Frontiers in Mathematics

M. Elena Luna-Elizarrarás
Michael Shapiro
Daniele C. Struppa
Adrian Vajiac

Bicomplex
Holomorphic Functions:
The Algebra,
Geometry and Analysis
of Bicomplex
Numbers
Frontiers in Mathematics

Advisory Editorial Board

Leonid Bunimovich (Georgia Institute of Technology, Atlanta)

William Y. C. Chen (Nankai University, Tianjin, China)
Benoît Perthame (Université Pierre et Marie Curie, Paris)
Laurent Saloff-Coste (Cornell University, Ithaca)
Igor Shparlinski (Macquarie University, New South Wales)
Wolfgang Sprössig (TU Bergakademie Freiberg)
Cédric Villani (Institut Henri Poincaré, Paris)
M. Elena Luna-Elizarrarás • Michael Shapiro
Daniele C. Struppa • Adrian Vajiac

Bicomplex Holomorphic
Functions
The Algebra, Geometry and Analysis
of Bicomplex Numbers
M. Elena Luna-Elizarrarás Michael Shapiro
Escuela Sup. de Física y Matemáticas Escuela Sup. de Física y Matemáticas
Instituto Politécnico Nacional Instituto Politécnico Nacional
Mexico City, Mexico Mexico City, Mexico

Daniele C. Struppa Adrian Vajiac

Schmid College of Science and Technology Schmid College of Science and Technology
Chapman University Chapman University
Orange, CA, USA Orange, CA, USA

ISSN 1660-8046 ISSN 1660-8054 (electronic)

Frontiers in Mathematics
ISBN 978-3-319-24866-0 ISBN 978-3-319-24868-4 (eBook)
DOI 10.1007/978-3-319-24868-4

Library of Congress Control Number: 2015954663

Mathematics Subject Classification (2010): 30G35, 32A30, 32A10

Springer Cham Heidelberg New York Dordrecht London

© Springer International Publishing Switzerland 2015
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now
known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the
editors give a warranty, express or implied, with respect to the material contained herein or for any errors
or omissions that may have been made.

Printed on acid-free paper

Springer International Publishing AG Switzerland is part of Springer Science+Business Media

(www.birkhauser-science.com)
Contents

Introduction 1

1 The Bicomplex Numbers 5

1.1 Definition of bicomplex numbers . . . . . . . . . . . . . . . . . . . 5
1.2 Versatility of different writings of bicomplex numbers . . . . . . . . 7
1.3 Conjugations of bicomplex numbers . . . . . . . . . . . . . . . . . 8
1.4 Moduli of bicomplex numbers . . . . . . . . . . . . . . . . . . . . . 9
1.4.1 The Euclidean norm of a bicomplex number . . . . . . . . . 11
1.5 Invertibility and zero-divisors in BC . . . . . . . . . . . . . . . . . 12
1.6 Idempotent representations of bicomplex numbers . . . . . . . . . 15
1.7 Hyperbolic numbers inside bicomplex numbers . . . . . . . . . . . 20
1.7.1 The idempotent representation of hyperbolic numbers . . . 23
1.8 The Euclidean norm and the product of bicomplex numbers . . . . 25

2 Algebraic Structures of the Set of Bicomplex Numbers 29

2.1 The ring of bicomplex numbers . . . . . . . . . . . . . . . . . . . . 29
2.2 Linear spaces and modules in BC . . . . . . . . . . . . . . . . . . . 30
2.3 Algebra structures in BC . . . . . . . . . . . . . . . . . . . . . . . 33
2.4 Matrix representations of bicomplex numbers . . . . . . . . . . . . 35
2.5 Bilinear forms and inner products . . . . . . . . . . . . . . . . . . . 37
2.6 A partial order on the set of hyperbolic numbers . . . . . . . . . . 41
2.6.1 Definition of the partial order . . . . . . . . . . . . . . . . . 41
2.6.2 Properties of the partial order . . . . . . . . . . . . . . . . . 42
2.6.3 D-bounded subsets in D. . . . . . . . . . . . . . . . . . . . . 44
2.7 The hyperbolic norm on BC . . . . . . . . . . . . . . . . . . . . . . 47
2.7.1 Multiplicative groups of hyperbolic and bicomplex
numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

3 Geometry and Trigonometric Representations of Bicomplex Numbers 51

3.1 Drawing and thinking in R4 . . . . . . . . . . . . . . . . . . . . . . 52
3.2 Trigonometric representation in complex terms . . . . . . . . . . . 57
3.3 Trigonometric representation in hyperbolic terms . . . . . . . . . . 62

v
vi Contents

3.3.1 Algebraic properties of the trigonometric representation of

bicomplex numbers in hyperbolic terms . . . . . . . . . . . 65
3.3.2 A geometric interpretation of the hyperbolic trigonometric
representation. . . . . . . . . . . . . . . . . . . . . . . . . . 68

