An Introduction To The Harmonic Series and Logarithmic Integ1

An Introduction to
The Harmonic Series

And Logarithmic Integrals
For High School Students

Up To Researchers
Ali Shadhar Olaikhan

Copyright © 2021 by Ali Shadhar Olaikhan.
All rights reserved. No part of this book may be reproduced, distributed, or trans-
mitted in any form or by any means, including photocopying, recording, or other
electronic or mechanical methods, without the prior written permission of the pub-
lisher, except as provided by United States of America copyright law. For permission
requests, e-mail the publisher at harmonicseries2021@gmail.com.
First edition published April 2021
Book cover designed by Islam Farid and Aqil Almosawi
LATEX class prepared by Elio A. Farina
ISBN 978-1-7367360-1-2 (eBook)
To my parents
Preface
The harmonic series and logarithmic integrals, which are strongly interrelated, are
not commonly found in the standard textbooks. Evaluating them can be challenging
to new learners, as it requires specific approaches and a good knowledge of special
functions such as the gamma function, the polygamma function, the beta function,
the polylogarithm function, and various other special functions and constants. It
also requires a lot of experience and patience, since it involves plenty of tricks and
time-consuming calculations.
The purpose of this book is to introduce the harmonic series in a way suitable for all
readers with a good knowledge of calculus, from high school students to researchers.
The book is the result of over five years of working on the harmonic series. As
I taught myself this topic, I struggled to find the proofs for most of the identities
required for evaluating the harmonic series. With the experience gained over years,
I managed to prove these identities in detail using only basic definitions and well–
known techniques, and without using contour integration or the residue theorem,
which require a deep understanding of complex analysis.
I would like to inform the reader that I borrowed a few proofs from some sites, mainly
from the Mathematics Stack Exchange site, adding more details and modifying them
my own way. Also, most of the text is written in equations, so the reader won’t find
much unnecessary verbiage in this book.
The book consists of four chapters. Chapter 1 presents some essential series trans-
formations and special functions and shows how these functions are related to each
other. It explains the definition and properties of each function and also derives many
special values needed for the calculations in chapters 3 and 4.
In chapter 2, the reader will find the derivations of plenty of useful identities: gen-
erating functions involving the harmonic number and series expansion of powers of
arcsin(x). Other identities are derived using the beta function, the Cauchy product,
Abel’s summation, and Fourier series.
Chapter 3 prepares all the integral results required to calculate the harmonic series in
chapter 4, including some new results. These were derived using algebraic identities,
integral manipulations and the beta function.
Chapter 4 shows how to calculate many types of harmonic series: non-alternating
series, alternating series, series with powers of 2 in the denominator, series with
powers of 2n + 1 in the denominator, series with rational argument, series with skew
iv
Preface v
harmonic number, series with central binomial coefficient, and many others. Several
solutions are presented using two different methods.
At the end of the book, I have provided a table of Mathematica commands for
approximating or evaluating limits, derivatives, integrals, and series, so that the
reader can verify any result of interest throughout the book.
More advanced and challenging problems about the harmonic series may be found
on my Mathematics Stack Exchange page, https://math.stackexchange.com
/users/432085/ali-shadhar. I decided not to include them in the book for the
sake of simplicity. To keep up to date with any new identities or results, you can
follow my Facebook group, Harmonic Series, https://www.facebook.com/gro
ups/178723409566339.
Finally, I would like to express my gratitude to my friend Cornel Ioan Vălean for
being a big motivation for me to explore the realm of the harmonic series through his
amazing problems and solutions, many of which are included in his book, (Almost)
Impossible Integrals, Sums, and Series, and for his valuable tips for writing this book.
I would also like to thank Elio Arturo and my brother Hasan Shadhar for their help in
using LaTeX. I extend my gratitude to my friends, Khalaf Ruhemi, Shivam Sharma,
and Hasan Hussein for all the support and encouragement they offered me while
writing this book. I also want to thank my parents, to whom I am dedicating this
book, for all their support.
Phoenix, Arizona, USA Ali Shadhar Olaikhan

April 2021
Contents
1 Series Transformations and Special Functions 1

1.1 Shifting the Sum Index . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Reversing the Order of the Sum Terms . . . . . . . . . . . . . . . . 2
1.3 Splitting a Sum Into its Odd and Even Parts . . . . . . . . . . . . . 3
1.4 Converting the Summand a2n to an . . . . . . . . . . . . . . . . . 4
1.5 Converting the Summand a2n+1 to an . . . . . . . . . . . . . . . . 4
1.6 Converting the Summand (−1)n a2n to in an . . . . . . . . . . . . . 5
1.7 Converting the Summand (−1)n a2n+1 to in an . . . . . . . . . . . 6
1.8 Converting a Sum to a Product . . . . . . . . . . . . . . . . . . . . 6
1.9 Double Summation . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.10 The Logarithm of a Complex Number . . . . . . . . . . . . . . . . 8
1.11 Gamma Function . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.11.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.11.2 Functional Equation . . . . . . . . . . . . . . . . . . . . . 10
1.11.3 Stirling’s Approximation . . . . . . . . . . . . . . . . . . . 11
1.11.4 Expressing Gamma Function as a Product . . . . . . . . . . 12
1.11.5 Euler’s Definition as an Infinite Product . . . . . . . . . . . 13
1.11.6 Euler’s Reflection Formula . . . . . . . . . . . . . . . . . . 14
1.11.7 Legendre Duplication Formula . . . . . . . . . . . . . . . . 16
1.12 Beta Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.12.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.12.2 Trigonometric Integral Representation . . . . . . . . . . . . 18
1.12.3 Improper Integral Representation . . . . . . . . . . . . . . 18
1.12.4 Powerful Integral Representation . . . . . . . . . . . . . . . 18
1.13 Riemann Zeta Function . . . . . . . . . . . . . . . . . . . . . . . . 19
1.13.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.13.2 Integral Representation . . . . . . . . . . . . . . . . . . . . 19
1.13.3 Evaluation of ζ(0) . . . . . . . . . . . . . . . . . . . . . . 20
1.13.4 Evaluation of ζ(2) . . . . . . . . . . . . . . . . . . . . . . 21
1.13.5 Evaluation of ζ(2n) . . . . . . . . . . . . . . . . . . . . . 22
1.14 Dirichlet Eta Function . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.14.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
vi
Contents vii

1.15 Dirichlet Beta Function . . . . . . . . . . . . . . . . . . . . . . . . 26
1.15.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1.15.3 Evaluation of β(2a) . . . . . . . . . . . . . . . . . . . . . 27
1.15.4 Evaluation of β(2a + 1) . . . . . . . . . . . . . . . . . . . 29
1.16 Polylogarithm Function . . . . . . . . . . . . . . . . . . . . . . . . 30
1.16.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
1.16.2 Dilogarithm Reflection Formula . . . . . . . . . . . . . . . 34
1.16.3 Landen’s Dilogarithm Identity . . . . . . . . . . . . . . . . 35
1.16.4 Dilogarithm Inversion Formula . . . . . . . . . . . . . . . . 36
1.16.5 Relation Involving Four Dilogarithm Functions . . . . . . . 36
1.16.6 Another Relation Involving Dilogarithm Functions . . . . . 37
1.16.7 Landen’s Trilogarithm Identity . . . . . . . . . . . . . . . . 37
1.16.8 Polylogarithm Inversion Formula . . . . . . . . . . . . . . . 39
1.17 Harmonic Number . . . . . . . . . . . . . . . . . . . . . . . . . . 40
1.17.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
1.17.2 Infinite Series Representation . . . . . . . . . . . . . . . . 43
1.18 Skew Harmonic Number . . . . . . . . . . . . . . . . . . . . . . . 44
1.18.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
1.18.2 Infinite Series Representation . . . . . . . . . . . . . . . . 46
1.19 Digamma Function . . . . . . . . . . . . . . . . . . . . . . . . . . 47
1.19.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
1.19.2 Digamma Reflection Formula . . . . . . . . . . . . . . . . 47
1.19.3 Digamma–Harmonic Number Identity . . . . . . . . . . . . 48
1.20 Polygamma Function . . . . . . . . . . . . . . . . . . . . . . . . . 50
1.20.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
1.20.2 Series Representation . . . . . . . . . . . . . . . . . . . . . 50
1.20.4 Evaluation of ψ (a) (1). . . . . . . . . . . . . . . . . . . . . 52
1.20.5 Evaluation of ψ (a) 12 . . . . . . . . . . . . . . . . . . . . 52
1.20.6 Evaluation of ψ (2a) 14

. . . . . . . . . . . . . . . . . . . 53

. . . . . . . . . . . . . . . . . . . 53
1.21 Catalan’s Constant . . . . . . . . . . . . . . . . . . . . . . . . . . 54
1.22 Euler–Mascheroni Constant . . . . . . . . . . . . . . . . . . . . . . 55
2 Generating Functions and Powerful Identities 56

2.1 Generating Functions . . . . . . . . . . . . . . . . . . . . . . . . . 57
P∞ (a)
2.1.1 Pn=1 Hn xn . . . . . . . . . . . . . . . . . . . . . . . . 57
∞ Hn n
2.1.2
P∞n=1 n x . . . . . . . . . . . . . . . . . . . . . . . . . 59
Hn n
2.1.3 n=1 n2 x . . . . . . . . . . . . . . . . . . . . . . . . . 59
P∞ Hn(2) n
2.1.4 n x . . . . . . . . . . . . . . . . . . . . . . . . 60
Pn=1
∞ 2 (2) n
2.1.5 n=1 (Hn − Hn )x . . . . . . . . . . . . . . . . . . . . 61
viii Contents
2 (2)
P∞ Hn −Hn
2.1.6 n=1 n xn . . . . . . . . . . . . . . . . . . . . . . 62
P∞ Hn2 n
2.1.7
Pn=1 n x . . . . . . . . . . . . . . . . . . . . . . . . . 63
∞ Hn n
2.1.8 n=1 n3 x . . . . . . . . . . . . . . . . . . . . . . . . . 63
P∞ Hn(2) n
2.1.9 n=1 n2 x . . . . . . . . . . . . . . . . . . . . . . . . 67
P∞ Hn(3) n
2.1.10 Pn=1 n x . . . . . . . . . . . . . . . . . . . . . . . . 67
∞ 3 n
2.1.11 H n x . . . . . . . . . . . . . . . . . . . . . . . . . 68
Pn=1
∞ Hn 2
n
2.1.12 n=1 n2 x . . . . . . . . . . . . . . . . . . . . . . . . . 68
P∞ (2) n
2.1.13 Hn Hn x . . . . . . . . . . . . . . . . . . . . . . 69
Pn=1
∞ 3 (2) (3)
2.1.14 n=1 (H n − 3H H
n n + 2Hn )xn . . . . . . . . . . . . . 70
P∞ Hn Hn(2) n
2.1.15 n=1 x . . . . . . . . . . . . . . . . . . . . . . . 70
P∞ Hn3 n n
2.1.16 n=1 n x . . . . . . . . . . . . . . . . . . . . . . . . . 72
P∞
2.1.17 Pn=1 (Hn − 6Hn Hn + 8Hn Hn(3) + 3(Hn(2) )2 − 6Hn(4) )xn
4 2 (2)
. . 72
∞
2.1.18 H n xn . . . . . . . . . . . . . . . . . . . . . . . . . 73
Pn=1
∞ Hn n
2.1.19 n x . . . . . . . . . . . . . . . . . . . . . . . . . 74
Pn=1
∞ Hn n
2.1.20 Pn=1 n2 x . . . . . . . . . . . . . . . . . . . . . . . . . 75
∞
n n . . . . . . . . . . . . . . . . . . . . . . .
2.1.21 n=1 H 2 x . . 78
P∞ Hn/2 n
2.1.22 n x . . . . . . . . . . . . . . . . . . . . . . . . 79
Pn=1
∞ Hn/2 n
2.1.23 n=1 n2 x . . . . . . . . . . . . . . . . . . . . . . . . 80
P∞ (2n n) n
2.1.24 n=1 4n Hn x . . . . . . . . . . . . . . . . . . . . . . . 82
P∞ (2n n ) Hn n
2.1.25 n=1 4n n x . . . . . . . . . . . . . . . . . . . . . . . 83
2n
P∞ ( n ) Hn n
2.1.26
Pn=1 4n n2 x . . . . . . . . . . . . . . . . . . . . . . . 84
∞ 2H2n −Hn 2n
2.1.27 n=1 n x . . . . . . . . . . . . . . . . . . . . . 85
P∞ H2n 2n+1
2.1.28 n=1 x . . . . . . . . . . . . . . . . . . . . . . 86
P∞ 2n+1
(−1)n H2n 2n+1
2.1.29 x . . . . . . . . . . . . . . . . . . . . 87
Pn=1
∞
2n+1
Hn −H2n 1
2n
2.1.30 n=1 n − 2n2 x . . . . . . . . . . . . . . . . . 88
2.2 Series Expansion of Powers of arcsin(z) . . . . . . . . . . . . . . . 89
2.2.1 Series Expansion of arcsin(z) . . . . . . . . . . . . . . . . 89
2.2.2 Series Expansion of arcsin(z)
√
1−z 2
. . . . . . . . . . . . . . . . . 90
2.2.3 Series Expansion of arcsin3 (z) . . . . . . . . . . . . . . . 91
2.2.4 Series Expansion of arcsin4 (z) . . . . . . . . . . . . . . . 92
2.3 Identities by Beta Function . . . . . . . . . . . . . . . . . . . . . . 95
2.3.1 Expressing Beta Function as a Product . . . . . . . . . . . . 95
2.3.2 Evaluation of Four Logarithmic Integrals . . . . . . . . . . 95
2.4 Identities by Cauchy Product . . . . . . . . . . . . . . . . . . . . . 99
2.4.1 Cauchy Product of Two Power Series . . . . . . . . . . . . 99
2.4.2 Cauchy Product of − ln(1 − x) Li2 (x) . . . . . . . . . . . . 99
2.4.3 Cauchy Product of Li22 (x) . . . . . . . . . . . . . . . . . . 100
Contents ix
2.4.5 Cauchy Product of Li2 (x) Li3 (x) . . . . . . . . . . . . . . 102

2.4.6 Cauchy Product of Li23 (x) . . . . . . . . . . . . . . . . . . 103
2.5 Identities by Abel’s Summation . . . . . . . . . . . . . . . . . . . . 105
2.5.1 Abel’s Summation . . . . . . . . . . . . . . . . . . . . . . 105
2.5.2 First Application . . . . . . . . . . . . . . . . . . . . . . . 107
2.5.3 Second Application . . . . . . . . . . . . . . . . . . . . . . 108
2.5.4 Third Application . . . . . . . . . . . . . . . . . . . . . . . 110
2.6 Identities By Fourier Series . . . . . . . . . . . . . . . . . . . . . . 112
2.6.1 Fourier Series . . . . . . . . . . . . . . . . . . . . . . . . . 112
2.6.2 Fourier Series of Even Function . . . . . . . . . . . . . . . 114
2.6.3 Fourier Series of Odd Function . . . . . . . . . . . . . . . 115
2.6.4 Fourier Series of cos(zx) . . . . . . . . . . . . . . . . . . . 116
2.6.5 Fourier Series of sin(zx) . . . . . . . . . . . . . . . . . . . 117
2.6.6 Fourier Series of ln(sin x) . . . . . . . . . . . . . . . . . . 118
2.6.7 Fourier Series of ln(cos x) . . . . . . . . . . . . . . . . . . 121
2.6.8 Fourier Series of ln(tan x) . . . . . . . . . . . . . . . . . . 122
π
2.6.9 Series Representation of sin(πz) . . . . . . . . . . . . . . . 122
2.6.10 Series Representation of cot(πz) . . . . . . . . . . . . . . . 122
2.6.11 Euler’s Product Formula of sin(πz) . . . . . . . . . . . . . 123
2.6.12 Series Representation of sec π2 z . . . . . . . . . . . . . . 124
2.6.13 Series Representation of sin(x) . . . . . . . . . . . . . . . 125
2.6.14 Series Representation of tan x ln(sin x) . . . . . . . . . . . 125
2.6.15 Series Representation of ln2 (2 cos x) . . . . . . . . . . . . 126
3 Logarithmic Integrals 128

3.1 Generalized Logarithmic Integrals . . . . . . . . . . . . . . . . . . 128
R 1 lna (x)
3.1.1 dx . . . . . . . . . . . . . . . . . . . . . . . . . . 128
R01 ln1−x
a
(x)
3.1.2 0 1+x
dx . . . . . . . . . . . . . . . . . . . . . . . . . . 129
R 1 lna ( 1−x
1+x )
3.1.3 0 x dx . . . . . . . . . . . . . . . . . . . . . . . . 130
R 1 ln( 1−x
1+x ) ln a−1
(x)
3.1.4 dx . . . . . . . . . . . . . . . . . . . . 131
R01 lna (1−x)x
3.1.5 0 1+x dx . . . . . . . . . . . . . . . . . . . . . . . . 131
R 12 lna (x)
3.1.6 dx . . . . . . . . . . . . . . . . . . . . . . . . . 132
R01 ln1−x
a
(1+x)
3.1.7 0 x dx . . . . . . . . . . . . . . . . . . . . . . . . 134
R 1 ln2a−1 ( 1−x x
)
3.1.8 1+x dx . . . . . . . . . . . . . . . . . . . . . . 135
R0∞ lna (1+x)
3.1.9 2 dx . . . . . . . . . . . . . . . . . . . . . . . . 136
R01 lna1+x
(1−x)
3.1.10 0 1+x2 dx . . . . . . . . . . . . . . . . . . . . . . . . 137
R ∞ 2a
3.1.11 0 ln1+x(x) 2 dx . . . . . . . . . . . . . . . . . . . . . . . . . 137
R ∞ Lia (−x)
3.1.12 0 1+x2 dx . . . . . . . . . . . . . . . . . . . . . . . . 139
R1
3.1.13 0 Li2a+1 1+x2
(−x)
dx . . . . . . . . . . . . . . . . . . . . . . . 140
x Contents
R1 ln2a (x) ln(1+x)

3.1.14 0 1+x2 dx . . . . . . . . . . . . . . . . . . . . . 141
R ∞ Lia (−x2 )
3.1.15 0 1+x2 dx . . . . . . . . . . . . . . . . . . . . . . . . 142
R 1 Li2a+1 (−x2 )
3.1.16 1+x2 dx . . . . . . . . . . . . . . . . . . . . . . . 143
R01 ln2a (x) arctan(x)
3.1.17 1−x2 dx . . . . . . . . . . . . . . . . . . . . . 144
R0∞ ln2a (x) ln(1+x)
3.1.18 0
√
x(1+x)
dx . . . . . . . . . . . . . . . . . . . . . 145
R 1 ln2a (x) ln(1+x2 )
3.1.19 2 dx . . . . . . . . . . . . . . . . . . . . . 147
R01 lna (x)1+x
lna (1−x)
3.1.20 0 x(1−x) dx . . . . . . . . . . . . . . . . . . . . . 147
R 12 lna (x) lna (1−x)
3.1.21 x(1−x) dx . . . . . . . . . . . . . . . . . . . . . 148
R01 lna (x) ln(1−x)
3.1.22 dx . . . . . . . . . . . . . . . . . . . . . . 149
R01 lna (x)1−x
ln(1−x)
3.1.23 x(1−x) dx . . . . . . . . . . . . . . . . . . . . . . 149
R01 lna (x) ln(1+x)
3.1.24 dx . . . . . . . . . . . . . . . . . . . . . . 150
R01 lna (x)1+x
ln(1+x)
3.1.25 x(1+x) dx . . . . . . . . . . . . . . . . . . . . . . 150
R01 lna (1−x) ln(1+x)
3.1.26 0 x dx . . . . . . . . . . . . . . . . . . . . 151
R 1 lna (x) ln( 1+x2 )
3.1.27 1−x dx . . . . . . . . . . . . . . . . . . . . . . 151
R01 lna (1−x) Li2 (x)
3.1.28 dx . . . . . . . . . . . . . . . . . . . . . 152
R0∞ ln2a−1x(x) ln(1+x)
3.1.29 x(1+x) dx . . . . . . . . . . . . . . . . . . . . 153
R01 x2n
3.1.30 dx . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
R01 1+x
xn
3.1.31 dx . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
R01 1+x2n−1
3.1.32 x arctanh(x)dx . . . . . . . . . . . . . . . . . . . 156
R01 n−1
3.1.33 0
x Li a (x)dx . . . . . . . . . . . . . . . . . . . . . . 156
3.2 Results of Logarithmic Integrals . . . . . . . . . . . . . . . . . . . 158
R π2
3.2.1 sin(2nx) cot(x)dx . . . . . . . . . . . . . . . . . . . . 158
R0π2
3.2.2 ln(sin x)dx . . . . . . . . . . . . . . . . . . . . . . . . 159
R0π2 2
3.2.3 ln (sin x)dx . . . . . . . . . . . . . . . . . . . . . . . 161
R0π2
3.2.4 ln(sin x) ln(cos x)dx . . . . . . . . . . . . . . . . . . . 162
R01 ln(x) ln(1−x)
3.2.5 0
√
x 1−x
dx . . . . . . . . . . . . . . . . . . . . . . 163
R ∞ ln2 (x) ln(1+x2 )
3.2.6 1+x2 dx . . . . . . . . . . . . . . . . . . . . . 165
R01 ln(1−x) ln(1+x)
3.2.7 x dx . . . . . . . . . . . . . . . . . . . . . 167
R01 ln(x) ln(1−x) ln(1+x)
3.2.8 dx . . . . . . . . . . . . . . . . . . . 168
R01 ln(1−x) lnx2 (1+x)
3.2.9 dx . . . . . . . . . . . . . . . . . . . . . 169
R01 ln2 (1−x)x ln(1+x)
3.2.10 0 dx . . . . . . . . . . . . . . . . . . . . . 170
R 1 ln3 (1−x)x ln(1+x)
3.2.11 0 dx . . . . . . . . . . . . . . . . . . . . . 170
R 1 ln(1−x) xln3 (1+x)
3.2.12 0 dx . . . . . . . . . . . . . . . . . . . . . 172
R 1 ln3 (1+x)x ln(x)
3.2.13 0 x dx . . . . . . . . . . . . . . . . . . . . . . 172
Contents xi
R1 ln(x) ln(1+x)
3.2.14 1−x dx . . . . . . . . . . . . . . . . . . . . . . 174
R01 ln(x) ln(1−x)
3.2.15 1+x dx . . . . . . . . . . . . . . . . . . . . . . 175
R01 ln(x) ln2 (1−x) R 1 2 ln(1−x)
3.2.16 0 1+x dx & 0 ln (x)1+x dx . . . . . . . . . . 175
R1 ln2 (1+x) R1 R1
3.2.17 0 1+x2
dx, 0 ln(1−x) ln(1+x)
1+x2
dx, & 0 ln(x)1+x
ln(1+x)
2 dx . . . 177
R1 ln(x) ln2 (1−x)
3.2.18 0
√
x(1−x)
dx . . . . . . . . . . . . . . . . . . . . . . 179
1
R ln2 (x) ln(1−x)
3.2.19 2
1−x dx . . . . . . . . . . . . . . . . . . . . . . 180
R01 ln2 (x) ln(1+x)
3.2.20 1+x dx . . . . . . . . . . . . . . . . . . . . . . 181
R01 ln3 (1−x) ln(x)
3.2.21 1
x dx . . . . . . . . . . . . . . . . . . . . . . 181
R 21 Li2 (−x)
3.2.22 1+x2 dx . . . . . . . . . . . . . . . . . . . . . . . . . 183
R01 ln(x) arctan x
3.2.23 1+x dx . . . . . . . . . . . . . . . . . . . . . . 183
R01 ln2 (x) arctan x
3.2.24 0 x(1+x2 ) dx . . . . . . . . . . . . . . . . . . . . . . 184
R1 Li22 (−x)
3.2.25 x dx . . . . . . . . . . . . . . . . . . . . . . . . . 186
R012 Li22 (−x)
3.2.26 x dx . . . . . . . . . . . . . . . . . . . . . . . . . 187
R01 ln2 (1−x) Li2 (x)
3.2.27 x dx . . . . . . . . . . . . . . . . . . . . . 188
R01 ln3 (1−x) Li2 (x)
3.2.28 x dx . . . . . . . . . . . . . . . . . . . . . 188
R01 ln4 (1−x) Li2 (x)
3.2.29 0 x dx . . . . . . . . . . . . . . . . . . . . . 188
4 Harmonic Series 189

4.1 Generalized
P∞Harmonic Series . . . . . . . . . . . . . . . . . . . . . 189
Hn/p
4.1.1 q . . . . . . . . . . . . . . . . . . . . . . . . . . 189
Pn=1
∞ Hn
n
4.1.2 n=1 nq . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
P∞ Hn
4.1.3 nq . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
Pn=1
∞ (−1)n Hn
4.1.4 n=1 n2q . . . . . . . . . . . . . . . . . . . . . . . . 198
P∞ (−1)n H n
4.1.5 n2q . . . . . . . . . . . . . . . . . . . . . . . . 201
Pn=1
∞ Hn/2
4.1.6 n=1 n2q . . . . . . . . . . . . . . . . . . . . . . . . . . 202
P∞ (−1) n
Hn/2
4.1.7 n=1 n2q . . . . . . . . . . . . . . . . . . . . . . . 202
P∞ ζ(q)−Hn(q)
4.1.8 n=1 n . . . . . . . . . . . . . . . . . . . . . . . 203
P∞ Hn(2)
4.1.9 n=1 n2q+1 . . . . . . . . . . . . . . . . . . . . . . . . . 205
P∞ Hn(2q+1)
4.1.10 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . . 207
P∞ 2
Hn
4.1.11 n2q+1 . . . . . . . . . . . . . . . . . . . . . . . . . 208
Pn=1
∞ Hn
4.1.12 n=1 (2n+1) q . . . . . . . . . . . . . . . . . . . . . . . . 210
P∞ (−1)n Hn(q)
4.1.13 n=1 n . . . . . . . . . . . . . . . . . . . . . . . 212
P∞ (−1)n Hn(2q+1)
4.1.14 n=1 2n+1 . . . . . . . . . . . . . . . . . . . . . 213
P∞ (−1) n
Hn
4.1.15 n=1 (2n+1)2q+1 . . . . . . . . . . . . . . . . . . . . . . . 214
4.2 Non–Alternating Harmonic Series . . . . . . . . . . . . . . . . . . 216
xii Contents
P∞ Hn
4.2.1 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
P∞ (2)
Hn
4.2.2 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . . . . 216
P∞ Hn2
4.2.3 n2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
Pn=1
∞ Hn H2n
4.2.4 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . . . 218
P∞ Hn(2)
4.2.5 n=1 n3 . . . . . . . . . . . . . . . . . . . . . . . . . . 220
P∞ Hn(3)
4.2.6 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . . . . 220
P∞ Hn2
4.2.7 n=1 n3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
P∞ Hn Hn(2)
4.2.8 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . . 222
P∞ Hn3
4.2.9 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
P∞ Hn(2)
4.2.10 n=1 n4 . . . . . . . . . . . . . . . . . . . . . . . . . . 225
P∞ Hn2
4.2.11 n=1 n4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
P∞ Hn(4)
4.2.12 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . . . . 226
P∞ (Hn(2) )2
4.2.13 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . . . 227
P∞ Hn Hn(3)
4.2.14 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . . 227
P∞ Hn2 Hn(2)
4.2.15 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . . 228
P∞ Hn4
4.2.16 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
P∞ Hn Hn(2)
4.2.17 n=1 n3 . . . . . . . . . . . . . . . . . . . . . . . . 230
P∞ Hn3
4.2.18 n=1 n3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
P∞ Hn(2)
4.2.19 n=1 n5 . . . . . . . . . . . . . . . . . . . . . . . . . . 232
P∞ Hn2
4.2.20 n=1 n5 . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
P∞ Hn(3)
4.2.21 n=1 n4 . . . . . . . . . . . . . . . . . . . . . . . . . . 234
P∞ Hn(4)
4.2.22 n=1 n3 . . . . . . . . . . . . . . . . . . . . . . . . . . 234
P∞ Hn2 Hn(2)
4.2.23 n=1 n3 . . . . . . . . . . . . . . . . . . . . . . . . 235
P∞ Hn(2)
4.2.24 n=1 n7 . . . . . . . . . . . . . . . . . . . . . . . . . . 236
4.3 Alternating Harmonic Series . . . . . . . . . . . . . . . . . . . . . 238
P∞ (−1)n Hn
4.3.1 . . . . . . . . . . . . . . . . . . . . . . . . 238
Pn=1 n
∞ (−1)n H2n
4.3.2 n=1 n . . . . . . . . . . . . . . . . . . . . . . . 238
P∞ (−1)n Hn
4.3.3 n2 . . . . . . . . . . . . . . . . . . . . . . . . 239
Pn=1
∞ (−1)n H2n
4.3.4 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . 240
P∞ (−1)n Hn(2)
4.3.5 n=1 n . . . . . . . . . . . . . . . . . . . . . . . 240
P∞ (−1)n Hn(3)
4.3.6 . . . . . . . . . . . . . . . . . . . . . . . 241
Pn=1 n
∞ (−1)n Hn
4.3.7 n=1 n3 . . . . . . . . . . . . . . . . . . . . . . . . 241
P∞ (−1)n Hn(2)
4.3.8 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . 243
P∞ (−1)n Hn2
4.3.9 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . . 243
Contents xiii
P∞ (−1)n Hn Hn (2)
4.3.10 n=1 n . . . . . . . . . . . . . . . . . . . . . . 244
P∞ (−1)n Hn3
4.3.11 . . . . . . . . . . . . . . . . . . . . . . . . 245
Pn=1
∞
n
(−1)n Hn
4.3.12 n=1 n4 . . . . . . . . . . . . . . . . . . . . . . . . 245
P∞ (−1)n Hn(2)
4.3.13 n=1 n3 . . . . . . . . . . . . . . . . . . . . . . . 247
P∞ (−1)n Hn2
4.3.14 n=1 n3 . . . . . . . . . . . . . . . . . . . . . . . . 247
P∞ (−1)n Hn(4)
4.3.15 n=1 n . . . . . . . . . . . . . . . . . . . . . . . 248
P∞ (−1)n Hn(3)
4.3.16 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . 249
P∞ (−1)n Hn Hn (2)
4.3.17 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . 249
P∞ (−1)n Hn3
4.3.18 n=1 n2 . .. . . . . . . . . . . . . . . . . . . . . . 251
4.4 Harmonic Series
P∞ Hn with Powers of 2 in the Denominator . . . . . . . 252
4.4.1 n=1 n2 n . . . . . . . . . . . . . . . . . . . . . . . . . . 252
P∞ Hn
4.4.2 n=1 n2 2n . . . . . . . . . . . . . . . . . . . . . . . . . . 252
P∞ Hn(2)
4.4.3 n=1 n2n . . . . . . . . . . . . . . . . . . . . . . . . . . 253
P∞ Hn2
4.4.4 n2n . . . . . . . . . . . . . . . . . . . . . . . . . . 254
Pn=1
∞ Hn
4.4.5 n=1 n3 2n . . . . . . . . . . . . . . . . . . . . . . . . . . 254
P∞ Hn(2)
4.4.6 n=1 n2 2n . . . . . . . . . . . . . . . . . . . . . . . . . . 255
P∞ Hn2
4.4.7 n=1 n2 2n . . . . . . . . . . . . . . . . . . . . . . . . . . 256
P∞ Hn(3)
4.4.8 n2n . . . . . . . . . . . . . . . . . . . . . . . . . . 256
Pn=1
∞ Hn
4.4.9 n=1 n4 2n . . . . . . . . . . . . . . . . . . . . . . . . . . 257
P∞ Hn(4)
4.4.10 n=1 n2n . . . . . . . . . . . . . . . . . . . . . . . . . . 258
P∞ Hn(2)
4.4.11 n=1 n3 2n . . . . . . . . . . . . . . . . . . . . . . . . . . 258
P∞ Hn(3)
4.4.12 n=1 n2 2n . . . . . . . . . . . . . . . . . . . . . . . . . . 259
P∞ Hn2
4.4.13 n=1 n3 2n . . . . . . . . . . . . . . . . . . . . . . . . . . 260
P∞ Hn Hn(2)
4.4.14 n=1 n2 2n . . . . . . . . . . . . . . . . . . . . . . . . 262
P∞ Hn3
4.4.15 n=1 n2 2n . . . . . . . . . . . . . . . . . . . . . . . . . . 265
4.5 Harmonic Series with Powers of 2n + 1 in the denominator . . . . . 266
P∞ (−1)n H2n+1
4.5.1 n=0 . . . . . . . . . . . . . . . . . . . . . . 266
P∞ (−1)2n+1 n
H2n+1
4.5.2 n=0 (2n+1)2 . . . . . . . . . . . . . . . . . . . . . . 266
P∞ (−1)n H2n+1 (2)
4.5.3 . . . . . . . . . . . . . . . . . . . . . . 267
Pn=0
∞ Hn
2n+1
4.5.4 n=1 (2n+1)2 . . . . . . . . . . . . . . . . . . . . . . . . 267
P∞ (−1)n Hn
4.5.5 n=0 (2n+1)2 . . . . . . . . . . . . . . . . . . . . . . . . 268
P∞ (−1)n Hn(2)
4.5.6 n=0 2n+1 . . . . . . . . . . . . . . . . . . . . . . . 269
P∞ (−1) n
H2n+1
4.5.7 n=0 (2n+1)3 . . . . . . . . . . . . . . . . . . . . . . 271
n
P∞ (−1) H2n+1 (2)
4.5.8 n=0 (2n+1)2 . . . . . . . . . . . . . . . . . . . . . . 272
xiv Contents
P∞ (2)
Hn
4.5.9 n=1 (2n+1)2 . . . . . . . . . . . . . . . . . . . . . . . . 272
P∞ 2
Hn
4.5.10 n=1 (2n+1)2 . . . . . . . . . . . . . . . . . . . . . . . . 275
P∞ (2)
Hn
4.5.11 n=1 (2n+1)3 . . . . . . . . . . . . . . . . . . . . . . . . 275
P∞ (3)
Hn
4.5.12 n=1 (2n+1)2 . . . . . . . . . . . . . . . . . . . . . . . . 276
P∞ (3) P∞ (−1)n Hn(3)
Hn
4.5.13 n=1 (2n+1)3 + 4 n=1 n3 . . . . . . . . . . . . . 276
4.6 Skew Harmonic Series
P∞ (−1)n H n . . . . . . . . . . . . . . . . . . . . . . . . 277
4.6.1 n=1 n . . . . . . . . . . . . . . . . . . . . . . . . 277
P∞ Hn
4.6.2 n=1 n3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
P∞ (−1)n H n
4.6.3 n=1 n3 . . . . . . . . . . . . . . . . . . . . . . . . 278
P∞ (−1)n H n Hn
4.6.4 . . . . . . . . . . . . . . . . . . . . . . 279
Pn=1
∞ H n Hn
n
4.6.5 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . . . 280
P∞ (−1)n H n Hn
4.6.6 n2 . . . . . . . . . . . . . . . . . . . . . . 281
Pn=1
∞ H 2n H2n
4.6.7 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . . 284
4.7 Harmonic Series with Rational Argument . . . . . . . . . . . . . . 285
P∞ (−1)n Hn/2
4.7.1 . . . . . . . . . . . . . . . . . . . . . . . 285
Pn=1
∞ Hn/2
n
4.7.2 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . . . . 286
P∞ (−1) n
Hn/2
4.7.3 n=1 n2 . . . . . . . . . . . . . . . . . . . . . . . 286
P∞ Hn/2
4.7.4 n=1 n3 . . . . . . . . . . . . . . . . . . . . . . . . . . 287
P∞ (−1) n
Hn/2
4.7.5 n=1 n3 . . . . . . . . . . . . . . . . . . . . . . . 288
P∞ Hn Hn/2
4.7.6 n=1 n2n . . . . . . . . . . . . . . . . . . . . . . . . 288
P∞ (−1) Hn Hn/2
4.7.7 n=1 2 . . . . . . . . . . . . . . . . . . . . . 289
P∞ (−1)n nHn/2
4.7.8 n=1 n4 . . . . . . . . . . . . . . . . . . . . . . . 290
4.8 Harmonic Series2nwith Binomial Coefficient in the Numerator . . . . 291
P∞ ( n ) Hn
4.8.1 n=1 4n n . . . . . . . . . . . . . . . . . . . . . . . . 291
P∞ (2n n ) (−1) Hn
n
4.8.2 n=1 4n n . . . . . . . . . . . . . . . . . . . . . 292
P∞ (2n n ) Hn
4.8.3 n=1 4n n2 . . . . . . . . . . . . . . . . . . . . . . . . 292
P∞ (2n n ) Hn
(2)
4.8.4 n=1 4n n . . . . . . . . . . . . . . . . . . . . . . . . 293
P∞ (2n n ) H
(2)
4.8.5 n=1 4n n . . . . . . . . . . . . . . . . . . . . .
2n
. . . 294
P∞ (2n n ) Hn
2
4.8.6 n=1 4n n . . . . . . . . . . . . . . . . . . . . . . . . 295
P∞ (2n n ) H 2
4.8.7 n=1 4n n2
n
. . . . . . . . . . . . . . . . . . . . . . . . 297
4.9 Harmonic Series
P∞ 4n Hn with Binomial Coefficient in the Denominator . . . 299
4.9.1 n=1 (2n) n2 . . . . . . . . . . . . . . . . . . . . . . . . 299
P∞ 4nn H2n
4.9.2 n=1 (2n) n2 . . . . . . . . . . . . . . . . . . . . . . . . 301
P∞ 4nn Hn
4.9.3 n=1 (2n) n3 . . . . . . . . . . . . . . . . . . . . . . . . 302
n
Contents xv
(2)
4n Hn
P∞
4.9.4 . . . . . . . . . . . . . . . . . . . . . . . . 304
n=1 ( ) 2
2n
n
n
P∞ n 2
4 Hn
4.9.5 n=1 (2n) n2 . . . . . . . . . . . . . . . . . . . . . . . . 305
P∞ 4nn H2n
4.9.6 n=1 (2n) n3 . . . . . . . . . . . . . . . . . . . . . . . . 307
n
Table of Mathematica Commands 312
References 314
Index 316
Notations
C The set of complex numbers

R The set of real numbers
Z The set of integers (Z = {. . . , −3, −2, −1, 0, 1, 2, 3, . . . })
Z≥0 The set of non-negative integers (Z≥0 = {0, 1, 2, . . . })
Z≤0 The set of non-positive integers (Z≤0 = {. . . , −2, −1, 0})
Z+ The set of positive integers (Z+ = {1, 2, 3, . . . })
Z− The set of negative integers (Z− = {. . . , −3, −2, −1})
R(z) The real part of a complex number z
J(z) The imaginary part of a complex number z
n! n factorial Qn
n! = 1 · 2 · 3 · · · (n − 1) · n = k=1 k, n ∈ Z+
a

b The central binomial coefficient
a
Γ(a+1)
b = Γ(b+1)Γ(a−b+1)
Γ The Gamma
R ∞ function
Γ(z) = 0 tz−1 e−t dt, R(z) > 0
B The Beta function
R1
B(a, b) = 0 xa−1 (1 − x)b−1 dx, R(a) > 0, R(b) > 0
ζ The Riemann zeta function P
∞
ζ(z) = 1 + 21z + 31z + · · · = k=1 k1z , R(z) > 1
η The Dirichlet eta function
P∞ k−1
η(z) = 1 − 21z + 31z − · · · = k=1 (−1) kz , R(z) > 0
β The Dirichlet beta function
P∞ (−1)k
β(z) = 1 − 31z + 51z − · · · = k=0 (2k+1) z, R(z) > 0
Lin The Polylogarithm function
2 3 P∞ k
Lin (z) = z + 2zn + 3zn + · · · = k=1 kzn , |z| ≤ 1
Hn The n-th harmonic number Pn
Hn = 1 + 21 + 13 + · · · + n1 = k=1 k1 , n ∈ Z+
(a)
Hn The n-th harmonic number of order a
(a) Pn
Hn = 1 + 21a + 31a + · · · + n1a = k=1 k1a , n ∈ Z+
Hn The skew harmonic number
n−1 Pn k−1
H n = 1 − 12 + 13 − · · · + (−1)n = k=1 (−1)k , n ∈ Z+
xvi
Notations xvii
ψ The Digamma function

d ln0 (Γ(n))
ψ(n) = dn ln(Γ(n)) = Γ(n)
ψ (a) The Polygamma function
da da+1
ψ (a) (n) = dn a ψ(n) = dna+1 ln(Γ(n))
ln The natural logarithm (loge )

e The base of the natural logarithm
e = limn→∞ (1 + 1/n)n = 2.7182818284590 . . .
γ The Euler–Mascheroni constant
γ = limn→∞ (Hn − ln(n)) = 0.5772156649015 . . .
G The Catalan’s constant P∞ (−1)k
G = 1 − 312 + 512 − · · · = k=0 (2k+1) 2 = 0.9159655941772 . . .
Chapter 1
Series Transformations and

Special Functions
1.1 Shifting the Sum Index
n
X n+c
X
ak = ak−c . (1.1)
k=m k=m+c
Pn
Proof. The index k in k=m ak ranges from m to n:
m ≤ k ≤ n.
Replace k by j − c,
m ≤ j − c ≤ n.
On solving this compound inequality, we get
m + c ≤ j ≤ n + c.
This indicates that if we replace the index k by j − c, the index j will range from
m + c to n + c:
Xn n+c
X
ak = aj−c .
k=m j=m+c
Replace j by k in the latter equality to finish the proof.
1
2 Chapter 1. Series Transformations and Special Functions
Hk
Example 1: Let ak = k+1 and m = 0 then shift the index by −1,
n n+1
X Hk X Hk−1
= .
k+1 k
k=0 k=1
Example 2: Let ak = Hk xk−1 and m = 3 then shift the index by +2,

n
X n−2
X
Hk xk−1 = Hk+2 xk+1 .
k=3 k=1
1.2 Reversing the Order of the Sum Terms
n
X n
X
ak = an−k+m . (1.2)
k=m k=m
Proof. Following the previous proof, we have
m ≤ k ≤ n.
Replace k by n − j + m,
m≤n+m−j ≤n
or
m ≤ j ≤ n.
This shows that if we replace the index k by n − j + m, the index j will range from
m to n as well:
Xn Xn
ak = an−j+m .
k=m j=m
The proof completes on replacing j by k in the latter equality.

This type of transformation reverses the order of the sum terms. To see that, let
m = 1 and n = 4 in (1.2), the LHS sum gives
4
X
ak = a1 + a2 + a3 + a4 ,
k=1
which is equivalent to the RHS sum:

4
X
a5−k = a4 + a3 + a2 + a1 ,
k=1
1.3. Splitting a Sum Into its Odd and Even Parts 3
but in reversed order.

1
Example 1: Put ak = k and m = 1,
n n
X 1 X 1
= . (1.3)
k n−k+1
k=1 k=1
Example 2: Put ak = k 2 and m = 3,

n
X n
X
k2 = (n − k + 3)2 .
k=3 k=3
1.3 Splitting a Sum Into its Odd and Even Parts

∞
X ∞
X ∞
X
an = a2n+1 + a2n . (1.4)
n=1 n=0 n=1
Proof.
∞
X
an = a1 + a2 + a3 + · · ·
n=1
= (a1 + a3 + a5 + · · · ) + (a2 + a4 + a6 + · · · )
X∞ X∞
= a2n+1 + a2n ,
n=0 n=1
and the proof is complete

1
Example 1: Put an = n2 ,
∞ ∞ ∞ ∞ ∞
X 1 X 1 X 1 X 1 1X 1
= + = +
n=1
n2 n=0
(2n + 1)2 n=1 (2n)2 n=0
(2n + 1)2 4 n=1 n2
or
∞ ∞
X 1 4X 1
2
= .
n=1
n 3 n=0 (2n + 1)2
Hn
∞ ∞ ∞
X Hn X H2n+1 X H2n
3
= 3
+ 3
.
n=1
n n=0
(2n + 1) n=1
(2n)
1.4 Converting the Summand a2n to an

∞ ∞ ∞
X 1X 1X
a2n = an + (−1)n an . (1.5)
n=1
2 n=1 2 n=1
Proof. Starting with the RHS,

∞
X ∞
X
an + (−1)n an = a1 + a2 + a3 + · · · + (−a1 + a2 − a3 + · · · )
n=1 n=1
= 2a2 + 2a4 + 2a6 + · · ·
∞
X
= 2(a2 + a4 + a6 + · · · ) = 2 a2n .
n=1
The proof finalizes on dividing both sides by 2. Following the same approach, we
also find
∞ ∞ ∞
X 1X 1X
a2n = an + (−1)n an . (1.6)
n=0
2 n=0 2 n=0
1
Example 1: Let an = (n+1)4 in (1.5),
∞ ∞ ∞
X 1 1X 1 1 X (−1)n
= + .
n=1
(2n + 1)4 2 n=1 (n + 1)4 2 n=1 (n + 1)4
Hn+1
Example 2: Let an = (n+3)3 in (1.6),
∞ ∞ ∞
X H2n+1 1 X Hn+1 1X Hn+1
3
= 3
+ (−1)n .
n=0
(2n + 3) 2 n=0 (n + 3) 2 n=0 (n + 3)3
1.5 Converting the Summand a2n+1 to an

∞ ∞ ∞
X 1X 1X
a2n+1 = an − (−1)n an . (1.7)
n=0
2 n=1 2 n=1
Proof.
∞
X ∞
X
an − (−1)n an = a1 + a2 + a3 + · · · − (−a1 + a2 − a3 + · · · )
n=1 n=1
= 2a1 + 2a3 + 2a5 + · · ·
1.6. Converting the Summand (−1)n a2n to in an 5
∞
X
= 2(a1 + a3 + a5 + · · · ) = 2 a2n+1 .
n=0
Divide both sides by 2 to complete the proof.

Let’s shift the index of the LHS sum in (1.8) by −1,
∞ ∞ ∞
X 1X 1X
a2n−1 = an − (−1)n an . (1.8)
n=1
2 n=1 2 n=1
Hn
Example 1: Set an = n3 in (1.7),
∞ ∞ ∞
X H2n+1 1 X Hn 1X Hn
3
= 3
− (−1)n 3 .
n=0
(2n + 1) 2 n=1 n 2 n=1 n
1
Example 2: Set an = n4 in (1.8),
∞ ∞ ∞
X 1 1X 1 1 X (−1)n
= − .
n=1
(2n − 1)4 2 n=1 n4 2 n=1 n4
1.6 Converting the Summand (−1)n a2n to in an
∞
X ∞
X
n
(−1) a2n = R in an . (1.9)
n=1 n=1
Proof.
∞
X
in an = ia1 + i2 a2 + i3 a3 + i4 a4 + i5 a5 + i6 a6 + · · ·
n=1
= ia1 − a2 − ia3 + a4 + ia5 − a6 + · · ·
= i(a1 − a3 + a5 − · · · ) + (−a2 + a4 − a6 + · · · )
∞
X ∞
X
=i (−1)n a2n+1 + (−1)n a2n , (1.10)
n=0 n=1
and the proof follows on comparing the real parts of both sides.
xn
∞ ∞
X x2n X
nx
n
(−1)n = R i .
n=1
(2n)3 n=1
n3
Hn+1
∞ ∞
X H2n+1 X Hn+1
(−1)n =R in .
n=1
(2n)2 n=1
n2
1.7 Converting the Summand (−1)n a2n+1 to in an

∞
X ∞
X
(−1)n a2n+1 = J in an . (1.11)
n=0 n=1
Proof. Compare the imaginary parts of both sides of (1.10).

1
Example 1: Let an = n3 ,
∞ ∞
X (−1)n X in
= J .
n=0
(2n + 1)3 n=1
n3
Hn
Example 2: Let an = (n+1)2 ,
∞ ∞
X H2n+1 X Hn
(−1)n = J in .
n=0
(2n + 2)2 n=1
(n + 1)2
1.8 Converting a Sum to a Product

r
X r
Y
ln(an ) = ln an . (1.12)
n=m n=m
Proof.
r
X
ln(an ) = ln(am ) + ln(am+1 ) + · · · + ln(ar )
n=m
r
Y
= ln(am × am+1 × · · · × ar ) = ln an .
n=m
Example 1: Let an = n,
r
X r
Y
ln(n) = ln n = ln(1 × 2 × 3 × · · · × r) = ln(r!).
n=1 n=1
1.9. Double Summation 7
Example 2: Let an = en ,
r r r
Y X X r(r + 1)
ln en = ln(en ) = n= .
n=1 n=1 n=1
2
1.9 Double Summation

∞ X
X m ∞ X
X ∞
a m bn = am bn . (1.13)
m=1 n=1 n=1 m=n
Proof.
∞ X
X m 1
X 2
X 3
X
am bn = a1 bn + a2 bn + a3 bn + · · ·
m=1 n=1 n=1 n=1 n=1
= a1 (b1 ) + a2 (b1 + b2 ) + a3 (b1 + b2 + b3 ) + · · ·
= b1 (a1 + a2 + · · · ) + b2 (a2 + a3 + · · · ) + b3 (a3 + a4 + · · · ) + · · ·
X ∞ X∞ X∞
= b1 am + b2 am + b3 am + · · ·
m=1 m=2 m=3
X∞ ∞
X ∞ X
X ∞
= bn am = am bn ,
n=1 m=n n=1 m=n
and the proof is complete. If we follow the same steps above, we also find
∞ m−1
X X ∞
X ∞
X
am bn = am bn . (1.14)
m=1 n=1 n=1 m=n+1
Example 1: Let am = pm and bn = pn ,

∞ X ∞ ∞
m
!
X X X
m+n n m
p = p p
m=1 n=1 n=1 m=n
{use the geometric series formula for the inner sum asuming |p| < 1}
∞ n ∞
X p 1 X 2 n
= pn = (p )
n=1
1−p 1 − p n=1
{use the geometric series formula again}
2
1 p p2
= 2
= .
1−p 1−p (1 − p)(1 − p2 )
Example 2: Let am = xm and bn = H n ,

∞ X ∞ ∞
m
!
X X X
m m
x Hn = Hn x
m=1 n=1 n=1 m=n
{use the geometric series formula for the inner sum asuming |x| < 1}
∞ n ∞
X x 1 X
= Hn = H n xn
n=1
1 − x 1 − x n=1
{substitute the result from (2.28)}

1 ln(1 + x) ln(1 + x)
= = .
1−x 1−x (1 − x)2
1.10 The Logarithm of a Complex Number
The logarithm of a complex number, z = x + iy, is given by

1 y
ln(x + iy) = ln(x2 + y 2 ) + i arctan , x > 0. (1.15)
2 x
Proof. We begin with converting to polar coordinates,
x + iy = r cos(θ) + ir sin(θ)
= r(cos(θ) + i sin(θ)).
On using Euler’s formula:

eix = cos x + i sin x, (1.16)
which can be proved by expanding cos x, sin x, and ex in Taylor series, we have
x + iy = reiθ .
Take the logarithm of both sides,
ln(x + iy) = ln reiθ = ln(r) + ln eiθ = ln(r) + iθ.

p
The proof completes on substituting r = x2 + y 2 and θ = arctan(y/x).
The constraint x > 0 in (1.15) shows that this rule is valid only when the complex
number is in the first or fourth quadrant of the complex plane.
In general, for positive x and y, where x = y 6= 0 and y 6= 0, we have
1 y
ln(x + iy) = ln(x2 + y 2 ) + i arctan ; (1.17)
2 x
1 h y i
ln(−x + iy) = ln(x2 + y 2 ) + i π − arctan ; (1.18)
2 x
1.10. The Logarithm of a Complex Number 9
1 h y i
ln(−x − iy) = ln(x2 + y 2 ) − i π − arctan ; (1.19)
2 x
1 y
ln(x − iy) = ln(x2 + y 2 ) − i arctan . (1.20)
2 x
Note that (1.19) and (1.20) follows from replacing i by −i in (1.18) and (1.17)
respectively. To prove (1.18), we sum it with its conjugate,
ln(−x + iy) + ln(x + iy) = ln(−x2 − y 2 ) = ln(−1) + ln(x2 + y 2 )
or
ln(−x + iy) = ln(−1) + ln(x2 + y 2 ) − ln(x + iy)

{substitute ln(−1) = iπ and recall the result from (1.17)}
1 y
= iπ + ln(x2 + y 2 ) − ln(x2 + y 2 ) − arctan
2 x
1 2 2
h y i
= ln(x + y ) + i π − arctan ,
2 x
which matches (1.18).
The value ln(−1) = iπ follows from using the identity:
ln(−x) = ln(x) + iπ, x > 0. (1.21)
To show that, set x = π in Euler’s formula in (1.16),
eiπ = −1,
take the log of both sides,

iπ = ln(−1),
add ln(x) to both sides,
ln(x) + iπ = ln(−1) + ln(x) = ln(−x).
Using the rules from (1.17) to (1.20) we find:

1 π
ln(1 + i) = ln(2) + i ; (1.22)
2 4
1 3π
ln(−1 + i) = ln(2) + i ; (1.23)
2 4
1 3π
ln(−1 − i) = ln(2) − i ; (1.24)
2 4
1 π
ln(1 − i) = ln(2) − i ; (1.25)
2 4
π
ln(i) = i ; (1.26)
2
π
ln(−i) = −i . (1.27)
2
1.11 Gamma Function

1.11.1 Definition
The gamma function is defined by

Z ∞
Γ(z) = tz−1 e−t dt, R(z) > 0. (1.28)
0
For a different integral form, set t = −n ln(x) in (1.28),

Z 1
Γ(z) = (−1)z−1 nz xn−1 lnz−1 (x)dx, R(z) > 0. (1.29)
0
Divide both sides of (1.29) by nz Γ(z),

1
(−1)z−1
Z
1
z
= xn−1 lnz−1 (x)dx, R(z) > 0. (1.30)
n Γ(z) 0
On using Γ(z) = (z − 1)! given in (1.33), we obtain

1
(−1)z−1
Z
1
= xn−1 lnz−1 (x)dx, z ∈ Z+ . (1.31)
nz (z − 1)! 0
1.11.2 Functional Equation
One of the key properties of the gamma function is
Γ(z + 1) = zΓ(z), z∈
/ Z≤0 . (1.32)
Proof. Replace z by z + 1 in (1.28),

Z ∞
Γ(z + 1) = tz e−t dt
0
{apply integration by parts (IBP)}
Z ∞
∞
= −tz e−t 0 +z tz−1 e−t dt
| {z } 0
0
= zΓ(z).
1.11. Gamma Function 11
Note that Z ∞
∞
Γ(1) = e−t dt = −e−t 0
= −0 + 1 = 1,
0
and by using Γ(z + 1) = zΓ(z), we see that:
Γ(2) = 1 · Γ(1) = 1 · 1 = 1!
Γ(3) = 2 · Γ(2) = 2 · 1! = 2!
Γ(4) = 3 · Γ(3) = 3 · 2! = 3!
Γ(5) = 4 · Γ(4) = 4 · 3! = 4!.
So in general we have
Γ(z) = (z − 1)!, z ∈ Z+ . (1.33)
1.11.3 Stirling’s Approximation
Stirling’s Approximation is an approximation for factorials:

√ 1
n! ∼ 2πnn+ 2 e−n , (1.34)
n!
where the sign ∼ means lim √ 1 = 1.
n→∞ 2πnn+ 2 e−n
The following proof may be found in [7, p. 277–278]:

Proof. Begin with the definition of the gamma function:
Z ∞
n! = Γ(n + 1) = tn e−t dt
0
n √ o
let t = n + x 2n
r !n
√ n+ 1 −n Z ∞ −x
√
2n 2
= 2n 2 e √n e 1+x dx
− 2 n
( r !n √ 2 )
2

n ln 1+x n
write 1 + x =e
n
√ n+ 1 −n Z ∞ √2 √
n ln 1+x n −x 2n
= 2n 2 e √ e dx.
n
− 2
√ 1
Divide both sides by 2nn+ 2 e−n then let n → ∞, we get
Z ∞ √2 √
n!

n ln 1+x n −x 2n
lim √ n+ 1 = lim √n e dx
n→∞ 2n 2 e −n n→∞ − 2
Z ∞ h √2 √ i
limn→∞ n ln 1+x n −x 2n
= e dx.
−∞
q
2 2x2
To find the remaining limit, let x n = y and so n = y2 ,
" r ! #
2 √ ln(1 + y) − y
lim n ln 1 + x − x 2n = 2x2 lim
n→∞ n y→0 y2
{apply L’Hopital’s rule, since we have the case 0/0}
1
1+y −1 −1
= 2x2 lim = x2 lim = −x2 .
y→0 2y y→0 1 + y
It follows that Z ∞
n! 2
lim √ n+ 12
= e−x dx.
n→∞ 2n e−n −∞
√
This integral is called the Gaussian Integral, which evaluates to π by using the
polar coordinates (see[39]). Thus,
n! √
lim √ 1 = π.
n→∞ 2nn+ 2 e−n
√
Divide both sides by π to finish the proof.
To evaluate the Gaussian Integral in a different way, split the integral at x = 0,
Z ∞ Z 0 Z ∞
−x2 2
e dx = + e−x dx
−∞ −∞ 0
Z 0 Z ∞ Z ∞ Z ∞
2 2 2 2
= e−x dx + e−x dx = e−x dx + e−x dx
−∞ 0 0 0
| {z }
x→−x
∞ √ ∞
1 use (1.41) √
Z Z
−x2 x= y 1
=2 e dx = y − 2 e−y dy = Γ = π.
0 0 2
1.11.4 Expressing Gamma Function as a Product
The gamma function is also expressed as

n
Γ(z + n + 1) Y 1
Γ(z) = , z∈
/ Z≤0 . (1.35)
z z+k
k=1
Proof. By the functional equation in (1.32), we have:
Γ(z + 1) = zΓ(z);
Γ(z + 2) = (z + 1) · Γ(z + 1) = (z + 1) · zΓ(z);

Γ(z + 3) = (z + 2) · Γ(z + 2) = (z + 2) · (z + 1) · zΓ(z).
This can be generalized to
Γ(z + n + 1) = (z + n) · (z + n − 1) · · · · (z + 1) · zΓ(z)
or
n
Γ(z + n + 1) Γ(z + n + 1) Γ(z + n + 1) Y 1
Γ(z) = = Qn = .
z(z + 1) · · · (z + n) z k=1 (z + k) z z+k
k=1
1.11.5 Euler’s Definition as an Infinite Product

Another form of the gamma function is
∞ z
1 Y 1 + k1
Γ(z) = , z∈
/ Z≤0 . (1.36)
z 1 + kz
k=1
Proof. Multiply both sides of (1.35):

n
Γ(z + n + 1) Y 1
Γ(z) =
z z+k
k=1
n!nz
by Γ(z+n+1) , we obtain
n
Γ(z)n!nz 1 Y n!nz
=
Γ(z + n + 1) z z+k
k=1
( n n n n
)
Y Y Y Y
write n! = k then use ak bk = ak bk
k=1 k=1 k=1 k=1
n n
1 Y knz 1 Y nz
= = .
z k+z z 1 + kz
k=1 k=1
Take the limit on both sides letting n 7→ ∞,

n
n!nz 1 Y nz
Γ(z) lim = lim . (1.37)
n→∞ Γ(z + n + 1) z n→∞ 1 + kz
k=1
For the LHS limit, use Stirling’s approximation for n! and Γ(z + n + 1),
√ 1
n!nz 2πnn+ 2 e−n nz
lim = lim √ 1
n→∞ Γ(z + n + 1) n→∞ 2π(n + z)n+z+ 2 e−n−z
1
nn+z+ 2
= ez lim 1
(n + z)n+z+ 2
n→∞
z+ 12 n
z n n
= e lim
n→∞ n + z n+z
z+ 12
n
lim z+ 21
n→∞ n + z z (1)
= ez z n = e = 1.
ez

lim 1 +
n→∞ n
Since
n−1 z n−1 Qn−1 z
Y 1 Y (k + 1)z k=1 (k + 1)
1+ = = Qn−1 z
k kz k=1 k
k=1 k=1
2z · 3z · · · (n − 1)z · nz
= = nz
1z · 2z · · · (n − 1)z
n−1 Qn
Y
k=1 ak
and since ak = , we have
an
k=1
n−1 z Qn z
z
Y 1 k=1 1 + k1
n = 1+ = z .
k=1
k 1 + n1
Substitute this result in the RHS limit,

1 z
n n Qn ∞ z
1 + k1

Y nz Y
k=1 1 + k
Y
lim = lim z = .
n→∞ 1 + kz n→∞ 1 + n1 1 + kz 1 + kz
k=1 k=1 k=1
The proof follows on plugging the two limits in (1.37).
1.11.6 Euler’s Reflection Formula
A well-known relationship is Euler’s reflection formula:

π
Γ(z)Γ(1 − z) = , z∈
/ Z. (1.38)
sin(πz)
Proof. Set a = z and b = 1 − z in (1.51):

∞
xa−1
Z
Γ(a)Γ(b)
B(a, b) = = dx,
Γ(a + b) 0 (1 + x)a+b
we obtain
Z ∞ z−1
x
Γ(z)Γ(1 − z) = dx (1.39)
0 1 +x
Z 1 Z ∞ z−1 Z 1 z−1 Z ∞ z−1
x x x
= + dx = dx + dx
0 1 1 + x 0 1 + x 1 +x
| 1 {z }
x→1/x
Z 1 z−1 Z 1 −z
x x
= dx + dx
0 1+x 0 1+x

1
expand in series in both integrals
1+x
∞ ∞
Z 1 ! Z 1 !
X X
z−1 n n −z n n
= x (−1) x dx + x (−1) x dx
0 n=0 0 n=0
{interchange integration and summation}
∞
X Z 1 X∞ Z 1
n n+z−1 n
= (−1) x dx + (−1) xn−z dx
n=0 0 n=0 0
∞ n ∞ n
X (−1) (−1)
X
= +
n=0
n+z n=0
n−z+1
{seperate the first term of the first sum and shift the index of the second}
∞ ∞ ∞
1 X (−1)n X (−1)n 1 X 2z(−1)n
= + − = −
z n=1 n + z n=1
n−z z n=1 n2 − z 2
{recall the identity in (2.130)}
π
= . (1.40)
sin(πz)
Moreover, setting z = 1/2 in (1.38) yields

1 π
Γ2 = =π
2 sin( π2 )
or
√

1
Γ = π. (1.41)
2
Remark: The interchange of integration and summation used in the proof above is
justified by Lebesgue’s dominated convergence theorem (see [13]):
∞
Z X ∞ Z
X
fn (x)dx = fn (x)dx, (1.42)
X n=1 n=1 X
where X is the integration interval.

This theorem will be frequently used throughout the book where it’s applicable. So
we are not going to reexplain why the interchange is justified if used again.
More results of the gamma function may be found in [38].
1.11.7 Legendre Duplication Formula
The Legendre duplication formula is given by

√
1 2 π Γ(2n)
Γ +n = . (1.43)
2 4n Γ(n)
The following proof may be found in [40]:

Proof. Letting a = b = n in the beta function in (1.47):
Z 1
Γ(a)Γ(b)
B(a, b) = = xa−1 (1 − x)b−1 dx
Γ(a + b) 0
yields
1 1
Γ2 (n)
Z Z
n−1 n−1 x= 1+u 1−2n
= x (1 − x) dx 2
= 2 (1 − u2 )n−1 du
Γ(2n) 0 −1
Z 0 Z 1
= 21−2n + (1 − u2 )n−1 du
−1 0
Z 0 Z 1
1−2n 2 n−1 1−2n
=2 (1 − u ) du +2 (1 − u2 )n−1 du
−1 0
| {z }
u→−u
Z 1 Z 1
1−2n 2 n−1 1−2n
=2 (1 − u ) du + 2 (1 − u2 )n−1 du
0 0
Z 1
1−2n
=2 2(1 − u2 )n−1 du (1.44)
0
√ Z 1
u= x 1
= 21−2n x− 2 (1 − x)n−1 dx
0
{use the defnition of the beta function in (1.47)}

1−2n 1
=2 B ,n
2
{recall the property of the beta function in (1.48)}
2 Γ( 12 )Γ(n)
= .
4n Γ( 12 + n)
√
The proof completes on substituting Γ(1/2) = π given in (1.41).
1.12. Beta Function 17
Further, let’s set u = cos(x) in (1.44),

π
4n Γ2 (n)
Z 2
sin2n−1 (x)dx = . (1.45)
0 4Γ(2n)
By the definition of the central binomial coefficient:

a Γ(a + 1)
= ,
b Γ(b + 1)Γ(a − b + 1)
we have

2n Γ(2n + 1)
=
n Γ2 (n + 1)
{use Γ(z + 1) = zΓ(z) given in (1.32)}
2nΓ(2n) 2Γ(2n)
= 2
=
(nΓ(n)) nΓ2 (n)
or
Γ2 (n) 2
= 2n
.
Γ(2n) n n
Plug this result in (1.45),
π
4n 1
Z 2
sin2n−1 (x)dx = 2n 2n .
(1.46)
0 n
1.12 Beta Function

1.12.1 Definition
The beta function is defined by
Z 1
B(a, b) = xa−1 (1 − x)b−1 dx, R(a) > 0, R(b) > 0. (1.47)
0
One of the key identities of the beta function is the Beta–Gamma identity:
Γ(a)Γ(b)
B(a, b) = . (1.48)
Γ(a + b)

Proof. Using the definition of the gamma function in (1.28), we have
Z ∞ Z ∞
Γ(a)Γ(b) = e−x xa dx e−y y b dy
0 0
{let x = zt and y = z(1 − t)}

Z 1 Z ∞
= e−z (zt)a−1 (z(1 − t))b−1 z dt dz
0 0
Z ∞ Z 1
−z a+b−1 a−1 b−1
= e z dz t (1 − t) dt
0 0
= (Γ(a + b)) (B(a, b)) ,
and the proof follows on dividing both sides by Γ(a + b).

