (a, y) − f (a, 0) 0
lim = lim = 0,
y→0 y−0 y→0 y
and if a = 0, then
y

f (a, y) − f (a, 0) 2 arctan a a
lim = lim a − y arctan = a.
y→0 y−0 y→0 y y
Thus, f is partially derivable with respect to the variable y at (a, 0) and

∂f
∂y (a, 0) = a.
∂f
Similarly, one can prove that f is partially derivable at (0, a) and ∂x (0, a) =
−a, ∂f
∂y (0, a) = 0.
In conclusion, f is partially derivable on R2 and we have
.
y
∂f 2x arctan −y if x=0
(x, y) = x
∂x −y if x=0
.
∂f −2y arctan x
+x if y = 0
(x, y) = y
∂y x if y = 0.
One can prove, based on the expressions of the partial derivatives of f written
above, that these are continuous functions on R2 , so f is of class C 1 .
(b) We have
∂f ∂f
∂ 2f ∂x (0, y) − ∂x (0, 0) −y
(0, 0) = lim = lim = −1
∂y∂x y→0 y−0 y→0 y
and
∂f ∂f
∂ 2f ∂y (x, 0) − ∂y (0, 0) x
(0, 0) = lim = lim = 1.
∂x∂y x→0 x−0 x→0 x
(c) We prove that f does not satisfy the conditions of Schwarz’s theorem. We have
. x2−y 2
∂ 2f if x = 0
(x, y) = x +y
2 2
∂y∂x −1 if x = 0.
Let (xn )n∈N be a sequence of real numbers which converges to 0. The

sequence ((xn , 0))n∈N converges to (0, 0) and
∂ 2f ∂ 2f
lim (xn , 0) = 1 = (0, 0).
n→∞ ∂y∂x ∂y∂x
∂2f
Therefore, the function ∂y∂x is not continuous at (0, 0). Similarly, one can
∂2f
prove that the function ∂x∂y is not continuous at (0, 0).
5.6. L(f ) = −2f and L(L(f )) = 4f .

2n n 2

5.9. ∂y∂ n ∂xf n (x, y) = ex+y Ckn (n − k)!(x k + y k ).
k=0
∂ 4p f
5.10. When n = 2p, p ≥ 0 is an integer, we have = 0, and when
∂y 2p ∂x 2p
(0, 0)
12 0
∂ 2(2p+1) f
n = 2p + 1, p ≥ 0 is an integer, we have ∂y 2p+1 ∂x 2p+1 (0, 0) = (2p+2)!
2 .
5.13. We have
⎛ ⎞
1 sin y
⎜ y x ⎟
J (f )(x, y) = ⎜
⎝ 1 + x2y2
⎟, (x, y) ∈ R2 ,
1 + x2y2 ⎠
π cos(π x − y) − cos(π x − y)
and
⎛ ⎞
1 0
J (f )(1, 0) = ⎝ 0 1⎠ .
−π 1
5.14. We have
⎛ ⎞
yexy−z xexy−z exy−z
J (f )(x, y, z) = ⎝ − xy−z ⎠,
exy−z + 1 exy−z + 1 e +1
2
2xyz(x + 1) yz−1 2 2 2 2
z(x + 1) ln(x + 1) y(x + 1) ln(x + 1)
yz yz
for all (x, y, z) ∈ R3 , and

1 1 −1 −1
J (f )(−1, 1, −1) = .
2 1 − ln 2 ln 2
D(x,y)
5.15. D(ρ,θ) = ρ.
D(x,y,z)
5.16. D(ρ,ϕ,θ) = ρ 2 sin ϕ.
D(x,y)
5.17. D(u,v) = 1
(1−u2 −v 2 )2
.
D(x,y,z)
5.18. D(u,v,w) = 24uvw(w − u)(v − u)(w − v).
5.19. We have

y(y − 1)x y−2 x y−1 (1 + y ln x)
H (f )(x, y) = , (x, y) ∈ (0, ∞) × R,
x y−1 (1 + y ln x) x y ln2 x
and it follows that

21
H (f )(1, −1) = .
10
5.20. We have
⎛ ⎞
2(y − z) 1 x − 2y + z
⎜ (z − x)3 (z − x)2 (z − x)3 ⎟
⎜ ⎟
⎜ 1 1 ⎟
H (f )(x, y, z) = ⎜ 0 − ⎟,
⎜ (z − x)2 (z − x)2 ⎟
⎝ x − 2y + z 1 2(x − y) ⎠
− −
(z − x)3 (z − x)2 (z − x)3
for all (x, y, z) ∈ A, and

⎛ ⎞
2 1 −3
H (f )(1, −1, 0) = ⎝ 1 0 −1⎠ .
−3 −1 4

yz xz xy
5.21. ∇f (x, y, z) = , , ;
1 + x 2 y 2 z2 1 + x 2 y 2 z2 1 + x 2 y 2 z2
2xyz(x 2 y 2 + y 2 z2 + x 2 z2 )
Δf (x, y, z) = − .
(1 + x 2 y 2 z2 )2
x y z
5.22. ∇f (x, y, z) = − 2 , − , − ;
x + y 2 + z2 x 2 + y 2 + z2 x 2 + y 2 + z2
1
Δf (x, y, z) = − 2 .
x + y 2 + z2
5.24. The only point which satisfies the property is (3, 3, 3).
5.25. The only point which satisfies the property is (5, 53 , 12 ).
y
5.26. div F (x, y, z) = 1 + x − 1+(x−yz) 2;

−z
rot F (x, y, z) = 1+(x−yz) 2 + e , 1 − 1+(x−yz)2 , y + 1 .
z 1
5.27. div F (x, y, z) = cos(yz) + exz + xy cos(xyz);

rot F (x, y, z) = (xz cos(xyz) − xy exz , −xy sin(yz) − yz cos(xyz),
yz exz +xz sin(yz)).
13.2 The Chain Rule
5.28.
(a) Let u(x) = sin x and v(x) = arctan x. Then F (x) = f (u(x), v(x)) and we have
1
F (x) = cos x · fu (u(x), v(x)) + · f (u(x), v(x)),
1 + x2 v
2 cos x
F (x) = cos2 x · fu2 (u(x), v(x)) + · f (u(x), v(x))
1 + x 2 uv
1
+ · f 2 (u(x), v(x)) − sin x · fu (u(x), v(x))
(1 + x 2 )2 v
2x
− · f (u(x), v(x)).
(1 + x 2 )2 v
(b) Let u(x) = ln(x 2 + 1) and v(x) = cos x. Then F (x) = f (u(x), v(x)) and we
have
2x
F (x) = · f (u(x), v(x)) − sin x · fv (u(x), v(x)),
x2 +1 u
4x 2 4x sin x
F (x) = · fu2 (u(x), v(x)) − 2 · fuv (u(x), v(x))
(x + 1)
2 2 x +1
2(x 2 − 1)
+ sin2 x · fv 2 (u(x), v(x)) − · f (u(x), v(x))
(x 2 + 1)2 u
− cos x · fv (u(x), v(x)).
5.29.
(a) Let u(x) = −x and v(x) = 3x. Then F (x) = f (u(x), v(x)) and we have that
F (x) = −fu3 (u(x), v(x)) + 9fu2 v (u(x), v(x))

−27fuv 2 (u(x), v(x)) + 27fv 3 (u(x), v(x)).
(b) Let u(x) = x and v(x) = x 2 . Then F (x) = f (u(x), v(x)) and we have that
F (x) = fu3 (u(x), v(x)) + 6xfu2 v (u(x), v(x)) + 12x 2 fuv 2 (u(x), v(x))
+8x 3 fv 3 (u(x), v(x)) + 6fuv (u(x), v(x)) + 12xfv 2 (u(x), v(x)).
5.30.
√
(a) Let u(x) = ex , v(x) = ln(x + x 2 + 1) and w(x) = x 2 . Then F (x) =
f (u(x), v(x), w(x)) and we have that
1
F (x) = ex · fu (u(x), v(x), w(x)) + · fv (u(x), v(x), w(x))
x2 + 1
+2x · fw (u(x), v(x), w(x)),
1
F (x) = e2x · fu2 (u(x), v(x), w(x)) + 2 · f 2 (u(x), v(x), w(x))
x +1 v
2 ex
+4x 2 · fw2 (u(x), v(x), w(x)) + · fuv (u(x), v(x), w(x))
x2 + 1
4x
+ · fvw (u(x), v(x), w(x)) + 4x ex · fuw (u(x), v(x), w(x))
x2 + 1
x
+ ex · fu (u(x), v(x), w(x)) − · fv (u(x), v(x), w(x))
(x 2 + 1) x 2 + 1
+2 · fw (u(x), v(x), w(x)).
2
(b) Let u(x) = 2x , v(x) = arccot x and w(x) = ex . Then F (x) =
f (u(x), v(x), w(x)) and we have that
1
F (x) = 2x ln 2 · fu (u(x), v(x), w(x)) − · fv (u(x), v(x), w(x))
1 + x2
2
+2x ex · fw (u(x), v(x), w(x)),
1
F (x) = 22x ln2 2 · fu2 (u(x), v(x), w(x)) + · f 2 (u(x), v(x), w(x))
(1 + x 2 )2 v
2 2x+1 ln 2
+4x 2 e2x · fw2 (u(x), v(x), w(x)) − · fuv (u(x), v(x), w(x))
1 + x2
2
4x ex 2
− · fvw (u(x), v(x), w(x)) + x ex 2x+2 ln 2 · fuw (u(x), v(x), w(x))
1 + x2
2x
+2x ln2 2 · fu (u(x), v(x), w(x)) + · fv (u(x), v(x), w(x))
(1 + x 2 )2
2
+2(2x 2 + 1) ex · fw (u(x), v(x), w(x)).
5.31.
(a) Let u(x, y) = x 2 − 2xy + 3y. Then F (x, y) = f (u(x, y)) and we have that
Fx (x, y) = 2(x − y) · f (u(x, y)),

Fy (x, y) = −(2x − 3) · f (u(x, y)),
Fx 2 (x, y) = 4(x − y)2 · f (u(x, y)) + 2 · f (u(x, y)),

Fxy (x, y) = −2(x − y)(2x − 3) · f (u(x, y)) − 2 · f (u(x, y)),
Fy 2 (x, y) = (2x − 3)2 · f (u(x, y)).
(b) Let u(x, y) = x sin y − y cos x. Then F (x, y) = f (u(x, y)) and we have that
Fx (x, y) = (sin y + y sin x) · f (u(x, y)),

Fy (x, y) = (x cos y − cos x) · f (u(x, y)),
Fx 2 (x, y) = (sin y + y sin x)2 · f (u(x, y)) + y cos x · f (u(x, y)),

Fxy (x, y) = (sin y + y sin x)(x cos y − cos x) · f (u(x, y))

+(cos y + sin x) · f (u(x, y)),
Fy 2 (x, y) = (x cos y − cos x)2 · f (u(x, y)) − x sin y · f (u(x, y)).
5.32.
(a) Let u(x, y, z) = x − y − z. Then F (x, y, z) = f (u(x, y, z)) and we have that
Fxyz (x, y, z) = f (u(x, y, z)),

Fyz2 (x, y, z) = −f (u(x, y, z)).
(b) Let u(x, y, z) = xyz. Then F (x, y, z) = f (u(x, y, z)) and we have that
Fx 3 (x, y, z) = y 3 z3 f (u(x, y, z)),
Fx 2 z (x, y, z) = xy 3 z2 f (u(x, y, z)) + 2y 2 zf (u(x, y, z)).
(c) Let u(x, y, z) = x 2 − xy + z. Then F (x, y, z) = f (u(x, y, z)) and we have that
Fxy 2 (x, y, z) = x 2 (2x − y)f (u(x, y, z)) + 2xf (u(x, y, z)),
Fxyz (x, y, z) = −x(2x − y)f (u(x, y, z)) − f (u(x, y, z)).
(d) Let u(x, y, z, t) = xyt − z + t. Then F (x, y, z, t) = f (u(x, y, z, t)) and we

have
Fxyt (x, y, z, t) = x(xy + 1)yt 2 f (u(x, y, z, t)) + (3xy + 1)tf (u(x, y, z, t))
+f (u(x, y, z, t)),
Fxzt (x, y, z, t) = −(xy + 1)ytf (u(x, y, z, t)) − yf (u(x, y, z, t)),
Fxt 2 (x, y, z, t) = (xy + 1)2 ytf (u(x, y, z, t)) + 2(xy + 1)yf (u(x, y, z, t)),
Fyzt (x, y, z, t) = −x(xy + 1)tf (u(x, y, z, t)) − xf (u(x, y, z, t)),

IV (x, y, z, t) = −x(xy + 1)yt 2 f IV (u(x, y, z, t)) − (3xy + 1)tf (u(x, y, z, t))
Fxyzt
−f (u(x, y, z, t)).
5.33.
(a) Let u(x, y) = xy − x + y and v(x, y) = x arctan y. We can write F (x, y) =

f (u(x, y), v(x, y)) and we have
Fx (x, y) = (y − 1) · fu (u(x, y), v(x, y)) + arctan y · fv (u(x, y), v(x, y)),
x
Fy (x, y) = (x + 1) · fu (u(x, y), v(x, y)) + · fv (u(x, y), v(x, y)),
1 + y2
Fx 2 (x, y) = (y − 1)2 · fu2 (u(x, y), v(x, y)) + 2(y − 1) arctan y · fuv (u(x, y), v(x, y))
+ arctan2 y · fv 2 (u(x, y), v(x, y)),
Fxy (x, y) = (x + 1)(y − 1) · fu2 (u(x, y), v(x, y))

x(y − 1)
+ + (x + 1) arctan y · fuv (u(x, y), v(x, y))
1 + y2
x arctan y
+ · fv 2 (u(x, y), v(x, y))
1 + y2
1
+fu (u(x, y), v(x, y)) + · fv (u(x, y), v(x, y)),
1 + y2
2x(x + 1)
Fy 2 (x, y) = (x + 1)2 · fu2 (u(x, y), v(x, y)) + · fuv (u(x, y), v(x, y))
1 + y2
x2 2xy
+ · fv 2 (u(x, y), v(x, y)) − · fv (u(x, y), v(x, y)).
(1 + y )
2 2 (1 + y 2 )2
(b) Let u(x, y) = sin(x −y) and v(x, y) = xy −ln(y 2 +1). We can write F (x, y) =
f (u(x, y), v(x, y)) and it follows that
Fx (x, y) = cos(x − y) · fu (u(x, y), v(x, y)) + y · fv (u(x, y), v(x, y)),

2y
Fy (x, y) = − cos(x − y) · fu (u(x, y), v(x, y)) + x − 2 · fv (u(x, y), v(x, y)),
y +1
Fx 2 (x, y) = cos2 (x − y) · fu2 (u(x, y), v(x, y)) + 2y cos(x − y) · fuv (u(x, y), v(x, y))
+y 2 · fv 2 (u(x, y), v(x, y)) − sin(x − y) · fu (u(x, y), v(x, y)),
Fxy (x, y) = − cos2 (x − y) · fu2 (u(x, y), v(x, y))

2y
+ x−y− 2 cos(x − y) · fuv (u(x, y), v(x, y)
y +1

2y
+y x − 2 · fv 2 (u(x, y), v(x, y))
y +1
+ sin(x − y) · fu (u(x, y), v(x, y)) + fv (u(x, y), v(x, y)),
Fy 2 (x, y) = cos2 (x − y) · fu2 (u(x, y), v(x, y))

2y
−2 x − 2 cos(x − y) · fuv (u(x, y), v(x, y))
y +1
2
2y
+ x− 2 · fv 2 (u(x, y), v(x, y))
y +1
2(y 2 − 1)
− sin(x − y) · fu (u(x, y), v(x, y)) + · fv (u(x, y), v(x, y)).
(y 2 + 1)2
5.34. To simplify the writing, we drop out the argument of the functions that appears
in the calculations.
(a) Let u(x, y, z) = x − yz and v(x, y, z) = x − y. Then
F (x, y, z) = f (u(x, y, z), v(x, y, z))
and we have that
Fy 2 = z2 fu2 + 2zfuv + fv 2 ,
Fyz = yzfu2 + yfuv − fu ,

Fxyz = yzfu3 + y(z + 1)fu2 v + yfuv 2 − fu2 − fuv .
(b) Let u(x, y, z) = x − z and v(x, y, z) = y − z. Then
F (x, y, z) = f (u(x, y, z), v(x, y, z))
and we have that
Fxz = −fu2 − fuv ,

Fyz = −fuv − fv 2 ,
Fxyz = −fu2 v − fuv 2 ,
Fxz2 = fu3 + 2fu2 v + fuv 2 .
(c) Let u(x, y, z) = xy − z and v(x, y, z) = xz − y. Then
F (x, y, z) = f (u(x, y, z), v(x, y, z))
and we have that
Fxy = xyfu2 + (xz − y)fuv − zfv 2 + fu ,
Fyz = −xfu2 + (x 2 + 1)fuv − xfv 2 ,
Fxyz = −xyfu3 + (x 2 y − xz + y)fu2 v + (x 2 z − xy + z)fuv 2

−xzfv 3 − fu2 + 2xfuv − fv 2 ,
Fy 2 z = −x 2 fu3 + x(x 2 + 2)fu2 v − (2x 2 + 1)fuv 2 + xfv 3 .
(d) Let u(x, y, z, t) = x − t and v(x, y, z, t) = yz − t. Then
F (x, y, z, t) = f (u(x, y, z, t), v(x, y, z, t))
and we have that
Fyt = −zfuv − zfv 2 ,

Fzt = −yfuv − yfv 2 ,
Fxyz = yzfuv 2 + fuv ,
Fxyt = −zfu2 v − zfuv 2 ,
Fxzt = −yfu2 v − yfuv 2 ,
Fxt 2 = fu3 + 2fu2 v + fuv 2 ,
IV
Fxyzt = −yzfuIV2 v 2 − yzfuv
IV
3 − fu2 v − fuv 2 .
5.35. To simplify the writing, we drop out the argument of the functions that appears
in the calculations.
(a) Let u(x, y, z) = x − y, v(x, y, z) = y − z and w(x, y, z) = xy − z. Then
F (x, y, z) = f (u(x, y, z), v(x, y, z), w(x, y, z))
and we have that
Fxy = −fu2 + xyfw2 + fuv + (x − y)fuw + yfvw + fw ,

Fxz = −yfw2 − fuv − fuw − yfvw ,
Fyz = −fv 2 − xfw2 + fuv + fuw − (x + 1)fvw ,
Fxyz = −xyfw3 + fu2 v + fu2 w − fuv 2 − (x − y + 1)fuvw − (x − y)fuw2
−yfv 2 w − (x + 1)yfvw2 − fvw − fw2 ,
Fyz2 = fv 3 + xfw3 − fuv 2 − 2fuvw − fuw2 + (x + 2)fv 2 w + (2x + 1)fvw2 .
(b) Let u(x, y, z) = y, v(x, y, z) = x − z and w(x, y, z) = x − yz. Then
and we have that