4 Lines and curves in BC 73

4.1 Straight lines in BC . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.1.1 Real lines in the complex plane . . . . . . . . . . . . . . . . 73
4.1.2 Real lines in BC . . . . . . . . . . . . . . . . . . . . . . . . 77
4.1.3 Complex lines in BC . . . . . . . . . . . . . . . . . . . . . . 77
4.1.4 Parametric representation of complex lines . . . . . . . . . 78
4.1.5 More properties of complex lines . . . . . . . . . . . . . . . 81
4.1.6 Slope of complex lines . . . . . . . . . . . . . . . . . . . . . 83
4.1.7 Complex lines and complex arguments of bicomplex
numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
4.2 Hyperbolic lines in BC . . . . . . . . . . . . . . . . . . . . . . . . . 88
4.2.1 Parametric representation of hyperbolic lines . . . . . . . . 91
4.2.2 More properties of hyperbolic lines . . . . . . . . . . . . . . 92
4.3 Hyperbolic and Complex Curves in BC . . . . . . . . . . . . . . . . 95
4.3.1 Hyperbolic curves . . . . . . . . . . . . . . . . . . . . . . . 95
4.3.2 Hyperbolic tangent lines to a hyperbolic curve . . . . . . . 97
4.3.3 Hyperbolic angle between hyperbolic curves . . . . . . . . . 97
4.3.4 Complex curves . . . . . . . . . . . . . . . . . . . . . . . . . 98
4.4 Bicomplex spheres and balls of hyperbolic radius . . . . . . . . . . 101
4.5 Multiplicative groups of bicomplex spheres . . . . . . . . . . . . . . 102

5 Limits and Continuity 107

5.1 Bicomplex sequences . . . . . . . . . . . . . . . . . . . . . . . . . . 107
5.2 The Euclidean topology on BC . . . . . . . . . . . . . . . . . . . . 110
5.3 Bicomplex functions . . . . . . . . . . . . . . . . . . . . . . . . . . 110

6 Elementary Bicomplex Functions 113

6.1 Polynomials of a bicomplex variable . . . . . . . . . . . . . . . . . 113
6.1.1 Complex and real polynomials. . . . . . . . . . . . . . . . . 113
6.1.2 Bicomplex polynomials . . . . . . . . . . . . . . . . . . . . 114
6.2 Exponential functions . . . . . . . . . . . . . . . . . . . . . . . . . 118
6.2.1 The real and complex exponential functions . . . . . . . . . 118
6.2.2 The bicomplex exponential function . . . . . . . . . . . . . 119
6.3 Trigonometric and hyperbolic functions of a bicomplex variable . . 123
6.3.1 Complex Trigonometric Functions . . . . . . . . . . . . . . 123
6.3.2 Bicomplex Trigonometric Functions . . . . . . . . . . . . . 124
6.3.3 Hyperbolic functions of a bicomplex variable . . . . . . . . 127
6.4 Bicomplex radicals . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Contents vii

6.5 The bicomplex logarithm . . . . . . . . . . . . . . . . . . . . . . . 128

6.5.1 The real and complex logarithmic functions. . . . . . . . . . 128
6.5.2 The logarithm of a bicomplex number . . . . . . . . . . . . 129
6.6 On bicomplex inverse trigonometric functions . . . . . . . . . . . . 131
6.7 The exponential representations of bicomplex numbers . . . . . . . 131

7 Bicomplex Derivability and Differentiability 135

7.1 Different kinds of partial derivatives . . . . . . . . . . . . . . . . . 135
7.2 The bicomplex derivative and the bicomplex derivability . . . . . . 137
7.3 Partial derivatives of bicomplex derivable functions . . . . . . . . . 144
7.4 Interplay between real differentiability and derivability of
bicomplex functions . . . . . . . . . . . . . . . . . . . . . . . . . . 152
7.4.1 Real differentiability in complex and hyperbolic terms. . . . 152
7.4.2 Real differentiability in bicomplex terms . . . . . . . . . . . 156
7.5 Bicomplex holomorphy versus holomorphy in two (complex or
hyperbolic) variables . . . . . . . . . . . . . . . . . . . . . . . . . . 159
7.6 Bicomplex holomorphy: the idempotent representation . . . . . . . 162
7.7 Cartesian versus idempotent representations in BC-holomorphy . 167

8 Some Properties of Bicomplex Holomorphic Functions 179

8.1 Zeros of bicomplex holomorphic functions . . . . . . . . . . . . . . 179
8.2 When bicomplex holomorphic functions reduce to constants . . . . 181
8.3 Relations among bicomplex, complex and hyperbolic holomorphies 185
8.4 Bicomplex anti-holomorphies . . . . . . . . . . . . . . . . . . . . . 186
8.5 Geometric interpretation of the derivative . . . . . . . . . . . . . . 188
8.6 Bicomplex Riemann Mapping Theorem . . . . . . . . . . . . . . . 190