Another identity of the beta function is the symmetry identity:
B(a, b) = B(b, a). (1.49)
To prove it, let 1 − x = y in (1.47),

Z 1 Z 1
B(a, b) = xa−1 (1 − x)b−1 dx = y b−1 (1 − y)a−1 dy = B(b, a).
0 0
1.12.2 Trigonometric Integral Representation
A different integral form of the beta function is

Z π
2
B(a, b) = 2 cos2a−1 (x) sin2b−1 (x)dx, R(a) > 0, R(b) > 0. (1.50)
0
Proof. Make the change of variable x = cos2 (y) in (1.47).
1.12.3 Improper Integral Representation
Another form of the beta function is

Z ∞
xa−1
B(a, b) = dx, R(a) > 0, R(b) > 0. (1.51)
0 (1 + x)a+b
y
Proof : Make the substitution x = 1+y in (1.47).
1.12.4 Powerful Integral Representation
We also have
1
xa−1 + xb−1
Z
B(a, b) = dx, R(a) > 0, R(b) > 0. (1.52)
0 (1 + x)a+b
1.13. Riemann Zeta Function 19
The following proof may be found in [28, p. 72]:

Proof. Let x → 1/x,
1 ∞
xa−1 + xb−1 xa−1 + xb−1
Z Z
dx = dx
0 (1 + x)a+b 1 (1 + x)a+b
1
xa−1 + xb−1
Z
add dx to both sides then divide by 2
0 (1 + x)a+b
1 1 xa−1 + xb−1 1 ∞ xa−1 + xb−1
Z Z
= dx + dx
2 0 (1 + x)a+b 2 1 (1 + x)a+b
Z 1 Z ∞ a−1
1 x + xb−1
= + dx
2 0 1 (1 + x)a+b
1 ∞ xa−1 + xb−1
Z
= dx
2 0 (1 + x)a+b
Z ∞
xa−1 1 ∞ xb−1
Z
1
= dx + dx
2 0 (1 + x)a+b 2 0 (1 + x)a+b
{use the definition of the beta function given in (1.51)}
1 1
= B(a, b) + B(b, a)
2 2
{use B(b, a) = B(a, b) given in (1.49)}
= B(a, b).
1.13 Riemann Zeta Function

1.13.1 Definition
The zeta function is defined by

∞
1 1 X 1
ζ(z) = 1 + z
+ z
+ · · · = , R(z) > 1. (1.53)
2 3 n=1
nz
1.13.2 Integral Representation
The integral form of the zeta function is given by
(−1)z−1 1
lnz−1 (x)
Z
ζ(z) = dx, R(z) > 1. (1.54)
Γ(z) 0 1−x
Proof : Take the summation for both sides of (1.31):

1
(−1)z−1
Z
1
= xn−1 lnz−1 (x)dx
nz Γ(z) 0
over n ≥ 1, we get
∞ ∞ Z
X 1 (−1)z−1 X 1 n−1 z−1
:= ζ(z) = x ln (x)dx
n=1
nz Γ(z) n=1 0
{interchange integration and summation }
∞
!
(−1)z−1 1 z−1
Z X
= ln (x) xn−1 dx
Γ(z) 0 n=1
{use the geometric series formula}
(−1)z−1 1 z−1
Z
1
= ln (x) dx,
Γ(z) 0 1−x
and the proof is finalized.
1.13.3 Evaluation of ζ(0)
1
ζ(0) = − . (1.55)
2
The Following solution may be found in [9]:

Solution Let’s consider (1.79):
Z ∞ z−1
(−1)z 1 lnz (x)
Z
x=e−t 1 t
η(z) = dx = dt.
Γ(z) 0 1 + x Γ(z) 0 1 + et
Substitute η(z) = (1 − 21−z )ζ(z) given in (1.75),

Z ∞ z−1
1 t
ζ(z) = dt
(1 − 21−z )Γ(z) 0 1 + et
 
∞
tz 1 ∞ tz et
 Z 
IBP 1  
=  + dt
(1 − 21−z )Γ(z)  z(1 + e t) z 0 (1 + e t )2 
{z 0}
 
|
0
{write zΓ(z) = Γ(z + 1)}
Z ∞
1 tz et
= dt.
(1 − 2 )Γ(z + 1) 0 (1 + et )2
1−z
Next, take the limit on both sides and let z → 0,

∞
∞
et
Z
1 1
lim ζ(z) = ζ(0) = − dt = =− .
z→0 0 (1 + et )2 1 + et 2
0
1.13.4 Evaluation of ζ(2)
π2
ζ(2) = . (1.56)
6
The following solution may be found in [3]:

Solution Squaring both sides of
Z 1
π dx
= ,
4 0 1 + x2
we have
1 1
π2
Z Z
dx dy
=
16 0 1 + x2 0 1 + y2
Z 1Z 1
dy dx
=
0 0 (1 + x2 )(1 + y 2 )
Z 1Z x
t=xy dt dx
=
0 0 x(1 + + t2 /x2 )x2 )(1
{change the order of integration}
1 1 1
Z Z
dx dt
=
2 0 t x(1 + x2 )(1 + t2 /x2 )
√ Z 1 Z 1
x= u 1 du
= 2
dt
2 0 t2 (1 + u)(u + t )

1 1 1 1
write = −
(1 + u)(u + t2 ) 1 − t2 u + t2 1+u
Z 1 Z 1 Z 1
1 1 du du
= − dt
2 0 1 − t2 t2 u + t2 t2 1 + u
1 1 1
Z
u=1
= 2
ln(u + t2 ) − ln(1 + u) u=t2 dt
2 0 1−t
1 1 1 1 + t2 2t2
Z
= ln − ln dt
2 0 1 − t2 2 1 + t2
1 1 1 (1 + t2 )2
Z
= ln dt
2 0 1 − t2 4t2

1+t2 1+x2
Z 1 ln 2t 1 t= 1−x
1+x
Z 1 ln 1−x2
= dt = dx
0 1− t2 2 0 x

Z 1 ln 1+y 1−y 2
√
1 1 ln (1−y)2
Z
x= y 1 1−y
= dy = dy
4 0 y 4 0 y
Z 1
ln(1 − y 2 ) 1 1 ln(1 − y)
Z
1
= dy − dy
4 0 y 2 0 y
| {z }
y 2 →y
1
ln(1 − y) 1 1 ln(1 − y)
Z Z
1
= dy − dy
8 0 y 2 0 y
3 1 ln(1 − y)
Z
=− dy
8 0 y
{expand ln(1 − y) in series}
∞
!
3 11 yn
Z X
=− − dy
8 0 y n=1
n
{interchange intergation and summation}
∞ ∞
3 X 1 1 n−1
Z
3X 1 3
= y dy = 2
= ζ(2).
8 n=1 n 0 8 n=1 n 8
π2 π2
So we have 16 = 38 ζ(2) or ζ(2) = 6 .
1.13.5 Evaluation of ζ(2n)
The following generalization is derived by Rob Johnson (see [14]):

n
(−1)n 2n X ζ(2n − 2k)π 2k
ζ(2n) = − π − (−1)k , n ∈ Z+ . (1.57)
2(2n)! (2k + 1)!
k=1
Proof. His proof starts with letting
ζ(2n)
fn = −2 . (1.58)
π 2n
Multiply both series by x2n then consider the summation over n ≥ 0,
∞ ∞
X X ζ(2n) 2n
fn x2n = −2 x
n=0 n=0
π 2n
{separate the first term then expand ζ(2n) in series}
∞ ∞
X x2n X 1
= −2ζ(0) − 2
n=1
π 2n k 2n
k=1
{write ζ(0) = −1/2 given in (1.55) and change the order of summations}
∞ ∞ n !
X X x2
=1−2
n=1
π2 k2
k=1
∞ x2 ∞
X
π 2 k2 x2
X
=1−2 2 =1−2
1 − πx2 k2 π 2 k 2 − x2
k=1 k=1
n x o
to get this sum, set z = in (2.131)
π
1 x
=1−2 − cot x .
2 2
As a result, we have
∞
X
fn x2n = x cot x. (1.59)
n=0
sin x
Multiply both sides of (1.59) by x ,
∞
sin x X
cos x = fn x2n .
x n=0
Expanding cos x and sin x in Taylor series:

∞ ∞
X (−1)n X (−1)n 2n+1
cos x = x2n , sin x = x
n=0
(2n)! n=0
(2n + 1)!
yields
∞ ∞ ∞
X (−1)n X (−1)n X
x2n = fn x2n
n=0
(2n)! n=0
(2n + 1)! n=0
{make use of the Cauchy product in (2.81)}
(−1)n

where = an and fn = bn
(2n + 1)!
∞ n
!
X X (−1)k
= fn−k x2n .
n=0
(2k + 1)!
k=0
Separate the first term of the inner sum,

∞ ∞ n
!
X (−1)n 2n
X X
k fn−k
x = fn + (−1) x2n . (1.60)
n=0
(2n)! n=0
(2k + 1)!
k=1
Compare the coefficients of x2n in both sides of (1.60),

n
(−1)n X fn−k
= fn + (−1)k
(2n)! (2k + 1)!
k=1
or
n
(−1)n X fn−k
fn = − (−1)k . (1.61)
(2n)! (2k + 1)!
k=1
Substitute (1.61) in (1.58),

n
ζ(2n) (−1)n X ζ(2n − 2k)
−2 2n
= +2 (−1)k ,
π (2n)! (2k + 1)!π 2n−2k
k=1
2n
and the proof finishes on multiplying both sides by − π2 .
Examples
π2
ζ(2) = ;
6
π4
ζ(4) = ;
90
π6
ζ(6) = ;
945
π8
ζ(8) = ;
9450
π 10
ζ(10) = ;
93555
691π 12
ζ(12) = .
638512875
Also note the following relations:
π4 5
ζ 2 (2) = = ζ(4); (1.62)
36 2
π6 35
ζ 3 (2) = = ζ(6); (1.63)
216 8
π6 7
ζ(2)ζ(4) = = ζ(6); (1.64)
4500 4
1.14. Dirichlet Eta Function 25
π8 7
ζ 2 (4) = = ζ(8); (1.65)
8100 6
π8 5
ζ(2)ζ(6) = = ζ(8); (1.66)
5670 3
π 10 33
ζ(2)ζ(8) = = ζ(10); (1.67)
56700 20
π 10 11
ζ(4)ζ(6) = = ζ(10); (1.68)
85050 10
π 12 875875
ζ 6 (2) = = ζ(12); (1.69)
46656 44224
π 12 7007
ζ 3 (4) = = ζ(12); (1.70)
729000 5528
π 12 715
ζ 2 (6) = = ζ(12); (1.71)
893025 691
12
π 3003
ζ(4)ζ(8) = = ζ(12); (1.72)
850500 2764
π 12 2275
ζ(2)ζ(10) = = ζ(12). (1.73)
561330 1382
1.14 Dirichlet Eta Function
1.14.1 Definition
The Dirichlet eta function is defined by

∞
X (−1)n−1
η(z) = , R(z) > 0. (1.74)
n=1
nz
1
This function is related to the zeta function. To show that, let an = nz in (1.5),
∞ ∞ ∞
X 1 X 1 X (−1)n
2 = +
n=1
(2n)z n=1
nz n=1 nz
∞ ∞ ∞
X 1 X 1 X (−1)n−1
21−z = −
n=1
nz n=1
nz n=1 nz
21−z ζ(z) = ζ(z) − η(z).
Rearrange the terms,

η(z) = (1 − 21−z )ζ(z). (1.75)
Examples
1
η(2) = ζ(2); (1.76)
2
3
η(3) = ζ(3); (1.77)
4
7
η(4) = ζ(4). (1.78)
8
The integral representation of eta function is
(−1)z−1 1
lnz−1 (x)
Z
η(z) = dx, R(z) > 0. (1.79)
Γ(z) 0 1+x

1
(−1)z−1
Z
1
z
= xn−1 lnz−1 (x)dx.
n Γ(z) 0
by (−1)n−1 then consider the summation over n ≥ 1,

∞ ∞ Z 1
X (−1)n−1 (−1)z−1 X
z
:= η(z) = (−1) n−1
xn−1 lnz−1 (x)dx
n=1
n Γ(z) n=1 0

∞
!
(−1)z−1 1 z−1
Z X
n−1
= ln (x) (−x) dx
Γ(z) 0 n=1
(−1)z−1 1 lnz−1 (x)
Z
= dx.
Γ(z) 0 1+x
1.15 Dirichlet Beta Function

1.15.1 Definition
The Dirichlet beta function is defined by
∞
X (−1)n 1 1
β(z) = z
= 1 − z + z − ··· , R(z) > 0. (1.80)
n=0
(2n + 1) 3 5
1.15. Dirichlet Beta Function 27
The integral form of the Dirichlet beta function is
(−1)z−1 1
lnz−1 (x)
Z
β(z) = dx, R(z) > 0. (1.81)
Γ(z) 0 1 + x2
Proof. By (1.30), we have

1
(−1)z−1
Z
1
= x2n lnz−1 (x)dx.
(2n + 1)z Γ(z) 0
Multiply both sides by (−1)n then consider the summation over n ≥ 0,

∞ ∞ Z 1
X (−1)n (−1)z−1 X
z
:= β(z) = (−1) n
x2n lnz−1 (x)dx
n=0
(2n + 1) Γ(z) n=0 0

∞
!
(−1)z−1 1 z−1
Z X
= ln (x) (−x2 )n dx
Γ(z) 0 n=0
(−1)z−1 1 lnz−1 (x)
Z
= dx.
Γ(z) 0 1 + x2
1.15.3 Evaluation of β(2a)
For a ∈ Z+ , we have
21−4a

1
β(2a) = − ψ (2a−1) − (1 − 2−2a )ζ(2a). (1.82)
(2a − 1)! 4
Proof. Replace z by 2a in (1.80),
∞
X (−1)n
β(2a) =
n=0
(2n + 1)2a
∞ ∞ ∞
( )
X X X
n
use (−1) an = 2 a2n − an given in (1.6)
n=0 n=0 n=0
∞ ∞
X 1 1X
=2 2a
− . (1.83)
n=0
(4n + 1) n=0
(2n + 1)2a
Evaluation of the first sum:

∞ ∞
X 1 1 X 1
2a
= 2a 2a
n=0
(4n + 1) 4 n=0
(n + 1/4)
{make use of the definition in (1.177)}
2−4a

(2a−1) 1
=− ψ . (1.84)
(2a − 1)! 4
1
Evaluation of the second sum: Set an = (n+1)a in (1.6),
∞ ∞ ∞
X 1 1X 1 1 X (−1)n
= +
n=0
(2n + 1)a 2 n=0 (n + 1)a 2 n=0 (n + 1)a
{shift the index n by −1 in both series on the RHS}
∞ ∞
1X 1 1 X (−1)n−1
= +
2 n=1 na 2 n=1 na
1 1
= ζ(a) + η(a)
2 2
1−a
{substitute η(a) = (1 − 2 )ζ(a) given in (1.75)}
= (1 − 2−a )ζ(a).
Thus, we have
∞
X 1
a
= (1 − 2−a )ζ(a). (1.85)
n=0
(2n + 1)
Replace a by 2a in (1.85),
∞
X 1
2a
= (1 − 2−2a )ζ(2a). (1.86)
n=0
(2n + 1)
The proof completes on plugging (1.84) and (1.86) in (1.83).
Examples
∞ 1
ψ (1) 14

(−1)n
Z
X ln(x) 3
β(2) = =− dx = − ζ(2); (1.87)
n=0
(2n + 1)2 0 1+x
2 48 4
∞
1 1 ln3 (x) ψ (3) 14

X (−1)n Z
15
β(4) = 4
=− 2
dx = − ζ(4); (1.88)
n=0
(2n + 1) 6 0 1 + x 768 16
∞
ψ (5) 14
n Z 1 5
X (−1) 1 ln (x) 63
β(6) = 6
=− 2
dx = − ζ(6); (1.89)
n=0
(2n + 1) 120 0 1 + x 245760 64
1.15. Dirichlet Beta Function 29
∞ 1
ln7 (x) ψ (7) 41

(−1)n
Z
X 1 255
β(8) = 8
=− dx = − ζ(8);
n=0
(2n + 1) 5040 0 1 + x2 165150720 256
(1.90)
∞ n Z 1 9
X (−1) 1 ln (x)
β(10) = 10
=− dx
n=0
(2n + 1) 362880 0 1 + x2
ψ (9) 14

1023
= − ζ(10). (1.91)
190253629440 1024
1.15.4 Evaluation of β(2a + 1)
For a ∈ Z≥0 , we have
4−a−1 π d2a
β(2a + 1) = lim1 2a csc(πs). (1.92)
(2a)! s→ 2 ds
Proof. Replace z by 2a + 1 in (1.81) then write Γ(2a + 1) = (2a)!,

∞ ∞
1
ln2a (x) ln2a (x)
Z Z Z
1 1
β(2a + 1) = 2
dx = − dx
(2a)! 0 1+x (2a)! 0 1 1 + x2
Z ∞ 2a Z ∞ 2a
1 ln (x) 1 ln (x)
= dx − dx
(2a)! 0 1 + x2 (2a)! 1 + x2
|1 {z }
x→1/x
∞ 2a
ln2a (x) 1
Z Z
1 ln (x) 1
= dx − dx
(2a)!
0 1 + x2 (2a)! 0 1 + x2
Z 1 2a
1 ln (x)
add β(2a + 1) := dx to both sides then divide by 2
(2a)! 0 1 + x2
Z ∞ 2a
1 ln (x)
= dx
2(2a)! 0 1 + x2
{recall the result from (3.53)}
4−a−1 d2a
= lim1 2a [π csc(πs)].
(2a)! s→ 2 ds
Examples
∞ Z 1
X (−1)n 1 π
β(1) = = 2
dx = ; (1.93)
n=0
2n + 1 0 1 + x 4
∞
(−1)n 1 1 ln2 (x) π3
X Z
β(3) = = dx = ; (1.94)
n=0
(2n + 1)3 2 0 1 + x2 32
∞
(−1)n 1
ln4 (x) 5π 5
Z
X 1
β(5) = 5
= 2
dx = ; (1.95)
n=0
(2n + 1) 24 0 1+x 1536
∞ Z 1 6
X(−1)n 1 ln (x) 16π 7
β(7) = = dx = ; (1.96)
n=0
(2n + 1)7 720 0 1 + x2 184320
∞ Z 1 8
X (−1)n 1 ln (x) 277π 9
β(9) = = dx = . (1.97)
n=0
(2n + 1)9 40320 0 1 + x2 8257536
1.16 Polylogarithm Function

1.16.1 Definition
The polylogarithm function is defined by
∞
X zn z2 z3
Lia (z) = a
= z + a + a + ··· , |z| ≤ 1, (1.98)
n=1
n 2 3
p
where |z| = x2 + y 2 is the modulus of the complex number, z = x + iy.
Let’s discuss the case a = 1:

∞
X zn
Li1 (z) = = − ln(1 − z). (1.99)
n=1
n
This function diverges when |z| = 1 and converges to − ln(2) when z = −1, and so
the case a = 1 is valid when z = −1, but invalid when |z| = 1. Actually the range
of a can be extended to the whole complex plane. To keep it simple, we will consider
only the case a ∈ Z+ , since that is all we need for later calculations.
Note that
∞ ∞
X (1)n X 1
Lia (1) = a
= = ζ(a) (1.100)
n=1
n n=1
na
and
∞
X (−1)n
Lia (−1) = a
= −η(a) = (21−a − 1)ζ(a). (1.101)
n=1
n
Examples
1
Li2 (−1) = − ζ(2); (1.102)
2
3
Li3 (−1) = − ζ(3); (1.103)
4
1.16. Polylogarithm Function 31
7
Li4 (−1) = − ζ(4); (1.104)
8
15
Li5 (−1) = − ζ(5); (1.105)
16
31
Li6 (−1) = − ζ(6). (1.106)
32
It’s also valid to set z = i in (1.98), since |i| = 1 assuming a = 2, 3, 4, · · · ,

∞
X in
Lia (i) =
n=1
na
∞ ∞ ∞
( )
X X X
use f (n) = f (2n + 1) + f (2n) given in (1.4)
n=1 n=0 n=1
∞ ∞
X i2n+1 i2n
X
= a
+
n=0
(2n + 1) n=1
(2n)a
∞ ∞
X (−1)n −a
X (−1)n
=i + 2
n=0
(2n + 1)a n=1
na
{use the definitions in (1.80) and (1.74)}
= iβ(a) − 2−a η(a)
{write η(a) = (1 − 21−a )ζ(a) given in (1.75)}
= iβ(a) + (21−2a − 2−a )ζ(a).
Thus,
Lia (i) = (21−2a − 2−a )ζ(a) + iβ(a). (1.107)
Examples
1 1
Li2 (i) = − ζ(2) + iβ(2) = − ζ(2) + iG; (1.108)
8 8
3 3 π3
Li3 (i) = − ζ(3) + iβ(3) = − ζ(3) + i ; (1.109)
32 32 32
7
Li4 (i) = − ζ(4) + iβ(4). (1.110)
128
π3
In the calculations above, we used β(2) = G and β(3) = 32 given in (1.206) and
(1.94) respectively.
To write the polylogarithm function in integral form, multiply both sides of (1.31):
1
(−1)a−1
Z
1
a
= tn−1 lna−1 (t)dt
n (a − 1)! 0
by z n then take the summation over ≥ 1,

∞ ∞ Z
X zn (−1)a−1 X 1 n n−1 a−1
:= Lia (z) = z t ln (t)dt
n=1
na (a − 1)! n=1 0
{switch integration and summation}
∞
!
(−1)a−1 1 lna−1 (t) X
Z
n
= (zt) dt
(a − 1)! 0 t n=1
a−1 1
lna−1 (t) (−1)a−1 1 z lna−1 (t)
Z Z
(−1) zt
= dt = dt.
(a − 1)! 0 t 1 − zt (a − 1)! 0 1 − zt
Therefore,
(−1)a−1 1
z lna−1 (t)
Z
Lia (z) = dt. (1.111)
(a − 1)! 0 1 − zt
This integral form extends the range of z to the whole complex plane except the real
line for x > 1. In other words, z ∈/ (1, ∞). Note that a 6= 1 when z = 1, since we
R1 1
will have 0 1−t dt, which is a divergent integral.
z
Let’s replace z by z−1 in (1.111),
(−1)a 1
z lna−1 (t)
Z
z
Lia = dt. (1.112)
z−1 (a − 1)! 0 1 − z + zt
Like the integral in (1.111), this integral also extends the range of z to thewhole

z
complex plane except the real line for x > 1 but notice here z 6= 1, since Lia z−1
is undefined for this value deducing that z ∈
/ [1, ∞).
Both of (1.111) and (1.112) can be found in [28, p. 4].
For a different integral form, we begin with
Z z
zn
= tn−1 dt.
n 0
Divide both sides by na−1 then consider the summation over n ≥ 1,

∞ ∞ Z z
X zn X 1
a
:= Lia (z) = a−1
tn−1 dt
n=1
n n=1
n 0

∞
Z z !
1 X tn
= dt
0 t n=1
na−1
{recall the definition in (1.98)}
Z z
Lia−1 (t)
= dt.
0 t
Thus, Z z
Lia−1 (t)
Lia (z) = dt. (1.113)
0 t
This integral form also extends the range of z to the whole complex plane. Note that
R1
a 6= 1 when z = 1, since we will have 0 1t dt, which is a divergent integral.
If we start with the integral representation in (1.111) for both Lia (z) and Lia (−z),
we find
(−1)a−1 1 a−1
Z
z z
Lia (z) + Lia (−z) = ln (t) − dt
(a − 1)! 0 1 − zt 1 + zt
(−1)a−1 1 a−1 2tz 2
Z
= ln (t) dt
(a − 1)! 0 1 − z 2 t2
√ Z 1 2 a−1
t= y 1−a (−1)a−1 z ln (y)
= 2 2
dy
(a − 1)! 0 1−z y
{to get this integral, replace z by z 2 in (1.111)}
= 21−a Lia (z 2 ).
Then, we have
Lia (z) + Lia (−z) = 21−a Lia (z 2 ). (1.114)
This relation extends the range of z to the whole complex plane.
n
For a different approach assuming |z| ≤ 1, put an = nz a in (1.5):
∞
X ∞
X ∞
X
2 a2n = an + (−1)n an ,
n=1 n=1 n=1
we get
∞ ∞ ∞
X z 2n X z n X (−z)n
2 = +
n=1
(2n)a n=1
na n=1 na
∞ ∞ ∞
X (z 2 )n X z n X (−z)n
21−a = +
n=1
na n=1
na n=1 na
21−a Lia (z 2 ) = Lia (z) + Lia (−z).
Examples
1
Li2 (z) + Li2 (−z) = Li2 (z 2 ); (1.115)
2
1
Li3 (z) + Li3 (−z) = Li3 (z 2 ); (1.116)
4
1
Li4 (z) + Li4 (−z) = Li4 (z 2 ). (1.117)
8
Let’s differentiate both sides of (1.98) with respect to z,
∞ ∞ ∞
∂ ∂ X zn X nz n−1 1 X zn Lia−1 (z)
Lia (z) = = = = .
∂z ∂z n=1 na n=1
n a z n=1
n a−1 z
Therefore,
∂ Lia−1 (z)
Lia (z) = . (1.118)
∂z z
1.16.2 Dilogarithm Reflection Formula
For z ∈ C, the following formula holds:
Li2 (z) + Li2 (1 − z) = ζ(2) − ln(z) ln(1 − z). (1.119)
Proof. Differentiate ln(z) ln(1 − z) then integrate,

Z
ln(z) ln(1 − z) = d(ln(z) ln(1 − z))
Z
ln(z) ln(1 − z)
= − + dz
1−z z
= − Li2 (1 − z) − Li2 (z) + C.
To find C, take the limit on both sides and let z → 0,
lim ln(z) ln(1 − z) = − Li2 (1) − Li2 (0) + C.

z→0
Since lim ln(z) ln(1 − z) = 0, we have C = Li2 (1) = ζ(2) and the proof is
z→0
complete. To show the limit is 0, we write
ln(1 − z)

lim ln(z) ln(1 − z) = lim z ln(z) lim ,
z→0 z→0 z→0 z
where, by using L’Hopital’s rule, the first limit is 0 and the second limit is −1.
If we put z = 1/2 in (1.119), we obtain

1 1
Li2 + Li2 = ζ(2) − ln2 (2)
2 2
or
1 1 1
Li2 = ζ(2) − ln2 (2). (1.120)
2 2 2
Furthermore, set z = i in (1.119),
Li2 (i) + Li2 (1 − i) = ζ(2) − ln(i) ln(1 − i). (1.121)
Substitute the values from (1.26), (1.25), and (1.108) in (1.121),

3 π
Li2 (1 − i) = ζ(2) − ln(2) + G i. (1.122)
8 4
Replacing i by −i in (1.122) yields
3 π
Li2 (1 + i) = ζ(2) + ln(2) + G i. (1.123)
8 4
1.16.3 Landen’s Dilogarithm Identity
For z ∈ C, z 6= 0, the following identity holds:

z−1 1
Li2 (1 − z) + Li2 = − ln2 (z). (1.124)
z 2
z−1

Proof. Differentiate Li2 zthen integrate,
Z
z−1 z−1
Li2 = d Li2
z z

∂ Lia−1 (z)
employ Lia (z) = given in (1.118)
∂z z
Z Z Z
ln(z) ln(z) ln(z)
= dz = − dz − dz
z(1 − z) z 1−z
1
= − ln2 (z) − Li2 (1 − z) + C.
2
The proof finishes on extracting C = 0 by putting z = 1.
Let’s set z = 1 − i in (1.124),

1−i 1
Li2 (i) + Li2 = − ln2 (1 − i).
2 2
Substitute the values from (1.108), (1.26), and (1.25),

1−i 5 1 π
Li2 = ζ(2) − ln2 (2) − G − ln(2) i. (1.125)
2 16 8 8
Replace i by −i,

1+i 5 1 π
Li2 = ζ(2) − ln2 (2) + G − ln(2) i. (1.126)
2 16 8 8
This result can again be found by setting z = −1 + i in (1.127).
1.16.4 Dilogarithm Inversion Formula
ln2 (z)

1
Li2 (−z) + Li2 − =− + 2 Li2 (−1). (1.127)
z 2
Proof.
ln 1+z
Z Z
1 1 z
Li2 − = d Li2 − = dz
z z z
Z Z
ln(1 + z) ln(z) 1
= dz − dz = − Li2 (−z) − ln2 (z) + C.
z z 2
To find the constant C, set z = 1,
1 2
Li2 (−1) = − Li2 (−1) − ln (1) + C
2
or
C = 2 Li2 (−1),
and we are done with the proof.
1.16.5 Relation Involving Four Dilogarithm Functions
For z ∈ C, z 6= −1, the following identity holds:

2z z 1+z
Li2 − Li2 + Li2
1+z 1+z 2

1
= Li2 (z) + Li2 + ln(2) ln(1 + z). (1.128)
2
Proof. Let

2z z 1+z
f (z) = Li2 − Li2 + Li2 ,
1+z 1+z 2
we have
Z
ln(1 − z)
Z
ln(2)
f (z) = d f (z) = − dz
1+z z
= ln(2) ln(1 + z) + Li2 (z) + C.
Setting z = 0 gives C = Li2 ( 12 ) and the proof is finalized.
1.16.6 Another Relation Involving Dilogarithm Functions
For z ∈ C, z 6= −1, the following identity holds:

2z z 1−z 1
Li2 − Li2 − Li2 − Li2 (z) + Li2
1+z 1+z 2 2

1+z
= ln(1 − z) ln . (1.129)
2
Proof. Let

2z z 1−z 1
f (z) = Li2 − Li2 − Li2 − Li2 (z) + Li2 ,
1+z 1+z 2 2
we have
Z
ln(1 − z) ln(1 + z)
Z
ln(2)
f (z) = d f (z) = − + dz
1+z 1−z 1−z
Z
ln(1 − z) ln(1 + z)
Z
ln(2)
= − dz + dz
1+z 1−z 1−z
Z
= d(ln(1 − z) ln(1 + z)) − ln(2) ln(1 − z)
= ln(1 − z) ln(1 + z) − ln(2) ln(1 − z)

1+z
= ln(1 − z) ln + C.
2
The proof completes on extracting C = 0 by setting z = 0.
1.16.7 Landen’s Trilogarithm Identity

z−1 1
Li3 (z) + Li3 (1 − z) + Li3 = ζ(3) + ln3 (z) + ζ(2) ln(z)
z 6
1
− ln2 (z) ln(1 − z). (1.130)
2
Proof.
Li2 z−1
Z
z−1 z−1
Z
z
Li3 = d Li3 = − dz
z z z(1 − z)
{make use of Landen’s dilogarithm identity in (1.124)}

ln2 (z)
Z
Li2 (1 − z)
= + dz
z(1 − z) 2z(1 − z)
Li2 (1 − z) Li2 (1 − z) 1 ln2 (z) 1 ln2 (z)
Z
= + + + dz
z 1−z 2 z 2 1−z
Li2 (1 − z) ln2 (z)
Z Z
1 1
= dz − Li3 (1 − z) + ln3 (z) + dz
x 6 2 1−z
| {z }
IBP
ln2 (z)
Z
1 3 1
= ln(z) Li2 (1 − z) − Li3 (1 − z) + ln (z) − dz.
6 2 1−z
For the remaining integral, force integration by parts twice,
ln2 (z) ln(1 − z) ln(z)

Z Z
dz = − ln(1 − z) ln2 (z) + 2 dz
1−z z
Z
Li2 (z)
= − ln(1 − z) ln2 (z) − 2 Li2 (z) ln(z) + 2 dz
z
= − ln(1 − z) ln2 (z) − 2 Li2 (z) ln(z) + 2 Li3 (z). (1.131)
Plug this integral back in,

z−1 1 1
Li3 = ln(z)[Li2 (1 − z) + Li2 (z)] + ln(1 − z) ln2 (z) + ln3 (z)
z 2 6
− Li3 (1 − z) − Li3 (z)
{make use of (1.119) for the first term}
1 1
= ζ(2) ln(z) − ln(1 − z) ln2 (z) + ln3 (z) − Li3 (1 − z) − Li3 (z) + C,
2 6
and the proof follows on extracting C = ζ(3) by setting z = 1.
Setting z = 1/2 in (1.130) yields

1 1 1 1
Li3 + Li3 + Li3 (−1) = ζ(3) − ln3 (2) − ζ(2) ln(2) + ln3 (2).
2 2 6 2
Substitute Li3 (−1) = − 43 ζ(3) given in (1.103),

1 7 1 1
Li3 = ζ(3) − ln(2)ζ(2) + ln3 (2). (1.132)
2 8 2 6
For another result, set z = i in (1.130) then consider the real parts of both sides,

i−1
R Li3 (i) + Li3 (1 − i) + Li3
i

1 3 1 2
= ζ(3) + R ln (i) + ζ(2) ln(i) − ln (i) ln(1 − i) .
6 2
i−1

Since Li3 i = Li3 (1 + i) and R Li3 (1 − i) = R Li3 (1 + i), we have
R {Li3 (i) + 2 Li3 (1 + i)}

1 3 1
= ζ(3) + R ln (i) + ζ(2) ln(i) − ln2 (i) ln(1 − i) .
6 2
Collect the results from (1.109), (1.26), and (1.25),

3 35
R Li3 (1 + i) = ln(2)ζ(2) + ζ(3) = R Li3 (1 − i). (1.133)
16 64
1.16.8 Polylogarithm Inversion Formula
For z ∈ C, z 6= 0, the following two identities hold:

a
ln2a−2n (z) ln2a (z)

1 X
Li2a (−z) + Li2a − =2 Li2n (−1) − ; (1.134)
z n=1
(2a − 2n)! (2a)!
a
ln2a−2n+1 (z) ln2a+1 (z)

1 X
Li2a+1 (−z) − Li2a+1 − =2 Li2n (−1) − .
z n=1
(2a − 2n + 1)! (2a + 1)!
(1.135)
Proof. Divide both sides of (1.127) by z then integrate from z = 1 to z,
ln3 (z)

1
Li3 (−z) − Li3 − =− + 2 Li2 (−1) ln(z). (1.136)
z 2·3
Repeat the same process,
ln4 (z) ln2 (z)

1
Li4 (−z) + Li4 − =− + 2 Li2 (−1) + 2 Li4 (−1), (1.137)
z 2·3·4 2
again,
ln5 (z) ln3 (z)

1
Li5 (−z) − Li5 − =− + 2 Li2 (−1) + 2 Li4 (−1) ln(z).
z 2·3·4·5 2·3
(1.138)
This can be generalized to (1.134) and (1.135) and the proof is finished.
Let’s put z = −1 + i in (1.136) then consider the real parts of both sides,

1+i 1
R Li3 (1 − i) − Li3 = R −ζ(2) ln(−1 + i) − ln3 (−1 + i) .
2 6
Substitute the values from (1.23) and (1.133),
ln3 (2) ln(2)

1+i 35 1−i
R Li3 = − ζ(2) + ζ(3) = R Li3 . (1.139)
2 48 32 64 2
For more values, let z = −2 in (1.127), (1.136), (1.137), and (1.138) using ln(−2) =
ln(2) + iπ, which follows from (1.21), we obtain:
3
Li2 (2) =ζ(2) − π ln(2)i ; (1.140)
2
7 3 π
Li3 (2) = ζ(3) + ln(2)ζ(2) − ln2 (2)i ; (1.141)
8 2 2
1 1 4 π
Li4 (2) = − Li4 + 2ζ(4) + ln2 (2)ζ(2) − ln (2) − ln3 (2)i ; (1.142)
2 24 6

1 1 3 1 π 4
Li5 (2) = Li5 + 2 ln(2)ζ(4) + ln (2)ζ(2) − ln5 (2) − ln (2)i.
2 3 120 24
(1.143)
The values of Li2 ( 12 ) and Li3 ( 21 ), given in (1.120) and (1.132), were used in (1.140)
and (1.141).
For more results of polylogarithm functions, check [16].
1.17 Harmonic Number

1.17.1 Definition
The n-th harmonic number of order a is defined by

n
X 1 1 1 1
Hn(a) = = 1 + a + a + ··· + a, n, a ∈ Z+ . (1.144)
ka 2 3 n
k=1
One of the properties of the harmonic number is
(a) 1
Hn(a) − Hn−1 = . (1.145)
na
To show that, subtract the following two harmonic numbers:
1 1 1 1
Hn(a) = 1 + + a + ··· + + a,
2a 3 (n − 1)a n
(a) 1 1 1
Hn−1 = 1 + a + a + ··· + .
2 3 (n − 1)a
1.17. Harmonic Number 41
With a = 1 in (1.144) and (1.145), we have

n
X 1 1 1 1
Hn(1) = Hn = = 1 + + + ··· + , (1.146)
k 2 3 n
k=1
1
Hn − Hn−1 = . (1.147)
n
Another property is
n
(a)
X 1
H2n − 2−a Hn(a) = . (1.148)
(2k − 1)a
k=1
To show that, begin with the RHS:

n
X 1 1 1 1
= 1 + a + a + ··· +
(2k − 1)a 3 5 (2n − 1)a
k=1

1 1 1 1
add and subtract a + a + a + · · · +
2 4 6 (2n)a

1 1 1 1 1 1 1
= 1 + a + a + ··· + − + + + · · · +
2 3 (2n)a 2a 4a 6a (2n)a

1 1 1 1 1 1 1
= 1 + a + a + ··· + − 1 + + · · · +
2 3 (2n)a 2a 2a 3a na
2n n
X 1 1 X 1 (a)
= − = H2n − 2−a Hn(a) .
ka 2a ka
k=1 k=1
Further, if we let
n
(a)
X 1
fn = H2n − 2−a Hn(a) =
(2k − 1)a
k=1
and so
n+1
(a) (a)
X 1
fn+1 = H2n+2 − 2−a Hn+1 = ,
(2k − 1)a
k=1
we have
n+1 n
X 1 X 1
fn+1 − fn = a
−
(2k − 1) (2k − 1)a
k=1 k=1
( n+1 n
)
X X
use f (k) = f (k) + f (n + 1) for the first sum
k=1 k=1
n n
X 1 1 X 1 1
= a
+ a
− = .
(2k − 1) (2n + 1) (2k − 1)a (2n + 1)a
k=1 k=1
(a) (a)
Thus, with fn = H2n − 2−a Hn , we have
1
fn+1 − fn = . (1.149)
(2n + 1)a
(a)
To write Hn in integral form, take the summation for both sides of (1.31):
1
(−1)a−1
Z
1
= xk−1 lna−1 (x)dx
ka (a − 1)! 0
from k = 1 to n,
n ∞ Z
X 1 (a) (−1)a−1 X 1 lna−1 (x)
:= Hn = dx
ka (a − 1)! n=1 0 1 − x
k=1
n
!
(−1)a−1 1 a−1
Z X
k−1
= ln (x) x dx
(a − 1)! 0
k=1
{use the geometric series}
(−1)a−1 1 a−1 1 − xn
Z
= ln (x) dx.
(a − 1)! 0 1−x
Therefore,
(−1)a−1 1
lna−1 (x)(1 − xn )
Z
Hn(a) = dx. (1.150)
(a − 1)! 0 1−x
This integral representation extends the range of n to the whole complex plane where
R(n) > −1. Note that setting a = 1 in (1.150) gives
1
1 − xn
Z
Hn = dx, R(n) > −1. (1.151)
0 1−x
Let’s break up the integrand in (1.150),
(−1)a−1 1 lna−1 (x) (−1)a−1 1 xn lna−1 (x)

Z Z
Hn(a) = dx − dx
(a − 1)! 0 1 − x (a − 1)! 0 1−x
{substitute the result from (3.1) for the first integral}
(−1)a−1 1 xn lna−1 (x)
Z
= ζ(a) − dx. (1.152)
(a − 1)! 0 1−x
1.17. Harmonic Number 43
Reorder the terms,

Z 1 n a−1
x ln (x)
dx = (−1)a−1 (a − 1)! ζ(a) − Hn(a) . (1.153)
0 1−x
1.17.2 Infinite Series Representation
(a)
Another form of Hn is
∞
X 1 1
Hn(a) = − , / Z− .
n∈ (1.154)
ka (k + n)a
k=1
Proof. Start with recalling (1.150)
(−1)a−1 lna−1 (x)(1 − xn )

1
Z
Hn(a) = dx
(a − 1)! 0 1−x

1
expand in series
1−x
∞
Z 1 !
(−1)a−1 a−1 n
X
k−1
= ln (x)(1 − x ) x dx
(a − 1)! 0 k=1
∞ Z
(−1)a−1 X 1 a−1
= ln (x)(xk−1 − xk+n−1 )dx
(a − 1)! 0
k=1
{make use of the result in (1.31)}
∞
X 1 1
= − .
ka (k + n)a
k=1
Setting a = 1 in (1.154) gives

∞ ∞
X 1 1 X n
Hn = − = . (1.155)
k k+n k(k + n)
k=1 k=1
To find the derivative of the harmonic number, rewrite (1.154) as

∞
X 1
= ζ(a) − Hn(a) . (1.156)
(k + n)a
k=1
Next, differentiate both sides of (1.154) with respect to n,

∞
∂ (a) X a
Hn = 0 + .
∂n (k + n)a+1
k=1
Employ the result in (1.156) for the latter sum,
∂ (a)
Hn = a ζ(a + 1) − Hn(a+1) . (1.157)
∂n
Moreover, bring back (1.155)
∞ r
X 1 1 X 1 1
Hn = − = lim −
k k + n r→∞ k k+n
k=1 k=1
and so
r
X 1 1
Hm = lim − .
r→∞ k k+m
k=1
Subtracting the two harmonic numbers gives

r ∞
X 1 1 X 1 1
Hn − Hm = lim − = − . (1.158)
r→∞ k + m k + n n=1 k + m k + n
k=1
ThePreason we inserted the limit in the calculations above is to avoid the divergence
∞
of k=1 k1 . Using the definition in (1.154), we also see that
∞
X 1 1
Hn(a) − Hm
(a)
= − . (1.159)
(k + m)a (k + n)a
k=1
1.18 Skew Harmonic Number

1.18.1 Definition
The n-th skew harmonic number is defined by

n
X (−1)k−1 1 1 (−1)n−1
Hn = =1− + − ··· + , n ∈ Z+ . (1.160)
k 2 3 n
k=1
One of the key properties of the skew harmonic number is
(−1)n−1
H n − H n−1 = . (1.161)
n
1.18. Skew Harmonic Number 45
To prove that, write (1.160) as
1 1 (−1)n (−1)n−1
Hn = 1 − + − ··· + + .
2 3 n−1 n
Replace n by n − 1,
1 1 (−1)n
H n−1 = 1 − + − ··· + .
2 3 n−1
Subtracting them yields (1.161).
Another property is
H 2n = H2n − Hn . (1.162)
To show that, we begin with the definition of H 2n :
2n
X (−1)k−1 1 1 1 1
H 2n = =1− + − ··· + −
k 2 3 2n − 1 2n
k=1
{split the terms into odd and even}

1 1 1 1 1 1 1
= 1 + + + ··· + − + + + ··· +
3 5 2n − 1 2 4 6 2n

1 1 1 1
add and subtract + + + · · · +
2 4 6 2n

1 1 1 1 1 1 1 1
= 1 + + + ··· + + −2 + + + ··· +
2 3 2n − 1 2n 2 4 6 2n

1 1 1 1 1 1 1
= 1 + + + ··· + + − 1 + + + ··· +
2 3 2n − 1 2n 2 3 n
2n n
X 1 X1
= − = H2n − Hn .
k k
k=1 k=1
A similar property is
H 2n+1 = H2n+1 − Hn , (1.163)
which can be proved by using the definition of H 2n+1 :

2n+1
X (−1)k−1
H 2n+1 =
k
k=1
( n+1 n
)
X X
use f (k) = f (k) + f (n + 1)
k=1 k=1
2n
X (−1)k−1 1 1
= + = H 2n +
k 2n + 1 2n + 1
k=1
{substitute (1.162)}
1
= H2n − Hn +
2n + 1

1
use H2n + = H2n+1
2n + 1
= H2n+1 − Hn .
The skew harmonic number can be written in integral form:

n n 1
(−1)k−1
X X Z
Hn = = (−1)k−1 xk−1 dx
k 0
k=1 k=1
Z 1 X n
!
k−1
= (−x) dx
0 k=1
1 Z 1 Z 1
1 − (−x)n (−x)n
Z
dx
= dx = − dx
0 1+x 0 1+x 0 1+x
Z 1
(−x)n
= ln(2) − dx.
0 1+x
Therefore,
1
(−x)n
Z
H n = ln(2) − dx. (1.164)
0 1+x
Like the integral in (1.151), this integral also extends the range of n to the whole
complex plane where R(n) > −1.
1.18.2 Infinite Series Representation
Another form of the skew harmonic number is

∞
X (−1)k+n
H n = ln(2) + , / Z− .
n∈ (1.165)
k+n
k=1
Proof. Begin with (1.164)

1
(−x)n
Z
H n = ln(2) − dx
0 1+x
1.19. Digamma Function 47

1
expand in series
1+x
Z 1 X∞
= ln(2) − (−x)n (−x)k−1 dx
0 k=1
X∞ Z 1
k+n−1
= ln(2) − (−1) xk+n−1 dx
k=1 0
∞
X (−1)k+n
= ln(2) + ,
k+n
k=1
and the prof is complete. Let’s multiply both sides of (1.165) by (−1)n ,
∞
X (−1)k
= (−1)n H n − ln(2) .

(1.166)
k+n
k=1
1.19 Digamma Function

1.19.1 Definition
The Digamma function is defined as
d Γ0 (n)
ψ(n) = ln(Γ(n)) = . (1.167)
dn Γ(n)
1.19.2 Digamma Reflection Formula
One of the key properties of the digamma function is
ψ(n) − ψ(1 − n) = −π cot(πn), n∈

/ Z. (1.168)
Proof. Take the logarithm of both sides of the Euler’s reflection formula in (1.38):
Γ(n)Γ(1 − n) = π csc(πn), n∈
/ Z,
we obtain
ln(Γ(n)) + ln(Γ(1 − n)) = ln(π) + ln(csc(πn)).
d
Next, differentiate both sides using dn ln(Γ(n)) = ψ(n) given in (1.167),
π csc(πn) cot(πn)
ψ(n) − ψ(1 − n) = − = −π cot(πn).
csc(πn)
1.19.3 Digamma–Harmonic Number Identity
The relation between the digamma function and the harmonic number is
ψ(n + 1) = Hn − γ, / Z− .
n∈ (1.169)

∞ n
1 Y 1 + k1
Γ(n) = , n∈
/ Z≤0
n
k=1
1 + nk
by n then use nΓ(n) = Γ(n + 1) given in (1.32),

∞ n
Y 1 + k1
Γ(n + 1) = n ,
/ Z− .
n∈
k=1
1 + k
Take the logarithm of both sides,

∞ n
Y 1 + k1
ln[Γ(n + 1)] = ln n

k=1
1 + k
∞ ∞
( )
Y X
use ln ak = ln(ak ) given in (1.12)
k=1 k=1
∞ n !
X 1 + k1
= ln
1 + nk

k=1
or
∞
X k+1 k+n
ln[Γ(n + 1)] = n ln − ln .
k k
k=1
d
Differentiate both sides with respect to n using dn ln(Γ(n + 1)) = ψ(n + 1),
∞
X k+1 1
ψ(n + 1) = ln − .
k k+n
k=1
1
Add and subtract k in the RHS then insert the limit, we arrive at
m
X k+1 1 1 1
ψ(n + 1) = lim ln − + −
m→∞ k k+n k k
k=1
{rearrange the terms}
"m X m
# "m #
X k+1 1 X1 1
= lim ln − + lim −
m→∞ k k m→∞ k k+n
k=1 k=1 k=1
1.19. Digamma Function 49
∞
" m m
#
X k+1 X 1 X 1 1
= lim ln − + −
m→∞ k k k k+n
k=1 k=1 k=1
∞
( m
)
X 1 X 1 1
use = Hm and − = Hn given in (1.146) and (1.155)
k k k+n
k=1 k=1
"m #
X k + 1
= lim ln − Hm + Hn .
m→∞ k
k=1
The remaining sum can be evaluated using the telescoping series:

m m
X k+1 X
ln = ln(k + 1) − ln(k)
k
k=1 k=1
= (ln(2) − ln(1)) + (ln(3) − ln(2)) + · · · + (ln(m) − ln(m − 1))
+(ln(m + 1) − ln(m)).
Notice that all terms cancel out but ln(m + 1):

m
X k+1
ln = ln(m + 1).
k
k=1
Substitute this sum back, we have
ψ(n + 1) = lim [ln(m + 1) − Hm ] + Hn

m→∞
{let m + 1 = p }
= lim [ln(p) − Hp−1 ] + Hn
p→∞

1
write Hp−1 = Hp − given in (1.147)
p
1
= − lim [Hp − ln(p)] + lim + Hn
p→∞ p→∞ p
{the first limit is the Euler–Mascheroni constant defined in (1.209)}

= −γ + 0 + Hn ,
Let’s set n = 0 in (1.169),

ψ(1) = −γ. (1.170)
Now set n = −1/2 in (1.169),
Z 1 1
1 − x− 2

1
ψ + γ = H− 12 = dx
2 0 1−x
√ Z 1
x=y dy
= −2 = −2 ln(2). (1.171)
0 1+y
Combining (1.170) and (1.171) yields

1
ψ − ψ(1) = −2 ln(2). (1.172)
2
Moreover, we have
1
ψ(n + 1) − ψ(n) = . (1.173)
n
To show that, make use of the identity in (1.169),
ψ(n + 1) − ψ(n) = Hn − γ − (Hn−1 − γ)

= Hn − Hn−1
{recall the relation in (1.147)}
1
= .
n
1.20 Polygamma Function

1.20.1 Definition
The Polygamma function is defined as
da da+1
ψ (a) (n) = ψ(n) = ln(Γ(n)). (1.174)
dna dna+1
Observe that
ψ (0) (n) = ψ(n).
1.20.2 Series Representation
Another series form of the polygamma function is

∞
X (−1)a+1 a!
ψ (a) (n + 1) = , / Z− .
n∈ (1.175)
(k + n)a+1
k=1
Proof. Differentiate both sides of (1.169):

∞
X 1 1
ψ(n + 1) = Hn − γ = − −γ
k k+n
k=1
1.20. Polygamma Function 51
a times with respect to n,

∞
!
da da X 1 1
a
ψ(n + 1) = ψ (a) (n + 1) = − −0
dn dna k k+n
k=1
∞
X (−1)a+1 a!
= .
(k + n)a+1
k=1
Replacing n by n − 1 in (1.175) gives

∞
X (−1)a+1 a!
ψ (a) (n) = . (1.176)
(k + n − 1)a+1
k=1
Shift the index by +1,

∞
X (−1)a+1 a!
ψ (a) (n) = , n∈
/ Z≤0 . (1.177)
(k + n)a+1
k=0
Another form of the polygamma function is

1
xn lna (x)
Z
ψ (a)
(n + 1) = − dx, / Z− .
n∈ (1.178)
0 1−x
Proof. Differentiating both sides of (1.169):

1
1 − xn
Z
ψ(n + 1) = Hn − γ = dx − γ
0 1−x
a times with respect to n,

Z 1
da 1 − xn
ψ (a) (n + 1) = a
dx − 0
dn 0 1 − x
{use differentiation under the integral sign theorem given in (2.78)}
Z 1 a Z 1 n a
∂ 1 − xn x ln (x)
= a 1−x
dx = − dx.
0 ∂n 0 1−x
Moreover, if we compare (1.178) and (1.153), we deduce that

ψ (a) (n + 1) = (−1)a a! Hn(a+1) − ζ(a + 1) . (1.179)
1.20.4 Evaluation of ψ (a) (1)
ψ (a) (1) = (−1)a−1 a!ζ(a + 1). (1.180)
Proof. Put n = 0 in (1.175),

∞
X (−1)a−1 a!
ψ (a) (1) = = (−1)a−1 a!ζ(a + 1).
k a+1
k=1
Examples
ψ (1) (1) = ζ(2); (1.181)

(2)
ψ (1) = −2ζ(3); (1.182)
ψ (3) (1) = 6ζ(4); (1.183)
(4)
ψ (1) = −24ζ(5); (1.184)
(5)
ψ (1) = 120ζ(6). (1.185)
1

1.20.5 Evaluation of ψ (a) 2

1
ψ (a) = (−1)a−1 a!(2a+1 − 1)ζ(a + 1). (1.186)
2
Proof. Put n = 1/2 in (1.177),

∞
(−1)a+1 a!
X
1
ψ (a) =
2 (k + 1/2)a+1
k=0
∞
a−1 a+1
X 1
= (−1) a!2 .
(2k + 1)a+1
k=0
This sum is given in (1.85) and the proof is finished.

Examples

1
ψ (1) = 3ζ(2); (1.187)
2

1
ψ (2) = −14ζ(3); (1.188)
2
1.20. Polygamma Function 53

(3) 1
ψ = 90ζ(4); (1.189)
2

1
ψ (4) = −744ζ(5); (1.190)
2

1
ψ (5) = 7560ζ(6). (1.191)
2
1

d2a

1
ψ (2a) = (22a − 21+4a )(2a)!ζ(2a + 1) − 22a−1 π lim1 2a [π csc(πs)].
4 s→ 2 ds
(1.192)
Proof. Replace 2a by 2a + 1 in (1.82),
2−1−4a (2a) 1

β(2a + 1) = − ψ − (1 − 2−1−2a )ζ(2a + 1). (1.193)
(2a)! 4
Combine (1.92) and (1.193) to finish the proof.

Examples

1
ψ (2) = −56ζ(3) − 2π 3 ; (1.194)
4

1
ψ (4) = −11904ζ(5) − 40π 5 ; (1.195)
4

1
ψ (6) = −5852160ζ(7) − 1952π 7 ; (1.196)
4

1
ψ (8) = −5274501120ζ(9) − 177280π 9 ; (1.197)
4

1
ψ (10) = −7606429286400ζ(11) − 25866752π 11 . (1.198)
4
3

d2a

(2a) 3 (2a) 1
ψ =ψ + lim1 2a
[π cot(πn)]. (1.199)
4 4 n→ 4 dn
ψ(n) − ψ(1 − n) = −π cot(πn).
2a times with respect to n,
d2a
ψ (2a) (n) − ψ (2a) (1 − n) = − [π cot(πn)].
dn2a
The proof follows on taking the limit on both sides and letting n → 1/4.
Examples

3 1
ψ (2) = ψ (2) + 4π 3 ; (1.200)
4 4

3 1
ψ (4) = ψ (4) + 80π 5 ; (1.201)
4 4

3 1
ψ (6) = ψ (6) + 3904π 7 ; (1.202)
4 4

3 1
ψ (8) = ψ (8) + 354560π 9 ; (1.203)
4 4

3 1
ψ (10) = ψ (10) + 354560π 11 . (1.204)
4 4
1.21 Catalan’s Constant

The Catalan’s constant, denoted by G, is defined as
∞
X (−1)n 1 1
G= 2
= 1 − 2 + 2 − ··· . (1.205)
n=0
(2n + 1) 3 5
The Catalan’s constant is a special case of the Dirichlet beta function in (1.80):
∞
X (−1)n
β(z) = ,
n=0
(2n + 1)z
G = β(2). (1.206)
In (1.87), we also see that Z 1
ln(x)
G=− dx (1.207)
0 1 + x2
and
ψ (1) 41

3
G= − ζ(2). (1.208)
48 4
1.22. Euler–Mascheroni Constant 55
For more integral and series representations of the Catalan’s constant, check [36].
1.22 Euler–Mascheroni Constant

The Euler–Mascheroni Constant is defined as
γ = lim (Hn − ln(n)) . (1.209)

n→∞
The Euler–Mascheroni constant, denoted by γ, is defined as the area bounded by the

two functions, y = 1/x and y = 1/bxc, where bxc is the floor function, over the
interval x ∈ [1, ∞). To get the form in (1.209), we calculate the bounded area over
the interval x ∈ [1, n] then we let integer n → ∞:
Z n Z n
dx dx
γ = lim −
n→∞ 1 bxc 1 x
Z n
dx 1 1 1
note that = 1 + + + ··· + = Hn−1
1 bxc 2 3 n−1
Z n
dx
= lim Hn−1 −
n→∞ 1 x
Z n
1 dx
we have Hn−1 = Hn − and = ln(n)
n 1 x

1
= lim Hn − − ln(n)
n→∞ n
1
= lim (Hn − ln(n)) − lim
n→∞ n→∞ n
= lim (Hn − ln(n)) − 0.
n→∞
For more representations of Euler–Mascheroni constant, see [37].

Chapter 2
Generating Functions and

Powerful Identities
Before we start deriving the generating functions, we need to prove the following
series identity:
∞ ∞
X 1 X
an xn = (an − an−1 )xn , a0 = 0. (2.1)
n=1
1 − x n=1
Proof. We begin with the fact that

1 x
1= − .
1−x 1−x
Multiply through by an xn then take the summation over n ≥ 1,
∞ ∞ ∞
X 1 X 1 X
an xn = an xn − an xn+1
n=1
1 − x n=1 1 − x n=1
{let the index n of the second sum start from 0, since a0 = 0}
∞ ∞
1 X 1 X
= an xn − an xn+1
1 − x n=1 1 − x n=0
{shift the index n of the second sum by −1}
∞ ∞
1 X 1 X
= an xn − an−1 xn
1 − x n=1 1 − x n=1
∞
1 X
= (an − an−1 )xn ,
1 − x n=1
56
2.1. Generating Functions 57
2.1 Generating Functions

P∞
2.1.1 n=1 Hn(a) xn
For |x| < 1, the following identity holds:

∞
X Lia (x)
Hn(a) xn = . (2.2)
n=1
1−x
(a) (a)
Proof. Since H0 = 0, it’s valid to set an = Hn in (2.1),
∞ ∞
X 1 X (a) (a)

Hn(a) xn = Hn − Hn−1 xn
n=1
1 − x n=1

(a) (a) 1
write Hn − Hn−1 = a given in (1.145)
n
∞ n
1 X x Lia (x)
= = ,
1 − x n=1 na 1−x
and the proof is complete.

(a)
Let the index n in (2.2) start from 0, since H0 = 0,
∞ ∞
Lia (x) X (a) n X (a) n−1
= Hn x = Hn−1 x . (2.3)
1−x n=0 n=1
The last step follows from shifting the index n by −1.

For a different proof, see [28, pp. 348–349].
If we set a = 1 in (2.2) and (2.3) using
∞
X xn
Li1 (x) = = − ln(1 − x),
n=1
n1
we have
∞
X ln(1 − x)
Hn xn = − , |x| < 1, (2.4)
n=1
1−x
∞
X ln(1 − x)
Hn−1 xn−1 = − , |x| < 1. (2.5)
n=1
1−x
58 Chapter 2. Generating Functions and Powerful Identities
Let’s integrate both sides of (2.5) from x = 0 to x,

Z ∞
xX
1 2
ln (1 − x) = Hn−1 xn−1 dx
2 0 n=1
∞ Z x ∞
X
n−1
X xn
= Hn−1 x dx = Hn−1 .
n=1 0 n=1
n
Therefore,
∞
X Hn−1 n
ln2 (1 − x) = 2 x , |x| ≤ 1, x 6= 1. (2.6)
n=1
n
The latter identity can be proved using the Cauchy product, as follows:
ln2 (1 − x) = (− ln(1 − x))(− ln(1 − x))

{expand both logs in series}
∞
! ∞ !
X xn X xn
=
n=1
n n=1
n

1 1
apply (2.80) where an = and bn =
n n
∞ n
!
X X1 1
= xn+1 ·
n=1
k n − k+1
k=1

1 1 1 1 1
write · = +
k n−k+1 n+1 k n−k+1
∞ n n
!
X
n+1 1 X 1 X 1
= x · +
n=1
n+1 k n−k+1
k=1 k=1
( n n
)
X 1 X1
use = given in (1.3)
n−k+1 k
k=1 k=1
∞ n
!
X
n+1 1 X 1
= x · 2
n=1
n+1 k
k=1
( n
)
X 1
use = Hn defined in (1.146)
k
k=1
∞
X Hn n+1
=2 x
n=1
n+1
{let the index n start from 0, since H0 = 0}
∞
X Hn n+1
=2 x
n=0
n +1
{shift the index n by −1}
∞
X Hn−1 n
=2 x .
n=1
n
P∞ Hn n
2.1.2 n=1 n x
For |x| ≤ 1, x 6= 1, the following identity holds:

∞
X Hn n 1
x = Li2 (x) + ln2 (1 − x). (2.7)
n=1
n 2
1
Proof. In (2.6), substitute Hn−1 = Hn − n given in (1.147),
∞ 1 ∞ ∞
X Hn − X Hn n X xn
ln2 (1 − x) = 2 n
xn = 2 x −2
n=1
n n=1
n n=1
n2
∞
X Hn n
=2 x − 2 Li2 (x).
n=1
n
The proof follows on adding 2 Li2 (x) to both sides then dividing by 2.
P∞ Hn n
2.1.3 n=1 n2 x
For |x| ≤ 1, the following identity holds:

∞
X Hn n
x = Li3 (x) − Li3 (1 − x) + ln(1 − x) Li2 (1 − x)
n=1
n2
1
+ ln(x) ln2 (1 − x) + ζ(3). (2.8)
2
Proof. Divide both sides of (2.7) by x then integrate,

∞
ln2 (1 − x)
Z Z Z
1 X Hn n Li2 (x) 1
x dx = dx + dx
x n=1 n x 2 x
ln2 (1 − x)
Z
1
= Li3 (x) + dx.
2 x
Since
∞ ∞ ∞
Hn x n
Z Z
1 X Hn n X Hn X
x dx = xn−1 dx = ,
x n=1 n n=1
n n=1
n n
we have
∞
ln2 (1 − x)
Z
X Hn n 1
x = Li3 (x) + dx.
n=1
n2 2 x
For the remaining integral, forcing integration by parts twice yields
ln2 (1 − x) ln(x) ln(1 − x)

Z Z
dx = ln(x) ln2 (1 − x) + 2 dx
x 1−x

Li2 (1 − x)
Z
2
= ln(x) ln (1 − x) + 2 ln(1 − x) Li2 (1 − x) + dx
1−x
= ln(x) ln2 (1 − x) + 2 ln(1 − x) Li2 (1 − x) − 2 Li3 (1 − x). (2.9)
Substitute this integral back and add the constant of integration,

∞
X Hn n 1
2
x = Li3 (x) − Li3 (1 − x) + ln(1 − x) Li2 (1 − x) + ln(x) ln2 (1 − x) + C.
n=1
n 2
Set x = 0,
0 = −ζ(3) + C =⇒ C = ζ(3).
Plugging in the value of the constant completes the proof.
P∞ (2)
Hn
2.1.4 n=1 n xn

∞ (2)
X Hn n
x = Li3 (x) + 2 Li3 (1 − x) − ln(1 − x) Li2 (1 − x)
n=1
n
−ζ(2) ln(1 − x) − 2ζ(3). (2.10)
Proof. Set a = 2 in (2.2),

∞
X Li2 (x)
Hn(2) xn = .
n=1
1−x
Divide both sides by x then integrate using

∞ ∞ ∞
xn
Z Z
1 X (2) n X
(2)
X
Hn x dx = Hn xn−1 dx = Hn(2) ,
x n=1 n=1 n=1
n
we have
∞ (2) Z Z Z
X Hn n Li2 (x) Li2 (x) Li2 (x)
x = dx = dx + dx
n=1
n x(1 − x) x 1−x
| {z }
IBP
2
ln (1 − x)
Z
= Li3 (x) − ln(1 − x) Li2 (x) − dx
x
= Li3 (x) + 2 Li3 (1 − x) − ln(1 − x)[Li2 (x) + ln(x) ln(1 − x)]
−2 ln(1 − x) Li2 (1 − x) + C
{make use of the reflection formula (1.124) for the second term}
= Li3 (x) + 2 Li3 (1 − x) − ln(1 − x) Li2 (1 − x) − ζ(2) ln(1 − x) + C.
Put x = 0,
0 = 2ζ(3) + C =⇒ C = −2ζ(3),
For a different method, substitute the result from (2.8) in (2.82).
P∞ 2
2.1.5 n=1 (Hn − Hn(2) )xn

∞
X ln2 (1 − x)
(Hn2 − Hn(2) )xn = . (2.11)
n=1
1−x
(2)
Proof. Set an = Hn2 − Hn in (2.1),
∞ ∞
X 1 X 2 (2)

(Hn2 − Hn(2) )xn = (Hn − Hn(2) ) − (Hn−1
2
− Hn−1 ) xn
n=1
1 − x n=1
( 2 )
2 1 (2) (2) 1
write Hn−1 = Hn − and Hn−1 = Hn − 2
n n
∞ ∞
1 xn

2 X 2 X Hn−1 n
= Hn − = x
1 − x n=1 n n 1 − x n=1 n
{susbstitute the result from (2.6)}
ln2 (1 − x)

2 1 2
= ln (1 − x) = ,
1−x 2 1−x
and the proof is finished.

(2)
Further, let the index in (2.11) start from 0, since Hn2 − Hn = 0,
∞
ln2 (1 − x) X 2
= (Hn − Hn(2) )xn
1−x n=0
{shift the index by −1}
∞
(2)
X
2
= (Hn−1 − Hn−1 )xn−1
n=1
∞ 2 !
X 1 1
= Hn − − Hn + 2 xn−1 .
(2)
n=1
n n
Expand and simplify, we obtain

∞
ln2 (1 − x) X

2Hn 2
= Hn2 − Hn(2) − + 2 xn−1 . (2.12)
1−x n=1
n n
Check [28, p. 355] for an alternative proof.
P∞ 2 −H (2)
Hn
2.1.6 n=1 n
n
xn

∞
X xn
(Hn2 − Hn(2) ) = 2 ln(1 − x) Li2 (1 − x) − 2 Li3 (1 − x)
n=1
n
1 3
+ ln(x) ln2 (1 − x) − ln (1 − x) + 2ζ(3). (2.13)
3
xn
xn−1 dx =
R
Proof. Divide both sides of (2.11) by x then integrate using n ,
∞ n
ln2 (1 − x)
Z
X x
(Hn2 − Hn(2) ) = dx
n=1
n x(1 − x)
ln2 (1 − x) ln2 (1 − x)
Z Z
= dx + dx
x 1−x
{substitute the result of the first integral from (2.9)}
= ln(x) ln2 (1 − x) + 2 ln(1 − x) Li2 (1 − x) − 2 Li3 (1 − x)
1
− ln3 (1 − x) + C.
3
On setting x = 0, we get C = 2ζ(3) and the proof is finalized.
P∞ 2
Hn n
2.1.7 n=1 n x

∞
X Hn2 n 1
x = Li3 (x) − ln(1 − x) Li2 (x) − ln3 (1 − x). (2.14)
n=1
n 3
Proof. The proof follows on combining the results from (2.10) and (2.13).
Check [28, p. 349–350] for another approach.
P∞ Hn n
2.1.8 n=1 n3 x
∞
X Hn n x
x = Li4 − Li4 (1 − x) + 2 Li4 (x) − ln(1 − x) Li3 (1 − x)
n=1
n3 x−1
1 1 1
+ζ(3) ln(1−x)+ ζ(2) ln2 (1−x)− ln(x) ln3 (1−x)+ ln4 (1−x)+ζ(4).
2 6 24
(2.15)
Proof. Replace z by 1 − x in (1.130),

x
Li3 (1 − x) + Li3 (x) + Li3
x−1
1 3 1
= ζ(3) + ln (1 − x) + ζ(2) ln(1 − x) − ln2 (1 − x) ln(x).
6 2
Divide both sides by x(1 − x) then integrate,

Z Li3 x
Li3 (1 − x)
Z Z
Li3 (x) x−1
dx + dx + dx
x(1 − x) x(1 − x) x(1 − x)
| {z } | {z } | {z }
I1 I2 I3
3
ln (1 − x) ln(1 − x)
Z Z Z
dx 1
= ζ(3) + dx +ζ(2) dx
x(1 − x) 6 x(1 − x) x(1 − x)
| {z } | {z } | {z }
I4 I5 I6
2
ln (1 − x) ln(x)
Z
1
− dx.
2 x(1 − x)
| {z }
I7
Evaluation of I1 :
Li3 (1 − x) Li3 (1 − x) Li3 (1 − x)
Z Z Z
I1 = dx = dx + dx
x(1 − x) x 1−x
| {z }
IBP
ln(x) Li2 (1 − x)
Z
= ln(x) Li3 (1 − x) + dx − Li4 (1 − x)
1−x
1
= ln(x) Li3 (1 − x) + Li22 (1 − x) − Li4 (1 − x).
2
Evaluation of I2 :
Z Z Z
Li3 (x) Li3 (x) Li3 (x)
I2 = dx = dx + dx
x(1 − x) x 1−x
| {z }
IBP
ln(1 − x) Li2 (x)
Z
= Li4 (x) − ln(1 − x) Li3 (x) + dx
x
1
= Li4 (x) − ln(1 − x) Li3 (x) − Li22 (x).
2
x
Evaluation of I3 : Substitute x−1 = y,

Z Li3 x Z
x−1 Li3 (y) x
I3 = dx = dy = Li4 (y) = Li4 .
x(1 − x) y x−1
Evaluation of I4 :
Z Z Z
dx dx dx
I4 = = + = ln(x) − ln(1 − x).
x(1 − x) x 1−x
Evaluation of I5 :
ln3 (1 − x) ln3 (1 − x) ln3 (1 − x)

Z Z Z
I5 = dx = dx + dx
x(1 − x) x 1−x
ln3 (1 − x)
Z
1
= dx − ln4 (1 − x).
x 4
Evaluation of I6 :
ln(1 − x) ln(1 − x) ln(1 − x)
Z Z Z
I6 = dx = dx + dx
x(1 − x) x 1−x
1
= − Li2 (x) − ln2 (1 − x)
2
{replace z by 1 − x in Landen’s formula in (1.124)}

x
= Li2 .
x−1
Evaluation of I7 :
ln2 (1 − x) ln(x)
Z
I7 = dx
x(1 − x)
ln2 (1 − x) ln(x) ln2 (1 − x) ln(x)
Z Z
= dx + dx .
1−x x
| {z } | {z }
I70 I700
Integrate I70 by parts,
ln2 (1 − x) ln(x) ln3 (1 − x)

Z Z
1 1
I70 = dx = − ln3 (1 − x) ln(x) + dx.
1−x 3 3 x
For I700 , expand ln2 (1 − x) in series given in (2.6),

∞
ln2 (1 − x) ln(x)
Z Z
X Hn−1
I700 = dx = 2 xn−1 ln(x)dx
x n=1
n
∞ 1
− ln(x) n xn

IBP
X H n n
= 2 x − 2
n=1
n n n
∞ ∞
X Hn n X Hn n
= 2 ln(x) 2
x −2 x − 2 ln(x) Li3 (x) + 2 Li4 (x)
n=1
n n=1
n3
= 2 Li4 (x) − 2 ln(x) Li3 (1 − x) + 2 ln(x) ln(1 − x) Li2 (1 − x)
∞
2 2
X Hn n
+ ln (x) ln (1 − x) + 2ζ(3) ln(x) − 2 x .
n=1
n3
Combining the results of I70 and I700 ,

1 3
I7 = 2 Li4 (x) − ln (1 − x) ln(x) − 2 ln(x) Li3 (1 − x) + ln2 (x) ln2 (1 − x)
3
+2 ln(x) ln(1 − x) Li2 (1 − x) + 2ζ(3) ln(x)
∞
ln3 (1 − x)
Z
X Hn n 1
−2 x + dx.
n=1
n3 3 x
R ln3 (1−x)
By collecting all integrals, the integral x dx nicely cancels out,
∞
X Hn n x
n=1
n3 x−1
1 4 1
+ ln (1 − x) − ln(x) ln3 (1 − x) + ζ(3) ln(1 − x)
24 6
1 2 1
+ Li2 (1 − x) + ln(x) ln(1 − x) Li2 (1 − x) + ln2 (x) ln2 (1 − x)
2 2

1 2 x
− Li2 (x) − ζ(2) Li2 .
2 x−1
The last two lines can be further simplified:

1 2 1
Li2 (1 − x) + ln(x) ln(1 − x) Li2 (1 − x) + ln2 (x) ln2 (1 − x)
2 2
1 2 x
− Li2 (x) − ζ(2) Li2
2 x−1
2
1 1 2 x
= Li2 (1 − x) + ln(x) ln(1 − x) − Li2 (x) − ζ(2) Li2
2 2 x−1
{use the dilogarithm reflection formula (1.119) for the first term}
2
1 1 x
= ζ(2) − Li2 (x) − Li22 (x) − ζ(2) Li2
2 2 x−1

5 x
= ζ(4) − ζ(2) Li2 (x) − ζ(2) Li2
4 x−1

5 x
= ζ(4) − ζ(2) Li2 (x) + Li2
4 x−1
{make use of Landen’s dilogarithm identity (1.124)}

5 1
= ζ(4) − ζ(2) − ln2 (1 − x)
4 2
5 1
= ζ(4) + ζ(2) ln2 (1 − x).
4 2
Thus,
∞
X Hn n x
n=1
n3 x−1
1 4 1
+ ln (1 − x) − ln(x) ln3 (1 − x) + ζ(3) ln(1 − x)
24 6
5 1
+ ζ(4) + ζ(2) ln2 (1 − x) + C.
4 2
Set x = 0, we get C = − 14 ζ(4) and this finalizes the proof.
P∞ (2)
Hn
2.1.9 n=1 n2
xn

∞ (2)
X Hn n x 1
2
x = −2 Li4 + 2 Li4 (1 − x) − Li4 (x) + Li22 (x)
n=1
n x−1 2
1 4
+2 ln(1 − x) Li3 (1 − x) − ln (1 − x) − ζ(2) ln2 (1 − x)
12
1
−2ζ(3) ln(1 − x) + ln(x) ln3 (1 − x) − 2ζ(4). (2.16)
3
Proof. Substitute the result from (2.15) in (2.83).
P∞ (3)
Hn
2.1.10 n=1 n
xn

∞ (3)
X Hn n 1
x = Li4 (x) − ln(1 − x) Li3 (x) − Li22 (x). (2.17)
n=1
n 2
Proof. Set a = 3 in (2.2),

∞
X Li3 (x)
Hn(3) xn = .
n=1
1−x
xn
xn−1 dx =
R
Divide both sides by x then integrate using n ,
∞ (3) Z
X Hn n Li3 (x)
x = dx
n=1
n x(1 − x)
Z Z
Li3 (x) Li3 (x)
= dx + dx
x 1−x
| {z }
IBP
ln(1 − x) Li2 (x)
Z
= Li4 (x) − ln(1 − x) Li3 (x) + dx
x
1
= Li4 (x) − ln(1 − x) Li3 (x) − Li22 (x) + C.
2
The proof completes on extracting C = 0 by setting x = 0.
P∞
2.1.11 n=1 Hn3 xn

∞
X 1 3
Hn3 xn = Li3 (x) + 3 Li3 (1 − x) + ln(x) ln2 (1 − x)
n=1
1−x 2

−3ζ(2) ln(1 − x) − ln3 (1 − x) − 3ζ(3) . (2.18)
Proof. Set an = Hn3 in (2.1),

∞ ∞
X 1 X 3
Hn3 xn = (Hn − Hn−13
)xn
n=1
1 − x n=1
∞ 3 !
1 X 3 1
= Hn − Hn − xn
1 − x n=1 n
∞
1 X 3Hn2

3Hn 1
= − 2 + 3 xn
1 − x n=1 n n n
" ∞ ∞ ∞
#
1 X H2 X Hn n X xn
n n
= 3 x −3 x + .
1−x n=1
n n=1
n2 n=1
n3
Gathering the results from (2.14) and (2.8) ends the proof.
A different method may be found in [28, p. 352–354].
P∞ 2
Hn n
2.1.12 n=1 n2 x

∞
X Hn2 n 1
2
x = Li4 (x) − 2 Li4 (1 − x) + 2 ln(1 − x) Li3 (1 − x) + Li22 (x)
n=1
n 2
1
− ln2 (1 − x) Li2 (1 − x) − ln(x) ln3 (1 − x) + 2ζ(4). (2.19)
3
Proof. Divide both sides of (2.14):