Fxy = −zfw2 + fuv + fuw − zfvw ,

Fxz = −fv 2 − yfw2 − (y + 1)fvw ,
Fyz = yzfw2 − fuv − yfuw + zfvw − fw ,
Fz2 = fv 2 + y 2 fw2 + 2yfvw ,
Fxy 2 = z2 fw3 + fu2 v + fu2 w − 2zfuvw − 2zfuw2 + z2 fvw2 ,
Fxyz = yzfw3 − fuv 2 − (y + 1)fuvw − yfuw2 + zfv 2 w + (y + 1)zfvw2

−fvw − fw2 ,
Fy 2 z = −yz2 fw3 − fu2 v − yfu2 w + 2zfuvw + 2yzfuw2 − z2 fvw2
+2zfw2 − 2fuw .
(c) Let u(x, y, z) = z, v(x, y, z) = x − y − z and w(x, y, z) = x. Then
and we have that
Fxy = −fv 2 − fvw ,

Fxz = −fv 2 + fuv + fuw − fvw ,
Fyz = fv 2 − fuv ,
Fz2 = fu2 + fv 2 − 2fuv ,
Fxyz = fv 3 − fuv 2 − fuvw + fv 2 w ,
Fxz2 = fv 3 + fu2 v + fu2 w − 2fuv 2 − 2fuvw + fv 2 w ,
3 = −fv 4 − fv 3 w .
IV IV IV
Fxy
(d) Let u(x, y, z, t) = xt, v(x, y, z, t) = y − z and w(x, y, z, t) = x − t. Then
F (x, y, z, t) = f (u(x, y, z, t), v(x, y, z, t), w(x, y, z, t))
and we have that
Fxy = tfuv + fvw ,

Fyt = xfuv − fvw ,
Fzt = −xfuv + fvw ,
Ft 2 = x 2 fu2 + fw2 − 2xfuw ,

Fxyz = −tfuv 2 − fv 2 w ,
Fxyt = xtfu2 v + (x − t)fuvw − fvw2 + fuv ,
Fyzt = −xfuv 2 + fv 2 w ,
Ft 3 = x 3 fu3 − fw3 − 3x 2 fu2 w + 3xfuw2 ,

IV
Fxyzt = −xtfuIV2 v 2 − (x − t)fuv
IV
2 w + fv 2 w 2 − fuv 2 .
IV
5.36. a − b = 0.
5.37. a + b = 0.
13.3 Homogeneous Functions. Euler’s Identity
5.38. Homogeneous functions of degree. (a) − 12 ; (b) −1; (c) 1; (d) −2; (e) 1.
13.4 Taylor’s Formula for Real Functions of Two Real Variables
5.40.
(a) f (x, y) = xy 2 − 2xy + 2y 2 + x − 4y + 2;

(b) f (x, y) = x 2 y + x 2 − 2xy − 2x + y + 1.
5.41. f (x, y) = (x −1)3 +(x −1)2 (y +1)+3(x −1)2 −(x −1)(y +1)+3(x −1)+7.
5.42. f (x, y) = (x + 1)2 (y − 2) + 4(y − 2)3 + 2(x + 1)(y − 2) + 4(y − 2)2 + 8.
5.43. (a) (M2 f )(x, y) = xy;
(b) (M3 f )(x, y) = 1 + x + 2! (x − y ) + 3! (x
1 2 2 1 3 − 3xy 2 );
2 2
(c) (M4 f )(x, y) = 1 + xy + x 2y .
5.44. We have
(a) (T2 f )(x, y) = 1 + [(x − 1) + (y + 1)]

+ 2!1 3(x − 1)2 + 4(x − 1)(y + 1) + (y + 1)2 ;

(b) (T2 f )(x, y) = 1 + 2!1 −(x − 1)2 − 2(x − 1)(y − 1) − (y − 1)2 .
√ √
5.45. Use the function f (x, y) = x 3 y and the point (x0 , y0 ) = (1, 1).
5.46. Use the function f (x, y) = x y and the point (x0 , y0 ) = (1, 3).
5.47. The coefficient of x n y n is 5n .
5.48. The coefficient of (x − 1)n (y − 1)n is Cn2n .
5.49. The coefficient of x n y n is Cn2n+1 . We have
∞
∞

1 1
f (x, y) = · = x n
(x + y) m
1−x 1−x−y
n=0 m=0
∞ ∞ m
= xn Ckm x m−k y k .
n=0 m=0 k=0
The coefficient of x n y n is Cnn + Cnn+1 + Cnn+2 + · · · + Cn2n . We observe that this

is the coefficient of x n from the expression (1 + x)n + (1 + x)n+1 + · · · + (1 + x)2n .
This sum can be written in the following way:
(1 + x)n + (1 + x)n+1 + · · · + (1 + x)2n

= (1 + x)n [1 + (1 + x) + (1 + x)2 + · · · + (1 + x)n ]
(1 + x)n+1 − 1
= (1 + x)n ·
x
1
= [(1 + x)2n+1 − (1 + x)n ].
x
Therefore, the coefficient we are looking for is the coefficient of x n+1 from the
expression
(1 + x)2n+1 − (1 + x)n ,
which is Cn+1
2n+1 = C2n+1 .
n
13.5 The Differential of Several Real Variable Functions
5.50. df (0, 1)(1, 2) = 2e + 1 and d2 f (0, 1)(1, 2) = 6e + 8.

5.51. df (1, 0)(2, 1) = 4e + 1 and d2 f (1, 0)(2, 1) = 22e + 4.
π −1, 1)(−1,
5.52. df (π, 2, 1) = −1 − π andd2 f (π,−1, 1)(−1, 2, 1) = 2 − 4π .
5.53. df 2 , 1, 0 (1, −2, 2) = −3 and d2 f π2 , 1, 0 (1, −2, 2) = 8.
5.54. d4 f (2, 1, 2)(−1, 1, 1) = 52 .
5.55. −28.
5.56. v = (1, −2, 2), df df
dv (1, 1, −1) = 1 and dv (−1, 1, −1) = 0.
13.6 Extrema of Several Real Variable Functions

√ √ √
5.57. We have x + 1 ≥ 2 x, y + 1 ≥ 2 y and x + y ≥ 2 xy. It follows that
xy 1
f (x, y) ≤ √ √ √ = = f (1, 1), ∀(x, y) ∈ (0, ∞)2 .
2 x · 2 y · 2 xy 8
Therefore, (1, 1) is a global maximum point of f .

5.58.
√ (a)√(a,a), a > 0, points of global minimum; (b) (0, 0) point of local
√ √ minimum,

3
3
, − 3
3
and − 3
3
, 3
3
saddle points; (c) (0, 0) saddle point, 1
2 , 1 point of

global maximum; (d) (−2, 1) point of local maximum; (e) 2π 3 , 2π
3 point of
local minimum; (f) (a, a, a), a > 0, points of global minimum; (g) (−1, 0, 0)
point of local minimum; (h) (1, 1, 1) point of global minimum; (i) (a, 0, a + 1),
a ∈ R, points of global minimum; (j) (0, 0, 0) saddle point, − π4 , − π4 , − π4 and
π π π
4 , 4 , 4 points of local maximum.
6 If 6
5.59. k ≤ 0 or l ≤ 0, then f does not have critical points. If k > 0 and l > 0, then
3 k2 3 l2
l , k is a point of global minimum.
5.60.
1
(a) Solution I. The minimum value is 180 . We have that
1 b2 2a b 1
(x 2 − a − bx)2 dx = a 2 + − − + ab + .
0 3 3 2 5
2
Let f (a, b) = a 2 + b3 − 2a − b + ab + 15 . A calculation shows that the
3 2
only critical point of f is − 16 , 1 and Sylvester’s criterion implies that
this is a local minimum point of f . We prove that this point is a global
minimum point of f , i.e., f (a, b) ≥ f − 16 , 1 , ∀(a, b) ∈ R2 . To prove
that the previous inequality holds we observe, since f is a polynomial of
degree 2 in two variables, that f coincideswith its Taylor’s polynomial
of degree 2 corresponding to f at the point − 16 , 1 . A calculation shows
that

1 2 1 1 1
f (a, b) = (T2 f )(a, b) = a + + a+ (b − 1) + (b − 1)2 +
6 6 3 180

1 1
≥ f − ,1 = , ∀(a, b) ∈ R2 .
6 180
The last inequality follows since x 2 + xy + 13 y 2 ≥ 0, for all x, y ∈ R.

Solution II. We consider the vector space C([0, 1], R), of continuous
real valued functions on the interval [0, 1], together with the inner product
1
defined by f, g = 0 f (x)g(x)dx.
(continued)
1
Let V = span {1, x} and observe that min (x 2 − a − bx)2 dx is the
a, b ∈ R 0
square of the distance, in C([0, 1], R), from the vector x 2 to the subspace
V . It follows that
1 G(1, x, x 2 ) 1
min (x 2 − a − bx)2 dx = d 2 (x 2 , V ) = = ,
a, b ∈ R 0 G(1, x) 180
where G(v1 , v2 , . . . , vn ) denotes the Gram determinant of vectors

v1 , v2 , . . . , vn .
8
(b) 175 .

5.61. max f (C) = f 2, 32 = 3.
5.62. min f (C) = f (4, 1) = 5.
5.63. Let L : R2 ×R → R, L(x, y, λ) = x2 +2y 2 +λ(2x +4y −3) be the Lagrange

function. The only critical point of L is 2 , 2 , − 2 . We obtain that 12 , 12 is a
1 1 1
conditional local minimum point of f .

5.64. Since C is compact and f is continuous, it follows, based on the Weierstrass
theorem, that f |C is bounded and it attains its minimum and maximum values on
C. The points where f |C attains its minimum and maximum values are located
between the conditional critical points of f relative to the set C.
Let L : R2 × R → R, L(x, y, λ) = x − y + λ(x 2 + y 2 − 2) be the
Lagrange function. The critical points of L are −1, 1, 12 and 1, −1, − 12 .
Since f (−1, 1) = −2 and f (1, −1) = 2, it follows that min f (C) = −2 and
max f (C) = 2.
5.65. min f (C) = − 12 and max f (C) = 12 .
5.66. Let L : R2 ×R → R, L(x, y, λ) = xy+λ(x+2y−4) be the Lagrange function.
The only critical point of L is (2, 1, −1). We obtain that (2, 1) is a conditional local
maximum point of f .
5.69. The set C is the triangular surface with vertices (π, 0, 0), (0, π, 0), and
(0, 0, π ). Since C is a compact set and f is continuous, it follows, based on the
Weierstrass theorem, that the function f |C is bounded and it attains its bounds on
C. The points where f |C attains its bounds are located between:
– The conditional critical points of f relative to the set
C0 = (x, y, z) ∈ (0, ∞)3 | x + y + z = π
(the interior of the triangular surface);

C1 = {(x, y, z)| x = 0, y, z > 0, y + z = π}
(the side of the triangle, without endpoints, situated in the plane yOz);
C2 = {(x, y, z)| y = 0, x, z > 0, x + z = π }
(the side of the triangle, without the endpoints, situated in the plane xOz);
C3 = {(x, y, z)| z = 0, x, y > 0, x + y = π}
(the side of triangle, without the endpoints, situated in the plane xOy);
– The points (π, 0, 0), (0, π, 0), (0, 0, π ) (the triangle vertices).
To determine the conditional critical points of f relative to the set C0 , we

consider the Lagrange function L : (0, ∞)3 × R → R, L(x, y, z, λ) = sin x +
sin y + sin z + λ(x + y + z − π ). The only critical point of L is π3 , π3 , π3 , − 12 .
consider the function g(y) = f (0, y, π − y) = 2 sin y, whose only critical point
is y = π2 . Therefore, 0, π2 , π2 is the conditional critical point of f relative to the
set C1 . Similarly, one can obtain the conditional critical points of f relative to √the

other two sides of the triangle: π2 , 0, π2 , π2 , π2 , 0 . Since f π3 , π3 , π3 = 3 2 3 ,
π π π
f 0, 2 , 2 = f 2 , 0, π2 = f π2 , π2 , 0 = 2, and f (π, 0, 0) = f (0, π, 0) =
√
f (0, 0, π ) = 0, it follows that min f (C) = 0 and max f (C) = 3 3
2 .
Remark. In any triangle ABC the following inequality holds true:

√
3 3
sin A + sin B + sin C ≤ .
2
The equality is valid if and only if the triangle ABC is equilateral.
5.70. Let L : R3 × R → R, L(x, y, z, λ) = x + y 2 + z3 + λ(x + 2y + 3z − a)

be the Lagrange function. The critical points of L are (a − 5, 1, 1, −1) and (a +
1, 1, −1, −1). We obtain that (a − 5, 1, 1) is a conditional local minimum point of
f , and (a + 1, 1, −1) is not a conditional local extremum point of f .
5.71. Let L : (0, ∞)3 × R → R, L(x, y, z, λ) = xyz + λ(x + y3 + 2z − a) be the
2
Lagrange function. The only critical point of L is a4 , a4 , a4 , − a43 . We obtain that

a a a
4 , 4 , 4 is a conditional local maximum point of f .
5.72. Let L : R3 × R → R, L(x, y, z, λ) = x(y + z) + 3yz + λ(xyz − 3) be
the Lagrange function. The only critical point of L is (3, 1, 1, −2). We obtain that
(3, 1, 1) is a conditional local minimum point of f .
5.73. Since C is compact and f is continuous, it follows, based on the Weierstrass
theorem, that f |C is bounded and attains its bounds on C.
The points where f |C attains its bounds are located between the conditional
critical points of f relative to the set C.
Let L : R3 × R2 → R, L(x, y, z, λ, μ) = x + y + z + λ(x − 2y+ z − 2) + μ(x 2 +
y 2 +z2 −1) be the Lagrange function. The critical points of L are 0, −1, 0, −1, 32

and 23 , − 13 , 23 , 1, − 32 . Since f (0, −1, 0) = −1 and f 23 , − 13 , 23 = 1, it follows
that min f (C) = −1 and max f (C) = 1.
5.74. Let L : R3 × R2 → R, L(x, y, z, λ, μ) = xyz + λ(x + y + z −
4) + μ(xy + yz + xz − 5) be the Lagrange function. The critical points
of L are (1, 1, 2, 1, −1), (1, 2, 1, 1, −1), (2, 1, 1, 1, −1), 53 , 53 , 23 , 25 9 , − 5
3 ,

3, 3, 3, 9 , −3 , 3 , 3 , 3 , 9 , − 3 . We obtain that (1, 1, 2), (1, 2, 1), and
5 2 5 25 5 2 5 5 25 5

(2, 1, 1) are the conditional local maximum points of f , and 53 , 53 , 23 , 53 , 23 , 53 ,

and 23 , 53 , 53 are the conditional local minimum points of f .
Remark. The set C is bounded, since
x 2 + y 2 + z2 = (x + y + z)2 − 2(xy + yz + xz) = 6,
for all (x, y, z) ∈ C, and it is closed, being the intersection of closed sets. Thus, C,
which is a circle in space, is compact and f is continuous and it follows, based on
the Weierstrass theorem, that f is bounded and attains its bounds on C. The points
where f |C attains its bounds are located between the conditional critical points of
f relative to the set C. Since
f (1, 1, 2) = f (1, 2, 1) = f (2, 1, 1) = 2,

5 5 2 5 2 5 2 5 5 50
f , , =f , , =f , , = ,
3 3 3 3 3 3 3 3 3 27
it follows that (1,1, 2), (1, 2,

1), and (2, 1, 1) are conditional
global maximum
points of f , and 53 , 53 , 23 , 53 , 23 , 53 , and 23 , 53 , 53 are the conditional global
minimum points of f .
5.75. C is compact and f is continuous. We have, based on the Weierstrass theorem,

that f |C is bounded and attains its bounds on C. The points where f |C attains its
bounds are located between:
– The critical points of f situated in the interior of C;

Csphere = (x, y, z) ∈ R3 | x 2 + y 2 + z2 = 1 .
Since f does not have critical points, we determine the conditional critical points of
f relative to the set Csphere .
Let L : R3 × R → R, L(x, y, z, λ) = 1 − x − y − z + λ(x 2 + y 2 + z2 − 1) be
the Lagrange function. The critical points of L are
√ √ √ √ √ √ √ √
3 3 3 3 3 3 3 3
, , , and − ,− ,− ,− .
3 3 3 2 3 3 3 2
√ √ √ √ √ √ √ √
Since f 3 3
3 , 3 , 3
3
= 1 − 3 and f − 33 , − 33 , − 33 = 1 + 3, we have
√ √
that min f (C) = 1 − 3 and max f (C) = 1 + 3.
5.76. The set C is compact and the function f is continuous. We have, based on the
Weierstrass theorem, that f |C is bounded and attains its bounds on C. The points
where f |C attains its bounds are located between:
– The critical points of f situated in the interior of C;

– The conditional
2 critical points of f relative to 3the set
C1 = (x, y, z) ∈ R3 | z = x 2 + y 2 , z < 1 ;
– The conditional
2 critical points of f relative to the
3 set
C2 = (x, y, z) ∈ R3 | x 2 + y 2 < 1, z = 1 ;
– The conditional
2 critical points of f relative to the
3 set
C3 = (x, y, z) ∈ R3 | x 2 + y 2 = 1, z = 1 .
The only critical point of f is (0, 0, 0), which is not located in the interior of C.
To determine the conditional critical point of f relative to the set C1 , we consider
the function g(x, y) = f (x, y, x +y ) = x + x y + xy + y + 2xy. The critical
2 2 3 2 2 3
points of g are (0, 0) and − 3 , − 3 . It follows that the conditional critical points
1 1

of f relative to the set C1 are (0, 0, 0) and − 13 , − 13 , 29 .
consider the function h(x, y) = f (x, y, 1) = 2xy + x + y. The only critical point
2 2
of h is − 12 , − 12 . We have − 12 + − 12 < 1 and it follows that − 12 , − 12 , 1
is the unique conditional critical point of f relative to the set C2 .
consider the Lagrange function L(x, √ y,√λ) = 2xy + x + y +√λ(x 2 + y√2
√ √
−
1). The critical points of L are 2 , 2 , −1 − 2 , − 2 , − 2 , −1 + 2 ,
2 2 2 2 2 2
√ √ √ √
4 , − 4 , 1, 1 , and − 4 , 4 , 1, 1 , which implies that the condi-
7−1 7+1 7+1 7−1
√ √ √ √
tional critical points of f relative to the set C3 are 22 , 22 , 1 , − 22 , − 22 , 1 ,
√ √ √ √
7−1
4 , − 7+1
4 , 1 , and − 7+1
4 , 7−1
4 , 1 .
Since

1 1 2 2 1 1 1
f (0, 0, 0) = 0, f − , − , = , f − ,− ,1 = − ,
3 3 9 27 2 2 2
√ √ √ √
2 2 √ 2 2 √
f , , 1 = 1 + 2, f − ,− , 1 = 1 − 2,
2 2 2 2
√ √ √ √
7−1 7+1 5 7+1 7−1 5
f ,− ,1 = − and f − , ,1 = − ,
4 4 4 4 4 4
√
it follows that min f (C) = − 54 and max f (C) = 1 + 2.
5.77. [132] The minimum value of the expression x + yz is 50.
Let a = x + yz. We analyze the following cases:
y=1 Then z = 160 − x and a = x + 160 − x = 160.

y=2 Because 2x + z = 160, it follows that x ≤ 79. Then
2 · 160 − a
a = x + 2(160 − 2x) ⇒ x = ≤ 79 ⇒ a ≥ 83.
3
For x = 79 and z = 2, we have a = 83.
y=3 Because 3x + z = 160, it follows that x ≤ 53. From a = x + 3(160 − 3x)
we obtain that a ≥ 56. For x = 53 and z = 1 we have a = 56.
we obtain that a ≥ 50. For x = 26 and z = 4 we have a = 50 .