9 Second Order Complex and Hyperbolic Differential Operators 193

9.1 Holomorphic functions in C and harmonic functions in R2 . . . . . 193
9.2 Complex and hyperbolic Laplacians . . . . . . . . . . . . . . . . . 194
9.3 Complex and hyperbolic wave operators . . . . . . . . . . . . . . . 197
9.4 Conjugate (complex and hyperbolic) harmonic functions . . . . . . 198

10 Sequences and Series of Bicomplex Functions 201

10.1 Series of bicomplex numbers . . . . . . . . . . . . . . . . . . . . . . 201
10.2 General properties of sequences and series of functions . . . . . . . 202
10.3 Convergent series of bicomplex functions . . . . . . . . . . . . . . . 204
10.4 Bicomplex power series . . . . . . . . . . . . . . . . . . . . . . . . . 205
10.5 Bicomplex Taylor Series . . . . . . . . . . . . . . . . . . . . . . . . 208

11 Integral Formulas and Theorems 211

11.1 Stokes’ formula compatible with the bicomplex Cauchy–Riemann
operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
11.2 Bicomplex Borel–Pompeiu formula . . . . . . . . . . . . . . . . . . 214
viii Contents

Bibliography 219

Index 226
Introduction

The best known extension of the field of complex numbers to the four-dimensional
setting is the skew field of quaternions, introduced by W.R. Hamilton in 1844,
[36], [37]. Quaternions arise by considering three imaginary units, i, j, k that an-
ticommute and such that ij = k. The beauty of the theory of quaternions is that
they form a field, where all the customary operations can be accomplished. Their
blemish, if one can use this word, is the loss of commutativity. While from a purely
algebraic point of view, the lack of commutativity is not such a terrible problem,
it does create many difficulties when one tries to extend to quaternions the fecund
theory of holomorphic functions of one complex variable. Within this context, one
should at least point out that several successful theories exist for holomorphicity
in the quaternionic setting. Among those the notion of Fueter regularity (see for
example Fueter’s own work [27], or [97] for a modern treatment), and the theory
of slice regular functions, originally introduced in [30], and fully developed in [31].
References [97] and [31] contain various quaternionic analogues of the bicomplex
results presented in this book.
It is for this reason that it is not unreasonable to consider whether a four-
dimensional algebra, containing C as a subalgebra, can be introduced in a way that
preserves commutativity. Not surprisingly, this can be done by simply considering
two imaginary units i, j, introducing k = ij (as in the quaternionic case) but now
imposing that ij = ji. This turns k into what is known as a hyperbolic imaginary
unit, i.e., an element such that k2 = 1. As far as we know, the first time that these
objects were introduced was almost contemporary with Hamilton’s construction,
and in fact J.Cockle wrote, in 1848, a series of papers in which he introduced a
new algebra that he called the algebra of tessarines, [15, 16, 17, 18]. Cockle’s work
was certainly stimulated by Hamilton’s and he was the first to use tessarines to
isolate the hyperbolic trigonometric series as components of the exponential series
(we will show how this is done later on in Chapter 6). Not surprisingly, Cockle
immediately realized that there was a price to be paid for commutativity in four
dimensions, and the price was the existence of zero-divisors. This discovery led
him to call such numbers impossibles, and the theory had no further significant
development for a while.
It was only in 1892 that the mathematician Corrado Segre, inspired by the
work of Hamilton and Clifford, introduced what he called bicomplex numbers in

© Springer International Publishing Switzerland 2015 1

M.E. Luna-Elizarrarás et al., Bicomplex Holomorphic Functions, F rontiers in Mathematics,
DOI 10.1007/978-3-319-24868-4_1
2 Introduction

[82], their algebra being equivalent to the algebra of tessarines. It was in his original
1 + ij 1 − ij
papers that Segre noticed that the elements and are idempotents
2 2
and play a central role in the theory of bicomplex numbers. Following Segre, a few
other mathematicians, in particular Spampinato [88, 89] and Scorza Dragoni [83],
developed the first rudiments of a function theory on bicomplex numbers.
The next major push in the study of bicomplex analysis was the work of
J.D.Riley, who in 1953 published his doctoral dissertation [57] in which he further
developed the theory of functions of bicomplex variables. But the most important
contribution was undoubtedly the work of G. B. Price, [56], where the theory of
holomorphic functions of a bicomplex variable (as well as multicomplex variables)
is widely developed. Until this monograph, the work of G. B. Price had to be
regarded as the foundational work in this theory.
In recent years, however, there has been a resurgence of interest in the study
of holomorphic functions on one and several bicomplex variables, as well as a sig-
nificant interest in developing functional analysis on spaces that have a structure
of modules over the ring of bicomplex numbers. Without any pretense of com-
pleteness, we refer in this book to [2, 12, 13, 14, 19, 20, 29, 32, 34, 45, 59, 61, 62,
63, 65, 96]. Most of this new work indicates a need for the development of the
foundations of the theory of holomorphy on the ring of bicomplex numbers, that
better expresses the similarities, and differences, with the classical theory of one
complex variable.
This is the explicit and intentional purpose of this book, which we have writ-
ten as an elementary, yet comprehensive, introduction to the algebra, geometry,
and analysis of bicomplex numbers.