∞
X Hn2 n 1
x = Li3 (x) − ln(1 − x) Li2 (x) − ln3 (1 − x)
n=1
n 3
xn
xn−1 dx =
R
by x then integrate using n ,
∞
Hn2 n ln3 (1 − x)
Z
X 1 2 1
x = Li4 (x) + Li2 (x) − dx.
n=1
n2 2 3 x
1
For the remaining integral, set 1 − x = y then expand 1−y in series,
∞ Z
ln3 (1 − x) ln3 (y)
Z Z X
dx = − dy = − y n−1 ln3 (y)dy
x 1−y n=1
∞ n n
yn yn

IBP
X
3 y 2 y
= − ln (y) − 3 ln (y) 2 + 6 ln(y) 3 − 6 4
n=1
n n n n
= ln3 (y) ln(1 − y) + 3 ln2 (y) Li2 (y) − 6 ln(y) Li3 (y) + 6 Li4 (y)
{substitute y = 1 − x back}
= ln (1 − x) ln(x) + 3 ln2 (1 − x) Li2 (1 − x) − 6 ln(1 − x) Li3 (1 − x)
3
+6 Li4 (1 − x). (2.20)
On plugging in this integral, we arrive at

∞
X Hn2 n 1 1
2
x = Li4 (x) + Li22 (x) − ln3 (1 − x) ln(x) − ln2 (1 − x) Li2 (1 − x)
n=1
n 2 3
+2 ln(1 − x) Li3 (1 − x) − 2 Li4 (1 − x) + C.
The proof follows on extracting C = 2ζ(4) by setting x = 0.
P∞
2.1.13 n=1 Hn Hn(2) xn

∞
X 1 1
Hn Hn(2) xn = Li3 (x) + Li3 (1 − x) + ln(x) ln2 (1 − x)
n=1
1−x 2

−ζ(2) ln(1 − x) − ζ(3) . (2.21)
(2)
Proof. Set an = Hn Hn in (2.1),
∞ ∞
X 1 X (2)

Hn Hn(2) xn = Hn Hn(2) − Hn−1 Hn−1 xn
n=1
1 − x n=1
∞
1 X (2) 1 (2) 1
= Hn Hn − Hn − Hn − 2 xn
1 − x n=1 n n
∞
!
(2)
1 X Hn Hn 1
= + − 3 xn
1 − x n=1 n2 n n
∞ ∞ (2)
1 X Hn n 1 X Hn n Li3 (x)
= x + x − .
1 − x n=1 n2 1 − x n=1 n 1−x
Collect the results from (2.8) and (2.10) to complete the proof.
Another approach may be found in [28, pp. 350–552].
P∞ 3
2.1.14 n=1 (Hn − 3Hn Hn(2) + 2Hn(3) )xn

∞
X ln3 (1 − x)
Hn3 − 3Hn Hn(2) + 2Hn(3) xn = − . (2.22)
n=1
1−x
(2) (3)
Proof. Set an = Hn3 − 3Hn Hn + 2Hn in (2.1),
∞
X
Hn3 − 3Hn Hn(2) + 2Hn(3) xn
n=1
∞
1 Xh 3 (2) (3)
i
= Hn − 3Hn Hn(2) + 2Hn(3) − Hn−1
3
+ 3Hn−1 Hn−1 − 2Hn−1 xn
1 − x n=1
∞
" ! #
(2) (2)
1 X Hn2 − Hn Hn 6
= 3 −6 + 3 xn
1 − x n=1 n n n
∞ xn ∞
1 X 1 X Hn n 6 Li3 (x)
= ·3 Hn2 − Hn(2) − ·6 x + ,
1−x n=1
n 1 − x n=1
n2 1−x
and the proof ends on collecting the results from (2.13) and (2.8).
P∞ (2)
Hn Hn
2.1.15 n=1 n
xn

∞ (2)
X Hn Hn n x
x = − Li4 + Li4 (1 − x) − ln(1 − x) Li3 (1 − x)
n=1
n x−1
1 1 1 4
+ ln2 (1 − x) Li2 (1 − x) + ln(x) ln3 (1 − x) − ln (1 − x) − ζ(4).
2 6 24
(2.23)

1 (2) (3)
Hn3 + 3Hn Hn + 2Hn
Z
xn−1 ln3 (1 − x)dx = − .
0 n
n
by −y then take the summation over n ≥ 1,
∞ ∞
!
(2) (3) Z 1 3
X Hn3 + 3Hn Hn + 2Hn n ln (1 − x) X
n
y =− (xy) dx
n=1
n 0 x n=1
Z 1 3
ln (1 − x) xy
=− dx
0 x 1 − xy
y ln3 (1 − x) y ln3 (t)
Z 1 Z 1
1−x=t
=− dx = − dt
0 1 − xy 0 1 − y + yt
{make use of (1.112) for the latter integral}

y
= 6 Li4 .
y−1
Therefore,
∞ (2) (3)
Hn3 + 3Hn Hn + 2Hn n

X y
y = −6 Li4 . (2.24)
n=1
n y−1
To establish another relation, divide both sides of (2.22):

∞
X ln3 (1 − x)
Hn3 − 3Hn Hn(2) + 2Hn(3) xn = −
n=1
1−x
yn
Ry
by x then integrate from x = 0 to y using 0
xn−1 dx = n ,
∞ (2) (3) Z y 3
X Hn3 − 3Hn Hn + 2Hn n ln (1 − x)
y =− dx
n=1
n 0 x(1 − x)
Z y 3 Z y 3
ln (1 − x) ln (1 − x)
=− dx − dx.
0 x 0 1−x
The first integral is calculated in (2.20) and the second integral is 1

4 ln4 (1 − y). Thus,
∞ (2) (3)
X Hn3 − 3Hn Hn + 2Hn n
y = − ln3 (1 − y) ln(y) − 3 ln2 (1 − y) Li2 (1 − y)
n=1
n
1 4
+6 ln(1 − y) Li3 (1 − y) − 6 Li4 (1 − y) + ln (1 − y) + 6ζ(4). (2.25)
4
The proof finalizes on combining (2.24) and (2.25) then dividing by 6.
P∞ 3
Hn n
2.1.16 n=1 n x

∞
Hn3 n

X x
x = −3 Li4 − 3 Li4 (1 − x) − 2 Li4 (x) + Li22 (x)
n=1
n x − 1
1 4
+3 ln(1 − x) Li3 (1 − x) + 2 ln(1 − x) Li3 (x) + ln (1 − x)
8
3 1
− ln2 (1 − x) Li2 (1 − x) − ln(x) ln3 (1 − x) + 3ζ(4). (2.26)
2 2
Proof. Combine (2.24) and (2.25) then divide by 2,

∞
Hn3 n

X x 1 3
x = −3 Li4 − ln3 (1 − y) ln(x) − ln2 (1 − x) Li2 (1 − x)
n=1
n x − 1 2 2
∞ (3)
1 4 X Hn n
+3 ln(1 − x) Li3 (1 − x) − 3 Li4 (1 − x) + ln (1 − x) + 3ζ(4) − 2 x .
8 n=1
n
Substitute the result from (2.17) to finalize the proof.
P∞ (2) (3) (2) (4)

2.1.17 4
n=1 (Hn
2
− 6Hn Hn + 8Hn Hn + 3(Hn )2 − 6Hn )xn

∞
ln4 (1 − x)
X 2
Hn4 − 6Hn2 Hn(2) + 8Hn Hn(3) + 3 Hn(2) − 6Hn(4) xn = .
n=1
1−x
(2.27)
2
(2) (3) (2) (4)
Proof. Put an = Hn4 − 6Hn2 Hn + 8Hn Hn + 3 Hn − 6Hn in (2.1),
∞
X 2
Hn4 − 6Hn2 Hn(2) + 8Hn Hn(3) (2)
+ 3 Hn (4)
− 6Hn xn
n=1
∞
1 Xh 4 4

(2)

− 6 Hn2 Hn(2) − Hn−1
2

= Hn − Hn−1 Hn−1
1 − x n=1
2 2
(3) (3) (2) (2)
+8 Hn Hn − Hn−1 Hn−1 + 3 Hn − Hn−1
i
(4)
−6 Hn(4) − Hn−1 xn

∞
!
 X (2)
1  4Hn 2Hn 6
= 6 + − 4 xn
1−x
 n=1 n3 n2 n
| {z }
S1

∞
!
(2) (3)
Hn3 3H 2

X 3Hn Hn 2Hn 3
− 2n + 4 n

+4 − + x .
n=1
n n n n n 
| {z }
S2
The sum S1 is the Cauchy product of Li22 (x) given in (2.83). For S2 , by collecting
the results from (2.14), (2.18), and (2.21), we have
∞
3Hn2

X
3 (2) (3) 3
Hn − 3Hn Hn + 2Hn − + 3 xn
n=1
n n
x ln3 (1 − x)
=− + 3 ln(1 − x) Li2 (x).
1−x
xn
Divide both sides of the latter identity by x then integrate using xn−1 dx =
R
n ,
∞
!
(2) (3)
X Hn3 3Hn Hn 2Hn 3Hn2 3
− + − 2 + 4 xn := S2
n=1
n n n n n
3
ln (1 − x) ln(1 − x) Li2 (x)
Z Z
=− dx + 3 dx
1−x x
1 3
= ln4 (1 − x) − Li22 (x) + C.
4 2
Setting x = 0 gives C = 0. Collect S1 and S2 to complete the proof.
For another proof, check [28, p. 355].
P∞
2.1.18 n=1 H n xn

∞
X ln(1 + x)
H n xn = . (2.28)
n=1
1−x
Proof.

ln(1 + x) 1
= (ln(1 + x))
1−x 1−x
{expand both function in Taylor series}

∞
! ∞ ! ∞
! ∞ !
X (−1)n−1 n X
n−1 1 X (−1)n−1 n X
n
= x x = x 1x
n=1
n n=1
x n=1 n n=1
(−1)n−1

0
apply (2.80) with an = and bn = 1 = n
n
∞ ∞
n
! n
!
1 X n+1 X (−1)k−1 X X (−1) k−1
= x (n − k + 1)0 = xn
x n=1 k n=1
k
k=1 k=1
( n
)
X (−1)k−1
use = H n defined in (1.160)
k
k=1
∞
X
= H n xn .
n=1
P∞ Hn n
2.1.19 n=1 n
x

∞
X Hn n 1−x 1
x = Li2 − Li2 (−x) − ln(2) ln(1 − x) − Li2 . (2.29)
n=1
n 2 2

∞
X ln(1 + x)
H n xn =
n=1
1−x
xn
xn−1 dx =
R
∞ Z Z Z
X Hn n ln(1 + x) ln(1 + x) ln(1 + x)
x = dx = dx + dx
n=1
n x(1 − x) x 1−x

1+x
write ln(1 + x) = ln + ln(2) in the second integral
2
ln 1+x
Z Z Z
ln(1 + x) 2 ln(2)
= dx − dx + dx
x 1−x 1−x
Z
1−x
= − Li2 (−x) − d Li2 − ln(2) ln(1 − x)
2

1−x
= − Li2 (−x) − Li2 − ln(2) ln(1 − x) + C.
2
1

The proof finishes on extracting C = Li2 2 by setting x = 0.
P∞ Hn n
2.1.20 n=1 n2 x
For |x| ≤ 1, x 6= −1, the following identity holds:

∞
X Hn n 2x x 1+x
x = Li3 − Li3 − Li3 − Li3 (−x)
n=1
n2 1+x 1+x 2

1 1 1
− Li3 (x) + Li3 + ln(1 + x) Li2 (x) + Li2 + ln(2) ln(1 + x) .
2 2 2
(2.30)
xn
xn−1 dx =
R
∞ 1−x
1

Li2 − Li2
Z
X Hn n 2 2
2
x = dx
n=1
n x
| {z }
IBP
ln(1 − x)
Z Z
Li2 (−x)
− dx − ln(2) dx
x x
Z
1−x 1 ln(x) ln(1 + x)
= ln(x) Li2 − Li2 − dx
2 2 1−x
ln(1 − x)
Z Z Z
ln(x) Li2 (−x)
+ ln(2) dx − dx − ln(2) dx
1−x x x
Z
1−x 1 ln(x) ln(1 + x)
= ln(x) Li2 − Li2 − dx
2 2 1−x
+ ln(2) Li2 (1 − x) − Li3 (−x) + ln(2) Li2 (x). (2.31)
For the remaining integral, set a = ln(x) and b = ln(1 + x) in the algebraic identity:
2ab = a2 + b2 − (a − b)2 ,

2 2 2 x
2 ln(x) ln(1 + x) = ln (x) + ln (1 + x) − ln .
1+x
Divide both sides by 1 − x then integrate,

Z Z 2 Z 2 Z ln2 x
ln(x) ln(1 + x) ln (x) ln (1 + x) 1+x
2 dx = dx + dx − dx .
1−x 1−x 1−x 1−x
| {z } | {z } | {z }
I1 I2 I3
1
For I1 , expand 1−x in series,
∞ Z
ln2 (x)
Z X
dx = xn−1 ln2 (x)dx
1−x n=1
∞ n
xn xn

IBP
X x
= ln2 (x) − 2 ln(x) 2 + 2 3
n=1
n n n
= − ln2 (x) ln(1 − x) − 2 ln(x) Li2 (x) + 2 Li3 (x). (2.32)
For I2 , substitute 1 + x = y,
ln2 (1 + x) ln2 (y)

Z Z
dx = dy
1−x 2−y
∞
( )
1 X y n−1
expand in Taylor series as
2−y n=1
2n
∞ Z
X 1
= n
y n−1 ln2 (y)dy
n=1
2
∞
yn yn yn

X 1 2
= ln (y) − 2 ln(y) + 2
n=1
2n n n2 n3
y y y
= − ln2 (y) ln 1 − − 2 ln(y) Li2 + 2 Li3
2 2 2
{substitute y = 1 + x back}

1−x 1+x 1+x
= − ln2 (1 + x) ln − 2 ln(1 + x) Li2 + 2 Li3 .
2 2 2
x
For I3 , substitute 1+x = t,

Z ln2 x
ln2 (t) ln2 (t) ln2 (t)
Z Z Z
1+x
dx = dt = 2 dt − dt
1−x (1 − t)(1 − 2t) 1 − 2t 1−t
= − ln2 (t) ln(1 − 2t) − 2 ln(t) Li2 (2t) + 2 Li3 (2t)
+ ln2 (t) ln(1 − t) + 2 ln(t) Li2 (t) − 2 Li3 (t)

2 1 − 2t
= − ln (t) ln − 2 ln(t) (Li2 (2t) − Li2 (t)) + 2 Li3 (2t) − 2 Li3 (t)
1−t

x
substitute t = back
1+x

2x x 2 x
= 2 Li3 − 2 Li3 − ln ln(1 − x)
1+x 1+x 1+x

x 2x x
−2 ln Li2 − Li2 .
1+x 1+x 1+x
Gather the three integrals then divide by 2,

Z
ln(x) ln(1 + x) x 2x 1+x
dx = Li3 − Li3 + Li3
1−x 1+x 1+x 2

x 2x x
+ Li3 (x) + ln Li2 − Li2
1+x 1+x 1+x

1+x 1
− ln(1 + x) Li2 − ln(x) Li2 (x) − ln2 (x) ln(1 − x)
2 2

1 2 1−x 1 2 x
− ln (1 + x) ln + ln ln(1 − x).
2 2 2 1+x
Substitute this integral in (2.31) then factor ln(1 + x), ln(x), and ln(2) out,
∞
X Hn n 2x x 1+x
x = Li3 − Li3 − Li3 − Li3 (−x)
n=1
n2 1+x 1+x 2

2x x 1+x
− Li3 (x) + ln(1 + x) Li2 − Li2 + Li2
1+x 1+x 2

2x x 1−x
− ln(x) Li2 − Li2 − Li2 − Li2 (x)
1+x 1+x 2

1 1 2 1−x
+ Li2 + ln(2)[Li2 (x) + Li2 (1 − x)] + ln (1 + x) ln
2 2 2

1 2 x 1 2
− ln ln(1 − x) + ln (x) ln(1 − x) + C
2 1+x 2
{substitute the relations from (1.128), (1.129), and (1.119) }

2x x 1+x
= Li3 − Li3 − Li3 − Li3 (−x)
1+x 1+x 2

1
− Li3 (x) + ln(1 + x) Li2 (x) + Li2 + ln(2) ln(1 + x)
2
− ln(x) [ln(1 − x) ln(1 + x) − ln(2) ln(1 − x)]

1 2 1−x
+ ln(2)[ζ(2) − ln(x) ln(1 − x)] + ln (1 + x) ln
2 2

1 x 1
− ln2 ln(1 − x) + ln2 (x) ln(1 − x) + C.
2 1+x 2
To extract the constant C, put x = 0,

1
C = Li3 − ln(2)ζ(2).
2
Therefore,
∞
X Hn n 2x x 1+x
x = Li3 − Li3 − Li3 − Li3 (−x)
n=1
n2 1+x 1+x 2

1 1
− Li3 (x) + Li3 + ln(1 + x) Li2 (x) + Li2 + ln(2) ln(1 + x)
2 2
− ln(x) [ln(1 − x) ln(1 + x) − ln(2) ln(1 − x)]

1 2 1−x
+ ln(2)[ζ(2) − ln(x) ln(1 − x)] + ln (1 + x) ln
2 2

1 2 x 1 2
− ln ln(1 − x) + ln (x) ln(1 − x) − ln(2)ζ(2).
2 1+x 2
The proof completes on simplifying the last three lines to − 21 ln(2) ln2 (1 + x).
A different proof may be found in [8, p. 9].
P∞
2.1.21 n=1 H n2 xn
The following identity is derived by Wolfgang Hintze (see [12]):

∞
X 2 ln(2)x 2 ln(1 − x)
H n2 xn = − − , |x| < 1. (2.33)
n=1
1 − x2 1 − x2
Proof. His proof starts with considering the integral form of the harmonic number,
∞ ∞ Z 1 ∞
n Z 1 !
X
n
X 1−y2 n 1 X
n √ n
H n2 x = dy x = x − (x y) dy
n=1 n=1 0 1−y 0 1−y n=1
Z 1 √ √ Z 1
1 x x y y=u 2u x xu
= − √ dy = 2
− du.
0 1−y 1−x 1−x y 0 1−u 1 − x 1 − xu
Let’s calculate this integral indefinitely:

Z
2u x xu
− du
1 − u2 1 − x 1 − xu
2xu2
Z Z
x 2u
= 2
du − du
1−x 1−u (1 − xu)(1 − u2 )
2xu2

x x
write = +
(1 − xu)(1 − u2 ) (1 − x)(1 − u) (1 + x)(1 + u)

2x
− in the second integral
(1 − x2 )(1 − xu)
x ln(1 − u2 )
Z Z Z
x du x du 2x du
=− − − + 2
1−x 1−x 1−u 1+x 1+u 1−x 1 − xu
x ln(1 − u2 ) x ln(1 − u) x ln(1 + u) 2 ln(1 − xu)
=− + − −
1−x 1−x 1+x 1 − x2
x ln(1 + u) x ln(1 + u) 2 ln(1 − xu)
=− − −
1−x 1+x 1 − x2
2x ln(1 + u) 2 ln(1 − xu)
=− − .
1 − x2 1 − x2
Thus, we have
∞ Z 1
X 2u x xu
H n2 xn = − du
n=1 0 1 − u2 1 − x 1 − xu
u=1
2x ln(1 + u) 2 ln(1 − xu) 2 ln(2)x 2 ln(1 − x)
=− − =− − .
1 − x2 1 − x2 u=0 1 − x2 1 − x2
P∞ Hn
2.1.22 n=1 n
2
xn
The following identity is also derived by Wolfgang Hintze (see [12]):

∞
H n2 n

X 1−x 1 1
x = Li2 − Li2 + 2 Li2 (x) + ln2 (1 − x)
n=1
n 2 2 2

1−x
+ ln(1 + x) ln , |x| ≤ 1, x 6= 1. (2.34)
2
Proof. His approach begins with dividing both sides of (2.33):

∞
X 2 ln(2)x 2 ln(1 − x)
H n2 xn = − −
n=1
1 − x2 1 − x2
xn
xn−1 dx =
R
by x then integrating using n ,
∞
H n2 n 2 ln(1 − x)
Z Z
X dx
x = −2 ln(2) 2
− dx
n=1
n 1 − x x(1 − x2 )
ln(1 − x) ln(1 − x) ln(1 − x)
Z Z Z Z
2 dx
= ln(2) − − 2 dx + dx − dx
1 − x2 x 1+x 1−x

1−x ln(1 − x)
Z
1
= ln(2) ln + 2 Li2 (x) + dx + ln2 (1 − x).
1+x 1+x 2
For the remaining integral, apply integration by parts,
ln(1 − x)
Z Z
ln(1 + x)
dx = ln(1 + x) ln(1 − x) + dx
1+x 1−x

1+x
write ln(1 + x) = ln + ln(2)
2
ln 1+x
Z Z
2 ln(2)
= ln(1 + x) ln(1 − x) + dx + dx
1−x 1−x

1−x
= ln(1 + x) ln(1 − x) + Li2 − ln(2) ln(1 − x)
2

1−x 1−x
= Li2 + ln(1 + x) ln .
2 2
Plug this integral back in,

∞
H n2 n

X 1−x 1
x = Li2 + 2 Li2 (x) + ln2 (1 − x)
n=1
n 2 2

1−x
+ ln(1 + x) ln + C,
2
and the proof follows on extracting C = − Li2 ( 12 ) by setting x = 0.
P∞ Hn
2.1.23 n=1 n2
2
xn

∞
H n2 n

X 2x x 1+x 1
2
x = Li3 − Li3 − Li3 + Li3
n=1
n 1+x 1+x 2 2

1 1 1 x
+ ln(1 + x) Li2 + Li2 (x) + Li2 − ln(1 + x) ln
1+x 2 2 2
1 1
− Li3 (1 − x2 ) + ln(1 − x2 ) Li2 (1 − x2 ) + ln(2)[Li2 (−x) − Li2 (x)]
2 2
1 3 1 2 2 1 1
+ Li3 (x) + ln (1 + x) + ln(x) ln (1 − x ) − ζ(3) + Li3 .
3 2 2 1+x
(2.35)
Proof. Bring back (2.34)

∞
H n2 n

X 1−x 1
x = Li2 − Li2 + 2 Li2 (x) − ln(2) ln(1 + x)
n=1
n 2 2
1
+ ln2 (1 − x) + ln(1 + x) ln(1 − x)
2
1 2 1 2 2 1 2
use ln (1 − x) + ln(1 + x) ln(1 − x) = ln (1 − x ) − ln (1 + x)
2 2 2

1−x 1
= Li2 − Li2 + 2 Li2 (x) − ln(2) ln(1 + x)
2 2
1 1
+ ln2 (1 − x2 ) − ln2 (1 + x).
2 2
Combine this identity and (2.29) then replace x by y,
∞ ∞
X H n2 n X Hn n 1
y − y = 2 Li2 (y) + Li2 (−y) + ln2 (1 − y 2 )
n=1
n n=1
n 2

1 1−y
− ln2 (1 + y) + ln(2) ln . (2.36)
2 1+y
Rx n
Divide both sides by y then integrate from y = 0 to x using 0 y n−1 dy = xn ,
∞ ∞
X H n2 n X Hn n
2
x − x
n=1
n n=1
n2
1 x ln2 (1 − y 2 ) 1 x ln2 (1 + y)
Z Z
= 2 Li3 (x) + Li3 (−x) + dy − dy
2 0 y 2 0 y
| {z } | {z }
I1 I2
+ ln(2)[Li2 (−x) − Li2 (x)]. (2.37)
For I1 , substitute 1 − y 2 = t,
x
ln2 (1 − y 2 ) 1
ln2 (t)
Z Z
1
I1 = dy = dt
0 y 2 1−x2 1−t
1
1 2
= − ln (t) ln(1 − t) − 2 ln(t) Li2 (t) + 2 Li3 (t)
2 1−x2
= ζ(3) + ln(x) ln2 (1 − x2 ) + ln(1 − x2 ) Li2 (1 − x2 ) − Li3 (1 − x2 ).
1
For I2 , substitute 1+y = t,
x
ln2 (1 + y) 1
ln2 (t) 1
ln2 (t) 1
ln2 (t)
Z Z Z Z
I2 = dy = dt = dt + dt
0 y 1
1+x
t(1 − t) 1
1+x
t 1
1+x
1−t
{recall the result from (2.32) for the second integral}
1
1 3
= ln (t) − ln2 (t) ln(1 − t) − 2 ln(t) Li2 (t) + 2 Li3 (t)
3 1
1+x

1 1
= 2ζ(3) − 2 Li3 − 2 ln(1 + x) Li2
1+x 1+x
2
+ ln(x) ln2 (1 + x) − ln3 (1 + x).
3
Plug in the results of I1 and I2 along with the result from (2.30) in (2.37), the proof
is finalized.
P∞ (2n
n)
2.1.24 n=1 4n
Hn xn

∞ 2n
√
X
n n 2 1+ 1−x
Hn x = √ ln √ . (2.38)
n=1
4n 1−x 2 1−x

R1 n
Proof. Using Hn = 0 1−y1−y dy, we have
∞ 2n
∞ Z 1
1 − yn

X
n
X 1 2n
n
Hn x = dy xn
n=1
4n n=1
4 n n 0 1 − y
∞ ∞
!
1 2n 2n
Z
1 X
n n
X
n n
= x − (xy) dy
0 1 − y n=1 4n n=1
4n
{recall Taylor series for the two sums}
Z 1
1 1 1
= √ −√ dy.
0 1−y 1−x 1 − xy
Let’s find both integrals indefinitely starting with the first one:
ln(1 − y)
Z
1
√ dy = − √ .
(1 − y) 1 − x 1−x
√
For the second integral, make the change of variable 1 − xy = t,
Z Z
1 dt 2 t
√ dy = 2 = √ arctanh √
(1 − y) 1 − xy 1 − x − t2 1−x 1−x
p
{substitute t = 1 − xy back}
√
2 1 − xy
=√ arctanh √
1−x 1−x

1+z
use 2 arctanh z = ln
1−z
√ √
1 1 − x + 1 − xy
=√ ln √ √
1−x 1 − x − 1 − xy
√ √
1 − x + 1 − xy
multiply the argument of the log by √ √
1 − x + 1 − xy
√ √
( 1 − x + 1 − xy)2

1
=√ ln .
1−x −x(1 − y)
Combine the two integrals,

∞ 2n
Z 1
X
n n 1 1 1
Hn x = √ −√ dy
n=1
4n 0 1−y 1−x 1 − xy
√ √ y=1
( 1 − x + 1 − xy)2

ln(1 − y) 1
= − √ −√ ln
1−x 1−x −x(1 − y) y=0
√ √ 2
y=1
1 ( 1 − x + 1 − xy)
= −√ ln
1−x −x y=0
√ 2
√
(1 + 1 − x)2

1 (2 1 − x)
= −√ ln − ln
1−x −x −x
√ 2 √
1 2 1−x 2 1+ 1−x
= −√ ln √ =√ ln √ .
1−x 1+ 1−x 1−x 2 1−x
P∞ (2n
n ) Hn
2.1.25 n=1 4n n
xn

∞ 2n
√
X
n Hn n 1− 1−x
x = 2 Li 2 √ . (2.39)
n=1
4n n 1+ 1−x
xn
xn−1 dx =
R
∞ 2n
√
1+ 1−x
Z
X
nHn n 2
x = √ ln √ dx
n=1
4n n 1−x 2 1−x

√ Z ln 1+y
y= 1−t ln(1 − t)
Z
1−x=y 2y 1+t
= −2 2
dy = −2 dt
1−y t
√
1−y 1− 1−x
= 2 Li2 (t) = 2 Li2 = 2 Li2 √ + C.
1+y 1+ 1−x
Setting x = 0 gives C = 0 and the proof is completed.
P∞ (2n
n ) Hn
2.1.26 n=1 4n n2
xn

∞
! !
2n
r r
X
n Hn n 2x 2x
x = −4 ln 1 + Li2
n=1
4n n2 1+x 1+x
r ! √ 2x
ln(1 − t) ln(1 + t)
Z 1+x
2x
+2 Li3 −4 dt. (2.40)
1+x 0 t
R y n−1Divideynboth sides of (2.39) by x then integrate from x = 0 to y using

Proof.
0
x dx = n ,
√
∞ 2n
Z y Li2 1−√1−x
X
n H n n 1+ 1−x
n n2
y =2 dx
n=1
4 0 x

√ Z √1−y u Li2 1−u
1−x=u 1+u
= −4 2
du
1 1 − u
q
2y
1−u
(1 − t) Li2 (t)
Z 1+y
1+u =t
= 2 dt
0 t(1 + t)

1−t 1 2
write = −
t(1 + t) t 1+t
q q
Z 1+y 2y Z 1+y 2y
Li2 (t) Li2 (t)
=2 dt − 4 dt
0 t 1+t
|0 {z }
IBP
q q
2y 2y
1+y 1+y
= 2 Li3 (t) − 4 ln(1 + t) Li2 (t)
0 0
q
2y
ln(1 − t) ln(1 + t)
Z 1+y
−4 dt
0 t
r r r
2y 2y 2y
= 2 Li3 − 4 ln 1 + Li2
1+y 1+y 1+y
q
2y
ln(1 − t) ln(1 + t)
Z 1+y
−4 dt.
0 t
The proof follows on replacing y by x.

P∞ 2H2n −Hn 2n
2.1.27 n=1 n
x

∞
X 2H2n − Hn 2n
x = 2 arctanh2 (x). (2.41)
n=1
n
Proof. We begin with

1 1−x
arctanh(x) = − ln .
2 1+x
Squaring both sides,

1 2 1−x
2
arctanh (x) = ln
4 1+x

1 2 1 2 1 2 1 2
use the algebric identity (a − b) = a + b − (a + b)
4 2 2 4
{with a = ln(1 − x) and b = ln(1 + x)}
1 2 1 1
= ln (1 − x) + ln2 (1 + x) − ln2 (1 − x2 )
2 2 4
{expand all squared logs in series given in (2.6)}
∞ ∞ ∞
X Hn−1 n X Hn−1 n 1 X Hn−1 2n
= x + (−1)n x − x
n=1
n n=1
n 2 n=1 n
∞ ∞ ∞
( )
X X X
n
use an + (−1) an = 2 a2n given in (1.5) for the first two sums
n=1 n=1 n=1
∞ ∞
X H2n−1 2n 1 X Hn−1 2n
=2 x − x
n=1
2n 2 n=1 n
∞ 1 ∞ 1 ∞
X H2n − 1 X Hn − 1 X 2H2n − Hn 2n
=2 2n
x2n − n
x2n = x ,
n=1
2n 2 n=1 n 2 n=1 n
If we replace x by ix in (2.41) then use arctanh2 (ix) = − arctan2 (x), we get

∞
X 2H2n − Hn 2n
(−1)n x = −2 arctan2 (x), |x| ≤ 1. (2.42)
n=1
n
Furthermore, by differentiating (2.41) and (2.42) with respect to x, we find

∞
X 2 arctanh(x)
(2H2n − Hn )x2n−1 = , (2.43)
n=1
1 − x2
∞
X 2 arctan(x)
(−1)n (2H2n − Hn )x2n−1 = − . (2.44)
n=1
1 + x2
P∞ H2n 2n+1
2.1.28 n=1 2n+1 x

∞
X H2n 2n+1 1
x = − arctanh(x) ln(1 − x2 ). (2.45)
n=1
2n + 1 2
Proof.
1
arctanh(x) ln(1 − x2 ) = {ln(1 + x) − ln(1 − x)}{ln(1 + x) + ln(1 − x)}
2
1 1
= ln2 (1 + x) − ln2 (1 − x)
2 2
{expand both squared logs in series given in (2.6)}
∞ ∞
X Hn−1 n X Hn−1 n
= (−1)n x − x
n=1
n n=1
n
∞ ∞ ∞
( )
X X X
n
use (−1) an − an = −2 a2n+1 given in (1.7)
n=1 n=1 n=0
∞
X H2n 2n+1
= −2 x
n=0
2n +1
{let the index start from 1, since H0 = 0}
∞
X H2n 2n+1
= −2 x ,
n=1
2n + 1

Let’s differentiate both sides of (2.45) with respect to x,
∞
X x arctanh(x) ln(1 − x2 )
H2n x2n = − .
n=1
1 − x2 2(1 − x2 )
P∞ (−1)n H2n 2n+1

2.1.29 n=1 2n+1
x

∞
X (−1)n H2n 2n+1 1
x = − arctan(x) ln(1 + x2 ). (2.46)
n=1
2n + 1 2
Proof. Since −1 = i2 , we have

∞
X ∞
X
2 (−1)n t2n H2n = 2 (i)2n t2n H2n
n=1 n=1
∞ ∞ ∞
( )
X X X
n
use 2 a2n = an + (−1) an given in (1.5)
n=1 n=1 n=1
∞
X ∞
X
= (it)n Hn + (−it)n Hn
n=1 n=1
{make use of the generating function in (2.4)}
ln(1 − it) ln(1 + it)
=− −
1 − it 1 + it
ln(1 − it) + ln(1 + it) + it(ln(1 − it) − ln(1 + it))
=−
1 + t2
2
ln(1 + t ) + it(−2i arctan t)
=−
1 + t2
2
ln(1 + t ) 2t arctan t
=− − .
1 + t2 1 + t2
Therefore,
∞
X ln(1 + t2 ) 2t arctan t
2 (−1)n t2n H2n = − − .
n=1
1 + t2 1 + t2
Rx x2n+1
Integrate both sides from t = 0 to x using 0
t2n dt = 2n+1 ,
∞ Z x
(−1)n H2n 2n+1 ln(1 + t2 ) 2t arctan t
X
2 x =− + dt
n=1
2n + 1 0 1 + t2 1 + t2
Z x
=− d(ln(1 + t2 ) arctan t)
0
2 x
= − ln(1 + t ) arctan t 0
= − ln(1 + x2 ) arctan x.
Divide both sides by 2 to finish the proof.

For a different proof, replace x by ix in (2.45) then use arctanh(ix) = i arctan(x).
P∞ Hn −H2n 1

2.1.30 n=1 n
− 2n2
x2n

∞
X Hn − H2n 1
− x2n = ln(1 − x) ln(1 + x). (2.47)
n=1
n 2n2
Proof (i). Put a = ln(1 − x) and b = ln(1 + x) in the algebraic identity:

1 1
ab = (a + b)2 − (a − b)2 ,
4 4
we have

1 2 2 1 2 1−x
ln(1 − x) ln(1 + x) = ln (1 − x ) − ln
4 4 1+x
1 2 2
= ln (1 − x2 ) − arctanh (x)
4
{expand the first squared log in series given in (2.6)}
and substitute the result of arctanh2 (x) given in (2.41)

∞ ∞
1 X Hn−1 2n 1 X 2H2n − Hn 2n
= x − x
2 n=1 n 2 n=1 n
∞ ∞
1 X Hn − n1 2n 1 X 2H2n − Hn 2n
= x − x
2 n=1 n 2 n=1 n
∞
X Hn − H2n 1
= − 2 x2n .
n=1
n 2n
Proof (ii). Replace y by −y in (2.28):

∞
ln(1 + y) X
= H n yn ,
1−y n=1
we get
∞
ln(1 − y) X
= (−1)n H n y n .
1+y n=1
Subtract the two generating functions,

∞ ∞
ln(1 − y) ln(1 + y) X X
− = (−1)n H n y n − H n yn .
1+y 1−y n=1 n=1
2.2. Series Expansion of Powers of arcsin(z) 89
Integrate both sides from y = 0 to x using:

Z x
ln(1 − y) ln(1 + y)
− dy = ln(1 − x) ln(1 + x),
0 1+y 1−y
we have
∞ ∞
X (−1)n H n xn+1 X H n xn+1
ln(1 − x) ln(1 + x) = −
n=1
n+1 n=1
n+1
∞ ∞ ∞
( )
X X X
n
use (−1) an − an = −2 a2n−1 given in (1.8)
n=1 n=1 n=1
∞
X H 2n−1 2n
= −2 x
n=1
2n
(−1)n

use H n−1 = H n + given in (1.161)
n
∞ 1
X H 2n + 2n
=− x2n ,
n=1
n
and the proof follows on using H 2n = H2n − Hn given in (1.162).

For a different proof, see [28, p. 334].
2.2 Series Expansion of Powers of arcsin(z)

For |z| ≤ 1, the following identity holds:

∞ 2n

X
n z 2n+1
arcsin(z) = . (2.48)
n=0
4n 2n + 1
Proof. Differentiate arcsin(z) then integrate,

Z Z
1
arcsin(z) = d(arcsin(z)) = √ dz
1 − z2
∞
( )
2n

1 X
n 2n
expand √ in Taylor series as z
1 − z2 n=0
4n
∞ ∞
!
2n 2n Z
Z X
X
n 2n n 2n
= z dz = z dz
n=0
4n n=0
4n
∞ 2n

X
n z 2n+1
= + C.
n=0
4n 2n + 1
Extracting C = 0 by setting z = 0 completes the proof.

√
2 1−z
For |z| < 1, the following identity holds:

∞
arcsin(z) 1 X 4n z 2n−1
√ = . (2.49)
2 n=1 2n

1 − z2 n
n

π
4n 1
Z 2
sin2n−1 (x)dx = 2n 2n

0 n
by z 2n−1 then take the summation over n ≥ 1,

∞ ∞
Z π2 !
1 X 4n z 2n−1 1 X
2 2 n
= (z sin (x)) dx
2 n=1 2n

n
n 0 z sin(x) n=1
{employ the geometric series formula}
Z π2 2 2
1 z sin (x)
= dx
0 z sin(x) 1 − z 2 sin2 (x)
Z π2
z sin(x)
= dx
0 1 − z + z 2 cos2 (x)
2
x= π2
1 z cos(x)
= −√ arctan √
1 − z2 1 − z 2 x=0

1 z arcsin(z)
=√ arctan √ = √ .
1−z 2 1−z 2 1 − z2

z
To justify the last step, differentiate arctan √1−z 2
then integrate back.
If we integrate both sides of (2.49) from z = 0 to z, we get
∞
1 X 4n z 2n
arcsin2 (z) = 2, |z| ≤ 1. (2.50)
2 n=1 2n
n
n
2.2.3 Series Expansion of arcsin3 (z)
For |z| ≤ 1, the following equality holds:

∞ 2n

1 (2) z 2n+1

(2)
X
3 n
arcsin (z) = 6 H2n − Hn . (2.51)
n=0
4n 4 2n + 1

Proof. Let arcsin(z) = x in (2.48):
∞ 2n

X
n z 2n+1
arcsin(z) =
n=0
4n 2n + 1
and write
(2n)!2

2n (2n)!
= 2
= 2 ,
n n! n! (2n)!
we have
∞ ∞
X (2n)!2 sin2n+1 (x) X 2
x= = bn fn (x), (2.52)
n=0
4n n!2 (2n + 1)! n=0
where
sin2n+1 (x) (2n)!
fn (x) = , bn = .
(2n + 1)! 2n n!
Note that
fn00 (x) = fn (x) − (2n + 1)2 fn (x), (2.53)

bn+1 = (2n + 1)bn . (2.54)
In light of (2.52), write

∞
X
x3 = an b2n fn (x). (2.55)
n=0
Assuming a0 = 0 allows the index n to start from 1,

∞
X
x3 = an b2n fn (x). (2.56)
n=1
Differentiate both sides of (2.56) with respect to x twice then divide by 6,

∞
1X
x= an b2n fn00 (x)
6 n=1
{substitute the result of fn00 (x) given in (2.53)}
∞ ∞
1X 1X
= an b2n fn−1 (x) − (2n + 1)2 an b2n fn (x)
6 n=1 6 n=1
{shift the index n by +1 in the first sum}
{and let n start from 0 in the second sum, since we assumed a0 = 0}
∞ ∞
1X 1X
= an+1 b2n+1 fn (x) − (2n + 1)2 an b2n fn (x)
6 n=0 6 n=0
{substitute bn+1 given in (2.54) in the first sum}
∞ ∞
1 X 1X
= (2n + 1)2 an+1 b2n fn (x) − (2n + 1)2 an b2n fn (x)
6 n=0 6 n=0
∞
X (2n + 1)2 [an+1 − an ] 2
= bn fn (x). (2.57)
n=0
6
By comparing the series in (2.52) and (2.57), we see that
(2n + 1)2 [an+1 − an ]

1=
6
or
6
an+1 − an = .
(2n + 1)2
Employing the generalization in (1.149), we find

(2) 1
an = 6 H2n − Hn(2) . (2.58)
4
(2)
Notice that an in (2.58) meets our assumption (a0 = 0), since H0 = 0.
Substitute (2.58) in (2.55),
∞
X (2) 1
x3 = 6 H2n − Hn(2) b2n fn (x).
n=0
4
Finally, substitute fn (x) and bn back and let x = arcsin(z) to finish the proof.
2.2.4 Series Expansion of arcsin4 (z)
For |z| ≤ 1, the following identity holds:

∞ (2)
4 3 X 4n Hn−1 z 2n
arcsin (z) = . (2.59)
2 n=1 2n n2

n

Proof. Set arcsin(z) = x in (2.50):

∞
1 X 4n z 2n
arcsin2 (z) = 2,
2 n=1 2n
n
n
we get
∞
2 1 X 4n sin2n (x)
x = .
2 n=1 2n n2

n
Since
2n (2n)! (2n)! (2n)!
= 2
= 2
= 2 ,
n (n!) (n(n − 1)!) n (n − 1)!2
we have
∞ ∞
1X n sin2n (x) 1X 2
x2 = (2 (n − 1)!)2 = b fn (x), (2.60)
2 n=1 (2n)! 2 n=1 n
where
sin2n (x)
fn (x) = , bn = 2n (n − 1)!.
(2n)!
Note that
fn00 (x) = fn−1 (x) − (2n)2 fn (x), (2.61)
bn+1 = 2nbn . (2.62)
In light of (2.60), write
∞
41X
x = an b2n fn (x). (2.63)
2 n=1
Assuming a1 = 0 allows the index n to start from 2,
∞
41X
x = an b2n fn (x). (2.64)
2 n=2
Differentiating both sides of (2.64) twice with respect to x then dividing by 12,
∞
1 X
x = 2
an b2n fn00 (x)
24 n=2
{substitute the result of fn00 (x) from (2.61)}
∞ ∞
1 X 1 X
= an b2n fn−1 (x) − an (2n)2 b2n fn (x)
24 n=2 24 n=2
{shift the index n of the first sum by +1 }
{and let n start from 1 in the second sum, since we assumed a1 = 0}
∞ ∞
1 X 1 X
= an+1 b2n+1 fn (x) − an (2n)2 b2n fn (x)
24 n=1 24 n=1
{substitute bn+1 given in (2.62) in the first sum}
∞ ∞
1 X 1 X
(2n)2 an+1 b2n fn (x) − (2n)2 an b2n fn (x)
24 n=1 24 n=1
∞
1 X (2n)2 [an+1 − an ] 2
= bn fn (x). (2.65)
2 n=1 12
By comparing the coefficients of fn (x) in (2.60) and (2.65), we see that
(2n)2 [an+1 − an ]
1=
12
or
3
an+1 − an = .
n2
Using (1.145), we find
(2)
an = 3Hn−1 . (2.66)
Notice that an in (2.66) meets our assumption (a1 = 0).
Substitute (2.66) in (2.63), we obtain
∞
3 X (2) 2
x4 = H b fn (x).
2 n=1 n−1 n
Plugging fn (x) and bn back in and letting x = arcsin(z) completes the proof.
By differentiating both sides of (2.51) and (2.59), we obtain
∞ 2n

arcsin2 (z)

X
n (2) 1 (2) 2n
√ =2 H − H z |z| < 1 (2.67)
1 − z2 n=1
4n 2n
4 n
and
∞ (2)
arcsin3 (z) 3 X 4n Hn−1 z 2n−1
√ = . |z| < 1. (2.68)
4 n=1 2n

1 − z2 n
n
2.3. Identities by Beta Function 95
2.3 Identities by Beta Function

2.3.1 Expressing Beta Function as a Product
For n ∈ Z+ , the following identity holds:

n−1
Y 1
B(m, n) = Γ(n) . (2.69)
k+m
k=0

n
Γ(z + n + 1) Y 1
Γ(z) =
z z+k
k=1
z
by Γ(z+n+1) and use Γ(z + 1) = zΓ(z), we obtain
n
Γ(z + 1) Y 1
= .
Γ(z + n + 1) z+k
k=1
Next, replace z by m − 1 then multiply both sides by Γ(n),

n n−1
Γ(m)Γ(n) Y 1 Y 1
:= B(m, n) = Γ(n) = Γ(n) ,
Γ(m + n) k+m−1 k+m
k=1 k=0
where the last form follows from shifting the index k by +1.
2.3.2 Evaluation of Four Logarithmic Integrals
For n ∈ Z+ , the following identities hold:

Z 1
Hn
xn−1 ln(1 − x)dx = − ; (2.70)
0 n
1 (2)
Hn2 + Hn
Z
xn−1 ln2 (1 − x)dx = ; (2.71)
0 n
1 (2) (3)
Hn3 + 3Hn Hn + 2Hn
Z
xn−1 ln3 (1 − x)dx = − ; (2.72)
0 n
1 (2) (3) (2) (4)
Hn4 + 6Hn2 Hn + 8Hn Hn + 2(Hn )2 + 6Hn
Z
xn−1 ln4 (1 − x)dx = .
0 n
(2.73)

Proof. Take the logarithm of both sides of (2.69), we have

n−1
Y 1
ln B(m, n) = ln Γ(n) + ln
k+m
k=0
n Y X o
use ln an = ln(an ) given in (1.12)
n−1
X
= ln Γ(n) − ln(k + m).
k=0
Differentiate both sides with respect to m,

∂ n−1
∂mB(m, n) X 1
=−
B(m, n) k+m
k=0
or
n−1
∂ X 1
B(m, n) = − B(n, m) . (2.74)
∂m k+m
k=0
Differentiate both sides of the latter identity with respect to m,

n−1 n−1
∂2 ∂ X 1 X 1
B(m, n) = − B(m, n) + B(m, n) .
∂m2 ∂m k+m (k + m)2
k=0 k=0
∂
Substitute the result ofB(n, m) from (2.74),
∂m
 !2 n−1 
2 n−1
∂ X 1 X 1
B(m, n) = B(n, m)  + . (2.75)
∂m2 k+m (k + m)2
k=0 k=0
Differentiate (2.75) twice with respect to m,

 !3 !
n−1 n−1
∂3 X 1 X 1
B(m, n) = − B(m, n)  +3
∂m3 k+m k+m
k=0 k=0
n−1
! n−1
#
X 1 X 1
2
+ 2 , (2.76)
(k + m) (k + m)3
k=0 k=0
 !4 !
n−1 n−1
∂4 X 1 X 1
B(m, n) = B(m, n)  +6
∂m4 k+m (k + m)2
k=0 k=0
n−1
!2 n−1
! n−1
!
X 1 X 1 X 1
+8
k+m k+m (k + m)3
k=0 k=0 k=0
2.3. Identities by Beta Function 97
n−1
! n−1
#
X 1 X 1
+3 +6 . (2.77)
(k + m)2 (k + m)4
k=0 k=0
Now set m = 1 in (2.74), (2.75), (2.76), and (2.77) using:
Γ(1)Γ(n) Γ(n)Γ(1) 1
B(1, n) = = =
Γ(n + 1) nΓ(n) n
and
n−1 n
X 1 X 1
a
= = Hn(a) ,
(k + 1) ka
k=0 k=1
we get

∂ Hn
B(m, n) =− ;
∂m m=1 n
2 (2)
Hn2 + Hn

∂
B(m, n) = ;
∂m2 m=1 n
3 (2) (3)
Hn3 + 3Hn Hn + 2Hn

∂
B(m, n) = − ;
∂m3 m=1 n
4 (2) (3) (2) (4)
H 4 + 6Hn2 Hn + 8Hn Hn + 2(Hn )2 + 6Hn

∂
4
B(m, n) = n .
∂m m=1 n
On the other hand, by using the definition of the beta function:

Z 1 Z 1
m−1 n−1 1−x→x
B(m, n) = x (1 − x) dx = xn−1 (1 − x)m−1 dx,
0 0
we have
Z 1
∂ ∂ n−1 m−1
B(m, n) = x (1 − x) dx .
∂m m=1 ∂m 0 m=1
A special case of Leibniz’s integral rule (see [43]) is differentiation under the integral
sign theorem:
Z b Z b
d ∂
f (x, m)dx = f (x, m)dx. (2.78)
dm a a ∂m
Since the beta function has three variables (m, n, x), using (2.78), we have
Z b Z b
∂ ∂
B(x, m, n)dx = B(x, m, n)dx. (2.79)
∂m a a ∂m
Using this rule, we reach

Z 1
∂ ∂ n−1
B(m, n) = x (1 − x)m−1 dx
∂m m=1 0 ∂m m=1
Z 1
n−1 m−1
= x ln(1 − x)(1 − x) dx
0 m=1
Z 1
= xn−1 ln(1 − x)dx;
0
2 Z 1 2
∂ ∂ n−1 m−1
B(m, n) = x (1 − x) dx
∂m2 m=1 0 ∂m
2
m=1
Z 1
= xn−1 ln2 (1 − x)(1 − x)m−1 dx
0 m=1
Z 1
= xn−1 ln2 (1 − x)dx;
0
3 Z 1 3
∂ ∂ n−1 m−1
B(m, n) = x (1 − x) dx
∂m3 m=1 0 ∂m
3
m=1
Z 1
= xn−1 ln3 (1 − x)(1 − x)m−1 dx
0 m=1
Z 1
= xn−1 ln3 (1 − x)dx;
0
4 Z 1 4
∂ ∂ n−1 m−1
B(m, n) = x (1 − x) dx
∂m4 m=1 0 ∂m
4
m=1
Z 1
= xn−1 ln4 (1 − x)(1 − x)m−1 dx
0 m=1
Z 1
= xn−1 ln4 (1 − x)dx,
0
and the proof follows on comparing the last eight results.

A different proof, with no use of the beta function or the gamma function, may be
found in [28, pp. 59–62].
2.4. Identities by Cauchy Product 99
2.4 Identities by Cauchy Product

2.4.1 Cauchy Product of Two Power Series
P∞ P∞
Let n=1 an xn and n=1 bn xn be two power series. The Cauchy product of
these two series is given by
∞
! ∞ ! ∞ n
!
X X X X
n n n+1
an x bn x = x ak bn−k+1 . (2.80)
n=1 n=1 n=1 k=1
Proof.
∞ ∞
! !
X X
an xn bn xn
n=1 n=1
= a1 x + a2 x2 + a3 x3 + · · · b1 x + b2 x2 + b3 x3 + · · ·

= x2 (a1 b1 ) + x3 (a1 b2 + a2 b1 ) + x4 (a1 b3 + a2 b2 + a3 b1 ) + · · ·

1
! 2
! 3
!
X X X
2 3 4
=x ak b2−k + x ak b3−k + x ak b4−k + · · ·
k=1 k=1 k=1
∞ n
!
X X
= xn+1 ak bn−k+1 .
n=1 k=1
Following the steps above also gives

∞
! ∞ ! ∞ n
!
X X X X
n n
an x bn x = xn ak bn−k . (2.81)
n=0 n=0 n=0 k=0
2.4.2 Cauchy Product of − ln(1 − x) Li2 (x)
For |x| ≤ 1, x 6= 1, the following identity holds

∞ ∞ (2)
X Hn n X Hn n
− ln(1 − x) Li2 (x) = 2 x + x − 3 Li3 (x). (2.82)
n=1
n2 n=1
n
Proof. We follow the same approach as in [28, p. 516]:

Expand Li2 (x) and ln(1 − x) in series,
∞
! ∞ !
X xn X xn
(Li2 (x))(− ln(1 − x)) =
n=1
n2 n=1
n

1 1
employ the Cauchy product in (2.80) where an = 2 and bn =
n n
∞ n
!
X X 1
= xn+1 2 (n − k + 1)
n=1
k
k=1
{use the partial fraction decomposition for the inner sum}
∞ n
!
X
n+1
X 1 1 1
= x + +
n=1
(n + 1)k 2 (n + 1)2 k (n + 1)2 (n − k + 1)
k=1
( n n n
)
X 1 X 1 X1
use 2
= Hn(2) and = = Hn given in (1.3)
k n−k+1 k
k=1 k=1 k=1
∞
!
(2)
X
n+1 Hn 2Hn
= x +
n=1
(n + 1) (n + 1)2
(2)
{let the index n start from 0, since H0 = H0 = 0}
∞
!
(2)
X Hn 2Hn
= xn+1 +
n=0
n + 1 (n + 1)2
∞ (2)
!
X
n Hn−1 2Hn−1
= x +
n=1
n n2
∞ (2)
!
1 2
X Hn − n2 2Hn −
= xn + n
n=1
n n2
∞ (2) ∞ ∞
X Hn n X Hn n X xn
= x +2 2
x − 3
n=1
n n=1
n n=1
n3
∞ (2) ∞
X Hn n X Hn n
= x +2 x − 3 Li3 (x).
n=1
n n=1
n2
2.4.3 Cauchy Product of Li22 (x)
For |x| ≤ 1, the following identity holds

∞ ∞ (2)
X Hn n X Hn n
Li22 (x) = 4 x + 2 x − 6 Li4 (x). (2.83)
n=1
n3 n=1
n2

∞ ∞ (2)
X Hn n X Hn n
− ln(1 − x) Li2 (x) = 2 x + x − 3 Li3 (x)
n=1
n2 n=1
n
xn
xn−1 dx =
R
∞ ∞ (2)
X Hn n X Hn n
2 x + x − 3 Li4 (x)
n=1
n3 n=1
n2
− ln(1 − x) Li2 (x)
Z
1
= dx = Li22 (x) + C.
x 2
The proof finishes on extracting C = 0 by setting x = 0.

∞ ∞ (2) ∞ (3)
X Hn n X Hn n X Hn
− ln(1 − x) Li3 (x) = 2 x + x + − 4 Li5 (x).
n=1
n3 n=1
n2 n=1
n
(2.84)
Proof.
∞ ∞
!!
X Xxn xn
(Li3 (x))(− ln(1 − x)) =
n=1
n3
n=1
n

1 1
employ (2.80) where an = 3 and bn =
n n
∞ n
!
X X 1
= xn+1 3
n=1
k (n − k + 1)
k=1
{make use of the partial fraction decomposition for the inner sum}
∞ n
X
n+1
X1 1 1
= x 3
+ 2 k2
+
n=1
(n + 1)k (n + 1) (n + 1)3 k
k=1

1
+
(n + 1)3 (n − k + 1)
∞
!
(3) (2)
X
n+1 Hn Hn 2Hn
= x + +
n=1
n + 1 (n + 1)2 (n + 1)3
n o
(3) (2)
let the index n start from 0, since H0 = H0 = H0 = 0
∞
!
(3) (2)
X Hn Hn 2Hn
= xn+1 + +
n=0
n + 1 (n + 1)2 (n + 1)3
∞ (3) (2)
!
X Hn−1 H 2Hn−1
= x n
+ n−1 +
n=1
n n2 n3
∞ (3) (2)
!
1 1 1
X
n Hn − n3 Hn − n2 Hn − n
= x + +2
n=1
n n n2
∞ (3) ∞ (2) ∞ ∞
X Hn X Hn n X Hn n X xn
= xn + 2
x + 2 3
x − 4
n=1
n n=1
n n=1
n n=1
n4
∞ (3) ∞ (2) ∞
X Hn n X Hn n X Hn n
= x + 2
x + 2 x − 4 Li4 (x).
n=1
n n=1
n n=1
n3
2.4.5 Cauchy Product of Li2 (x) Li3 (x)

∞ ∞ (2) ∞ (3)
Li2 (x) Li3 (x) = 6 4
x + 3 3
x + x − 10 Li5 (x).
n=1
n n=1
n n=1
n2
(2.85)
Proof (i).
∞ ∞
! !
X xn X xn
Li3 (x) Li2 (x) =
n=1
n3 n=1
n2
∞ n
!
X X 1
= xn+1
n=1
k 3 (n − k + 1)2
k=1
∞ n
X X 1 2 1
= xn+1 + +
n=1
(n + 1)2 k 3 (n + 1)3 k 2 (n + 1)3 (n − k + 1)2
k=1

3 3
+ +
(n + 1)4 k (n + 1)4 (n − k + 1)
∞
!
(3) (2)
X
n+1 Hn Hn Hn
= x +3 +6
n=1
(n + 1)2 (n + 1)3 (n + 1)4
∞ (3) (2)
!
X H n−1 H n−1 H n−1
= xn +3 3 +6 4
n=1
n2 n n
∞ (3) (2)
!
1 1 1
X
n Hn − n3 Hn − n2 Hn − n
= x +3 +6
n=1
n2 n3 n4
∞ (3) ∞ (2) ∞ ∞
X Hn n X Hn n X Hn n X xn
= 2
x + 3 3
x + 6 x − 10
n=1
n n=1
n n=1
n4 n=1
n5
∞ ∞ (2) ∞ (3)
= 6 x + 3 x + x − 10 Li5 (x).
n=1
n4 n=1
n3 n=1
n2
Proof (ii).
Z
Li2 (x) Li3 (x) = d(Li2 (x) Li3 (x))
Z
1
Li22 (x) − ln(1 − x) Li3 (x) dx

=
x
{substitute the results from (2.83) and (2.84)}
∞ ∞ ∞
!
Z (2) (3)
1 X Hn n X Hn n X Hn n
= 6 x +3 x + x − 10 Li4 (x) dx
x n=1
n3 n=1
n2 n=1
n
xn
Z
interchange integration and summation then use xn−1 dx =
n
∞ ∞ (2) ∞ (3)
X Hn n X Hn n X Hn
=6 4
x +3 3
x + − 10 Li5 (x) + C.
n=1
n n=1
n n=1
n2
The proof finalizes on finding C = 0.
2.4.6 Cauchy Product of Li23 (x)

∞ ∞ (2) ∞ (3)
Li23 (x) = 12 x +6 x +2 x −20 Li6 (x). (2.86)
n=1
n5 n=1
n4 n=1
n3

∞ ∞ (2) ∞ (3)
Li2 (x) Li3 (x) = 6 4
x +3 3
x + x − 10 Li5 (x)
n=1
n n=1
n n=1
n2
by x then integrate,
∞ ∞ (2) ∞ (3)
6 x + 3 x + x − 10 Li6 (x)
n=1
n5 n=1
n4 n=1
n3
Z
Li2 (x) Li3 (x) 1
= dx = Li23 (x) + C,
x 2
and the proof follows as C = 0.

∞ ∞ (2) ∞ (3)
− ln(1 − x) Li4 (x) = 2 x + x + x
n=1
n4 n=1
n3 n=1
n2
∞ (4)
X Hn n
+ x − 5 Li5 (x). (2.87)
n=1
n

Proof.
∞ ∞
! !
X xn X xn
(Li4 (x))(− ln(1 − x)) =
n=1
n4 n=1
n
∞ n
!
X
n+1
X 1
= x
n=1
k 4 (n − k + 1)
k=1
∞
!
(4) (3) (2)
X Hn Hn Hn Hn
= xn+1 + 2
+ 3
+2
n=1
n + 1 (n + 1) (n + 1) (n + 1)4
∞ (4) (3) (2)
!
X H n−1 H H H n−1
= xn + n−12
+ n−1 +2 4
n=1
n n n3 n
∞ (4) (3) (2)
!
1 1 1 1
X
n Hn − n4 Hn − n3 Hn − n2Hn − n
= x + + +2
n=1
n n2 n3 n4
∞
!
(4) (3) (2)
X Hn Hn Hn Hn 5
= xn + 2
+ 3 +2 4 − 5
n=1
n n n n n
∞ ∞ (2) ∞ (3) ∞ (4)
X Hn n X Hn n X Hn n X Hn n
=2 x + x + x + x − 5 Li5 (x),
n=1
n4 n=1
n3 n=1
n2 n=1
n
and we are done with the proof.

Applying the Cauchy product, we also find, for |x| ≤ 1, the following identities:
∞ ∞ (2) ∞ (3)
Li2 (x) Li4 (x) = 8 x + 4 x + 2 x
n=1
n5 n=1
n4 n=1
n3
2.5. Identities by Abel’s Summation 105
∞ (4)
X Hn n
+ x − 15 Li6 (x); (2.88)
n=1
n2
∞ ∞ (2) ∞ (3)
Li3 (x) Li4 (x) = 20 x + 10 x + 4 x
n=1
n6 n=1
n5 n=1
n4
∞ (4)
X Hn n
+ x − 35 Li7 (x); (2.89)
n=1
n3
∞ ∞ (2) ∞ (3)
Li24 (x) = 40 x + 20 x + 8 x
n=1
n7 n=1
n6 n=1
n5
∞ (4)
X Hn n
+2 x − 70 Li8 (x). (2.90)
n=1
n4
Note that (2.90) follows from dividing both sides of (2.89) by x then integrating.
2.5 Identities by Abel’s Summation

2.5.1 Abel’s Summation
Pn Pn Pn
Given two finite sums k=1 ak and k=1 bk , define An = i=1 ai . Then
n
X n−1
X
ak bk = An bn + Am bm−1 − Ak (bk+1 − bk ) . (2.91)
k=m k=m
Pn
Proof. By the given sum, An = i=1 ai , one can write
k
X
Ak = ai = a1 + a2 + ... + ak−1 + ak
i=1
and so
k−1
X
Ak−1 = ai = a1 + a2 + ... + ak−1 .
i=1
Subtracting the two sums yields
Ak − Ak−1 = ak . (2.92)
Multiply both sides of (2.92) by bk then take the summation from k = m to n,

n
X n
X
a k bk = (Ak − Ak−1 ) bk
k=m k=m
n
X n
X
= Ak bk − Ak−1 bk .
k=m k=m
n
X n−1
X
For the first sum, use the fact that f (k) = f (n) + f (k),
k=m k=m
n
X n−1
X
Ak bk = An bn + Ak bk ,
k=m k=m
n
X n
X
and for the second sum, use f (k) = f (m) + f (k),
k=m k=m+1
n
X n
X
Ak−1 bk = Am−1 bm + Ak−1 bk
k=m k=m+1
{shift the index k by +1}

n−1
X
= Am−1 bm + Ak bk+1 .
k=m
Combining the two sums, we obtain

n
X n−1
X n−1
X
ak bk = An bn + Am−1 bm + Ak bk − Ak bk+1
k=m k=m k=m
n−1
X
= An bn + Am−1 bm − Ak (bk+1 − bk ) ,
k=m
For the two cases m = 0 and m = 1, we have Am−1 = 0. So, (2.91) becomes:
n
X n−1
X
ak bk = An bn − Ak (bk+1 − bk ) , (2.93)
k=0 k=0
n
X n−1
X
ak bk = An bn − Ak (bk+1 − bk ) . (2.94)
k=1 k=1
Also note that the index k in the RHS of (2.94) can start from 0, since A0 = 0. Thus,
n
X n−1
X n
X
ak bk = An bn − Ak (bk+1 − bk ) , An = ai . (2.95)
k=1 k=0 i=1
You may find in [6, Theorem 2.20, p. 55] a proof for a similar formula:
n
X n
X
ak bk = An bn+1 + Am−1 bm − Ak (bk+1 − bk ) .
k=m k=m
2.5.2 First Application
For integers p, q ≥ 2, the following identity holds:

∞ (p) ∞ (q)
X H k
X H k
+ = ζ(p)ζ(q) + ζ(p + q). (2.96)
kq kp
k=1 k=1
1 (p)
Proof. Let ak = kq and bk = Hk in (2.95),
n (p) n
! n−1 k
!
X H X 1 X X 1
(p) (p)

k
= Hn(p) − Hk+1 − Hk
kq i=1
iq i=1
iq
k=1 k=0

(p) 1
(p)
use Hk+1
= + Hk given in (1.145)
(k + 1)p
n−1
X (q)
(q) (p) 1
= Hn Hn − Hk
(k + 1)p
k=0
{shift the index k by −1}
n (q)
X Hk−1
= Hn(q) Hn(p) −
kp
k=1
X H (q)
n
− 1
k kq
= Hn(q) Hn(p) −
kp
k=1
n (q)
!
X Hk 1
= Hn(q) Hn(p) − − p+q
kp k
k=1
n (q)
X H
= Hn(q) Hn(p) + ζ(q + p) − k
.
kp
k=1
Reorganize the terms, we have

n (p) n (q)
X H X H
k
+ k
= Hn(q) Hn(p) + ζ(q + p).
kq kp
k=1 k=1
Next, take the limit on both sides letting n → ∞,

∞ (p) ∞ (q)
X H X H n o
k
+ k
= lim Hn(q) Hn(p) + ζ(q + p) ,
kq kp n→∞
k=1 k=1
and the proof follows on using

n ∞
X 1 X 1
lim Hn(q) = lim = = ζ(p).
n→∞ n→∞ kp kp
k=1 k=1
For a different approach, see [28, p. 358].

Setting q = p in (2.96) yields
∞ (p)
X H k ζ 2 (p) + ζ(2p)
= . (2.97)
kp 2
k=1
Examples
∞ (2)
X H k ζ 2 (2) + ζ(4) 7
= = ζ(4); (2.98)
k2 2 4
k=1
∞ (3)
X H k ζ 2 (3) + ζ(6)
= ; (2.99)
k3 2
k=1
∞ (4)
X Hk ζ 2 (4) + ζ(8) 13
= = ζ(8), (2.100)
k4 2 12
k=1
where we used ζ 2 (2) = 25 ζ(4) and ζ 2 (4) = 76 ζ(8) given in (1.62) and (1.65).
2.5.3 Second Application
For integer p ≥ 2, the following identity holds:

2
∞ (p) ∞
X Hk X H
(p)
ζ 3 (p) − ζ(3p)
k
− = (2.101)
kp k 2p 3
k=1 k=1
2
1 (p)
Proof. Let ak = kp and bk = Hk in (2.95),
2
n (p) n
! n−1 k
!
Hk 1 2 1 2 2
(p) (p)
X X X X
= Hn(p) − Hk+1 − Hk
kp i=1
ip
i=1
i p
k=1 k=0

(p) (p) 1
use Hk+1 = Hk + given in (1.145)
(k + 1)p
n−1 (p)
!
2 2Hk 1
(p)
X
= Hn(p) Hn(p) − Hk p
+
(k + 1) (k + 1)2p
k=0
(p)
n
!
3 X
(p) 2H k−1 1
= Hn(p) − Hk−1 + 2p
kp k
k=1
n (p)
!
2Hk − k2p
3 X

(p) (p) 1 1
= Hn − Hk − p + 2p
k kp k
k=1
2
n (p) n
3 X Hk X Hk
(p)
= Hn(p) − 2 + 3 − Hn(3p) .
kp k 2p
k=1 k=1
Rearranging the terms,

2
n (p) n
X Hk X H
(p) 3
3 −3 k
= Hn(p) − Hn(3p) .
kp k 2p
k=1 k=1
Take the limit on both sides letting n → ∞,

2
∞ (p) ∞
X Hk X H
(p) 3
3 −3 k
= lim Hn(p) − Hn(3p) ,
kp k 2p n→∞
k=1 k=1
and the proof completes on using lim Hn(3p) = ζ(3p).

n→∞
Another approach may be found in [28, p. 359].
Examples
2
∞ (2) ∞
X Hk X H
(2)
ζ 3 (2) − ζ(6) 9
k
− = = ζ(6); (2.102)
k2 k4 3 8
k=1 k=1
2
∞ (3) ∞
X Hk X Hk
(3)
ζ 3 (3) − ζ(9)
3
− 6
= ; (2.103)
k k 3
k=1 k=1
2
∞ (4) ∞
X Hk X Hk
(4)
ζ 3 (4) − ζ(12) 493
4
− 8
= = ζ(12), (2.104)
k k 3 5528
k=1 k=1
35 7007
where we used ζ 3 (2) = 8 ζ(6) and ζ 3 (4) = 5528 ζ(12) given in (1.63) and (1.70)
respectively.
2.5.4 Third Application
∞ (q) ∞ (p)
X Hk −p −p q−1
XH
k
= (1 − 2 )ζ(q)ζ(p) + (2 − 2 )
(2k + 1)p kq
k=1 k=1
∞ (p)
X (−1)k Hk
−2q−1 . (2.105)
n=1
kq
Proof. We follow the same technique as in [31]:

1 (q)
Let ak = (2k−1)p and bk = Hk − ζ(q) in (2.95),
n (q)
X H − ζ(q)k
(2k − 1)p
k=1
n n−1 k
!
X 1 X X 1
(q) (q)

= (Hn(q) − ζ(q)) p
− Hk+1 − Hk .
i=1
(2i − 1) i=1
(2i − 1)p
k=0
Let n → ∞ and write lim Hn(q) = ζ(q),

n→∞
∞ ∞
(q) k
!
X H − ζ(q)
k
X X 1
(q) (q)

=0− Hk+1 − Hk
(2k − 1)p i=1
(2i − 1)p
k=1 k=0
( k
)
X 1 (p) −p (p)
use = H2k − 2 Hk given in (1.148)
i=1
(2i − 1)p

(q) (q) 1
and Hk+1 − Hk =
(k + 1)q
∞ (p) ∞ (p)
X Hk X H2k
= 2−p −
(k + 1)q (k + 1)q
k=0 k=0
{shift both indexes by −1}
∞ (p) ∞ (p)
X Hk−1 X H2k−2
= 2−p −
kq kq
k=1 k=1
∞ (p) 1 ∞ H (p) − 1
− 1
−p
X H − k kp
X 2k (2k)p (2k−1)p
=2 −
kq kq
k=1 k=1
∞ (p) ∞ (p) ∞
X H X H X 1
= 2−p k
− 2k
+
kq kq k q (2k − 1)p
k=1 k=1 k=1
{use (1.5) for the second sum}
∞ (p) ∞ (p) ∞ (p) ∞
X H X H X (−1)k H X 1
= 2−p k
− 2q−1 k
− 2q−1 k
+
nq kq kq k q (2k − 1)p
k=1 k=1 k=1 k=1
∞ (p) ∞ (p) ∞
X H X (−1)k H X 1
= 2−p − 2q−1 k
− 2q−1 k
+ . (2.106)
kq kq k q (2k − 1)p
k=1 k=1 k=1
On the other hand,

∞ (q) ∞ (q) ∞
X H − ζ(q)
k
X Hk X 1
= − ζ(q)
(2k − 1)p (2k − 1) p (2k − 1)p
k=1 k=1 k=1
{seperate the first term of the first sum}
∞ (q) ∞
X Hk X 1
=1+ p
− ζ(q)
(2k − 1) (2k − 1)p
k=2 k=1
{shift both indexes by +1}
∞ (q) ∞
X Hk+1 X 1
=1+ − ζ(q)
(2k + 1)p (2k + 1)p
k=1 k=0
∞ (q) 1 ∞
X Hk + (k+1) q X 1
=1+ p
− ζ(q)
(2k + 1) (2k + 1)p
k=1 k=0
∞ (q) ∞ ∞
X Hk X 1 X 1
=1+ + − ζ(q)
(2k + 1)p q
(k + 1) (2k + 1) p (2k + 1)p
k=1 k=1 k=0
{shift the index of the second sum by −1}
∞ (q) ∞ ∞
X Hk X 1 X 1
=1+ p
+ q p
− ζ(q)
(2k + 1) k (2k − 1) (2k + 1)p
k=1 k=2 k=0
∞ ∞
( )
X 1 X 1
use the fact that =1+
k q (2k − 1)p k q (2k − 1)p
k=1 k=2
{and recall the result of the latter sum from (1.85)}
∞ (q) ∞
X Hk X 1
= + − ζ(q)(1 − 2−p )ζ(p). (2.107)
(2k + 1)p k q (2k − 1)p
k=1 k=1
Combining (2.106) and (2.107) completes the proof.