We also analyze the following cases:
z=1 Then xy = 159 = 3 · 53. Therefore, we can have
x = 1, y = 159 ⇒ a = 160,
x = 3, y = 53 ⇒ a = 56,
x = 53, y =3 was analyzed,
x = 159, y =1 was analyzed.
z=2 Then xy = 158 = 2 · 79. Therefore, we can have
x = 1, y = 158 ⇒ a = 517,
x = 2, y = 79 ⇒ a = 160,
z=3 Then xy = 157. Therefore, we can have
x = 1, y = 157 ⇒ a = 472,
x = 157, y = 1 was analyzed.
z=4 Then xy = 156 = 22 · 3 · 13. Therefore, we can have
x = 1, y = 156 ⇒ a = 625,
x = 2, y = 78 ⇒ a = 314,
x = 3, y = 52 ⇒ a = 211,
x = 4, y = 39 ⇒ a = 160,
x = 6, y = 26 ⇒ a = 110,
x = 12, y = 13 ⇒ a = 64,
x = 13, y = 12 ⇒ a = 61,

If y ≥ 10 and z ≥ 5, then a = x + yz > 50. Therefore, the minimum value of

the expression x + yz is 50.
5.78. Solution I. Since 2(x 2 + y 2 ) ≥ (x + y)2 , for all x, y ∈ R, we have that
/ / |x + y| |x + y|
/ 2/
/(x + y)e−x −y / ≤ (x+y)2 = |x+y|2 .
2
e 2 e 2
√ √
Using the substitution |x + y| = t 2, we have to prove that 2te−t ≤ √1e ,
2
∀t ≥ 0.
Solution II. We determine the global extremum values of the function f : R2 →
R, f (x, y) = (x + y)e−x −y . We consider the compact set
2 2
D(0, r) = (x, y) ∈ R2 | x 2 + y 2 ≤ r 2 , r > 0,
and we determine the extremum values of f on D(0, r). We have
fx (x, y) = e−x
2 −y 2
fy (x, y) = e−x
2 −y 2
(1 − 2x(x + y)) and (1 − 2y(x + y)).

The critical points of f are 1 1
2, 2 and − 12 , − 12 , which are local extremum
points of f and

1 1 1 1 1 1
f , =√ , f − ,− = −√ .
2 2 e 2 2 e
We determine the extremum values of f on the boundary of D(0, r). Let
L(x, y, λ) = (x + y)−(x
2 +y 2 )
− λ(x 2 + y 2 − r 2 ), λ ∈ R.
The conditional critical points can be obtained by solving the system Lx = 0, Ly =

√
0, x 2 + y 2 = r 2 . It follows that x = y = ± r 2 2 . Since
√ √
r 2 r 2 √ 1
= r 2e−r ≤ √ ,
2
f ,
2 2 e
it follows that max(x,y)∈R2 |f (x, y)| = √1 .

e
Implicit Functions
14
14.1 Implicit Functions of One Real Variable Defined by an

Equation
6.1. y (0) = 1, y (0) = 12 , y (0) = 11 4 .

6.2. y (0) = −π 2 , y (0) = 4π 3 .
6.3. y (0) = 0, y (0) = − 23 , y (0) = − 23 .
6.4. y (0) = 0, y (0) = π + 1.
6.5. y (0) = 2, y (0) = 4.
6.6. y (1) = 1, y (1) = 0.
√
6.7. Let (x = 0. We have x0 +
√0 , y0 ) be an arbitrary point on the√curve,√with x0 y0 √
√
y0 = c. We differentiate the equation x + y(x) = c with respect to x,
6
and we obtain that y (x) = − y(x) x . The equation of the tangent to the curve at the
point (x0 , y0 ) is given by
7
y0
y − y0 = − (x − x0 ).
x0
A calculation shows that the lengths of the segments are

√ √
x = x0 + x0 y0 and y = y0 + x0 y0 .
√ √ √
It follows that x + y = x0 + 2 x0 y0 + y0 = ( x0 + y0 )2 = c.
When x0 = c and y0 = 0, we have y− (c) = 0. In this case x = c and y = 0,
which implies x + y = c.
When x0 = 0 and y0 = c, we have y+ (0) = −∞. In this case x = 0 and y = c,
x + y = c.
which implies
6.8. y 3a2 = −1, y 3a 2 = − 3a
32
, y 3a
2 = − 3a
512
2.
√ √ √
6.9. y a 2
= −1, y a 2
= −6a2.
√2 2
a 3 a
6.10. 2 ,2 .
6.13.
(a) y (1) = − 43 , y (1) = − 68

9 ;
(b) y (−1) = 3 , y (−1) = − 68
4
9 .
6.14. y (1) = −1, y (1) = −4.

Equation
6.15. zx (−1, 1) = 0, zy (−1, 1) = 1, zx 2 (−1, 1) = 0, zxy (−1, 1) = 1,

zy 2 (−1, 1) = −1.
6.16. zx (1, −1) = 12 , zy (1, −1) = −1, zx 2 (1, −1) = − 11
8 , zxy (1, −1) = 2 ,
1
zy 2 (1, −1) = − 52 .
6.17. zx (−1, 2) = −4, zy (−1, 2) = −1, zx 2 (−1, 2) = −12, zxy (−1, 2) = 2,
zy 2 (−1, 2) = 2.

System of Equations
6.18. y (1) = −2, z (1) = −3, y (1) = −9, z (1) = 2.

6.19. y (1) = 4, z (1) = −14, y (1) = −32, z (1) = 293.
6.20. y (−1) = −3, z (−1) = −3, y (−1) = 16, z (−1) = 1.

System of Equations
6.21. ux (1, −1) = −1, uy (1, −1) = 1, vx (1, −1) = −2π +2, vy (1, −1) = π −2.
6.22. ux (−1, 2) = 1, uy (−1, 2) = 0, vx (−1, 2) = 3, vy (−1, 2) = 1.
Challenges, Gems, and Mathematical
Beauties 15
7.1. (a) x 2 ; (b) xy; (c) 0.

7.2. The sequence diverges.
7.3. k k . Use Stirling’s formula.
7.4. 12 . Use that lim n (Hn − ln n − γ ) = (see Problem 1.31).1
2
n→∞ √
7.5. (b) Use the inequality in part (a) with x = n k, k = 1, 2, . . . , n.
7.6. ea−2b .
7.7. 1.
7.8.
√
= 2n sin 1 + 2n sin 2 + · · · + 2n sin n . Since 2n sin i ≤
n
(a) The limit equals 2. Let xn √
2n , we get that xn ≤ 2 n n. Fix i = 1, . . . , n. We have 2sin i ≤ xn and
it follows that 2sin i ≤ lim infn→∞ xn ≤ lim supn→∞ xn ≤ 2. Therefore,
2 = lim supi→∞ 2sin i ≤ lim infn→∞ xn ≤ lim supn→∞ xn ≤ 2.
(b) The limit equals 2sup xn .
√ √
1 1 a+ a 2 −1 a 2 +b2
7.9. (a) (b)
2; 16 ; (c) 2 ; (d) 4 .
7.10. We have
ln n ln n
1+ 1
3 + ··· + 1
2n−1 H2n − 12 Hn
lim = lim
n→∞ 1
2 + 4 + · · · + 2n
1 1 n→∞ 1
2 Hn
ln n
H2n
= lim 2 −1
n→∞ Hn
ln n
H2n
= lim 1 + 2 −1
n→∞ Hn

H2n
2 lim ln n Hn −1
=e n→∞
ln n
2 lim (H2n −Hn )
=e n→∞ Hn
= 4,
since lim ln n
= 1 and lim (H2n − Hn ) = ln 2.
n→∞ Hn n→∞
7.12. A. (a), (c), and (d) Use Taylor’s formula. B. (a) Use Cauchy–d’Alembert’s
criterion. A. (b) and B. (b) use the technique for calculating limits of sequences of
indeterminate form 1∞ . In our case, we have the following. n
Let (xn )n∈N , xn ∈ R, ∀n ∈ N, be such that lim xn = x and lim xxn = L. If
n→∞ n→∞
lim n(xn − x) exists, then lim n(xn − x) = x ln L.
n→∞ n→∞
A. (e) The challenge: Use a similar technique to that given in the solution of
problem 1.14 in [26].
7.13. We use Abel’s summation formula, with ak = 1 and bk = k12 + (k+1)
1
2 + ···,
and we have that
n n
1 1 1 1 1
+ + ··· =n + + ··· + .
k 2 (k + 1)2 (n + 1)2 (n + 2)2 k
k=1 k=1
It follows that
n n
1 1 1 1 1
+ + · · · −ln n = n + + ··· + −ln n
k2 (k + 1)2 (n + 1)2 (n + 2)2 k
k=1 k=1

and the result follows since lim n 1
(n+1)2
+ 1
(n+2)2
+ · · · = 1.
n→∞
7.14.
x n+1 −2x+1
(a) Let fn (x) = x n + x n−1 + · · · + x − 1 = Since fn (x) = nx n−1 +
.
x−1
· · ·+1 > 0, x ≥ 0, we have that fn strictly increases on (0, ∞). Since fn 12 =

− 21n < 0 and fn (1) = n − 1 > 0, n ≥ 2, we have that xn ∈ 12 , 1 .
(b) We have
xnn+2 − 2xn + 1 xn (2xn − 1) − 2xn + 1

fn+1 (xn ) = = = 2xn − 1 > 0,
xn − 1 xn − 1

since xn > 12 . Also, fn+1 12 = − 2n+11
< 0 and this implies xn+1 ∈ 12 , xn .

(c) We prove that xn ∈ 12 , 12 + 2n+1
1
. It suffices to show that fn 12 + 2n+1
1
> 0.
We have
n+1
1
1+ 1
−1
1 1 1 2 2n
fn + n+1 = · > 0,
2n − 1
2 2 2n−1 1
n+1 √
since 1 + 1
2n < 2, n ≥ 2. The preceding inequality is equivalent to n+1
2−
1 > 1
2n , ∀n ≥ 2. The inequality is verified for n = 2. Let n ≥ 3. We have,
1
based on Lagrange’s Mean Value Theorem, that 2 n+1 − 20 = ln 2 θn
n+1 2 , for some

θn ∈ 0, n+1
1
. It follows that
1 ln 2 θn ln 2 1
2 n+1 − 1 = 2 > > n, n ≥ 3,
n+1 n+1 2
since 2n ln 2 = 2n−1 ln 4 > 2n−1 ≥ n + 1, ∀n ≥ 3.

(d) We have

1
lim 2 xn −
n
= lim 2n−1 (2xn − 1) = lim 2n−1 xnn+1
n→∞ 2 n→∞ n→∞
1 1 lim (n+1)(2xn −1) 1

= lim (2xn )n+1 = en→∞ = ,
4 n→∞ 4 4
since (see the solution of part (c)) 0 < (n + 1) (2xn − 1) < n+1
2n .
7.15. Use that:

n2 1 1 2 1 1
(a) = + + + ;
k 2 (n − k)2 k2 (n − k)2 n k n−k

n3 1 1 3 1 1 6 1 1
(b) 3 = 3+ + + + 2 + .
k (n − k)3 k (n − k)3 n k2 (n − k)2 n k n−k
1 n+1
7.16. (a) e n+2 ; (b) e n+2 .
2 2
7.17. (a) − x2 ; (b) x2 f (0) − (f (0))2 ef (0)x .
7.18. (a) f (1)

2 ; (b) 3 (1 − ln 2)f (1).
2
7.19.
aqn+k+1 aqn+k+1+p · · · aqn+k+1+s(n−1)p
(i) Let xn = , n ∈ N.
aqn+k aqn+k+p · · · aqn+k+s(n−1)p
2 , for all n ∈ N. Hence
We have an an+2 < an+1
-
p−1 -
p−1 -
p−1
p−i p−i
-
p−1
2(p−i)
i i 2i
an+i−1 an+i+1 < an+i and an+i−1 an+i+1 < an+i ,
i=1 i=1 i=1 i=1
for all n ∈ N. Therefore,

p−1 p p−1 p
an an+p < an+p−1 and an an+p < an+1 , n ∈ N.
Taking into account these inequalities, we get
p p−1
aqn+k+1+s(n−1)p p aqn+k+1 aqn+k+1+s(n−1)p
p−1
< xn < p ,
aqn+k aqn+k+s(n−1)p aqn+k
p
for all n ∈ N\{1}. It follows, based on Squeeze Theorem, that lim xn = ps+q q .
6 n→∞
Therefore, lim xn = ps+q
p
n→∞ q .
(ii) Let yn = , n ∈ N. Using the limit from part
(n!)s
(i), it follows that
q
yn+1 p ps + q
lim = [(ps + q)r]s .
n→∞ yn q
Now, according to the Cauchy–d’Alembert criterion, we are able to write

q
√ yn+1 p ps + q
lim n
yn = lim = [(ps + q)r]s .
n→∞ n→∞ yn q

(iii) Let zn = , n ∈ N. We have
nns
zn+1 yn+1 1
= · 0 n 1s , n ∈ N.
zn yn 1 + n1
It follows that
q
s
zn+1 p ps + q (ps + q)r
lim = .
n→∞ zn q e
Now, according to the Cauchy–d’Alembert criterion, we are able to write

q
s
√ zn+1 p ps + q (ps + q)r
lim n
zn = lim = .
n→∞ n→∞ zn q e
7.20. We have
3
55n Cn2n [(5n + 5)(5n + 10) · · · (10n)]5
xn = =
C5n n 2n
10n C5n C4n
(5n + 1)(5n + 2) · · · (10n)

(5n + 5)(5n + 10) · · · (10n) 4 (5n + 4)(5n + 9) · · · (10n − 1) 3
= ·
(5n + 4)(5n + 9) · · · (10n − 1) (5n + 3)(5n + 8) · · · (10n − 2)

(5n + 3)(5n + 8) · · · (10n − 2) 2 (5n + 2)(5n + 7) · · · (10n − 3)
× · ,
(5n + 2)(5n + 7) · · · (10n − 3) (5n + 1)(5n + 6) · · · (10n − 4)
for all n ∈ N.
Choosing a = r = 1, p = 5, q = 5, and s = 1 in part (i) of Problem 7.19, we
obtain
(5n + k + 1)(5n + k + 6) · · · (10n + k − 4) √
5
lim = 2, k ∈ {1, 2, 3, 4}.
n→∞ (5n + k)(5n + k + 5) · · · (10n + k − 5)
So,
√
5
√
5
√
5
√
5
lim xn = ( 2)4 ( 2)3 ( 2)2 2 = 4.
n→∞

1
7.21. Having in view the recurrence relation nxn+1 = n + xn , for n ∈ N, we
p
can write that
-
n -
n
pj xj +1 = (pj + 1)xj
j =1 j =1
and
n n
pj xj +1 = [p(j − 1)xj + (p + 1)xj ],
j =1 j =1
for all n ∈ N. We get
(p + 1)(2p + 1) · · · (np + 1)
xn+1 = x1
p(2p) · · · (np)
and
pn
x1 + x2 + · · · + xn = xn+1 ,
p+1
for all n ∈ N. Therefore,

xαn + xαn+1 + · · · + xβn

pβn p(αn − 1)
= xβn+1 − xαn
p+1 p+1
pβn (p + 1)(2p + 1) · · · (βnp + 1) p(αn − 1)
= · x1 − xαn
p+1 p(2p) · · · (βnp) p+1
pβn (αnp + 1)((αn + 1)p + 1) · · · (βnp + 1) p(αn − 1)
= · xαn − xαn ,
p+1 (αnp)((αn + 1)p) · · · (βnp) p+1
for all n ∈ N.
The above relations, obtained for the sequence (xn )n∈N , hold for p = 1 too.
Hence
yn+1 = (n + 1)y1
and
(β 2 − α 2 )n + α + β
yαn + yαn+1 + · · · + yβn = yαn ,
2α
for all n ∈ N.
It follows that
yαn (xαn + xαn+1 + · · · + xβn )
xαn (yαn + yαn+1 + · · · + yβn )
2pαβn
=
(p + 1)[(β 2 − α 2 )n + α + β]
(αnp + 1)((αn + 1)p + 1) · · · (βnp − (β − α)p + 1)
×
(αnp)((αn + 1)p) · · · (βnp − (β − α)p)
(βnp − (β − α)p + p + 1) · · · (βnp + 1)
×
(βnp − (β − α)p + p) · · · (βnp)
2pα(αn − 1)
− ,
(p + 1)[(β 2 − α 2 )n + α + β]
for all n ∈ N.
Choosing a = r = 1, k = 0, q = αp, and s = β − α in part (i) of Problem 7.19,
we obtain
7
(αnp + 1)((αn + 1)p + 1) · · · (βnp − (β − α)p + 1) p β
lim = .
n→∞ (αnp)((αn + 1)p) · · · (βnp − (β − α)p) α
So,
6
β β p
yαn (xαn + xαn+1 + · · · + xβn ) 2p α −1 α
lim = · 2 .
n→∞ xαn (yαn + yαn+1 + · · · + yβn ) p+1 β
α − 1
7.22. Let (zn )n∈N be a sequence, with z1 = 0, defined by the recurrence relation
nzn+1 = (n + 1)zn , for n ∈ N. Using Problem 7.21, we can write that
yαn (xαn + xαn+1 + · · · + xβn )

lim
n→∞ xαn (yαn + yαn+1 + · · · + yβn )

zαn (xαn + xαn+1 + · · · + xβn ) yαn (zαn + zαn+1 + · · · + zβn )
= lim ·
n→∞ xαn (zαn + zαn+1 + · · · + zβn ) zαn (yαn + yαn+1 + · · · + yβn )
λ+1 2 λ+1
β λ β β λ
2λ α − 1 μ+1 α − 1 λ(μ + 1) α −1
= · 2 · · μ+1 = · μ+1 .
λ+1 β 2μ μ(λ + 1)
α −1 β μ
−1 β μ
−1
α α
6
p p b
7.23. We use Problem 7.21, and we obtain that the limit is p+1 b a −a .
(m)
7.24. For m ∈ {1, 2, . . . , p}, let (xn )n∈N be the sequence defined by
-
n−1
n + jp + m
xn(m) = .
n + jp + m − 1
j =0
Clearly, xn(1) = xn . As one can observe
(1)
-
m
n + (n + k − 1)p + m + 1
xn+m = xn(m+1) , m ∈ {1, 2, . . . , p − 1}.
n + (n + k − 1)p + m
k=1
Thus,
(1) (1) (p)