We describe now the structure of this work. Chapter 1 introduces the funda-
mental properties of bicomplex numbers, their definitions, and the different ways
in which they can be written. In particular, we show how hyperbolic numbers can
be recognized inside the set of bicomplex numbers. The algebraic structure of this
set is described in detail in the next chapter, where we define linear spaces and
modules on BC and we introduce a partial order on the set of hyperbolic num-
bers. Maybe the most important contribution in this chapter is the definition of
a hyperbolic-valued norm on the ring of bicomplex numbers. This norm will have
great importance in all future applications of bicomplex numbers. In Chapter 3
we move into geometry, and we spend considerable time in discussing how to visu-
alize the 4-dimensional geometry of bicomplex numbers. We also discuss the way
in which the trigonometric representation of complex numbers can be extended to
the ring of bicomplex numbers. In Chapter 4 we remain in the geometric realm
and discuss lines in BC; in particular we study real, complex, and hyperbolic lines
in BC. We then extend this analysis to the study of hyperbolic and complex curves
in BC. With Chapter 5 we abandon geometry and begin the study of analysis of
bicomplex functions. We discuss here the notion of limit in the bicomplex con-
text, which will be necessary when we study holomorphy in the bicomplex setting.
Introduction 3

Chapter 6 is devoted to a careful and detailed study of the elementary bicomplex

functions such as polynomials, exponentials, trigonometric (and inverse trigono-
metric) functions, radicals, and logarithms. This chapter is particularly interesting
because, while it follows rather closely the exposition one would expect for com-
plex functions, it also shows the significant, and interesting, differences that arise
in this setting. Chapter 7 is, in some sense, the core of the book, as it explores the
notions of bicomplex derivability and differentiability. It is in this chapter that the
different ways in which bicomplex numbers can be written play a fundamental role.
The fundamental properties of bicomplex holomorphic functions are studied in de-
tail in Chapter 8. As one will see throughout the book, bicomplex holomorphic
functions play an interesting role in understanding constant coefficients second
order differential operators (both complex and hyperbolic). This role is explored
in detail in Chapter 9. In Chapter 10 we discuss the theory of bicomplex Taylor
series. Finally, this book ends with a chapter in which we show the way in which
the Stokes’ formula can be used to obtain new and intrinsically interesting integral
formulas in the bicomplex setting.

Acknowledgments. This work has been made possible by frequent exchanges be-
tween the Instituto Politécnico Nacional in Mexico, D.F., and Chapman University
in Orange, California. The authors express their gratitude to these institutions for
facilitating their collaboration. A very special thank you goes to M. J. C. Robles–
Casimiro, who skillfully prepared all the drawings that are included in this volume.
Chapter 1

The Bicomplex Numbers

1.1 Definition of bicomplex numbers

We start directly by defining the set BC of bicomplex numbers by

BC := {z1 + jz2  z1 , z2 ∈ C},
where C is the set of complex numbers with the imaginary unit i, and where i
and j = i are commuting imaginary units, i.e., ij = ji, i2 = j2 = −1. Thus bicom-
plex numbers are “complex numbers with complex coefficients”, which explains
the name of bicomplex, and in what follows we will try to emphasize the simi-
larities between the properties of complex and bicomplex numbers. As one might
expect, although the bicomplex numbers share some structures and properties of
the complex numbers, there are many deep and even striking differences between
these two types of numbers.
Bicomplex numbers can be added and multiplied. If Z = z1 + jz2 and W =
w1 + jw2 are two bicomplex numbers, the formulas for the sum and the product
of two bicomplex numbers are:
Z + W := (z1 + w1 ) + j(z2 + w2 ) (1.1)
and
Z · W := (z1 + jz2 )(w1 + jw2 ) = (z1 w1 − z2 w2 ) + j(z1 w2 + z2 w1 ). (1.2)
Of course there is no need to memorize these formulas; we have just to multiply
term-by-term and take into account that j2 = −1.
The commutativity of the product of the two imaginary units together with
definitions (1.1) and (1.2) readily imply that both operations possess the usual
properties:
Z + W = W + Z, Z + (W + Y ) = (Z + W ) + Y,
DOI 10.1007/978-3-319-24868_4

© Springer International Publishing Switzerland 2015 5

M.E. Luna-Elizarrarás et al., Bicomplex Holomorphic Functions, F rontiers in Mathematics,
DOI 10.1007/978-3-319-24868-4_2
6 Chapter 1. The Bicomplex Numbers

that is, the addition is commutative and associative;

Z · W = W · Z, Z · (W · Y ) = (Z · W ) · Y,

which means that the multiplication is commutative and associative;

Z · (W + Y ) = Z · W + Z · Y ,

that is, the multiplication distributes over addition.