2.6 Identities By Fourier Series
2.6.1 Fourier Series
Let f (x) be a function with a period of 2p and integrable on the interval [−p, p].
Then its Fourier series is given by
∞ ∞
X nπx X nπx
f (x) = a0 + an cos + bn sin , (2.108)
n=1
p n=1
p
where
Z p
1
a0 = f (x)dx,
2π −p
1 p 1 p
Z Z
nπx nπx
an = f (x) cos dx, bn = f (x) sin dx.
π −p p π −p p
py

Proof. Suppose f π is a 2π–periodic function and expand it in cosine and sine
series,
py ∞
X ∞
X
f = An cos(ny) + Bn sin(ny),
π n=0 n=0
where An and Bn are the coefficients of the two series. Separate the first term of
both series and use cos(0) = 1 and sin(0) = 0,
py ∞
X ∞
X
f = A0 + An cos(ny) + Bn sin(ny). (2.109)
π n=1 n=1
To find A0 , integrate both sides of (2.109) from y = −π to π,

Z π py Z π Z π ∞
X Z π ∞
X
f dy = A0 dy + An cos(ny)dy + Bn sin(ny)dy
−π π −π −π n=1 −π n=1

π ∞
X Z π X∞ Z π
= A0 y + An cos(ny)dy + An sin(ny)dy.
−π n=1 −π n=1 −π
Since
( Ra
Z a
2 0 f (x)dx if f (x) is even function (f (−x) = f (x)),
f (x)dx =
−a 0 if f (x) is odd function (f (−x) = −f (x))
(2.110)
2.6. Identities By Fourier Series 113
and cos(ny) is even an function and sin(ny) is an odd function, we have

Z π py ∞
X Z π
f dy = 2πA0 + 2 An cos(ny)dy
−π π n=1 0
∞ π
X sin(ny)
= 2πA0 + 2 An = 2πA0 + 0.
n=1
n 0
Divide both sides by 2π,

Z π
1 py
A0 = f dy. (2.111)
2π −π π
To find An , multiply both sides of (2.109) by cos(ny) then integrate from y = −π

to π,
Z π
py
f cos(ny)dy
−π π
Z π X∞ Z π X∞ Z π
= A0 cos(ny)dy + An cos2 (ny)dy + Bn sin(ny) cos(ny)dy
−π n=1 −π n=1 −π
{the last integral is 0, since the integrand is an odd function}

π ∞ π
sin(ny) X 2ny + sin(2ny)
= A0 + An
n −π n=1
4n −π
∞
2A0 sin(nπ) X sin(2nπ)
= + An +π
n n=1
2n
{write sin(nπ) = sin(2nπ) = 0, since n is an integer}
X∞
=0+π An .
n=1
Divide both sides by π,

Z π
1 py
An = f cos(ny)dy. (2.112)
π −π π
To find Bn , multiply both sides of (2.109) by sin(ny) then integrate from x = −π to

π,
Z π
py
f sin(ny)dy
−π π
Z π X∞ Z π X∞ Z π
= A0 sin(ny)dy + An cos(ny) sin(ny)dy + Bn sin2 (ny)dy
−π n=1 −π n=1 −π
{the first two integrals are 0, since their integrand is an odd function}
∞ π ∞ ∞
X 2ny − sin(2ny) X sin(2nπ) X
= Bn = Bn π − =π Bn ,
n=1
4n −π n=1
2n n=1
where we used sin(2nπ) = 0 for integer n. Divide both sides by π,
1 π py
Z
Bn = f sin(ny)dy. (2.113)
π −π π
Plugging the results from (2.111), (2.112), and (2.113) in (2.109) yields
py Z π ∞ Z π
1 py X 1 py
f = f dy + f cos(ny)dy cos(ny)
π 2π −π π n=1
π −π π
∞
1 π py
X Z
+ − f sin(ny)dy sin(ny). (2.114)
n=1
π −π π
py
Substitute π = x in (2.114),
Z p ∞ Z p
1 X 1 nπx nπx
f (x) = f (x)dx + f (x) cos dx cos
2p −p p −p p p
| {z } n=1 | {z }
ao an
∞ Z p
X 1 nπx nπx
+ f (x) sin dx sin
n=1 |
p −p p p
{z }
bn
∞ ∞
X nπx X nπx
= a0 + an cos + bn sin .
n=1
p n=1
p
f py has a period of 2π and so f py = f πp (y + 2π) =

Since we assumed π π
f py py

π + 2p , and since we substituted π = x, we have f (x) = f (x + 2p), which
indicates that f (x) has a period of 2p and the proof is finished.
2.6.2 Fourier Series of Even Function

Let f (x) be an even function with a period of 2p and integrable on the interval
[−p, p]. Then its Fourier series is given by
∞
X nπx
f (x) = a0 + an cos , (2.115)
n=1
p
(continued)
where
Z p Z p
1 2 nπx
a0 = f (x)dx, an = f (x) cos dx.
π 0 π 0 p
Proof. Let’s recall the definitions of a0 , an , and bn in (2.108):

Z p
1
a0 = f (x)dx
2π −p
{the integrand is an even function}
1 p
Z
= f (x)dx.
π 0
Z p
1 nπx
an = f (x) cos dx
π −p p
2 p
Z
nπx
= f (x) cos dx.
π 0 p
Z p
1 nπx
bn = f (x) sin dx
π −p p
{the integrand is an odd function}
= 0.
The proof follows on plugging a0 , an , and bn in (2.108).
2.6.3 Fourier Series of Odd Function
Let f (x) be an odd function with a period of 2p and integrable on the interval
[−p, p]. Then its Fourier series is given by
∞
X nπx
f (x) = bn sin , (2.116)
n=1
p
where Z p
2 nπx
bn = f (x) sin dx.
π 0 p
Proof. We follow the previous approach:

Z p
1
a0 = f (x)dx
2π −p
= 0.
Z p
1 nπx
an = f (x) cos dx
π −p p
= 0.
Z p
1 nπx
bn = f (x) sin dx
π −p p
2 p
Z
nπx
= f (x) sin dx,
π 0 p
and the proof completes on plugging a0 , an , and bn in (2.108).
2.6.4 Fourier Series of cos(zx)
The following identity holds:

∞
" #
2z sin(πz) 1 X (−1)n cos(nx)
cos(zx) = − , z∈
/ Z. (2.117)
π 2z 2 n=1 n2 − z 2
Proof. Since cos(zx) is an even function, we recall (2.115)

∞
X nπx
f (x) = a0 + an cos ,
n=1
p
where Z p Z p
1 2 nπx
a0 = f (x)dx, an = f (x) cos dx.
π 0 π 0 p
Since cos(zx) = cos(zx + 2π), which indicates that the period of the function is 2π
and so p = π, its Fourier expansion is given by
∞
X
cos(zx) = a0 + an cos(nx). (2.118)
n=1
Let’s find a0 and an :

Z π
1 sin(πz)
a0 = cos(zx)dx = .
π 0 πz
2 π
Z
an = cos(zx) cos(nx)dx
π 0
{make use of 2 cos(x) cos(y) = cos(x − y) + cos(x + y)}
1 π
Z
= [cos((z − n)x) + cos((z + n)x)]dx
π 0
π
1 sin((z − n)x) sin((z + n)x)
= +
π z−n z+n
0
1 sin((z − n)π) sin((z + n)π)
= +
π z−n z+n
{use sin(x ± y) = sin(x)cos(y) ± cos(x) sin(y)}

2 n cos(πz) sin(πn) − z sin(πz) cos(πn)
=
π n2 − z 2
{write cos(nπ) = (−1)n and sin(nπ) = 0, since n is an integer}
2(−1)n z sin(πz)
=− .
π(n2 − z 2 )
Substituting the results of a0 and an in (2.118) completes the proof.
2.6.5 Fourier Series of sin(zx)
The following equality holds:

∞
2 sin(πz) X (−1)n n sin(nx)
sin(zx) = − , z∈
/ Z. (2.119)
π n=1
n2 − z 2
Proof. Since sin(zx) is an odd function, we recollect (2.116)

∞ Z p
X nπx 2 nπx
f (x) = bn sin , bn = f (x) sin dx.
n=1
p π 0 p
Since sin(zx) = sin(zx + 2π), which indicates that the period of the function is 2π
and so p = π, its Fourier expansion is given by
∞
X
sin(zx) = bn sin(nx). (2.120)
n=1
Let’s find bn :
2 π
Z
bn = sin(zx) sin(nx)dx
π 0
{make use of 2 sin(x) sin(y) = cos(x − y) − cos(x + y)}
1 π
Z
= [cos((z − n)x) − cos((z + n)x)]dx
π 0
π
1 sin((z − n)x) sin((z + n)x)
= −
π z−n z+n
0
1 sin((z − n)π) sin((z + n)π)
= −
π z−n z+n
{use sin(x ± y) = sin(x)cos(y) ± cos(x) sin(y)}

2 z cos(πz) sin(πn) − n sin(πz) cos(πn)
=
π n2 − z 2
{write cos(nπ) = (−1)n and sin(nπ) = 0 for integer n}
2(−1)n n sin(πz)
=− .
π(n2 − z 2 )
Plugging bn in (2.120) yields the proof.
2.6.6 Fourier Series of ln(sin x)
For 0 < x < π, we have

∞
X cos(2nx)
ln(sin x) = − ln(2) − . (2.121)
n=1
n
Proof (i). Since ln | sin x| is an even function and has a period of π as ln | sin x| =
ln | sin(x + π)| and so p = π/2. Thus, based on (2.115), its Fourier expansion is
given by
X∞
ln | sin x| = a0 + an cos(2nx), (2.122)
n=1
where
Z π Z π
2 2 4 2
a0 = ln | sin x|dx, an = ln | sin x| cos(2nx)dx.
π 0 π 0
We have
Z π
2 2
a0 = ln | sin x|dx
π 0
{note that ln | sin x| = ln(sin x) for 0 < x < π}

Z π2
2
= ln(sin x)dx
π 0
{this integral is given in (3.107)}
2 π
= − ln(2) = − ln(2)
π 2
and
Z π Z π2
4 2 4
an = ln | sin x| cos(2nx)dx = ln(sin x) cos(2nx)dx
π 0 π 0
π Z π2
IBP 4 2
1
= sin(2nx) ln(sin x) − sin(2nx) cot(x)dx
2πn 0 2n 0
| {z }
0
{the latter integral is given in (3.106)}
1 π π
=− =− .
2n 2 4n
Plug a0 and an in (2.122),
∞
X cos(2nx)
ln | sin x| = − ln(2) − .
n=1
n
The proof follows on noticing that ln | sin x| = ln(sin x) for 0 < x < π.
Proof (ii). By considering the real parts of Euler’s formula in (1.16), we have
cos(x) = Reix .
Therefore,
∞ ∞ ∞
X cos(2nx) X e2inx X (e2ix )n
=R =R
n=1
n n=1
n n=1
n
2ix

= −R ln 1 − e
{write e2ix = cos(2x) + i sin(2x)}
= −R ln (1 − cos(2x) − i sin(2x))

1 y
use ln(x + iy) = ln(x2 + y 2 ) + i arctan , x > 0 given in (1.15)
2 x
{since 1 − cos(2x) > 0 for 0 < x < π}

1 − sin(2x)
= −R ln[(1 − cos(2x))2 + sin2 (2x)] + i arctan
2 1 − cos(2x)
{use 1 − cos(2x) = 2 sin2 x and sin(2x) = 2 sin x cos x}

1 −2 sin x cos x
= −R ln[4 sin4 x + 4 sin2 x cos2 x] + i arctan
2 2 sin2 x

1
= −R ln[4 sin2 x(sin2 x + cos2 x)] + i arctan (− cot x)
2
n π π o
note that arctan(− cot x) = arctan − tan −x =− −x
2 2
1 π
= −R ln(2 sin x)2 − i −x
2 2
n π o
= R − ln(2 sin x) + i −x (2.123)
2
= − ln(2 sin x) = − ln(2) − ln(sin x),
and the proof is complete.
Further, by considering the imaginary parts in (2.123), we see that

∞ ∞
π X e2inx X sin(2nx)
−x=J = .
2 n=1
n n=1
n
Replace x by x/2
∞
X sin(nx) π x
= − . (2.124)
n=1
n 2 2
The latter identity is very useful. To show that, integrate both sides,
∞
X cos(nx) π x2
− = x − + C.
n=1
n2 2 4
To find the constant, C, set x = 0,

∞
X 1 π2
C=− = −ζ(2) = − .
n=1
n2 6
Therefore,
∞
X cos(nx) x2 π π2
= − x + , 0 ≤ x ≤ 2π. (2.125)
n=1
n2 4 2 6
Integrating the latter identity from x = 0 to x gives
∞
X sin(nx) x3 π 2 π2
= − x + x, 0 ≤ x ≤ 2π, (2.126)
n=1
n3 12 4 6
and so on.
2.6.7 Fourier Series of ln(cos x)
π
For |x| < 2, we have
∞
X (−1)n cos(2nx)
ln(cos x) = − ln(2) − . (2.127)
n=1
n
π
Proof (i). Let x = 2 − y in (2.121),
∞
π X cos(πn − 2ny)
ln sin −y = − ln(2) − .
2 n=1
n
Use sin( π2 − y) = cos y and cos(a − b) = cos(a) cos(b) + sin(a) sin(b),

∞
X cos(πn) cos(2ny) + sin(πn) sin(2ny)
ln(cos y) = − ln(2) − .
n=1
n
The proof follows on using cos(πn) = (−1)n and sin(πn) = 0 for integer n.
Proof (ii).
∞ ∞ ∞
X (−1)n cos(2nx) X (−1)n e2inx X (−e2ix )n
=R =R
n=1
n n=1
n n=1
n
= −R ln 1 + e2ix = −R ln (1 + cos(2x) + i sin(2x))

1 y
use ln(x + iy) = ln(x2 + y 2 ) + i arctan , x > 0 given in (1.15)
2 x
{since 1 + cos(2x) > 0 for |x| < π/2}

1 sin(2x)
= −R ln[(1 + cos(2x))2 + sin2 (2x)] + i arctan
2 1 + cos(2x)
{use 1 + cos(2x) = 2 cos2 x and sin(2x) = 2 sin x cos x}

1 2 sin x cos x
= −R ln[4 cos4 x + 4 sin2 x cos2 x] + i arctan
2 2 cos2 x

1
= −R ln[4 cos2 x(cos2 x + sin2 x)] + i arctan (tan x)
2

1
= −R ln(2 cos x)2 + ix
2
= R {− ln(2 cos x) − ix} (2.128)
= − ln(2 cos x) = − ln(2) − ln(cos x).
2.6.8 Fourier Series of ln(tan x)

π
For 0 < x < 2, we have
∞
X cos((4n + 2)x)
ln(tan x) = −2 . (2.129)
n=0
2n + 1
cos(2nx)
Proof. Set an = n in (1.7):
∞
X ∞
X ∞
X
2 a2n+1 = an − (−1)n an ,
n=0 n=1 n=1
we have
∞ ∞ ∞
X cos(4n + 2)x) X cos(2nx) X (−1)n cos(2nx)
2 = −
n=0
2n + 1 n=1
n n=1
n
{collect the results from (2.121) and (2.127)}
= − ln(2) − ln(sin x) − (− ln(2) − ln(cos x))
= − ln(sin x) + ln(cos x)

sin x
= − ln = − ln(tan x).
cos x
π
2.6.9 Series Representation of sin(πz)

∞
π 1 X 2z(−1)n
= − , z∈
/ Z. (2.130)
sin(πz) z n=1 n2 − z 2
Proof. Set x = 0 in (2.117).
2.6.10 Series Representation of cot(πz)

∞
X 1 1 π
2 − z2
= 2− cot(πz), z∈
/ Z. (2.131)
n=1
n 2z 2z
Proof. Set x = π in (2.117) and use cos(πn) = (−1)n for integer n.

Moreover, if we replace z by iz in (2.131) and use cot(iz) = −i coth(z), we obtain

∞
X 1 1 π
2 + z2
=− 2 + coth(πz), z 6= Z. (2.132)
n=1
n 2z 2z
2.6.11 Euler’s Product Formula of sin(πz)

∞ 2 −1
πz Y n − z2
= , z∈
/ Z. (2.133)
sin(πz) n=1 n2

∞
X 1 1 π
= 2− cot(πz)
n=1
n2 −z 2 2z 2z
by 2z then integrate from z = 0 to x,

∞ Z x Z x
X 2z 1
dz = − π cot(πz) dz.
n=1 0 n − z2
2
0 z
The LHS: Since

x −1
x x
n2 2z n2 − z 2 n2 − x2
Z Z
2z
dz = dz = − ln = ln ,
0 n2 − z 2 0 n − z 2 n2
2 n2 n2
0
we have
∞ Z x ∞ 2 −1 ∞ 2 −1
X 2z X n − x2 Y n − x2
dz = ln = ln ,
n=1 0 n2 − z 2 n=1
n2 n=1
n2
where the last step follows from using (1.12).

The RHS:
Z x x
1
− π cot(πz) dz = ln(z) − ln sin(πz)
0 z 0
x
z x z
= ln = ln − lim ln
sin(πz) sin(πx) z→0 sin(πz)
0

z 1
use lim =
z→0 sin(πz) π

x πx
= ln + ln(π) = ln .
sin(πx) sin(πx)
Equating the two sides,

∞ 2 −1
n − x2

Y πx
ln = ln .
n=1
n2 sin(πx)
Exponentiating both sides with base e then replacing x by z completes the proof.
If we replace z by iz in (2.133) and use sin(iz) = −i sinh(z), we obtain
∞ 2 −1
πz Y n + z2
=− , z ∈ C. (2.134)
sinh(πz) n=1
n2
π

2.6.12 Series Representation of sec 2
z

∞
π 4 X n in
sec z = ·J , z∈
/ Z. (2.135)
2 π n=1
n2 − z 2
Proof. Set x = π/2 in (2.119), we obtain

∞
sin π2 z 2 X (−1)n n sin π2 n

=− .
sin(πz) π n=1 n2 − z 2
The LHS:
sin π2 z sin π2 z

1 1 π
= = = sec z .
2 sin π2 z cos π π

sin(πz) 2z 2 cos 2 z 2 2
The RHS: Since

∞
X π
an sin n = a1 (1) + a2 (0) + a3 (−1) + a4 (0) + · · ·
n=1
2
= a1 − a3 + a5 − · · ·
∞
X X∞
= (−1)n a2n+1 = J in an ,
n=0 n=1
where the last result follows from (1.11), we have

∞ ∞ ∞
(−1)n n sin π2 n

X X n (−i)n X n in
= J = −J .
n=1
n2 − z 2 n=1
n2 − z 2 n=1
n2 − z 2
In the last step , we used
J(x − iy) = −J(x + iy).
The proof finalizes on equating the two sides.

If we replace z by iz then use sec(iz) = sech(z), we conclude
∞
π 4 X n in
sech z = ·J , z ∈ Z. (2.136)
2 π n=1
n + z2
2
2.6.13 Series Representation of sin(x)

π
For |x| < 2, the following equality holds:
∞
8 X (−1)n n sin(2nx)
sin(x) = − . (2.137)
π n=1 4n2 − 1
Proof. The proof follows from (2.119) on setting z = 1/2 then replacing x by 2x.
2.6.14 Series Representation of tan x ln(sin x)
For 0 < x < π, we have

∞ Z 1
X 1 − t n−1
tan x ln(sin x) = − t dt sin(2nx). (2.138)
n=1 0 1+t

Proof. By the definition of tanh(x), we have
∞ ∞ 1
1 − e−2y −2ny 1 − t n−1
Z Z Z
e−2y =t 1
tanh(y)e−2ny dy = e dy = t dt.
0 0 1 + e−2y 2 0 1+t
Multiply both sides by 2 sin(2x) then take the summation over n ≥ 1,

∞ Z 1
X 1 − t n−1
t dt sin(2nx)
n=1 0 1+t
∞
Z ∞ !
X
−2ny
=2 tanh(y) sin(2nx)e dy
0 n=1
∞
!
Z ∞ X
=2 tanh(y) J e2inx e−2ny dy
0 n=1
∞
!
Z ∞
2ix−2y n
X
=2 tanh(y) J e dy
0 n=1
∞
e2ix−2y
Z
=2 tanh(y) J dy
0 1 − e2ix−2y
Z ∞
1 sin(2x)
=2 tanh(y) dy
0 2 cosh(2y) − cos(2x)
Z 1
sinh(y)
= sin(2x) 2 dx
0 cosh(y)(2 cosh (y) − 1 − cos(2x))
{set cosh(y) = 1/t}
Z 1
t
= sin(2x) 2
dt
0 2 − (1 + cos(2x))t
{use 1 + cos(2x) = 2 cos2 (x)}
Z 1
1 t
= sin(2x) 2 (x)t2
dt
2 0 1 − cos
t=1
ln(1 − cos2 (x)t2 )

1
= sin(2x) −
2 2 cos2 x t=0
1 ln(sin x)
= − · 2 sin x cos x = − tan x ln(sin x).
2 cos2 x
Note that the value x = π/2 is included in the domain of tan x ln(sin x), since
lim tan x ln(sin x) = 0 by L’Hopital’s rule.
x→0
Further, by letting x → π2 −x in (2.138) and using sin( π2 −x) = cos(x), tan( π2 −x) =
cot(x), and sin(2n( π2 − x)) = −(−1)n sin(2nx) for integer n, we get
∞ Z 1
X 1 − t n−1 π
(−1)n t dt sin(2nx) = cot x ln(cos x), |x| < . (2.139)
n=1 0 1+t 2
Also note the fact lim cot x ln(cos x) = 0 justifies why the value x = 0 is included
x→0
in the domain of the function.
2.6.15 Series Representation of ln2 (2 cos x)
π
For |x| < 2, we have
∞
X Hn−1
ln2 (2 cos x) = x2 + 2 (−1)n cos(2nx). (2.140)
n=1
n
Proof. Since cos x = Reix , we have

∞ ∞
X Hn−1 X Hn−1 2inx
(−1)n cos(2nx) = R (−1)n e
n=1
n n=1
n
∞
X Hn−1
=R (−e2ix )n
n=1
n
{replace x by −e2ix in the generating function in (2.6)}
1
= R ln2 (1 + e2ix )
2
{recall the result of ln(1 + e2ix ) from (2.128)}
1 2
= R [ln(2 cos x) + ix]
2
1
= R ln2 (2 cos x) + 2ix ln(2 cos x) − x2

2
1 x2
= ln2 (2 cos x) − ,
2 2
and the proof is completed.
If we replace x by π2 −x in (2.140) using cos( π2 −x) = sin(x) and (cos(2n( π2 −x))) =
(−1)n cos(2nx) for integer n, we have
π 2 ∞
X Hn−1
ln2 (2 sin x) = −x +2 cos(2nx), 0 < x < π. (2.141)
2 n=1
n
Chapter 3
Logarithmic Integrals
3.1 Generalized Logarithmic Integrals

R1 lna (x)
3.1.1 0 1−x
dx
For a ∈ Z+ , the following identity holds:

1
lna (x)
Z
dx = (−1)a a!ζ(a + 1). (3.1)
0 1−x
1
Proof. Expand 1−x in series then interchange integration and summation,
∞ Z 1
1
lna (x)
Z X
dx = xn−1 lna (x)dx
0 1−x n=1 0
{make use of (1.31)}
∞
a
X 1
= (−1) a! a+1
= (−1)a a!ζ(a + 1).
n=1
n
Examples
Z 1
ln(x)
dx = −ζ(2); (3.2)
0 1−x
1
ln2 (x)
Z
dx = 2ζ(3); (3.3)
0 1−x
1
ln3 (x)
Z
dx = −6ζ(4); (3.4)
0 1−x
128
3.1. Generalized Logarithmic Integrals 129
1
ln4 (x)
Z
dx = 24ζ(5); (3.5)
0 1−x
1
ln5 (x)
Z
dx = −120ζ(6). (3.6)
0 1−x
R1 lna (x)
3.1.2 0 1+x
dx

1
lna (x)
Z
dx = (−1)a a!(1 − 2−a )ζ(a + 1). (3.7)
0 1+x
Proof. Multiply both sides of 1

1+x = 1
1−x − 2x
1−x2 by lna (x) then integrate,
1
lna (x) 1
lna (x) 1
2x lna (x)
Z Z Z
dx = dx − dx
0 1+x 0 1−x 1 − x2
|0 {z }
x2 →x
Z 1 a Z 1 a
ln (x) ln (x)
= dx − 2−a dx
0 1 − x 0 1−x
Z 1 a
−a ln (x)
= (1 − 2 ) dx
0 1−x
= (−1)a a!(1 − 2−a )ζ(a + 1),
Examples
Z 1
ln(x) 1
dx = − ζ(2); (3.8)
0 1+x 2
1
ln2 (x)
Z
3
dx = ζ(3); (3.9)
0 1+x 2
1
ln3 (x)
Z
21
dx = − ζ(4); (3.10)
0 1+x 4
1
ln4 (x)
Z
45
dx = ζ(5); (3.11)
0 1+x 2
1
ln5 (x)
Z
465
dx = − ζ(6). (3.12)
0 1+x 4
130 Chapter 3. Logarithmic Integrals
R1 lna ( 1−x
1+x )
3.1.3 0 x
dx

Z 1 lna 1−x
1+x
dx = (−1)a a!(2 − 2−a )ζ(a + 1). (3.13)
0 x
1−x
Proof. By making the substitution 1+x = y,

Z 1 lna 1−x
1+x lna (y)
Z 1
dx = 2
2
dy (3.14)
x 0 1−y
0
1 1 y
write = −
1 − y2 1 − y 1 − y2
Z 1 a
y lna (y)
Z 1
ln (y)
=2 dy − 2 dy
0 1−y 1 − y2
|0 {z }
y 2 →y
a
1
lna (y) 1
Z Z
ln (y)
=2 dy − 2−a dy
0 1−y 0 1−y
Z 1 a
ln (y)
= (2 − 2−a ) dy
0 1−y
{make use of (3.1)}
= (2 − 2−a )(−1)a a!ζ(a + 1).
Examples

1−x
Z 1 ln 3
1+x
dx = − ζ(2); (3.15)
0 x
2

2
Z 1 ln 1−x
1+x 7
dx = ζ(3); (3.16)
0 x 2

Z 1 ln3 1−x
1+x 45
dx = − ζ(4); (3.17)
0 x 4

Z 1 ln4 1−x
1+x 93
dx = ζ(5); (3.18)
0 x 2

Z 1 ln5 1−x
1+x 945
dx = − ζ(6). (3.19)
0 x 4
R1 ln( 1−x
1+x )
lna−1 (x)
3.1.4 0 x
dx

Z 1 ln 1−x lna−1 (x)
1+x
dx = (−1)a (a − 1)(2 − 2−a )ζ(a + 1). (3.20)
0 x
Proof. Start with applying integration by parts,

Z 1 ln 1−x lna−1 (x) 1
lna (x) 2 1 lna (x)

1−x
Z
1+x
dx = ln + dx
0 x a 1 + x 0 a 0 1 − x2
| {z }
0
Z 1 a
2 ln (x)
= dx.
a 0 1 − x2
The latter integral appeared in (3.14). Collecting its result gives the proof.
Examples

1−x
Z 1 ln 1+x 3
dx = − ζ(2); (3.21)
0 x 2

1−x
Z 1 ln 1+x ln(x) 7
dx = ζ(3); (3.22)
0 x 4

Z 1 ln 1−x 2
1+x ln (x) 15
dx = − ζ(4); (3.23)
0 x 4

Z 1 ln 1−x
1+x ln3 (x) 93
dx = ζ(5); (3.24)
0 x 8

Z 1 ln 1−x ln4 (x)
1+x 189
dx = − ζ(6). (3.25)
0 x 4
R1 lna (1−x)
3.1.5 0 1+x
dx

1
lna (1 − x)
Z
1
dx = (−1)a a! Lia+1 . (3.26)
0 1+x 2
Proof. We begin with the change of variable 1 − x = t,
lna (1 − x)
Z 1 Z 1 a
ln (t)
dx = dt
0 1+x 0 2−t
∞ n−1
( )
1 X t
expand in series as then swab integration and summation
2−t n=1
2n
∞ Z 1
X 1
= n
tn−1 lna (t)dt
n=1
2 0

∞
X 1 1
= (−1)a a! n na+1
= (−1)a a! Lia+1 .
n=1
2 2
Examples
1
ln(1 − x)
Z
1 1 1
dx = − Li2 = − ζ(2) + ln2 (2); (3.27)
0 1+x 2 2 2
1 2
ln (1 − x)
Z
1 7 1
dx = 2 Li3 = ζ(3) − ln(2)ζ(2) + ln3 (2); (3.28)
0 1+x 2 4 3
Z 1 3
ln (1 − x) 1
dx = −6 Li4 ; (3.29)
0 1 + x 2
Z 1 4
ln (1 − x) 1
dx = 24 Li5 . (3.30)
0 1+x 2
In (3.27) and (3.28), we used:

1 1 1
Li2 = ζ(2) − ln2 (2),
2 2 2

1 7 1 1
Li3 = ζ(3) − ln(2)ζ(2) + ln3 (2),
2 8 2 6
which are given in (1.120) and (1.132) respectively.
R1 lna (x)
3.1.6 0
2
1−x
dx

1 a
lna (x)
Z
2 X
k a a−k 1
dx = (−1) k! (− ln(2)) Lik+1 . (3.31)
0 1−x k 2
k=0

Proof. Substituting x = y2 ,
1
lna (x) 1
[ln(y) − ln(2)]a
Z 2
Z
dx = dy
0 1−x 0 2−y
( a
)
a
X a a−k k
use the binomial theorem (x − y) = (−y) x
k
k=0
a Z 1 k !
X a a−k ln (y)
= (− ln(2)) dy
k=0
k 0 2−y

1
expand in series then swab integration and summation
2−y
∞
n
!
1 1 i−1 k
Z
X a a−k
X
= (− ln(2)) y ln (y)dy
k i=1
2i 0
k=0
{make use of (1.31) for the integral}
∞
a !
X a a−k k
X 1
= (− ln(2)) (−1) k!
k i=1
2i ik+1
k=0
a
X a 1
= (− ln(2))a−k (−1)k k! Lik+1 .
k 2
k=0
Examples
Z 1
ln(x)
2 1 1
dx = − ln(2) Li1 − Li2
0 1 − x 2 2
1 1
= − ζ(2) − ln2 (2); (3.32)
2 2
Z 12 2
ln (x) 1 1 1
dx = ln2 (2) Li1 + 2 ln(2) Li2 + 2 Li3
0 1 − x 2 2 2
7 1
= ζ(3) + ln3 (2); (3.33)
4 3
1
ln3 (x)
Z
2
3 1 2 1 1
dx = − ln (2) Li1 − 3 ln (2) Li2 − 6 ln(2) Li3
0 1−x 2 2 2

1
−6 Li4
2

3 2 21 1 4 1
= ln (2)ζ(2) − ln(2)ζ(3) − ln (2) − 6 Li4 ; (3.34)
2 4 2 2
1
ln4 (x)
Z
2 1 1 1
dx = ln4 (2) Li1 + 4 ln3 (2) Li2 + 12 ln2 (2) Li3
0 1−x 2 2 2

1 1
+24 ln(2) Li4 + 24 Li5
2 2

21 2 3 5 1
= ln (2)ζ(3) − 4 ln (2)ζ(2) + ln (2) + 24 ln(2) Li4
2 2

1
+24 Li5 . (3.35)
2
In the calculations above, we used the value Li1 12 = ln(2), which follows from

(1.99) on setting z = 1/2.
R1 lna (1+x)
3.1.7 0 x
dx

a
1
lna (1 + x) lna+1 (2)
Z
X a 1
dx = +a!ζ(a+1)− k! lna−k (2) Lik+1 .
0 x a+1 k 2
k=0
(3.36)

1
Proof. Putting 1+x = y,
lna (1 + x)
1
lna (y)
Z Z 1
a
dx = (−1) dy
x 1 y(1 − y)
0 2
Z 1 a Z 1 a
ln (y) ln (y)
= (−1)a dy + (−1)a dy
1
2
y 1
2
1−y
| {z }
R1 R 1/2
0
− 0
Z 21 a
lna+1 (2)
Z 1 a
ln (y) ln (y)
= (−1)a (−1)a + (−1)a dy − (−1)a dy
a+1 0 1 − y 0 1−y
{recall the result from (3.1) for the first integral}
Z 12 a
lna+1 (2) a ln (y)
= + a!ζ(a + 1) − (−1) dy,
a+1 0 1−y
and the proof follows on substituting the result from (3.31).

Examples
Z 1
ln(1 + x) 1
dx = ζ(2); (3.37)
0 x 2
1
ln2 (1 + x)
Z
1
dx = ζ(3); (3.38)
0 x 4
1
ln3 (1 + x)
Z
21 3 1
dx = 6ζ(4) − ln(2)ζ(3) + ln2 (2)ζ(2) − ln4 (2)
0 x 4 2 4

1
−6 Li4 ; (3.39)
2
1
ln4 (1 + x)
Z
21 2 4
dx = 24ζ(5) − ln (2)ζ(3) + 4 ln3 (2)ζ(2) − ln5 (2)
0 x 2 5

1 1
−24 ln(2) Li4 − 24 Li5 . (3.40)
2 2
x
R1 ln2a−1 ( 1−x )
3.1.8 0 1+x
dx

Z 1 2a−1 x " a
#
ln ( 1−x ) ln2a (2) X ln2a−2n (2)
dx = (2a−1)! 2η(2a) − −2 η(2n)
0 1+x (2a)! n=1
(2a − 2n)!
(3.41)
x
Proof. Start with the change of variable 1−x = y,
1 ln2a−1 ( 1−x
x
) ∞
ln2a−1 (y)
Z Z
dx = dy
0 1+x 0 (1 + y)(1 + 2y)
∞
1
ln2a−1 (y) ln2a−1 (y)
Z Z
= dy + dy
0 (1 + y)(1 + 2y) (1 + y)(1 + 2y)
|1 {z }
y→1/y
2a−1
1
ln2a−1 (y)
1
Z Z
ln (y)
= dy − dy
0 (1 + y)(1 + 2y) 0 (1 + y)(2 + y)
{decompose both integrands}
1 2a−1 Z 1 1 2a−1 Z 1 2a−1
2 ln (y)
Z
2 ln (y) ln (y)
= dy + 1 dy − 2 dy
0 1 + 2y 0 1 + 2 y 0 1 + y
{make use of (1.111) for the first two integrals}
{and make use of (1.79) for the last integral}

1
= (2a − 1)! Li2a (−2) + Li2a − + 2(2a − 1)!η(2a).
2
Set z = 2 in (1.134) then write Li2n (−1) = −η(2n) to complete the proof.
Examples
1 ln3 ( 1−x
x
)
Z
1 4
dx = −3 ln2 (2)ζ(2) − ln (2); (3.42)
0 1+x 4
1 ln5 ( 1−x
x
)
Z
1 6
dx = −105 ln2 (2)ζ(4) − 5 ln4 (2)ζ(2) − ln (2); (3.43)
0 1+x 6
1 ln7 ( 1−x
x
) ln8 (2)
Z
9765 2 735 4
dx = − ln (2)ζ(6) − ln (2)ζ(4) − 7 ln6 (2)ζ(2) − .
0 1+x 2 2 8
(3.44)
R∞ lna (1+x)
3.1.9 0 1+x2
dx

Z ∞ a
ln (1 + x)
dx = (−1)a a! J{Lia+1 (1 + i)}. (3.45)
0 1 + x2

Proof. Let x = 1/y,

∞ Z ∞ lna y
lna (1 + x)
Z
1+y
dx = dy
0 1 + x2 0 1 + y2
lna (x)
y Z 1
1+y =x
= 2 2
dx
0 x + (1 − x)

1 1+i
write 2 = J
x + (1 − x)2 1 − (1 + i)x
Z 1 a
(1 + i) ln (x)
=J dx.
0 1 − (1 + i)x
The proof follows on replacing z by 1 + i in (1.111).

Examples
Z ∞
ln(1 + x)
dx = − J{Li2 (1 + i)}; (3.46)
0 1 + x2
Z ∞
ln2 (1 + x)
dx = 2 J{Li3 (1 + i)}; (3.47)
0 1 + x2
∞
ln3 (1 + x)
Z
dx = −6 J{Li4 (1 + i)}. (3.48)
0 1 + x2
R1 lna (1−x)
3.1.10 0 1+x2
dx

1
lna (1 − x)
Z
a 1+i
dx = (−1) a! J Lia+1 . (3.49)
0 1 + x2 2
Proof.
lna (1 − x)
1
lna (y)
Z Z 1
1−x=y
dx = dy
0 1 + x2 0 1 + (1 − y)
2

1 i
notice that = J
1 + (1 − y)2 1 + i − iy
Z 1 a
i ln (y)
=J dy
0 1 + i − iy
{ to get this inegral, set z = −i in (1.112)}

−i
= (−1)a a! J Lia+1
−i − 1

1+i
= (−1)a a! J Lia+1 .
2
Examples
1
ln(1 − x)
Z
1+i
dx = − J Li2 ; (3.50)
0 1 + x2 2
Z 1 2
ln (1 − x) 1+i
dx = 2 J Li3 ; (3.51)
0 1 + x2 2
Z 1 3
ln (1 − x) 1+i
2
dx = −6 J Li4 . (3.52)
0 1+x 2
R∞ ln2a (x)
3.1.11 0 1+x2
dx

∞
ln2a (x) d2a
Z
−2a−1
dx = 2 π lim csc(πs). (3.53)
0 1 + x2 s→ 12 ds
2a
Proof. Differentiate both sides of Euler’s reflection formula in (1.38):

Z ∞ s−1
π y
= dy
sin(πs) 0 1+y
2a times with respect to s,

Z ∞ s−1
d2a π d2a y
= 2a dy
ds2a sin(πs) ds 0 1 +y
{ use differentiation under the integral sign theorem given in (2.78)}
Z ∞ 2a s−1 Z ∞ 2a
∂ y ln (y)y s−1
= dy = dy.
0 ∂s2a 1 + y 0 1+y
Next, take the limit on both sides letting s → 1/2,

∞ √ ∞
d2a ln2a (y) ln2a (x)
Z Z
π y=x
lim1 2a
= √ dy = 22a+1 dx.
s→ 2 ds sin(πs) 0 y(1 + y) 0 1 + x2
Dividing both sides by 22a+1 completes the proof.
Examples
∞
ln2 (x) π3
Z
2
dx = ; (3.54)
0 1+x 8
∞
ln4 (x) 5π 5
Z
dx = ; (3.55)
0 1 + x2 32
∞
ln6 (x) 61π 7
Z
dx = ; (3.56)
0 1 + x2 128
∞
ln8 (x) 1385π 9
Z
dx = . (3.57)
0 1 + x2 512
Further, the case 2a + 1 leads to 0:

Z ∞
ln2a+1 (x)
dx = 0. (3.58)
0 1 + x2
To show that, let x → 1/x,

∞ ∞
ln2a+1 (x) ln2a+1 (x)
Z Z
dx = − dx.
0 1 + x2 0 1 + x2
R∞ ln2a+1 (x)
Adding 0 1+x2 dx to both sides then dividing by 2 yields (3.58).
R∞ Lia (−x)
3.1.12 0 1+x2
dx
For integer a ≥ 2, the following identity holds:

Z ∞
Lia (−x)
dx = (2−2a − 2−a−1 )πζ(a) − aβ(a + 1). (3.59)
0 1 + x2
Proof. Replace z by −x in (1.111),
(−1)a 1
x lna−1 (y)
Z
Lia (−x) = dy.
(a − 1)! 0 1 + xy
Divide both sides by 1 + x2 then integrate,

Z ∞ Z ∞
x lna−1 (y)
Z 1
(−1)a

Lia (−x) 1
dx = dy dx
0 1 + x2 0 1 + x2 (a − 1)! 0 1 + xy
{change the order of the integration}
Z 1
a
Z ∞
(−1) a−1 x
= ln (y) dx dy
(a − 1)! 0 0 (1 + x2 )(1 + xy)
{compute the inner integral by partial fraction decomposition}
Z 1
(−1)a

a−1 π y ln(y)
= ln (y) − dy
(a − 1)! 0 2 1 + y2 1 + y2
y lna−1 (y)
Z 1 Z 1 a
(−1)a π (−1)a ln (y)
= dy − dy
2(a − 1)! 0 1 + y2 (a − 1)! 0 1 + y 2
| {z }
√
y= x
(−1)a π 1
lna−1 (x) (−1)a 1
lna (y)
Z Z
= a+1 dx − dy.
2 (a − 1)! 0 1+x (a − 1)! 0 1 + y2
Gathering the results from (3.7) and (1.80), the proof is finalized.
Examples
∞
π3
Z
Li2 (−x)
dx = − − 2β(3); (3.60)
1 + x2 96
Z ∞0
Li3 (−x) 3π
2
dx = − ζ(3) − 3β(4); (3.61)
0 1 + x 64
Z ∞
Li4 (−x) 7π 5
2
dx = − − 4β(5); (3.62)
0 1+x 23040
Z ∞
Li5 (−x) 15π
2
dx = − ζ(5) − 5β(6). (3.63)
0 1 + x 1024
R1 Li2a+1 (−x)
3.1.13 0 1+x2
dx

Z 1
Li2a+1 (−x)
dx = (2−4a−3 − 2−2a−3 )πζ(2a + 1) − aβ(2a + 2)
0 1 + x2
X a
+ (1 − 21−2n )ζ(2n)β(2a − 2n + 2). (3.64)
n=1
Proof.
Z 1 Z ∞ Z ∞
Li2a+1 (−x) Li2a+1 (−x)
dx = − dx
0 1 + x2 0 1 1 + x2
Z ∞ Z ∞
Li2a+1 (−x) Li2a+1 (−x)
= 2
dx − dx
0 1 + x 1 1 + x2
| {z }
x7→1/x
Z ∞ Z 1
Li2a+1 (−x) Li2a+1 (−1/x)
= dx − dx
0 1 + x2 0 1 + x2
Z 1
Li2a+1 (−x)
0 1 + x2
1 ∞ Li2a+1 (−x) 1 1 Li2a+1 (−x) − Li2a+1 (−1/x)
Z Z
= dx + dx.
2 0 1 + x2 2 0 1 + x2
For the first integral, replace a by 2a + 1 in (3.59),

Z ∞
Li2a+1 (−x)
dx = (2−4a−2 −2−2a−2 )πζ(2a+1)−(2a+1)β(2a+2). (3.65)
0 1 + x2
For the second integral, divide both sides of (1.135):

a
ln2a+1 (x)

1 X Li2n (−1)
Li2a+1 (−x)−Li2a+1 − =− +2 ln2a−2n+1 (x)
x (2a + 1)! n=1
(2a − 2n + 1)!
by 1 + x2 then integrate from x = 0 to 1, we get

1
Li2a+1 (−x) − Li2a+1 (−1/x)
Z
dx
0 1 + x2
a
ln2a+1 (x)
Z 1 Z 1 2a−2n+1
1 X Li2n (−1) ln (x)
=− dx + 2 dx
(2a + 1)! 0 1 + x2 n=1
(2a − 2n + 1)! 0 1 + x 2
use (1.81) for the two integrals and Li2n (−1) = (21−2n − 1)ζ(2n)

a
X
= β(2a + 2) + 2 (1 − 21−2n )ζ(2n)β(2a − 2n + 2), (3.66)
n=1
and the proof follows on combining (3.65) and (3.66).

Examples
Z 1
Li3 (−x) 3π 1
2
dx = − ζ(3) − β(4) + ζ(2)β(2); (3.67)
0 1+x 128 2
Z 1
Li5 (−x) 15π 1 7
2
dx = − ζ(5) − 2β(6) + ζ(2)β(4) + ζ(4)β(2); (3.68)
0 1+x 2048 2 8
Z 1
Li7 (−x) 63π 1 7
2
dx = − ζ(7) − 3β(8) + ζ(2)β(6) + ζ(4)β(4)
0 1+x 32768 2 8
31
+ ζ(6)β(2); (3.69)
32
Z 1
Li9 (−x) 255π 1 7
2
dx = − ζ(9) − 4β(10) + ζ(2)β(8) + ζ(4)β(6)
0 1 + x 524288 2 8
31 127
+ ζ(6)β(4) + ζ(8)β(2). (3.70)
32 128
R1ln2a (x) ln(1+x)

3.1.14 0 1+x2
dx

1
ln2a (x) ln(1 + x)
Z
(2a)! ln(2)
dx = β(2a + 1) − a(2a)!β(2a + 2)
0 1 + x2 2
Xa
+(2a)! (1 − 21−2n )ζ(2n)β(2a − 2n + 2). (3.71)
n=1
Proof. Replace z by −x and a by 2a + 1 in (1.111),

1
−x ln2a (y)
Z
(2a)! Li2a+1 (−x) = dy.
0 1 + xy

1 1 1
−x ln2a (y)
Z Z Z
Li2a+1 (−x) 1
(2a)! dx = dy dx
0 1 + x2 0 1 + x2 0 1 + xy
1 Z 1
−x
Z
= ln2a (y) 2
dx dy
0 0 (1 + x )(1 + xy)
{ evaluate the inner integral by partial fraction decomposition}

Z 1
2a ln(1 + y) ln(2) 1 π y
= ln (y) − − dy
0 1 + y2 2 1 + y2 4 1 + y2
Z 1 2a
ln(2) 1 ln2a (y) π 1 y ln2a (y)
Z Z
ln (y) ln(1 + y)
= dy − dy − dy,
0 1 + y2 2 0 1+y
2 4 0 1 + y2
where
1
ln2a (y)
Z
dy = (2a)!β(2a + 1)
0 1 + y2
follows from (1.81) and
y ln2a (y) √y=x −2a−1 1 ln2a (x)

Z 1 Z
dy = 2 dx
0 1 + y2 0 1+x
= (2−2a−1 − 2−4a−1 )(2a)!ζ(2a + 1)
follows from (3.7).

Putting together these two integrals along with (3.64), the proof is complete.
Examples
1
ln2 (x) ln(1 + x)
Z
dx = ζ(2)β(2) + ln(2)β(3) − 2β(4); (3.72)
0 1 + x2
1
ln4 (x) ln(1 + x)
Z
dx = 21ζ(4)β(2) + 12ζ(2)β(4) + 12 ln(2)β(5) − 48β(6);
0 1 + x2
(3.73)
Z 1 6
ln (x) ln(1 + x) 1395
2
dx = ζ(6)β(2) + 630ζ(4)β(4) + 360ζ(2)β(6)
0 1+x 2
+360 ln(2)β(7) − 2160β(8). (3.74)
R∞ Lia (−x2 )
3.1.15 0 1+x2
dx
For integer a ≥ 2, the following identity holds:

Z ∞
Lia (−x2 )
dx = (1 − 2a−1 )πζ(a). (3.75)
0 1 + x2
Proof. Replace z by −x2 in (1.111),
(−1)a 1
x2 lna−1 (y)
Z
Lia (−x2 ) = dy,
(a − 1)! 0 1 + x2 y
from which, it follows that

Z ∞ Z ∞ Z 1 2 a−1
Lia (−x2 ) (−1)a

1 x ln (y)
dx = dy dx
0 1 + x2 0 1 + x2 (a − 1)! 0 1 + x2 y
{ change the order of integration}
1 Z ∞
(−1)a x2
Z
= lna−1 (y) dx dy
(a − 1)! 0 0 (1 + x2 )(1 + x2 y)
Z 1
(−1)a

π 1
= lna−1 (y) √ dy
(a − 1)! 0 2y+ y
√ Z 1 a−1
y=x (−1)a 2a−1 π ln (x)
= dx
(a − 1)! 0 1+x
{recall the result in (3.7)}
= (1 − 2a−1 )πζ(a).
R1 Li2a+1 (−x2 )
3.1.16 0 1+x2
dx

1
Li2a+1 (−x2 )
Z
dx = (2−1 − 22a−1 )πζ(2a + 1) + 22a β(2a + 2)
0 1 + x2
X a
+22a+1 2−2n (1 − 21−2n )ζ(2n)β(2a − 2n + 2). (3.76)
n=1
Proof.
1 Z ∞ Z ∞
Li2a+1 (−x2 ) Li2a+1 (−x2 )
Z
2
dx = − dx
0 1+x 0 1 1 + x2
Z ∞ Z ∞
Li2a+1 (−x2 ) Li2a+1 (−x2 )
= 2
dx − dx
0 1+x 1 1 + x2
| {z }
x→1/x
∞ 1
Li2a+1 − x12

Z 2 Z
Li2a+1 (−x )
= dx − dx
0 1 + x2 0 1 + x2
1
Li2a+1 (−x2 )
Z
0 1 + x2
1 ∞ Li2a+1 (−x2 ) 1 1 Li2a+1 (−x2 ) − Li2a+1 − x12
Z Z
= dx + dx,
2 0 1 + x2 2 0 1 + x2
where the first integral

Z ∞
Li2a+1 (−x2 )
dx = (1 − 22a )πζ(2a + 1) (3.77)
0 1 + x2
follows from (3.75) on replacing a by 2a + 1. For the second integral, replace z by

x2 in the polylogarithm inversion formula in (1.135),

1
Li2a+1 (−x2 ) − Li2a+1 − 2
x
2a+1 2a+1 a 2a−2n+1
2 ln (x) X2 Li2n (−1) 2a−2n+1
=− +2 ln (x).
(2a + 1)! n=1
(2a − 2n + 1)!
Thus,
Li2a+1 (−x2 ) − Li2a+1 − x12

1
Z
dx
0 1 + x2
a
22a+1 1
ln2a+1 (x) 22a−2n+1 Li2n (−1) 1 ln2a−2n+1 (x)
Z X Z
=− dx + 2 dx
(2q + 1)! 0 1 + x2 n=1
(2a − 2n + 1)! 0 1 + x2
{ make use of (1.81) for the two integrals}
Xa
= 22a+1 β(2a + 2) − 2 22a−2n+1 Li2n (−1)β(2a − 2n + 2). (3.78)
n=1
Gather (3.77) and (3.78) and write Li2n (−1) = (21−2n − 1)ζ(2n) to complete the
proof.
R1 ln2a (x) arctan(x)

3.1.17 0 1−x2
dx

1
ln2a (x) arctan(x)
Z
dx = (2a)!(2−2 + 2−2a−3 − 2−2a−2 )πζ(2a + 1)
0 1 − x2
a
(2a)! X
− β(2a + 2) − (2a)! 2−2n (1 − 21−2n )ζ(2n)β(2a − 2n + 2).
2 n=1
(3.79)
Proof. Replace z by −x2 and a by 2a + 1 in (1.111),

1
−x2 ln2a (y)
Z
2
(2a)! Li2a+1 (−x ) = dy.
0 1 + x2 y

1
−x2 ln2a (y)
Z 1 Z 1
Li2a+1 (−x2 )
Z
1
(2a)! dx = dy dx
0 1 + x2 0 1+x
2 2 2
0 (1 + x )(1 + x y)
Z 1 Z 1
−x2

= ln2a (y) 2 2
dx dy
0 0 (1 + x )(1 + x y)
{evaluate the inner integral by partial fraction}
Z 1 √
2a π 1 arctan y
= ln (y) −√ dy
0 4 1−y y(1 − y)
Z 1 2a √
π 1 ln2a (y) ln (y) arctan( y)
Z
= dy − √ dy
4 0 1−y 0 y(1 − y)
√
{use (3.1) for the first integral and let y = x in the second one}
Z 1 2a
(2a)!π 2a+1 ln (x) arctan(x)
= ζ(2a + 1) − 2 dx
4 0 1 − x2
Reorder the terms,

Z 1 2a Z 1
ln (x) arctan(x) (2a)!π Li2a+1 (−x2 )
22a+1 2
dx = ζ(2a + 1) − (2a)! dx.
0 1−x 4 0 1 + x2
Plug in the result from (3.76) then divide both sides by 22a+1 to finish the proof.
Examples
1
ln2 (x) arctan(x)
Z
7π 1
2
dx = ζ(3) − ζ(2)β(2) − β(4); (3.80)
0 1−x 16 4
1
ln4 (x) arctan(x)
Z
93π 21
dx = ζ(5) − ζ(4)β(2) − 3ζ(2)β(4) − 12β(6);
0 1 − x2 16 16
(3.81)
1
ln6 (x) arctan(x)
Z
5715π 1395 315
dx = ζ(7) − ζ(6)β(2) − ζ(4)β(4)
0 1 − x2 32 128 8
−90ζ(2)β(6) − 360β(8). (3.82)
R∞ ln2a (x) ln(1+x)

3.1.18 0
√
x(1+x)
dx

Z∞
ln2a (x) ln(1 + x) d2a ψ(1 − m) + γ
√ dx = −π lim1 2a
. (3.83)
x(1 + x) m→ 2 dm sin(mπ)
0
Proof. We follow the same approach as in [17]:
Reduce n by m in the beta function in (1.51):

Z ∞
xm−1 Γ(m)Γ(n)
m+n
dx = B(m, n) = ,
0 (1 + x) Γ(m + n)
we have ∞
xm−1 Γ(m)Γ(n − m)
Z
n
dx = .
0 (1 + x) Γ(n)
Differentiate both sides 2a times with respect to m and once with respect to n,
Z∞
∂ 2a ∂ Γ(m)Γ(n − m) ∂ 2a ∂ xm−1
= dx
∂m2a ∂n Γ(n) ∂m2a ∂n (1 + x)n
0
Z∞ 2a
∂ ∂ xm−1
= dx
∂m2a ∂n (1 + x)n
0
Z∞
ln2a (x) ln(1 + x)xm−1
=− dx.
(1 + x)n
0
Now take the limit on both sides letting m → 1/2 and n → 1,

Z∞
ln2a (x) ln(1 + x) ∂ 2a ∂ Γ(m)Γ(n − m)
− √ dx = lim
x(1 + x) m→1/2 ∂m2a ∂n Γ(n)
0 n→1
∂ 2a

∂ Γ(n − m)
= lim Γ(m)
m→1/2 ∂m2a ∂n Γ(n)
n→1
∂ 2a

Γ(n − m)[ψ(n − m) − ψ(n)]
= lim Γ(m)
m→1/2 ∂m2a Γ(n)
n→1
{evaluate the limit when n → 1 and use ψ(1) = −γ given in (1.170)}
∂ 2a
= lim Γ(m)Γ(1 − m)[ψ(1 − m) + γ]
m→1/2 ∂m2a

π
use Γ(m)Γ(1 − m) = given in (1.38)
sin(mπ)

∂ d
and write as , since we have one variable left
∂m dm
d2a ψ(1 − m) + γ
= π lim1 2a
.
m→ 2 dm sin(mπ)
R1 ln2a (x) ln(1+x2 )

3.1.19 0 1+x2
dx
For a ∈ Z≥0 , the following identity holds:

1
ln2a (x) ln(1 + x2 )
Z
dx = −(2a + 1)!β(2a + 2)
0 1 + x2
d2a ψ(1 − m) + γ
−4−a−1 π lim1 2a
. (3.84)
m→ 2 dm sin(mπ)
Proof.
Z ∞ Z ∞ 2a
1
ln2a (x) ln(1 + x2 ) ln (x) ln(1 + x2 )
Z
dx = − dx
0 1 + x2 0 1 1 + x2
Z ∞ 2a ∞
ln (x) ln(1 + x2 ) ln2a (x) ln(1 + x2 )
Z
= dx − dx
0 1 + x2 1 + x2
|1 {z }
x→1/x
∞ 2a 2 1 2a
ln (x) ln(1 + x2 )
Z Z
ln (x) ln(1 + x )
= 2
dx − dx
0 1+x 0 1 + x2
Z 1 2a
ln (x) ln(x)
+2 dx
0 1 + x2
ln2a (x) ln(1 + x2 )
1
Z
0 1 + x2
Z ∞ 2a Z 1 2a+1
1 ln (x) ln(1 + x2 ) ln (x)
= 2
dx + dx
2 0 1+x 0 1 + x2
{set x2 → x in the first integral and use (1.81) for the second one}
Z ∞ 2a
ln (x) ln(1 + x)
= 4−a−1 √ dx − (2a + 1)!β(2a + 2).
0 x(1 + x)
The remaining integral is given in (3.83).
R1 lna (x) lna (1−x)

3.1.20 0 x(1−x)
dx
The following equality holds

1
lna (1 − x) lna (x) 1
lna (1 − x) lna (x)
Z Z
dx = 2 dx. (3.85)
0 x(1 − x) 0 x
Proof.
1
Z
dx
0 x(1 − x)
Z 1 a
1
ln (1 − x) lna (x)
Z
= dx + dx
1−x x
|0 {z } 0
1−x→x
ln (1 − x) lna (x)
a Z 1 a
1
ln (1 − x) lna (x)
Z
= dx + dx
0 x 0 x
Z 1 a
ln (1 − x) lna (x)
=2 dx.
0 x
R1 lna (x) lna (1−x)

3.1.21 0
2
x(1−x)
dx
For a ∈ Z+ , the following equality holds

1
lna (x) lna (1 − x) 1
lna (x) lna (1 − x)
Z 2
Z
dx = dx. (3.86)
0 x(1 − x) 0 x
Proof. Making the substitution 1 − x → x,

1
lna (1 − x) lna (x) 1
lna (x) lna (1 − x)
Z 2
Z
dx = dx
0 x(1 − x) 1
2
(1 − x)x
( 1
)
Z 2
0 x(1 − x)
1
lna (1 − x) lna (x) 1 1 lna (1 − x) lna (x)
Z Z
1 2
= dx + dx
2 0 x(1 − x) 2 12 x(1 − x)
Z 12 Z 1 ! a
1 ln (1 − x) lna (x)
= + dx
2 0 1
2
x(1 − x)
1 1 lna (1 − x) lna (x)
Z
= dx
2 0 (1 − x)x
Z 1 a
ln (1 − x) lna (x)
= dx,
0 x


3.1.22 0 1−x
dx

∞
" #
1
lna (x) ln(1 − x)
Z X Hn
dx = (−1)a a! ζ(a + 2) − . (3.87)
0 1−x n=1
na+1
Proof. Multiply both sides of the generating function in (2.5):

∞
ln(1 − x) X
=− Hn−1 xn−1
1−x n=1
by lna (x) then integrate from x = 0 to 1, we obtain

∞
1
lna (x) ln(1 − x)
Z X Z 1
dx = − Hn−1 xn−1 lna (x)dx
0 1−x n=1 0

∞ ∞
(−1)a a! Hn − n1
X X
a
=− Hn−1 a+1
= −(−1) a!
n=1
n n=1
na+1
"∞ ∞
# " ∞
#
X Hn X 1 X Hn
a a
= −(−1) a! − = (−1) a! ζ(a + 2) − .
n=1
na+1 n=1 na+2 n=1
na+1

3.1.23 0 x(1−x)
dx

∞
1
lna (x) ln(1 − x)
Z X Hn
dx = (−1)a−1 a! . (3.88)
0 x(1 − x) n=1
na+1
Proof. Multiply both sides of the generating function in (2.4):

∞
ln(1 − x) X
=− Hn xn
1−x n=1
lna (x)
by x then integrate from x = 0 to 1,
∞
1
lna (x) ln(1 − x)
Z X Z 1
dx = − Hn xn−1 lna (x)dx
0 x(1 − x) n=1 0
∞ ∞
(−1)a a!

X X Hn
=− Hn = (−1)a−1 a! .
n=1
na+1 n=1
n a+1
R1 lna (x) ln(1+x)

3.1.24 0 1+x
dx

∞
Z 1 a " #
ln (x) ln(1 + x) a −a−1
X (−1)n Hn
dx = (−1) a! (1 − 2 )ζ(a + 2) + .
0 1+x n=1
na+1
(3.89)
Proof. Replace x by −x in (2.5),

∞
ln(1 + x) X
= (−1)n Hn−1 xn−1 .
1+x n=1
Multiply both sides by lna (x) then integrate from x = 0 to 1,

∞
lna (x) ln(1 + x)
1
Z X Z 1
dx = (−1)n Hn−1 xn−1 lna (x)dx
0 1 + x n=1 0
∞
"∞ #
1 X (−1)n Hn
n Hn − n
X
a a
= (−1) a! (−1) = (−1) a! + η(a + 2)
n=1
na+1 n=1
na+1
{use η(s) = (1 − 21−s )ζ(s) given in (1.75)}
"∞ #
X (−1)n Hn
a −a−1
= (−1) a! + (1 − 2 )ζ(a + 2) .
n=1
na+1
R1 lna (x) ln(1+x)

3.1.25 0 x(1+x)
dx
For a ∈ Z≥0 , the following equality holds

∞
1
lna (x) ln(1 + x) (−1)n Hn
Z X
dx = (−1)a−1 a! . (3.90)
0 x(1 + x) n=1
na+1
Proof. Replace x by −x in (2.4),

∞
ln(1 + x) X
=− (−1)n Hn xn .
1+x n=1
lna (x)
Multiply both sides by x then integrate from x = 0 to 1,
∞
1
lna (x) ln(1 + x)
Z X Z 1
dx = − n
(−1) Hn xn−1 lna (x)dx
0 x(1 + x) n=1 0
∞
X (−1)n Hn
= (−1)a−1 a! .
n=1
na+1
R1 lna (1−x) ln(1+x)

3.1.26 0 x
dx

∞ (a+1)
1
lna (1 − x) ln(1 + x)
Z X Hn
dx = (−1)a a! . (3.91)
0 x n=1
n2n

1
lna (x)(1 − xn )
Z
(−1)a a!Hn(a+1) = dx
0 1−x
n
by n2 then take the summation over n ≥ 1,
∞ ∞
!
(a+1) Z 1 a
a
X Hn ln (x) X 1 − xn
(−1) a! n
= dx
n=1
n2 0 1−x n=1
n2n
∞ ∞
!
1
lna (x) xn
Z X 1 X
= − dx
0 1−x
n=1
n2n n=1 n2n
Z 1 a
ln (x) x
= ln(2) + ln 1 − dx
0 1−x 2
lna (x) ln(2 − x)
1 Z 1 a
ln (1 − x) ln(1 + x)
Z
1−x→x
= dx = dx.
0 1−x 0 x
R1 lna (x) ln( 1+x

2 )
3.1.27 0 1−x
dx

∞ (a+1)
1
lna (x) ln 1+x

(−1)n Hn
Z X
2
dx = (−1)a a! . (3.92)
0 1−x n=1
n

1
lna (x)(1 − xn )
Z
(−1)a a!Hn(a+1) = dx
0 1−x
(−1)n
by n then consider the summation over n ≥ 1,
∞ ∞
!
n (a+1) Z 1 a n
X (−1) H n ln (x) X (−1)
(−1)a a! = (1 − xn ) dx
n=1
n 0 1−x n=1
n
∞ ∞
Z 1 a !
ln (x) X (−1)n X (−x)n
= − dx
0 1−x n=1
n n=1
n
Z 1 a
ln (x) ln 1+x
Z 1 a
ln (x) 2
= (− ln(2) + ln(1 + x))dx = dx.
0 1−x 0 1−x
R1 lna (1−x) Li2 (x)

3.1.28 0 x
dx
lna (1 − x) Li2 (x)

1
Z
dx = (−1)a [a! + (a + 1)!]ζ(a + 3)
0 x
∞ ∞ (2)
X Hn X Hn
+(−1)a a!ζ(2)ζ(a + 1) − (−1)a (a + 1)! − (−1) a
a! .
n=1
na+2 n=1
na+1
(3.93)
Proof.
lna (1 − x) Li2 (x)

1 Z 1 a
ln (y) Li2 (1 − y)
Z
1−x=y
dx = dy
0 x 0 1−y
{recall the dilogarithm reflection formula given in (1.119)}
Z 1 a
ln (y)
= (ζ(2) − ln(y) ln(1 − y) − Li2 (y))dy
0 1−y
Z 1 a Z 1 a+1 Z 1 a
ln (y) ln (y) ln(1 − y) ln (y) Li2 (y)
= ζ(2) dy − dy − dy.
0 1 − y 0 1 − y 0 1−y
The first two integrals are calculated in (3.1) and (3.87). For the third integral, expand
Li2 (y)
1−y in series given in (2.3),
∞
1
lna (y) Li2 (y)
Z Z 1
(2)
X
dy = Hn−1 y n−1 lna (y)dy
0 1−y n=1 0
∞ ∞ (2)
(−1)a a! Hn − n12

(2)
X X
= Hn−1 = (−1)a a!
n=1
na+1 n=1
na+1
∞ ∞ ∞
" # " #
(2) (2)
X Hn X 1 X Hn
= (−1)a a! − a
= (−1) a! − ζ(a + 3) .
n=1
na+1 n=1
na+3 n=1
na+1
Group the three integrals to finalize the proof.
R∞ ln2a−1 (x) ln(1+x)

3.1.29 0 x(1+x)
dx
∞
ln2a−1 (x) ln(1 + x)
Z
dx = (2a − 1)!(1 + 2a)(1 − 2−2a )ζ(2a + 1)
0 x(1 + x)
∞
X (−1)n Hn
+2(2a − 1)! . (3.94)
n=1
n2a
Proof. Replace a by 2a − 1 in (3.90),

∞ Z 1 2a−1
X (−1)n Hn 1 ln (x) ln(1 + x) 1
2a
= dx = Ia . (3.95)
n=1
n (2a − 1)! 0 x(1 + x) (2a − 1)!
Let’s find Ia :
∞ ∞
ln2a−1 (x) ln(1 + x) ln2a−1 (x) ln(1 + x)
Z Z
Ia = dx − dx
0 x(1 + x) x(1 + x)
|1 {z }
x7→1/x
∞
ln2a−1 (x) ln(1 + x) 1
ln2a−1 (x) ln(1 + x) 1
ln2a (x)
Z Z Z
= dx + dx − dx.
0 x(1 + x) 0 1+x 0 1+x
By adding
ln2a−1 (x) ln(1 + x)

1
Z
Ia := dx
0 x(1 + x)
ln2a−1 (x) ln(1 + x)
Z 1 Z 1 2a−1
ln (x) ln(1 + x)
= dx − dx
0 x 0 1+x
R1 ln2a−1 (x) ln(1+x)
to both sides, the integral 0 1+x dx nicely cancels out,
∞
ln2a−1 (x) ln(1 + x) 1
ln2a−1 (x) ln(1 + x) 1
ln2a (x)
Z Z Z
2Ia = dx + dx − dx
0 x(1 + x) x 1+x
|0 {z } 0
IBP
∞
ln2a−1 (x) ln(1 + x) 1 + 2a 1 ln2a (x)
Z Z
= dx − dx
0 x(1 + x) 2a 0 1+x
{make use of (3.7) for the second integral}
∞
ln2a−1 (x) ln(1 + x)
Z
1 + 2a
= dx − (2a)!(1 − 2−2a )ζ(2a + 1).
0 x(1 + x) 2a
(2a)!
Divide both sides by 2 and use 2a = (2a − 1)!,
∞
ln2a−1 (x) ln(1 + x)
Z
1 1 + 2a
Ia = dx − (2a − 1)!(1 − 2−2a )ζ(2a + 1).
2 0 x(1 + x) 2
The proof completes on plugging Ia in (3.95).
x2n
R1
3.1.30 0 1+x
dx

1
x2n
Z
1
dx = ln(2) + Hn − H2n , R(n) > − . (3.96)
0 1+x 2
Proof. Forcing integration by parts,

1 Z 1
x2n
Z
2n 1
dx = ln(1 + x)x 0 − 2n x2n−1 ln(1 + x)dx
0 1+x 0
{expand ln(1 + x) in series}
∞
(−1)k−1 1 2n+k−1
X Z
= ln(2) − 2n x dx
k 0
k=1
∞
X (−1)k
= ln(2) + 2n
k(k + 2n)
k=1
∞ ∞ ∞
( )
X X X
k
employ (−1) ak = 2 a2k − ak given in (1.5)
k=1 k=1 k=1
∞ ∞
X 1 X 1
= ln(2) + 4n − 2n
2k(2k + 2n) k(k + 2n)
k=1 k=1
∞ ∞
X n X 2n
= ln(2) + −
k(k + n) k(k + 2n)
k=1 k=1
{recall the definition of the harmonic number in (1.155)}
= ln(2) + Hn − H2n ,

If we compare the first and last lines of the proof above, we see that
Z 1
ln(2) − 2n x2n−1 ln(1 + x)dx = ln(2) + Hn − H2n
0
or
1
H2n − Hn
Z
x2n−1 ln(1 + x)dx = . (3.97)
0 2n
An alternative proof for (3.97) may be found in [25].
xn
R1
3.1.31 0 1+x
dx

1
xn
Z
1
dx = H n2 − H n−1 , R(n) > −1. (3.98)
0 1+x 2 2
Proof. We begin with the definition of the harmonic number in (1.151),

n 1
Z 1 n−1
1 − x2 1−x 2
Z
H n2 − H n−1 = dx − dx
2
0 1−x 0 1−x
Z 1 n−1 n Z 1 n
x 2 − x2 x=y 2 y − y n+1
= dx = 2 dy
0 1−x 0 1 − y2
Z 1 n Z 1 n
y (1 − y) y
=2 2
dy = 2 dy.
0 1 − y 0 1 +y
For a different form, use Hn = ψ(n + 1) + γ given in (1.169),

n+2 n+1
H 2 − H n−1 = ψ
n +γ− ψ +γ
2 2 2

n+2 n+1
=ψ −ψ .
2 2
Substituting this in (3.98) yields
1
xn
Z
1 n+2 1 n+1
dx = ψ − ψ . (3.99)
0 1+x 2 2 2 2
Moreover, if we replace n by 2n in (3.98) and compare it with (3.96), we deduce
Hn− 21 = 2H2n − Hn − 2 ln(2). (3.100)

R1
3.1.32 0
x2n−1 arctanh(x)dx

1
2H2n − Hn
Z
1
x2n−1 arctanh(x)dx = , R(n) > − . (3.101)
0 4n 2
1 1
Proof. We have arctanh(x) = 2 ln(1 + x) − 2 ln(1 − x). Then
Z 1
x2n−1 arctanh(x)dx
0
1
1 1 2n−1
Z Z
1 2n−1
= x ln(1 + x)dx − x ln(1 − x)dx
2 0 2 0
1 H2n − Hn 1 H2n
= · + ·
2 2n 2 2n
2H2n − Hn
= .
4n
R1
3.1.33 0
xn−1 Lia (x)dx
1 a−1
ζ(a − k + 1)
Z
Hn X
xn−1 Lia (x)dx = (−1)a−1 − (−1)k . (3.102)
0 na nk
k=1
Proof. Let’s evaluate the following integrals:

First integral:
1
xn 1 1 n−1
Z Z
IBP 1
xn−1 Li2 (x)dx = Li2 (x) 0 + x ln(1 − x)dx
0 n n 0
ζ(2) Hn
= − 2. (3.103)
n n
Second integral:
1
xn 1 1 n−1
Z Z
1
xn−1 Li3 (x)dx = Li3 (x) 0 − x Li2 (x)dx
0 n n 0
ζ(3) ζ(2) Hn
= − 2 + 3. (3.104)
n n n
Third integral:
1
xn 1 1 n−1
Z Z
1
xn−1 Li4 (x)dx = Li4 (x) 0 − x Li3 (x)dx
0 n n 0
ζ(4) ζ(3) ζ(2) Hn
= − 2 + 3 − 4. (3.105)
n n n n
So in general we have
1
ζ(a) ζ(a − 1)
Z
ζ(2) Hn
xn−1 Lia (x)dx = − 2
+ .... + (−1)a a−1 + (−1)a−1 a
0 n n n n
a−1
X ζ(a − k + 1) Hn
= (−1)k−1 k
+ (−1)a−1 a ,
n n
k=1

3.2 Results of Logarithmic Integrals

Rπ
3.2.1 2
0
sin(2nx) cot(x)dx
Show that Z π
2 π
sin(2nx) cot(x)dx = , n ∈ Z+ . (3.106)
0 2
Solution Since cos x = Reix , which follows from Euler’s formula, we have
n−1 n−1
( n−1 )
2ix k
X X X
(2k+1)ix ix

cos((2k + 1)x) = R e =R e e
k=0 k=0 k=0
1 − e2ixn

= R eix
1 − e2ix
{make use of eix = cos x + i sin x and simplify}
sin2 (nx)

sin(2nx) sin(2nx)
=R +i = .
2 sin x sin x 2 sin x
Separate the first term of the sum,

n−1 n−1
sin(2nx) X X
= cos((2k + 1)x) = cos x + cos((2k + 1)x).
2 sin x
k=0 k=1
Multiply both sides by 2 cos x,

n−1
X
sin(2nx) cot x = 2 cos2 x + 2 cos((2k + 1)x) cos x.
k=1
Use 2 cos2 x = 1 + cos(2x) and 2 cos a cos b = cos(a − b) + cos(a + b),

n−1
X
sin(2nx) cot x = 1 + cos(2x) + cos(2kx) + cos((2k + 2)x).
k=1
Now integrate both sides from x = 0 to π/2,

Z π
2
sin(2nx) cot xdx
0
Z π Z π n−1
2 2 X
= (1 + cos(2x))dx + [cos(2kx) + cos((2k + 2)x)]dx
0 0 k=1
3.2. Results of Logarithmic Integrals 159
{reverse the order of integration and summation}

π n−1 Z π
sin(2x) 2 X 2
=x+ + [cos(2kx) + cos((2k + 2)x)]dx
2 0 k=1 0
n−1 π
π sin(π) 1 X sin(2kx) sin((k + 1)x) 2
= + + +
2 2 2 k k+1 0
k=1
n−1 n−1
π 1 X sin(kπ) 1 X sin((k + 1)π)
= +0+ + .
2 2 k 2 k+1
k=1 k=1
For n ∈ Z+ , we have
n−1
X sin(kπ) sin(π) sin(2π) sin((n − 1)π)
= + + ··· + = 0 + 0 + · · · + 0 = 0,
k 1 2 n−1
k=1
and the same applies to the second sum. Therefore, the integral is π/2.
Rπ
3.2.2 0
2
ln(sin x)dx
Show that Z π
2 π
ln(sin x)dx = − ln(2). (3.107)
0 2
Rb Rb
Solution (i) By the rule a f (x)dx = a f (a + b − x)dx, which can be proved by
setting a + b − x = y then changing y to x, we have
Z π Z π Z π
2 2 π 2
ln(sin x)dx = ln sin −x dx = ln(cos x)dx.
0 0 2 0
R π
Add the integral 0
2
ln(sin x)dx to both sides,
Z π Z π Z π
2 2 2
2 ln(sin x)dx = ln(sin x)dx + ln(cos x)dx
0 0 0
Z π
2
= ln(sin x cos x)dx
0

sin(2x)
use sin x cos x =
2
Z π2 Z π2
= ln(sin(2x))dx − ln(2)dx
|0 {z } 0
2x→x
Z π
1 π
ln(sin x)dx − ln(2)
=
0 2 2
Z π2 Z π !
1 π
= + ln(sin x)dx − ln(2)
2 0 π
2
2
Z π2 Z π
1 1 π
= ln(sin x)dx + ln(sin x)dx − ln(2)
2 0 2 π2 2
| {z }
x=π−y
Z π Z π
1 2 1 2 π
= ln(sin x)dx + ln(sin y)dy − ln(2)
2 0 2 2
|0 {z }
y=x
Z π
2 π
= ln(sin x)dx − ln(2).
0 2
Therefore, Z π Z π
2 2 π
2 ln(sin x)dx = ln(sin x)dx − ln(2),
0 0 2
R π
and the solution completes on subtracting 2
0
ln(sin x)dx from both sides.
Solution (ii) Differentiate both sides of the beta function in (1.50):

Z π
2 1 1 Γ(a)Γ(b)
cos2a−1 (x) sin2b−1 (x)dx = B(a, b) =
0 2 2 Γ(a + b)
with respect to b then use the rule given in (2.79),

Z π
2 1 ∂ Γ(a)Γ(b)
2 cos2a−1 (x) sin2b−1 (x) ln(sin x)dx =
0 2 ∂b Γ(a + b)
{use Γ0 (x) = Γ(x)ψ(x) given in (1.167)}
1 Γ(a)Γ(b)(ψ(b) − ψ(a + b))
= .
2 Γ(a + b)
Now set a = b = 1/2 then divide through by 2,

Z π2
Γ2 12 ψ 12 − ψ(1)

ln(sin x)dx =
0 4 Γ(1)
√
{write Γ(1/2) = π and recall the result from (1.172)}
π π
= (−2 ln(2)) = − ln(2).
4 2
Rπ
3.2.3 0
2
ln2 (sin x)dx
Show that π
π3
Z 2 π 2
ln2 (sin x)dx = ln (2) + . (3.108)
0 2 24
Solution (i) We have
ln2 (2 sin x) = ln2 (2) + 2 ln(2) ln(sin x) + ln2 (sin x)
or
ln2 (sin x) = ln2 (2 sin x) − 2 ln(2) ln(sin x) − ln2 (2).
Integrate both sides from x = 0 to π/2,
Z π Z π Z π Z π
2 2 2 2
ln2 (sin x)dx = ln2 (2 sin x)dx − 2 ln(2) ln(sin x)dx − ln2 (2)dx.
0 0 0 0
The third integral is π2 ln2 (2) and the second integral is − π2 ln(2) given in (3.107).
For the first one, integrate both sides of (2.141):
π 2 ∞
X Hn−1
ln2 (2 sin x) = −x +2 cos(2nx)
2 n=1
n
from x = 0 to π/2 then change the order of integration and summation,

Z π Z π
π 2 ∞ Z π2
2 2 X Hn−1
ln2 (2 sin x)dx = − x dx + 2 cos(2nx)dx
0 0 2 n=1
n 0
π ∞ π
1 π 3 2 X Hn−1 sin(2nx) 2
=− −x +2 ·
3 2 0 n=1
n 2n 0
3 ∞
π Hn−1 sin(nπ)
X
= +2 ·
24 n=1
n 2n
{the sum evaluates to 0, since sin(nπ) = 0 for integer n}
π3
= .
24
Combining the three integrals completes the solution.