--
p−1 m
n + (n + k − 1)p + m + 1
xn(1) xn+1 · · · xn+p−1 = xn(1) xn(2) · · · xn .
n + (n + k − 1)p + m
m=1 k=1
(1) (2) (p)

However, the product xn xn · · · xn telescopes as
(p)
--
n−1 p
n + jp + m - n + (j + 1)p
n−1
xn(1) xn(2) · · · xn = = = p + 1.
n + jp + m − 1 n + jp
j =0 m=1 j =0
Therefore,
(1) (1)
--
p−1 m
n + (n + k − 1)p + m + 1
xn(1) xn+1 · · · xn+p−1 = (p + 1) .
n + (n + k − 1)p + m
m=1 k=1
√
It follows that ( lim xn )p = p + 1, and so lim xn = p
p + 1.
n→∞ n→∞
7.25. We have
∞ √
p ∞
n 1 1 1
lim p = lim √ √ − √
n→∞
k=1 j =1 k (n + k)p−j +1
p j n→∞ p np−1
k=1
p
k p
n+k
n
1 1
= lim √ √
n→∞ p np−1 p
k
k=1
n
1 1
= lim 6
n→∞ n p k
k=1 n
1 1
= √
p
dx
0 x
p
= .
p−1
7.26.
(a) The limit equals 1/4. It is known that lim Sn = ln(4/π ) (see [140]). We have,
n→∞
based on Stolz–Cesàro lemma, the 0/0 case, that
S2n − ln π4 S2n+2 − S2n

lim = lim
n→∞ 1
(2n)2
n→∞ 1
(2n+2)2
− 1
(2n)2
(2n+1)(2n+3)
1
2n+1 − 2n+2 + ln
1
(2n+2)2
= lim
n→∞ 1
(2n+2)2
− (2n)2
1
1
=− ,
4
where the last equality follows by l’Hôpital’s rule.
Also, an application of Stolz–Cesàro lemma, the 0/0 case, gives
S2n+1 − ln π4 S2n+3 − S2n+1

lim = lim
n→∞ 1
(2n+1)2
n→∞ 1
(2n+3)2
− 1
(2n+1)2
(2n+3) 2
− 2n+2
1
+ 1
2n+3 + ln (2n+2)(2n+4)
= lim
n→∞ 1
(2n+3)2
− 1
(2n+1)2
1
= .
4
Consequently,
/ /
/ / / 4/
/ 4/ /Sn − ln π/ 1
lim n2 //Sn − ln // = lim 1
= .
n→∞ π n→∞
2
4
n
5
(b) In a similar way as in part (a) we get that the limit is 12 .
7.27.
(a) The limit equals 1. Observe that
2 xn 1 xn 2 1
dx = dx + 1 − dx.
0 1 + xn 0 1 + xn 1 1 + xn
We have
1 xn 1 1
0< dx ≤ x n dx =
0 1 + xn 0 n+1
xn 1
and it follows that lim dx = 0.
n→∞ 0 1 + x n
n
On the other hand, 1 + x n ≥ 2x 2 , and we have that
2 1 2 1 1 n

dx ≤ dx = 1 − 21− 2 , n > 2,
1 + xn n−2
n
1 1 2x 2
1 2
which shows that lim dx = 0.
n→∞ 1 1 + x n
(b) First we observe, since lim f (x) exists, that f is bounded (prove it!). We have
x→∞
2 1 2
f (x n )dx = f (x n )dx + f (x n )dx.
0 0 1
An application of the Bounded Convergence Theorem shows that, the details are
1 2
left to the reader, lim f (x n )dx = f (0) and lim f (x n )dx = f (∞).
n→∞ 0 n→∞ 1
A challenge. Prove that the preceding two limits hold by using an −δ
argument.
7.28. Solution I. Let > 0. Since f is continuous at 1, we have that there exists
δ > 0 such that |f (x) − f (1)| < , for x ∈ (δ, 1]. We have
1 δ 1
x y x f (y)dy = x y x (f (y) − f (1))dy + x y x (f (y) − f (1))dy
a a δ
x
+ f (1)(1 − a x+1 ).
x+1
On the one hand,

/ /
/
/x
1 / x
/ y (f (y) − f (1))dy //≤
x
1 − δ x+1 ≤
δ x+1
which implies that

/ /
/ 1 /
lim / x y (f (y) − f (1))dy //≤ .
x
x→∞ / δ
1
Since > 0 was arbitrarily taken, we get that lim x δ y x (f (y) − f (1))dy = 0.
x→∞
On the other hand,
/ /
/ δ / 2x
/x y (f (y) − f (1))dy //≤
x
||f || δ x+1 − a x+1 → 0, x → ∞.
/ x+1
a
1
It follows, based on the previous calculations, that lim x a y x f (y)dy = f (1) and
x→∞
the problem is solved.
Solution II. Approximate f by a polynomial function.
1 3
1 4 4 1
x (t 2 −t+1)x dt = x (t 2 −t+1)x dt+x (t 2 −t+1)x dt+x (t 2 −t+1)x dt.
1 3
0 0 4 4
Using the substitution t 2 − t + 1 = y, we get that

1
4 1 yx
x (t 2 − t + 1)x dt = x √ dy
0 13
16
4y − 3
and it follows, based on Problem 7.28, that

1
4 1 yx
lim x (t 2 − t + 1)x dt = lim x √ dy = 1.
x→∞ 0 x→∞ 13
16
4y − 3
On the other hand,

3 1 3 x
4 2 4 13
x (t − t + 1) dt = x
2 x
(t − t + 1) dt + x
2 x
(t − t + 1) dt ≤ 2x
2 x
,
1
4
1
4
1
2
16
3
which implies that lim x 1
4
(t 2 − t + 1)x dt = 0.
x→∞ 4
The substitution t 2 − t + 1 = y shows that
1 1 yx
x (t 2 − t + 1)x dt = x √ dy
3
4
13
16
4y − 3

1 1 yx
lim x (t 2 − t + 1)x dt = lim x √ dy = 1.
x→∞ 3
4
x→∞ 13
16
4y − 3
7.30. Solution due to I. Gavrea. The limit equals 1. We have

1 3
1 2e 2e 1
x t dt = x
tx
t dt + x
tx
t tx dt + x t tx dt.
1 3
0 0 2e 2e
1 1
2e
Let α = 1
2e and let f (t) = t t , t ∈ 1
0, 2e , with f (0) = 1. Using the
substitution t t = y ⇒ dt = 1
y(ln f −1 (y)+1)
dy, we get that
1
2e 1 yx
x t tx dt = −x dy
0 α y(ln f −1 (y) + 1)

1
2e 1 yx 1
lim x t tx dt = − lim x dy = − = 0.
x→∞ 0 x→∞ α y(ln f −1 (y) + 1) 1 · (ln f −1 (1) + 1)
3
2e
Let β = 3
2e . We have
3 1 3
2e e 2e
x t dt = x
tx
t dt + x
tx
t tx dt ≤ x α x + β x
1 1 1
2e 2e e
3
and this implies that lim x 12e t tx dt = 0.
0
x→∞ 1 2e
Let g(t) = t t , t ∈ 2e
3
, 1 . Using the substitution t t = y ⇒ dt = 1
y(ln g −1 (y)+1)
dy,
we get that
1 1 yx
x t tx dt = x dy
3
2e β y(ln g −1 (y) + 1)

1 1 yx 1
lim x t tx dt = lim x −1
dy = −1
= 1.
x→∞ 3
2e
x→∞ β y(ln g (y) + 1) 1 · (ln g (1) + 1)
2
7.31. The limit equals ln a . We have
1 1 t (1−t)x
2 1
x a t (t−1)x
dt = 2x dt
0 0 a
1 yx
t (1−t)=y 4 1 dy
= 2x √
0 a 1 − 4y
x 1 xu
1−4y=u x 1 4 a4
= √ du
2 a 0 u
x 1
u=v 2 1 4 xv 2
= x a 4 dv
a 0
xv 2
√ x
4 =t x 4 at
= x √ dt.
a 4 0 t
It follows that
1 √ x
x 4 at
lim x a t (t−1)x
dt = lim x √ dt
x→∞ 0 x→∞ a 4 0 t
b at
x √ dt
4 =b 0 t
= 2 lim
b→∞ ab
√
b
∞
ab
√
l’Hôpital’s rule ∞ b
= 2 lim b √ b 1
b→∞ a ln a b−a · 2√b
b
√
b
= 2 lim √
b→∞ ln a b − 1
√
2 b
2
= .
ln a
b at
∞ at
When we applied l’Hôpital’s rule, we have used that lim 0
√
t
dt = 0
√
t
dt =
b→∞
∞.
∞ ∞ k+1
{x}n e−x dx = {x}n e−x dx
0 k=0 k
∞ k+1
= (x − k)n e−x dx
k=0 k
∞ 1
x−k=y
= y n e−(k+y) dy
k=0 0
∞ 1
= e−k y n e−y dy
k=0 0
e 1
= y n e−y dy.
e−1 0
It follows that
√
∞ n
e 1
{x}n e−x dx y n e−y dy = 1,
n
lim n
= lim √
n→∞ 0
n
n→∞ e−1 0
6
1
since lim n
0 y n e−y dy = 1. The preceding limit follows based on the inequalities
n→∞
1 1 1
≤ y n e−y dy ≤ .
e(n + 1) 0 n+1
7.33. The limit equals 1e . On the one hand,
∞ ∞
xn e−x dx < x n e−x dx = n!.
0 0
On the other hand,

n+1 n+1 ∞
nn 1
1− = nn e−x dx = xn e−x dx < xn e−x dx.
en e n n 0
These imply that

7 √
1n 1 1 n ∞ n
n!
1− < xn e−x dx < ,
e e n 0 n
√
n
and the result follows since lim n
n!
= 1e .
n→∞
7.34. The limit equals ln(a + b). Integrate by parts and observe that
1 √ n2 √ an 1 x n+1
n2 xn
n n
ax + b − 1 dx = a+b−1 − dx.
0 n+1 n + 1 0 n (ax + b)n−1
7.35. Using the substitution x n = y, we have that
1 √ 1 1
1
n2 ( 1 + x n − 1)dx = n e n ln(1+y) − 1 y n −1 dy
n
0 0
1 ln(1 + y) √
= n
y eθn (y) dy,
0 y

where θn (y) ∈ 0, n1 ln(1 + y) .
ln(1+y) √
Let fn : [0, 1] → R, fn (y) = y
n y eθn (y) . We have that lim fn (y) =
n→∞
ln(1+y)
y and |fn (y)| ≤ 2 ln(1+y)
which is integrable over [0, 1]. It follows, based on
y ,
Lebesgue Dominated Convergence Theorem, that
1 ln(1 + y) √ 1 ln(1 + y) π2
lim n
y eθn (y) dy = dy = .
n→∞ 0 y 0 y 12
Similarly, one can prove that

1 √
lim n2 1 − 1 − x n dx = ζ (2).
n
n→∞ 0
For a more√general problem, the reader is referred to [26,

6 problem 1.60, p. 11].
√ 6π √ π √
7.36. (a) a+ b 2 ; (b) af 2 + bf (0) π
2.
b
7.37. The limit equals a f (x)dx. Since f is Riemann integrable, we have that f is
bounded, i.e. there exists M > 0 such that |f (x)| ≤ M, ∀x ∈ [a, b]. We have
/ /
/ b b f (x) /
/ f (x)dx − dx /
/ 1 + cos x cos(x + 1) · · · cos(x + n) /
a a
b | cos x cos(x + 1) · · · cos(x + n)|
≤M dx
a |1 + cos x cos(x + 1) · · · cos(x + n)|
b | cos x cos(x + 1) · · · cos(x + n)|
≤M dx,
a 1 − | cos x cos(x + 1) · · · cos(x + n)|
since |1 + cos x cos(x + 1) · · · cos(x + n)| ≥ 1 − | cos x cos(x + 1) · · · cos(x + n)|.

An application of the AM − GM inequality shows that
n+1
cos2 x + cos2 (x + 1) + · · · + cos2 (x + n)
cos2 x cos2 (x + 1) · · · cos2 (x + n) ≤ .
n+1
A calculation shows
n + 1 sin(n + 1) cos(n + 2x)
cos2 x + cos2 (x + 1) + · · · + cos2 (x + n) = + ,
2 2 sin 1
and it follows that
n+1
1 1 2
| cos x cos(x + 1) · · · cos(x + n)| ≤ + .
2 2(n + 1) sin 1
Since the function h(x) = x

1−x , x ∈ [0, 1) is strictly increasing, we have that
/ /
/ b b f (x) /
/ f (x)dx − dx //
/
a a 1 + cos x cos(x + 1) · · · cos(x + n)
n+1
2
2 + 2(n+1) sin 1
1 1
≤ M · (b − a) · n+1 .
2
1 − 2 + 2(n+1) sin 1
1 1
Passing to the limit, as n → ∞, in the previous inequality, the result follows.

7.38.
(a) The limit equals (k), where denotes the Gamma function. See [108].
(b) The limit equals (k + 1), where denotes the Gamma function.
We have
∞ x k sin x ∞ x k sin x e−(n+1)x
nk+1 dx = nk+1 dx
0 e(n+1)x − enx 0 1 − e−x
1
(− ln t)k sin(ln t)t n
e−x =t
= −nk+1 dt
0 1−t
√ 1
t n =y
1
k+1 sin ln y
n
√
= (− ln y) · √ · n
y · n
√ dy.
0 ln n y 1− n y
√
sin(ln n y ) √ 1
Let fn (y) = (− ln y)k+1 · √
ln n y · n y· n√
1− n y , y ∈ (0, 1).
We have that lim fn (y) = (− ln y)k , y ∈ (0, 1), and
n→∞
/ √ /
/ k+1 sin ln y
n
√ 1 / (− ln y)k+1
/
|fn (y)| =/ (− ln y) · √ · n
y · n
√ /≤ ,
ln n y 1− n y / 1−y
/ / 1
since / sinx x / ≤ 1, ∀x ∈ R, and 1−n√n y ≤ 1−y , for all n ∈ N. The last inequality
1
follows from the fact that the function g : [0, 1] → R, g(x) = 1−y
x
x is an increasing
function when y ∈ (0, 1). Since
1 (− ln y)k+1 ∞ t k+1 e−t ∞ ∞
y=e−t
dy = dt = t k+1 e−t e−it dt
0 1−y 0 1 − e−t 0 i=0
∞ ∞ ∞ ∞
1
t k+1 e−(i+1)t dt x k+1 e−x dx
(i+1)t=x
= =
(i + 1)k+2 0
i=0 0 i=0
= (k + 2)ζ (k + 2),

k+1
we have that the positive function y → (− ln
1−y
y)
is integrable over [0, 1].
It follows, based on Lebesgue Dominated Convergence Theorem, that
∞
x k sin x
lim nk+1 dx
n→∞ 0 − enx e(n+1)x
√ 1
1 sin ln n y √
= lim (− ln y)k+1 · √ · n
y · n
√ dy
n→∞ 0 ln n y 1− n y
1
= (−1)k lnk y dy
0
∞
y=e−t
= t k e−t dt
0
= (k + 1).
y
7.39. Use the substitution x = sinh 2n and apply Lebesgue Dominated Convergence
Theorem.
7.40.
*n+
(a) The limit equals 2
π. Let k = π . This implies that kπ ≤ n < (k + 1)π . We
have
n kπ n
| sin x|dx = | sin x|dx + | sin x|dx
0 0 kπ
k−1 (i+1)π n
= | sin x|dx + | sin x|dx
i=0 iπ kπ
π n
x−iπ =y
= k sin y dy + | sin x|dx
0 kπ
n
= 2k + | sin x|dx.
kπ
It follows that
1 n 2k 1 n
| sin x|dx = + | sin x|dx.
n 0 n n kπ
We have
2k 2 n n 2 2 n 2
= − = − →
n n π π π n π π
and
1 n n − kπ
0< | sin x|dx ≤ → 0,
n kπ n
since 0 ≤ n−kπ
n < πn .
2
(b) The limit equals 3π . Use a technique similar to that in the solution of part (a).
7.41.
* +
(a) Let k = πn . This implies that kπ ≤ n < (k + 1)π . Using the substitution
nx = t, we have that
1 1 n
f (| sin nx|) dx = f (| sin t|) dt
0 n 0
1 kπ 1 n
= f (| sin t|) dt + f (| sin t|) dt
n 0 n kπ
k−1 (i+1)π n
1 1
= f (| sin t|) dt + f (| sin t|) dt
n iπ n kπ
i=0
k π 1 n
t−iπ =y
= f (sin y) dy + f (| sin t|) dt.
n 0 n kπ
We have
k 1 n n 1 1 n 1
= − = − → .
n n π π π n π π
Since f is Riemann integrable, we have that f is bounded and it follows that
/ /
/ 1 n / n − kπ
0 <// f (| sin x|) dx //≤ ||f || → 0,
n kπ n
since 0 ≤ n−kπ
n < πn .
(b) This part can be solved similarly as in part (a).
The identity
n n n
n(n + 1)(2n + 1)
min {i, j } = i2 = , n∈N
6
i=1 j =1 i=1
can be proved by mathematical induction.
√
7.43. ln
√
k. Make the substitution x n = t and apply Stolz–Cesàro Theorem.
k
7.44. 2−12 .
15.3 Convergence and Evaluation of Series
7.45. The series converges for α > 1 and diverges for α ∈ (0, 1]. We have, based on
Lagrange Mean Value Theorem, that
ln n α
√ α lnα+β n αθn ln n
n
n − 1 lnβ n = e n − 1 lnβ n = e , θn ∈ 0, .
nα n
∞ lnα+β n
Therefore, the series behaves like nα .
n=2
7.46. The series converges. Observe the sequence sin n1 converges decreas-
n∈N
ingly to 0 and
sin n2 sin n+1

sin 1 + sin 2 + · · · + sin n = 2
, n ∈ N.
sin 12
Now the result follows based on Dirichlet’s criterion for the convergence of series.
Dirichlet’s Criterion If (an )n≥1 is a decreasing sequence of real numbers,

convergent to 0, and (bn )n≥1 is a sequence
n of real numbers with the property
that
∞ there exists M > 0 such that | k=1 bk | ≤ M, ∀n ≥ 1, then the series
a
n=1 n nb converges.
7.47. (b) First we prove that the following identity holds:
A Fractional Part Identity If n ∈ N, then

√ n √
(2 + 3) = 1 − (2 − 3)n ,
We have
. √ n √
(2 + 3) = an + 3bn
√ √ an , bn ∈ N.
(2 − 3)n = an − 3bn
√ √ √
Now we observe that (2 + 3)n = 2an − (an − 3bn ) = 2an − (2 − 3)n ⇒
√ n √ n √ n
(2 + 3) = −(2 − 3) = 1 − (2 − 3) .
We are ready to solve part (b) of the problem. We have

/ / / /
/ √ n / / " √ n # √ n /
/ sin π 2 + 3 / =/ sin π 2 + 3 + 2 + 3 /
/ / / /
/ /
/ √ n /
=// sin π 2 + 3 /
/
/ /
/ √ n /
=// sin π 1 − 2 − 3 /
/
/ /
/ √ n /
=// sin π 2 − 3 /
/
√ n
≤π 2− 3 .
√ n √ n
Since ∞ n=1 2 − 3 converges, it follows that ∞n=1 sin π(2 + 3) con-
verges absolutely.
7.48. The series converges. We have, based on Lagrange Mean Value Theorem, that
sin n sin n sin n

1−n n =1−e n ln n
=− ln n eθn ,
n
where θn is between 0 and sinn n ln n.