The bicomplex numbers 0 = 0 + 0 · j and 1 = 1 + 0 · j play the roles of the
usual zero and one:
0 + Z = Z + 0 = Z,
1·Z =Z ·1=Z.
Until now, we have used the denotation C for the field of complex numbers.
Working with bicomplex numbers, the situation becomes more subtle since inside
the set BC there are more than one subset which has the “legitimate” right to bear
the name of the field of complex numbers; more exactly, there are two such subsets.
One of them is the set of those bicomplex numbers with z2 = 0 : Z = z1 + j0 = z1 ;
we will use the notation C(i) for it. Since j has the same characteristic property
j2 = −1, then another set of complex numbers inside BC is C(j) := {z1 + jz2 |
z1 , z2 ∈ R }. Of course, C(i) and C(j) are isomorphic fields but coexisting inside
BC they are different. We will see many times in what follows that there is a
certain asymmetry in their behavior.
The set of hyperbolic numbers D can be defined intrinsically (independently
of BC) as the set 
D := {x + ky  x, y ∈ R},
where k is a hyperbolic imaginary unit, i.e., k2 = 1, commuting with both real
numbers x and y. In some of the existing literature, hyperbolic numbers are also
called duplex, double or bireal numbers.
Addition and multiplication operations of the hyperbolic numbers have the
obvious definitions, we just have to replace k2 by 1 whenever it occurs. For exam-
ple, for two hyperbolic numbers z1 = x1 + ky1 and z2 = x2 + ky2 their product
is
z1 · z2 = (x1 x2 + y1 y2 ) + k(x1 y2 + x2 y1 ) .
Working with BC, a hyperbolic unit k arises from the multiplication of the
two imaginary units i and j: k := ij. Thus, there is a subset in BC which is
isomorphic as a ring to the set of hyperbolic numbers: the set

D = {x + ijy x, y ∈ R}

inherits all the algebraic definitions, operations and properties from BC.
1.2. Versatility of different writings of bicomplex numbers 7

The following subset of D:


D+ := {x + ky x2 − y 2 ≥ 0, x ≥ 0 }

will be especially useful later. We will call its elements “non-negative hyperbolic
numbers”; the set

D+ \ {0} = {x + ky x2 − y 2 ≥ 0, x > 0 }

will be called the set of “positive hyperbolic numbers”. Such definitions of “non-
negativeness” and of “positiveness” for hyperbolic numbers do not look intuitively
clear but later on we will give them other descriptions clarifying the reason for such
names. It turns out that the non-negative hyperbolic numbers play with respect to
all hyperbolic numbers a role deeply similar to that of real non-negative numbers
with respect to all real numbers.
The set
 
D− := {x + ky x2 − y 2 ≥ 0, x ≤ 0 } = {z − z ∈ D+ }

will bear the name of non-positive hyperbolic numbers; and of course the set

D− \ {0} = {x + ky x2 − y 2 ≥ 0, x < 0 }

is the set of negative hyperbolic numbers. Clearly, there are hyperbolic numbers
which are neither non-negative nor non-positive.

1.2 Versatility of different writings of bicomplex

numbers
A bicomplex number defined as Z = z1 +jz2 admits several other forms of writing,
or representations, which show different aspects of this number and which will
help us to understand better the structure of the set BC. First of all, if we write
z1 = x1 + iy1 , z2 = x2 + iy2 with real numbers x1 , y1 , x2 , y2 , then any bicomplex
number can be written in the following different ways:

Z = (x1 + iy1 ) + j (x2 + iy2 ) =: z1 + j z2 (1.3)

= (x1 + jx2 ) + i (y1 + jy2 ) =: ζ1 + i ζ2 (1.4)
= (x1 + ky2 ) + i (y1 − kx2 ) =: z1 + i z2 (1.5)
= (x1 + ky2 ) + j (x2 − ky1 ) =: w1 + j w2 (1.6)
= (x1 + iy1 ) + k(y2 − ix2 ) =: w1 + k w2 (1.7)
= (x1 + jx2 ) + k(y2 − jy1 ) =: ω1 + k ω2 (1.8)
= x1 + iy1 + jx2 + ky2 , (1.9)
8 Chapter 1. The Bicomplex Numbers

where z1 , z2 , w1 , w2 ∈ C(i), ζ1 , ζ2 , ω1 , ω2 ∈ C(j), and z1 , z2 , w1 , w2 ∈ D. Equa-

tion (1.9) says that any bicomplex number can be seen as an element of R4 ;
meanwhile formulas (1.3) and (1.7) allow us to identify Z with elements in C2 (i)
and formulas (1.4) and (1.8) with elements in C2 ( j); similarly formulas (1.5) and
(1.6) identify Z with elements in D2 := D × D.