Z π
2 1 1 Γ(a)Γ(b)
cos2a−1 (x) sin2b−1 (x)dx = B(a, b) =
0 2 2 Γ(a + b)
with respect to b twice,

π
1 ∂ 2 Γ(a)Γ(b)
Z 2
4 cos2a−1 (x) sin2b−1 (x) ln2 (sin x)dx =
0 2 ∂b2 Γ(a + b)
1 Γ(a)Γ(b) h i
= (ψ(b) − ψ(a + b))2 + ψ (1) (b) − ψ (1) (a + b) .
2 Γ(a + b)
Setting a = b = 1/2 then dividing by 4 gives

Z π2 " 2 #
2 1
1 Γ ( ) 1 1
ln2 (sin x)dx = 2
ψ − ψ(1) + ψ (1) − ψ (1) (1)
0 8 Γ(1) 2 2
{recall the results from (1.172), (1.187), and (1.181)}
π π π3
= (−2 ln(2))2 + 3ζ(2) − ζ(2) = ln2 (2) + .
8 2 24
∂ 2 Γ(a)Γ(b)
The Mathematica command for lim is
a→1/2 ∂b2 Γ(a + b)
b→1/2
Normal[Series[D[Gamma[a]Gamma[b]/Gamma[a+b],{a,0},{b,2}]
,{a,1/2,0},{b,1/2,0}]]//FullSimplify//Expand
Rπ
3.2.4 0
2
ln(sin x) ln(cos x)dx
Show that π
π3
Z 2 π 2
ln(sin x) ln(cos x)dx = ln (2) − . (3.109)
0 2 48
Solution (i) Let a = ln(sin x) and b = ln(cos x) in the algebraic identity

1 2 1 2 1
ab = a + b − (a − b)2 ,
2 2 2
we have
1 2 1 1
ln(sin x) ln(cos x) = ln (sin x) + ln2 (cos x) − ln2 (tan x).
2 2 2
Integrate both sides from x = 0 to π/2,
Z π
2
ln(sin x) ln(cos x)dx
0
Z π Z π Z π
1 2 1 2 1 2
= ln2 (sin x)dx + ln2 (cos x)dx − ln2 (tan x)dx.
2 0 2 0 2 0
Rb
The second integral is equivalent to the first one by using the rule a
f (x)dx =
Rb
a
f (a + b − x)dx. For the third integral, let tan x = y, we have
π π
∞
ln2 (y)
Z Z Z
2 2 1
ln(sin x) ln(cos x)dx = ln2 (sin x)dx − dy.
0 0 2 0 1 + y2
These two integrals are given in (3.108) and (3.54).

Solution (ii) Differentiate both sides of (1.50) with respect to a and b,
π
1 ∂ 2 Γ(a)Γ(b)
Z 2
4 cos2a−1 (x) sin2b−1 (x) ln(cos x) ln(sin x)dx =
0 2 ∂a∂b Γ(a + b)
{use Γ0 (n) = Γ(n)ψ(n) given in (1.167) and ψ 0 (n) ≡ ψ (1) (n)}
1 Γ(a)Γ(b) h i
= (ψ(a) − ψ(a + b))(ψ(b) − ψ(a + b)) − ψ (1) (a + b) .
2 Γ(a + b)
Setting a = b = 1/2 then dividing by 4 gives

Z π2 " 2 #
1 Γ2 12 1
ln(sin x) ln(cos x)dx = ψ − ψ(1) − ψ (1) (1)
0 8 Γ(1) 2
{recall the results form (1.172) and (1.181)}
π2 π3

π π
= (−2 ln(2))2 − = ln2 (2) − .
8 6 2 48
∂ 2 Γ(a)Γ(b)
a→1/2 ∂a∂b Γ(a + b)
b→1/2
,{a,1/2,0},{b,1/2,0}]]//FullSimplify//Expand
R1 ln(x) ln(1−x)
3.2.5 0
√
x 1−x
dx
Show that
1
ln(x) ln(1 − x)
Z
√ dx = 7ζ(3) − 6 ln(2)ζ(2). (3.110)
0 x(1 − x)
√
Solution (i) Make the change of variable 1 − x = y,
1 1
ln(x) ln(1 − x) ln(y) ln(1 − y 2 )
Z Z
√ dx = 4 dy
0 x(1 − x) 1 − y2
0
Z 1 ln 1−x ln 4x
y= 1−x
1+x 1+x (1+x)2
= 2 dx
0 x

1−x 1−x
Z 1 ln 1+x
Z 1 ln(x) ln 1+x
Z 1
ln2 (1 + x)
= 4 ln(2) dx + 2 dx + 4 dx
0 x 0 x 0 x
1
ln(1 − x) ln(1 + x)
Z
−4 dx.
0 x
All these integrals are given in (3.21), (3.22), (3.38), and (3.115) respectively.

Z 1
Γ(a)Γ(b)
xa−1 (1 − x)b−1 dx = B(a, b) =
0 Γ(a + b)
with respect to a and b,

1
∂ 2 Γ(a)Γ(b)
Z
xa−1 (1 − x)b−1 ln(x) ln(1 − x)dx = .
0 ∂a∂b Γ(a + b)
Next, take the limit on both sides letting b → 0,

1
xa−1 ln(x) ln(1 − x) ∂ 2 Γ(a)Γ(b)
Z
dx = lim
0 1−x b→0 ∂a∂b Γ(a + b)
Γ(a)Γ(b) h i
= lim (ψ(a) − ψ(a + b))(ψ(b) − ψ(a + b)) − ψ (1) (a + b)
b→0 Γ(a + b)

Γ(1 + b) 1
write Γ(b) = and ψ(b) = ψ(1 + b) −
b b
{given in (1.32) and (1.173)}
Γ(a)Γ(1 + b)
= lim ×
b→0 Γ(a + b)
(ψ(a) − ψ(a + b))(ψ(1 + b) − 1b − ψ(a + b)) − ψ (1) (a + b)

lim
b→0 b

Γ(a)Γ(1 + b) Γ(a)Γ(1)
note that lim = =1
b→0 Γ(a + b) Γ(a)
(ψ(a) − ψ(a + b))(ψ(1 + b) − 1b − ψ(a + b)) − ψ (1) (a + b)

= lim
b→0 b
{multiply by b/b}

(ψ(a) − ψ(a + b))(bψ(1 + b) − 1 − bψ(a + b)) − bψ (1) (a + b)
= lim
b→0 b2
{now we can apply L’Hopital’s rule, since we have 0/0}
1 n (1)
= lim ψ (a + b) [bψ(a + b) − bψ(1 + b)] + [ψ(a) − ψ(a + b)]
b→0 2b
o
[ψ(1 + b) − ψ(a + b) − ψ (1) (a + b) + bψ (1) (1 + b)] − bψ (2) (a + b)
{apply L’Hopital’s rule again, since we have 0/0}

1 n (1)
= lim 2ψ (a + b)[ψ(a + b) + bψ (1) (a + b) − ψ(1 + b) − bψ (1) (1 + b)]
b→0 2
+ψ (2) (a + b)[bψ(a + b) − bψ(1 + b) − 1] + [ψ(a) − ψ(a + b)]

o
[2ψ (2) (a + b) + bψ (2) (a + b) − 2ψ (2) (1 + b) − bψ (2) (1 + b)] − bψ (3) (a + b)
1 n (1) o
= 2ψ (a)[ψ(a) − ψ(1)] − ψ (2) (a)
2
{substitute ψ(1) = −γ given in (1.170)}
1 n (1) o
= 2ψ (a)[ψ(a) + γ] − ψ (2) (a) .
2
Thus,
Z 1
xa−1 ln(x) ln(1 − x) 1
dx = ψ (1) (a)[ψ(a) + γ] − ψ (2) (a), R(a) > −1.
0 1−x 2
(3.111)
Set a = 1/2 in (3.111),
1
ln(x) ln(1 − x)
Z
1 1 1 1
√ dx = ψ (1) ψ + γ − ψ (2)
0 x(1 − x) 2 2 2 2
= 7ζ(3) − 6 ln(2)ζ(2).
Of course solution (i) is much easier than solution (ii), but the latter is more useful
because it involves the identity (3.111), which allows us to adjust the power of ln(x)
in the numerator by differentiating both sides with respect to a.
Moreover, by setting x = sin2 u in (3.111), we obtain
Z π
2 1 (1) 1
sin2a−2 (u) tan u ln(sin u) ln(cos u)du =
ψ (a)[ψ(a) + γ] − ψ (2) (a).
0 8 16
(3.112)
R1
Remark: We know that the beta function, B(a, b) = 0 xa−1 (1 − x)b−1 dx, is
defined for R(a) > 0. However, its derivative in (3.111) is defined for R(a) > −1
due to the analytic continuation (see[35]).
R∞ ln2 (x) ln(1+x2 )

3.2.6 0 1+x2
dx
Show that
∞
ln2 (x) ln(1 + x2 ) π3
Z
7
2
dx = ln(2) + πζ(3). (3.113)
0 1+x 4 4
Solution Make the substitution x2 → x,

Z ∞ 2
ln (x) ln(1 + x2 ) 1 ∞ ln2 (x) ln(1 + x)
Z
dx = √ dx. (3.114)
0 1 + x2 8 0 x(1 + x)
This integral can be calculated using the same technique as in [17]:

Replace b by b − a in the beta function in (1.52):
Z ∞
xa−1 Γ(a)Γ(b)
a+b
dx = B(a, b) = ,
0 (1 + x) Γ(a + b)
we get
∞
xa−1 Γ(a)Γ(b − a)
Z
dx = .
0 (1 + x)b Γ(b)
Differentiate twice with respect to a and once with respect to b,
Z∞
ln2 (x) ln(1 + x)xa−1 ∂ 2 ∂ Γ(a)Γ(b − a)
b
dx = − 2 .
(1 + x) ∂a ∂b Γ(b)
0
Now take the limit on both sides letting a → 1/2 and b → 1,

Z∞
ln2 (x) ln(1 + x) ∂ 2 ∂ Γ(a)Γ(b − a)
√ dx = − lim
x(1 + x) a→1/2 ∂a2 ∂b Γ(b)
0 b→1
∂2

∂ Γ(b − a)
= − lim Γ(a)
a→1/2 ∂a2 ∂b Γ(b)
b→1
∂2

Γ(b − a)[ψ(b − a) − ψ(b)]
= − lim Γ(a)
a→1/2 ∂a2 Γ(b)
b→1
{evaluate the limit when b → 1}
∂2
= − lim Γ(a)Γ(1 − a)[ψ(1 − a) − ψ(1)]
a→1/2 ∂a2

π
use Γ(a)Γ(1 − a) = given in (1.38)
sin(aπ)

∂ d
and write as , since we have one variable left
∂a da
d2 π
= − lim 2
[ψ(1 − a) − ψ(1)]
a→1/2 da sin(aπ)
π
= − lim [2ψ (2) (1 − a) + 4π cot(aπ)ψ (1) (1 − a)
a→1/2 2 sin(aπ)
+π 2 csc2 (aπ)(ψ(1 − a) − ψ(1))(cos(2aπ) + 3)]


π (2) 1 π
(1) 1
=− π
2ψ + 4π cot ψ
2 sin 2
2 2 2
π 1
+π 2 csc2 ψ − ψ(1) (cos(π) + 3)
2 2

π 1 1
=− 2ψ (2) + π2 ψ − ψ(1) (−1 + 3)
2 2 2
{recall the results from (1.188) and (1.172)}
= 2π 3 ln(2) + 14πζ(3).
Substituting this value in (3.114) completes the solution.

∂ 2 ∂ Γ(a)Γ(b − a)
a→1/2 ∂a2 ∂b Γ(b)
b→1
Normal[Series[D[Gamma[a]Gamma[b-a]/Gamma[b],{a,2},{b,1}]
,{a,1/2,0},{b,1,0}]]//FullSimplify//Expand
R1 ln(1−x) ln(1+x)
3.2.7 0 x
dx
Show that
1
ln(1 − x) ln(1 + x)
Z
5
dx = − ζ(3). (3.115)
0 x 8
Solution Let a = ln(1 − x) and b = ln(1 + x) in the algebraic identity

1 1
ab = (a + b)2 − (a − b)2 ,
4 4
we have
1 1 1−x
ln(1 − x) ln(1 + x) = ln2 (1 − x2 ) − ln2 .
4 4 1+x
Divide both sides by x then integrate from x = 0 to 1,

Z 1
ln(1 − x) ln(1 + x) 1
Z 1
ln2 (1 − x2 ) 1
Z 1 ln2 1−x
1+x
dx = dx − dx
0 x 4 x 4 x
|0 {z } 0
1−x2 =y

Z 1 ln2 1−x
1 1 ln2 (y)
Z
1+x
= dy − dx.
8 0 1−y 0 x
{gather the results from (3.3) and (3.16)}
5
= − ζ(3),
8
which matches (3.115).
R1 ln(x) ln(1−x) ln(1+x)

3.2.8 0 x
dx
Show that
Z 1
ln(x) ln(1 − x) ln(1 + x) 1 27 7
dx = 2 Li4 − ζ(4) + ln(2)ζ(3)
0 x 2 16 4
1 2 1 4
− ln (2)ζ(2) + ln (2). (3.116)
2 12
Solution (i) Let a = ln(1 − x) and b = ln(1 + x) in the algebraic identity

1 1 1
(a + b)2 − a2 − b2 ,
ab =
2 2 2
1 2 1 2 1
ln(1 − x) ln(1 + x) = ln (1 − x ) − ln (1 − x) − ln2 (1 + x).
2
2 2 2
ln(x)
Multiply both sides by x then integrate,
1
ln(x) ln(1 − x) ln(1 + x) 1
ln(x) ln2 (1 − x2 )
Z Z
1
dx = dx
0 x 2 x
|0 {z }
1−x2 =y
2
1
ln(x) ln (1 − x) 1 1 ln(x) ln2 (1 + x)
Z Z
1
− dx − dx
2 x 2 0 x
|0 {z } | {z }
1−x=y IBP
2
1
ln(1 − y) ln (y) ln(1 − y) ln2 (y)
1
Z Z
1 1
= dy − dy
08 1−y 2 0 1−y
1 1 ln2 (x) ln(1 + x)
Z
1 2 2 1
− ln (x) ln (1 + x) 0 + dx
4 2 0 1+x
ln(1 − y) ln2 (y) 1 1 ln2 (x) ln(1 + x)
Z 1 Z
3
=− dy + dx
8 0 1−y 2 0 1+x
{set a = 2 in (3.87) to get the first integral}
∞
1 1 ln2 (x) ln(1 + x)
Z
3 3 X Hn
= − ζ(4) + + dx,
4 4 n=1 n3 2 0 1+x
and the solution finalizes on gathering the results from (4.6) and (3.143).
Solution (ii) Multiply both sides of (2.47):

∞
X H2n − Hn 1
ln(1 − x) ln(1 + x) = − + 2 x2n
n=1
n 2n
ln(x)
1
ln(1 − x) ln(1 + x) ln(x)
Z
dx
0 x
∞ Z 1
X H2n − Hn 1
=− + 2 x2n−1 ln(x)dx
n=1
n 2n 0
∞
X H2n − Hn 1 1
=− + 2 −
n=1
n 2n (2n)2
∞ ∞
X H2n 1 X Hn 1
=2 3
− 3
+ ζ(4)
n=1
(2n) 4 n=1
n 8
{make use of (1.5) for the first sum}
∞ ∞
X (−1)n Hn 3 X Hn 1
= 3
+ 3
+ ζ(4).
n=1
n 4 n=1
n 8
These two sums are given in (4.143) and (4.5).
R1 ln(1−x) ln2 (1+x)

3.2.9 0 x
dx
Show that
1
ln(1 − x) ln2 (1 + x)
Z
3
dx = − ζ(4). (3.117)
0 x 8
Solution Put a = ln(1 − x) and b = ln(1 + x) in the algebraic identity

1 1 1
ab2 = (a + b)3 + (a − b)3 − a3 ,
6 6 3
ln(1 − x) ln2 (1 + x)
1
Z
dx
0 x

3 1−x
1
ln3 (1 − x2 ) 1 1 ln 1+x 1 1 ln3 (1 − x)
Z Z Z
1
= dx + dx − dx
6 x 6 0 x 3 0 x
|0 {z } | {z }
1−x2 =y 1−x=y

3 1−x
ln3 (y)
1
1 1 ln 1+x 1 1 ln3 (y)
Z Z Z
1
= dy + dx − dy
12 0 1−y 6 0 x 3 0 1−y

3 1−x
1 1 ln3 (y) 1 1 ln 1+x
Z Z
=− dy + dx.
6 0 1−y 4 0 x
The solution completes on putting together the results from (3.4) and (3.17).
R1 ln2 (1−x) ln(1+x)

3.2.10 0 x
dx
Show that
ln2 (1 − x) ln(1 + x)
Z 1
1 5 7
dx = 2 Li4 − ζ(4) + ln(2)ζ(3)
0 x 2 8 4
1 2 1 4
− ln (2)ζ(2) + ln (2). (3.118)
2 12
Solution Set a = ln(1 − x) and b = ln(1 + x) in

1 1 1
a2 b = (a + b)3 − (a − b)3 − b3 ,
6 6 3
we have
ln2 (1 − x) ln(1 + x)
1
Z
dx
0 x

3 1−x
1
ln3 (1 − x2 ) 1 1 ln 1+x 1 1 ln3 (1 + x)
Z Z Z
1
= dx − dx − dx
6 x 6 0 x 3 0 x
|0 {z }
1−x2 =y

1
Z 1 3
ln (y) 1
Z 1 ln3 1−x
1+x 1
Z 1
ln3 (1 + x)
= dy − dx − dx.
12 0 1−y 6 0 x 3 0 x
These three integrals are given in (3.4), (3.17), and (3.39).
R1 ln3 (1−x) ln(1+x)

3.2.11 0 x
dx
Show that
Z 1 3
ln (1 − x) ln(1 + x) 1 1 81
dx = 6 Li5 + 6 ln(2) Li4 − ζ(5)
0 x 2 2 16
21 21 2 1
− ζ(2)ζ(3) + ln (2)ζ(3) − ln3 (2)ζ(2) + ln5 (2). (3.119)
8 8 5

Solution Let I denote our integral. Making use of the algebraic identity
1 4 1 4 1 5 1
a3 b = a − b + (a + b)4 − (a − b)4 − (a − b)3 b,
4 4 16 16 2
we have
1
ln4 (1 − x) 1
ln4 (1 + x) 1
ln4 (1 − x2 )
Z Z Z
1 1 1
I= dx − dx + dx
4 x 4 x 16 x
|0 {z } 0
|0 {z }
1−x=y 1−x2 =y

5
Z 1 ln4 1−x
1+x 1
Z 1 ln3 1−x
1+x ln(1 + x)
− dx − dx
16 0 x 2 0 x
| {z }
(1−x)/(1+x)=y

9
Z 1 4
ln (y) 1 ln (1 + x)
Z
51 4 Z 1 ln4 1−x
1+x
= dy − dx − dx
32 0 1−y 4 0 x 16 0 x
Z 1 3
ln (y) ln 1+y

2
+ dy .
1 − y2
|0 {z }
J
Collect the results from (3.5), (3.40), and (3.18),

1 1 441 21 2
I = 6 Li5 + 6 ln(2) Li4 − ζ(5) + ln (2)ζ(3)
2 2 32 8
1
ln3 (2)ζ(2) + ln5 (2) + J. (3.120)
5
ln( 1+y
2 )
1+y
1 ln( 2 )
For the integral J, write 1−y 2 = − 12 ln(2)
1+y + 2 1−y + 1 ln(1+y)
2 1+y ,
3 1+y
1
ln3 (y) 1 1 ln (y) ln 2
Z Z
1
J = − ln(2) dy + dy
2 1+y 2 0 1−y
|0 {z } | {z }
J1 J2
Z 1 3
1 ln (y) ln(1 + y)
+ dy.
2 1+y
|0 {z }
J3
∞ (4)
X (−1)n Hn
Setting a = 3 in (3.92) gives J2 = −6 , and setting a = 3 in (3.89)
n=1
n
∞
45 X (−1)n Hn
gives J3 = − ζ(5) − 6 .
8 n=1
n4
Collecting the values of J2 and J3 along with the value of J1 given in (3.10),
∞ (4) ∞
21 45 X (−1)n Hn X (−1)n Hn
J= ln(2)ζ(4) − ζ(5) − 3 −3
8 16 n=1
n n=1
n4
{recall the relation involving the first sum from (4.155)}
∞ ∞
3 21 X Hn X (−1)n Hn
= − ζ(2)ζ(3) − 4
− 9
4 8 n=1 n n=1
n4
{ substitute the results from (4.6) and (4.150)}
279 21
= ζ(5) − ζ(2)ζ(3). (3.121)
32 8
On plugging (3.121) in (3.120), the solution is completed.
R1 ln(1−x) ln3 (1+x)

3.2.12 0 x
dx
Show that
ln(1 − x) ln3 (1 + x)
Z 1
1 1 3
dx = −6 Li5 − 6 ln(2) Li4 + ζ(5)
0 x 2 2 4
21 21 2 1
+ ζ(2)ζ(3) − ln (2)ζ(3) + ln3 (2)ζ(2) − ln5 (2). (3.122)
8 8 5
Solution By employing a b3 = 18 (a + b)4 − 18 (a − b)4 − a3 b, we have
ln(1 − x) ln3 (1 + x)
1
Z
dx
0 x

4 1−x
ln4 (1 − x2 ) 1 1 ln 1+x
1 Z 1 3
ln (1 − x) ln(1 + x)
Z Z
1
= dx − dx − dx
8 x 8 0 x x
|0 {z } 0
1−x2 =y

1
Z 1 4
ln (y) 1
Z 1 ln4 1−x
1+x
Z 1
ln3 (1 − x) ln(1 + x)
= dy − dx − dx.
16 0 1−y 8 0 x 0 x
These integrals are calculated in (3.5), (3.18), and (3.119).
R1 ln3 (1+x) ln(x)

3.2.13 0 x
dx
Show that
Z 1 3
ln (1 + x) ln(x) 1 1 99
dx = −12 Li5 − 12 ln(2) Li4 + ζ(5)
0 x 2 2 16
21 2 3 2 5
+3ζ(2)ζ(3) − ln (2)ζ(3) + 2 ln (2)ζ(2) − ln (2). (3.123)
4 5
1−t
Solution Put x = t then replace t by x,
Z 1 3
ln3 (1 + x) ln(x)
1
ln (x) ln( 1−x
x )
Z
dx = − dx
0 x 1
2
x(1 − x)
Z 1 4 Z 1 4 Z 1 3
ln (x) ln (x) ln (x) ln(1 − x)
= dx + dx − dx
1
2
x 1
2
1 − x 1
2
x
| {z }
IBP
1
ln3 (x) ln(1 − x)
Z
− dx
1 1−x
|2 {z }
1−x→x
!
4
1 1
ln4 (x)
Z Z
1 ln (x) 1 5 1
= ln5 (2) + dx − ln (2) + dx
5 1
2
1−x 4 4 1
2
1−x
Z 12 3
ln (1 − x) ln(x)
− dx
x
|0 {z }
R1 R1
0
− 1/2
1 4 Z 1 3
ln (1 − x) ln(x)
Z
1 5 3 ln (x)
=− ln (2) + dx − dx
20 4 1 1−x 0 x
|2 {z } | {z }
R1 R 1/2 1−x→x
0
− 0
ln3 (1 − x) ln(x)
1
Z
+ dx
1
2
x
Z 1
3 1 ln4 (x) 3 2 ln4 (x)
Z
1
= − ln5 (2) + dx − dx
20 4 0 1−x 4 0 1−x
Z 1 3 Z 1 3
ln (x) ln(1 − x) ln (1 − x) ln(x)
− dx + dx
0 1 − x 1
2
x
{recall the relation involving the last integral from (3.146)}
Z 1
3 1 ln4 (x) 1 2 ln4 (x)
Z
1 5 93
= ln (2) − ζ(5) + dx − dx
10 16 4 0 1−x 2 0 1−x
1 1 ln3 (x) ln(1 − x)
Z
− dx.
2 0 1−x
The first and second integrals are given in (3.5) and (3.35).
To get the latter, set a = 3 in (3.87),
∞
1
ln3 (x) ln(1 − x)
Z X Hn
dx = −6ζ(5) + 6
0 1−x n=1
n4
{this sum is given in (4.6)}
= 12ζ(5) − 6ζ(2)ζ(3). (3.124)
Gathering the three integrals completes the proof.

A different approach involving the derivative of the beta function may be found in
[27].
R1 ln(x) ln(1+x)
3.2.14 0 1−x
dx
Show that Z 1
ln(x) ln(1 + x) 3
dx = ζ(3) − ln(2)ζ(2). (3.125)
0 1−x 2
Solution By integration by parts,

1 1
ln(1 − x) ln(1 + x)
Z Z
Li2 (x)
dx = ln(2)ζ(2) + dx. (3.126)
0 1+x 0 x
R1
On the other hand, by using (1.111), we have Li2 (x) = 0
− x1−xu
ln(u)
du. Thus,
Z 1 Z 1 Z 1
Li2 (x) 1 x ln(u)
dx = − du dx
0 1+x 0 1+x 0 1 − xu
Z 1 Z 1
−x
= ln(u) dx du
0 0 (1 + x)(1 − ux)
{evaluate the inner integral by partial fraction decomposition}
Z 1
ln(2) ln(1 − u) ln(1 − u)
= ln(u) + − du
0 1+u u 1+u
Z 1 Z 1 Z 1
ln(u) ln(u) ln(1 − u) ln(u) ln(1 − u)
= ln(2) du + du − du
0 1 + u 0 u 0 1+u
| {z } | {z }
IBP IBP
Z 1 Z 1 2 Z 1
ln(u) 1 ln (u) ln(u) ln(1 + u)
= ln(2) du + du − du
0 1 + u 2 0 1 − u 0 1−u
Z 1
ln(1 − u) ln(1 + u)
+ du
0 u
Z 1 Z 1
1 ln(u) ln(1 + u) ln(1 − u) ln(1 + u)
= − ln(2)ζ(2) + ζ(3) − du + du,
2 0 1 − u 0 u
(3.127)
and the solution completes on combining (3.126) and (3.127).

Moreover, plugging (3.115) in (3.126) gives

Z 1
Li2 (x) 1
dx = ζ(3) − ln(2)ζ(2). (3.128)
0 1+x 2
R1 ln(x) ln(1−x)
3.2.15 0 1+x
dx
Show that
1
ln(x) ln(1 − x)
Z
13 3
dx = ζ(3) − ln(2)ζ(2). (3.129)
0 1+x 8 2

1
ln(x) ln(1 − x)
Z
dx
0 1+x
Z 1
1 ln(1 − x) ln(x)
= ln(1 + x) ln(x) ln(1 − x) 0 − ln(1 + x) − dx
| {z } 0 x 1−x
0
1 1
ln(1 − x) ln(1 + x)
Z Z
ln(x) ln(1 + x)
= dx − dx,
0 1−x 0 x
Group the results from (3.125) and (3.115) to end the solution.
R1 ln(x) ln2 (1−x) R1 ln2 (x) ln(1−x)

3.2.16 0 1+x
dx & 0 1+x
dx
Show that
ln(x) ln2 (1 − x)
Z 1
1 11 1
dx = −6 Li4 + ζ(4) − ln4 (2); (3.130)
0 1 + x 2 4 4
Z 1 2
ln (x) ln(1 − x) 1 1
dx = −4 Li4 + ζ(4) + ln2 (2)ζ(2) − ln4 (2).
0 1 + x 2 6
(3.131)
Solution Let
1
ln(x) ln2 (1 − x)
Z
P = dx;
0 1+x
1
ln2 (x) ln(1 − x)
Z
Q= dx.
0 1+x
First relation: Using the identity

1 1 1
ab2 − a2 b = (a − b)3 − a3 + b3 ,
3 3 3
with a = ln(x) and b = ln(1 − x), we have

Z 1 ln3 x
1 1 ln3 (x) 1 1 ln3 (1 − x)
Z Z
1 1−x
P −Q= dx − dx + dx
3 0 1+x 3 0 1+x 3 0 1+x
{gather the results from (3.42), (3.10), and (3.29)}

1 7 1 4
= −2 Li4 + ζ(4) − ln2 (2)ζ(2) − ln (2). (3.132)
2 4 12
Second relation: By integration by parts,

1 Z 1 2
ln(x) ln(1 − x) ln(1 + x) ln (1 − x) ln(1 + x)
Z
P =2 dx − dx . (3.133)
1−x x
|0 {z } | 0
{z }
Y Z
The integral Z is already given in (3.118).

For the integral Y , use 4ab = (a+b)2 −(a−b)2 with a = ln(1−x) and b = ln(1+x),

Z 1 2 2 Z 1 ln(x) ln2 1−x
ln(x) ln (1 − x ) 1+x
4Y = dx − dx .
1 − x 1 − x
|0 {z } |0 {z }
Y1 Y2
1 1+x 1 x
To evaluate Y1 , use 1−x = 1−x2 = 1−x2 + 1−x2 ,
1
ln(x) ln2 (1 − x2 ) x ln(x) ln2 (1 − x2 )
1
Z Z
Y1 = dx + dx
0 1 − x2 0 1 − x2
1
ln(x) ln2 (1 − x) 1 1 ln(x) ln2 (1 − x)
Z Z
x2 →x 1
= √ dx + dx
4 0 x(1 − x) 4 0 1−x
| {z }
1−x→x
1 2 1 2
ln(x) ln (1 − x) ln(1 − x) ln (x)
Z Z
1 1
= √ dx + dx
4 0 x(1 − x) 4 x
|0 {z }
IBP
1 2 1 3
ln(x) ln (1 − x)
Z Z
1 1 ln (x)
= √ dx + dx
4 0 x(1 − x) 12 0 1 − x
17
= − ζ(4) + 7 ln(2)ζ(3) − 3 ln2 (2)ζ(2).
4
1−x
For Y2 , let 1+x → x,

Z 1 ln(x) ln2 1−x
1+x
Y2 = dx
0 x(1 + x)

Z 1 ln 1−x
1+x ln2 (x) Z 1
ln(1 − x) ln2 (x)
Z 1
ln(1 + x) ln2 (x)
= dx − dx + dx
0 x 1+x 1+x
|0 {z } 0
Q

1 15 7 1
= 4 Li4 − ζ(4) + ln(2)ζ(3) − ln2 (2)ζ(2) + ln4 (2) − Q.
2 2 2 6
Grouping the results of Y1 and Y2 ,

1 13 7 1
4Y = −4 Li4 + ζ(4) + ln(2)ζ(3) − 2 ln2 (2)ζ(2) − ln4 (2) + Q.
2 4 2 6
Substitute the results of Y and Z in (3.133),

1 9 1
2P − Q = −8 Li4 + ζ(4) − ln2 (2)ζ(2) − ln4 (2). (3.134)
2 2 3
Combining (3.132) and (3.134) completes the solution.
R1 ln2 (1+x) R1 R1
3.2.17 0 1+x2
dx, 0 ln(1−x) ln(1+x)
1+x2
dx, & 0 ln(x)1+x
ln(1+x)
2 dx
Show that
Z 1 2
ln (1 + x) 7π 3 3π 2
2
dx = J Li3 (1 + i) − − ln (2) − 2 ln(2)G; (3.135)
0 1+x 64 16
Z 1
ln(1 − x) ln(1 + x) π3
2
dx = J Li3 (1 + i) − − ln(2)G; (3.136)
0 1+x 32
Z 1
ln(x) ln(1 + x) 5π 3 3π 2
2
dx = 3 J Li3 (1 + i) − − ln (2) − 2 ln(2)G.
0 1 + x 64 16
(3.137)
Solution
1−x
First relation: By the substitution x → 1+x ,

Z 1 2
ln (x)
Z 1 ln2 1−x
1+x
dx = dx
0 1 + x2 0 1 + x2
1
ln2 (1 − x) 1
ln(1 − x) ln(1 + x) 1
ln2 (1 + x)
Z Z Z
= dx − 2 dx + dx.
0 1 + x2 0 1 + x2 0 1 + x2
Group the results from (1.94) and (3.51),

1
ln2 (1 + x) 1
ln(1 − x) ln(1 + x)
Z Z
dx − 2 dx
0 1 + x2 1 + x2 0
3π 3 3π 2
= J Li3 (1 + i) − − ln (2). (3.138)
64 16
Second relation: Using the same substitution,

1 Z 1 ln 1−x ln 2x
ln(x) ln(1 − x)
Z
1+x 1+x
dx = dx
0 1 + x2 0 1 + x2
1 Z 1 2 Z 1
ln(x) ln(1 − x) ln(1 − x) ln(1 + x)
Z
ln (1 + x)
= dx + dx − dx
0 1 + x2 0 1 + x2 0 1 + x2

Z 1 Z 1 ln 1−x
ln(x) ln(1 + x) 1+x
− dx + ln(2) dx.
0 1 + x2 0 1 + x2
R1 ln(x) ln(1−x)
Notice that 0 1+x2 dx cancels out from both sides.
1−x
For the last integral, let x → 1+x ,

1−x
Z 1 ln 1+x
Z 1
ln(x)
dx = dx = −G,
0 1+ x2 0 1 + x2
where the last step is given in (1.207). Substitute this integral back,
1
ln2 (1 + x) 1
ln(1 − x) ln(1 + x) 1
Z Z Z
ln(x) ln(1 + x)
dx − dx − dx
0 1 + x2 0 1 + x2 0 1 + x2
= ln(2)G. (3.139)
Third relation:
Z ∞ Z ∞ 2
1
ln2 (1 + x)
Z
ln (1 + x)
dx = − dx
0 1 + x2 0 1 1 + x2
Z ∞ 2 Z ∞ 2
ln (1 + x) ln (1 + x)
= 2
dx − dx
0 1+x 1 + x2
|1 {z }
x→1/x
∞ 2 1
ln2 1+x
Z Z
ln (1 + x) x
= dx − dx
0 1 + x2 0 1 + x2

2 1+x
expand ln
x
∞
ln2 (1 + x) ln2 (1 + x)
1 1
Z Z Z
ln(x) ln(1 + x)
= dx − dx + 2 dx
0 1 + x2 0 1 + x2 0 1 + x2
Z 1 2
ln (x)
− dx.
0 1 + x2
Collect the results from (3.47) and (1.94),

Z 1 2 Z 1
ln (1 + x) ln(x) ln(1 + x) π3
2
dx − 2
dx = J Li3 (1 + i) − . (3.140)
0 1+x 0 1+x 32
Combine the three relations in (3.138), (3.139), and (3.140) to finalize the solution.
R1 ln(x) ln2 (1−x)

3.2.18 0
√
x(1−x)
dx
Show that
ln(x) ln2 (1 − x)
Z 1
√ dx = −15ζ(4) +28 ln(2)ζ(3)−12 ln2 (2)ζ(2). (3.141)
0 x(1 − x)
√
Solution Make the change of variable 1 − x = y,
ln(x) ln2 (1 − x)
1
ln(y) ln2 (1 − y 2 )
Z Z 1
√ dx = 4 dy
0 x(1 − x) 1 − y2
0
Z 1 ln 1−x ln2 4x
y= 1−x
1+x 1+x (1+x)2
= 2 dx
0 x

Z 1 ln 1−x Z 1 ln 1−x ln(x)
1+x 1+x
= 8 ln2 (2) dx + 8 ln(2) dx
0 x 0 x
ln(x) ln2 (1 + x)
Z 1 Z 1
ln(x) ln(1 − x) ln(1 + x)
−8 dx + 8 dx
0 x 0 x
Z 1 Z 1 2
ln(1 − x) ln(1 + x) ln (1 + x)
−16 ln(2) dx + 16 ln(2) dx
0 x 0 x
ln(1 − x) ln2 (1 + x)
Z 1 Z 1 3
ln (1 + x)
+8 dx − 8 dx.
0 x 0 x
All these integrals are given in (3.21), (3.22), (3.116), (3.115), (3.38), (3.117), and
(3.39). Regarding the fourth integral, integrate it by parts,
1
ln(x) ln2 (1 + x) 1
ln2 (x) ln(1 + x)
Z Z
dx = − dx,
0 x 0 1+x
which is given in (3.143).

A different method is by differentiating the beta function in (1.47):
1
ln(x) ln2 (1 − x) ∂ 3 Γ(a)Γ(b)
Z
√ dx = lim ,
0 x(1 − x) a→1/2 ∂a∂b2 Γ(a + b)
b→0
but lengthy calculations will be involved and we had better let Mathematica do it.
∂ 3 Γ(a)Γ(b)
a→1/2 ∂a∂b2 Γ(a + b)
b→0
,{a,1/2,0},{b,0,0}]]//FullSimplify//Expand
R1 ln2 (x) ln(1−x)

3.2.19 0
2
1−x
dx
Show that
1
ln2 (x) ln(1 − x)
Z 2 1 1
dx = − ln4 (2) − ζ(4). (3.142)
0 1−x 4 4

1
ln2 (x) ln(1 − x)
Z 2
dx
0 1−x
Z 12 2
1 1 ln (1 − x) ln(x)
= − ln2 (1 − x) ln(x) 02 + dx
2 x
|0 {z }
1−x→x
1
ln2 (x) ln(1 − x)
Z
1
= − ln4 (2) + dx
2 1
2
1−x
( 1
)
2
ln (x) ln(1 − x)
Z 2
0 1−x
1 1 ln2 (x) ln(1 − x)
Z
1
= − ln4 (2) + dx
4 2 0 1−x
{set a = 2 in (3.87) to get the integral}
∞
1 X Hn
= − ln4 (2) + ζ(4) −
2 n=1
n3
1 1
= − ln4 (2) − ζ(4).
4 4
R1 ln2 (x) ln(1+x)

3.2.20 0 1+x
dx
Show that
Z 1 2
ln (x) ln(1 + x) 1 15 7
dx = 4 Li4 − ζ(4) + ln(2)ζ(3) − ln2 (2)ζ(2)
0 1 + x 2 4 2
1 4
+ ln (2). (3.143)
6
x
Solution By making the change of variable y = 1+x ,

Z 1 2
ln (x) ln(1 + x)
Z 1
2 ln2
ln(1 − y)y
1−y
dx = − dy
0 1+x 0 1−y
1 Z 12 2 Z 21 3
ln(y) ln2 (1 − y) ln (y) ln(1 − y) ln (1 − y)
Z 2
=2 dy − dy − dy .
1−y 1−y 1−y
|0 {z } |0 {z } |0 {z }
I1 I2 I3
For I1 , integrate by parts,

1
ln3 (1 − y) 1 1 ln3 (x)
Z Z
1 1 2 1
I3 = − ln4 (2) + dy = − ln4 (2) + dx
3 3 y 3 3 12 1 − x
|0 {z } | {z }
1−y=x R1 R 1/2
0
− 0
Z 1
1 1 ln3 (x) 1 2 ln3 (x)
Z
1
= − ln4 (2) + dx − dx
3 3 0 1−x 3 0 1−x
{gather the results from (3.4) and (3.142)}

1 7 1 1
= 2 Li4 − 2ζ(4) + ln(2)ζ(3) − ln2 (2)ζ(2) − ln4 (2).
2 4 2 6
The value of I2 is given in (3.142), and I3 = − 14 ln4 (2).

The solution finalizes on combining the three integrals.
R1 ln3 (1−x) ln(x)

3.2.21 1
x
dx
2
Show that
Z 1 3
ln (1 − x) ln(x) 1 1 3
dx = 6 Li5 + 6 ln(2) Li4 + ζ(5)
1
2
x 2 2 16
21 2 9 5
−3ζ(2)ζ(3) + ln (2)ζ(3) − ln3 (2)ζ(2) + ln (2). (3.144)
8 20
Solution First, we denote the integral by I. Let a = ln(1 − x) and b = ln(x) in

the algebraic identity
4a3 b = a4 + b4 − (a − b)4 − 4ab3 + 6a2 b2 ,
ln4 (1 − x)
1 Z 1 4 Z 1
1 4 1−x
Z
ln (x)
4I = dx + dx − ln dx
1 x 1 x 1 x x
|2 {z } | 2
{z } | 2
{z }
1−x→x 1
ln5 (2) (1−x)/x=y
5
3
1
ln(1 − x) ln (x) 1
ln (1 − x) ln2 (x)
2
Z Z
−4 dx + 6 dx
1 x 1 x
|2 {z } | 2
{z }
IBP J
1
Z 4 Z 1 4 Z 1 4
2 ln (x) ln (x) ln (y) 4
= dx − dx − dy − ln5 (2) + J
0 1−x 1
2
1−x 0 1+y 5
| {z }
R1 R 1/2
0
− 0
1
Z 4 Z Z 1 4
1 4
2 ln (x) ln (x) ln (y) 4
=2 dx − dx − dy − ln5 (2) + J
0 1−x 0 1 − x 0 1 + y 5
Z 12 4
93 4 5 ln (x)
= − ζ(5) − ln (2) + 2 dx + J. (3.145)
2 5 0 1−x
Let’s compute J:
1
ln2 (1 − x) ln2 (x) 1
ln3 (x) ln(1 − x)
Z Z
IBP
J =6 dx = 2 ln5 (2) + 4 dx
1
2
x 1 1−x
|2 {z }
1−x→x
1
3
ln (1 − x) ln(x)
Z 2
= 2 ln5 (2) + 4 dx
x
|0 {z }
R1 R1
0
− 1/2
1 3 Z 1 3
ln (1 − x) ln(x) ln (1 − x) ln(x)
Z
5
= 2 ln (2) + 4 dx −4 dx
x 1 x
|0 {z } | 2
{z }
1−x→x I
1 3
ln (x) ln(1 − x)
Z
= 2 ln5 (2) + 4 dx − 4I.
0 1−x
Plugging the result of J in (3.145),

1
ln4 (x) 1
ln3 (x) ln(1 − x)
Z Z
3 5 93 1 2 1
I= ln (2) − ζ(5) + dx + dx.
20 16 4 0 1−x 2 0 1−x
(3.146)
Gather the results from (3.124) and (3.35) to finish the solution.
R1 Li2 (−x)
3.2.22 0 1+x2
dx
Show that
Z 1
Li2 (−x) 7π 3 3π 2 3
2
dx = + ln (2) + ln(2)G − 3 J Li3 (1 + i). (3.147)
0 1+x 96 16 2
R1 x ln(y)
Solution Write Li2 (−x) = 0 1+xy dy, which follows from replacing x by −x in
(1.111),
Z 1 Z 1 Z 1
Li2 (−x) 1 x ln(y)
dx = dy dx
0 1 + x2 0 1 + x2 0 1 + xy
Z 1Z 1
x
= ln(y) 2
dx dy
0 0 (1 + x )(1 + xy)
{evaluate the inner integral by partial fraction decomposition}
Z 1
π y ln(2) 1 ln(1 + y)
= ln(y) + − dy
0 4 1 + y2 2 1 + y2 1 + y2
π 1 y ln(y) ln(2) 1 ln(y)
Z Z Z 1
ln(y) ln(1 + y)
= 2
dy + 2
dy − dy
4 0 1+y 2 0 1+y 0 1 + y2
| {z }
√
y= x
Z 1 Z 1 Z 1
π ln(x) ln(2) ln(y) ln(y) ln(1 + y)
= dx + dy − dy. (3.148)
16 0 1+x 2 0 1 + y2 0 1 + y2
and the solution finalizes on recalling the results from (3.8), (1.207), and (3.137).
R1 ln(x) arctan x
3.2.23 0 1+x
dx
Show that
1
π3
Z
ln(x) arctan x 1
dx = ln(2)G − . (3.149)
0 1+x 2 64

Z 1 Z 1
ln(x) arctan x
dx = d(Li2 (−x) + ln(x) ln(1 + x)) arctan x
0 1+x 0
Z 1
1 Li2 (−x) + ln(x) ln(1 + x)
= Li2 (−x) + ln(x) ln(1 + x)) arctan x − dx
0
0 1 + x2
Z 1 1
π3
Z
Li2 (−x) ln(x) ln(1 + x)
=− − 2
dx − dx
48 0 1 + x 0 1 + x2
{recall the relation involving the first integral from (3.148)}
Z 1
π3 ln(2) 1 ln(y)
Z
π ln(x)
=− − dx − 2
dy.
48 16 0 1 + x 2 0 1+y
The remaining integrals are given in (3.8) and (1.207).
R1 ln2 (x) arctan x

3.2.24 0 x(1+x2 )
dx
Show that
1
ln2 (x) arctan x π3
Z
7π
dx = ln(2) − ζ(3) + β(4). (3.150)
0 x(1 + x2 ) 16 64

Solution Let I denote our integral,
∞ ∞
ln2 (x) arctan x ln2 (x) arctan x
Z Z
I= dx − dx
0 x(1 + x2 ) x(1 + x2 )
|1 {z }
x→1/x
Z ∞ 2 Z 1 2
ln (x) arctan x x ln (x) arctan(1/x)
= 2
dx − dx
0 x(1 + x ) 0 1 + x2
n π o
write arctan(1/x) = − arctan x
2
Z ∞ 2
π 1 x ln2 (x) x ln2 (x) arctan x
Z Z 1
ln (x) arctan x
= dx − dx + dx
0 x(1 + x2 ) 2 0 1 + x2 0 1 + x2
let x2 → x in the second integral

x 1 1
and write = − in the third one
1 + x2 x x(1 + x2 )
Z ∞ 2 Z 1 2 Z 1 2
ln (x) arctan x π ln (x) ln (x) arctan x
= 2
dx − dx + dx − I
0 x(1 + x ) 16 0 1 + x 0 x
{add I to both sides and integrate the third integral by parts then divide by 2}
∞
ln2 (x) arctan x
Z 1 2
1 1 ln3 (x)
Z Z
1 π ln (x)
= dx − dx − dx
2 0 x(1 + x2 ) 32 0 1 + x 6 0 1 + x2
1 ∞ ln2 (x) arctan x
Z
3π
= dx − ζ(3) + β(4). (3.151)
2 0 x(1 + x2 ) 64
R1 x
For the remaining integral, write arctan x = 0 1+x2 y 2
dy,
∞ Z ∞
ln2 (x) arctan x ln2 (x)
Z Z 1
x
dx = dy dx
0 x(1 + x2 ) 0 x(1 + x2 ) 0 1+x y
2 2
y2

1 1 1
write = −
(1 + x2 )(1 + x2 y 2 ) 1 − y 2 1 + x2 1 + x2 y 2
{then change the order of integration}
 
Z 1 Z ∞ 2 Z ∞ 2 2
1  ln (x) y ln (x) 
= dx − dx dy
 
1 − y 2  0 1 + x2 1 + x 2 y2

0
} |0

| {z {z }
x=t xy=t
∞ 2 ∞
1
y ln2 (t/y)
Z Z Z
1 ln (t)
= dt − dt dy
0 1 − y2 0 1 + t2 0 1 + t2
{write ln2 (t/y) = ln2 (t) − 2 ln(y) ln(t) + ln2 (y)}
Z 1 Z ∞ 2 Z ∞
1 ln (t) y ln(y) ln(t)
= 2
(1 − y) 2
dt + 2 dt
0 1 − y 0 1 + t 0 1 + t2
Z ∞
y ln2 (y)

− dt dy
0 1 + t2
{recall the result of the first integral from (3.54)}
{and note that the second integral is 0 by using (3.58)}
πy ln2 (y)
Z 1
π3

1
= 2
(1 − y) − dy
0 1−y 8 2
π3 1 1 − y π 1 y ln2 (y)
Z Z
= dy − dy
8 0 1 − y2 2 0 1 − y2
| {z }
y 2 =x
π3 1
ln2 (x) 1
Z Z
dy π
= − dx
8 0 1 + y 16 0 1 − x
π3 π
= ln(2) − ζ(3).
8 8
Plugging this result in (3.151) completes the solution.
R1 Li2
2 (−x)
3.2.25 0 x
dx
Show that
1
Li22 (−x)
Z
3 17
dx = ζ(2)ζ(3) − ζ(5). (3.152)
0 x 4 16
Solution Bu using (1.115):

1
Li2 (−x) = Li2 (x2 ) − Li2 (x),
2
we get
1 Z 1 2
Li22 (−x)
Z
1 1 2
dx = Li2 (x ) − Li2 (x) dx
0 x 0 x 2
1 1 Li22 (x2 )
Z 1 Z 1 2
Li2 (x2 ) Li2 (x)
Z
Li2 (x)
= dx − dx + dx
4 0 x 0 x 0 x
| {z }
x2 →x
1 1 Z 1 2
Li22 (x) Li2 (x2 ) Li2 (x)
Z Z
1 Li2 (x)
= dx − dx + dx
80 x 0 x 0 x
9 1 Li22 (x)
Z 1
Li2 (x2 ) Li2 (x)
Z
= dx − dx
8 0 x 0 x
expand Li2 (x) and Li2 (x2 ) in series

∞ Z 1 ∞ Z 1
9X 1 n−1
X 1
= x Li2 (x)dx − x2n−1 Li2 (x)dx
8 n=1 n2 0 n=1
n 2
0
{recall the result in (3.103) for both integrals}

∞ X ∞
9 X 1 ζ(2) Hn 1 ζ(2) H2n
= − 2 − −
8 n=1 n2 n n n=1
n2 2n (2n)2
∞ ∞
9 9 X Hn 1 X H2n
= ζ(2)ζ(3) − − ζ(2)ζ(3) + 4
8 8 n=1 n4 2 n=1
(2n) 4
{make use of (1.5) for the latter sum}

∞ ∞ ∞
!
5 9 X Hn 1 X Hn 1 X (−1)n Hn
= ζ(2)ζ(3) − +4 +
8 8 n=1 n4 2 n=1 n4 2 n=1 n4
∞ ∞
5 7 X Hn X (−1)n Hn
= ζ(2)ζ(3) + + 2 . (3.153)
8 8 n=1 n4 n=1
n4
R1 Li2
2 (−x)
3.2.26 0
2
x
dx
This following integral is proposed by Cornel Vălean (see[34]):

1
Li22 (x)
Z
2 1 1 27 5
dx = −2 Li5 − 2 ln(2) Li4 + ζ(5) − ln(2)ζ(4)
0 x 2 2 32 8
7 7 1 7 5
+ ζ(2)ζ(3) − ln2 (2)ζ(3) + ln3 (2)ζ(2) − ln (2). (3.154)
8 8 2 60
Solution By applying integration by parts twice, we have

1 Z 12
Li22 (x) Li3 (x) ln(1 − x)
Z 2 1
dx = Li3 (x) Li2 (x) 02 + dx
0 x 0 x
Z 12
1 1 1 Li4 (x)
= Li3 Li2 + Li4 (x) ln(1 − x) 0 +
2
dx
2 2 0 1−x
Z 1
1 1 1 2 Li (x)
4
= Li3 Li2 − ln(2) Li4 + dx
2 2 2 0 1 −x
∞
( )
Li4 (x) X (4) n−1
use = Hn−1 x given in (2.3) for the latter integral
1−x n=1
X ∞ Z 21
1 1 1 (4)
= Li3 Li2 − ln(2) Li4 + Hn−1 xn−1 dx
2 2 2 n=1 0
( Z 21 )
n−1 1 (4) (4) 1
write x dx = n and Hn−1 = Hn − 4
0 n2 n
∞ (4)
Hn − n14
X
1 1 1
= Li3 Li2 − ln(2) Li4 +
2 2 2 n=1
n2n
X ∞ (4)
1 1 1 1 Hn
= Li3 Li2 − ln(2) Li4 − Li5 +
2 2 2 2 n=1
n2n
{set a = 3 in (3.91) to get the integral representation of the sum}

1 1 1 1
= Li3 Li2 − ln(2) Li4 − Li5
2 2 2 2
Z 1 3
1 ln (1 − x) ln(1 + x)
− dx,
6 0 x
and the solution completes on collectingthe result from (3.119) along with writing
Li2 21 = 21 ζ(2) − 21 ln2 (2) and Li3 12 = 78 ζ(3) − 12 ln(2)ζ(2) + 16 ln3 (2) given
in (1.120) and (1.132).
R1 ln2 (1−x) Li2 (x)

3.2.27 0 x
dx
Show that
1
ln2 (1 − x) Li2 (x)
Z
dx = 2ζ(2)ζ(3) − ζ(5). (3.155)
0 x
Solution Set a = 2 in (3.93),

∞ ∞ (2)
1
ln2 (1 − x) Li2 (x)
Z X Hn X Hn
dx = 8ζ(5) + 2ζ(2)ζ(3) − 6 − 2 .
0 x n=1
n4 n=1
n3
Putting together the results from (4.6) and (4.102) completes the solution.
R1 ln3 (1−x) Li2 (x)

3.2.28 0 x
dx
Show that
1
ln3 (1 − x) Li2 (x)
Z
1
dx = − ζ(6) − 6ζ 2 (3). (3.156)
0 x 2

∞ ∞ (2)
1
ln3 (1 − x) Li2 (x)
Z
81 X Hn X Hn
dx = − ζ(6) + 24 + 6 .
0 x 2 n=1
n5 n=1
n4
Collecting the results from (4.7) and (4.114) finalizes the solution.
R1 ln4 (1−x) Li2 (x)

3.2.29 0 x
dx
Show that
Z 1 4
ln (1 − x) Li2 (x)
dx = 24ζ(2)ζ(5) + 72ζ(3)ζ(4) − 96ζ(7). (3.157)
0 x
ln4 (1 − x) Li2 (x)

Z 1
dx
0 x
∞ ∞ (2)
X Hn X Hn
= 24ζ(2)ζ(5) + 144ζ(7) − 120 − 24 ,
n=1
n6 n=1
n5
and solution finishes on gathering the results from (4.127) and (4.8).
Chapter 4
Harmonic Series
4.1 Generalized Harmonic Series

P∞ Hn
p
4.1.1 n=1 nq
For p, q ∈ Z+ , q ≥ 2, the following identity holds:

∞ ∞ q−2
X H np X Hpn X
= (−1)q p − (−p)−j ζ(q − j)ζ(j + 1). (4.1)
n=1
nq n=1
(pn) q
j=1
Proof.
∞ ∞
!
X H np 1X 1 H np
= n
n=1
nq p n=1 nq−1 p
{replace n by n/p in the identity (2.70)}

∞ Z 1
1X 1 n
p −1 ln(1 − x)dx
= − x
p n=1 nq−1 0
∞
Z 1 n
!
1 ln(1 − x) X x p
=− dx
p 0 x n=1
nq−1
1 1 ln(1 − x)
Z
1
=− Liq−1 (x p ) dx
p 0 x
Z 1
x=y p ln(1 − y p ) Liq−1 (y)
= − dy
0 y
{expand ln(1 − y p ) in series}
189
190 Chapter 4. Harmonic Series
∞
1 1 pn−1
X Z
= y Liq−1 (y)dy
n=1
n 0
 
∞ q−2
X 1 H pn
X ζ(q − j)
= (−1)q q−1
− (−1)j j

n=1
n (pn) j=1
(pn)
∞ q−2 ∞
!
q
X Hpn X
−j
X 1
= (−1) p q
− (−p) ζ(q − j) j+1
n=1
(pn) j=1 n=1
n
∞ q−2
X Hpn X
= (−1)q p q
− (−p)−j ζ(q − j)ζ(j + 1),
n=1
(pn) j=1
P∞ Hn
4.1.2 n=1 nq
For integer q ≥ 2, the generalized Euler sum holds:

∞ q−2
X Hn q+2 1X
= ζ(q + 1) − ζ(q − j)ζ(j + 1). (4.2)
n=1
nq 2 2 j=1

Proof. Start with expanding ζ(q − j) and ζ(j + 1) in series,
∞
! ∞ !
X 1 X 1
ζ(q − j)ζ(j + 1) = .
m=1
mq−j n=1
nj+1
Next, take the summation for both sides from j = 1 to q − 2,

q−2 q−2 ∞
! ∞ !
X X X 1 X 1
S= ζ(q − j)ζ(j + 1) =
j=1 j=1 m=1
mq−j n=1
nj+1
{change the order of summations}
∞ X q−2
∞ X
X 1
=
m=1 n=1 j=1
mq−j nj+1
{break up the middle sum}
∞ m−1 ∞
! q−2
X X X X 1
= + an=m +
m=1 n=1 n=m+1 j=1
mq−j nj+1
4.1. Generalized Harmonic Series 191
{pull out the terms for n = m}

q−2
∞ X ∞ m−1 ∞
! q−2
X 1 X X X X 1
= q+1
+ + q−j nj+1
m=1 j=1
m m=1 n=1 n=m+1 j=1
m
 
q−2 ∞ ∞ ∞ q−2
! m−1
!
X X 1 X X X 1 X mj 
= + +
j=1 m=1
mq+1 m=1 n=1 n=m+1
mq n j=1 nj
q−2 ∞ ∞
m−1
!
X X X X 1 1
= ζ(q + 1) + + q−1 (n − m)
− q−1 (n − m)
j=1 m=1 n=1 n=m+1
nm mn
∞ m−1
X X 1 1
= (q − 2)ζ(q + 1) + −
m=1 n=1
nmq−1 (n − m) mnq−1 (n − m)
∞ ∞
X X 1 1
+ q−1 (n − m)
− q−1 (n − m)
. (4.3)
m=1 n=m+1
nm mn
∞ m−1
X X ∞
X ∞
X
By using the f (m, n) = f (m, n) given in (1.14), the first double
m=1 n=1 n=1 m=n+1
sum in (4.3) becomes
∞ m−1
X X 1 1
−
m=1 n=1
nmq−1 (n − m) mnq−1 (n − m)
∞ ∞
X X 1 1
= q−1 (n − m)
− q−1 (n − m)
n=1 m=n+1
nm mn
{swab the variables n and m}
∞ ∞
X X 1 1
= −
m=1 n=m+1
mnq−1 (m − n) nmq−1 (m − n)
{write m − n = −(n − m) in both terms}
∞ ∞
X X 1 1
= q−1 (n − m)
− q−1 (n − m)
.
m=1 n=m+1
nm mn
Plug this sum back in (4.3),

∞ ∞
X X 1 1
S = (q − 2)ζ(q + 1) + 2 q−1 (n − m)
− q−1 (n − m)
m=1 n=m+1
nm mn
{shift the index n by +m}
∞ X ∞
X 1 1
= (q − 2)ζ(q + 1) + 2 −
m=1 n=1
(n + m)m n m(n + m)q−1 n
q−1
∞ X
∞ ∞ X∞
X 1 X 1
= (q − 2)ζ(q + 1) + 2 q−1 n
−2 .
m=1 n=1
(n + m)m m=1 n=1
m(n + m)q−1 n
| {z } | {z }
S1 S2
For S1 ,
∞ X
∞ ∞ ∞
!
X 1 X 1 X m
S1 = q−1 n
= q
m=1 n=1
(n + m)m m=1
m n=1
n(n + m)
{recall the definition of Hm in (1.155)}
∞
X Hm
= .
m=1
mq
For S2 ,
∞
∞ X
X 1
S2 =
m=1 n=1
m(n + m)q−1 n

n+m
multiply by
n+m
∞ X ∞
X n+m
=
m=1 n=1
m(n + m)q n
∞
∞ X ∞ X ∞
X 1 X 1
= q
+
m=1 n=1
m(n + m) m=1 n=1
(n + m)q n
{swap the variables m and n in the first double sum}
{and change the order of summations in the second double sum}
∞ X ∞
X 1
=2
n=1 m=1
n(n + m)q
{shift the index m by −n}
∞ ∞
X X 1
=2 q
n=1 m=n+1
nm
∞ ∞
( )
X X
use f (m) = f (m) − f (n) for the inner sum
m=n+1 m=n
∞ ∞ ∞ X
∞ ∞
!
X X 1 1 X 1 X 1
=2 q
− q+1 =2 q
− 2 q+1
n=1 m=n
nm n n=1 m=n
nm n=1
n
∞ X
∞ ∞ X
( m
)
X X
use f (n, m) = f (n, m) given in (1.13) for the first term
n=1 m=n m=1 n=1
∞ X ∞
m m
!
X 1 X 1 X 1
=2 q
− 2ζ(q + 1) = 2 q
− 2ζ(q + 1)
m=1 n=1
nm m=1
m n=1
n
∞
X Hm
=2 − 2ζ(q + 1).
m=1
mq
Combine S1 and S2 , we have

q−2 ∞
X X Hm
S= ζ(q − j)ζ(j + 1) = (q + 2)ζ(q + 1) − 2 .
j=1 m=1
mq
Reorganize the terms then divide by 2 to finish the proof.
Examples
∞
X Hn
= 2ζ(3); (4.4)
n=1
n2
∞
X Hn 5
3
= ζ(4); (4.5)
n=1
n 4
∞
X Hn
= 3ζ(5) − ζ(2)ζ(3); (4.6)
n=1
n4
∞
X Hn 7 1
5
= ζ(6) − ζ 2 (3); (4.7)
n=1
n 4 2
∞
X Hn
= 4ζ(7) − ζ(2)ζ(5) − ζ(3)ζ(4); (4.8)
n=1
n6
∞
X Hn
= 5ζ(9) − ζ(2)ζ(7) − ζ(3)ζ(6) − ζ(4)ζ(5). (4.9)
n=1
n8
Further, the following identity holds for only odd q ≥ 3:
∞ q−2
X Hn 1X
q
= − (−1)i ζ(q − i)ζ(i + 1). (4.10)
n=1
n 2 i=1
Proof. Replace a by q − 1 in (3.102),
1 q−2
ζ(q − i)
Z
n−1 qHn X
x Liq−1 (x)dx = (−1) q−1 − (−1)i .
0 n i=1
ni
Divide both sides by n then consider the summation over n ≥ 1,

∞ q−2 ∞
X Hn X X 1
(−1)q q
− (−1) i
ζ(q − i) i+1
n=1
n i=1 n=1
n
∞ q−2
X
q Hn X
= (−1) − (−1)i ζ(q − i)ζ(i + 1)
n=1
nq i=1
∞
!
1 Z 1
Liq−1 (x) X xn Liq−1 (x) ln(1 − x)
Z
= dx = − dx
0 x n=1
n 0 x
{expand Liq−1 (x) in series}
∞ Z 1
X 1
=− q−1
xn−1 ln(1 − x)dx
n=1
n 0
{make use of the result in (2.70)}

∞ X ∞
X 1 Hn Hn
=− q−1
− = .
n=1
n n n=1
nq
Reorder the terms,

∞ q−2
X Hn q
X
q
(1 − (−1) ) = − (−1)i ζ(q − i)ζ(i + 1).
n=1
n i=1
For even q, the LHS sum becomes zero. For odd q, we have
∞ q−2
X Hn X
q
(2) = − (−1)i ζ(q − i)ζ(i + 1).
n=1
n i=1
Divide through by 2 to finish the proof.
P∞ Hn
4.1.3 n=1 nq
For integer q ≥ 2, the following identity holds:

∞
X Hn 1−q

−q q
= (2 − 2 ) ln(2)ζ(q) + 1 − 2 − ζ(q + 1)
n=1
nq 2
q−2
1X
+ (1 − 21−q+j )(1 − 2−j )ζ(q − j)ζ(j + 1). (4.11)
2 j=1
Proof. We follow Rob Johnson’s approach in (4.2), but here we begin with expanding
η(q − j) and η(j + 1) in series,
q−2 q−2 ∞
! ∞ !
X X X (−1)m−1 X (−1)n−1
S= η(q − j)η(j + 1) =
j=1 j=1 m=1
mq−j n=1
nj+1
{change the order of summations}
∞ X ∞ Xq−2
X (−1)m+n
=
m=1 n=1 j=1
mq−j nj+1
{break up the middle sum}
∞ m−1 ∞
! q−2
X X X X (−1)m+n
= + an=m +
m=1 n=1 n=m+1 j=1
mq−j nj+1
{pull out the terms for n = m}
q−2
∞ X ∞ m−1 ∞
! q−2
X 1 X X X X (−1)m+n
= q+1
+ +
m=1 j=1
m m=1 n=1 n=m+1 j=1
mq−j nj+1
 
q−2 ∞ ∞ ∞ q−2
! m−1
!
X X 1 X X X (−1)m+n X mj 
= + +
j=1 m=1
mq+1 m=1 n=1 n=m+1
mq n j=1
nj
q−2 ∞ ∞
m−1
!
X X X X (−1)m+n (−1)m+n
= ζ(q + 1) + + −
j=1 m=1 n=1 n=m+1
nm (n − m) mnq−1 (n − m)
q−1
∞ m−1
X X (−1)m+n (−1)m+n
= (q − 2)ζ(q + 1) + −
m=1 n=1
nm (n − m) mnq−1 (n − m)
q−1
∞ ∞
X X (−1)m+n (−1)m+n
+ − .
m=1 n=m+1
nmq−1 (n − m) mnq−1 (n − m)
∞ m−1
X X ∞
X ∞
X
By using f (m, n) = f (m, n) given in (1.14), the first double
m=1 n=1 n=1 m=n+1
sum in the latter equality becomes:
∞ m−1
X X (−1)m+n (−1)m+n
−
m=1 n=1
nm (n − m) mnq−1 (n − m)
q−1
∞ ∞
X X (−1)m+n (−1)m+n
= −
n=1 m=n+1
nm (n − m) mnq−1 (n − m)
q−1
{swab the variables n and m}

∞ ∞
X X (−1)n+m (−1)n+m
= −
m=1 n=m+1
mn (m − n) nmq−1 (m − n)
q−1
∞ ∞
X X (−1)n+m (−1)n+m
= − .
m=1 n=m+1
nmq−1 (n − m) mnq−1 (n − m)
Thus,
∞ ∞
X X (−1)n+m (−1)n+m
S = (q − 2)ζ(q + 1) + 2 −
m=1 n=m+1
nmq−1 (n − m) mnq−1 (n − m)
{shift the index n by +m}
∞ X ∞
X (−1)n (−1)n
= (q − 2)ζ(q + 1) + 2 −
m=1 n=1
(n + m)m n m(n + m)q−2 n
q−2
∞
∞ X ∞ X∞
X (−1)n X (−1)n
= (q − 2)ζ(q + 1) + 2 −2 .
m=1 n=1
(n + m)mq−1 n m=1 n=1
m(n + m)q−1 n
| {z } | {z }
S1 S2
For S1 ,
∞ X
∞ ∞ ∞
!
X (−1)n X 1 X (−1)n m
S1 = q−1 n
= q
m=1 n=1
(n + m)m m=1
m n=1
n(n + m)
∞ ∞ ∞
!
X 1 X (−1)n X (−1)n
= −
m=1
mq n=1
n n=1
n+m
{recall the result of the second sum from (1.166)}
∞
X 1
− ln(2) − (−1)m H m − ln(2)

= q
m=1
m
∞
X (−1)m H m
= − ln(2)ζ(q) − ln(2)η(q) − .
m=1
mq
For S2 ,
∞ X
∞
X (−1)n
S2 =
m=1 n=1
m(n + m)q−1 n

n+m
multiply by
n+m
∞ X ∞
X (−1)n (n + m)
=
m=1 n=1
m(n + m)q n
∞ X
∞ ∞ X ∞
X (−1)n X (−1)n
= q
+
m=1 n=1
m(n + m) m=1 n=1
(n + m)q n
{swap the variables m and n in the first double sum}
{and change the order of summations in the second double sum}
∞ X ∞
X (−1)m + (−1)n
=
n=1 m=1
n(n + m)q
{shift the index m by −n}
∞ ∞
X X (−1)m−n + (−1)n
=
n=1 m=n+1
nmq
∞ ∞
( )
X X
use f (m) = f (m) − f (n)
m=n+1 m=n
∞ ∞
!
X X (−1)m−n + (−1)n 1 + (−1)n
= q
−
n=1 m=n
nm nq+1
∞ X ∞ ∞
X (−1)m−n + (−1)n X 1 + (−1)n
= −
n=1 m=n
nmq n=1
nq+1
∞ X
∞ ∞ X
( m
)
X X
use f (n, m) = f (n, m) given in (1.13) for the first term
n=1 m=n m=1 n=1
∞ X m
X (−1)m−n + (−1)n
= − ζ(q + 1) + η(q + 1)
m=1 n=1
nmq
∞ m
!
X 1 X (−1)m−n + (−1)n
= − ζ(q + 1) + η(q + 1)
m=1
mq n=1 n
{recall the definition of H n in (1.160) for the inner sum}
∞
X 1
−(−1)m H m − H m − ζ(q + 1) + η(q + 1)

= q
m=1
m
∞ ∞
X (−1)m H m X Hm
=− q
− − ζ(q + 1) + η(q + 1).
m=1
m m=1
mq
∞
X (−1)m H m
By combining S1 and S2 , the sum cancels out,
m=1
mq
q−2
X
η(q − j)η(j + 2) = (q)ζ(q + 1) − 2 ln(2)[η(q) + ζ(q)]
j=1
∞
X Hm
−2η(q + 1) + 2 ,
m=1
mq
and the proof follows on using η(s) = (1 − 21−s )ζ(s) given in (1.75).
For other proofs, check [19, Theorem 3.5, p. 9] and [11, Theorem 7.1 (i), p. 32].
Examples
∞
X Hn 3 1
2
= ln(2)ζ(2) − ζ(3); (4.12)
n=1
n 2 4
∞
X Hn 7 5
3
= ln(2)ζ(3) − ζ(4); (4.13)
n=1
n 4 16
∞
X Hn 15 3 17
4
= ln(2)ζ(4) + ζ(2)ζ(3) − ζ(5); (4.14)
n=1
n 8 8 16
∞
X Hn 31 9 49
5
= ln(2)ζ(5) + ζ 2 (3) − ζ(6); (4.15)
n=1
n 16 32 64
∞
X Hn 63 21 15 129
6
= ln(2)ζ(6) + ζ(3)ζ(4) + ζ(2)ζ(5) − ζ(7). (4.16)
n=1
n 32 32 32 64
P∞ (−1)n Hn
4.1.4 n=1 n2q
For q ∈ Z+ , the following identity holds:

∞
X (−1)n Hn 22q+1 q − 2q − 1
2q
=− ζ(2q + 1)
n=1
n 22q+1
2q−2
1 X
2 − 2−j − (−2)1−j − 2j−2q+1 ζ(2q − j)ζ(j + 1).