This implies, based on Lagrange Mean Value Theorem, that
sin n sin n
1−n n =− ln n eθn
n
sin n sin n
=− ln n eθn − 1 − ln n
n n
sin n sin n
=− ln n · θn · eθn − ln n,
n n
where θn is between 0 and θn .

It follows that
∞ sin n
∞
sin n
∞
sin n
1−n n =− ln n · θn · eθn − ln n.
n n
n=2 n=2 n=2
| sin n|
Observe that |θn | ≤ |θn | ≤ n ln n ≤ ln n
n < 2 and we have
∞ / / ∞
/ sin n / 2 ln2 n
/ /≤ e
/ n ln n · θn · e
θn
/ ,
n2
n=2 n=2

which implies that the series ∞ n=2n ln n · θn · e converges absolutely.
sin n θn
On the other hand, the series ∞ sin n

n=2 n ln n converges based on Dirichlet’s
criterion.
7.49. Solution I. The series converges absolutely. Observe that
/ /
/ 1 xn / 1 xn
/ f (x)dx / ≤ ||f || dx
/ 1 + x + x2 + · · · + xn / 1 + x + x2 + · · · + xn
0 0
1 xn
≤ ||f || n dx
0 (n + 1)x 2
2||f ||
= ,
(n + 1)(n + 2)
n
since the AM − GM inequality implies that 1 + x + x 2 + · · · + x n ≥ (n + 1)x 2 .
Solution II. Prove that (see [26, problem 1.39 (b)])
1 xn
lim n2 f (x)dx = f (1)ζ (2),
n→∞ 0 1 + x + x2 + · · · + xn
∞ 1
and compare the series with n=1 n2 .
7.50.
(a) Solution I. The series diverges. Using the substitution x = nt, we have that
∞
∞
π n sin2 x ∞ sin2 x
− dx = dx
2 0 x2 x2
n=1 n=1 n
∞ ∞
1 sin2 (nt)
= dt
n 1 t2
n=1
∞

∞ 1 sin2 (nt)
= dt
1 t2 n
n=1
= ∞.
We used that
∞ ∞
sin2 (nt) 1 − cos(2nt)
= = ∞,
n 2n
n=1 n=1

since the harmonic series diverges and the series ∞ n=1
cos(2nt)
n converges, based
on Dirichlet’s criterion.
Solution II. The series diverges. Let f : [1, ∞) → (0, ∞), f (x) =
∞ sin2 y
y2
dy. Observe that f decreases and we have, based on Cauchy’s
x

integral
criterion,that the series ∞n=1 f (n) behaves the same like the integral
∞ ∞ sin2 y
1 x y2
dy dx. The domain of integration of the preceding integral is
the infinite triangle with vertex (1, 1) and sides x = 1 and y = x. Changing the
order of integration, Tonelli’s Theorem applies, we get that

∞ ∞ sin2 y ∞ y sin2 y
dy dx = dx dy
1 x y2 1 1 y2
∞ sin2 y
= (y − 1)dy
1 y2
∞ sin2 y ∞ sin2 y
= dy − dy
1 y 1 y2
= ∞,
∞ 2 ∞ 2
since 1 siny y dy = ∞ and 1 siny 2 y dy < ∞.
(b) The series converges. See the solution of part (a).
7.51.
(a) We need the following inequality due to Bernoulli:
Bernoulli’s inequality. If x ∈ [0, 1] and α ∈ [0, 1], then (1−x)α ≤ 1−αx.
The inequality to prove reads

7
1 n
<p−p p
.
n+1 n+1
We have, based on Bernoulli’s inequality, that

7 1
n 1 p 1 1
p
= 1− <1− · ,
n+1 n+1 p n+1
and the preceding inequality follows.

7.52. We have
∞ ∞ ∞
{x} e−x dx = xe−x dx − x e−x dx
0 0 0
∞ k+1
=1− x e−x dx
k=0 k
∞ k+1
=1− k e−x dx
k=0 k
∞
1
=1− 1− ke−k
e
k=1
1
1
=1− 1− · e
2
e
1− 1
e
e−2
= .
e−1
∞ e (e + 1)
Similarly, one can prove that xe−x dx = .
0 2 (e − 1)2
∞ ∞
1 (−1)n
7.53. = 1 and = −e−1 .
(n − 1)! + n! (n + (−1)n )!
n=1 n=0
∞ ∞
n n
7.54. = 1 and = 2 cosh 1.
(n + 1)! (n + (−1)n )!
n=1 n=1
7.55. Use that
1 1 1 1
= − − .
(n + k − 1)(n + k)(n + k)! (n + k − 1)(n + k − 1)! (n + k)(n + k)! (n + k)!
7.56. Calculate the 2nth partial sum of the series.

7.57. For the first series, calculate the 2nth partial sum of the series or use part (b)
of Problem 3.23. For the second series, use part (d) of Problem 3.23.
7.58. Calculate the 4nth partial sum of the series and use Wallis formula.
7.59.
(a) The integral can be calculated using integration by parts. An alternative solution
is based on the formula
1
(1 − x)n x k−1 dx = B (n + 1, k) ,
0
where B denotes the beta function of Euler.

(b) We have
∞ ∞ 1
1 1
= (1 − x)n x k−1 dx
k(k + 1)(k + 2) · · · (k + n) n! 0
k=1 k=1
∞

1 1
= (1 − x) n
x k−1
dx
n! 0 k=1
1 1
= (1 − x)n−1 dx
n! 0
1
= .
n · n!
15.4 Harmonic Series
7.60. We have
1 ln x ln(1 − x) 1 ln x ln(1 − x)
dx = 2 dx
0 x(1 − x) 0 x
1 ∞
ln x xn
= −2 dx
0 x n
n=1
∞ 1
1
= −2 x n−1 ln x dx
n 0
n=1
∞
1
=2
n3
n=1
= 2ζ (3).
On the other hand, using the generating function of the nth harmonic number
(see part (a) of Problem 3.63), we have that
∞

1 ln x ln(1 − x) 1 ln x
dx = − Hn x n dx
0 x(1 − x) 0 x
n=1
∞ 1
=− Hn x n−1 ln x dx
n=1 0
∞
Hn
= .
n2
n=1
7.61.

1
(a) Observe that n(n+k) = k1 n1 − n+k1
.
∞ ∞ ∞
Hk 1
=
k2 nk(n + k)
k=1 k=1 n=1
∞ ∞ 1
1
= x n+k−1 dx
nk 0
k=1 n=1
1 ∞ xk
∞
x n−1
= dx
0 k=1 k n
n=1
1 ln2 (1 − x)
= dx
0 x
1 ln2 x
= dx
0 1−x
= 2ζ (3).
(e) Observe that
Hn2 H2 H2 Hn+1 1
= n − n+1 + 2 − .
n(n + 1) n n+1 (n + 1)2 (n + 1)3
7.62.

Use part (a) of the problem and Abel’s summation formula with:
(b) an = Hn2 and bn = ζ (2) − 1 − 212 − · · · − n12 − n1 ;
(c) an = Hn2 and bn = ζ (3) − 1 − 213 − · · · − n13 .
7.63.
(a) Use Abel’s summation formula with

an = n1 and bn = ζ (2) − 112 − 212 − · · · − n12 .
(b) We have, based on Abel’s summation formula with an = 1
n and bn = 1
n2
+
1
(n+1)2
+ (n+2)
1
2 + · · · , that
∞
1 1 1 1 1 1
+ + + ··· = lim Hn + + ···
n n2 (n + 1)2 (n + 2)2 n→∞ (n + 1)2 (n + 2)2
n=1
∞
Hn
+
n2
n=1
∞
7.61 (b) Hn
=
n2
n=1
= 2ζ (3).
An alternative solution is based on part (a) of the problem.

7.64. We apply Abel’s summation formula with
π4 H1 H2 Hn
an = Hn and bn = − 3 − 3 − ··· − 3
72 1 2 n

n
and we have, since Hk = (n + 1)Hn+1 − (n + 1), that
k=1
∞
π4 H1 H2 Hn
Hn − 3 − 3 − ··· − 3
72 1 2 n
n=1

π4 H1 H2 Hn+1
= lim [(n + 1)Hn+1 − (n + 1)] − 3 − 3 − ··· −
n→∞ 72 1 2 (n + 1)3
∞
Hn+1
+ [(n + 1)Hn+1 − (n + 1)]
(n + 1)3
n=1
∞

2
Hn+1 Hn+1
= −
(n + 1)2 (n + 1)2
n=1
17
= ζ (4) − 2ζ (3).
4
The last equality follows based on the Sandham–Yeung series, discussed in
Chap. 8, and part (b) of Problem 7.61.
7.67. Add and subtract the series in part (a) of Problem 3.76 and part (b) of Problem
7.66.
7.68.
(a) We use the formula (prove it!)
1 1 (−1)n−1 1 xn
ln + 1 − + ··· + = (−1)n−1 dx, n ≥ 1.
2 2 n 0 1+x
15.5 Series with Factorials 501
We have
∞ ∞ 1
1 1 (−1)n−1 xn
Hn ln + 1 − + · · · + = (−1)n−1 Hn dx
2 2 n 0 1+x
n=1 n=1
1 ∞
1
=− (−x)n Hn dx
0 1+x
n=1
3.63 (a)
1 ln(1 + x)
= dx
0 (1 + x)2
1 − ln 2
= .
2
7.70.
(a) Solve this part of the problem by direct computations.

(b) Use part (a) of the problem and part C of Problem 3.4.
7.71. Calculate the nth partial sum of the series, use part (a) of Problem 7.94 and
the definition of the Glaisher–Kinkelin constant.
15.5 Series with Factorials
7.72. (b) We have, based on part (a), that

∞ ∞ ∞ ∞
(n − 1)!(m − 1) (n − 1)! (m − 1)! m!
= −
(n + m)! n (n + m − 1)! (n + m)!
n=1 m=1 n=1 m=1
∞
(n − 1)! 1
= ·
n n!
n=1
∞
1
=
n2
n=1
= ζ (2).
7.73. For the solution of this problem see [116].

7.74.
(a) We have, based on part (a) of Problem 7.72, that

∞ ∞
(i − 1)!(j − 1)!
(−1)i−1
(i + j )!
i=1 j =1
∞ ∞
(i − 1)! (j − 1)! j!
= (−1)i−1 −
i (i + j − 1)! (i + j )!
i=1 j =1
∞
(i − 1)! 1
= (−1)i−1 ·
i i!
i=1
∞
(−1)i−1
=
i2
i=1
ζ (2)
= .
2
(b) Use symmetry and part (a) of the problem.
Parts (c) and (d) can be solved using the formula in part (a) of Problem 7.72.
15.6 Series of Functions
7.75. Let α ∈ (0, 1) and let p, q > 1, with p > α5 , such that 1
p + 1
q = 1. Let >0
and let δ > 0 be given by
? @
p
δ= p p .
2 q ζ q ((2 − α)q) ζ (αp − 4)
Let x, y ∈ R such that |x − y| ≤ δ. We have

/ ∞ /
/ cos(n4 x) − cos(n4 y) //
|f (x) − f (y)| =// /
n2
n=1
∞ / /
1 // n4 (x − y) /
/
≤2 2 / sin /
n 2
n=1
∞ / /p p1 ∞
1
1 // n4 (x − y) /
/ 1 q
≤2 sin ,
nαp / 2 / n(2−α)q
n=1 n=1
where the last inequality follows based on Hölder’s inequality for series.
On the other hand,
/ /p / / / 4 /
/ / / / / / n4
/ sin n (x − y) / ≤/ sin n (x − y) /≤ sin / n (x − y)
4 4 4
/≤ |x − y| ≤ n δ,
/ 2 / / 2 / / 2 / 2 2
since | sin x| ≤ |x|, ∀x ∈ R.

It follows that
∞
1 ∞
1
p q
δ 1
|f (x) − f (y)| ≤ 2
2nαp−4 n(2−α)q
n=1 n=1
1 1 1 1
≤ δ p 2 q ζ p (αp − 4)ζ q ((2 − α)q)
≤ ,
and this implies that f is uniformly continuous on R.

7.76. For > 0 there exists n0 ∈ N such that |an − l| < , ∀n ≥ n0 .
We have
∞ n0 −1 ∞
−x xn xn xn
e an − l = e−x (an − l) + e−x (an − l)
n! n! n=n
n!
n=0 n=0 0
and it follows that

/ ∞ / n0 −1 ∞
/ −x xn / |x|n |x|n
/e an − l // ≤ e−x |an − l| + e−x |an − l|
/ n! n! n!
n=0 n=0 n=n 0
n0 −1
|x|n
≤ e−x |an − l| + e−x · e|x| .
n!
n=0
Passing to the limit, as x → ∞, in the previous inequality, we have that

/ ∞ / n0 −1
/ xn / |x|n
lim / e−x an − l // ≤ lim e−x |an − l| + = ,
x→∞ / n! x→∞ n!
n=0 n=0
∞ xn
and, since > 0 was arbitrary, it follows that lim e−x n=0 an n! = l.
x→∞
7.77.
(a) We assume that sup an = 1. Otherwise, we replace the sequence an by supanan .

We assume that the series ∞ n=1 an converges. This implies that an → 0 and it
x
follows that sup an is finite.

Let > 0. Since the tail of a convergent series converges to 0, we have that

∞
there exists n0 ∈ N such that akx < , ∀n ≥ n0 .
k=n
We have
∞
1
x 1 x
1
a1 + a2x + · · · + anx x ≤ anx ≤ a1x + a2x + · · · + anx + x , n ≥ n0 .
n=1
Passing, to the limit as x → ∞, in the previous inequalities, we get that

∞
1
x
max ak ≤ lim anx ≤ 1, n ≥ n0 , (15.1)
k=1,...,n x→∞
n=1
since
1
lim a1x + a2x + · · · + anx x = max ak
x→∞ k=1,...,n
1
and lim a1x + a2x + · · · + anx + x = 1, when ai ∈ (0, 1), i = 1, . . . , n and
x→∞
> 0.
Passing to the limit, as n → ∞, in (15.1), we get that
∞
1
x
1 = sup an = lim max ak ≤ lim anx ≤ 1,
n→∞ k=1,...,n x→∞
n=1
∞ 1
which implies that lim x
n=1 an
x = 1.
x→∞
1
(b) lim ζ (x) = 1.
x
x→∞

Remark. If the series ∞ x
n=1 an does not converge, the problem is no longer valid. If

an = ln(n+1) , then n=1 an = ∞
1 ∞ x
n=1 lnx (n+1) = ∞, ∀x > 0 and supn≥1 ln(n+1) =
1 1
1
ln 2 .
7.78. We have, if x = 0, that
1 1 1
xt
dt = e−tx ln t dt
0 t 0
∞

1 (−tx ln t)n
= dt
0 n!
n=0
∞ 1
xn
= (−1)n t n lnn tdt
n! 0
n=0
∞
xn
=
(n + 1)n+1
n=0
1
∞ x n
= .
x n
n=1
We used in our calculations the formula

1 n!
x m lnn x dx = (−1)n , m, n ≥ 0.
0 (m + 1)n+1
It follows that, for x > 0, one has

1 x1
∞ x n x
1
1 1 1
1 1 x x1
= xx xt
dt = xx dt .
n 0 t 0 tt
n=1
This implies that

1
∞ x n x 1 1 x x1
lim = lim dt = max f (t),
x→∞ n x→∞ 0 tt t∈[0,1]
n=1
where f is the continuous function defined by

.
1 if t = 0
f (t) = 1
tt if t ∈ (0, 1].
−1
A calculation shows that f attains its maximum when t = 1
e and max f (t) = ee .
t∈[0,1]
7.79. Part (a) follows by direct computations, see the solution of Problem 7.78. For
part (b) of the problem see the solution of Problem 7.30.
7.80. We have
∞
xn 1 1+x x
= ln + ln(1 − x 2 ) − 1, x ∈ (−1, 1).
(−1)n n + 1 2x 1 − x 2
n=2
7.81. We have, for x ∈ (−1, 1), that

∞ ∞ ∞
1
Hn − 1 − x − x2 − · · · − xn = Hn x n+m
1−x
n=1 n=1 m=1
∞ ∞
= xm Hn x n
m=1 n=1
3.63 (a) x
= − ln(1 − x).
(1 − x)2
7.82. Solution I. We apply Abel’s summation formula with an = 1 and bn =
∞

∞
x n+m−2 x n+m−1
ex − 1 − 1!
x
− · · · − (n+m−2)! and we have, since bn −bn+1 = (n+m−1)! ,
m=1 m=1
that
∞ ∞
x x2 x n+m−2
e −1− −
x
− ··· −
1! 2! (n + m − 2)!
n=1 m=1
∞ ∞
x n+m−1
= n
(n + m − 1)!
n=1 m=1
∞
xn x n+1
= n + + ···
n! (n + 1)!
n=1
∞
n(n + 1) x n
= ·
2 n!
n=1
x 2 + 2x x
= e ,
2
where the third equality follows based on Abel’s summation formula with an = n
n x n+1
and bn = xn! + (n+1)! + ···.
Solution II. Use part (a) of Problem 4.69.
7.83. We integrate by parts and we have
z ln(1 − t) /z z ln t
−Li2 (z) = dt = ln t ln(1 − t)/0 + dt
0 t 0 1−t
1 ln t 1 ln t
= ln z ln(1 − z) + dt − dt
0 1−t z 1−t
π2 1−z ln(1 − u)
= ln z ln(1 − z) − − du
6 0 u
π2
= ln z ln(1 − z) − + Li2 (1 − z).
6
(b) Let z = sin2 θ and we have, based on part (a), that

∞
sin2n θ + cos2n θ π2
2
= − 4 ln | sin θ | ln | cos θ |.
n 6
n=1
7.85. We have, based on the formula (see part (a) of Problem 4.69)
x (x − t)n t x x2 xn
e dt = ex − 1 − − − ··· − , ∀n ∈ N, x ∈ R,
0 n! 1! 2! n!
that
∞ ∞
1 1 1 2 1 1 (1 − y)n y
n! e − 1 − − − · · · − = (1 − x)n ex dx e dy
1! 2! n! n!
n=0 n=0 0 0
∞
1 1 [(1 − x)(1 − y)]n
= e x+y
dxdy
0 0 n!
n=0
1 1
= ex+y · e(1−x)(1−y) dxdy
0 0
1 1
=e exy dxdy
0 0
1 ex − 1
=e dx
0 x
1 ∞ x n−1
=e dx
0 n=1 n!
∞
1
=e .
n · n!
n=1
7.86. Formula (7.1) can be proved by differentiation and then by using part (a) of
Problem 3.63.
7.87. Divide by x = 0 and integrate from 0 to x the power series in Problem 3.4 C.
The power series can also be proved by differentiation and by comparing to the
power series in Problem 3.4 C.
7.88. Use Abel’s summation formula
with2
n 2
an = 1 and bn = ln2 (1 − x) − x + x2 + · · · + xn .
15.7 Pearls of Series with Tails of Zeta Function Values
7.90.
(a) Use Abel’s summation formula with
an = Hn and bn = ζ (2) − 1 − 212 − · · · − 1
n2
− n1 .
(2)
(b) Let Hn = 1 + 1
22
+ ··· + 1
n2
. Use that
2
1 1
Hn(2) ζ (2) − Hn −
(2)
= − ζ (2) − Hn (2)
+ ζ (2) ζ (2) − Hn −
(2)
n n
1
+ ζ (2) − Hn(2) .
n
(a) an = n and bn = ζ (3) − 1 − 213 − · · · − n13 − 2n1 2 ;

(b) an = n2 and bn = ζ (4) − 1 − 214 − · · · − n14 − 3n1 3 ;
(c) an = n3 and bn = ζ (5) − 1 − 215 − · · · − n15 − 4n1 4 .
7.94.

an = nHn and bn = ζ (2) − 1 − 212 − · · · − 1
n + 1
2n2
.
(c) Use Abel’s summation formula with
an = nHn and bn = ζ (3) − 1 − 213 − · · · − 1
n3
− 1
2n2
.
(d) Use Abel’s summation formula with
an = nHn and bn = ζ (4) − 1 − 214 − · · · − 1
n4
.
7.95.