1.3 Conjugations of bicomplex numbers

The structure of BC (there are two imaginary units of complex type and one
hyperbolic unit in it) suggests three possible conjugations on BC:
(i) Z := z 1 + j z 2 (the bar-conjugation);
†
(ii) Z := z1 − j z2 (the † -conjugation);
  †
(iii) Z ∗ := Z = (Z † ) = z 1 − j z 2 (the ∗ -conjugation),
where z 1 , z 2 are usual complex conjugates to z1 , z2 ∈ C(i).
Let us see how these conjugations act on the complex numbers in C(i) and
in C( j) and on the hyperbolic numbers in D. If Z = z1 ∈ C(i), i.e., z2 = 0, then
Z = z1 = x1 + iy1 and one has:
Z = z 1 = x1 − iy1 = z1∗ = Z ∗ , Z † = z1† = z1 = Z,
that is, both the bar-conjugation and the ∗-conjugation, restricted to C(i), coincide
with the usual complex conjugation there, and the †-conjugation fixes all elements
of C(i).
If Z = ζ1 belongs to C( j), that is, ζ1 = x1 + jx2 , then one has:
ζ 1 = ζ1 , ζ1∗ = x1 − jx2 = ζ † ,
that is, both the ∗-conjugation and the †-conjugation, restricted to C( j) coincide
with the usual conjugation there. In order to avoid any confusion with the notation,
from now on we will identify the conjugation on C( j) with the †-conjugation. Note
also that any element in C( j) is fixed by the bar-conjugation.
Finally, if Z = x1 + ijy2 ∈ D, that is, y1 = x2 = 0, then
Z = x1 − ijy2 = Z † , Z ∗ = Z,
thus, the bar-conjugation and the †-conjugation restricted to D coincide with the
intrinsic conjugation there. We will use the bar-conjugation to denote the latter.
Note that any hyperbolic number is fixed by the ∗-conjugation.
Using formulas (1.4)–(1.9), the bicomplex conjugations (i)–(iii) defined above
can be written as

(i’) Z = ζ1 − i ζ2 = z1 − i z2 = w1 + j w2 = w1 − k w2 = ω1 − k ω2
= x 1 − i y1 + j x 2 − k y2 ;
1.4. Moduli of bicomplex numbers 9

(ii’) Z † = ζ1† + i ζ2† = z1 + i z2 = w1 − j w2 = w1 − k w2 = ω1† − k ω2†

= x 1 + i y1 − j x 2 − k y2 ;

(iii’) Z ∗ = ζ1† − i ζ2† = z1 − i z2 = w1 − j w2 = w1 + k w2 = ω1† + k ω2†

= x 1 − i y1 − j x 2 + k y2 .

Each conjugation is an additive, involutive, and multiplicative operation on BC:

† ∗
(Z + W ) = Z + W , (Z + W ) = Z † + W † , (Z + W ) = Z ∗ + W ∗ (1.10)
 † † ∗
Z = Z, Z = Z, (Z ∗ ) = Z (1.11)
† † † ∗ ∗ ∗
(Z · W ) = Z · W , (Z · W ) = Z · W , (Z · W ) = Z · W . (1.12)

1.4 Moduli of bicomplex numbers

In the complex case the modulus of a complex number is intimately related with
the complex conjugation: by multiplying a complex number by its conjugate one
gets the square of its modulus. Applying this idea to each of the three conjugations
introduced in the previous section, three possible “moduli” arise in accordance
with the formulas for their squares:

• |Z|2i := Z · Z † = z12 + z22

 
= |ζ1 |2 − |ζ2 |2 + 2 Re (ζ1 ζ2† )i
   
= |z1 |2hyp + |z2 |2hyp + z1 z2 − z1 z2 j
   
= |w1 |2hyp − |w2 |2hyp + w1 w2 + w1 w2 i

= w12 − w22
   
= |ω1 |2 − |ω2 |2 − 2 Im ω1† ω2 i ∈ C(i);
 
• |Z|2j := Z · Z = |z1 |2 − |z2 |2 + 2 Re (z1 z 2 )j
= ζ12 + ζ22
   
= |z1 |2hyp − |z2 |2hyp + z1 z2 + z1 z2 j
   
= |w1 |2hyp + |w2 |2hyp + w1 w2 − w1 w2 i
 
= |w1 |2 − |w2 |2 + (w2 w1 − w1 w2 ) k
= ω12 − ω22 ∈ C(j);
10 Chapter 1. The Bicomplex Numbers
 