+ (4.17)
4 j=1

1
(−x)n
Z
H n = ln(2) − dx
0 1+x
n
(−1)
by n2q then take the summation over n ≥ 1, we get
∞ ∞ Z 1
(−1)n H n (−1)n (−x)n
X X
= ln(2) − dx
n=1
n2q n=1
n2q 0 1+x
∞ ∞ Z 1 n
X (−1)n X 1 x
= ln(2) 2q
− 2q
dx
n=1
n n=1
n 0 1 +x
{make use of the identity in (3.96) for the integral}
∞
X 1
= − ln(2)η(2q) − 2q
(ln(2) + H n2 − Hn )
n=1
n
∞ ∞
X H n2 X Hn
= − ln(2)η(2q) − ln(2)ζ(2q) − 2q
+ . (4.18)
n=1
n n=1
n2q
To get the first sum, set p = 2 and replace q by 2q in (4.1),

∞ ∞ 2q−2
X H n2 X H2n X
2q
= 2 2q
− (−2)−j ζ(2q − j)ζ(j + 1)
n=1
n n=1
(2n) j=1

∞ ∞ 2q−2
X Hn (−1)n Hn
X X
= + − (−2)−j ζ(2q − j)ζ(j + 1). (4.19)
n=1
n2q n=1 n2q j=1
Substitute (4.19) in (4.18) then rearrange the terms,

∞ ∞ 2q−2
X (−1)n H n X (−1)n Hn X
2q
+ 2q
= (−2)−j ζ(2q − j)ζ(j + 1)
n=1
n n=1
n j=1
− ln(2)(η(2q) + ζ(2q)). (4.20)
Hn
To establish another relation, let an = n2q in (1.5):
∞ ∞ ∞
X 1X 1X
a2n = an + (−1)n an ,
n=1
2 n=1 2 n=1
we obtain
∞ ∞ ∞
X (−1)n H n X H n X H 2n
2q
+ 2q
= 2 2q
n=1
n n=1
n n=1
(2n)
{substitute H 2n = H2n − Hn given in (1.162)}
∞ ∞
X H2n X Hn
=2 2q
− 2
n=1
(2n) n=1
(2n)2q
{employ (1.5) for the first sum}
∞ ∞ ∞
X (−1)n Hn X Hn X Hn
= 2q
+ 2q
− 2
n=1
n n=1
n n=1
(2n)2q
∞ ∞
X (−1)n Hn 1−2q
X Hn
= 2q
+ (1 − 2 ) 2q
.
n=1
n n=1
n

∞ ∞ ∞ ∞
X (−1)n H n X (−1)n Hn 1−2q
X Hn X H n
− = (1 − 2 ) − . (4.21)
n=1
n2q n=1
n2q n=1
n2q n=1 n2q
Take the difference of (4.20) and (4.21) then divide by 2,

∞ 2q−2
X (−1)n Hn 1 1 X
2q
= − ln(2)(ζ(2q) + η(2q)) + (−2)−j ζ(2q − j)ζ(j + 1)
n=1
n 2 2 j=1
∞ ∞
1 X Hn 1 − 21−2q X Hn
+ − .
2 n=1 n2q 2 n=1
n2q
The proof follows on substituting the results from (4.11) and (4.2) and using η(s) =
(1 − 21−s )ζ(s).
Check [4, II.1, pp. 4–7] and [23] for other proofs with a different closed form, which
is well-known in the mathematical literature.
Examples
∞
X (−1)n Hn 5
2
= ζ(3); (4.22)
n=1
n 8
∞
X (−1)n Hn 1 59
4
= ζ(2)ζ(3) − ζ(5); (4.23)
n=1
n 2 32
∞
X (−1)n Hn 1 7 377
6
= ζ(2)ζ(5) + ζ(3)ζ(4) − ζ(7); (4.24)
n=1
n 2 8 128
∞
X (−1)n Hn 1 31 7 2039
8
= ζ(2)ζ(7) + ζ(3)ζ(6) + ζ(4)ζ(5) − ζ(9); (4.25)
n=1
n 2 32 8 512
∞
X (−1)n Hn 1 127 7 31
10
= ζ(2)ζ(9) + ζ(3)ζ(8) + ζ(4)ζ(7) + ζ(5)ζ(6)
n=1
n 2 128 8 32
10229
−ζ(11); (4.26)
2048
∞
X (−1)n Hn 1 511 7 127
12
= ζ(2)ζ(11) + ζ(3)ζ(10) + ζ(4)ζ(9) + ζ(5)ζ(8)
n=1
n 2 512 8 128
31 49139
+ ζ(6)ζ(7) − ζ(13). (4.27)
32 8192
P∞ (−1)n H n
4.1.5 n=1 n2q
For q ∈ Z+ , the following equality holds:

∞
X (−1)n H n 22q+1 q − 2q − 1
2q
= (21−2q − 2) ln(2)ζ(2q) + ζ(2q + 1)
n=1
n 22q+1
2q−2
1 X j−2q+1
− (−2)1−j + 2−j − 2 ζ(2q − j)ζ(j + 1).

+ 2 (4.28)
4 j=1
Proof. Combine (4.20) and (4.21) then divide by 2,

∞ 2q−2
X (−1)n H n 1 1 X
2q
= − ln(2)(ζ(2q) + η(2q)) + (−2)−j ζ(2q − j)ζ(j + 1)
n=1
n 2 2 j=1
∞ ∞
1 X Hn 1 − 21−2q X Hn
− + .
2 n=1 n2q 2 n=1
n2q

Examples
∞
X (−1)n H n 5 3
2
= ζ(3) − ln(2)ζ(2); (4.29)
n=1
n 8 2
∞
X (−1)n H n 59 15 3
4
= ζ(5) − ln(2)ζ(4) − ζ(2)ζ(3); (4.30)
n=1
n 32 8 4
∞
X (−1)n H n 377 63 15 3
6
= ζ(7) − ln(2)ζ(6) − ζ(2)ζ(5) − ζ(3)ζ(4); (4.31)
n=1
n 128 32 16 4
∞
X (−1)n H n 2039 255 63 3
8
= ζ(9) − ln(2)ζ(8) − ζ(2)ζ(7) − ζ(3)ζ(6)
n=1
n 512 128 64 4
15
− ζ(4)ζ(5); (4.32)
16
∞
X (−1)n H n 10239 1023 255 3
= ζ(11) − ln(2)ζ(10) − ζ(2)ζ(9) − ζ(3)ζ(8)
n=1
n10 2048 512 256 4
63 15
− ζ(4)ζ(7) − ζ(5)ζ(6); (4.33)
64 16
∞
X (−1)n H n 49139 4095 1023
= ζ(13) − ln(2)ζ(12) − ζ(2)ζ(11)
n=1
n12 8192 2048 1024
3 255 63 15
− ζ(3)ζ(10) − ζ(4)ζ(9) − ζ(6)ζ(7) − ζ(5)ζ(8). (4.34)
4 256 64 16
P∞ Hn
4.1.6 n=1 n2q
2

∞
X H n2 22q+1 + 2q + 1
2q
= ζ(2q + 1)
n=1
n 22q+1
2q−2
1 X
(−2)1−j − 2−j − 2j−2q+1 ζ(2q − j)ζ(j + 1).

+ (4.35)
4 j=1
Proof. Substitute the results from (4.2) and (4.17) in (4.19).
Examples
∞
X H n2 11
2
= ζ(3); (4.36)
n=1
n 8
∞
X H n2 37 1
4
= ζ(5) − ζ(2)ζ(3); (4.37)
n=1
n 32 4
∞
X H n2 135 1 1
6
= ζ(7) − ζ(2)ζ(5) − ζ(3)ζ(4); (4.38)
n=1
n 128 16 4
∞
X H n2 521 1 1 1
8
= ζ(9) − ζ(2)ζ(7) − ζ(3)ζ(6) − ζ(4)ζ(5); (4.39)
n=1
n 512 64 4 16
∞
X H n2 2059 ζ(2) 1 1 1
10
= ζ(11) − ζ(9) − ζ(3)ζ(8) − ζ(4)ζ(7) − ζ(5)ζ(6).
n=1
n 2048 256 4 64 16
(4.40)
P∞ (−1)n H n
4.1.7 n=1 n2q
2

∞
X (−1)n H n2 2q + 3 − 22q+1
= ζ(2q + 1)
n=1
n2q 22q+1
2q−2
1 X
(−2)1−j − 2−j − 21−2q (2j − 2) ζ(2q − j)ζ(j + 1).

− (4.41)
4 j=1
Hn
Proof. Set an = 2
n2q in (1.5):
∞
X ∞
X ∞
X
2 a2n = an + (−1)n an ,
n=1 n=1 n=1
we get
∞ ∞ ∞
X Hn X H n2 X Hn
2 2q
= 2q
+ (−1)n 2q2
n=1
(2n) n=1
n n=1
n
or
∞ ∞ ∞
X (−1)n H n2 1−2q
X Hn X H n2
= 2 − .
n=1
n2q n=1
n2q n=1 n2q
On gathering the results from (4.2) and (4.35), the proof is complete.
Examples
∞
X (−1)n H n2 3
2
= − ζ(3); (4.42)
n=1
n 8
∞
X (−1)n H n2 1 25
4
= ζ(2)ζ(3) − ζ(5); (4.43)
n=1
n 8 32
∞
X (−1)n H n2 1 7 119
6
= ζ(2)ζ(5) + ζ(3)ζ(4) − ζ(7); (4.44)
n=1
n 32 32 128
∞
X (−1)n H n2 1 31 7 501
8
= ζ(2)ζ(7) + ζ(3)ζ(6) + ζ(4)ζ(5) − ζ(9);
n=1
n 128 128 128 512
(4.45)
∞ n n
X (−1) H 2 1 127 7 31
10
= ζ(2)ζ(9) + ζ(3)ζ(8) + ζ(4)ζ(7) + ζ(5)ζ(6)
n=1
n 512 512 512 512
2035
− ζ(11). (4.46)
2048
P∞ (q)
ζ(q)−Hn
4.1.8 n=1 n
For integer q ≥ 2, the following equality holds:

∞ (q) q−2
X ζ(q) − Hn q 1X
= ζ(q + 1) − ζ(q − j)ζ(j + 1). (4.47)
n=1
n 2 2 j=1
Proof (i). Divide both sides of (1.153):
(−1)q−1 1
xn lnq−1 (x)
Z
ζ(q) − Hn(q) = dx
(q − 1)! 0 1−x
by n then take the summation over n ≥ 1,

∞ ∞
!
(q)
ζ(q) − Hn (−1)q−1 1 lnq−1 (x) X xn
X Z
= dx
n=1
n (q − 1)! 0 1 − x n=1
n
(−1)q−1 1 − lnq−1 (x) ln(1 − x)
Z
= dx
(q − 1)! 0 1−x
∞
X Hn
= − ζ(q + 1).
n=1
nq
Substitute the generalized Euler sum given in (4.2) to complete the proof.
(q) 1
Proof (ii). Let bk = ζ(q) − Hk and ak = k in (2.95):
n
X n−1
X n
X
ak bk = An bn − Ak (bk+1 − bk ) , An = ai ,
k=1 k=0 i=1
we have
n (q) n n−1 k
!
X ζ(q) − H X 1 X X 1
(q) (q)

k
= (ζ(q) − Hn(q) ) − −Hk+1 + Hk
k i=1
i i=1
i
k=1 k=0
n−1
X −1
= (ζ(q) − Hn(q) )Hn − (Hk ) .
(k + 1)q
k=0
Let n 7→ ∞ and use the lim Hn(q) = ζ(q),

n→∞
∞ (q) ∞
X ζ(q) − H k
X Hk
=0+
k (k + 1)q
k=1 k=0
∞ ∞ ∞
X Hk−1 X Hk − k1 X Hk
= = = − ζ(q + 1).
kq kq kq
k=1 k=1 k=1
Examples
∞ (2)
X ζ(2) − Hn
= ζ(3); (4.48)
n=1
n
∞ (3)
X ζ(3) − Hn 1
= ζ(4); (4.49)
n=1
n 4
∞ (4)
X ζ(4) − Hn
= 2ζ(5) − ζ(2)ζ(3); (4.50)
n=1
n
∞ (5)
X ζ(5) − Hn 3 1
= ζ(6) − ζ 2 (3); (4.51)
n=1
n 4 2
∞ (6)
X ζ(6) − Hn
= 3ζ(7) − ζ(2)ζ(5) − ζ(3)ζ(4). (4.52)
n=1
n
P∞ (2)
Hn
4.1.9 n=1 n2q+1

∞ (2)
X Hn 1
2q+1
= ζ(2)ζ(2q + 1) − (q + 2)(2q + 1)ζ(2q + 3)
n=1
n 2
2q 2q−1
2q + 1 X 1 X
+ ζ(2q − j + 2)ζ(j + 1) − k(−1)k ζ(2q − k + 1)ζ(k + 2).
4 j=1 2
k=1
(4.53)

Z 1
Hn
xn−1 ln(1 − x)dx = −
0 n
with respect to n using the derivative of the harmonic number given in (1.157),
(2) Z 1
Hn Hn ζ(2) d
+ − = xn−1 ln(1 − x)dx
n2 n n dn 0
Z 1
∂ n−1
= x ln(1 − x)dx
0 ∂n
Z 1
= xn−1 ln(x) ln(1 − x)dx.
0
Thus, we have
(2) Z 1
Hn Hn ζ(2)
+ − = xn−1 ln(x) ln(1 − x)dx. (4.54)
n2 n n 0
Next, divide both sides by na then take the summation over n ≥ 1,

∞ ∞ ∞
!
(2) Z 1
X Hn X Hn ln(x) ln(1 − x) X xn
a+2
+ a+1
− ζ(2)ζ(a + 1) = dx
n=1
n n=1
n 0 x n=1
na
Z 1
ln(x) ln(1 − x) Lia (x)
= dx
0 x
{expand ln(1 − x) in series}
∞
1 1 n−1
X Z
=− x ln(x) Lia (x)dx
n=1
n 0
∞
1 1 ∂ n−1
X Z
=− x Lia (x)dx
n=1
n 0 ∂n
{use differentiation under the intgral sign theorem given in (2.78)}
∞ Z 1
X 1 d
=− xn−1 Lia (x)dx
n=1
n dn 0

∞
" a−1
#
1 d a−1 Hn k ζ(a − k + 1)
X X
=− (−1) − (−1)
n=1
n dn na nk
k=1
∞
" ! a−1 #
(2)
1 Hn aHn ζ(2) k ζ(a − k + 1)
X X
a−1
=− (−1) − a − a+1 + a + k(−1)
n=1
n n n n nk+1
k=1
∞ (2) ∞
X Hn X Hn
= −(−1)a a+1
− a(−1) a
a+2
+ (−1)a ζ(2)ζ(a + 1)
n=1
n n=1
n
a−1
X
− k(−1)k ζ(a − k + 1)ζ(k + 2).
k=1
Reorder the term,

∞ (2) ∞
X Hn X Hn
(1 + (−1)a ) a+1
= −(1 + a(−1) a
)
n=1
n n=1
na+2
a−1
X
+(1 + (−1)a )ζ(2)ζ(a + 1) − k(−1)k ζ(a − k + 1)ζ(k + 2).
k=1
Replace a by 2q then divide by 2,

∞ (2) ∞
X Hn 1 + 2q X Hn
= − + ζ(2)ζ(2q + 1)
n=1
n2q+1 2 n=1 n2q+2
2q−1
1 X
− k(−1)k ζ(2q − k + 1)ζ(k + 2). (4.55)
2
k=1
The proof completes on substituting the generalized Euler sum in (4.2).
Examples
∞ (2)
X Hn 9
3
= 3ζ(2)ζ(3) − ζ(5); (4.56)
n=1
n 2
∞ (2)
X Hn
= 5ζ(2)ζ(5) + 2ζ(3)ζ(4) − 10ζ(7).; (4.57)
n=1
n5
∞ (2)
X Hn 35
7
= 7ζ(2)ζ(7) + 2ζ(3)ζ(6) + 4ζ(4)ζ(5) − ζ(9); (4.58)
n=1
n 2
∞ (2)
X Hn
= 9ζ(2)ζ(9) + 2ζ(3)ζ(8) + 6ζ(4)ζ(7) + 4ζ(5)ζ(6) − 27ζ(11); (4.59)
n=1
n9
∞ (2)
X Hn
= 11ζ(2)ζ(11) + 2ζ(3)ζ(10) + 8ζ(4)ζ(9) + 4ζ(5)ζ(8) + 6ζ(6)ζ(7)
n=1
n11
77
− ζ(13). (4.60)
2
P∞ (2q+1)
Hn
4.1.10 n=1 n2

∞ (2q+1)
X Hn (q + 2)(2q + 1)
= 1+ ζ(2q + 3)
n=1
n2 2
2q 2q−1
2q + 1 X 1 X
− ζ(2q − j + 2)ζ(j + 1) + k(−1)k ζ(2q − k + 1)ζ(k + 2).
4 j=1 2
k=1
(4.61)
Proof. Set p = 2q + 1 and q = 2 in (2.96):

∞ (p) ∞ (q)
X H k
X H k
+ = ζ(p)ζ(q) + ζ(p + q),
kq kp
k=1 k=1
we have
∞ (2q+1) ∞ (2)
X Hn X Hn
+ = ζ(2)ζ(2q + 1) + ζ(2q + 3).
n=1
n2 n=1
n2q+1
Substituting the result from (4.53) finishes the proof.

Examples
∞ (3)
X Hn 11
2
= ζ(5) − 2ζ(2)ζ(3); (4.62)
n=1
n 2
∞ (5)
X Hn
= 11ζ(7) − 4ζ(2)ζ(5) − 2ζ(3)ζ(4); (4.63)
n=1
n2
∞ (7)
X Hn 37
2
= ζ(9) − 6ζ(2)ζ(7) − 2ζ(3)ζ(6) − 4ζ(4)ζ(5); (4.64)
n=1
n 2
∞ (9)
X Hn
= 28ζ(11) − 8ζ(2)ζ(9) − 2ζ(3)ζ(8) − 6ζ(4)ζ(7) − 4ζ(5)ζ(6); (4.65)
n=1
n2
∞ (11)
X Hn 79
2
= ζ(13) − 10ζ(2)ζ(11) − 2ζ(3)ζ(10) − 8ζ(4)ζ(9) − 4ζ(5)ζ(8)
n=1
n 2
−6ζ(6)ζ(7). (4.66)
P∞ Hn2
4.1.11 n=1 n2q+1

∞
X Hn2 1 1
2q+1
= (q + 2)(2q + 5)ζ(2q + 3) − ζ(2)ζ(2q + 1)
n=1
n 6 3
 
2q−1 k−1
1 X X
− (−1)k ζ(2q − k + 1) (k + 2)ζ(k + 2) − 2 ζ(k − j + 1)ζ(j + 1)
6 j=1
k=1
2q
2q + 5 X
− ζ(2q − j + 2)ζ(j + 1). (4.67)
12 j=1

1 (2)
Hn2 + Hn
Z
xn−1 ln2 (1 − x)dx =
0 n
a
by n then consider the summation,
∞ ∞ ∞
!
(2) Z 1 2
X Hn2 X Hn ln (1 − x) X xn
+ = dx
n=1
na+1 n=1 na+1 0 x n=1
na
Z 1 2
ln (1 − x) Lia (x)
= dx
0 x
{expand ln2 (1 − x) in series given in (2.6)}
∞
Hn−1 1 n−1
X Z
=2 x Lia (x)dx
n=1
n 0
{recall the result of the integral from (3.102)}

∞
" a−1
#
X Hn − n1 a−1 Hn
X
k ζ(a − k + 1)
=2 (−1) − (−1) .
n=1
n na nk
k=1
Distributing and rearranging the terms,

∞ ∞ (2) ∞
X Hn2 X Hn X Hn
(1 + 2(−1)a ) a+1
= − a+1
+ 2(−1) a
a+2
n=1
n n=1
n n=1
n
∞
a−1
!
X
k
X Hn
+2 (−1) ζ(a − k + 1) ζ(k + 2) − .
n=1
nk+1
k=1
P∞ (2)
Hn
Set a = 2q and recall the relation involving n=1 n2q+1 from (4.53),
∞ ∞
X Hn2 2q + 5 X Hn
3 = − ζ(2)ζ(2q + 1)
n=1
n2q+1 2 n=1 n2q+2
2q−1 ∞
" #
X
k k+4 X Hn
+ (−1) ζ(2q − k + 1) ζ(k + 2) − 2 .
2 n=1
nk+1
k=1
Substituting the generalized Euler sum given in (4.2) completes the proof.
Examples
∞
X Hn2 7
3
= ζ(5) − ζ(2)ζ(3); (4.68)
n=1
n 2
∞
X Hn2 5
5
= 6ζ(7) − ζ(2)ζ(5) − ζ(3)ζ(4); (4.69)
n=1
n 2
∞
X Hn2 55 7 1 5
7
= ζ(9) − ζ(2)ζ(7) − ζ(3)ζ(6) + ζ 3 (3) − ζ(4)ζ(5); (4.70)
n=1
n 6 2 3 2
∞
X Hn2 9 5 7
9
= 11ζ(11) − ζ(2)ζ(9) − ζ(3)ζ(8) − ζ(4)ζ(7) − ζ(5)ζ(6)
n=1
n 2 2 2
+ζ 2 (3)ζ(5); (4.71)
∞
X Hn2 35 5
11
= ζ(13) − ζ(2)ζ(11) + ζ(3)ζ 2 (5) + ζ 2 (3)ζ(7) − ζ(4)ζ(9)
n=1
n 2 2
9 7 11
− ζ(5)ζ(8) − ζ(6)ζ(7) − ζ(3)ζ(10). (4.72)
2 2 2
P∞ Hn
4.1.12 n=1 (2n+1)q

∞
X Hn
q
= (21−q − 2) ln(2)ζ(q) + q(1 − 2−q−1 )ζ(q + 1)
n=1
(2n + 1)
q−2
1 X j+1
− (2 − 1)(2−j − 2−q )ζ(q − j)ζ(j + 1). (4.73)
2 j=1
Hn
Proof. Set an = nq in (1.4):
∞
X ∞
X ∞
X
an = a2n+1 + a2n ,
n=1 n=0 n=1
we obtain
∞ ∞ ∞
X Hn X H 2n+1 X H 2n
q
= q
+ .
n=1
n n=0
(2n + 1) n=1
(2n)q
For the first sum, write H 2n+1 = H2n+1 − Hn given in (1.163),

∞ ∞ ∞
X H 2n+1 X H2n+1 X Hn
q
= q
−
n=0
(2n + 1) n=0
(2n + 1) n=0
(2n + 1)q
{and let the index start from 1 in the second sum, since H0 = 0}
∞ ∞ ∞
1 X Hn+1 1 X (−1)n Hn+1 X Hn
= + −
2 n=0 (n + 1)q 2 n=0 (n + 1)q n=1
(2n + 1)q
{shift the index n of the first and second sums by −1}
∞ ∞ ∞
1 X Hn 1 X (−1)n Hn X Hn
= − − .
2 n=1 nq 2 n=1 nq n=1
(2n + 1)q
For the second sum, write H 2n = H2n − Hn given in (1.162),

∞ ∞ ∞
X H 2n X H2n X Hn
q
= q
− q
n=1
(2n) n=1
(2n) n=1
(2n)
∞ ∞ ∞
1 X Hn 1 X (−1)n Hn −q
X Hn
= + − 2
2 n=1 nq 2 n=1 nq n=1
nq
∞ ∞
1 X (−1)n Hn
X
1 −q Hn
= + − 2 .
2 n=1 nq 2 n=1
nq
Combining the two sums, we arrive at

∞ ∞ ∞
X Hn −q
X Hn X H n
= (1 − 2 ) − ,
n=1
(2n + 1)q n=1
nq n=1
nq
and the proof follows on collecting the results from (4.2) and (4.11).
Examples
∞
X Hn 7 3
2
= ζ(3) − ln(2)ζ(2); (4.74)
n=1
(2n + 1) 4 2
∞
X Hn 45 7
3
= ζ(4) − ln(2)ζ(3); (4.75)
n=1
(2n + 1) 32 4
∞
X Hn 31 15 21
4
= ζ(5) − ln(2)ζ(4) − ζ(2)ζ(3); (4.76)
n=1
(2n + 1) 8 8 16
∞
X Hn 315 31 49
5
= ζ(6) − ln(2)ζ(5) − ζ 2 (3); (4.77)
n=1
(2n + 1) 128 16 64
∞
X Hn 381 63 93 105
6
= ζ(7) − ln(2)ζ(6) − ζ(2)ζ(5) − ζ(3)ζ(4).
n=1
(2n + 1) 64 32 64 64
(4.78)
(q)
P∞ (−1)n Hn
4.1.13 n=1 n

∞ (q)
X (−1)n Hn q
= (1 − 21−q ) ln(2)ζ(q) − ζ(q + 1)
n=1
n 2
q−2
1X
+ (1 − 21−q+j )(1 − 2−j )ζ(q − j)ζ(j + 1). (4.79)
2 j=1
(−1)q−1 1
xn lnq−1 (x)
Z
Hn(q) = ζ(q) − dx,
(q − 1)! 0 1−x
(−1)n
∞ ∞ ∞
!
(q)
(−1)n Hn (−1)n (−1)q−1 1 lnq−1 (x) X (−x)n
X X Z
= ζ(q) − dx
n=1
n n=1
n (q − 1)! 0 1 − x n=1
n
(−1)q−1 1 lnq−1 (x)
Z
= − ln(2)ζ(q) − (− ln(1 + x)) dx
(q − 1)! 0 1 − x
∞
( )
ln(1 + x) X n
write = H n x given in (2.28)
1−x n=1
∞ Z 1
(−1)q−1 X
= − ln(2)ζ(q) + Hn xn lnq−1 (x)dx
(q − 1)! n=1 0
∞
X Hn
= − ln(2)ζ(q) +
n=1
(n + 1)q
{let the index start from 0, since H 0 = 0}
∞
X Hn
= − ln(2)ζ(q) +
n=0
(n + 1)q
{shift the index by −1}
∞
X H n−1
= − ln(2)ζ(q) +
n=1
nq
(−1)n

use H n−1 = H n + given in (1.161)
n
∞ ∞
X H n X (−1)n
= − ln(2)ζ(q) + +
n=1
nq n=1
nq+1
∞
X Hn
= − ln(2)ζ(q) + − η(q + 1).
n=1
nq
Substituting the result from (4.11) and writing η(q + 1) = (1 − 2−q )ζ(q + 1) finalizes
the proof.
Using integral manipulations, Cornel Vălean managed to provide a different proof,
which you may find in [26]. He also derived, in the same reference, a closed form for
∞ (q)
X (−1)n H2n
.
n=1
n
Examples
∞ (2)
X (−1)n Hn 1
= ln(2)ζ(2) − ζ(3); (4.80)
n=1
n 2
∞ (3)
X (−1)n Hn 3 19
= ln(2)ζ(3) − ζ(4); (4.81)
n=1
n 4 16
∞ (4)
X (−1)n Hn 7 3
= ln(2)ζ(4) + ζ(2)ζ(3) − 2ζ(5); (4.82)
n=1
n 8 8
∞ (5)
X (−1)n Hn 15 9 111
= ln(2)ζ(5) + ζ 2 (3) − ζ(6); (4.83)
n=1
n 16 32 64
∞ (6)
X (−1)n Hn 15 21 31
= ζ(2)ζ(5) + ζ(3)ζ(4) + ln(2)ζ(6) − 3ζ(7). (4.84)
n=1
n 32 32 32
(2q+1)
P∞ (−1)n Hn
4.1.14 n=1 2n+1

∞ (2q+1)
X (−1)n Hn
= (2−1 − 22q−1 )πζ(2q + 1) + 22q β(2q + 2)
n=1
2n + 1
q
X
+ 22q−2n+1 (1 − 21−2n )ζ(2n)β(2q − 2n + 2). (4.85)
n=1
Proof. Replace a by 2q + 1 and x by −x2 In (2.2),

∞
X Li2q+1 (−x2 )
(−1)n Hn(2q+1) x2n = .
n=1
1 + x2
R1 1
Integrate both sides from x = 0 to 1 using 0
x2n dx = 2n+1 ,
∞ (2q+1) Z 1
X (−1)n Hn Li2q+1 (−x2 )
= dx.
n=1
2n + 1 0 1 + x2
This integral is given in (3.76).
Examples
∞ (3)
X (−1)n Hn 3π
= 4β(4) + ζ(2)β(2) − ζ(3); (4.86)
n=1
2n + 1 2
∞ (5)
X (−1)n Hn 7 15π
= 16β(6) + 4ζ(2)β(4) + ζ(4)β(2) − ζ(5); (4.87)
n=1
2n + 1 4 2
∞ (7)
X (−1)n Hn 31 63π
= 64β(8) + 16ζ(2)β(6) + 7ζ(4)β(4) + ζ(6)β(2) − ζ(7);
n=1
2n + 1 16 2
(4.88)
∞ (9)
X (−1)n Hn 31
= 256β(10) + 64ζ(2)β(8) + 28ζ(4)β(6) + ζ(6)β(4)
n=1
2n + 1 4
127 255π
+ ζ(8)β(2) − ζ(9); (4.89)
64 2
∞ (11)
X (−1)n Hn
= 1024β(12) + 256ζ(2)β(10) + 112ζ(4)β(8) + 31ζ(6)β(6)
n=1
2n + 1
127 511 1023π
+ ζ(8)β(4) + ζ(10)β(2) − ζ(11). (4.90)
16 256 2
P∞ (−1)n Hn
4.1.15 n=1 (2n+1)2q+1
For q ∈ Z≥0 , the following identity holds:

∞
X (−1)n Hn 4−q−1 π d2q ψ(1 − m) + γ
2q+1
= (2q + 1)β(2q + 2) + lim1 .
n=1
(2n + 1) (2q)! m→ 2 dm2q sin(mπ)
(4.91)
Proof. By using (1.31), we have

Z 1
1 1
= x2n ln2q (x)dx.
(2n + 1)2q+1 (2q)! 0
Multiply both sides by (−1)n Hn then take the summation over n ≥ 1,

∞ ∞
Z 1 !
X (−1)n Hn 1 2q
X
2q+1
= ln (x) Hn (−x2 )n dx
n=1
(2n + 1) (2q)! 0 n=1
{replace x by −x2 in the generating function in (2.4)}
Z 1
ln(1 + x2 )

1 2q
= ln (x) − dx
(2q)! 0 1 + x2
Z 1 2q
1 ln (x) ln(1 + x2 )
=− dx,
(2q)! 0 1 + x2
and this integral is given in (3.84).

Examples
∞
X (−1)n Hn π
= β(2) + ; (4.92)
n=1
2n + 1 4
∞
X (−1)n Hn 7π π3
= 3β(4) − ζ(3) − ln(2); (4.93)
n=1
(2n + 1)3 16 16
∞
X (−1)n Hn 31π 7π 3 5π 5
= 5β(6) − ζ(5) − ζ(3) − ln(2); (4.94)
n=1
(2n + 1)5 64 128 768
∞
X (−1)n Hn 127π 31π 3 35π 5 61π 7
7
= 7β(8) − ζ(7) − ζ(5) − ζ(3) − ln(2);
n=1
(2n + 1) 256 512 6144 92160
(4.95)
∞
X (−1)n Hn 511π 127π 3 155π 5
= 9β(10) − ζ(9) − ζ(7) − ζ(5)
n=1
(2n + 1)9 1024 2048 24576
427π 7 277π 9
− ζ(3) − ln(2), (4.96)
737280 4128768
where in the calculations, we used ψ (a) 12 = (−1)a−1 a!(2a+1 − 1)ζ(a + 1) given

in (1.186).
4.2 Non–Alternating Harmonic Series

P∞ Hn
4.2.1 n=1 n2
Show that
∞
X Hn
= 2ζ(3). (4.97)
n=1
n2
Solution Setting a = 1 in (3.88),

∞ Z 1
X Hn ln(x) ln(1 − x)
2
= dx
n=1
n 0 x(1 − x)
{set a = 1 in ( 3.85)}
Z 1
ln(x) ln(1 − x)
=2 dx
0 x
{expand ln(1 − x) in series}
∞
1 1 n−1
X Z
= −2 x ln(x)dx
n=1
n 0
∞
IBP
X 1 1
= −2 − 2 = 2ζ(3).
n=1
n n
For a different approach, set x = 1 in (2.8). Also check (4.4).
P∞ (2)
Hn
4.2.2 n=1 n2
Show that
∞ (2)
X Hn 17
2
= ζ(4). (4.98)
n=1
n 4
Solution Put x = 1 in (2.83),

∞ ∞ (2)
X Hn X Hn
Li22 (1) =4 + 2 − 6 Li4 (1).
n=1
n3 n=1
n2
The solution completes on recalling the result from (4.5). Note that Lia (1) = ζ(a)
given in (1.100) and Li22 (1) = ζ 2 (2) = 25 ζ(4) given in (1.62).
Check (2.98) for a different solution.
4.2. Non–Alternating Harmonic Series 217
P∞ 2
Hn
4.2.3 n=1 n2
Show that
∞
X Hn2 17
2
= ζ(4). (4.99)
n=1
n 4
Solution (i) Divide both sides of (2.71):

1 (2)
Hn2 + Hn
Z
xn−1 ln2 (1 − x)dx =
0 n

∞ ∞
!
(2) Z 1 2
X Hn2 + Hn ln (1 − x) X xn
= dx
n=1
n2 0 x n=1
n
Z 1 3 Z 1 3
ln (1 − x) 1−x=y ln (y)
=− dx = − dy.
0 x 0 1−y

∞ ∞ (2) Z 1 3
X Hn2 X Hn ln (y)
2
= − 2
− dy.
n=1
n n=1
n 0 1−y
These two terms are calculated in (3.4) and (2.98).

∞
1 2 X Hn−1 n
ln (1 − x) = x
2 n=1
n
ln(1−x)
∞
1
ln3 (1 − x)
Z Z 1
1 X Hn−1
dx = xn−1 ln(1 − x)dx
2 0 x n=1
n 0
{recall the result from (2.70) for the integral}

∞ ∞ ∞
Hn − n1 Hn X Hn2
X
X Hn
= − = − .
n=1
n n n=1
n3 n=1
n2
Rearrange the terms then let 1 − x = y in the integral,

∞ ∞
Hn2 1 1 ln3 (y)
Z
X X Hn
= − dy.
n=1
n2 n=1
n3 2 0 1−y
The solution finalizes on substituting the values from (4.5) and (3.4).
Solution (iii) Multiply both sides of (2.11):
∞
ln2 (1 − x) X 2
= (Hn − Hn(2) )xn
1−x n=1
R1
by − ln(x)
x then integrate using − 0
xn−1 ln(x)dx = 1
n2 ,
∞ (2) Z 1 2
X Hn2 − Hn ln (1 − x) ln(x)
2
= − dx
n=1
n 0 (1 − x)x
Z 1 2
1−x=y ln (y) ln(1 − y)
= − dy
0 y(1 − y)
{set a = 2 in (3.88)}
∞
X Hn
=2 .
n=1
n3
Reorder the terms,

∞ ∞ (2) ∞
X Hn2 X Hn X Hn
2
= 2
+ 2 .
n=1
n n=1
n n=1
n3
Substitute the results from (2.98) and (4.5) to finish the solution.
For different methods of evaluating (2.98) and (4.99), check [22].
P∞ Hn H2n
4.2.4 n=1 n2
Show that
∞
X Hn H2n 1 13 7 1
2
= 4 Li 4 + ζ(4) + ln(2)ζ(3) − ln2 (2)ζ(2) + ln4 (2).
n=1
n 2 8 2 6
(4.100)
Solution Replace n by 2n in (2.70),

Z 1
H2n
x2n−1 ln(1 − x)dx = − .
0 2n
Multiply both sides by 2 2H2nn−Hn then take the summation over n ≥ 1,

∞ ∞ ∞
Z 1 !
2
X Hn H2n X H2n ln(1 − x) X 2H2n − Hn 2n
2
−2 2
= 2 x dx
n=1
n n=1
n 0 x n=1
n
{recall the generating function from (2.41)}
1 Z 1
ln(1 − x) ln(1 − x) 1−x
Z
2 2

= 4 arctanh (x) dx = ln dx
0 x 0 x 1+x

1 1 2
use a(a − b)2 = (a + b)3 + (a − b)3 + a3 − 2a2 b
6 6 3
{with a = ln(1 − x) and b = ln(1 + x)}

3 Z 1 ln3 1−x
1 2
ln (1 − x ) 2 1 ln3 (1 − x)
Z Z
1 1 1+x
= dx + dx + dx
6 x 6 0 x 3 0 x
|0 {z } | {z }
1−x2 →x 1−x→x
1 2
ln (1 − x) ln(1 + x)
Z
−2 dx
0 x

3 1−x
ln3 (x) 1 1 ln 1+x
1 Z 1 2
ln (1 − x) ln(1 + x)
Z Z
3
= dx + dx − 2 dx.
4 0 1−x 6 0 x 0 x
Put together the results from (3.4), (3.17), and (3.118),

∞ ∞ 2
X Hn H2n X H2n
− 2
n=1
n2 n=1
n2

2 1 41 7 1 1 4
= − Li4 − ζ(4) − ln(2)ζ(3) + ln2 (2)ζ(2) − ln (2). (4.101)
3 2 48 12 6 36
For the remaining sum, we have

∞ 2 ∞ 2
X H2n X H2n
= 4
n=1
n2 n=1
(2n)2
{make use of the series trasnformation in (1.5)}
∞ ∞
X Hn2 X (−1)n Hn2
=2 + 2
n=1
n2 n=1
n2

1 27 7 1
= 4 Li4 + ζ(4) + ln(2)ζ(3) − ln2 (2)ζ(2) + ln4 (2).
2 8 2 6
Plugging this sum in (4.101) completes the solution.

P∞ (2)
Hn
4.2.5 n=1 n3
Show that
∞ (2)
X Hn 9
3
= 3ζ(2)ζ(3) − ζ(5). (4.102)
n=1
n 2
Solution Setting x = 1 in (2.85) gives

∞ (2) ∞ (3) ∞
X Hn X Hn X Hn
3 3
+ 2
= ζ(2)ζ(3) + 10ζ(5) − 6
n=1
n n=1
n n=1
n4
{plug in the result from (4.6)}
= 7ζ(2)ζ(3) − 8ζ(5). (4.103)
Now let’s set p = 2 and q = 3 in (2.96) to have

∞ (2) ∞ (3)
X Hn X Hn
+ = ζ(2)ζ(3) + ζ(5). (4.104)
n=1
n3 n=1
n2
Take the difference of the two relations in (4.103) and (4.104) then divide by 2 to end
the solution. Also see (4.56) for another method.
P∞ (3)
Hn
4.2.6 n=1 n2
Show that
∞ (3)
X Hn 11
2
= ζ(5) − 2ζ(2)ζ(3). (4.105)
n=1
n 2
Solution Combine the two relations in (4.103) and (4.104) after multiplying the
latter by −3.
P∞ Hn2
4.2.7 n=1 n 3
Show that
∞
X Hn2 7
3
= ζ(5) − ζ(2)ζ(3). (4.106)
n=1
n 2
Solution (i) Multiply both sides of (2.11):

∞
X ln2 (1 − x)
(Hn2 − Hn(2) )xn =
n=1
1−x
ln2 (x) R1
by 2x then integrate using 1
2 0
xn−1 ln2 (x)dx = 1
n3 ,
∞ (2)
Hn2 − Hn 1 1 ln2 (x) ln2 (1 − x)
X Z
= dx
n=1
n3 2 0 x(1 − x)
{set a = 2 in (3.85)}
Z 1 2
ln (1 − x) ln2 (x)
= dx
0 x
∞ ∞
Hn−1 1 n−1 2 Hn − n1
Z
X IBP
X 2
=2 x ln (x)dx = 2 3
n=1
n 0 n=1
n n
∞ ∞ ∞
X Hn X 1 X Hn
=4 4
− 4 5
= 4 − 4ζ(5).
n=1
n n=1
n n=1
n4
Reorganize the terms,

∞ ∞ (2) ∞
X Hn2 X Hn X Hn
3
= 3
+ 4 − 4ζ(5),
n=1
n n=1
n n=1
n4
and these two series are given in (4.102) and (4.6).
Solution (ii) Divide both sides of (2.6):

∞
X Hn−1 n 1
x = ln2 (1 − x)
n=1
n 2
yn
Ry
xn−1 dx = n ,
∞
1 y ln2 (1 − x)
Z
X Hn−1 n
y = dx.
n=1
n2 2 0 x
Multiply both sides of the latter equality by − ln(1−y)

y then integrate from x = 0 to 1
R 1 n−1 Hn
using − 0 y ln(1 − y)dy = n ,
∞ ∞ ∞ ∞
X Hn−1 Hn X (Hn − n1 )Hn X Hn2 X Hn
= = −
n=1
n3 n=1
n3 n=1
n3 n=1
n4
1 1 y ln2 (1 − x) ln(1 − y)
Z Z
=− dxdy
2 0 0 xy
1 1 ln2 (1 − x)
Z 1
ln(1 − y)
Z
=− dy dx
2 0 x x y
Z 1 2
1 ln (1 − x)
=− (Li2 (x) − ζ(2)) dx
2 0 x
Z 1 2
ln (1 − x) 1 1 ln2 (1 − x) Li2 (x)
Z
1
= ζ(2) dx − dx
2 x 2 0 x
|0 {z }
1−x=y
Z 1 2
1 1 ln2 (1 − x) Li2 (x)
Z
1 ln (y)
= ζ(2) dy − dx
2 0 1−y 2 0 x
∞ ∞ (2)
1 X Hn2 1 X Hn
= ζ(2)ζ(3) − − .
2 n=1 n3 2 n=1 n3

∞ ∞ ∞ (2)
X Hn2 2 2 X Hn 1 X Hn
= ζ(2)ζ(3) + − .
n=1
n3 3 3 n=1 n4 3 n=1 n3

Also check (4.68) for another solution.
P∞ (2)
Hn Hn
4.2.8 n=1 n2
Show that
∞ (2)
X Hn Hn
= ζ(5) + ζ(2)ζ(3). (4.107)
n=1
n2
Solution (i) Divide both sides of (2.72):

1 (2) (3)
Hn3
Z
Hn Hn Hn
− xn−1 ln3 (1 − x)dx = +3 +2
0 n n n
by n then take the summation,

∞ ∞ ∞ ∞
!
(2) (3) Z 1 3
X Hn3 X Hn Hn X Hn ln (1 − x) X xn
2
+3 2
+2 2
=− dx
n=1
n n=1
n n=1
n 0 x n=1
n
1
ln3 (1 − x) 1
ln4 (x)
Z Z
1−x→x
=− (− ln(1 − x)) dx = dx = 24ζ(5),
0 x 0 1−x
where the last integral is given in (3.5). Thus,

∞ ∞ (2) ∞ (3)
X Hn3 X Hn Hn X Hn
+ 3 + 2 = 24ζ(5). (4.108)
n=1
n2 n=1
n2 n=1
n2
Now multiply both sides (2.22)

∞
ln3 (1 − x) X n 3
− = x Hn − 3Hn Hn(2) + 2Hn(3)
1−x n=1
R1
by − ln(x)
xn−1 ln(x)dx = 1
n2 ,
∞ ∞ (2) ∞ (3)
X Hn3 X Hn Hn X Hn
− 3 + 2
n=1
n2 n=1
n2 n=1
n2
ln3 (1 − x) ln(x)
1 Z 1 3
ln (y) ln(1 − y)
Z
1−x=y
= dx = dy
0 x(1 − x) 0 y(1 − y)
∞
X Hn
=6
n=1
n4
= 18ζ(5) − 3ζ(2)ζ(3).
Then, we have
∞ ∞ (2) ∞ (3)
X Hn3 X Hn Hn X Hn
− 3 + 2 = 18ζ(5) − 6ζ(2)ζ(3). (4.109)
n=1
n2 n=1
n2 n=1
n2
Take the difference of the two relations in (4.109) and (4.108) then divide by 6 to
finish the solution.
Solution (ii) Divide both sides of the generating function in (2.6):

∞
X Hn−1 n 1
y = ln2 (1 − y)
n=1
n 2
Rx xn
by y then integrate from y = 0 to x using 0
y n−1 dy = n ,
∞
1 x ln2 (1 − y)
Z
X Hn−1 n
x = dy. (4.110)
n=1
n2 2 0 y
Next, we consider the following double sum,

∞ X
∞ ∞ ∞ Z 1
X Hn−1 X X Hn−1
2 2
= − xn+k−1 ln(x)dx
n=1
n (n + k) n=1
n2 0
k=1 k=1
∞
! ∞ !
Z1 X X Hn−1
k−1 n
=− ln(x) x x dx
0 n=1
n2
k=1
{recall the result of the second sum from (4.110)}
Z 1 Z x 2
1 1 ln (1 − y)
=− ln(x) dy dx
0 1−x 2 0 y
1 1 ln2 (1 − y)
Z Z 1
ln(x)
=− dx dy
2 0 y y 1−x
Z 1 2
1 ln (1 − y)
=− (− Li2 (1 − y)) dy
2 0 y
Z 1 2
1−y=x 1 ln (x) Li2 (x)
= dx
2 0 1−x

Li2 (x)
expand in series given in (2.3)
1−x
∞ Z 1
1 X (2)
= Hn−1 xn−1 ln2 (x)dx
2 n=1 0
∞ (2) ∞ (2)
X Hn−1 X Hn
= = − ζ(5). (4.111)
n=1
n3 n=1
n3
On the other hand, by setting a = 2 in (1.156), we have

∞
X 1
= ζ(2) − Hn(2) .
(n + k)2
k=1
Hn−1
Multiply both sides by n then take the summation over n ≥ 1,
∞ X
∞ ∞
X Hn−1 X Hn−1 (2)

= ζ(2) − Hn
n=1 k=1
n2 (n + k)2 n=1
n2
∞ ∞ (2) ∞ (2)
X Hn−1 X Hn Hn X Hn
= ζ(2) − + . (4.112)
n=1
n2 n=1
n2 n=1
n3
Combine (4.112) and (4.111),

∞ (2) ∞
X Hn Hn X Hn−1
2
= ζ(5) + ζ(2)
n=1
n n=1
n2
{set x = 1 in (4.110)}
Z 1 2
1 ln (1 − x)
= ζ(5) + ζ(2) dx
2 0 x
Z 1 2
1−x=y 1 ln (y)
= ζ(5) + ζ(2) dy.
2 0 1−y
This integral is calculated in (3.3).
P∞ 3
Hn
4.2.9 n=1 n2
Show that
∞
X Hn3
= 10ζ(5) + ζ(2)ζ(3). (4.113)
n=1
n2
Solution Combining (4.108) and (4.109) then dividing by 2 yields,

∞ ∞ (3)
X Hn3 X Hn
= 21ζ(5) − 3ζ(2)ζ(3) − 2 ,
n=1
n2 n=1
n2
and the solution finishes on recalling the result from (4.168).

For a different approach, see [28, pp. 401–402].
P∞ (2)
Hn
4.2.10 n=1 n4
Show that
∞ (2)
X Hn 1
4
= ζ 2 (3) − ζ(6). (4.114)
n=1
n 3
Solution Set x = 1 in (2.86),

∞ ∞ (2) ∞ (3)
X Hn X Hn X Hn
ζ 2 (3) = 12 + 6 + 2 − 20ζ(6).
n=1
n6 n=1
n4 n=1
n3
Gathering the results from (2.99) and (4.8) finalizes the solution.
P∞ 2
Hn
4.2.11 n=1 n4
Show that
∞
X Hn2 97
4
= ζ(6) − 2ζ 2 (3). (4.115)
n=1
n 24
Solution Divide both sides of (2.71):

1 (2)
Hn2 + Hn
Z
xn−1 ln2 (1 − x)dx =
0 n
3
by n then take the summation over n ≥ 1,
∞ (2) Z 1 2 ∞ Z 1 2
X Hn2 + Hn ln (1 − x) X xn ln (1 − x) Li3 (x)
4
= 3
dx = dx
n=1
n 0 x n=1
n 0 x
∞ Z 1
X Hn−1
=2 xn−1 Li3 (x)dx
n=1
n 0

∞
X Hn 1 ζ(3) ζ(2) Hn
=2 − 2 − 2 + 3
n=1
n n n n n
∞ ∞ ∞ ∞
X Hn X Hn X Hn X Hn2
= 2ζ(3) − 2ζ(2) − 2 + 2 .
n=1
n2 n=1
n3 n=1
n5 n=1
n4
Reorganize the terms, we arrive at

∞ ∞ ∞ ∞ ∞ (2)
X Hn2 X Hn X Hn X Hn X Hn
= −2ζ(3) + 2ζ(2) + 2 − .
n=1
n4 n=1
n2 n=1
n3 n=1
n5 n=1
n4
All these sums are given in (4.4), (4.5), (4.7), and (4.114) respectively.
The two series in (4.114) and (4.115) may be found evaluated differently in [22].
P∞ (4)
Hn
4.2.12 n=1 n2
Show that
∞ (4)
X Hn 37
2
= ζ(6) − ζ 2 (3). (4.116)
n=1
n 12
Solution Putting p = 2 and q = 4 in (2.96),

∞ (2) ∞ (4)
X Hn X Hn
+ = ζ(2)ζ(4) + ζ(6).
n=1
n4 n=1
n2
gather the value from (4.114) to finalize the solution. Note that ζ(2)ζ(4) = 47 ζ(6)
given in (1.64).
(2) 2

P∞ Hn
4.2.13 n=1 n2
Show that 2
∞ (2)
X Hn 35
= ζ(6) − ζ 2 (3). (4.117)
n=1
n2 24
Solution Substitute the value from (4.114) in (2.102).
P∞ (3)
Hn Hn
4.2.14 n=1 n2
Show that
∞ (3)
X Hn Hn 227 3
2
= ζ(6) − ζ 2 (3). (4.118)
n=1
n 48 2
Solution Multiply both sides (2.70):

Z 1
Hn
xn−1 ln(1 − x)dx = − .
0 n
(3)
Hn
∞ ∞
!
(3) Z 1 (3)
X Hn Hn ln(1 − x) X Hn n
2
= − x dx
n=1
n 0 x n=1
n
{recall the generating function in (2.17)}
1
ln(1 − x)
Z
1
− Li4 (x) − ln(1 − x) Li3 (x) − Li22 (x) dx
0 x 2
Z 1 2 Z 1
ln (1 − x) Li3 (x) ln(1 − x) Li4 (x)
= dx − dx
0 x 0 x
1 1 ln(1 − x) Li22 (x)
Z
+ dx
2 0 x
{expand Li3 (x) and ln(1 − x) in series in first and second integral}
∞ Z 1 ∞ Z 1
1
X X 1 1 1
=− 4
x n−1
ln(1 − x)dx + 3
xn−1 ln2 (1 − x)dx − Li32 (x) 0
n=1
n 0 n=1
n 0 6
{make use of (2.70) and (2.71) for the first two integrals}
∞ ∞
!
(2)
1 Hn2
X
X 1 Hn Hn 1
=− 4
− + 3
+ − ζ 3 (2)
n=1
n n n=1
n n n 6
∞ ∞ ∞ (2)
X Hn X Hn2 X Hn 35
= 5
+ 4
+ 4
− ζ(6).
n=1
n n=1
n n=1
n 48
Gather the results from (4.115), (4.114) and (4.7) to finish the solution. Note that we
used ζ 3 (2) = 35
8 ζ(6) given in (1.63).
Also check [28, p. 414–419] for an alternative solution.
P∞ 2H (2)
Hn n
4.2.15 n=1 n2
Show that
∞ (2)
X Hn2 Hn 41
2
= ζ(6) + 2ζ 2 (3). (4.119)
n=1
n 12
Solution Multiply both sides in (2.72):

1 (2) (3)
Hn3 + 3Hn Hn + 2Hn
Z
− xn−1 ln3 (1 − x)dx =
0 n
Hn
∞ ∞ (2) ∞ (3)
X Hn4 X Hn2 Hn X Hn Hn
+ 3 + 2
n=1
n2 n=1
n2 n=1
n2
∞
Z 1 3 !
ln (1 − x) X Hn n
=− x dx
0 x n=1
n
Z 1 3
ln (1 − x) 1 2
=− ln (1 − x) + Li2 (x) dx
0 x 2
Z 1 5 Z 1 3
1 ln (1 − x) ln (1 − x) Li2 (x)
=− dx − dx
2 0 x 0 x
| {z }
1−x=y
1
ln5 (y) 1
ln3 (1 − x) Li2 (x)
Z Z
1
=− dy − dx.
2 0 1−y 0 x
Gathering the results from (3.6) and (3.156),

∞ ∞ (2) ∞ (3)
X Hn4 X Hn2 Hn X Hn Hn 121
2
+ 3 2
+ 2 2
= ζ(6) + 6ζ 2 (3). (4.120)
n=1
n n=1
n n=1
n 2
Next, divide both sides of (2.22):

∞
ln3 (1 − x) X n 3
− = x Hn − 3Hn Hn(2) + 2Hn(3)
1−x n=1
yn
Ry
xn−1 dx = n ,
∞ (2) (3)
y
ln3 (1 − x) Hn3 − 3Hn Hn + 2Hn n
Z X
− dx = y .
0 x(1 − x) n=1
n
R1
Multiply both sides by − ln(1−y)
y then integrate using − 0
y n−1 ln(1 − y)dy = Hn
n ,
∞ ∞ (2) ∞ (3) Z 1Z y 3
X Hn4 X Hn2 Hn X Hn Hn ln (1 − x) ln(1 − y)
2
− 3 2
+ 2 2
= dxdy
n=1
n n=1
n n=1
n 0 0 xy(1 − x)
1 3 Z 1 Z 1 3
ln (1 − x) ln(1 − y) ln (1 − x)
Z
= dy dx = (Li2 (x) − ζ(2)) dx
0 x(1 − x) x y 0 x(1 − x)

1 1 1
write = +
x(1 − x) x 1−x
Z 1 3 Z 1 3
ln (1 − x) ln (1 − x)
= (Li2 (x) − ζ(2)) dx + (Li2 (x) − ζ(2)) dx
0 x 1−x
|0 {z }
IBP
3 3
1
ln (1 − x) Li2 (x) 1
ln (1 − x) 1 1 ln5 (1 − x)
Z Z Z
= dx − ζ(2) dx − dx
0 x x 4 0 x
|0 {z } | {z }
1−x=y 1−x=y
3 3
1
ln (1 − x) Li2 (x) 1 1
ln5 (y)
Z Z Z
ln (y) 1
= dx − ζ(2) dy − dy.
0 x 0 1−y 4 0 1−y
Put together the results from (3.156), (3.4), and (3.6),

∞ ∞ (2) ∞ (3)
X Hn4 X Hn2 Hn X Hn Hn
2
− 3 2
+ 2 2
= 40ζ(6) − 6ζ 2 (3). (4.121)
n=1
n n=1
n n=1
n
The solution completes on combining (4.120) and (4.121) then dividing by 6.
P∞ 4
Hn
4.2.16 n=1 n2
Show that
∞
X Hn4 979
2
= ζ(6) + 3ζ 2 (3). (4.122)
n=1
n 24
Solution Combine (4.120) and (4.121) then divide by 2,

∞ ∞ (3)
X Hn4 101 X Hn Hn
2
= ζ(6) − 2 2
.
n=1
n 4 n=1
n
Plug in the result form (4.118) to finish the solution.

For different methods of computing (4.122) and (4.119), see [28, pp. 421–427].
P∞ (2)
Hn Hn
4.2.17 n=1 n3
Show that
∞ (2)
X Hn Hn 101 5
3
=− ζ(6) + ζ 2 (3). (4.123)
n=1
n 48 2

1 (2) (3)
Hn3 + 3Hn Hn + 2Hn
Z
xn−1 ln3 (1 − x)dx = −
0 n

∞ ∞ (2) ∞ (3)
X Hn3 X Hn Hn X Hn
+ 3 + 2
n=1
n3 n=1
n3 n=1
n3
∞
!
ln3 (1 − x) X xn
1 Z 1 3
ln (1 − x) Li2 (x)
Z
=− 2
dx = − dx. (4.124)
0 x n=1
n 0 x
On the other hand, multiply both sides of (2.22):

∞
X ln3 (1 − x)
Hn3 − 3Hn Hn(2) + 2Hn(3) xn = −
n=1
1−x
ln(x) R1
2 0
xn−1 ln2 (x)dx = 1
n3 ,
∞ ∞ (2) ∞ (3)
X Hn3 X Hn Hn X Hn
− 3 + 2
n=1
n3 n=1
n3 n=1
n3
Z 1 2
ln (x) ln3 (1 − x) 1 1 ln3 (y) ln2 (1 − y)
Z
1 1−x=y
=− dx = − dy
2 0 x(1 − x) 2 0 y(1 − y)
ln2 (1 − y)

1−y
∞ Z 1
1 X
=− 2
Hn − Hn (2)
y n−1 ln3 (y)dy
2 n=1 0
∞ 3!
1X
2 (2)
=− Hn − Hn − 4
2 n=1 n
∞ ∞ (2)
X Hn2 X Hn
=3 − 3 . (4.125)
n=1
n4 n=1
n4
Take the difference of the two relations in (4.124) and (4.125) then divide by 6,
∞ (2) ∞ (2) ∞
1 X Hn2 1 1 ln3 (1 − x) Li2 (x)
Z
X Hn Hn 1 X Hn
= − − dx.
n=1
n3 2 n=1 n4 2 n=1 n4 6 0 x
and the solution completes on gathering the results from (4.114), (4.115), and (3.156).
For another approach, check [28, p. 411–414] .
P∞ 3
Hn
4.2.18 n=1 n3
Show that
∞
X Hn3 93 5
3
= ζ(6) − ζ 2 (3). (4.126)
n=1
n 16 2

∞ ∞ ∞ (2) ∞ (3)
X Hn3 3 X Hn2 3 X Hn X Hn
= − − 2
n=1
n3 2 n=1 n4 2 n=1 n4 n=1
n3
1 1 ln3 (1 − x) Li2 (x)
Z
− dx.
2 0 x
These terms are calculated in (4.114), (4.115), (2.99), and (3.156).

A different solution may be found in [22].
P∞ (2)
Hn
4.2.19 n=1 n5
Show that
∞ (2)
X Hn
= 5ζ(2)ζ(5) + 2ζ(3)ζ(4) − 10ζ(7). (4.127)
n=1
n5

∞
X 1
Hn(3) = ζ(3) − .
(n + k)3
k=1
Divide both sides by n4 then take the summation over n ≥ 1,

∞ ∞ ∞
" #
(3)
X Hn X 1 X 1
= ζ(3) −
n=1
n4 n=1
n4 (n + k)3
k=1
{distribute then change the order of summation}
∞
"∞ #
X X 1
= ζ(3)ζ(4) −
n4 (n + k)3
k=1 n=1

1
decompose 4 by partial fraction
n (n + k)3
∞ ∞
"
X X 10 1 1 6 4 3
= ζ(3)ζ(4) − − 6 − + 5 2+ 5 2
− 4 3
k n n + k k n k (n + k) k n
k=1 n=1

1 1
+ 4 3
+ 3 4
k (n + k) k n
∞
( )
X 1 1
use − = Hk given in (1.155) for the first term and use
n=1
n n+k
(∞ )
X 1 (a)
= ζ(a) − Hk given in (1.154) for the third and fifth terms
n=1
(k + n)a
∞
" (2)
X 10Hk 6ζ(2) ζ(2) − Hk 3ζ(3)
= ζ(3)ζ(4) − − 6 + 5
+ 4 5
−
k k k k4
k=1
(3)
#
ζ(3) − Hk ζ(4)
+ + 3 .
k4 n
(3)
P∞ Hk
On rearranging the terms, the sum k=1 k4 cancels out from both sides,
∞ (2) ∞
X H k 5 1 X Hk
= ζ(2)ζ(5) − ζ(3)ζ(4) − ,
k5 2 2 k6
k=1 k=1
and the remaining sum is given in (4.8). For an alternative solution, see (4.69).
P∞ 2
Hn
4.2.20 n=1 n5
Show that
∞
X Hn2 5
5
= 6ζ(7) − ζ(2)ζ(5) − ζ(3)ζ(4). (4.128)
n=1
n 2

1 (2)
Hn2 + Hn
Z
xn−1 ln2 (1 − x)dx =
0 n
4
∞ ∞ ∞
!
(2) Z 1 2
X Hn2 X Hn ln (1 − x) X xn
+ = dx
n=1
n5 n=1
n5 0 x n=1
n4
Z 1 2
ln (1 − x) Li4 (x)
= dx
0 x
∞
Hn−1 1 n−1
X Z
=2 x Li4 (x)dx
n=1
n 0
{recall the result from (3.105) for the integral}

∞
X Hn 1 ζ(4) ζ(3) ζ(2) Hn
=2 − 2 − 2 + 3 − 4 .
n=1
n n n n n n
Reorder the terms, we reach

∞ ∞ (2)
X Hn2 X Hn
3 +
n=1
n5 n=1
n5
∞ ∞ ∞ ∞
X Hn X Hn X Hn X Hn
=2 6
+ 2ζ(4) 2
− 2ζ(3) 3
+ 2ζ(2) − 2ζ(2)ζ(5).
n=1
n n=1
n n=1
n n=1
n4
Gather the results from (4.127), (4.4), (4.5), (4.6), and (4.8) to end the solution.
Also check [28, p. 396–398] for a different method.
P∞ (3)
Hn
4.2.21 n=1 n4
Show that
∞ (3)
X Hn
= 18ζ(7) − 10ζ(2)ζ(5). (4.129)
n=1
n4

∞ ∞ (2) ∞ (3)
Li23 (x) = 12 x + 6 x + 2 x − 20 Li6 (x)
n=1
n5 n=1
n4 n=1
n3
R1 1
by x then integrate using 0
xn−1 dx = n,
∞ ∞ (2) ∞ (3) Z 1 2
X Hn X Hn X Hn Li3 (x)
12 6
+6 5
+2 4
− 20ζ(7) = dx
n=1
n n=1
n n=1
n 0 x
{expand Li3 (x) in series}
∞ Z 1
X 1
= 3
xn−1 Li3 (x)dx
n=1
n 0

∞ ∞
X 1 ζ(3) ζ(2) Hn X Hn
= − 2 + 3 = ζ(3)ζ(4) − ζ(2)ζ(5) + .
n=1
n3 n n n n=1
n6

∞ (3) ∞ ∞ (2)
X Hn 1 1 11 X Hn X Hn
= ζ(3)ζ(4) − ζ(2)ζ(5) + 10ζ(7) − − 3 .
n=1
n4 2 2 2 n=1 n6 n=1
n5
These two series are given in (4.127) and (4.8).
P∞ (4)
Hn
4.2.22 n=1 n3
Show that
∞ (4)
X Hn
= ζ(3)ζ(4) + 10ζ(2)ζ(5) − 17ζ(7). (4.130)
n=1
n3
Solution Setting p = 3 and q = 4 in (2.96),

∞ (4) ∞ (3)
X Hn X Hn
= ζ(4)ζ(3) + ζ(7) .
n=1
n3 n=1
n4
The latter sum is computed in (4.129).
P∞ 2H (2)
Hn n
4.2.23 n=1 n3
Show that
∞ (2)
X Hn2 Hn 19
3
= ζ(3)ζ(4) − 2ζ(2)ζ(5) − 7ζ(7). (4.131)
n=1
n 2
Solution Divide both sides in (2.73):

1 (2) (3) (2) (4)
Hn4 + 6Hn2 Hn + 8Hn Hn + 2(Hn )2 + 6Hn
Z
xn−1 ln4 (1 − x)dx =
0 n
by n2 then take the summation,

∞ 2
X 1 4 2 (2) (3)

(2) (4)
Hn + 6H H
n n + 8H H
n n + 3 H n + 6Hn
n=1
n3
∞
Z 1 4 ! Z 1 4
ln (1 − x) X xn ln (1 − x) Li2 (x)
= 2
dx = dx. (4.132)
0 x n=1
n 0 x
To establish another relation, multiply both sides of (2.27):

∞
ln4 (1 − x)
X 2
4 2 (2) (3) (2) (4)
Hn − 6Hn Hn + 8Hn Hn + 3 Hn − 6Hn xn =
n=1
1−x
ln2 (x) R1
2 0
xn−1 ln2 (x)dx = 1
n3 , we obtain
∞ 2
X 1 4 2 (2) (3)

(2) (4)
H n − 6H H
n n + 8H n Hn + 3 Hn − 6H n
n=1
n3
1 1 ln2 (x) ln4 (1 − x)
Z 1 2
ln (1 − y) ln4 (y)
Z
1−x=y 1
= dx = dx
2 0 x(1 − x) 2 0 y(1 − y)
ln2 (1 − y)

1−y
∞ Z 1
1 X 2
= Hn − Hn(2) y n−1 ln4 (y)dy
2 n=1 0
∞ ∞
X H2 ∞ (2)
1X 4! n
X Hn
= Hn2 − Hn(2) = 12 − 12 . (4.133)
2 n=1 n5 n=1
n5 n=1
n5
Take the difference of (4.132) and (4.133) then divide by 12,

∞ (2) Z 1 4 ∞ (4) ∞ ∞ (2)
X Hn2 Hn 1 ln (1 − x) Li2 (x) X Hn X Hn2 X Hn
= dx − − + .
n=1
n3 12 0 x n=1
n3 n=1
n5 n=1
n5
Substitute the results from (3.157), (4.130), (4.128), and (4.127) to complete the
solution. Check [28, p. 456] for another solution.
P∞ (2)
Hn
4.2.24 n=1 n7
Show that
∞ (2)
X Hn 35
7
= 7ζ(2)ζ(7) + 2ζ(3)ζ(6) + 4ζ(4)ζ(5) − ζ(9). (4.134)
n=1
n 2

∞
X 1
Hn(2) = ζ(2) − .
(n + k)2
k=1
Divide both sides by n7 then consider the summation,

∞ ∞ ∞
" #
(2)
X Hn X 1 X 1
= ζ(2) −
n=1
n7 n=1
n7 (n + k)2
k=1
∞ ∞
" #
X X 1
= ζ(2)ζ(7) −
n7 (n + k)2
k=1 n=1
∞ ∞
"
X X 7 1 1 6 1 5
= ζ(2)ζ(7) − 8
− − 7 2− 7 2
+ 6 3
k n n + k k n k (n + k) k n
k=1 n=1

4 3 2 1
− 5 4+ 4 5− 3 6+ 2 7
k n k n k n k n
∞
X 7Hk 6ζ(2) 1
(2)
5ζ(3)
= ζ(2)ζ(7) − − − ζ(2) − H k +
k8 k7 k7 k6
k=1

4ζ(4) 3ζ(5) 2ζ(6) ζ(7)
− 5 + − + 2
k k4 k3 k
∞ ∞ (2)
X Hk X H k
= ζ(2)ζ(7) − 7 + 6ζ(2)ζ(7) − − 3ζ(3)ζ(6) + ζ(4)ζ(5).
k8 k7
k=1 k=1
Reorganizing the terms yields

∞ (2) ∞
X Hn X Hk
2 7
= 7ζ(2)ζ(7) − 3ζ(3)ζ(6) + ζ(4)ζ(5) − 7 .
n=1
n k8
k=1
The latter sum is given in (4.9). Another approach may be found in (4.70).
Remark: For integers p and q, where p > 1, q 6= p, and p + q is even > 6, there
∞ (p)
X Hn
does not exist a closed form for the series .
n=1
nq
4.3 Alternating Harmonic Series

P∞ (−1)n Hn
4.3.1 n=1 n
Show that
∞
X (−1)n Hn 1 1
= ln2 (2) − ζ(2). (4.135)
n=1
n 2 2

∞ Z 1
X (−1)n Hn ln(1 + x)
=− dx
n=1
n 0 x(1 + x)
y Z 12
x= 1−y ln(1 − y) 1 1
= − dy = Li2 (y) 0 = − Li2
2
,
0 y 2
and the solution finalizes on using Li2 12 = 12 ζ(2) − 12 ln2 (2) given in (1.120).