Use Abel’s summation formula with:
(b) an = (−1)n n2 and bn = ζ (3) − 1 − 213 − · · · − 1
n3
− 2n1 2 ;
(c) an = (−1)n n2 and bn = ζ (4) − 1 − 214 − · · · − 1
n4
.
7.97. Use Abel’s summation formula

with 2
an = 1 and bn = ζ (3) − 1 + 23 + · · · +
2 1 1
n3
.
7.98. See [61].
7.99.
(a) A solution is given in [27]. An alternative solution is based on the

application of Abel’s summation formula with an = n1 and bn =
2 ∞ Hn(2)
ζ (2) − 1 − 1
22
− ··· − 1
, combined with the formulae
n2 n=1 n3 =
(2)
3ζ (2)ζ (3)− 92 ζ (5) [22, Theorem 3.1, p. 22] and ∞n=1
Hn Hn
n2
= ζ (2)ζ (3)+ζ (5)
(2)
[109, Theorem 6, p. 210], where Hn = 1 + 1
22
+ ··· + 1
n2
.
(b) Use part (a).

2
(a) an = n and bn = ζ (3) − 1 − 213 − · · · − n13 ;
2
(b) an = n2 and bn = ζ (3) − 1 − 213 − · · · − n13 ;
2
(c) an = n3 and bn = ζ (3) − 1 − 213 − · · · − n13 − 1
4n4
.
7.102.
(a) We need the following formulae which can be proved by using Abel’s summa-
tion formula:
∞
ζ (2) − 1 − 1
22
− ··· − 1
n2
• = ζ (3);
n
n=1
∞
ζ (3) − 1 − 1
1
23
− ··· − 1
n3
• ζ (4); =
n 4
n=1
∞
1 1
• ζ (3) − 1 − 3 − · · · − 3 = ζ (2) − ζ (3).
2 n
n=1
Now we are ready to solve part

(a) of the problem. We apply
Abel’s summation
formula with an = n and bn = ζ (2) − 1 − 22 − · · · − n2 ζ (3) − 1 − 213 − · · ·
1 1

− n13 and we have that
∞
1 1 1 1
n ζ (2) − 1 − 2 − · · · − 2 ζ (3) − 1 − 3 − · · · − 3
2 n 2 n
n=1
∞

n(n + 1) ζ (2) − 1 − 12 − · · · − 12 ζ (3) − 1 − 13 − · · · − 13 1
2 n 2 n
= + −
2 (n + 1)3 (n + 1)2 (n + 1)5
n=1
∞

1 ζ (2) − 1 − 12 − · · · − 12 ζ (3) − 1 − 13 − · · · − 13 1
2 n 2 n
= n + −
2 (n + 1)2 n+1 (n + 1)4
n=1
⎛ ⎞
1
∞ ζ (2) − 1 − 12 − · · · − 1 ζ (3) − 1 − 13 − · · · − 1
1
(n+1)2 (n+1)3
n⎝ ⎠
2 2
= + +
2 (n + 1)2 n+1 (n + 1)4
n=1
∞

1 ζ (2) − 1 − 12 − · · · − 12 ζ (3) − 1 − 13 − · · · − 13 1
2 m 2 m
= (m − 1) + + 4
2 m2 m m
m=2
∞

1 ζ (2) − 1 − 12 − · · · − 12 ζ (3) − 1 − 13 − · · · − 13 1
2 m 2 m
= (m − 1) + +
2 m2 m m4
m=1
∞ ζ (2) − 1 − 1 − · · · − 1 ∞ ζ (2) − 1 − 1 − · · · − 1
1 22 m2 1 22 m2
= −
2 m 2 m2
m=1 m=1
∞ ∞ ζ (3) − 1 − 1 − · · · − 1
1 1 1 1 23 m3
+ ζ (3) − 1 − 3 − · · · − 3 −
2 2 m 2 m
m=1 m=1
∞ ∞
1 1 1 1
+ −
2 m3 2 m4
m=1 m=1
1 3 1 1 1 1
= ζ (3) − ζ (4) + (ζ (2) − ζ (3)) − ζ (4) + ζ (3) − ζ (4)
2 8 2 8 2 2
1
= (ζ (2) + ζ (3) − 2ζ (4)) .
2
(b) We need the following formulae:

∞
1 1
• ζ (4) − 1 − 4 − · · · − 4 = ζ (3) − ζ (4);
2 n
n=1
∞
ζ (4) − 1 − 1
24
− ··· − 1
n4
• = 2ζ (5) − ζ (2)ζ (3);
n
n=1
∞
ζ (2) − 1 − 1
22
− ··· − 1
n2 9
• = ζ (5) − 2ζ (2)ζ (3).
n3 2
n=1
The first two equalities can be proved by using Abel’s summation formula, and
the last series equality is a consequence of the series formula
∞
1+ 1
22
+ ··· + 1
n2 9
= − ζ (5) + 3ζ (2)ζ (3),
n3 2
n=1
which follows based on Theorem 3.1 in [22].

We apply Abel’s summation formula with an = n and

1 1 1 1
bn = ζ (2) − 1 − 2 − · · · − 2 ζ (4) − 1 − 4 − · · · − 4 ,
2 n 2 n
and we have that

∞
1 1 1 1
n ζ (2) − 1 − 2 − · · · − 2 ζ (4) − 1 − 4 − · · · − 4
2 n 2 n
n=1
∞

n(n + 1) ζ (2) − 1 − 12 − · · · − 12 ζ (4) − 1 − 14 − · · · − 14 1
2 n 2 n
= + −
2 (n + 1)4 (n + 1)2 (n + 1)6
n=1
∞

1 ζ (2) − 1 − 12 − · · · − 12 ζ (4) − 1 − 14 − · · · − 14 1
2 n 2 n
= n + −
2 (n + 1)3 n+1 (n + 1)5
n=1
∞

1 ζ (2) − 1 − 12 − · · · − 12 ζ (4) − 1 − 14 − · · · − 14 1
2 m 2 m
= (m − 1) + + 5
2 m3 m m
m=1
∞ ζ (2) − 1 − 1 − · · · − 1 ∞ ζ (2) − 1 − 1 − · · · − 1
1 22 m2 1 22 m2
= −
2 m2 2 m3
m=1 m=1
∞ ∞ ζ (4) − 1 − 1 − · · · − 1
1 1 1 1 24 m4
+ ζ (4) − 1 − 4 − · · · − 4 −
2 2 m 2 m
m=1 m=1
∞ ∞
1 1 1 1
+ −
2m4 2 m5
m=1 m=1

3 1 9 1 1
= ζ (4) − ζ (5) − 2ζ (2)ζ (3) + (ζ (3) − ζ (4)) − (2ζ (5) − ζ (2)ζ (3))
8 2 2 2 2
1 1
+ ζ (4) − ζ (5)
2 2
3 15 3 1
= ζ (2)ζ (3) − ζ (5) + ζ (4) + ζ (3).
2 4 8 2
7.105. (b) We have

∞ 2
1 1
+ + ···
nk (n + 1)k
n=1
∞ 2
1 1
= ζ (k) +
2
+ + ···
nk (n + 1)k
n=2
∞ 2
n−1=m 1 1
= ζ 2 (k) + + + ···
(m + 1)k (m + 2)k
m=1
∞ 2
1 1 1
= ζ 2 (k) + + + ··· −
mk (m + 1)k mk
m=1
∞ 2 ∞
1 1 1 1 1
= ζ (k) +
2
+ + ··· −2 + + ···
m k (m + 1)k mk m k (m + 1)k
m=1 m=1
+ ζ (2k)
and it follows that

∞
1 1 1 ζ 2 (k) + ζ (2k)
+ + ··· = .
nk nk (n + 1)k 2
n=1
To prove the equality in Remark 7.7, we note that
∞ ∞
1+ 1
2k
+ ··· + 1
nk 1 1 1
+ + + ···
nk nk nk (n + 1)k
n=1 n=1
∞ ∞
1 1 1 1
= + 1 + k + k + ···
n2k nk 2 3
n=1 n=1
= ζ (2k) + ζ (k), 2
and the equality follows based on part (b) of the problem.

7.106. We have
∞
1 1 1 1
+ + · · · + + · · ·
n2 (n + 1)2 (n + 1)2 (n + 2)2
n=1
∞
2
1 1 1 1 1
= + + ··· − 2 + + ···
n2 (n + 1)2 n n2 (n + 1)2
n=1
∞ 2 ∞
1 1 1 1 1
= + + ··· − + + ···
n2 (n + 1)2 n2 n2 (n + 1)2
n=1 n=1
7
= 3ζ (3) − ζ (4),
4
where the last equality follows based on formula (7.2) and part (a) of Problem 7.105.
7.107. (b) Shifting the index of summation.

We have
∞ 3 ∞ 3
1 1 3 (k) + 1 1
+ + · · · = ζ + + · · ·
nk (n + 1)k nk (n + 1)k
n=1 n=2
∞ 3
n−1=m 3 1 1
= ζ (k) + + + · · ·
(m + 1)k (m + 2)k
m=1
∞
1 1 1 3
= ζ 3 (k) + + + ··· −
mk (m + 1)k mk
m=1
∞ 3
(a−b)3 =a 3 −3ab(a−b)−b3 1 1
= ζ 3 (k) + + + · · ·
mk (m + 1)k
m=1
∞
1 1 1 1 1
−3 + + ··· + + ··· − ζ (3k),
mk mk (m + 1)k (m + 1)k (m + 2)k
m=1
7.108.
(a) We need the series formula
∞
1 1 1
+ + ··· = 2ζ (3), (15.2)
n n2 (n + 1)2
n=1
which can be proved using Abel’s summation formula with an = 1

n and bn =
1
n2
+ (n+1)
1
2 + ···.
We have
∞ 2
1 1 1
+ + ··· −
n2 (n + 1)2 n
n=1
∞ 2
1 1
= + + ···
n2 (n + 1)2
n=1
∞
1 1 1
−2 + + ··· + ζ (2)
n n2 (n + 1)2
n=1
= ζ (2) − ζ (3),
where the last equality follows based on formulae (7.2) and (15.2).
(b) We apply Abel’s summation formula with an = n and
2
1 1 1
bn = + + ··· − ,
n2 (n + 1)2 n
and we have, since

1 1 1 1 1 1
bn − bn+1 = 2 + + · · · − 2− − ,
n (n + 1)
2 n2 (n + 1)2 n n n+1
that
∞
1 1 1 2
n + + ··· −
n2 (n + 1)2 n
n=1
∞
1 1 1 1 1 1 1
= 2 + + · · · − 2 − −
2 n n2 (n + 1)2 n n n+1
n=1
∞ ∞ ∞ ∞
1 1 1 1 1 1 1 1 1
= + + ··· − − −
n n2 (n + 1)2 2 n3 2 n2 2 n(n + 1)
n=1 n=1 n=1 n=1
3 1 1
= ζ (3) − ζ (2) − ,
2 2 2
where the last equality follows based on formula (15.2).

2
(a) an = 2n − 1 and bn = 1
3 + 1
(n+1)3
+ ··· ;
n
2
(b) an = 2n − 1 and bn = 1
n2
+ 1
(n+1)2
+ ··· − 1
n2
.
Parts (a) and (b) of the problem are solved in [20] and [80].
7.110.
(a) We use Abel’s summation formula with an = 3n2 − 3n + 1 and bn =

2
1
n3
+ (n+1)
1
3 + ··· , and we have, since
n
1 2 2 1
ak = n3 and bn − bn+1 = + + ··· − ,
n3 n3 (n + 1)3 n3
k=1
that
∞ 2 ∞
1 1 2 2 1
(3n2 − 3n + 1) + + ··· = + + ··· −
n3 (n + 1)3 n3 (n + 1)3 n3
n=1 n=1
∞
1 1
=2 + + ··· − ζ (3)
n3 (n + 1)3
n=1
= 2ζ (2) − ζ (3).
We used the formula

∞
1 1
+ + ··· = ζ (2).
n3 (n + 1)3
n=1
More generally, for k > 2, one has that

∞
1 1
+ + ··· = ζ (k − 1).
nk (n + 1)k
n=1
(b) Use Abel’s summation formula with an = 4n3 − 6n2 + 4n − 1 and bn =

2
1
n4
+ (n+1)
1
4 + ··· .
(a) an = Hn and bn = n13 + (n+1)

1
3 + ···;
2
(b) an = n and bn = n13 + (n+1)
1
3 + · · · .
7.112. We will be using the following formula (see part (a) of Problem 7.66):
1 Hn
x n−1 ln(1 − x)dx = − , n ≥ 1. (15.3)
0 n
See [26, p. 206] for information about the history of this formula. We have
∞ ∞
Hn 1 1 Hn 1 1
+ + ··· = ζ (2) + + + ···
n n2 (n + 1)2 n n2 (n + 1)2
n=1 n=2
∞ 1
1 1
= ζ (2) − x n−1
ln(1 − x) ζ (2) − 1 − 2 − · · · − dx
2 (n − 1)2
n=2 0
1 ∞
1 1
= ζ (2) − ln(1 − x) ζ (2) − 1 − 2 − · · · − x n−1 dx
0 2 (n − 1)2
n=2
1 ∞
1 1
= ζ (2) − ln(1 − x) ζ (2) − 1 − 2 − · · · − 2 x m dx
0 2 m
m=1
1 ln(1 − x)
= ζ (2) − (xζ (2) − Li2 (x)) dx,
0 1−x
(15.4)
where the last equality follows based on part C of Problem 3.4.
We calculate the preceding integral by parts, with f (x) = xζ (2) − Li2 (x),
f (x) = ζ (2) + ln(1−x)
x , g (x) = − ln(1−x)
1−x , and g(x) = 2 ln (1 − x), and we
1 2
have that
1 ln(1 − x)
− (xζ (2) − Li2 (x)) dx
0 1−x
/
ln2 (1 − x) //1 1 ln2 (1 − x) ln3 (1 − x)
= (xζ (2) − Li2 (x)) / − ζ (2) + dx
2 0 0 2 2x
ζ (2) 1 1 1 ln3 (1 − x)
=− ln2 (1 − x)dx − dx
2 0 2 0 x
π4
= −ζ (2) + ,
30
(15.5)
1 1 ln3 (1−x) 4
since 0 ln (1 − x)dx = 2 and
2
= 0 x dx − π15 .
It follows, based on (15.4) and (15.5), that
∞
Hn 1 1 π4
+ + ··· = ,
n n2 (n + 1)2 30
n=1

7.113.

Hn
(a) Let xn = n(n+1) ζ (2) − 1 − 1
22
− ··· − 1
n2
and yn = ζ (2)−1− 212 −· · ·− n12 .
Observe that
Hn yn Hn+1 yn+1 Hn+1 yn+1 1

xn = − − + + , n ≥ 1.
n n+1 (n + 1)3 (n + 1)2 (n + 1)4
1
7.114. Part (a) follows based on the power series expansion of the function 1−x .

∞
(−1)n−1 1 1
ζ (2) − 1 − − ··· − 2
n 22 n
n=1
∞ 1
(−1)n xn
= ln x dx
n 0 1−x
n=1
1 ∞
ln x (−x)n
= dx
0 1−x n
n=1
1 ln x ln(1 + x)
=− dx
0 1−x
1−x=t
1 ln(1 − t) ln(2 − t)
= − dt
0 t

1 ln(1 − t) ln 2 + ln 1 − 2t
=− dt
0 t

1 ln(1 − t) 1 ln(1 − t) ln 1 − 2t
= − ln 2 dt − dt
0 t 0 t
∞ t n
1 ln(1 − t)
= ln 2ζ (2) + 2
dt
0 t n
n=1
∞ 1
1
= ln 2ζ (2) + t n−1 ln(1 − t)dt
n2n 0
n=1
∞
(15.3) Hn
= ln 2ζ (2) −
n2 2n
n=1
π2
= ln 2 − ζ (3),
4
since
∞
Hn π2
= ζ (3) − ln 2.
n2 2n 12
n=1
The previous formula follows directly from the generating function of the
sequence Hn2n (see [109, formula 28, p. 217])
∞
Hn n 1
z = ln2 (1 − z) ln z + ln(1 − z)Li2 (1 − z) − Li3 (1 − z) + Li3 (z) + ζ (3),
n2 2
n=1
for |z| < 1.