• |Z|2k := Z · Z ∗ = |z1 |2 + |z2 |2 − 2 Im (z1 z 2 )k
 
= |ζ1 |2 + |ζ2 |2 − 2 Im (ζ1 ζ2† )k
= z21 + z22
= w21 + w22
 
= |w1 |2 + |w2 |2 + (w2 w1 + w1 w2 ) k
 
= |ω1 |2 + |ω2 |2 + (ω1 ω2∗ + ω2 ω1∗ ) k ∈ D,
where for a complex number z (in C(i) or C(j)) we denote by |z| its usual modulus
and for a hyperbolic number z = a + kb we use the notation |z|2hyp = a2 − b2 .
Unlike what happens in the complex case, these moduli are not R+ -valued.
The first two moduli are complex-valued (in C(i) and C( j) respectively), while the
last one is hyperbolic-valued.
√
The value of |Z|i = Z · Z † , being the square root of a complex number, is
†
√ for the complex number z = Z · Z , if z
determined by the following convention:
non-negative real number, then z denotes its non-negative value; otherwise,
is a √
the z denotes the value of the square root of z in the upper half-plane. In many
standard references, this latter one is also called the “principal” square root of z.
Although in general |Z|i is a C(i)-complex number, nevertheless if Z is in
C( j), then its C(i)-complex modulus |Z|i coincides with the usual modulus of the
complex number ζ1 = x1 + jx2 : since z1 = x1 + i0, z2 = x2 + i0, then

|Z|i = x21 + x22 = |ζ1 | .

Hence the restriction of the quadratic form z12 + z22 onto the real two-dimensional
plane C(j) determines the usual Euclidean structure on this plane.
We make similar conventions for the C(j)-valued modulus

|Z|j = Z · Z.

We note again that in the special case when Z = z1 = xi + iy1 , we get:


|Z|j = x21 + y12 = |z1 | ,

hence the restriction of the quadratic form

ZZ = ζ12 + ζ22

onto the real two-dimensional plane C(i) determines the usual Euclidean structure
on this plane.
We observe here a kind of a “dual” relation between the two types of complex
moduli and the respective complex numbers: if Z = z1 ∈ C(i), then

|Z|i = |z1 |i = z12 ,
1.4. Moduli of bicomplex numbers 11

which, in general, is not equal to |z1 | but is equal to z1 or −z1 ; but somewhat
paradoxically, if Z = ζ1 ∈ C(j), then |Z|i = |ζ1 |. 
Similarly, the |Z|j of C(j)-numbers, Z = ζ1 , is |Z|j = |ζ1 |j = ζ12 , meanwhile
if Z = z1 ∈ C(i), then |z1 |j = |z1 |. We will refer to |Z|i , |Z|j as the C(i)- and
C(j)-valued moduli of the bicomplex number Z respectively.
The last modulus introduced has its square, | · |2k , which is hyperbolic-valued,
and later we will show that the modulus itself can be chosen hyperbolic-valued.
For its square the following holds:
 
|Z|2k = |z1 |2 + |z2 |2 + k (−2 Im(z1 z 2 ) ) =: x + k y ,

where x and y satisfy x2 − y 2 ≥ 0 (this is proved using the fact that | Im(z1 z 2 ) | ≤
|z1 | · |z2 |). Thus |Z|2k ∈ D+ .
We will specify the value of the square root of a hyperbolic number later on.
Although these moduli are not real-valued, nevertheless they preserve, for-
tunately, an important property related with the multiplication; specifically, we
have:

|Z · W |2i = |Z|2i · |W |2i ,

|Z · W |2j = |Z|2j · |W |2j ,
|Z · W |2k = |Z|2k · |W |2k .

1.4.1 The Euclidean norm of a bicomplex number

Since all the above moduli are not real valued, we will consider also the Euclidean
norm on BC when it is seen as

C2 (i) := C(i) × C(i) = { (z1 , z2 ) | z1 + j z2 ∈ BC } ,

or as
C2 (j) = { (ζ1 , ζ2 ) | ζ1 + i ζ2 ∈ BC } ,

or as
R4 = { (x1 , y1 , x2 , y2 ) | (x1 + i y1 ) + j(x2 + i y2 ) ∈ BC } .

The Euclidean norm |Z| is related with the properties of bicomplex numbers via
the D+ -valued modulus:
   
|Z| = |z1 |2 + |z2 |2 = |ζ1 |2 + |ζ2 |2 = Re (|Z|2k ) = x21 + y12 + x22 + y22 ,

and it is again direct to prove that

√
|Z · W | ≤ 2 |Z| · |W | . (1.13)
12 Chapter 1. The Bicomplex Numbers

Indeed, for Z = z1 + jz2 and W = w1 + jw2 one has:

|Z · W |2 = |z1 w1 − z2 w2 |2 + |z1 w2 + z2 w1 |2
2 2
≤ (|z1 ||w1 | + |z2 ||w2 |) + (|z1 ||w2 | + |z2 ||w1 |)
= |z1 |2 |w1 |2 + |z2 |2 |w2 |2 + 2|z1 ||w1 ||z2 ||w2 |
+ |z1 |2 |w2 |2 + |z2 |2 |w1 |2 + 2|z1 ||w1 ||z2 ||w2 |
≤ |z1 |2 |w1 |2 + |z1 |2 |w2 |2 + |z2 |2 |w2 |2 + |z2 |2 |w1 |2
+ |z1 |2 |w1 |2 + |z2 |2 |w2 |2 + |z1 |2 |w2 |2 + |z2 |2 |w1 |2
 2 
= 2 |z1 | |w1 |2 + |z1 |2 |w2 |2 + |z2 |2 |w2 |2 + |z2 |2 |w1 |2
  