For a different approach, set x = −1 in (2.7) and use Li2 (−1) = − 12 ζ(2).
P∞ (−1)n H2n
4.3.2 n=1 n
Show that
∞
X (−1)n H2n 5 1
= − ζ(2) + ln2 (2). (4.136)
n=1
n 8 4

∞ ∞
X (−1)n H2n X (−1)n H2n
=2
n=1
n n=1
2n

Hn
let an = in the series tramsformation in (1.9)
n
∞ n
X i Hn
= 2R
n=1
n
{set x = i in the generating function in (2.7)}

1 2
= 2 R Li2 (i) + ln (1 − i) .
2
The values of these two terms are given in (1.108) and (1.25).
4.3. Alternating Harmonic Series 239
Solution (ii) Set x = 1 in (2.42):

∞
X 2H2n − Hn 2n
(−1)n x = −2 arctan2 (x),
n=1
n
we get
∞
X 2H2n − Hn
(−1)n = −2 arctan2 (1)
n=1
n
π 2 π2 3
= −2 =− = − ζ(2).
4 8 4
Substitute the result from (4.135) to finalize the solution.
P∞ (−1)n Hn
4.3.3 n=1 n2
Show that
∞
X (−1)n Hn 5
2
= − ζ(3). (4.137)
n=1
n 8
Solution (i) Setting x = −1 in (2.8),

∞
X (−1)n Hn 1
2
= Li3 (−1) − Li3 (2) + ln(2) Li2 (2) + ln(−1) ln2 (2) + ζ(3).
n=1
n 2
The values of Li3 (−1), Li3 (2), and Li2 (2) are given in (1.103), (1.141), and (1.140).
Solution (ii) Replace x by −x in (2.7),
∞
X (−)n Hn n 1
x = Li2 (−x) + ln2 (1 + x).
n=1
n 2
R1 1
Divide both sides by x then integrate using 0
xn−1 dx = n,
∞
1 1 ln2 (1 + x)
Z 1
(−1)n Hn
Z
X Li2 (−x)
= dx + dx
n=1
n2 0 x 2 0 x
1 1 ln2 (1 + x)
Z
= Li3 (−1) + dx.
2 0 x
This integral is calculated in (3.38) and Li3 (−1) = − 34 ζ(3) given in (1.103).
An alternative solution may be found in [28, pp. 508–509] and (4.22).
P∞ (−1)n H2n
4.3.4 n=1 n2
Show that
∞
X (−1)n H2n 23
2
= ζ(3) − πG. (4.138)
n=1
n 16
Solution
∞ ∞
X (−1)n H2n X (−1)n H2n
2
= 4
n=1
n n=1
(2n)2

Hn
let an = 2 in (1.9)
n
∞ n
X i Hn
= 4R
n=1
n2

1
= 4 R Li3 (i) − Li3 (1 − i) + ln(1 − i) Li2 (1 − i) + ln(i) ln2 (1 − i) + ζ(3) .
2
By using the values in (1.122), (1.25), and (1.26), we find:
π2
3
2 π π 2
ln(i) ln (1 − i) = ln(2) − − ln (2) i; (4.139)
8 32 8
π2 π3

π 1 π 2
ln(1 − i) Li2 (1 − i) = − G − ln(2) − ln(2)G + + ln (2) i.
4 32 2 64 8
(4.140)
Collect the values from (1.109), (4.139), and (4.140) to finish the solution.
(2)
P∞ (−1)n Hn
4.3.5 n=1 n
Show that
∞ (2)
X (−1)n Hn 1
= ln(2)ζ(2) − ζ(3). (4.141)
n=1
n 2
Solution (i) Setting x = −1 produces

∞ ∞ (2)
X (−1)n Hn X (−1)n Hn
− ln(2) Li2 (−1) = 2 + − 3 Li3 (−1).
n=1
n2 n=1
n
The solution completes on collecting the values from (4.137), (1.102), and (1.103).
Solution (ii) Put a = 1 in (3.92),

∞ (2)
ln(x) ln 1+x
Z 1
X (−1)n Hn 2
=− dx
n=1
n 0 1−x
Z 1 Z 1
ln(x) ln(1 + x) ln(x)
=− dx + ln(2) dx.
0 1 − x 0 1 −x
(3)
P∞ (−1)n Hn
4.3.6 n=1 n
Show that
∞ (3)
X (−1)n Hn 3 19
= ln(2)ζ(3) − ζ(4). (4.142)
n=1
n 4 16
Solution Setting x = −1 in (2.17),

∞ (3)
X (−1)n Hn 1
= Li4 (−1) − ln(2) Li3 (−1) − Li22 (−1).
n=1
n 2
The values of these polylogarithm functions are given (1.104), (1.103), and (1.102).
For another solution, see (4.81).
P∞ (−1)n Hn
4.3.7 n=1 n3
Show that
∞
(−1)n Hn

X 1 11 7 1 1
3
= 2 Li4 − ζ(4)+ ln(2)ζ(3)− ln2 (2)ζ(2)+ ln4 (2).
n=1
n 2 4 4 2 12
(4.143)
Solution Put a = 3 in (3.90),

∞
(−1)n Hn 1 1 ln2 (x) ln(1 + x)
X Z
=− dx
n=1
n3 2 0 x(1 + x)

Z 1 2 y
y
x= 1−y 1 2 ln 1−y ln(1 − y)
= dy
2 0 y
1 Z 1 Z 12
ln2 (y) ln(1 − y) 1 2 ln3 (1 − y) ln(y) ln2 (1 − y)
Z
1 2
= dy + dy − dy .
2 y 2 0 y y
|0 {z } | {z } |0 {z }
I1 I2 I3
For I1 + I2 , integrate I1 by parts and let 1 − y = x in I2 ,

1
ln3 (x) 1
ln3 (x)
Z Z
1 4 1 2
I1 + I2 = ln (2) + dx + dx
3 3 0 1−x 1 1−x
|2 {z }
R1 R 1/2
0
− 0
1
Z 1 3 Z 3
1 4 ln (x) 2 ln (x) 2
= ln (2) + dx − dx
3 0 1 − x 3 0 1−x

1 7 2
= 4 Li4 − 6ζ(4) + ln(2)ζ(3) − ln2 (2)ζ(2) + ln4 (2).
2 2 3
For I3 ,
1
ln2 (y) ln(1 − y)
Z
IBP 1 4 2
I3 = ln (2) + dy
2 0 1−y
1
ln(y) ln2 (1 − y)
Z
1−y→y 1 4
= ln (2) + dy
2 1
2
y
( 1
)
ln(y) ln2 (1 − y)
Z 2
add I3 := dy to both sides then divide by 2
0 y
Z 1
1 2 ln(y) ln2 (1 − y) 1 1 ln(y) ln2 (1 − y)
Z
1
= ln4 (2) + dy + dy
4 2 0 y 2 21 y
1 1 ln(y) ln2 (1 − y)
Z
1
= ln4 (2) + dy
4 2 0 y
| {z }
1−y=x
1 1 ln(1 − x) ln2 (x)

Z
1
= ln4 (2) + dx
4 2 0 1−x
{set a = 2 in (3.87) to get the latter integral}
∞
1 4 X Hn
= ln (2) + 2ζ(4) − 2
2 n=1
n3
1 1
= ln4 (2) − ζ(4).
4 4
Gather the three integrals to complete the solution. A different method is by putting
x = −1 in (2.15). For another approach, check [29].
∞
X (−1)n Hn
Remark: For odd q > 3, there does not exist a closed form for .
n=1
nq
(2)
P∞ (−1)n Hn
4.3.8 n=1 n2
Show that
∞ (2)
(−1)n Hn

X 1 51 7 1
2
= −4 Li4 + ζ(4)− ln(2)ζ(3)+ln2 (2)ζ(2)− ln4 (2).
n=1
n 2 16 2 6
(4.144)
Solution Put x = −1 in (2.83),

∞ (2) ∞
X (−1)n Hn 1 2 X (−1)n Hn
2
= 3 Li 4 (−1) + Li2 (−1) − 2 .
n=1
n 2 n=1
n3
Substituting the results from (4.143), (1.102), and (1.104) completes the solution.
Also Check [28, pp. 505–506] for a different method.
P∞ (−1)n Hn
2
4.3.9 n=1 n2
Show that
∞
(−1)n Hn2

X 1 41 7 1 1
2
= 2 Li 4 − ζ(4)+ ln(2)ζ(3)− ln2 (2)ζ(2)+ ln4 (2).
n=1
n 2 16 4 2 12
(4.145)
Solution Replace x by −x in (2.6),

∞
X Hn−1 n 1
(−1)n x = ln2 (1 + x).
n=1
n 2
Multiplying both sides by − ln(1−x)

x then integrating from x = 0 to 1,
∞
1
ln2 (1 + x) ln(1 − x) (−1)n Hn−1 1
Z Z
1 X
− dx = −xn−1 ln(1 − x)dx
2 0 x n=1
n 0

∞ 1 ∞ ∞
Hn − (−1)n Hn2 X (−1)n Hn
X
X Hn
= (−1)n n
= − .
n=1
n n n=1
n2 n=1
n3

∞ ∞
(−1)n Hn2 (−1)n Hn 1 1 ln2 (1 + x) ln(1 − x)
X X Z
= − dx.
n=1
n2 n=1
n3 2 0 x
The values of these two terms are computed in (3.117) and (4.143).
Also check [28, pp. 506–508] for a different solution.
(2)
P∞ (−1)n Hn Hn
4.3.10 n=1 n
Show that
∞ (2)
(−1)n Hn Hn

X 1 7 1
= −2 Li4 + ζ(4) − ln(2)ζ(3) + ln2 (2)ζ(2)
n=1
n 2 8 4
1 4
− ln (2). (4.146)
12
Solution Multiply both sides of (2.72):

1 (2) (3)
Hn3 + 3Hn Hn + 2Hn
Z
xn−1 ln3 (1 − x)dx = −
0 n
by (−1)n−1 then take the summation,

∞ ∞ (2) ∞ (3)
X (−1)n Hn3 X (−1)n Hn Hn X (−1)n Hn
+3 +2
n=1
n n=1
n n=1
n
∞
ln3 (1 − x)
Z 1 X Z 1
= ln3 (1 − x) (−x)n−1 dx = dx
0 n=1 0 1+x

1
= −6 Li4 . (4.147)
2
On the other hand, replace x by −x in (2.22),

∞
X ln3 (1 + x)
(−1)n Hn3 − 3Hn Hn(2) + 2Hn(3) xn = − .
n=1
1+x
R1 1
xn−1 dx = n,
∞ ∞ (2) ∞ (3)
X (−1)n Hn3 X (−1)n Hn Hn X (−1)n Hn
−3 +2
n=1
n n=1
n n=1
n
Z 1 3 Z 1 3 Z 1 3
ln (1 + x) ln (1 + x) ln (1 + x)
=− dx = dx − dx
0 x(1 + x) 0 1+x 0 x
{the first integral is ln4 (2)/4 and the second one is given in (3.39)}

1 21 3 1
= 6 Li4 − 6ζ(4) + ln(2)ζ(3) − ln2 (2)ζ(2) + ln4 (2). (4.148)
2 4 2 2
Take the difference of (4.147) and (4.148) then divide by 6 to finalize the solution.
P∞ (−1)n Hn
3
4.3.11 n=1 n
Show that
∞
X (−1)n Hn3 5 9 3 1
= − ζ(4) + ln(2)ζ(3) − ln2 (2)ζ(2) + ln4 (2). (4.149)
n=1
n 8 8 4 4
Solution Adding (4.147) and (4.148),

∞ ∞ (3)
X (−1)n Hn3 X (−1)n Hn
2 +4
n=1
n n=1
n
21 3 1
= −6ζ(4) + ln(2)ζ(3) − ln2 (2)ζ(2) + ln4 (2).
4 2 2
The remaining sum is calculated in (4.142).
P∞ (−1)n Hn
4.3.12 n=1 n4
Show that
∞
X (−1)n Hn 1 59
4
= ζ(2)ζ(3) − ζ(5). (4.150)
n=1
n 2 32
Solution Putting a = 3 in (3.90) gives

∞
(−1)n Hn 1 1 ln3 (x) ln(1 + x)
X Z
= dx. (4.151)
n=1
n4 6 0 x(1 + x)
Let’s calculate the integral:

∞ ∞
1
ln3 (x) ln(1 + x) ln3 (x) ln(1 + x) ln3 (x) ln(1 + x)
Z Z Z
dx = dx − dx
0 x(1 + x) 0 x(1 + x) x(1 + x)
|1 {z }
x→1/x
∞ 3 3
1 1
ln4 (x)
Z Z Z
ln (x) ln(1 + x) ln (x) ln(1 + x)
= dx + dx − dx.
0 x(1 + x) 0 1+x 0 1+x
R1 ln3 (x) ln(1+x) R1 3 R 1 3 ln(1+x)

Add 0 x(1+x) dx := 0 ln (x) ln(1+x)
x dx − 0 ln (x)1+x dx to both sides
R 1 ln3 (x) ln(1+x)
and notice 0 1+x dx nicely cancels out from both sides, we obtain
ln3 (x) ln(1 + x)

1
Z
2 dx
0 x(1 + x)
∞
ln3 (x) ln(1 + x)
Z Z 1 4 Z 1 3
ln (x) ln (x) ln(1 + x)
= dx − dx + dx
0 x(1 + x) 0 1 + x 0 x
| {z }
IBP
Z ∞ 3 Z 1 4
ln (x) ln(1 + x) 5 ln (x)
= dx − dx.
x(1 + x) 4 1+x
|0 {z } 0
I
To evaluate I, make the substitution x = 1−tt ,

Z 1 ln3 t ln(t) Z 1 4 Z 1 3
1−t ln (t) ln (t) ln(1 − t)
I= dt = dt − 3 dt
0 1−t 0 1−t 0 1−t
Z 1 2
ln (t) ln2 (1 − t) ln(t) ln3 (1 − t)
Z 1
+3 dt − dt
0 1−t 0 1−t
| {z } | {z }
IBP IBP
1
ln4 (t) ln3 (t) ln(1 − t)
1
Z Z
= dt − 3 dt
0 1−t 0 1−t
Z 1 3
ln (1 − t) ln(t) 1 1 ln4 (1 − t)
Z
+2 dt − dt
0 t 4 0 t
| {z } | {z }
1−t→t 1−t→t
1 4 1 3
ln (t) ln(1 − t)
Z Z
3 ln (t)
= dt − dt.
4 0 1−t 0 1−t
Substitute the result of I back,
ln3 (x) ln(1 + x) 1

Z
dx
0 x(1 + x)
3 1 ln4 (x) 5 1 ln4 (x) 1 1 ln3 (x) ln(1 − x)
Z Z Z
= dx − dx − dx
8 0 1−x 8 0 1+x 2 0 1−x
{gather the results from (3.5), (3.11), and (3.124)}
177
= 3ζ(2)ζ(3) + ζ(5).
16
Plugging this integral in (4.151) finalizes the solution. For another method, see [29].
(2)
P∞ (−1)n Hn
4.3.13 n=1 n3
Show that
∞ (2)
X (−1)n Hn 11 5
3
= ζ(5) − ζ(2)ζ(3). (4.152)
n=1
n 32 8

∞ ∞ (2)
X (−1)n Hn n X (−1)n Hn n
Li22 (−x) =4 3
x +2 2
x − 6 Li4 (−x).
n=1
n n=1
n
R1 1
xn−1 dx = n,
1 ∞ ∞ (2)
Li22 (−x) (−1)n Hn (−1)n Hn
Z X X 45
dx = 4 4
+ 2 3
+ ζ(5).
0 x n=1
n n=1
n 8
Substitute the relation involving the integral from (3.153) then rearrange the terms,
∞ (2) ∞ ∞
X (−1)n Hn 45 5 7 X Hn X (−1)n Hn
= ζ(5) + ζ(2)ζ(3) + − .
n=1
n3 16 16 16 n=1 n4 n=1
n4
These two series are computed in (4.6) and (4.150).
P∞ (−1)n Hn
2
4.3.14 n=1 n3
Show that
∞
(−1)n Hn2

X 1 1 19 11
3
= −4 Li5 − 4 ln(2) Li4 + ζ(5) + ζ(2)ζ(3)
n=1
n 2 2 32 8
7 2 2 5
− ln2 (2)ζ(3) + ln3 (2)ζ(2) − ln (2). (4.153)
4 3 15
Solution Replace x by −x in (2.12) then multiply through by −1,

∞
ln2 (1 + x)

X
2 n
(2) 2Hn 2
(−1) Hn − Hn − + 2 xn−1 = − .
n=1
n n 1+x
R1
Multiply both sides by 1
2 ln2 (x) then integrate using 1
2 0
xn−1 ln2 (x)dx = 1
n3 ,
∞
X
n 2Hn 2 1
(−1) Hn2 − Hn(2) − + 2
n=1
n n n3
1
ln2 (1 + x) ln2 (x) 1
ln(x) ln3 (1 + x)
Z Z
1 IBP 1
=− dx = dx.
2 0 1+x 3 0 x
Distribute then reorder the terms,
∞ ∞ (2) ∞
X (−1)n Hn2 X (−1)n Hn X (−1)n Hn
3
= −2 Li5 (−1) + 3
+ 2
n=1
n n=1
n n=1
n4
1 1 ln(x) ln3 (1 + x)
Z
+ dx.
3 0 x
These terms are calculated in (1.105), (4.152), (4.150), and (3.123) respectively.
Check [28, pp. 517–519] for an alternative approach.
(4)
P∞ (−1)n Hn
4.3.15 n=1 n
Show that
∞ (4)
X (−1)n Hn 7 3
= −2ζ(5) + ln(2)ζ(4) + ζ(2)ζ(3). (4.154)
n=1
n 8 8
Solution Set a = 4 then replace x by −x in (2.2),

∞
X Li4 (−x)
(−1)n Hn(4) xn = .
n=1
1+x
R1 1
xn−1 dx = n,
∞ (4) Z 1 Z 1 Z 1
X (−1)n Hn Li4 (−x) Li4 (−x) Li4 (−x)
= dx = dx − dx
n 0 x(1 + x) 0 x 1+x
n=1 |0 {z }
IBP
Z 1
ln(1 + x) Li3 (−x)
= Li5 (−1) − ln 2 Li4 (−1) + dx
0 x
Z 1 2
IBP Li2 (−x)
= Li5 (−1) − ln(2) Li4 (−1) − Li2 (−1) Li3 (−1) + dx
0 x
Z 1 2
15 7 3 Li2 (−x)
= − ζ(5) + ln(2)ζ(4) − ζ(2)ζ(3) + dx
16 8 8 0 x
{recall the relation involving the latter integral from (3.153)}
∞ ∞
15 7 1 7 X Hn X (−1)n Hn
= − ζ(5) + ln(2)ζ(4) + ζ(2ζ(3) + 4
+ 2 . (4.155)
16 8 4 8 n=1 n n=1
n4
The solution finalizes on recalling the results from (4.6) and (4.150).
For different approaches, see [28, p. 516] and (4.82).
(3)
P∞ (−1)n Hn
4.3.16 n=1 n2
Show that
∞ (3)
X (−1)n Hn 21 3
2
= ζ(5) − ζ(2)ζ(3). (4.156)
n=1
n 32 4

∞ (3)
X (−1)n Hn n 1
x = Li4 (−x) − ln(1 + x) Li3 (−x) − Li22 (−x).
n=1
n 2
R1 1
xn−1 dx = n,
∞ (3)
X (−1)n Hn
n=1
n2
1 1 1
Li22 (−x)
Z Z Z
Li4 (−x) ln(1 + x) Li3 (−x) 1
= dx − dx − dx
0 x x 2 x
|0 {z } 0
IBP
3 1 Li22 (−x)
Z
= Li5 (−1) + Li2 (−1) Li3 (−1) − dx
2 0 x
3 1 Li22 (−x)
Z
3 15
= ζ(2)ζ(3) − ζ(5) − dx.
8 16 2 0 x
The latter integral is given in (3.152). Also, check [28, pp. 513–515] for different
methods for both series in (4.156) and (4.152).
(2)
P∞ (−1)n Hn Hn
4.3.17 n=1 n2
Show that
∞ (2)
(−1)n Hn Hn

X 1 1 23 7
2
= 4 Li 5 + 4 ln(2) Li 4 − ζ(5) + ln2 (2)ζ(3)
n=1
n 2 2 8 4
2 15 2 5
− ln3 (2)ζ(2) − ζ(2)ζ(3) + ln (2). (4.157)
3 16 15

1 (2) (3)
Hn3 + 3Hn Hn + 2Hn
Z
xn−1 ln3 (1 − x)dx = −
0 n
n
by − (−1)
n then consider the summation,
∞ ∞ (2) ∞ (3)
X (−1)n Hn3 X (−1)n Hn Hn X (−1)n Hn
+ 3 + 2
n=1
n2 n=1
n2 n=1
n2
∞
Z 1 3 !
ln (1 − x) X (−x)n
=− dx
0 x n=1
n
Z 1 3
ln (1 − x) ln(1 + x)
= dx. (4.158)
0 x
On the other hand, replace x by −x in (2.22),

∞
X ln3 (1 + x)
(−1)n Hn3 − 3Hn Hn(2) + 2Hn(3) xn = − .
n=1
1+x
R1
Multiply through by − ln(x)
xn−1 ln(x)dx = 1
n2 ,
∞ ∞ (2) ∞ (3)
X (−1)n Hn3 X (−1)n Hn Hn X (−1)n Hn
− 3 + 2
n=1
n2 n=1
n2 n=1
n2
Z 1 3
ln (1 + x) ln(x)
= dx
0 x(1 + x)
Z 1 3 Z 1 3
ln (1 + x) ln(x) ln (1 + x) ln(x)
= dx − dx
0 x 1+x
|0 {z }
IBP
1
ln3 (1 + x) ln(x) 1
ln4 (1 + x)
Z Z
1
= dx + dx. (4.159)
0 x 4 0 x
Take the difference of the two relations in (4.158) and (4.159) then divide by 6,
∞ (2)
1 1 ln3 (1 − x) ln(1 + x)
Z 1 4
(−1)n Hn Hn
Z
X 1 ln (1 + x)
2
= dx + dx
n=1
n 6 0 x 24 0 x
1 1 ln3 (1 + x) ln(x)
Z
+ dx.
6 0 x
These three integrals are given in (3.119), (3.40), and (3.123).

P∞ (−1)n Hn
3
4.3.18 n=1 n2
Show that
∞
(−1)n Hn3

X 1 1 9 27
2
= −6 Li5 − 6 ln(2) Li 4 + ζ(5) + ζ(2)ζ(3)
n=1
n 2 2 4 16
21 2 1
− ln (2)ζ(3) + ln3 (2)ζ(2) − ln5 (2). (4.160)
8 5

∞
(−1)n Hn3 1 1 ln3 (1 − x) ln(1 + x) 1 1 ln4 (1 + x)
X Z Z
= dx − dx
n=1
n2 2 0 x 8 0 x
∞ (3)
1 1 ln3 (1 + x) ln(x) (−1)n Hn
Z X
− dx − 2 .
2 0 x n=1
n2
These terms are given in (3.119), (3.40), (3.123), and (4.156) respectively.
To see the two series in (4.152) and (4.160) computed differently, check [28, pp.
520–523].
4.4 Harmonic Series with Powers of 2 in the

Denominator
P∞ Hn
4.4.1 n=1 n2n
Show that
∞
X Hn 1
n
= ζ(2). (4.161)
n=1
n2 2
Solution (i) Set x = 1/2 in (2.7),

∞
X Hn 1 1
n
= Li2 + ln2 (2),
n=1
n2 2 2
and the value of Li2 ( 12 ) is given in (1.120).
Solution (ii) Set a = 0 in (3.91),

∞ Z 1
X Hn ln(1 + x) 1 1
n
= dx = − Li2 (−x) 0
= − Li2 (−1) = ζ(2).
n=1
n2 0 x 2
P∞ Hn
4.4.2 n=1 n2 2n
Show that
∞
X Hn 1
2 n
= ζ(3) − ln(2)ζ(2). (4.162)
n=1
n 2 2
Solution (i) Set x = 1/2 in (2.8),

∞
X Hn 1 1
2 n
= − ln(2) Li2 − ln3 (2) + ζ(3).
n=1
n 2 2 2
Plug in the value of Li2 ( 12 ) given in (1.120).
Solution (ii) By integration by parts,

Z 1
2 ln(2) 1
− xn−1 ln(x)dx = + 2 n.
0 n2n n 2
4.4. Harmonic Series with Powers of 2 in the Denominator 253
Multiply both sides by Hn then take the summation over n ≥ 1,

∞ ∞ ∞
Z 12 !
X Hn X Hn ln(x) X
n
ln(2) n
+ 2 2n
=− Hn x dx
n=1
n2 n=1
n 0 x n=1
Z 21
ln(x) ln(1 − x)
= dx
0 x(1 − x)
Z 1
ln(x) ln(1 − x)
= dx
0 x
Z 1 2
IBP 1 ln (x)
= dx.
2 0 1−x
Therefore,
∞ ∞
1 1 ln2 (x)
Z
X Hn X Hn
2 2n
= dx − ln(2) n
.
n=1
n 2 0 1 − x n=1
n2
A different approach may be found in [28, p. 500].
P∞ (2)
Hn
4.4.3 n=1 n2n
Show that
∞ (2)
X Hn 5
n
= ζ(3). (4.163)
n=1
n2 8
Solution (i) Put x = 1/2 in (2.82) then reorganize the terms,

∞ (2) ∞
X Hn 1 1 X Hn
n
= 3 Li 3 + ln(2) Li2 − 2 .
n=1
n2 2 2 n=1
n2 2n
Collect the values from (4.162), (1.120), and (1.132).
Solution (ii) Set a = 1 in (3.91),

∞ (2) Z 1
X Hn ln(1 − x) ln(1 + x)
n
= − dx.
n=1
n2 0 x

P∞ Hn2
4.4.4 n=1 n2n
Show that
∞
X Hn2 7
n
= ζ(3). (4.164)
n=1
n2 8

1 (2)
Hn2
Z
Hn
xn−1 ln2 (1 − x)dx = +
0 n n
by 2n then consider the summation over n ≥ 1,

∞ ∞ ∞ n
!
(2) Z 1 2
X Hn2 X Hn ln (1 − x) X x
n
+ n
= dx
n=1
n2 n=1
n2 0 x n=1
2
1 2 x Z 1 2
ln (1 − x)
Z
2 1−x=y ln (y)
= x dx = dy.
0 x 1− 2 0 1+y

∞ Z 1 2 ∞ (2)
X Hn2 ln (y) X Hn
= dy − .
n=1
n2n 0 1+y n=1
n2n
Substituting the results from (3.9) and (4.163) completes the solution.
An alternative solution is by setting x = 1/2 in the generating function in (2.14).
P∞ Hn
4.4.5 n=1 n3 2n
Show that
∞
X Hn 1 1 1 1 4
3 n
= Li4 + ζ(4) − ln(2)ζ(3) + ln (2). (4.165)
n=1
n 2 2 8 8 24

1
ln2 (2) 2 ln(2)
Z 2 2
xn−1 ln2 (x)dx = + 2 n + 3 n.
0 n2n n 2 n 2
Multiply through by Hn then take the summation over n ≥ 1,

∞ ∞ ∞
X Hn X Hn X Hn
ln2 (2) n
+ 2 ln(2) 2 2n
+ 2 3 2n
n=1
n2 n=1
n n=1
n
∞
Z 21 2 !
ln (x) X n
= Hn x dx
0 x n=1
Z 21 2
ln (x) ln(1 − x)
=− dx
0 x(1 − x)
Z 1/2 2 Z 12 2
ln (x) ln(1 − x) ln (x) ln(1 − x)
=− dx − dx
x 1−x
|0 {z } 0
IBP
1 1
ln3 (x) ln2 (x) ln(1 − x)
Z Z
1 1 2 2
= − ln4 (2) − dx − dx.
3 3 0 1−x 0 1−x
Reordering the terms,

∞ ∞ ∞
X Hn 1 4 1 2 X Hn X Hn
3 n
= − ln (2) − ln (2) n
− ln(2)
n=1
n 2 6 2 n=1
n2 n=1
n2 2n
Z 1 Z 1
1 2 ln3 (x) 1 2 ln2 (x) ln(1 − x)
− dx − dx.
6 0 1−x 2 0 1−x
Grouping the results from (4.161), (4.162), (3.34), and (3.142) finalizes the solution.
Check [28, pp. 500–501] for another solution.
P∞ (2)
Hn
4.4.6 n=1 n2 2n
Show that
∞ (2)
X Hn 1 1 1 1 1 4
2 n
= Li4 + ζ(4) + ln(2)ζ(3) − ln2 (2)ζ(2) + ln (2).
n=1
n 2 2 16 4 4 24
(4.166)
Solution let x = 1/2 in (2.83),

∞ (2) ∞
X Hn 1 1 2 1 X Hn
2 n
= 3 Li 4 − Li2 − 2 ,
n=1
n 2 2 2 2 n=1
n3 2 n
and the solution finishes on collecting the values from (4.165) and (1.120).
P∞ Hn2
4.4.7 n=1 n2 2n
Show that
∞
Hn2

X 1 37 7 1 1 4
2 2n
= − Li4 + ζ(4) − ln(2)ζ(3) + ln2 (2)ζ(2) − ln (2).
n=1
n 2 16 4 4 24
(4.167)

1 (2)
Hn2
Z
Hn
xn−1 ln2 (1 − x)dx = +
0 n n
n
∞ ∞ ∞
!
(2) Z 1 2
X Hn2 X Hn ln (1 − x) X (x/2)n
+ = dx
n=1
n2 2n n=1 n2 2n 0 x n=1
n
Z 1 2
ln (1 − x) ln(1 − x/2)
=− dx
0 x
Z 1 2
ln (y) ln 1+y

1−x=y 2
= − dy
0 1−y
{set a = 2 in (3.92)}
∞ (3)
X (−1)n Hn
= −2 .
n=1
n
Then, we have
∞ ∞ (2) ∞ (3)
X Hn2 X Hn X (−1)n Hn
= − − 2 .
n=1
n2 2n n=1
n2 2n n=1
n
These two sums are calculated in (4.166) and (4.142).
P∞ (3)
Hn
4.4.8 n=1 n2n
Show that
∞ (3)
X Hn 1 5 7 1 1 4
n
= Li4 − ζ(4) + ln(2)ζ(3) − ln2 (2)ζ(2) + ln (2).
n=1
n2 2 16 8 4 24
(4.168)
Solution (i) Take x = 1/2 in the generating function in (2.17),

∞ (3)
X Hn 1 1 1 2 1
= Li4 + ln(2) Li3 − Li2 ,
n=1
n2n 2 2 2 2
where the values of Li3 ( 21 ) and Li2 ( 12 ) are given in (1.120) and (1.132).
∞ (3)
1 1 ln2 (1 − x) ln(1 + x)
Z
X Hn
= dx,
n=1
n2n 2 0 x
and this integral is given in (3.118).
P∞ Hn
4.4.9 n=1 n4 2n
Show that
∞
X Hn 1 1 1 1
4 n
= 2 Li5 + ln(2) Li4 − ln3 (2)ζ(2) + ln2 (2)ζ(3)
n=1
n 2 2 2 6 2
1 1 1 1 5
− ln(2)ζ(4) − ζ(2)ζ(3) + ζ(5) + ln (2). (4.169)
8 2 32 40
Solution By integration by parts, we have

1
6 ln(2) 3 ln2 (2) ln3 (2)
Z 2 6
− xn−1 ln3 (x)dx = + + + .
0 n4 2 n n3 2n n2 2n n2n
Multiply both sides by Hn then consider the summation over n ≥ 1,

∞ ∞ ∞ ∞
X Hn X Hn 2
X Hn 3
X Hn
6 4 2n
+ 6 ln(2) 3 2n
+ 3 ln (2) 2 2n
+ ln (2) n
n=1
n n=1
n n=1
n n=1
n2
∞
Z 12 3 !
ln (x) X n
= −Hn x dx
0 x n=1
{substitute the generating function in (2.4)}
Z 21 3
ln (x) ln(1 − x)
= dx
0 x(1 − x)
Z 12 3 Z 12 3
ln (x) ln(1 − x) ln (x) ln(1 − x)
= dx + dx
x 1−x
|0 {z } |0 {z }
IBP 1−x→x
1
ln4 (x) 1
ln3 (1 − x) ln(x)
Z Z
1 1 2
= − ln5 (2) + dx + dx
4 4 0 1−x 1
2
x
{recall the relation involving of the latter integral from (3.146)}
Z 1
1 2 ln4 (x) 1 1 ln3 (x) ln(1 − x)
Z
93 1 5
= − ζ(5) − ln (2) − dx + dx
16 10 2 0 1−x 2 0 1−x
{plug in the results from (3.35) and (3.124)}

1 1 3 21 2
= 12 Li5 + 12 ln(2) Li4 + ζ(5) + ln (2)ζ(3) − 2 ln3 (2)ζ(2)
2 2 16 4
2
−3ζ(2)ζ(3) + ln5 (2).
5
Put together the results from (4.161), (4.162), and (4.165) to finish the solution.
For a different approach, see [28, pp. 501–502].
P∞ (4)
Hn
4.4.10 n=1 n2n
The following sum is proposed by Cornel Vălean (see[33]):

∞ (4)
X Hn 1 1 81 21
n
= 6 Li5 + 6 ln(2) Li4 − ζ(5) − ζ(2)ζ(3)
n=1
n2 2 2 16 8
21 2 1
+ ln (2)ζ(3) − ln3 (2)ζ(2) + ln5 (2). (4.170)
8 5
Solution Put a = 3 in (3.91), we obtain

∞ (4)
1 1 ln3 (1 − x) ln(1 + x)
Z
X Hn
= − dx.
n=1
n2n 6 0 x
This integral is calculated in (3.119) and the solution is complete.
P∞ (2)
Hn
4.4.11 n=1 n3 2n
The following sum is proposed by Cornel Vălean (see [34]):

∞ (2)
X Hn 1 1 23 1
3 n
= −2 Li5 − 3 ln(2) Li 4 + ζ(5) − ln(2)ζ(4)
n=1
n 2 2 2 64 16
23 23 2 7 3 13 5
+ ζ(2)ζ(3) − ln (2)ζ(3) + ln (2)ζ(2) − ln (2). (4.171)
16 16 12 120
Solution (i) Let x = 1/2 in (2.87),

∞ (2) ∞ (3) ∞ ∞ (4)
X Hn X Hn n 1 1 X Hn X Hn
+ x = 5 Li 5 +ln(2) Li 4 −2 − .
n=1
n3 2n n=1 n2 2n 2 2 n=1
n4 2n n=1 n2n
(4.172)
Now put x = 1/2 in (2.85),
∞ (2) ∞ (3) ∞
X Hn X Hn 1 1 1 X Hn
3 3 2n
+ 2 2n
= Li2 Li 3 + 10 Li5 − 6 4 2n
.
n=1
n n=1
n 2 2 2 n=1
n
(4.173)
Take the difference of (4.172) and (4.173),
∞ (2)
X Hn 5 1 1 1 1 1 1
3 n
= Li5 − ln(2) Li4 + Li2 Li3
n=1
n 2 2 2 2 2 2 2 2
∞ ∞ (4)
X Hn 1 X Hn
−2 + .
n=1
n4 2 n 2 n=1 n2n
Gather the values from (1.120), (1.132), (4.169), and (4.170) to finalize the solution.
Solution (ii) Divide both sides of (2.83):

∞ ∞ (2)
X Hn n X Hn n
Li22 (x) = 4 x + 2 x − 6 Li4 (x)
n=1
n3 n=1
n2
1
1
xn−1 dx =
R
by x then integrate from x = 0 to 1/2, using 2
0 n2n ,
∞ (2) Z 1 ∞
1 2 Li22 (x)

X Hn X Hn 1
3 2n
= dx − 2 4 2n
+ 3 Li5 .
n=1
n 2 0 x n=1
n 2
The first two terms are calculated in (3.154) and (4.169).
P∞ (3)
Hn
4.4.12 n=1 n2 2n
The following sum is also proposed by Cornel Vălean (see [34]):

∞ (3)
X Hn 1 1 81 5
2 n
= 4 Li 5 + 3 ln(2) Li 4 − ζ(5) + ln(2)ζ(4)
n=1
n 2 2 2 64 16
7 7 5 3 11 5
− ζ(2)ζ(3) + ln2 (2)ζ(3) − ln (2)ζ(2) + ln (2). (4.174)
8 8 12 120
Solution Combine (4.172) and (4.173),

∞ (3) ∞ (4)
X Hn 5 1 3 1 1 1 1 3 X Hn
= Li5 + ln(2) Li4 − Li2 Li3 − .
n=1
n2 2n 2 2 2 2 2 2 2 2 n=1 n2n
Put together the values from (1.120), (1.132), and (4.170) to finalize the solution.
P∞ Hn2
4.4.13 n=1 n3 2n
Show that
∞
Hn2

X 1 1 279 37
3 n
= −2 Li5 − ln(2) Li4 + ζ(5) − ln(2)ζ(4)
n=1
n 2 2 2 64 16
9 7 2 1 3 1 5
− ζ(2)ζ(3) + ln (2)ζ(3) + ln (2)ζ(2) − ln (2). (4.175)
16 16 12 40

∞
X ln2 (1 − x)
(Hn2 − Hn(2) )xn = .
n=1
1−x
2
ln (x)
by x then integrate from x = 0 to 1/2,
1 ∞ Z 12
ln2 (1 − x) ln2 (x)
Z 2 X
dx = Hn2 − Hn(2) xn−1 ln2 (x)dx
0 x(1 − x) n=1 0
∞ ln2 (2) 2 ln(2)
IBP
X 2
= Hn2 − Hn(2) + +
n=1
n2n n2 2n n3 2 n
∞ ∞ (2)
Hn2 − Hn

X
2 (2) ln(2) 2 X
= ln(2) (Hn − Hn ) + + 2
n=1
n2n n2 2 n n=1
n3 2n
( Z 12 )
ln(2) 2 n−1
write + 2 n =− x (ln(2) + 2 ln(x))dx
n2n n 2 0
∞ ∞
Z 21 !
(2)
X
(2) 2 n−1
X Hn2 − Hn
= ln(2) (Hn − Hn ) x (ln(2) + 2 ln(x))dx + 2
n=1 0 n=1
n3 2n
{reverse the order of integration and summation}
∞ ∞
1
!
(2)
Hn2 − Hn
Z 2 ln(2) + 2 ln(x) X (2) 2 n
X
= ln(2) (Hn − Hn )x dx + 2
0 x n=1 n=1
n3 2 n
{recall the generation function in (2.11)}
1 ∞ (2)
ln2 (1 − x) Hn2 − Hn
Z
ln(2) + 2 ln(x)
2 X
= ln(2) − dx + 2
0 x 1−x n=1
n3 2n
Z 12 2 Z 21
ln (1 − x) ln(x) ln2 (1 − x)
= − ln2 (2) dx − 2 ln(2) dx
0 x(1 − x) 0 x(1 − x)
∞ ∞ (2)
X Hn2 X Hn
+2 − 2
n=1
n3 2 n n=1
n3 2n
Reorganizing the terms, we have

∞ ∞ (2) Z 1
X Hn2 X Hn 1 2 ln2 (1 − x) ln2 (x)
= + dx
n=1
n3 2 n n=1
n3 2n 2 0 x(1 − x)
| {z }
I1
1 1
2
ln(x) ln (1 − x) ln2 (1 − x)
Z Z
2 1 2
+ ln(2) dx + ln2 (2) dx . (4.176)
0 x(1 − x) 2 0 x(1 − x)
| {z } | {z }
I2 I3
For I1 , set a = 2 in (3.86),

1
ln2 (1 − x) ln2 (x) 1
ln2 (1 − x) ln2 (x)
Z 2
Z
I1 = dx = dx
0 x(1 − x) 0 x
∞
Hn−1 1 n−1 2
X Z
=2 x ln (x)dx
n=1
n 0
∞ ∞
Hn − n1 2!

X X Hn
=2 3
= 4 − 4ζ(5)
n=1
n n n=1
n4
= 8ζ(5) − 4ζ(2)ζ(3). (4.177)
For I2 ,
1
ln(x) ln2 (1 − x) 1
ln(1 − y) ln2 (y)
Z 2
Z
1−x=y
I2 = dx = dy
0 x(1 − x) 1 y(1 − y)
|2 {z }
R1 R 1/2
0
− 0
Z 21
ln(1 − y) ln2 (y)
1
ln(1 − y) ln2 (y)
Z
= dy − dy
0 y(1 − y) 0 y(1 − y)
{set a = 2 in (3.88) to get the first integral}

1 1 1
and write = + in the second one
y(1 − y) y 1−y
∞ Z 12 Z 12
X Hn ln(1 − y) ln2 (y) ln(1 − y) ln2 (y)
= −2 − dy − dy
n3 y 1−y
n=1 |0 {z } 0
IBP
∞ Z 1 Z 21
X Hn 1 4 1 2 ln3 (y) ln(1 − y) ln2 (y)
= −2 − ln (2) − dy − dy
n=1
n3 3 3 0 1−y 0 1−y

1 9 7 1 1 4
= 2 Li4 − ζ(4) + ln2 (2)ζ(3) + − ln2 (2)ζ(2) + ln (2).
2 4 4 2 12
For I3 ,
1
ln2 (1 − x) ln2 (y)
Z 2
Z 1
1−x=y
I3 = dx = dy
0 x(1 − x) 1 (1 − y)y
2
Z 1 2 Z 1 2
ln (y) ln (y)
= dy + dy
1
2
y 1
2
1−y
| {z }
R1 R 1/2
0
− 0
1
Z 1 2 Z 2
1 ln (y) 2 ln (y)
= ln3 (2) + dy − dy
3 0 1−y 0 1−y
1
= ζ(3).
4
Substitute the results of I1 , I2 , and I3 along with the result from (4.171) in (4.176)
to complete the solution. For a different method, check [21].
P∞ (2)
Hn Hn
4.4.14 n=1 n2 2n
The following sum is proposed by Cornel Vălean (see [?]):

∞ (2)
X Hn Hn 1 1 31 1
2 n
= 2 Li 5 + ln(2) Li4 − ζ(5) + ln(2)ζ(4)
n=1
n 2 2 2 32 8
1 1 3 1 5
+ ζ(2)ζ(3) − ln (2)ζ(2) + ln (2). (4.178)
8 12 40

1 (2) (3)
Hn3 + 3Hn Hn + 2Hn
Z
− xn−1 ln3 (1 − x)dx =
0 n
by n2n then consider the summation over n ≥ 1,

∞ ∞ ∞ ∞
!
(2) (3) Z 1 3
X Hn3 X Hn Hn X Hn ln (1 − x) X (x/2)n
2 2n
+3 2 2n
+2 2 2n
=− dx
n=1
n n=1
n n=1
n 0 x n=1
n
Z 1 3
1 3
ln (y) ln 1+y

ln (1 − x) ln(1 − x/2)
Z
1−x=y 2
= dx = dy
0 x 0 1−y
{set a = 3 in (3.92)}
∞ (4)
X (−1)n Hn
= −6 .
n=1
n
Substituting the result of the latter sum from (4.154),

∞ ∞ (2) ∞ (3)
X Hn3 X Hn Hn X Hn 21 9
2 2n
+ 3 2 2n
+ 2 2 2n
= 12ζ(5) − ln(2)ζ(4) − ζ(2)ζ(3).
n=1
n n=1
n n=1
n 4 4
(4.179)
Next, divide both sides of (2.22):

∞
ln3 (1 − x) X n 3
− = x Hn − 3Hn Hn(2) + 2Hn(3)
1−x n=1
1
1
xn−1 dx =
R
by x, then integrate using 0
2
n2n ,
∞ (2) (3) Z 12 3
X Hn3 − 3Hn Hn + 2Hn ln (1 − x)
n
= − dx. (4.180)
n=1
n2 0 x(1 − x)
On the other hand, multiply both sides of (2.22) by − ln(x)

x then integrate,
1
ln3 (1 − x) ln(x)
Z 2
dx
0 x(1 − x)
∞ Z 12 !
X
3 (2) (3) n−1
= Hn − 3Hn Hn + 2Hn − x ln(x)dx
n=1 0
∞
IBP
X ln(2) 1
= Hn3 − 3Hn Hn(2) + 2Hn(3) + 2 n .
n=1
n2n n 2
Distribute then rearrange the terms,

∞ ∞ (2) ∞ (3)
X Hn3 X Hn Hn X Hn
− 3 + 2
n=1
n2 2 n n=1
n2 2n n=1
n2 2n
1 ∞ (2) (3)
ln3 (1 − x) ln(x) Hn3 − 3Hn Hn + 2Hn
Z 2 X
= dx − ln(2)
0 x(1 − x) n=1
n2n
( Z 12 3 )
ln (1 − x)
the latter sum is equal to − dx given in (4.180)
0 x(1 − x)
Z 21 3 Z 21 3
ln (1 − x) ln(x) ln (1 − x)
= dx + ln(2) dx .
x(1 − x) x(1 − x)
|0 {z } |0 {z }
I1 I2
For I1 ,
1 1
ln3 (1 − x) ln(x) ln3 (1 − x) ln(x)
Z 2
Z 2
dx + dx
0 x 1−x
|0 {z }
IBP
Z 1! 3 1
1
ln (1 − x) ln(x) ln4 (1 − x)
Z Z
1 1 2
= − dx + ln5 (2) + dx
0 1 x 4 4 x
2 |0 {z }
1−x→x
1 3 1 3
ln (1 − x) ln(x) ln (1 − x) ln(x)
Z Z
1
= dx − dx + ln5 (2)
x 1 x 4
|0 {z } 2
1−x→x
1
ln4 (x)
Z
1
+ dx
4 1 1−x
|2 {z }
R1 R 1/2
0
− 0
3
1
ln (x) ln(1 − x) ln3 (1 − x) ln(x) 1
Z Z
1
= dx − dx + ln5 (2)
0 1 − x 1
2
x 4
Z 1 4 Z 1
1 ln (x) 1 2 ln4 (x)
+ dx − dx
4 0 1−x 4 0 1−x
{recall the relation involving the second integral from (3.146)}
Z 1
1 1 ln3 (x) ln(1 − x) 1 2 ln4 (x) 1 1 ln4 (x)
Z Z
= dx − dx + dx
2 0 1−x 2 0 1−x 4 0 1−x
3 1 5
− ζ(5) + ln (2)
8 10
{put together the results from (3.124), (3.35), and (3.5)}

1 1 285 21 2
= −12 Li5 − 12 ln(2) Li4 + ζ(5) − 3ζ(2)ζ(3) − ln (2)ζ(3)
2 2 16 4
2
+2 ln3 (2)ζ(2) − ln5 (2).
5
For I2 ,
1
ln3 (1 − x) 1
ln3 (y) 1
ln3 (y) 1
ln3 (y)
Z 2
Z Z Z
1−x=y
dx = dy = dy + dy
0 x(1 − x) 1
2
y(1 − y) 1
2
y 1 1−y
|2 {z }
R1 R 1/2
0
− 0
1
Z 1 3 Z 3
1 ln (y) ln (y)2
= − ln4 (2) + dy − dy
4 0 1 − y 0 1−y

1 21 3 1
= 6 Li4 − 6ζ(4) + ln(2)ζ(3) − ln2 (2)ζ(2) + ln4 (2).
2 4 2 4
Combine the results of I1 and I2 ,

∞ ∞ (2) ∞ (3)
Hn3

X X Hn Hn X Hn 1 1
2 2n
− 3 2 2n
+ 2 2 2n
= −12 Li5 − 6 ln(2) Li 4
n=1
n n=1
n n=1
n 2 2
285 1 3 5
+ ζ(5) − 6 ln(2)ζ(4) − 3ζ(2)ζ(3) + ln3 (2)ζ(2) − ln (2). (4.181)
16 2 20
Taking the difference of (4.179) and (4.181) then dividing by 6 finishes the solution.
P∞ Hn3
4.4.15 n=1 n2 2n
The following sum is proposed by Cornel Vălean (see [?]):

∞
Hn3

X 1 1 279 25
2 2n
= −14 Li5 − 9 ln(2) Li4 + ζ(5) − ln(2)ζ(4)
n=1
n 2 2 16 4
7 7 13 3 31 5
− ζ(2)ζ(3) − ln2 (2)ζ(3) + ln (2)ζ(2) − ln (2). (4.182)
8 4 12 120
Solution Combining (4.179) and (4.181) yields

∞ ∞ (3)
Hn3

X X Hn 1 1 477
2 2 2n
+4 2 2n
= −12 Li5 − 6 ln(2) Li4 + ζ(5)
n=1
n n=1
n 2 2 16
45 21 1 3 5
− ln(2)ζ(4) − ζ(2)ζ(3) + ln3 (2)ζ(2) − ln (2).
4 4 2 20
Plug in the result from (4.174) then divide by 2 to finish the solution.
4.5 Harmonic Series with Powers of 2n + 1 in the

Denominator
P∞ (−1)n H2n+1
4.5.1 n=0 2n+1
Show that
∞
X (−1)n H2n+1 π
= G − ln(2). (4.183)
n=0
2n + 1 8
Hn
Solution Set an = n in (1.11):
∞
X ∞
X
n
(−1) a2n+1 = J in an ,
n=0 n=1
we have
∞ ∞ n
X (−1)n H2n+1 X i Hn
=J
n=0
2n + 1 n=1
n
{let x = i in the generating function in (2.7)}

1 2
= J Li2 (i) + ln (1 − i) .
2
Gather the values from (1.108) and (1.25) to finish the solution.
P∞ (−1)n H2n+1
4.5.2 n=0 (2n+1)2
Show that
∞
X (−1)n H2n+1 π 2 1
2
= −J Li3 (1 − i) − ln (2) − ln(2)G. (4.184)
n=0
(2n + 1) 16 2
Hn
Solution Let an = n2 in (1.11),
∞ ∞ n
2
= J
n=0
(2n + 1) n=1
n2

1 2
= J Li3 (i) − Li3 (1 − i) + ln(1 − i) Li2 (1 − i) + ln(i) ln (1 − i) + ζ(3) .
2
Gather the values from (1.109), (4.140), (1.26), and (1.25) to finish the solution.
4.5. Harmonic Series with Powers of 2n + 1 in the denominator 267
(2)
P∞ (−1)n H2n+1
4.5.3 n=0 2n+1
Show that
∞ (2)
X (−1)n H2n+1 17π 3 π 2 1
= 2 J Li3 (1 − i) + + ln (2) + ln(2)G. (4.185)
n=0
2n + 1 192 8 2
(2)
Hn
Solution Set an = n in (1.11),
∞ (2) ∞ n (2)
=J
n=0
2n + 1 n=1
n
{employ the generating function in (2.10)}
= J {Li3 (i) + 2 Li3 (1 − i) − ln(1 − i) Li2 (1 − i) − ζ(2) ln(1 − i) − 2ζ(3)} .
These terms are given in (1.109), (4.140), and (1.25).

P∞ Hn
4.5.4 n=1 (2n+1)2
Show that
∞
X Hn 7 3
2
= ζ(3) − ln(2)ζ(2). (4.186)
n=1
(2n + 1) 4 2
Solution Replace x by x2 in (2.4),

∞
X ln(1 − x2 )
Hn x2n = − .
n=1
1 − x2
Multiply both sides by − ln(x) then integrate using:

Z 1
1
− x2n ln(x)dx = ,
0 (2n + 1)2
we obtain
∞ Z 1
X Hn ln(x) ln(1 − x2 )
= dx
n=1
(2n + 1)2 0 1 − x2
√ Z 1
x= y 1 ln(y) ln(1 − y)
= √ dy.
4 0 y 1−y
This integral is calculated in (3.110).

A different way to calculate this sum may be found in (4.74).
P∞ (−1)n Hn
4.5.5 n=0 (2n+1)2
Show that
∞
X (−1)n Hn 3π 3 π
2
= 2 J Li3 (1 − i) + + ln2 (2) − ln(2)G. (4.187)
n=0
(2n + 1) 32 8
1
Solution Substitute H2n = H2n+1 − 2n+1 in (3.96),
1
x2n
Z
1
ln(2) + Hn − H2n+1 + = dx.
2n + 1 0 1+x
(−1)n
Multiply both sides by (2n+1)2 then take the summation over n ≥ 0,
∞ ∞ ∞ ∞
X (−1)n X (−1)n Hn X (−1)n H2n+1 X (−1)n
ln(2) + − +
n=0
(2n + 1)2 n=0 (2n + 1)2 n=0 (2n + 1)2 n=0
(2n + 1)3
∞
Z 1 !
1 X (−1)n x2n
= dx
0 1+x n=0
(2n + 1)2
{multiply the sum by x/x}
∞
!
1
1 X (−1)n x2n+1
Z
1
= dx
0 1+x x n=0 (2n + 1)2
∞ ∞
( )
X X
n n
use (−1) a2n+1 = J i an given in (1.11)
n=0 n=1
∞ n n
!
Z 1
1 1 X i x
= J dx
0 1+x x n=1 n2
Z 1
1 Li2 (ix)
= J dx
0 1+x x
{make use of (1.111) for Li2 (ix)}
Z 1 Z 1
1 i ln(y)
= J − dy dx
0 1+x 0 1 − ixy

i 1
use the fact that J =
1 − ixy 1 + x2 y 2
Z 1 Z 1
1 ln(y)
= − dy dx
0 1+x 0 1 + x2 y 2
Z 1Z x
xy=t ln(x/t)
= dt dx
0 0 x(1 + x)(1 + t2 )

Z 1 Z 1
1 ln(x/t)
= 2
dx dt
0 1+t t x(1 + x)
{evaluate the inner integral by partial fraction decoposition}
Z 1
π2

1 1 2
= 2
Li2 (−t) + ln (t) + ln(2) ln(t) + dt
0 1+t 2 12
1 1 ln2 (t)
Z 1 Z 1
π 2 1 dt
Z Z
Li2 (−t) ln(t)
= dt + dt + ln(2) dt +
0 1 + t2 2 0 1 + t2 0 1+t
2 12 0 1 + t2

1
expand in series in the second and third integrals
1 + t2
Z 1 ∞ Z 1
Li2 (−t) 1X
= 2
dt + (−1) n
x2n ln2 (t)dt
0 1 + t 2 n=0 0
∞ Z 1
X π2 π
+ ln(2) (−1)n x2n ln(t)dt +
n=0 0 12 4
Z 1 ∞ n ∞
X (−1)n
Li2 (−t) X (−1) π3
= dt + − ln(2) + .
0 1 + t2 n=0
(2n + 1)3 n=0
(2n + 1)2 48
Reorganize the terms,

∞ ∞ ∞
X (−1)n Hn X (−1)n H2n+1 X (−1)n
= − 2 ln(2)
n=0
(2n + 1)2 n=0
(2n + 1)2 n=0
(2n + 1)2
Z 1
Li2 (−t) π3
+ 2
dt + . (4.188)
0 1+t 48
Put together the results from (4.184), (1.205), and (3.147) to end the solution.
(2)
P∞ (−1)n Hn
4.5.6 n=0 2n+1
∞ (2)
X (−1)n Hn 5π 3 π
= 4 J Li3 (1 − i) + + ln2 (2) + 2 ln(2)G. (4.189)
n=0
2n + 1 48 4
Solution We begin with expanding arctan x in series in the following integral:

1 ∞
(−1)n 1 x2n ln(x)
Z Z
ln(x) arctan x X
dx = dx
0 x(1 + x) n=0
2n + 1 0 1+x
∞
(−1)n 1 ∂ 1 x2n
X Z
= dx
n=0
2n + 1 0 ∂n 2 1 + x
∞ Z 1 2n
1 X (−1)n d x
= dx
2 n=0 2n + 1 dn 0 1 + x
∞
1 X (−1)n d
= (Hn − H2n + ln 2)
2 n=0 2n + 1 dn
{use the derivative of the harmonic number given in (1.157)}
∞
1 X (−1)n (2)
= 2H2n − Hn(2) − ζ(2)
2 n=0 2n + 1
∞ (2) ∞ (2) ∞
X (−1)n H2n 1 X (−1)n Hn 1 X (−1)n
= − − ζ(2)
n=0
2n + 1 2 n=0 2n + 1 2 n=0
2n + 1

(2) (2) 1
write H2n = H2n+1 − in the first sum
(2n + 1)2
∞ (2) ∞ (2) ∞ ∞
X (−1)n H2n+1 1 X (−1)n Hn X (−1)n 1 X (−1)n
= − − 3
− ζ(2) .
n=0
2n + 1 2 n=0 2n + 1 n=0
(2n + 1) 2 n=0
2n + 1
(4.190)
On the other hand, we have

Z 1 Z 1 Z 1
ln(x) arctan x ln(x) arctan x ln(x) arctan x
dx = dx − dx
0 x(1 + x) 0 x 0 1+x
{expand arctan x in series in the first integral}
∞
X (−1)n Z 1 Z 1
ln(x) arctan x
= x2n ln(x)dx − dx
n=0
2n + 1 0 0 1+x
∞ Z 1
X (−1)n ln(x) arctan x
=− 3
− dx. (4.191)
n=0
(2n + 1) 0 1+x
Combine (4.190) and (4.191),

∞ (2) ∞ (2) ∞ Z 1
X (−1)n Hn X (−1)n H2n+1 X (−1)n ln(x) arctan x
=2 −ζ(2) +2 dx.
n=0
2n + 1 n=0
2n + 1 n=0
2n + 1 0 1+x
Gather the results from (4.185), (1.93), and (3.149) to finish the solution.
P∞ (−1)n H2n+1
4.5.7 n=0 (2n+1)3
Show that
∞
X (−1)n H2n+1 35π
3
= 2β(4) − ζ(3). (4.192)
n=0
(2n + 1) 128

∞
arctan x 1X
= (−1)n (Hn − 2H2n )x2n−1
1 + x2 2 n=0
by x ln2 (x) then integrate from x = 0 to 1,

∞
1
x ln2 (x) arctan x
Z Z 1
1X
dx = (−1) n
(H n − 2H 2n ) x2n ln2 (x)dx
0 1 + x2 2 n=0 0
∞
1X 2
= (−1)n (Hn − 2H2n )
2 n=0 (2n + 1)3
∞ ∞
X (−1)n Hn X (−1)n H2n
= 3
− 2
n=0
(2n + 1) n=0
(2n + 1)3

1
use H2n = H2n+1 − in the last series
2n + 1
∞ n ∞ ∞
X (−1) Hn X (−1)n H2n+1 X (−1)n
= − 2 + 2 .
n=0
(2n + 1)3 n=0
(2n + 1)3 n=0
(2n + 1)4

∞ ∞ ∞
X (−1)n H2n+1 X (−1)n Hn X (−1)n
2 = + 2
n=0
(2n + 1)3 n=0
(2n + 1)3 n=0
(2n + 1)4
x ln2 (x) arctan x
Z 1
− dx
0 1 + x2
∞
( Z 1 2 )
X (−1)n ln (x) arctan x
write 2 = dx,
n=0
(2n + 1)4 0 x
{which follows from expanding arctan x in series then integrating}
∞ Z 1 2
x ln2 (x) arctan x
Z 1
X (−1)n Hn ln (x) arctan x
= + dx − dx
n=0
(2n + 1)3 0 x 0 1 + x2
∞ Z 1
(−1)n Hn

X 1 x
= 3
+ − 2
ln2 (x) arctan x dx
n=0
(2n + 1) 0 x 1 + x
∞ n Z 1 2
X (−1) Hn ln (x) arctan x
= 3
+ dx.
n=0
(2n + 1) 0 x(1 + x2 )
Collect the results from (4.93) and (3.150) then divide by 2 to finish the solution.
(2)
P∞ (−1)n H2n+1
4.5.8 n=0 (2n+1)2
Show that
∞ (2)
X (−1)n H2n+1 35π π2
2
= −β(4) + ζ(3) − G. (4.193)
n=0
(2n + 1) 64 48
Solution Set x = i in (2.83) then take the imaginary parts of both sides,
∞ ∞ (2)
X Hn i X Hn i
J Li22 (i) = 4 J x + 2 J x − 6 J Li4 (i).
n=1
n3 n=1
n2
∞
X ∞
X
Use J in an = (−1)n a2n+1 given in (1.11),
n=1 n=0
∞ (2) ∞
X (−1)n H2n+1 1 2
X (−1)n H2n+1
2
= 3 J Li4 (i) + J Li2 (i) − 2 .
n=0
(2n + 1) 2 n=0
(2n + 1)3
Using the value from (1.108), we have
π2

5
Li22 (i) = ζ(4) − G2 − G i.
128 24
Collect this result along with these from (1.110) and (4.192), the solution is finished.
P∞ (2)
Hn
4.5.9 n=1 (2n+1)2
Show that
∞ (2)
X Hn 1 121 1
= 8 Li4 − ζ(4)+7 ln(2)ζ(3)−2 ln2 (2)ζ(2)+ ln4 (2).
n=1
(2n + 1)2 2 16 3
(4.194)

Z 1
1
=− x2n ln(x)dx.
(2n + 1)2 0
(2)
Multiply both sides by Hn then take the summation over n ≥ 1,
∞ (2) ∞ Z 1
X Hn X
(2)
2
= − Hn x2n ln(x)dx
n=1
(2n + 1) n=1 0
∞
Z 1 !
X
=− ln(x) (x2 )n Hn(2) dx
0 n=1
{replace x by x2 in the generating function in (2.2)}
Z 1
ln(x) Li2 (x2 )
=− dx
0 1 − x2
write Li2 (x2 ) = 2 Li2 (x) + 2 Li2 (−x) given in (1.115)

Z 1 Z 1
ln(x) Li2 (x) ln(x) Li2 (−x)
= −2 2
dx − 2 dx
0 1−x 0 1 − x2
Z 1 Z 1 Z 1
ln(x) Li2 (x) ln(x) Li2 (x) ln(x) Li2 (−x)
=− dx − dx − dx
1−x 1+x 1−x
|0 {z } |0 {z } |0 {z }
I1 I2 I3
Z 1
ln(x) Li2 (−x)
− dx .
0 1+x
| {z }
I4
For I1 ,
Z 1
ln(x) Li2 (x)
I1 = dx
0 1−x
∞
( )
Li2 (x) X (2) n−1
write = Hn−1 x given in (2.3)
1−x n=1
∞ Z 1 ∞ (2)
X (2)
X Hn−1
= Hn−1 xn−1 ln(x)dx = −
n=1 0 n=1
n2
∞ (2) 1 ∞ (2)
X Hn − n2
X Hn
=− =− + ζ(4).
n=1
n2 n=1
n2
For I2 ,
Z 1
ln(x) Li2 (x)
I2 = dx
0 1+x
{expand 1/(1 + x) in series}

∞
X Z 1
n
=− (−1) xn−1 ln(x) Li2 (x)dx
n=1 0
∞
!
(2)
X
n Hn 2Hn 2ζ(2)
=− (−1) + 3 − .
n=1
n2 n n2
For I3 ,
Z 1
ln(x) Li2 (−x)
I3 = dx
0 1−x
{expand Li2 (−x) in series }
∞
(−1)n 1 xn ln(x)
X Z
= dx
n=1
n2 0 1−x
∞
X (−1)n (2)
= H n − ζ(2) .
n=1
n2
For I4 ,
Z 1
ln(x) Li2 (−x)
I4 = dx
0 1+x
∞
( )
Li2 (x) X (2) n−1
replace x by −x in = Hn−1 x given in (2.3)
1−x n=1
∞ Z 1
(2)
X
=− (−1)n Hn−1 xn−1 ln(x)dx
k=1 0
∞ (2) ∞ (2) 1
Hn−1 n Hn −
X X
n2
= (−1)n = (−1)
n=1
n2 n=1
n2
∞ (2)
X (−1)n Hn 7
= 2
+ ζ(4).
n=1
n 8
Gathering the four integrals,

∞ (2) ∞ ∞ (2) ∞ (2)
X Hn X (−1)n Hn X (−1)n Hn X Hn 5
2
= 2 3
− 2
+ 2
− ζ(4).
n=1
(2n + 1) n=1
n n n=1
n 8
k=1
Combining the results from (4.143), (4.144), and (2.98) completes the solution.
Solution (ii) Set p = q = 2 in the third application of Abel’s summation in (2.105),

∞ (2) ∞ (2) ∞ (2)
X Hn 5 1 X Hn 1 X (−1)n Hn
= ζ(4) − − ,
n=1
(2n + 1)2 8 4 n=1 n2 2 n=1 n2
and the solution finalizes on substituting the results from (4.144) and (2.98).
P∞ Hn2
4.5.10 n=1 (2n+1)2
Show that
∞
Hn2

X 1 61 1
= 8 Li4 − ζ(4) + ln2 (2)ζ(2) + ln4 (2). (4.195)
n=1
(2n + 1)2 2 16 3
Solution Replace x by x2 in (2.11),

∞
ln2 (1 − x2 ) X 2
= (Hn − Hn(2) )x2n .
1 − x2 n=1
Multiply both sides by − ln(x) then integrate from x = 0 to 1,

∞ (2)
ln(x) ln2 (1 − x2 )
Z 1
X Hn2 − Hn
= − dx
n=1
(2n + 1)2 0 1 − x2
√
1 1 ln(y) ln2 (1 − y)
Z
x=y
= − √ dy.
4 0 y(1 − y)
Reorder the terms,

∞ ∞ (2)
Hn2 1 1 ln(y) ln2 (1 − y)
Z
X X Hn
= − √ dy.
n=1
(2n + 1)2 n=1
(2n + 1)2 4 0 y(1 − y)
Substitute the results from (4.194) and (3.141) to complete the solution.
P∞ (2)
Hn
4.5.11 n=1 (2n+1)3
Show that
∞ (2)
X Hn 49 93
3
= ζ(2)ζ(3) − ζ(5). (4.196)
n=1
(2n + 1) 8 8
Solution Set q = 2 and p = 3 in (2.105),

∞ (2) ∞ (3) ∞ (3)
X Hn 7 15 X Hn X (−1)n Hn
= ζ(2)ζ(3) − − 2 .
n=1
(2n + 1)3 8 8 n=1 n2 n=1
n2
The last two sums are given in (4.105) and (4.156).
P∞ (3)
Hn
4.5.12 n=1 (2n+1)2
Show that
∞ (3)
X Hn 31
2
= ζ(5) − 8ζ(2)ζ(3). (4.197)
n=1
(2n + 1) 2
Solution Set q = 3 and p = 2 in (2.105),

∞ (3) ∞ (2) ∞ (2)
X Hn 3 15 X Hn X (−1)n Hn
= ζ(2)ζ(3) − − 4 .
n=1
(2n + 1)2 4 4 n=1 n3 n=1
n3
The last two sums are given in (4.102) and (4.152).
(3) (3)
P∞ Hn P∞ (−1)n Hn
4.5.13 n=1 (2n+1)3 +4 n=1 n3
Show that
∞ (3) ∞ (3)
X Hn X (−1)n Hn 17 31
3
+ 4 3
= − ζ 2 (3) − ζ(6). (4.198)
n=1
(2n + 1) n=1
n 16 16
Solution Set q = p = 3 in (2.105),

∞ (3) ∞ (3) ∞ (3)
X Hn 7 2 31 X Hn X (−1)n Hn
= ζ (3) − − 4 .
n=1
(2n + 1)3 8 8 n=1 n3 n=1
n3
The first sum is given in (2.99).