For an alternative solution of part (b) see [61, Lemma 4, (d)].
∞
1 1
(−1) Hn ζ (2) − 1 − 2 − · · · − 2
n
2 n
n=1
∞ 1 xn
=− (−1)n Hn ln xdx
0 1−x
n=1
1 ∞
ln x
=− Hn (−x)n dx
0 1−x
n=1
3.63 (a)
1 ln x ln(1 + x)
= dx
0 1 − x2
1 1 ln x ln(1 + x) 1 1 ln x ln(1 + x)
= dx + dx
2 0 1−x 2 0 1+x
7ζ (3) π 2 ln 2
= − ,
16 8
since (see the solution of Problem 7.114)
1 ln x ln(1 + x) π2
dx = ζ (3) − ln 2
0 1−x 4
and
1 ln x ln(1 + x) 1 1 ln2 (1 + x) ζ (3)
dx = − dx = − ,
0 1+x 2 0 x 8
where the last equality is a formula due to Ramanujan [6, pp. 291–292].
7.116. See [105].
π2
(a) an = 1 and bn = 8 −1− 1
32
− ··· − 1
(2n−1)2
− 1
4n ;
π2
(b) an = n and bn = 8 −1− 1
32
− ··· − 1
(2n−1)2
− 1
4n ;
15.8 Exotic Zeta Series 519
π2
(c) an = (−1)n n and bn = 8 −1− 1
32
− ··· − 1
(2n−1)2
− 1
4n ;
(d) Use that
π2 1 1 1 x 2n
− 1 − 2 − ··· − =− ln x dx, n ≥ 1;
8 3 (2n − 1)2 0 1 − x2
π2
(e) an = 1
(2n−1)2
and bn = 8 −1− 1
32
− ··· − 1
(2n−1)2
;
2
π4 π2
(f) an = 1 and bn = 64 − 1+ 1
32
+ ··· + 1
(2n−1)2
− 16n , and parts (d) and (e)
of the problem;
2
π2
(g) an = 1 and bn = 8 −1− 1
32
− ··· − 1
(2n−1)2
, and parts (d) and (e) of the
problem.
15.8 Exotic Zeta Series
7.121. (c) Using the formula
1 (k − 1)!
x m lnk−1 x dx = (−1)k−1 , m ∈ N, k ∈ N,
0 (m + 1)k
we have that
∞ ∞ 1
1 1 (−1)k−1
+ = lnk−1 x x 6n + x 6n+4 dx
(6n + 1)k (6n + 5)k (k − 1)! 0
n=0 n=0
(−1)k−1 1 ∞
= lnk−1 x x 6n + x 6n+4 dx
(k − 1)! 0 n=0
(−1)k−1 1 + x 4 k−1
1
= ln x dx
(k − 1)!
0 1−x
6

(−1)k−1 1 k−1 1 x2
= ln x − dx
(k − 1)! 0 1 − x2 1 − x6
1 k−1
(−1)k−1 1 ln x
= 1− k dx
(k − 1)! 3 0 1 − x 2
1 ∞
(−1)k−1 1
= 1− k lnk−1 x x 2n dx
(k − 1)! 3 0 n=0
∞
1 1
= 1− k
3 (2n + 1)k
n=0

1 1
= 1− k 1 − k ζ (k).
2 3

7.122. (e) We have
n ∞
1
ζ (3) + ζ (5) + · · · + ζ (2n + 1) =
k 2i+1
i=1 k=1
n ∞
1
=n+
k 2i+1
i=1 k=2
∞ n
1
=n+
k 2i+1
k=2 i=1
∞
1 1 − k 2n
1
=n+ · 2
k k −1
k=2
∞ ∞
1 1
=n+ −
k(k 2 − 1) (k 2 − 1)k 2n+1
k=2 k=2
∞
1 1
=n+ − .
4 (k 2 − 1)k 2n+1
k=2
It follows that
∞
1 1
n+ − ζ (3) − ζ (5) − · · · − ζ (2n + 1) = (15.6)
4 (k 2 − 1)k 2n+1
k=2
and
∞
1 1
lim n + − ζ (3) − ζ (5) − · · · − ζ (2n + 1) = lim = 0.
n→∞ 4 n→∞ (k 2 − 1)k 2n+1
k=2
We have, based on formula (15.6), that

∞ ∞ ∞
1 1
n + − ζ (3) − ζ (5) − · · · − ζ (2n + 1) =
4 k(k 2 − 1)k 2n
n=1 n=1 k=2
∞ ∞
1 1
=
k(k 2 − 1) k 2n
k=2 n=1
15.9 Special Differential Equations 521
∞
1
=
k(k 2 − 1)2
k=2
1
= .
16
The other parts of the problem can be solved either by using Abel’s summation
formula, or by using a similar technique to the one given in the solution of part (e).
7.123. Use Abel’s summation formula with an = 1 and bn = 12 − ζ (2) − ζ (4) −
· · · − ζ (2n) + ζ (3) + ζ (5) + · · · + ζ (2n + 1).
15.9 Special Differential Equations

k x k+1
x
7.124. (a) f (x) = f (0)e f 2 (0) , x ∈ R; (b) f (x) = f (0)e (k+1)f 2 (0) , x ∈ R.
7.125.
2f (0)x
(a) f (x) = 2f (0) 1+e
e
2f (0)x , x ∈ R.
2 (0)x
(b) f (x) = f (0)ef , x ∈ R. We have
f (x) = f 2 (x)f (−x), x ∈ R

f (−x) = f 2 (−x)f (x), x ∈ R.
We multiply the first equation by f (−x) and the second equation by f (x)
and we have that f (x)f (−x) − f (−x)f (x) = 0, ∀x ∈ R. It follows that
(f (x)f (−x)) = 0, ∀x ∈ R, and this shows that f (x)f (−x) = C , ∀x ∈ R.
When x = 0 we have C = f 2 (0), and it follows that f (x)f (−x) = f 2 (0). Thus,
f (x) = f 2 (x)f (−x) = f 2 (0)f (x), ∀x ∈ R. The solution of this differential
2
equation with separable variables is f (x) = f (0)ef (0)x , x ∈ R.
√ 2x √ 2 x3
7.126. (a) f (x) = 2 √ e 4x , x ∈ R; (b) f (x) = 2 6 e 3 4 3 , x ∈ R.
1+e 1+e 3
x
√ 3
e2f (0)x
7.127. f (x) = 3 2f (0) √
3 3
, x ∈ R.
1+e6f (0)x
7.128. f (x) = ax + b, ∀x ∈ R, a, b ∈ R.
7.129. f (x) = ax 2 + bx + c, ∀x ∈ R, a, b, c ∈ R.
7.130. f (x) = ax 2 + bx + c, ∀x ∈ R, a, b, c ∈ R.
7.131. f (x) = ax + b, ∀x ∈ R, a, b ∈ R.
7.132. f (x) = sin x + cos x, ∀x ∈ R.
15.10 Inequalities
7.133. Use that
1 ∞
e−(ai +aj )x dx.
2 2
=
ai + aj2
2
0
7.134. We have
n n
ij
S= ai aj
i+j −1
i=1 j =1
n n 1
= ij ai aj x i+j −2 dx
i=1 j =1 0
n
⎛ n
⎞
1
= iai x i−1 ⎝ j aj x j −1 ⎠ dx
0 i=1 j =1
n
2
1
= iai x i−1 dx
0 i=1
1
= f 2 (x)dx,
0

n
where f (x) = iai x i−1 . It follows, based on Hölder’s inequality for integrals,
i=1
that
/ / / /
√ 1 1 / 1 / / n
/
S= f 2 (x)dx ≥ |f (x)|dx ≥// f (x)dx //=// ai // .
0 0 0 i=1
15.11 Fabulous Integrals
7.135. We have

xa (1−x)a
1 ln(x a + (1 − x)a )
1
2 ln(1 − x)a 1 + (1−x)a 1 ln x a 1 + xa
dx = dx + dx
0 x 0 x 1
2
x
(continued)

xa
1
2 ln(1 − x)
1
2 ln 1 + (1−x)a
=a dx + dx
0 x 0 x

(1−x)a
1 ln x 1 ln 1 + xa
+a dx + dx
1
2
x 1
2
x
(15.7)
1
ln 1 + xa
1 ln2 2 2 (1−x)a
= −aLi2 −a + dx
2 2 0 x

(1−x)a
1 ln 1 + xa
+ dx.
1
2
x
Using the substitution x

1−x = t, we have that

xa
1
2 ln 1 + (1−x)a 1 ln(1 + t a )
dx = dt
0 x 0 t (1 + t)
(15.8)
1 ln(1 + t a ) 1 ln(1 + t a )
= dt − dt.
0 t 0 1+t
On the other hand, the substitution 1−x

x = t shows that

(1−x)a
1 ln 1 + xa 1 ln(1 + t a )
dx = dt. (15.9)
1
2
x 0 1+t
It follows, based on (15.7), (15.8), and (15.9), that

1 ln(x a + (1 − x)a ) 1 ln2 2 1 ln(1 + t a )
dx = −aLi2 −a + dt
0 x 2 2 0 t
t a =y π2 1 1 ln(1 + y)
= −a + dy
12 a 0 y

π2 1
= −a .
12 a
7.136. A more general version of this integral was proposed by O. Furdui as a

problem in [47] and two solutions by A. Kotronis and O. Kouba have been published
in [101].
7.138. Using the substitution x1 = t, we have that
1 ∞ ∞ k+1
xm 1 1
" # dx = dt = dt
0 1 1 t m+2 t t m+2 t
x k=1 k
∞
ζ (m + 2) 1 1
= − .
m+1 m+1 k(k + 1)m+1
k=1

Let Sm+1 = ∞ k=1
1
k(k+1)m+1
. Observe that Sm+1 = Sm − (ζ (m + 1) − 1).
7.139. (a) We have
∞ 1 1
∞ n+1 1 1

e−1− − ··· − dx = e−1− − ··· − dx
1 1! x! 1! x!
n=1 n
∞
1 1
= e − 1 − − ··· −
1! n!
n=1
∞
(∗) n
=
(n + 1)!
n=1
= 1.
We used at step (*) Abel’s summation formula with an = 1 and bn = e − 1 − 1!1 −

· · · − n!
1
.
The other parts of the problem can be solved similarly. The calculation of the
series that are equal to the integrals which appear in parts (b)–(f) can be found in
[59].
Remark.
∞ Problem 7.139 refers to the calculation of improper integrals of the
form 1 f (x)dx, which verify the equality
∞ ∞
f (x)dx = f (n).
1 n=1
(continued)
The preceding
∞ equality can be viewed as sophomore’s dream for integrals
of the form 1 f (x)dx (see also Problem 3.52).
7.143.
(a) Let k be the degree of f . We integrate by parts and we have that

∞ ∞
e−x f (x)dx = f (0) + e−x f (x)dx
0 0
∞
= f (0) + f (0) + e−x f (x)dx
0
= ···
= f (0) + f (0) + · · · + f (k) (0).
∞ n
(b) First we observe that if n ≥ 0 is an integer, then 0 e−x xn! dx = 1. Let Sn =
n
f (i) (0). A calculation shows that
i=0
∞
Sn = e−x Tn (x)dx,
0
(n)
where Tn (x) = f (0) + f 1!(0) x + f 2!(0) x 2 + · · · + f n!(0) x n is the nth degree
Taylor polynomial of f at 0. ∞
Now we prove that the integral 0 e−x f (x)dx converges. We need to
A
show that lim 0 e−x f (x)dx exists and is finite. We apply Bolzano–Cauchy
A→∞
criterion
for proving that the limit of a function exists and is finite. Let > 0.
Since ∞
n=0 f (n) (0) converges absolutely, we get that there exists n ∈ N
0
such that ∞ |f (n) (0)| < , for all n ≥ n 0 . On the other hand, since
∞ −x n=n 0 2
e T n (x)dx converges, its value being calculated by formula in part (a)
0 0
A −x
of the problem, we get that lim 0 e Tn0 (x)dx exists and it follows that, for
A→∞
> 0 there exists δ > 0 such that for all A , A > δ one has
/ /
/ A /
/ e −x
Tn0 (x)dx //< .
/ 2
A
We have, for A , A > δ, that

/ / / /
/ A / A / A /
/ e−x f (x)dx // ≤ e−x |f (x) − Tn0 (x)|dx+ // e−x Tn0 (x)dx //
/
A A A
/ ∞ /
A / f (k) (0) k //
< e−x / x / dx +
/ k! 2
A k=n0 +1
∞ A xk
≤ |f (k) (0)| e−x dx +
A k! 2
k=n0 +1
∞ ∞ xk
≤ |f (k) (0)| e−x dx +
0 k! 2
k=n0 +1
∞
≤ |f (k) (0)| +
2
k=n0 +1
≤ ,
A A
which proves that lim e−x f (x)dx exists and lim e−x f (x)dx =
∞ −x A→∞ 0 A→∞ 0
0 e f (x)dx.

∞ ∞
Now we show that f (n) (0) = 0 e−x f (x)dx. We calculate
n=0
/ / / /
/ ∞ / / ∞ /
/ −x
e f (x)dx − Sn // =// e −x
(f (x) − Tn (x))dx //
/
0 0
∞
≤ e−x |f (x) − Tn (x)|dx
0
∞ ∞
−x f (k) (0) k
≤ e x dx
0 k!
k=n+1
∞
≤ |f (k) (0)|
k=n+1
≤ ,
∞
which implies that lim Sn = 0 e−x f (x)dx and the problem is solved.
n→∞
Remark.
(a) An example of a function which verifies the statement of the problem

is f (x) = e−αx , α ∈ (−1, 1), and a function which has a power series
expansion at 0, with radius of convergence R =(n) ∞, for which the problem
fails to hold is f (x) = sin x, the series ∞ n=0 f (0) being divergent. It is
interesting to note that if f does not have a power series expansion at 0, or
the Maclaurin series has radius of convergence R = 0, then the problem
fails to hold. The function f (x) = e−1/x , when x = 0 and f (0) = 0, has
2
1
(n) (0) = 0 = ∞ e−x− x 2 dx.
f (n) (0) = 0 and ∞ n=0 f 0
(b) A generalization. Similarly, one can prove that if f is a function which has
aMaclaurin series expansion with radius of convergence R = ∞ such that
∞ (n) (x) converges absolutely, then ∞ e−t f (x + t)dt converges
n=0 f 0
and
∞ ∞
f (n) (x) = e−t f (x + t)dt.
n=0 0
7.144.
(a) Use a similar technique to that in the solution of Problem 7.143, with f replaced
by 12 (f (x) + f (−x)) (the even part of f ).
(b) Use a similar technique to that in the solution of Problem 7.143, with f replaced
by 12 (f (x) − f (−x)) (the odd part of f ).
References
1. Andreescu, T.: Problem U457, Mathematical Reflections 5, 2018, 3

2. Andreoli, M.: Problem 819, Coll. Math. J. 37 (1), 2006, 60
3. Baranenkov, G., Chostak, R., Démidovitch, B., et al., Démidovitch, B (éd.): Recueil
d’exercices et de problèmes d’analyse mathématique, Cinquième édition, Éditions Mir,
Moscou, 1974
4. Bataille, M.: Solution to problem 3965, Crux Mathematicorum 41 (7), 2015, 311–313
5. Becker, M.: Problem 874, Coll. Math. J. 39 (2), 2008, 154
6. Berndt, B.C.: Ramanujan’s Notebooks, Part 1, Springer-Verlag, New York, 1985
7. Biler, P., Witkowski, A.: Problems in Mathematical Analysis, Marcel Dekker, Inc., New York
and Basel, 1990
8. Borwein, D., Borwein, J.M.: On an intriguing integral and some series related to ζ (4), Proc.
Amer. Math. Soc. 123 (4), 1995, 1191–1198
9. Borwein, J.M., Borwein, P.B.: Pi and the AGM, John Wiley & Sons, New York, 1987
10. Bromwich, T.J.I’A.: An Introduction to the Theory of Infinite Series, Third Edition, American
Mathematical Society, Providence, Rhode Island, 1991
11. Castellanos, D.: The ubiquitous π , Math. Mag. Part I, 61 (2), 1988, 67–98; Part II, 61 (3),
1988, 148–163
12. Chen, H.: A new ratio test for positive monotone series, Coll. Math. J. 44 (2), 2013, 139–141
13. Comtet, L.: Advanced Combinatorics. The Art of Finite and Infinite Expansions, D. Reidel
Publishing Company, Dordrecht, 1974
14. De Doelder, P.J.: On some series containing ψ(x) − ψ(y) and (ψ(x) − ψ(y))2 for certain
values of x and y, J. Comput. Appl. Math. 37, 1991, 125–141
15. de Souza, P. N., Silva, J.-N.: Berkeley Problems in Mathematics, Third Edition, Springer,
New York, 2004
16. Dutta, R.: Solution to problem H-790, Fibonacci Q. 56 (2), 2018, 190–191
17. Edgar, T.: Proof without words: Sums of reciprocals of binomial coefficients, Math. Mag. 89
(3), 2016, 212–213
18. Eggleton, R., Kustov, V.: The product and the quotient rules revisited, Coll. Math. J. 42 (4),
2011, 323–326
19. Eydelzon, A.: Problem 12186, Amer. Math. Monthly 127 (5), 2020, 462
20. Fanego, B.S.: Problem 355, Soluciones, Gac. R. Soc. Mat. Esp. 22 (3), 2019, 539–540 http://
gaceta.rsme.es/abrir.php?id=1543
21. Finch, S.R.: Mathematical Constants. Encyclopedia of Mathematics and Its Applications 94,
Cambridge University Press, New York, 2003
22. Flajolet, P., Salvy, B.: Euler sums and contour integral representations, Exp. Math. 7 (1), 1998,
15–35
23. Freitas, P.: Integrals of polylogarithmic functions, recurrence relations, and associated Euler
sums, Math. Comp. 74, 2005, 1425–1440
in Mathematics, https://doi.org/10.1007/978-3-030-77139-3
530 References
24. Furdui, O.: From Lalescu’s sequence to a Gamma function limit, Austral. Math. Soc. Gaz. 35
(5), 2008, 339–344
25. Furdui, O.: Series involving products of two harmonic numbers, Math. Mag. 84 (5), 2011,
371–377
26. Furdui, O.: Limits, Series, and Fractional Part Integrals. Problems in Mathematical Analysis,
Springer, New York, 2013
27. Furdui, O.: The evaluation of a quadratic and a cubic series with trigamma function, Applied
Math. E Notes, 15, 2015, 187–196
28. Furdui, O.: Two surprising series with harmonic numbers and the tail of ζ (2), Gazeta Ma-
tematică, Seria A, 33 (112) (1-2), 2015, 1–8 https://ssmr.ro/gazeta/gma/2015/gma1-2-2015-
continut.pdf
29. Furdui, O.: A note on problem 7 of day 2 of IMC 2015, Gazeta Matematică, Seria A, 33 (112)
(1-2), 2015, 41–44 https://ssmr.ro/gazeta/gma/2015/gma1-2-2015-continut.pdf
30. Furdui, O.: Harmonic series with polygamma functions, J. Classical Anal. 8 (2), 2016, 123–
130
31. Furdui, O.: Problem 11810, Amer. Math. Monthly 122 (1), 2015, 75
32. Furdui, O.: Problem 11941, Amer. Math. Monthly 123 (9), 2016, 942
33. Furdui, O.: Problem 893, Coll. Math. J. 40 (1), 2009, 55–56
34. Furdui, O.: Solution to problem 874, Coll. Math. J. 40 (2), 2009, 137–138
35. Furdui, O.: Problem 1042, Coll. Math. J. 46 (1), 2015, 61
36. Furdui, O.: Problem 1117, Coll. Math. J. 49 (1), 2018, 60
37. Furdui, O.: Problem 4010, Crux Mathematicorum 41 (1), 2015, 28
38. Furdui, O.: Problem 158, EMS Newsletter March 2016, 62
39. Furdui, O.: Problem H-790, Fibonacci Q. 54 (2), 2016, 186
40. Furdui, O.: Problem 341, Gac. R. Soc. Mat. Esp. 21 (1), 2018, 110
41. Furdui, O.: Problem 346, Gac. R. Soc. Mat. Esp. 21 (2), 2018, 331
42. Furdui, O.: Problem 433, Gazeta Matematică, Seria A, 33 (112) (1-2), 2015, 45 https://ssmr.
ro/gazeta/gma/2015/gma1-2-2015-continut.pdf
45. Furdui, O.: Problem Q1014, Math. Mag. 84 (4), 2011, 297
46. Furdui, O.: Problem Q1040, Math. Mag. 87 (2), 2014, 152
47. Furdui, O.: Problem 53, Problems, Mathproblems, 2 (4), 2012, 119 http://www.
mathproblems-ks.org/?wpfb_dl=16
48. Furdui, O.: Problem 58, Problems, Mathproblems, 3 (1), 2013, 118 http://www.
49. Furdui, O.: Problem 5306, Sch. Sci. Math. 114 (4), 2014, 2
55. Furdui, O., Ivan, M., Sîntămărian, A.: A note on a UTCN SEEMOUS selection test problem,
Gazeta Matematică, Seria A, 34 (113) (3-4), 2016, 34–39 https://ssmr.ro/gazeta/gma/2016/
gma3-4-2016-continut.pdf
56. Furdui, O., Ivan, D.M., Sîntămărian, A.: Problem 11982, Amer. Math. Monthly 124 (5), 2017,
465
57. Furdui, O., Ivan, D.M., Sîntămărian, A.: Problem 1097, Coll. Math. J. 48 (2), 2017, 138
58. Furdui, O., Sîntămărian, A.: Quadratic series involving the tail of ζ (k), Integral Transforms
Spec. Funct. 26 (1) 2015, 1–8
References 531
59. Furdui, O., Sîntămărian, A.: Exotic series with fractional part function, Gazeta Matematică,
Seria A, 35 (114) (3-4), 2017, 1–12 https://ssmr.ro/gazeta/gma/2017/gma3-4-2017-continut.
pdf
60. Furdui, O., Sîntămărian, A.: A new proof of the quadratic series of Au–Yeung, Gazeta Ma-
tematică, Seria A, 37 (116), (1-2), 2019, 1–6 https://ssmr.ro/gazeta/gma/2019/gma1-2-2019-
continut.pdf
61. Furdui, O., Sîntămărian, A.: Pearls of quadratic series, Gazeta Matematică, Seria A, 38 (117),
62. Furdui, O., Sîntămărian, A.: Problem 12012, Amer. Math. Monthly 124 (10), 2017, 971
69. Furdui, O., Sîntămărian, A.: Problem 1154, Coll. Math. J. 50 (3), 2019, 224
70. Furdui, O., Sîntămărian, A.: Problem 1163, Coll. Math. J. 50 (5), 2019, 379
71. Furdui, O., Sîntămărian, A.: Problem 3965, Crux Mathematicorum 40 (7), 2014, 300
72. Furdui, O., Sîntămărian, A.: Problem 4368, Crux Mathematicorum 44 (7), 2018, 303
73. Furdui, O., Sîntămărian, A.: Problem 308, Gac. R. Soc. Mat. Esp. 19 (3), 2016, 590
77. Furdui, O., Sîntămărian, A.: Problem 474, Gazeta Matematică, Seria A, 36 (115) (1-2), 2018,
48 https://ssmr.ro/gazeta/gma/2018/gma1-2-2018-continut.pdf
80. Furdui, O., Sîntămărian, A.: Solution to problem 474, Gazeta Matematică, Seria A, 37 (116)
81. Furdui, O., Sîntămărian, A.: Problem 2068, Math. Mag. 91 (2), 2019, 151
82. Furdui, O., Sîntămărian, A.: Problem 825, The Pentagon 77 (2), 2018, 29
83. Furdui, O., Sîntămărian, A.: Problem 5504, Sch. Sci. Math. 118 (5), 2018, 2
87. Furdui, O., Sîntămărian, A.: Problem 5558, Sch. Sci. Math. 119 (6), 2019, 1–2
92. Furdui, O., Vălean, C.: Evaluation of series involving the product of the tail of ζ (k) and
ζ (k + 1), Mediterr. J. Math. 13 2016, 517–526
93. Garaschuk, K. (ed.): Problem CC95, Crux Mathematicorum 39 (9), 2013, 393
94. Garnir, H.G.: Fonctions de variables réelles, Tomme I, Librairie Universitaire, Lauvain &
Gauthiers-Villars, Paris, 1963
95. Gavrea, I., Ivan, M.: On an extension of Pólya–Szegö formula, Mediterr. J. Math. 13 (5),
2016, 3409–3416
96. Gelbaum, B.R., Olmsted, J.M.H.: Theorems and Counterexamples in Mathematics, Springer-
Verlag, New York, 1990
97. Grigorieva, E.: Methods of Solving Sequences and Series Problems, Birkhäuser/Springer,
Cham, 2016
532 References

98. Holland, F.: limm→∞ m k=0 (k/m) = e/(e − 1), Math. Mag. 83 (1), 2010, 51–54
m
99. Ioachimescu, A.G.: Problem 16, Gazeta Matematică 1 (2), 1895, 39

100. Jeffrey, A., Dai, H-H.: Handbook of Mathematical Formulas and Integrals, 4th ed., Elsevier
Academic Press, Amsterdam, 2008
101. Kotronis, A., Kouba, O.: Problem 53, Solutions, Mathproblems, 3 (1), 2013, 126–128 http://
www.mathproblems-ks.org/?wpfb_dl=9
102. Krijan, I.: Problem 6, IMC, Day 2, Blagoevgrad, Bulgaria, 30 July 2015
103. Lalescu, T.: Problem 579, Gazeta Matematică 6 (6), 1900-1901, 148
104. Levin, V.I.: About a problem of S. Ramanujan (in Russian), Uspekhi Mat. Nauk. 5 (3), 1950,
161–166
105. Levy, M.: Problem 5406, Solution 3, Solutions, Sch. Sci. Math. 116 (8), 2016, 17–18
106. Loya, P.: Amazing and Aesthetic Aspects of Analysis, Springer, New York, 2017
107. Mabry, R.: Problem 893, Problems and Solutions, Coll. Math. J., 41 (1), 2010, 67–69
108. Martínez Fernández, A.R.: Solution to problem 387, Gac. R. Soc. Mat. Esp. 24 (1), 2021,
136–137
109. Mező, I.: Nonlinear Euler sums, Pac. J. Math., 272 (1), 2014, 201–226
110. Mincu, G., Pop, V., Rus, M.: Traian Lalescu national mathematics contest for university
students, 2018 edition, Gazeta Matematică, Seria A, 36 (115) (3-4), 2018, 24–37 https://ssmr.
111. Mortici, C.: A power series approach to some inequalities, Amer. Math. Monthly 119 (2),
2012, 147–151
112. Nahin, P.J.: Inside Interesting Integrals, Springer, New York, 2015
113. Papacu, N.: Problem 27079, Gazeta Matematică, Seria B, 120 (5), 2015, 269
114. Pârşan, L.: Problema 17263, Gazeta Matematică, Seria B, 83 (6), 1978, 250
115. Perfetti, P.: Problem 3815, Crux Mathematicorum 39 (2), 2013, 92
116. Perfetti, P.: Solution to problem 58, Mathproblems, 3 (2), 2013, 142–143 http://www.
117. Plaza, Á.: Proof without words: Arctangent of two and the golden ratio, Math. Mag. 90 (3),
2017, 179
118. van der Poorten, A.J.: Some wonderful formulae. . . Footnotes to Apéry’s proof of the
irrationality of ζ (3), Séminaire Delange-Pisot-Poitou. Théorie des nombres, tome 20 (2),
1978-1979, 1–7
119. Pop, V., Furdui, O.: Square Matrices of Order Two. Theory, Applications, and Problems,
Springer, Cham, 2017
120. Popa, D.: Problem C: 1789, Gazeta Matematică, Seria B, 101 (2), 1996, 124
121. Pólya, G., Szegö, G.: Problems and Theorems in Analysis I. Series. Integral Calculus. Theory
of Functions, Springer-Verlag, Berlin, 1998
122. Rădulescu, T.-L.T., Rădulescu, V.D., Andreescu, T.: Problems in Real Analysis: Advanced
Calculus on the Real Axis, New York, 2009
123. Ricardo, H.: Solution to problem 1154, Coll. Math. J. 51 (3), 2020, 231
124. Rizzoli, I.: O teoremă Stolz–Cesaro, Gazeta Matematică, Seria B, 95 (10-11-12), 1990, 281–
284
125. Sandham, H.F.: Advanced Problem 4305, Amer. Math. Monthly 55 (7), 1948, 431
126. Senum, G.I.: Problem E 3352, Amer. Math. Monthly 96 (9), 1989, 838
127. Singh, J.: A noninductive proof of de Moivre’s formula, Amer. Math. Monthly 125 (1), 2018,
80
128. Sîntămărian, A.: Probleme selectate cu şiruri de numere reale (in Romanian), Editura U. T.
Press, Cluj-Napoca, 2008
129. Sîntămărian, A.: A Generalization of Euler’s Constant, Editura Mediamira, Cluj-Napoca,
2008
130. Sîntămărian, A.: Some convergent sequences and series, Int. J. Pure Appl. Math. 57 (6), 2009,
885–902
131. Sîntămărian, A.: Some applications of a limit problem, Gazeta Matematică, Seria B, 115 (11),
2010, 517–520
References 533
132. Sîntămărian, A.: Solution to problem CC95, Crux Mathematicorum 40 (9), 2014, 372–373
133. Sîntămărian, A.: Problem 11528, Amer. Math. Monthly 117 (8), 2010, 742
134. Sîntămărian, A.: Problem 1083, Coll. Math. J. 47 (4), 2016, 301
135. Sîntămărian, A.: Problem 3760, Crux Mathematicorum 38 (6), 2012, 243, 244
136. Sîntămărian, A.: Problem 266, Gac. R. Soc. Mat. Esp. 18 (1), 2015, 121
137. Sîntămărian, A.: Problem C: 3226, Gazeta Matematică, Seria B, 112 (10), 2007, 554, 556
138. Sîntămărian, A.: Problem Q1030, Math. Mag. 86 (2), 2013, 148
139. Sîntămărian, A., Furdui, O.: Problem Q1055, Math. Mag. 88 (5), 2015, 378
140. Sondow, J.: Double integrals for Euler’s constant and ln π4 and an analog of Hadjicostas’s
formula, Amer. Math. Monthly 112 (1), 2005, 61–65
141. Stadler, A.: Solution to problem 5306, Sch. Sci. Math. 114 (7), 2014, 11-13
142. Stoica, G.: Problem 1991, Math. Mag. 89 (2), 2016, 147
143. Trif, T.: Probleme de calcul diferenţial şi integral în Rn (in Romanian), Casa Cărţii de Ştiinţă,
Cluj-Napoca, 2003
144. Ulrich, A.: Solution to problem 346, Gac. R. Soc. Mat. Esp. 22 (2), 2019, 318–319 http://
gaceta.rsme.es/abrir.php?id=1517
145. Vălean, C.I.: Problem 5406, Sch. Sci. Math. 116 (5), 2016
146. Vălean, C.I., Furdui, O.: Reviving the quadratic series of Au–Yeung, J. Class. Anal. 6 (2)
2015, 113-118
147. Vălean, C.I., Levy, M.: Euler sum involving tail, ResearchGate, 2019
148. Weisstein, E. W.: Apéry’s constant http://mathworld.wolfram.com/AperysConstant.html
149. Weisstein, E. W.: Riemann Zeta Function zeta(2) http://mathworld.wolfram.com/
RiemannZetaFunctionZeta2.html
150. Whittaker, E.T., Watson, G.N.: A Course of Modern Analysis, Fourth Edition, University
Press, Cambridge, 1927
151. Woeginger, G.: Problem 3, IMC, Day 2, Blagoevgrad, Bulgaria, 31 July 2011
152. Xu, C.: A unique area property of the quadratic function, Math. Mag. 87 (1), 2014, 52–56
Index
A F
Abel’s summation formula, 38 Fibonacci sequence, 9
A ζ (3) series, 200 A fractional part identity, 493
A series of Goldbach, 201, 239
G
B
Generating function, 93
Bernoulli’s inequality, 496
the Bernoulli numbers, 95
Botez–Catalan identity, 53
the Fibonacci sequence, 95
the harmonic number On , 94
C the Lucas sequence, 96
Catalan integral, 353 the nth harmonic number Hn , 93
Cauchy–d’Alembert’s criterion, 8 the sequence (cos n)n≥1 , 97
Cauchy product, 93 the sequence (En )n≥0 , 95
Cauchy’s integral criterion, 50 the sequence (sin n)n≥1 , 96
Chain rule, 159 the skew-harmonic number Hn− , 95
Cloitre’s factorial series, 206 Glaisher–Kinkelin constant, 206
Cubic series, 223, 224 Gram determinant, 465
D H
de Moivre’s formula, 123 Homogeneous function, 165
Derivable function by a direction, 171 Hurwitz zeta function, 229
Differential of a function of several variables,
170
Differential operators
I
the divergence, 158
Ioachimescu’s constant, 12
the gradient, 158
Irrational numbers, 138
the Laplacian, 158
the rotation or the curl, 158
Digamma function, 229
Dirichlet’s criterion, 493 L
Lagrange Mean Value Theorem for real
functions of two real variables, 168
E Lagrange’s Mean Value Theorem
Euler’s constant, 11 for a real function of one variable, 137
Euler’s formula, 123 Legendre’s formula, 203
Euler’s identity, 165 Leibniz’s criterion for alternating series, 52
Euler’s series, 29 Leibniz’s formula, 133
in Mathematics, https://doi.org/10.1007/978-3-030-77139-3
536 Index
Local extremum points S

local maximum, 172 Series with Apéry’s constant, 201
local minimum, 172 Series with four factorials, 208
one real variable function, 148 Series with the floor function, 70
several real variables function, 172 Series with three factorials, 208
Logarithmic integral, 328 Sophomore’s dream, 90, 525
Sophomore’s dream for series, 90
Stirling’s formula, 36
M Stolz–Cesàro Theorem
Maclaurin formula the 00 case, 11
for elementary functions, 138 the ∞
one real variable function, 136 ∞ case, 10
Sylvester’s criterion, 173
two real variables function, 168
Maclaurin inequalities, 150
Matrix
Hessian matrix, 157 T
Jacobian matrix, 156 Tannery’s Theorem for Series, 274
Taylor’s formula
one real variable function, 136
N two real variables function, 166
Nonlinear Euler sum, 200 with the remainder in integral form,
137
Tetralogarithm Li4 , 385
O
The ∞-norm on Rn , 46
Oloa’s harmonic series, 54
The p-norm on Rn , 46
The generalized Leibniz formula, 135
Q Traian Lalescu’s sequence, 9
Quadratic form, 172
indefinite, 172
negative definite, 172 U
positive definite, 172 Uniformly continuous function, 208
Quadratic series, 225, 228
R W
Ramanujan’s series, 32 Wallis formula, 41
Riemann’s series, 51 Weierstrass product formula, 318
Riemann zeta function, 13 Wolstenholme’s limit, 14

Sharpening Mathematical Analysis Skills: Alina Sîntămărian Ovidiu Furdui

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sharpening Mathematical Analysis Skills: Alina Sîntămărian Ovidiu Furdui

Uploaded by

Copyright:

Available Formats

Problem Books in Mathematics

ISSN 0941-3502 ISSN 2197-8506 (electronic)

The library is the friend that all those who

in a natural way: either as generalizations or motivated by known results that have

This book is addressed to undergraduate students with a strong background in

Cluj-Napoca, Romania Alina Sîntămărian

Part I Theory and Problems

1 Sequences of Real Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

4.5 Taylor’s Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136

9 Sequences of Real Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253

9.3 Wolstenholme Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270

14.3 Implicit Functions of One Real Variable Defined by a

N The set of natural numbers (N = {1, 2, 3, . . .})

e limn→∞ (1 + 1/n) = 2.71828 . . .

γ limn→∞ (1 + 1/2 + · · · + 1/n − lnnn) = 0.57721 ...

Hn The nth harmonic number Hn = 1 + 1/2 + · · · + 1/n, n ∈ N

ζ (2, a) The Hurwitz zeta function ζ (2, a) = n=0 1/(n + a) , a =

{1}, or Lin (z) = 0 Lin−1 (t)/tdt, n ∈ N \ {1, 2}

C k (Ω, R) The set of all functions f : Ω ⊆ Rn → R which have partial

If we prove that γ is irrational do you give us

This chapter collects problems on sequences of real numbers. The problems

1.1 Limits of Sequences

The Constant e The sequence of rational numbers (en )n≥1 defined by

is strictly increasing and bounded, hence convergent. Its limit is denoted by

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 3

1.4. Let (xn )n≥1 be the sequence defined by xn = 1

Some Algebraic Identities If a, b ∈ R and n ∈ N, n ≥ 2, then the following

• a n − bn = (a − b)(a n−1 + a n−2 b + · · · + abn−2 + bn−1 );

1.5. Let k, m ∈ N, k, m ≥ 2. Calculate

Three Sums of Natural Numbers If n ≥ 1 is an integer, then:

1.8. Let {x} be the fractional part of x. Prove that:

Remark 1.2. If k ∈ N, k ≥ 2, and (xn )n∈N is a bounded sequence of real numbers

1.12. [32] Prove that:

1.13. [73] Prove that:

1.14. [A. Sîntămărian, 2016] Calculate

(a) Let (xn )n≥1 be the sequence defined by

1.16. [128, pp. 82, 83], [130, Remark 4] A Challenging Limit

(qn + 1)(qn + p + 1) · · · (qn + np + 1)

(b) If k ∈ N, k ≥ 2, study the convergence of the sequence (xn )n≥0 , with x0 ≥ 0,

Cauchy–d’Alembert’s Criterion Let (an )n≥1 be a sequence of positive real

1.23. [103] Traian Lalescu’s Sequence Calculate

The Fibonacci Sequence

[117] An Equality with the Golden Section

1.24. Two special sums of binomial coefficients. Prove that:

(a) C0n + C2n+1 + C4n+2 + · · · + C2n

1.25. Compositions of homographic functions.

(a) Let f : (0, ∞) → R, f (x) = 3x+2

1.2 Applications of Stolz–Cesàro Theorem, the ∞/∞ and the

(a) 0 < b1 < b2 < . . . < bn < . . . and lim bn = ∞;

This theorem, the discrete version of l’Hôpital’s Theorem, is mentioned in

1.28. Let (xn )n≥1 , with x1 > 0, be the sequence

(a) lim an = lim bn = 0;

1.31. Euler’s Constant Let (xn )n≥1 be the sequence defined by

The constant γ = 0.5772156649 . . . is called Euler’s constant, being

lim n(2Hn − Hn2 − γ ).

(b) A generalization. Let k ≥ 2 be an integer. Calculate

1.33. Let Hn be the nth harmonic number and let On = 1 + 13 + · · · + 2n−1

1.35. Ioachimescu’s Constant Let (xn )n≥1 be the sequence defined by

√ that (xn )n≥1 converges and, denoting by I its limit, calculate

The constant I = 0.539645491 . . . is called Ioachimescu’s constant.

founders of the famous Romanian magazine Gazeta Matematică, has pro-

1.38. [D. M. Bătineţu-Giurgiu, 1991]

1.39. Let x ∈ R, x > 1.

where a is the floor of a.

where a is the floor of a.