4.6. Skew Harmonic Series 277
4.6 Skew Harmonic Series

P∞ (−1)n H n
4.6.1 n=1 n
Show that
∞
X (−1)n H n 1 1
= − ζ(2) − ln2 (2). (4.199)
n=1
n 2 2

∞
X ln(1 − x)
(−1)n H n xn = . (4.200)
n=1
1+x
R1 1
xn−1 dx = n,
∞ Z 1 Z 1 Z 1
X (−1)n H n ln(1 − x) ln(1 − x) ln(1 − x)
= dx = dx − dx
n 0 x(1 + x) x 1+x
n=1 |0 {z } 0
1−x=y
1 1
ln(1 − x)
Z Z
ln(y)
= dy − dx.
0 1−y 0 1+x
Put together the results from (3.2) and (3.27) to complete the solution.
For an alternative solution, set x = −1 in the generating function in (2.29).
P∞ Hn
4.6.2 n=1 n3
Show that
∞
X Hn 1 3 1 1 4
3
= 2 Li4 − ζ(4) − ln2 (2)ζ(2) + ln (2). (4.201)
n=1
n 2 2 2 12
ln2 (x)
Solution Multiply both sides of (2.28) by 2x then integrate using
Z 1
1 1
xn−1 ln2 (x)dx = ,
2 0 n3
∞
1 1 ln(1 + x) ln2 (x)
Z
X Hn
= dx
n=1
n3 2 0 x(1 − x)
1
ln(1 + x) ln2 (x) 1 1 ln(1 + x) ln2 (x)
Z Z
1
= dx + dx
2 x 2 0 1−x
|0 {z } | {z }
IBP IBP
1
ln3 (x)
Z 1
ln(x) ln(1 + x) ln(1 − x)
Z
1
=− dx + dx
6 0 1+x 0 x
1 1 ln(1 − x) ln2 (x)
Z
+ dx.
2 0 1+x
Put together the results from (3.10), (3.116), and (3.130) to finish the solution.
For a different method, see (4.13).
P∞ (−1)n H n
4.6.3 n=1 n3
Show that
∞
(−1)n H n

X 1 3 1 1 4
3
= 2 Li4 − ζ(4) − ln2 (2)ζ(2) + ln (2). (4.202)
n=1
n 2 2 2 12
ln2 (x)
Solution (i) Multiply both sides of (4.200) by 2x then integrate using
Z 1
1 1
xn−1 ln2 (x)dx = ,
2 0 n3
∞
(−1)n H n 1 1 ln(1 − x) ln2 (x)
X Z
= dx
n=1
n3 2 0 x(1 + x)
1 1 ln(1 − x) ln2 (x) 1 1 ln(1 − x) ln2 (x)
Z Z
= dx − dx
2 0 x 2 0 1+x
| {z }
IBP
1
ln3 (x) 1
ln(1 − x) ln2 (x)
Z Z
1 1
= dx − dx,
6 0 1−x 2 0 1+x
and the solution completes on putting together the results from (3.4) and (3.131).
∞ ∞ ∞ ∞
X (−1)n H n X (−1)n Hn X H n 3 X Hn
= − + .
n=1
n3 n=1
n3 n=1
n3 4 n=1 n3
These sums are given in (4.143), (4.201) and (4.5).

P∞ (−1)n H n Hn
4.6.4 n=1 n
Show that
∞
X (−1)n H n Hn 1 1
= ln3 (2) − ln(2)ζ(2) − ζ(3). (4.203)
n=1
n 3 4

1
(−x)n
Z
H n = ln(2) − dx
0 1+x
n
(−1) Hn
by n then take the summation,
∞ ∞ ∞
Z 1 !
X (−1)n H n Hn X (−1)n Hn 1 X Hn n
= ln(2) − x dx
n=1
n n=1
n 0 1+x n=1
n
∞ Z 1
(−1)n Hn

X 1 1 2
= ln(2) − ln (1 − x) + Li2 (x) dx
n=1
n 0 1+x 2
∞
1 1 ln2 (1 − x)
X (−1)n Hn Z Z 1
Li2 (x)
= ln(2) − dx − dx
n=1
n 2 0 1 + x 0 1+x
| {z }
IBP
∞
(−1)n Hn 1 1 ln2 (1 − x)
X Z
= ln(2) − dx
n=1
n 2 0 1+x
Z 1
ln(1 + x) ln(1 − x)
− ln(2)ζ(2) − dx.
0 x
The solution finalizes on combining the results from (4.135), (3.28), and (3.115).
Solution (ii) Multiply both sides of (4.200) by − ln(1−x)

x then integrate using
R1
− 0 xn−1 ln(1 − x)dx = Hnn ,
∞ Z 1 2
X (−1)n H n Hn ln (1 − x)
=− dx
n=1
n 0 x(1 + x)
ln2 (1 − x)
1 Z 1 2 Z 1 2 Z 1 2
ln (1 − x) ln (1 − x)
Z
ln (y)
= dx − dx = dx − dy.
0 1+x x 1+x 0 1−y
|0 {z } 0
1−x=y
These two integrals are calculated in (3.28) and (3.3).

P∞ H n Hn
4.6.5 n=1 n2
Show that
∞
X H n Hn 1 43 3 1
2
= −3 Li4 + ζ(4) + ln2 (2)ζ(2) − ln4 (2). (4.204)
n=1
n 2 16 4 8
The following solution can be found in [18]:

1
(−x)n
Z
H n = ln(2) − dx
0 1+x
Hn
by n2 then take the summation,
∞ ∞ ∞
Z 1 !
X H n Hn X Hn 1 X Hn
2
= ln(2) 2
− 2
(−x)n dx.
n=1
n n=1
n 0 1 + x n=1
n
For the integral, utilize the generating function in (2.8) for the sum,
∞
Z 1 ! Z 1 Z 1
1 X Hn (−x)n Li3 (−x) Li3 (1 + x)
2
dx = dx − dx
0 1+x n 1+x 1+x
n=1 |0 {z } |0 {z }
I1 I2
1
1 1 ln(−x) ln2 (1 + x)
Z Z Z 1
ln(1 + x) Li2 (1 + x) ζ(3)
+ dx + dx + dx .
1+x 2 0 1+x 1+x
|0 {z } | {z } | 0 {z }
I3 I4 I5
For I1 ,
Z 1
IBP 1 ln(1 + x) Li2 (−x)
I1 = ln(1 + x) Li3 (−x) 0
− dx
0 x
3 1 1 3 5
= − ln(2)ζ(3) + Li22 (−x) 0
= − ln(2)ζ(3) + ζ(4).
4 2 4 16
For I2 ,
1
I2 = Li4 (1 + x) 0
= Li4 (2) − ζ(4).
For I3 ,
Z
IBP 1 Li3 (1 + x)
I3 = Li3 (1 + x) ln(1 + x) 0
− dx
1+x
1
= ln(2) Li3 (2) − Li4 (1 + x) 0
= ln(2) Li3 (2) − Li4 (2) + ζ(4).
For I4 ,
Z 1
IBP 2 1 Li2 (1 + x) ln(1 + x)
I4 = − Li2 (1 + x) ln (1 + x) 0
+2 dx
0 1+x
IBP 2
= − ln (2) Li2 (2) + 2 ln(2) Li3 (2) − 2 Li4 (2) + 2ζ(4).
For I5 ,
1
I5 = ζ(3) ln(1 + x) 0
= ln(2)ζ(3).
Group all integrals (I1 to I5 ) along with the result from (4.4),
∞
X H n Hn 7 53 1
2
= ln(2)ζ(3) − ζ(4) + 3 Li4 (2) − 2 ln(2) Li3 (2) + ln2 (2) Li2 (2).
n=1
n 4 16 2
Collect the values of Li4 (2), Li3 (2), and Li2 (2) given in (1.142), (1.141), and (1.140)
respectively to finalizes the solution.
P∞ (−1)n H n Hn
4.6.6 n=1 n2
Show that
∞
(−1)n H n Hn

X 1 29 3 1
2
= 3 Li 4 − ζ(4) − ln2 (2)ζ(2) + ln4 (2).
n=1
n 2 16 4 8
(4.205)
(−1)n Hn
Solution Multiply both sides of (1.164) by n2 then take the summation,
∞ ∞ ∞
Z 1 !
X (−1)n H n Hn X (−1)n Hn 1 X Hn n
= ln(2) − x dx.
n=1
n2 n=1
n2 0 1 + x n=1 n2
For the integral, make use of the generating function in (2.8),

∞
Z 1 ! Z 1 Z 1
1 X Hn xn Li3 (x) Li3 (1 − x)
2
dx = dx − dx
0 1 + x n=1
n 0 1 + x 0 1+x
| {z } | {z }
I1 I2
1 1 Z 1 2
ln(1 − x) Li2 (1 − x) ln(x) ln (1 − x)
Z Z
1 ζ(3)
+ dx + dx + dx .
1+x 2 0 1+x 1 +x
|0 {z } | {z } | 0
{z }
I3 I4 I5
1
For I1 , expand 1+x in series,
Z 1 ∞ Z 1
Li3 (x) X
n
I1 = dx = − (−1) xn−1 Li3 (x)dx
0 1+x n=1 0

∞
X ζ(3) ζ(2) Hn
=− (−1)n − 2 + 3
n=1
n n n
∞
5 X (−1)n Hn
= ln(2)ζ(3) − ζ(4) − .
4 n=1
n3
For I2 ,
1 1
Li3 (1 − x)
Z Z
1−x=y Li3 (y)
I2 = dx = dy
0 1+x 0 2−y
∞
( )
1 X y n−1
expand in series as
2−y n=1
2n
∞ Z 1
X 1
= n
y n−1 Li3 (y)dy
n=1
2 0
{recall the result of the integral from (3.104) }

∞
X 1 ζ(3) ζ(2) Hn
= − +
n=1
2n n n2 n3
X ∞
1 Hn
= ln(2)ζ(3) − ζ(2) Li2 + n n3
.
2 n=1
2
For I3 ,
1 Z 1
ln(1 − x) Li2 (1 − x)
Z
1−x=y ln(y) Li2 (y)
I3 = dx = dy
0 1+x 0 2−y
∞ Z 1
X 1
= n
y n−1 ln(y) Li2 (y)dy (4.206)
n=1
2 0
∞ Z 1
X 1 ∂ n−1
= n
y Li2 (y)dy
n=1
2 0 ∂n
∞ Z 1
X 1 d
= n dn
y n−1 Li2 (y)dy
n=1
2 0

∞
X 1 d ζ(2) Hn
= −
n=1
2n dn n n2
{use the derivative of the harmonic number given in (1.157)}
∞
!
(2)
X 1 2Hn Hn 2ζ(2)
= + 2 −
n=1
2n n3 n n2
∞ ∞ (2)
X Hn X Hn 1
=2 n n3
+ n n2
− 2ζ(2) Li2 . (4.207)
n=1
2 n=1
2 2
For I4 ,
ln(x) ln2 (1 − x)
1
ln(1 − y) ln2 (y)
Z Z 1
1−x=y
I4 = dx = dy
0 1+x 0 2−y
∞ Z 1
X 1
= n
y n−1 ln(1 − y) ln2 (y)dy
n=1
2 0
∞ Z 1 2
X 1 ∂
= n 2
y n−1 ln(1 − y)dy
n=1
2 0 ∂n
∞ Z 1
X 1 d2
= n dn2
y n−1 ln(1 − y)dy
n=1
2 0
∞
1 d2

X Hn
= −
n=1
2n dn2 n
∞
!
(2) (3)
X 1 2ζ(3) 2ζ(2) 2Hn 2Hn 2Hn
= + − 3 − −
n=1
2n n n2 n n2 n
∞ ∞ (2) ∞ (3)
1 X Hn X Hn X Hn
= 2 ln(2)ζ(3) + 2ζ(2) Li2 −2 − 2 − 2 .
2 n=1
2n n3 n=1
2n n2 n=1
2n n
(4.208)
For I5 ,
I5 = ζ(3) ln(1 + x)|10 = ln(2)ζ(3).
Combine the results of all integrals (I1 to I5 ) and rearrange the terms,
∞
X (−1)n H n Hn 5
2
= ζ(4) − 2 ln(2)ζ(3)
n=1
n 4
∞ ∞ ∞ (3)
X (−1)n Hn X (−1)n Hn X Hn
+ ln(2) + + .
n=1
n2 n=1
n3 n=1
2n n
Grouping the results from (4.137), (4.143), and (4.168) finishes the solution.
P∞ H 2n H2n
4.6.7 n=1 n2
Show that
∞
X H 2n H2n 7
2
= ζ(4). (4.209)
n=1
n 4
H n Hn
Solution Put an = n2 in (1.5):
∞ ∞ ∞
X 1X 1X
a2n = an + (−1)n an ,
n=1
2 n=1 2 n=1
we have
∞ ∞ ∞
X H 2n H2n 1 X H n Hn 1 X (−1)n H n Hn
= + .
n=1
(2n)2 2 n=1 n2 2 n=1 n2
Gather the results from (4.204) and (4.205) then multiply through by 4.
4.7. Harmonic Series with Rational Argument 285
4.7 Harmonic Series with Rational Argument

P∞ (−1)n H n
4.7.1 n=1 n
2
Show that
∞
X (−1)n H n2 1
= ln2 (2) − ζ(2). (4.210)
n=1
n 2
Solution (i) Replace n by n/2 in (3.97),

Z 1 Hn − H n2
xn−1 ln(1 + x)dx = . (4.211)
0 n
Multiply both sides by (−1)n then consider the summation,

∞ ∞ ∞
!
(−1)n Hn X (−1)n H n2
Z 1
X ln(1 + x) X
− = (−x)n dx
n=1
n n=1
n 0 x n=1
Z 1
ln(1 + x) −x
= dx
0 x 1+x
Z 1
ln(1 + x) 1
=− dx = − ln2 (2).
0 1 + x 2
Recall the result from (4.135) to finalize the solution.
Solution (ii) Consider (2.70)

Z 1 Z 1
Hn x=y 2
=− xn−1 ln(1 − x)dx = −2 y 2n−1 ln(1 − y 2 )dy.
n 0 0
Replace n by n/2 to have
H n2
Z 1
=− y n−1 ln(1 − y 2 )dy. (4.212)
n 0
Next, multiply both sides by (−1)n then take the summation,

∞ ∞
!
(−1)n H n2
Z 1
X ln(1 − y 2 ) X n
=− (−y) dy
n=1
n 0 y n=1
Z 1 Z 1
ln(1 − y 2 ) ln(1 − y 2 )

−y
=− dy = dy
0 y 1+y 0 1+y
{write ln(1 − y 2 ) = ln(1 − y) + ln(1 + y)}
1 1
ln(1 − y)
Z Z
ln(1 + y)
= dy + dy.
0 1+y 0 1+y
2
The first integral is 21 ln (2) and the second one is given in (3.27).
For a different approach, set x = −1 in the generating function in (2.34).
P∞ Hn
4.7.2 n=1 n2
2
Show that
∞
X H n2 11
2
= ζ(3). (4.213)
n=1
n 8
Solution Divide both sides of (4.212) by n then take the summation,

∞ Z 1 ∞
X H n2 ln(1 − y 2 ) X y n
= − dy
n=1
n2 0 y n=1
n
Z 1
ln(1 − y 2 ) ln(1 − y)
= dy
0 y
{write ln(1 − y 2 ) = ln(1 − y) + ln(1 + y)}
Z 1 2 Z 1
ln (1 − y) ln(1 + y) ln(1 − y)
= dy + dy
y y
|0 {z } 0
1−y=x
1
ln2 (x) 1
ln(1 + y) ln(1 − y)
Z Z
= dx + dy,
0 1−x 0 y
and these two integrals are given in (3.3) and (3.115).

For a different approach, see (4.36).
P∞ (−1)n H n
4.7.3 n=1 n2
2
Show that
∞
X (−1)n H n2 3
2
= − ζ(3). (4.214)
n=1
n 8
(−1)n
Solution Multiply both sides of (4.212) by n then take the summation,
∞ ∞
(−1)n H n2
Z 1
X ln(1 − y 2 ) X (−y)n
= − dy
n=1
n2 0 y n=1
n
1
ln(1 − y 2 ) ln(1 + y)
Z
= dy
0 y
{write ln(1 − y 2 ) = ln(1 − y) + ln(1 + y)}
Z 1 2 Z 1
ln (1 + y) ln(1 + y) ln(1 − y)
= dy + dy,
0 y 0 y
Gathering the results from (3.38) and (3.115) finishes the solution.
Check (4.42) for a different approach.
P∞ Hn
4.7.4 n=1 n3
2
Show that
∞
H n2

X 1 11 7 1 1 4
3
= −2 Li4 + ζ(4) − ln(2)ζ(3) + ln2 (2)ζ(2) − ln (2).
n=1
n 2 4 4 2 12
(4.215)

Z 1
Hn − H n2
xn−1 ln(1 + x)dx =
0 n
by n2 then consider the summation,

∞ ∞ ∞
Z 1 !
X Hn X H n2 ln(1 + x) X xn
− = dx
n=1
n3 n=1
n3 0 x n=1
n2
Z 1
ln(1 + x) Li2 (x)
= dx
0 x
{expand ln(1 + x) in series}
∞
X (−1)n−1 Z 1
= xn−1 Li2 (x)dx
n=1
n 0

∞
(−1)n−1 ζ(2) Hn
X
= − 2
n=1
n n n
∞
X (−1)n Hn
= ζ(2)η(2) + .
n=1
n3
P∞ Hn
Since n=1 n3 = 54 ζ(4) (given in (4.5)) and ζ(2)η(2) = 54 ζ(4), we have
∞ ∞
X H n2 X (−1)n Hn
= − .
n=1
n3 n=1
n3
The latter sum is calculated in (4.143).
P∞ (−1)n H n
4.7.5 n=1 n3
2
Show that
∞
(−1)n H n2

X 1 27 7 1 1
3
= 2 Li4 − ζ(4)+ ln(2)ζ(3)− ln2 (2)ζ(2)+ ln4 (2).
n=1
n 2 8 4 2 12
(4.216)
(−1)n
Solution Multiply both sides of (4.211) by n2 then consider the summation,
∞ ∞ ∞
(−1)n Hn X (−1)n H n2
Z 1
X ln(1 + x) X (−x)n
− = dx
n=1
n3 n=1
n3 0 x n=1
n2
Z 1
ln(1 + x) Li2 (−x) 1 5
=− dx = Li22 (−1) = ζ(4),
0 x 2 8
and the solution is finalized on plugging in the result from (4.143).
P∞ Hn H n
4.7.6 n=1 n2
2
Show that
∞
Hn H n2

X 1 97 21 3 1
2
= −3 Li4 + ζ(4)− ln(2)ζ(3)+ ln2 (2)ζ(2)− ln4 (2).
n=1
n 2 16 8 4 8
(4.217)
Hn
Solution Multiply both sides of (4.211) by n then consider the summation,
∞ ∞ ∞
Z 1 !
X Hn2 X Hn H n2 ln(1 + x) X Hn n
− = x dx
n=1
n2 n=1
n2 0 x n=1
n
Z 1
ln(1 + x) 1 2
= ln (1 − x) + Li2 (x) dx
0 x 2
Z 1 2 Z 1
1 ln(1 + x) ln (1 − x) ln(1 + x) Li2 (x)
= dx + dx.
2 0 x 0 x
Apply integration by parts in the second integral,

1 Z 1
Li2 (−x) ln(1 − x)
Z
ln(1 + x) Li2 (x)
dx = − Li2 (−x) Li2 (x)|10 − dx
0 x 0 x
∞
(−1)n 1 n−1
Z
5 X
= ζ(4) − x ln(1 − x)dx
4 n=1
n2 0
∞
5 X (−1)n Hn
= ζ(4) + .
4 n=1
n3
Substituting this integral back yields

∞ ∞ ∞
X Hn H n2 5 X Hn2 X (−1)n Hn
= − ζ(4) + −
n=1
n2 4 n=1
n2 n=1
n3
ln(1 + x) ln2 (1 − x)
Z 1
1
− dx.
2 0 x
The solution finalizes on collecting the results from (4.99), (4.143), and (3.118).
P∞ (−1)n Hn H n
4.7.7 n=1 n2
2
Show that
∞
(−1)n Hn H n2

X 1 21 35 5
2
= 5 Li4 − ζ(4) + ln(2)ζ(3) − ln2 (2)ζ(2)
n=1
n 2 4 8 4
5 4
+ ln (2). (4.218)
24
(−1)n Hn
Solution Multiply both sides of (4.211) by n then take the summation,
∞ ∞ ∞
!
(−1)n Hn2 X (−1)n Hn H n2
Z 1
X ln(1 + x) X Hn n
− = (−x) dx
n=1
n2 n=1
n2 0 x n=1
n
{replace x by −x in (2.7) to get the sum}
Z 1
ln(1 + x) 1 2
= ln (1 + x) + Li2 (−x) dx
0 x 2
Z 1 3 Z 1
1 ln (1 + x) ln(1 + x) Li2 (−x)
= dx + dx
2 0 x 0 x
1 1 ln3 (1 + x)
Z
1
= dx − Li22 (−x)|10
2 0 x 2
1
ln3 (1 + x)
Z
1 5
= dx − ζ(4).
2 0 x 16
Reorder the terms,

∞ ∞
(−1)n Hn H n2 (−1)n Hn2 1 1 ln3 (1 + x)
Z
X X 5
2
= 2
− dx + ζ(4).
n=1
n n=1
n 2 0 x 16
Substituting the results from (4.145) and (3.39) finishes the solution.
P∞ (−1)n H n
4.7.8 n=1 n4
2
Show that
∞
X (−1)n H n2 1 25
4
= ζ(2)ζ(3) − ζ(5). (4.219)
n=1
n 8 32
(−1)n
Solution Multiply both sides of (4.211) by n4 then take the summation,
∞ ∞ ∞
(−1)n Hn X (−1)n H n2
Z 1
X ln(1 + x) X (−x)n
− = dx
n=1
n4 n=1
n4 0 x n=1
n3
Z 1 Z 1 2
ln(1 + x) Li3 (−x) IBP 3 Li2 (−x)
= dx = − ζ(2)ζ(3) + dx.
0 x 8 0 x
Recall the relation involving the latter integral from (3.153),

∞ ∞ ∞
X (−1)n H n2 1 7 X Hn X (−1)n Hn
= − ζ(2)ζ(3) − − .
n=1
n4 4 8 n=1 n4 n=1
n4
Collecting the results from (4.6) and (4.150) completes the solution.
For a different solution, see (4.43).
4.8. Harmonic Series with Binomial Coefficient in the Numerator 291
4.8 Harmonic Series with Binomial Coefficient in the

Numerator
P∞ (2n
n ) Hn
4.8.1 n=1 4n n
Show that
∞ 2n

X
n Hn
= 2ζ(2). (4.220)
n=1
4n n
Solution (i) By Taylor series,

∞ 2n
∞ 2n

1 X
n
X
√ = n
x =1+ n
xn
1 − x n=0 4n n=1
4n
or
∞ 2n

X 1
n
xn = √
− 1. (4.221)
n=1
4n
1 −x
R1
Multiply through by − ln(1−x)
x then integrate using − 0 xn−1 ln(1 − x)dx = Hnn ,
∞ 2n
1 Z 1
ln(1 − x) ln(1 − x)
Z
X
n Hn
=− √ dx + dx
4n n x 1−x x
n=1 |0 {z } |0 {z }
√
1−x=y 1−x=y
Z 1 Z 1
ln(y) ln(y)
= −4 2
dy + dy
0 1−y 0 1−y

2 1 1
write = +
1 − y2 1−y 1+y
Z 1 Z 1
ln(y) ln(y)
=− dy − 2 dy,
0 1−y 0 1+y
and these two integrals are given in (3.2) and (3.8).
Solution (ii) Setting x = 1 in (2.39):

∞ 2n
√
X
n Hn n 1− 1−x
x = 2 Li2 √ ,
n=1
4n n 1+ 1−x
we have
∞ 2n

X
n Hn
= 2 Li2 (1) = 2ζ(2).
n=1
4n n
P∞ (2n
n ) (−1) Hn
n
4.8.2 n=1 4n n
Show that
∞ 2n
√
X
n (−1)n Hn
= 2 Li2 2 2 − 3 , (4.222)
n=1
4n n
Proof. Set x = −1 in (2.39),

∞ 2n
√ ! √
X
n (−1)n Hn 1− 2
= 2 Li2 √ = 2 Li2 2 2 − 3 .
n=1
4n n 1+ 2
P∞ (2n
n ) Hn
4.8.3 n=1 4n n2
Show that
∞ 2n

X
n Hn 9
= ζ(3) − 4 ln(2)ζ(2). (4.223)
n=1
4n n 2 2
Solution (i) Divide both sides of (4.221) by x then integrate,

∞ 2n

n Z Z
n x dx dx
X
n
= √ − .
n=1
4 n x 1−x x
√
To evaluate the first integral, set 1 − x = y,
Z Z
dx 2
√ =− dy
1−x 1 − y2
√
1−y 1− 1−x
= ln = ln √
1+y 1+ 1−x
√
1+ 1−x
multiply the argument of the log by √
1+ 1−x

x
= ln √
(1 + 1 − x)2
√
= ln(x) − 2 ln(1 + 1 − x).
Substitute this integral back and note that the second integral is ln(x),
∞ 2n

X
n x
n √
n n
= −2 ln(1 + 1 − x) + C,
n=1
4
where C = 2 ln(2) if we set x = 0. Then, we have

∞ 2n

X
n x
n √
n
= −2 ln(1 + 1 − x) + 2 ln(2). (4.224)
n=1
4 n
R1
xn−1 ln(1 − x)dx = Hn
n ,
∞ 2n
1 √ Z 1
1 − x) ln(1 − x) ln(1 − x)
Z
X
n Hn ln(1 +
=2 dx −2 ln(2) dx
4n n2 x x
n=1 |0 {z } | 0
{z }
√
1−x=y 1−x=y
Z 1 Z 1
y ln(1 + y) ln(y) ln(y)
=8 2
dy − 2 ln(2) dy
0 1−y 0 1−y
2y 1 1
{write = − in the first integral}
1 − y2 1−y 1+y
Z 1 Z 1 Z 1
ln(1 + y) ln(y) ln(1 + y) ln(y) ln(y)
=4 dy − 4 dy −2 ln(2) dy
0 1−y 1 + y 1 −y
|0 {z } 0
IBP
Z 1 Z 1 2 Z 1
ln(1 + y) ln(y) ln (1 + y) ln(y)
=4 dy + 2 dy − 2 ln(2) dy.
0 1−y 0 y 0 1−y
Putting together the results from (3.125), (3.38), and (3.2) finalizes the solution.
Solution (ii) Set x = 1 in (2.40),

∞ 2n
Z 1
X
n Hn ln(1 + t) ln(1 − t)
n n2
= −4 ln(2) Li2 (1) + 2 Li3 (1) − 4 dt.
n=1
4 0 t
The remaining integral is given in (3.115).
P∞ (2n
n ) Hn
(2)
4.8.4 n=1 4n n
Show that
∞ 2n
(2)
X
n Hn 3
= ζ(3). (4.225)
n=1
4n n 2

(2) Z 1
Hn Hn ζ(2)
+ − = xn−1 ln(x) ln(1 − x)dx
n2 n n 0
(2n
n)
by 4n then take the summation over n ≥ 1,
∞ 2n
∞ 2n
(2) ∞ 2n

X
nHn X n Hn X
n 1
+ − ζ(2)
n=1
4nn2 n=1
4n n n=1
4n n
∞
!
Z 1 2n

ln(x) ln(1 − x) X n n
= x dx
0 x n=1
4n
Z 1
ln(x) ln(1 − x) 1
= √ − 1 dx
0 x 1−x
Z 1 Z 1
ln(x) ln(1 − x) ln(x) ln(1 − x)
= √ dx − dx
0 x 1 − x 0 x
| {z } | {z }
1−x→x IBP
1 1 2
ln(x) ln(1 − x)
Z Z
1 ln (x)
= √ dx − dx.
0 x(1 − x) 2 0 1−x
= 6ζ(3) − 6 ln(2)ζ(2),
and the solution completes on collecting the result from (4.223) and writing
∞ 2n

n 1
X
= 2 ln(2),
n=1
4n n
which follows from (4.224) on setting x = 1.
P∞ (2n
n ) H2n
(2)
4.8.5 n=1 4n n
Show that
∞ 2n
(2)
X
n H2n 31
= 2πG + ζ(3). (4.226)
n=1
4n n 8

∞ 2n

arcsin2 (x)

X (2) 1 (2)
√ =2 n
n
H2n − Hn x2n .
1 − x2 n=1
4 4
R1 1
by x then integrate using 2 0
x2n−1 dx = n,
∞ 2n (2) (2) Z π2
H2n − 14 Hn 1

arcsin2 (x)
Z
x=sin u
X
n
√ = dx = u2 csc u du
n=1
4n n 0 x 1−x
2
0
u π2 Z π2 u
IBP 2
= u ln tan −2 u ln tan du
2 0 0 2
| {z }
0
{use the Fourier series of ln(tan x) given in (2.129)}
∞
Z π2 !
X cos((2n + 1)u)
= −2 u −2 du
0 n=0
2n + 1
∞ Z π2
X 1
=4 u cos((2n + 1)u)du
n=0
2n + 1 0
∞
X 1 π cos(nπ) sin(nπ) 1
=4 − −
n=0
2n + 1 2(2n + 1) (2n + 1)2 (2n + 1)2
{write cos(nπ) = (−1)n and sin(nπ) = 0, since n is an integer}
∞ ∞
X (−1)n X 1
= 2π 2
+4
n=0
(2n + 1) n=0
(2n + 1)3
{the first sum is the definition of the Catalan’s constant (see 1.205)}
{and the second sum can be obtained from (1.85)}
7
= 2πG + ζ(3).
2
Thus,
∞ 2n
(2) ∞ 2n
(2)
X
n H2n 7 1 X n Hn
= 2πG + ζ(3) + .
n=1
4n n 2 4 n=1 4n n
The latter sum is given in (4.225).
P∞ (2n
n ) Hn
2
4.8.6 n=1 4n n
Show that
∞ 2n

X
n Hn2 21
= ζ(3). (4.227)
n=1
4n n 2

1 (2)
Hn2 + Hn
Z
xn−1 ln2 (1 − x)dx =
0 n
1 2n

by 4n n then take the summation,
∞ ∞ ∞
(2) !
2n 2n Z 1 2 2n

X
n Hn2 X n Hn ln (1 − x) X
n n
+ = x dx
n=1
4n n n=1
4 n n 0 x n=1
4n
Z 1 2
ln (1 − x) 1
= √ − 1 dx
0 x 1−x
Z 1 2 Z 1 2
ln (1 − x) ln (1 − x)
= √ dx − dx
0 x 1−x 0 x
| {z } | {z }
√
1−x=y 1−x=y
2
1 1
ln2 (y)
Z Z
ln (y)
=8 dy − dy
0 1 − y2 0 1−y
1
ln2 (y) 1
ln2 (y)
Z Z
=4 dy + 3 dy.
0 1+y 0 1−y

∞ 2n

2 Z 1 2 Z 1 2 ∞ 2n
(2)
n Hn ln (y) ln (y) n Hn
X X
=4 dy + 3 dy − .
n=1
4n n 0 1+y 0 1−y n=1
4n n
Gathering the results from (3.9), (3.3), and (4.225) completes the solution.

∞ 2n
√
X
n n 2 1+ 1−x
Hn x = √ ln √ , |x| < 1.
n=1
4n 1−x 2 1−x
R1
by − ln(1−x)
n ,
∞ 2n
1 √
Hn2

ln(1 − x) 1+ 1−x
X Z
n
= −2 √ ln √ dx
n=1
4n n 0 x 1−x 2 1−x
√ Z 1
1−x=y ln(y) 1+y
= −8 2
ln dy
0 1−y 2y
Z 1 2 Z 1
y= 1−x
1+x ln (1 − x) ln(1 − x) ln(1 + x)
= 4 dx −4 dx
x x
|0 {z } 0
1−x=y
1 2 1
ln(1 − x) ln(1 + x)
Z Z
ln (y)
=4 dx − 4 dx.
0 1−y 0 x

P∞ (2n
n ) Hn
2
4.8.7 n=1 4n n2
Show that
∞ 2n

H2

X 1 4
n n
= 32 Li4 −14ζ(4)+7 ln(2)ζ(3)−8 ln2 (2)ζ(2)+ ln4 (2).
n=1
4n n2 2 3
(4.228)

∞ 2n
√
X
n Hn n 1− 1−x
x = 2 Li2 √
n=1
4n n 1+ 1−x
R1
by − ln(1−x)
x then integrate using − xn−1 ln(1 − x)dx = Hnn ,
0
√
∞ 2n
Z 1 ln(1 − x) Li2 1−√1−x
X
n Hn2 1+ 1−x
= −2 dx
n=1
4n n2 0 x

1−y
√
1−x=y
Z 1 y ln(y) Li2 1+y
= −8 dy
1 − y2
0

1−y Z 1 (1 − x) ln 1−x Li2 (x)
1+y =x 1+x
= −4 dx
0 x(1 + x)
Z 1
1 2 1−x
= −4 − ln Li2 (x)dx
0 x 1+x 1+x
Z 1 Z 1
ln(1 − x) Li2 (x) ln(1 + x) Li2 (x)
= −4 dx +4 dx
x x
|0 {z } |0 {z }
I1 I2
1 Z 1
ln(1 − x) Li2 (x)
Z
ln(1 + x) Li2 (x)
−8 dx +8 dx .
1+x 1+x
|0 {z } |0 {z }
I3 I4
For I1 ,
1 1 5
I1 = − Li22 (x)|10 = − ζ 2 (2) = − ζ(4).
2 2 4
For I2 , expand ln(1 + x) in series,
∞
(−1)n−1 1 n−1
X Z
I3 = x Li2 (x)dx
n=1
n 0

∞ ∞
(−1)n−1 ζ(2) Hn (−1)n Hn
X X
= − 2 = ζ(2)η(2) +
n=1
n n n n=1
n3
{collect the result from (4.143)}

1 3 7 1 1 4
= 2 Li4 − ζ(4) + ln(2)ζ(3) − ln2 (2)ζ(2) + ln (2).
2 2 4 2 12
For I3 , perform integration by parts,
1 1 ln2 (1 + x) ln(1 − x)
Z
1 2
I3 = ln (2)ζ(2) + dx
2 2 0 x
1 3
= ln2 (2)ζ(2) − ζ(4).
2 16
For I4 , make use of the dilogarithm reflection formula in (1.119),
1
ln(1 − x)[ζ(2) − ln(x) ln(1 − x) − Li2 (1 − x)]
Z
I4 = dx
0 1+x
ln(x) ln2 (1 − x)
Z 1 Z 1 Z 1
ln(1 − x) ln(1 − x) Li2 (1 − x)
= ζ(2) dx − dx − dx
0 1 + x 0 1 + x 0 1+x
{these three integrals are given in (3.27), (4.208), and (4.207)}
∞ (3) ∞ (2)
5 1 X Hn X Hn
= − ζ(4) − 2 ln(2)ζ(3) + ln2 (2)ζ(2) + 2 +
4 2 n=1
n2n n=1
n2 2 n

1 29 1 1
= 3 Li4 − ζ(4) − ln2 (2)ζ(2) + ln4 (2).
2 16 4 8
The solution completes on grouping the four integrals.

4.9. Harmonic Series with Binomial Coefficient in the Denominator 299
4.9 Harmonic Series with Binomial Coefficient in the

Denominator
P∞ 4n Hn
4.9.1 n=1 (2n) n2
n
Show that
∞
X 4n Hn 7
2n n2 = 6 ln(2)ζ(2) + 2 ζ(3). (4.229)

n=1 n
Solution (i) Let a = b = n in the beta function in (1.52):

Z 1 a−1
x + xb−1
dx = B(a, b),
0 (1 + x)a+b
we have
1
2xn−1 Γ2 (n)
Z
2
dx = = 2n
0 (1 + x)2n Γ(2n) n n
or
1 1 n
xn−1
Z Z
1 1 x
2n
= dx = dx. (4.230)
n n 0 (1 + x)2n 0 x (1 + x)2
4n Hn
Multiply both sides by then consider the summation over n ≥ 1,
n
∞ Z 1 ∞ n !
4n Hn

X 1 X Hn 4x
2n n2 = (1 + x)2
dx

n=1 n 0 x n=1
n
{make use of (2.7) for the sum}
Z 1
1 4x 1 2 4x
= Li2 + ln 1 − dx
0 x (1 + x)2 2 (1 + x)2
Z 1
IBP 2 + 2x 1−x
= ln(x) ln dx
0 x(1 − x) 1+x
Z 1
2 4 1−x
= + ln(x) ln dx
0 x 1−x 1+x
1 Z 1 Z 1
ln(x) ln(1 − x) ln(x) ln(1 − x)
Z
ln(x) ln(1 + x)
=2 dx + 4 dx −2 dx
0 x 1 − x x
|0 {z } 0
1−x→x
Z 1
ln(x) ln(1 + x)
−4 dx
0 1−x
1 Z 1 Z 1
ln(x) ln(1 − x)
Z
ln(x) ln(1 + x) ln(x) ln(1 + x)
=6 dx −2 dx −4 dx
x x 1−x
|0 {z } | 0
{z } 0
IBP IBP
1
ln2 (x) 1
ln2 (x) 1
Z Z Z
ln(x) ln(1 + x)
=3 dx + dx − 4 dx,
0 1−x 0 1+x 0 1−x
and the solution completes on grouping the results from (3.3), (3.9), and (3.125).
Solution (ii) Set x = 1 in (2.40),

∞ Z 1
X 4n Hn ln(1 − t) ln(1 + t)
2n n2
= −4 ln(2) Li 2 (1) + 2 Li 3 (1) − 4 dt.
n=1 n 0 t

√
Solution (iii) Replace z by x in (2.49),
√ ∞ 1
arcsin( x) 1 X 4n xn− 2
√ = .
2 n=1 2n

1−x n
n
R1
Multiply both sides by −2 ln(1−x)
√
x
then integrate using − 0
xn−1 ln(1−x)dx = Hn
n ,
∞ Z 1 √
X 4n Hn arcsin( x) ln(1 − x)
2n n2
= −4 √ √ dx
n=1 n 0 x 1−x
√ Z π2
x=sin u
= −8 u ln(cos u)du
0
{use the Fourier series of ln(cos u) given in (2.127)}
Z π2 " ∞
#
X (−1)n cos(2nu)
= −8 u − ln(2) − du
0 n=1
n
Z π2 ∞ Z π2
X (−1)n
= 8 ln(2) u du + 8 u cos(2nu)du
0 n=1
n 0
2 ∞
(−1)n

π X 1 cos(nπ) π sin(nπ)
= 8 ln(2) +8 − 2+ +
8 n=1
n 4n 4n2 4n
{we have cos(nπ) = (−1)n and sin(nπ) = 0 for integer n}
∞ ∞
X (−1)n X 1
= 6 ln(2)ζ(2) − 2 3
+2 3
n=1
n n=1
n
= 6 ln(2)ζ(2) − 2 Li3 (−1) + 2ζ(3).
The solution completes on writing Li3 (−1) = − 34 ζ(3) given in (1.103).

P∞ 4n H2n
4.9.2 n=1 (2n) n2
n
Show that
∞
X 4n H2n 35
2n
2 = 3 ln(2)ζ(2) + ζ(3). (4.231)
n=1 n
n 4

∞
arcsin(x) 1 X 4n x2n−1
√ =
2 n=1 2n

1 − x2 n
n
R1 H2n
by −4 ln(1 − x) then integrate using − 0
x2n−1 ln(1 − x)dx = 2n ,
∞ Z 1
X 4n H2n arcsin x ln(1 − x)
2n
2
= −4 √ dx
n=1 n
n 0 1 − x2
Z π2
x=sin u
= −4 u ln(1 − sin u)du
0
Z π
2 π u
= −4 u ln 2 sin2 − du
0 4 2
Z π
π u
4 − 2 =t
4
= (16t − 4π) ln(2 sin2 t)dt
0
Z π Z π
4 4
= ln(2) (16t − 4π)dt + 2 (16t − 4π) ln(sin t)dt
0 0
{use the Fourier serier of ln(sin t) given in (2.121)}
∞
Z π Z π4 " #
4 X cos(2nt)
= ln(2) (16t − 4π)dt + 2 (16t − 4π) − ln(2) − dt
0 0 n=1
n
Z π4 ∞ Z π4
X 1
= − ln(2) (16t − 4π)dt − 2 (16t − 4π) cos(2nt)dt
0 n=1
n 0
∞ π
π2

X 1 4 cos(2nt) 2π sin(2nt) 8t sin(2nt) 4
= − ln(2) − −2 − +
2 n=1
n n2 n n 0
∞
" #
nπ

X 1 4 cos 4
2
= 3 ln(2)ζ(2) − 2 2
− 2
n=1
n n n
∞
X cos nπ
∞
X 1
2
= 3 ln(2)ζ(2) − 8 3
+8
n=1
n n=1
n3
∞ ∞
( )
X nπ X
note that an cos = 0 − a2 + 0 + a4 + · · · = (−1)n a2n
n=1
2 n=1
∞ ∞
X (−1)n X 1
= 3 ln(2)ζ(2) − 8 3
+ 8 3
n=1
(2n) n=1
n
∞ ∞
X (−1)n X 1
= 3 ln(2)ζ(2) − 3
+ 8 3
n=1
n n=1
n
= 3 ln(2)ζ(2) − Li3 (−1) + 8ζ(3)
3
= 3 ln(2)ζ(2) + ζ(3) + 8ζ(3)
4
35
= 3 ln(2)ζ(2) + ζ(3).
4
P∞ 4n Hn
4.9.3 n=1 (2n) n3
n
Show that
∞
4n Hn

X 1 1
2n n3
= −8 Li4 + ζ(4) + 8 ln2 (2)ζ(2) − ln4 (2). (4.232)
n=1 n
2 3
√
Solution Set z = x in (2.50),
∞
√ 1 X 4n xn
arcsin2 ( x) = 2.
2 n=1 2n
n
n
R1
n ,
∞ Z 1 √
X 4n Hn arcsin2 ( x) ln(1 − x)
2n n3 = 2 dx

n=1 n 0 x
√ Z π2
x=sin θ
= −8 x2 cot x ln(cos x)dx. (4.233)
0
To compute this integral, recall the Fourier series of cot x ln(cos x) given in (2.139):
∞ Z 1
X 1 − t n−1
cot x ln(cos x) = (−1)n t dt sin(2nx), 0 < x < π.
n=1 0 1+t
On multiplying both sides by x2 then integrating from x = 0 to π/2, we have

Z π
2
x2 cot x ln(cos x)dx
0
 
∞ Z 1 Z π
X 1 − t n−1  2 2
(−1)n

= t dt  x sin(2nx)dx
0 1+t  
n=1 |0 {z }
IBP
∞ Z 1
X 1 − t n−1 cos(nπ) 3ζ(2) cos(nπ)
= (−1)n t dt −
n=1 0 1+t 4n3 4n

1 π sin(nπ)
− 3+
4n 4n2
{note that cos(nπ) = (−1)n and sin(nπ) = 0 for integer n}
∞ Z 1
(−1)n 3ζ(2)(−1)n

X 1 − t n−1 1
= (−1)n t dt − − + 0
n=1 0 1+t 4n3 4n 4n3
{change the order of integration and summation}
∞
!
1 1 1−t tn 3ζ(2)tn (−t)n
Z X
= − − dt
4 0 t(1 + t) n=1 n3 n n3
1 1 1
Z
2
= − (Li3 (t) + 3ζ(2) ln(1 − t) − Li3 (−t)) dt
4 0 t 1+t
1 1 Li3 (t) − Li3 (−t) 1 1 Li3 (t) − Li3 (−t)
Z Z
= dt − dt
4 0 t 2 0 1+t
| {z } | {z }
I1 I2
1 Z 1
ln(1 − t) ln(1 − t)
Z
3 3
+ ζ(2) dt − ζ(2) dt .
4 t 2 1+t
|0 {z } |0 {z }
I3 I4
For I1 ,
7 15
I1 = Li4 (1) − Li4 (−1) = ζ(4) + ζ(4) = ζ(4).
8 8
For I2 ,
Z 1 Z 1
7
IBP ln(1 + t) Li2 (t) ln(1 + t) Li2 (−t)
I2 = ln(2)ζ(3) − dt + dt
4 0 t 0 t
∞
(−1)n 1 n−1
Z
7 X 1 1
= ln(2)ζ(3) + t Li2 (t)dt − Li22 (−t) 0
4 n=1
n 0 2
∞
(−1)n ζ(2) Hn

7 X 5
= ln(2)ζ(3) + − 2 − ζ(4)
4 n=1
n n n 16
∞
7 5 X (−1)n Hn 5
= ln(2)ζ(3) − ζ(4) − 3
− ζ(4)
4 4 n=1
n 16

1 25 1 1 4
= −2 Li4 − ζ(4) + ln2 (2)ζ(2) − ln (2).
2 16 2 12
For I3 ,
I3 = − Li2 (1) = −ζ(2).
For I4 ,
1 1 ∞ Z 1
ln(1 − t) 1−x=y
Z Z
ln(y) X 1
I4 = dt = dy = n
y n−1 ln(y)dy
0 1+t 0 2−y n=1
2 0
∞
X 1 1 1 2 1
=− 2 n
= − Li2 = ln (2) − ζ(2).
n=1
n 2 2 2 2
Combining all four integrals reveals

Z π
2 1 1 1 4
2
x cot x ln(cos x)dx = Li4 − ζ(4) − ln2 (2)ζ(2) + ln (2).
0 2 8 24
Plugging this integral in (4.233) completes the solution.
(2)
P∞ 4n Hn
4.9.4 n=1 (2n) n2
n
Show that
∞ (2)
4n Hn

X 1 1
2n
2 = 8 Li4 − ζ(4) + 4 ln2 (2)ζ(2) + ln4 (2). (4.234)
n=1 n
n 2 3

(2) Z 1
Hn Hn ζ(2)
+ − = xn−1 ln(x) ln(1 − x)dx
n2 n n 0
4n
by n(2n
then consider the summation,
n)
∞ ∞ (2) ∞
X 4n Hn X 4n Hn X 4n 1
2n n3 + 2n − ζ(2) 2n n2
n2

n=1 n n=1 n n=1 n
∞
Z 1 !
ln(x) ln(1 − x) X 4n xn
= 2n n
dx
0 x n=1 n
√
{replace z by x in (2.49) to get the sum} (4.235)
1 √ √
ln(x) ln(1 − x) 2 x arcsin x
Z
= √ dx
0 x 1−x
√ Z π2
x=sin θ
= 16 θ ln(sin θ) ln(cos θ)dθ.
0
π
Let θ → 2 − θ using cos( π2 − θ) = sin θ and sin( π2 − θ) = cos θ,
Z π Z π
2 2 π
θ ln(sin θ) ln(cos θ)dθ = − θ ln(cos θ) ln(sin θ)dθ
0 0 2
Z π Z π2
π 2
= ln(cos θ) ln(sin θ)dθ − θ ln(cos θ) ln(sin θ)dθ.
2 0 0
( Z π
)
2
add θ ln(sin θ) ln(cos θ)dθ to both sides then divide by 2
0
Z π
π 2
= ln(sin θ) ln(cos θ)dθ
4 0
3 15
= ln2 (2)ζ(2) − ζ(4). (4.236)
4 32
Therefore,
∞ ∞ (2) ∞
X 4n Hn X 4n Hn X 4n 1 15
2n n3
+ 2n
2
− ζ(2) 2n
2 = 12 ln2 (2)ζ(2) − ζ(4).
n=1 n n=1 n
n n=1 n
n 2
We have
∞
4n 1
2
X π 15
ζ(2) 2n n2
= ζ(2) = ζ(4)
n=1 n
2 2
follows from (2.50) on setting z = 1. On collecting this result along with (4.232),
the solution is finalized. A different solution may be found in [28, p. 334]
2
P∞ 4n Hn
4.9.5 n=1 2n 2
( )n n
Show that
∞
4n Hn2

X 1 81
2n n2
= −24 Li4 + ζ(4) + 12 ln2 (2)ζ(2) − ln4 (2). (4.237)
n=1 n
2 2

1 (2)
Hn2 + Hn
Z
xn−1 ln2 (1 − x)dx =
0 n
n
4 1
by then take the summation,
(2n
n)
n
∞ ∞ ∞
!
(2) Z 1 2
X 4n Hn2 X 4n Hn ln (1 − x) X 4n xn
2n n2 +
2n
2 = 2n n
dx
n=1 n n=1 n
n 0 x n=1 n
√
{replace z by x in (2.49) to get the sum}
Z 1 2 √ √
ln (1 − x) x arcsin x
= 2 √ dx
0 x 1−x
√ Z π2
x=sin u
= 16 u ln2 (cos u)du
0
Z π
u= π
π
2 −x
2
= 16 − x ln2 (sin x)dx
0 2
Z π Z π
2 2
2
= 8π ln (sin x)dx − 16 x ln2 (sin x)dx,
0 0
where
π
ln2 (t) arcsin t 1
Z 2
Z
sin x=t
x ln2 (sin x)dx √ = dt
0 0 1 − t2
Z 1
IBP ln(t) arcsin2 t
= − dt
0 t
{recall the series expansion of arcsin2 t in (2.50)}
∞ Z 1 ∞
1 X 4n 1 2n−1 1 X 4n 1
=− t ln(t)dt = 4.
2 n=1 2n n2 0 8 n=1 2n

n n
n
For the latter sum, set x = 1 in (2.59):

∞ (2) ∞ (2) ∞
2 X 4n Hn−1 x2n X 4n Hn x2n X 4n x2n
arcsin4 x = 2n = 2n − 2n n4 ,
n2 n2

3 n=1 n n=1 n n=1 n
we obtain
∞ ∞ (2)
X 4n 1 X 4n Hn 15
2n n4
= 2n
2
− ζ(4). (4.238)
n=1 n n=1 n
n 4
Then, we have
π ∞ (2)
1 X 4n Hn
Z 2 15
x ln2 (sin x)dx = − ζ(4).
8 n=1 2n 2

0 n
n 32
Substitute this integral back then rearrange the terms,

∞ Z π2 ∞ (2)
X 4n Hn2 15 2
X 4n Hn
2n n2 = ζ(4) + 8π ln (sin x)dx − 3 2n .
n2

n=1 n
2 0 n=1 n
The solution ends on gathering the results from (3.108) and (4.234).
P∞ 4n H2n
4.9.6 n=1 (2n) n3
n
Show that
∞
4n H2n

X 1 65 5
2n
3
= −20 Li4 + ζ(4)+8 ln2 (2)ζ(2)− ln4 (2). (4.239)
n=1 n
n 2 8 6
Solution Differentiate both sides of (4.230):

Z 1 n
1 1 x
= dx
n 2n (1 + x)2

n 0 x
with respect to n,
Z 1 n
d 1 d 1 x
= dx
dn n 2n
n
dn 0 x (1 + x)2
Z 1 n
1 ∂ x
= dx
0 x ∂n (1 + x)2
Z 1 n
1 x x
= ln dx.
0 x (1 + x)2 (1 + x)2
Let’s find the derivative of n 12n : By the definition of the central binomial coefficient:
(n)

a Γ(a + 1)
= ,
b Γ(b + 1)Γ(a − b + 1)
we have
1 1 Γ2 (n + 1)
2n =
· .
n n
n Γ(2n + 1)
Use Γ(n + 1) = nΓ(n) given in (1.32),
1 1 n2 Γ2 (n) Γ2 (n)
2n =
· = .
n n
n 2nΓ(2n) 2Γ(2n)
Differentiate both sides,
d 1 d Γ2 (n)
=
dn n 2n

n
dn 2Γ(2n)
{use Γ0 (n) = Γ(n)ψ(n) given in (1.167)}
2Γ(2n)Γ2 (n)ψ(n) − 2Γ(2n)Γ2 (n)ψ(2n)
=
2Γ2 (2n)
Γ2 (n)
= (ψ(n) − ψ(2n))
Γ(2n)
{use ψ(n + 1) = Hn − γ given in (1.169)}
2
= (Hn−1 − γ − H2n−1 + γ) 2n
n n

1
write Hn−1 = Hn −
n

1 1 2
= Hn − − H2n +
2n n 2n

n n
2Hn 2H2n 1
= 2n − 2n − 2 2n .
n n n n n n
Therefore, we have
Z 1 n
2Hn 2H2n 1 1 x x
2n − 2n − 2 2n = ln
(1 + x)2 (1 + x)2
dx. (4.240)

n n n n n n 0 x
4n
Now multiply both sides of (4.240) by 2n2 then consider the summation,
∞ ∞ ∞
X 4n Hn X 4n H2n 1 X 4n 1
2n n3 − 2n
3 −
2 n=1 2n
4
n=1 n n=1 n
n n
n
| {z }
S
 n 
∞ 4x
1 11
Z X
x (1+x)2
= ln   dx
2 0 x (1 + x)2 n=1
n 2
Z 1
1 1 x 4x
= ln Li2 dx
2 0 x (1 + x)2 (1 + x)2
1
2(x − 1) 1−x
Z
IBP 5 1 1 2
= − ζ(4) − ln (x) + 2 Li2 (−x) ln dx
4 2 0 2 x(1 + x) 1+x
1 1 ln2 (x) ln(1 − x) 1 1 ln2 (x) ln(1 + x)
Z Z
5
= − ζ(4) + dx − dx
4 2 0 x 2 0 x
| {z } | {z }
I1 I2
1 1
ln(1 − x) Li2 (−x)
Z Z
ln(1 + x) Li2 (−x)
+2 dx −2 dx
x x
|0 {z } |0 {z }
I3 I4
1 2 1 2
ln (x) ln(1 − x)
Z Z
ln (x) ln(1 + x)
− dx + dx
1+x 1+x
|0 {z } |0 {z }
I5 I6
1 1
ln(1 − x) Li2 (−x)
Z Z
ln(1 + x) Li2 (−x)
−4 dx +4 dx .
1+x 1+x
|0 {z } |0 {z }
I7 I8
For I1 , expand ln(1 − x) in series,

∞ ∞
1 1 n−1 2
Z
X X 1
I1 = − x ln (x)dx = −2 = −2ζ(4).
n=1
n 0 n=1
n4
For I2 , expand ln(1 + x) in series,

∞ ∞
−1)n 1 n−1 2 (−1)n
Z
X X 7
I2 = − x ln (x)dx = −2 4
= ζ(4).
n=1
n 0 n=1
n 4
For I3 , expand Li2 (−x) in series,

∞ ∞
(−1)n 1 n−1 (−1)n Hn
X Z X
I3 = 2
x ln(1 − x)dx = − .
n=1
n 0 n=1
n3
For I4 ,
1 5
I4 = − Li22 (−1) = − ζ(4).
2 16
I5 is given in (3.131).
For I6 , set a = 2 in (3.89),

∞
7 X (−1)n Hn
I6 = ζ(4) + 2 .
4 n=1
n3
Li2 (−x)
For I7 , expand 1+x in series given in (2.3),
∞ Z 1
(2)
X
n
I7 = − (−1) Hn−1 xn−1 ln(1 − x)dx
n=1 0
∞ (2)
X (−1)n Hn−1 Hn
=
n=1
n
∞ (2) ∞
X (−1)n Hn Hn X (−1)n Hn
= − .
n=1
n n=1
n3
For I8 , apply integration by parts,
1 1 ln3 (1 + x)
Z
1 2
I8 = − ln (2)ζ(2) + dx
4 2 0 x

21 1 1 1
= 3ζ(4) − ln(2)ζ(3) + ln2 (2)ζ(2) − ln4 (2) − 3 Li4 .
8 2 8 2
For S, multiply both sides of (4.54):

(2) Z 1
Hn Hn ζ(2)
2
+ − = xn−1 ln(x) ln(1 − x)dx
n n n 0
4n
by n(2n
then take the summation,
n)
∞ ∞ (2) ∞
X 4n Hn X 4n Hn X 4n 1
2n n3
+ 2n
2
− ζ(2) 2n n2

n=1 n n=1 n
n n=1 n
∞
Z 1 !
ln(x) ln(1 − x) X 4n xn
= 2n n
dx
0 x n=1 n
{make use of (2.49) for the sum}
1 √ √
ln(x) ln(1 − x) 2 x arcsin x
Z
= √ dx
0 x 1−x
√ Z π2
x=sin θ
= 16 θ ln(sin θ) ln(cos θ)dθ
0
15
= 12 ln2 (2)ζ(2) − ζ(4). (4.241)
2
By setting z = 1 in (2.50), we find

∞
X 4n 1 π2
2n n2 = 2 = 3ζ(2).

n=1 n
Substitute this result in (4.241),

∞ ∞ (2)
X 4n Hn X 4n Hn
2n n3
= − 2n
2
+ 12 ln2 (2)ζ(2). (4.242)
n=1 n n=1 n
n
Adding (4.242) and (4.238) yields

∞ ∞
X 4n 1 2 15 X 4n Hn
S := 2n n4
= 12 ln (2)ζ(2) − ζ(4) − 2n n3 .

n=1 n
4 n=1 n
Collecting all integrals (I1 to I8 ) along with S yields

∞
4n H2n

X 1 67 21
2n
3
= 8 Li 4 − ζ(4) + ln(2)ζ(3) − 7 ln2 (2)ζ(2)
n=1 n
n 2 8 2
∞ (2) ∞ ∞
1 X (−1)n Hn Hn X (−1)n Hn 3 X 4n Hn
+ ln4 (2) + 4 −4 + 3.
3 n=1
n n=1
n3 2 n=1 2n
n
n
Substituting the results from (4.146), (4.143), and (4.232) completes the solution.
P∞
Remark: Usually the two Mathematica commands for approximating n=1 f (n):
NSum[f(n),{n,1,Infinity}]
NSum[f(n),{n,1,Infinity},WorkingPrecision->10]
don’t give the right approximation for series involving the binomial coefficient due to
the slow convergence. The following replacement works fine and with high accuracy:
major=Normal@Series[f(n),{n,Infinity,12}];
majorsum=Sum[major,{n,Infinity}];
majorsum+NSum[f(n)-major,{n,1,Infinity},NSumTerms->20,
WorkingPrecision->20,Method->"WynnEpsilon"]
On reaching the end of the book, I would like to say that there are still a wide range
of results about the harmonic series left to be discovered by the reader by employing
and manipulating the identities provided in the second chapter. Even though the book
has presented different solutions for several problems, there are still more paths to
take to reach the same results, since the realm of harmonic series is full of hidden
secrets and magic.
Table of Mathematica
Commands
R{expression} ComplexExpand[Re[expression]]
J{expression} ComplexExpand[Im[expression]]
sin(x) Sin[x]
sin(x) Cos[x]
tan(x) Tan[x]
sec(x) Sec[x]
csc(x) Csc[x]
cot(x) Cot[x]
arcsin(x) ArcSin[x]
arccos(x) ArcCos[x]
arctan(x) ArcTan[x]
a

b Binomial[a,b]
Γ(x) Gamma[x]
ζ(x) Zeta[x]
β(x) DirichletBeta[x]
η(x) DirichletEta[x]
ln(x) Log[x]
Lia (x) PolyLog[a,x]
Hx HarmonicNumber[x]
(a)
Hx HarmonicNumber[x,a]
Hx Log[2]-(-1)ˆx LerchPhi[-1,1,x+1]
ψ(x) PolyGamma[0,x]
ψ (a) (x) PolyGamma[a,x]
limx→a f (x) Limit[f(x),{x->a}]
limx→a f (x, y) Limit[f(x,y),{x,y}->{a,b}]
y→b
da
dxa f (x) D[f(x),{x,a}]
∂5
∂x2 ∂y 3 f (x, y) D[f(x,y),{x,2},{y,3}]
5
limx→a ∂x∂2 ∂y3 f (x, y) Normal[Series[D[f(x,y),{x,2},{y,3}],{x,a,0}
y→b
,{y,b,0}]]//FullSimplify//Expand
312
Table of Mathematica Commands 313
Rb
f (x)dx Integrate[f(x),{x,a,b}]
Rab
a
f (x) ≈ NIntegrate[f(x),{x,a,b}] (or)
NIntegrate[f(x),{x,a,b},WorkingPrecision
P∞ ->12]
Pn=1 f (n) Sum[f(n),{n,1,Infinity}]
∞
n=1 f (n) ≈ NSum[f(n),{n,1,Infinity}] (or)
NSum[f(n),{n,1,Infinity},WorkingPrecision
->12] (or)
major=Normal@Series[f(n),{n,Infinity,12}];
majorsum=Sum[major,{n,Infinity}];
majorsum+NSum[f(n)-major,{n,1,Infinity},
NSumTerms->20,WorkingPrecision->20,Method
->"WynnEpsilon"]
γ EulerGamma
G Catalan
e E
π
√ Pi
−1 I
References
1. Amrik, S.N.: Sum of series involving central binomial coefficient and harmonic number (2019).
https://arxiv.org/pdf/1806.03998
2. Au, K.C.: Linear relations between logarithmic integrals of high weight and some closed-form
evaluations (2019). https://arxiv.org/pdf/1910.12113.pdf
3. Balendran, S.: Mathematics Stack Exchange (2020). https://math.stackexchange.com/q/3959007
4. Bastien, G.: Elementary methods for evaluating Jordan’s sums (2013). https://arxiv.org/pdf/1301.7
662.pdf
5. Betal, K.C.: Mathematics Stack Exchange (2019). https://math.stackexchange.com/q/3260868
6. Bonar, D.D., Khoury, M.J.: Real Infinite Series, Classroom Resource Materials. The Mathematical
Association of America, Washington, DC (2006)
7. Krantz, S.G.: Real Analysis And Foundations. Chapman & Hall/CRC Press, Washington, DC (2005)
8. Boyadzhi, K.N.: Power series with skew-harmonic numbers, dilogarithms, and double integrals
(2017). https://arxiv.org/ftp/arxiv/papers/1701/1701.04484.pdf
9. D’Aurizio, J: Mathematics Stack Exchange (2016). https://math.stackexchange.com/q/1786846
10. Dutta, R.: Evaluation of a cubic Euler sum (2016). https://www.researchgate.net/publication/31170
3645
11. Flajolet, P., Salvy, B.: Euler Sums and Contour Integral Representations. Exp. Math. 7, 15–35 (1998)
12. Hintze, W.: Mathematics Stack Exchange (2020). https://math.stackexchange.com/q/3581958
13. Ian: Mathematics Stack Exchange (2015). https://math.stackexchange.com/q/1335512
14. Johnson, R.: Mathematics Stack Exchange (2012).https://math.stackexchange.com/q/116212
15. Johnson, R.: Mathematics Stack Exchange (2013).https://math.stackexchange.com/q/469785
16. Lewin, L.: Polylogarithms and Associated Functions. North-Holland, New York (1981)
17. Mathematics Stack Exchange (2020). https://math.stackexchange.com/q/3538399
18. Mathematics Stack Exchange (2020). https://math.stackexchange.com/q/3522967
19. Rao, S.: A formula of S. Ramanujan. Journal of Number Theory. 25, 1–19 (1987)
20. Song, M.: Mathematics Stack Exchange (2019). https://math.stackexchange.com/q/3325928
21. Song, M.: Mathematics Stack Exchange (2019). https://math.stackexchange.com/q/3337600
22. Vălean, C.I.: A master theorem of series and an evaluation of a cubic harmonic series, (2017).
https://www.researchgate.net/publication/317143428
23. Vălean, C.I.: A new powerful strategy of calculating a class of alternating Euler sums (2019).
24. Vălean, C.I.: A new proof for a classical quadratic harmonic series (2016). https://www.researchgate
.net/publication/305699445
314
References 315
25. Vălean, C.I.: A note on two elementary logarithmic integrals, (2020). https://www.researchgate.net
/publication/342703290
26. Vălean, C.I.: A simple strategy of calculating two alternating harmonic series generalizations (2019).
27. Vălean, C.I.: A new perspective on the evaluation of a logarithmic integral, (2020). https://www.rese
archgate.net/publication/339024876
28. Vălean, C.I.: Almost (Impossible) Integrals, Sums, and Series. Springer, New York (2019)
29. Vălean, C.I.: An easy approach to two classical Euler sums, (2020). https://www.researchgate.net/p
ublication/339290253
30. Vălean, C.I.: Mathematics Stack Exchange (2021). https://math.stackexchange.com/q/3997602
35. Weisstein, E.W.: Analytic Continuation. https://mathworld.wolfram.com/AnalyticContinuation.html
36. Weisstein, E.W.: Catalan’s Constant. http://mathworld.wolfram.com/CatalansConstant.html
37. Weisstein, E.W.: Euler–Mascheroni Constant. https://mathworld.wolfram.com/Euler-MascheroniCo
nstant.html
38. Weisstein, E.W.: Gamma Function. https://mathworld.wolfram.com/GammaFunction.html
39. Weisstein, E.W.: Gaussian Integral. https://mathworld.wolfram.com/GaussianIntegral.html
40. Weisstein, E.W.: Legendre Duplication Formula. https://mathworld.wolfram.com/LegendreDuplicat
ionFormula.html
41. WikiBooks: Series developments. https://de.wikibooks.org/wiki/Formelsammlung_Mathematik:
_Reihenentwicklungen#Arkusfunktionen
42. Wikipedia: Beta function. https://en.wikipedia.org/wiki/Beta_function
43. Wikipedia: Differentiation under the integral sign. https://en.wikipedia.org/wiki/Leibniz_integral_r
ule
Index
A Polylogarithm, 30
Abel’s summation, 105, 110, 204, 275 Riemann zeta, 19, 22
Analytic continuation, 165 Functional equation, 10
C
I
Cauchy product, 23, 58, 73, 99, 100, 103, 104
Identity
Central binomial coefficient, 17, 83, 299, 302,
Algebraic, 167, 170, 182
305, 307
Beta symmetry, 18
Constant
Catalan’s, 54, 177, 178, 238, 240, 266, Beta–Gamma, 17
267, 272 Digamma–Harmonic Number, 48, 155,
Euler–Mascheroni, 49, 55 308
Landen’s dilogarithm, 35, 38, 66
Landen’s trilogarithm, 37, 63
E
Euler’s definition of gamma, 13 Trigonometric, 117, 118, 305
Infinite product, 13, 124
Integral
F
Finite product, 6 Caussian, 12
Formula Divergent, 32
Digamma reflection, 47 Double, 21, 224
Dilogarithm inversion, 36 Generalized improper logarithmic, 145
Dilogarithm reflection, 34, 61, 66, 152 Generalized improper polylogarithmic,
Euler’s, 8, 119, 158 139
Euler’s product, 123 Generalized inverse hyperbolic tangent,
Euler’s reflection, 14, 47, 138, 146, 166 156
Geometric series, 20, 23, 26, 90, 158, Generalized inverse tangent, 144
254 Generalized logarithmic, 141
Geometric sries, 7 Generalized polylogarithmic, 140
Legendre duplication, 16 Integration by parts, 10, 20, 69, 153, 154, 175,
Polylogarithm inversion, 39, 144 187, 303, 309
Polylogarithm symmetry, 33
Geometric series, 263 L
Function L’Hopital’s rule, 34, 126, 164
Beta, 17, 95, 161, 166, 299
Digamma, 47, 160
Dirichlet beta, 26 N
Dirichlet eta, 25 Number
Gamma, 10 Complex, 8
Generating, 57, 59, 61, 63, 68, 70, 72, Harmonic, 40, 154, 156
75, 78, 80, 82, 86, 88, 280 Rational harmonic, 78, 202, 285
Polygamma, 50 Skew harmonic, 44, 73, 194, 201, 277
316
Index 317
P Taylor, 8, 23, 74, 89, 291

Partial fraction decomposition, 100, 101, 142, Telescoping, 49
145, 174, 183 Stirling’s approximation, 11, 13
Sum
S Double, 7, 192, 196, 224
Series Generalized Euler, 204
Divergent, 44 Triple, 190, 195
Fourier, 112, 302
Generalized alternating harmonic, 198
Generalized alternating rational T
harmonic, 202 Theorem
Generalized alternating skew harmonic, Binomial, 133
201 Differentiation under the integral sign,
Generalized rational harmonic, 202 51, 97, 138, 146, 205, 307
Generalized skew harmonic, 194 Lebesgue’s dominated convergence, 15

An Introduction To The Harmonic Series and Logarithmic Integ1

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

An Introduction To The Harmonic Series and Logarithmic Integ1

Uploaded by

Copyright:

Available Formats

An Introduction to

The Harmonic Series

For High School Students

Ali Shadhar Olaikhan

Phoenix, Arizona, USA Ali Shadhar Olaikhan

1 Series Transformations and Special Functions 1

1.14.2 Integral Representation . . . . . . . . . . . . . . . . . . . . 26

2 Generating Functions and Powerful Identities 56

2.4.5 Cauchy Product of Li2 (x) Li3 (x) . . . . . . . . . . . . . . 102

3 Logarithmic Integrals 128

R1 ln2a (x) ln(1+x)

4 Harmonic Series 189

Table of Mathematica Commands 312

C The set of complex numbers

ψ The Digamma function

ln The natural logarithm (loge )

Series Transformations and

1.1 Shifting the Sum Index

Replace j by k in the latter equality to finish the proof.

Example 2: Let ak = Hk xk−1 and m = 3 then shift the index by +2,

1.2 Reversing the Order of the Sum Terms

Proof. Following the previous proof, we have

The proof completes on replacing j by k in the latter equality.

which is equivalent to the RHS sum:

but in reversed order.

Example 2: Put ak = k 2 and m = 3,

1.3 Splitting a Sum Into its Odd and Even Parts

and the proof is complete

1.4 Converting the Summand a2n to an

Proof. Starting with the RHS,

1.5 Converting the Summand a2n+1 to an

Divide both sides by 2 to complete the proof.

1.6 Converting the Summand (−1)n a2n to in an

1.7 Converting the Summand (−1)n a2n+1 to in an

Proof. Compare the imaginary parts of both sides of (1.10).

1.8 Converting a Sum to a Product

1.9 Double Summation

Example 1: Let am = pm and bn = pn ,

Example 2: Let am = xm and bn = H n ,

1.10 The Logarithm of a Complex Number

The logarithm of a complex number, z = x + iy, is given by

Proof. We begin with converting to polar coordinates,

On using Euler’s formula:

Take the logarithm of both sides,

ln(x + iy) = ln reiθ = ln(r) + ln eiθ = ln(r) + iθ.

ln(−x + iy) + ln(x + iy) = ln(−x2 − y 2 ) = ln(−1) + ln(x2 + y 2 )

ln(−x + iy) = ln(−1) + ln(x2 + y 2 ) − ln(x + iy)

The value ln(−1) = iπ follows from using the identity:

ln(−x) = ln(x) + iπ, x > 0. (1.21)

To show that, set x = π in Euler’s formula in (1.16),

take the log of both sides,

ln(x) + iπ = ln(−1) + ln(x) = ln(−x).

Using the rules from (1.17) to (1.20) we find:

1.11 Gamma Function

The gamma function is defined by

For a different integral form, set t = −n ln(x) in (1.28),

Divide both sides of (1.29) by nz Γ(z),

On using Γ(z) = (z − 1)! given in (1.33), we obtain

1.11.2 Functional Equation

One of the key properties of the gamma function is

Proof. Replace z by z + 1 in (1.28),

and by using Γ(z + 1) = zΓ(z), we see that:

1.11.3 Stirling’s Approximation