= 2 |z1 |2 + |z2 |2 |w1 |2 + |w2 |2
= 2|Z|2 |W |2 ,
where first, we used the triangle inequality and then we used the fact that given
any two real numbers a and b, then 2ab ≤ a2 + b2 .
We will obtain below more properties of the interplay between the Euclidean
norm and the product of bicomplex numbers.

1.5 Invertibility and zero-divisors in BC

We know already that
Z · Z † = |Z|2i ∈ C(i) , (1.14)
Z ·Z = |Z|2j ∈ C( j) , (1.15)
∗
Z ·Z = |Z|2k ∈D (1.16)
(compare with the complex situation where z · z = |z|2 ).
Let us analyze (1.14). If Z = 0 but |Z|i = 0, then Z is obviously a zero-divisor
since Z † is also different from zero. But if |Z|i = 0 the number Z is invertible.
Indeed, in this case, dividing both sides of (1.14) over the right-hand side one gets:
Z†
Z· = 1,
|Z|2i
thus the inverse of an invertible bicomplex number Z is
Z†
Z −1 = ,
|Z|2i
similarly to what happens in the complex case. We have therefore obtained a
complete description both of the invertible elements and non-invertible elements
in BC.
In a complete analogy we analyze formulas (1.15) and (1.16) arriving at the
following conclusions:
1.5. Invertibility and zero-divisors in BC 13

1. A bicomplex number Z = 0 is invertible if and only if |Z|j = 0 or, equivalently,

|Z|k is not a zero-divisor and in this case the inverse of Z is

Z Z∗
Z −1 = = .
|Z|2j |Z|2k

2. A bicomplex number Z = 0 is a zero-divisor if and only if |Z|j = 0 or,

equivalently, |Z|k is a zero-divisor.
Let us see what all this means working with specific representations of a
bicomplex number. Assume that Z is given as Z = z1 + jz2 , then |Z|2i = z12 + z22 .
In this case Z is invertible if and only if z12 + z22 = 0 and the inverse of Z is

Z† z1 − jz2
Z −1 = = 2 .
z12 + z22 z1 + z22

If both z1 and z2 are non-zero but the sum z12 + z22 = 0, then the corresponding
bicomplex number Z = z1 + jz2 is a zero-divisor. This is equivalent to z12 = −z22 ,
i.e.,
z1 = ±iz2 , (1.17)
and thus all zero-divisors in BC are of the form:

Z = λ(1 ± ij), (1.18)

where λ runs the whole set C(i) \ {0}.

One wonders if the description (1.18) of zero-divisors depends on the form of
writing Z and what happens if Z = ζ1 + iζ2 with ζ1 , ζ2 ∈ C( j). In this case

Z · Z † = 0 ⇐⇒ | ζ1 |=| ζ2 | and Re(ζ1 ζ2† ) = 0. (1.19)

At first sight, we have something quite different from (1.17). Note however that
Re(ζ1 ζ2† ) is the Euclidean inner product in R2 , hence (1.19) means that ζ1 and ζ2
are orthogonal in C(j) and with the same magnitude (i.e., with the same modulus
of complex numbers), and thus

ζ1 = ± j ζ2 .

Hence, a zero-divisor Z = ζ1 + i ζ2 becomes

Z = ζ1 ± ijζ1 = ζ1 (1 ± i j) (1.20)

with ζ1 running in C(j) \ {0}.

Observe that (1.20) can be obtained as well by recalling that

Z · Z = ζ12 + ζ22 = 0
14 Chapter 1. The Bicomplex Numbers

which uses yet another conjugation, not the †-conjugation but the bar-conjugation.
It is possible to give several other descriptions of the set of zero-divisors
using all the three conjugations as well as formulas (1.3)–(1.9). This we leave as
an exercise to the reader.
We denote the set of all zero-divisors in BC by S, and we set S0 := S ∪ {0}.
We can summarize this discussion as follows.
Theorem 1.5.1. Let Z = 0, then the following are equivalent.
1. The bicomplex number Z is invertible.
2. Z is not a zero-divisor.
3. Z · Z † = 0.
4. Z · Z = 0.
5. Z · Z ∗ ∈ S0 .
6. |Z|i = 0.
7. |Z|j = 0.
8. |Z|k ∈ S0 .
9. If Z is given as Z = z1 + jz2 , then z12 + z22 = 0.
10. If Z is given as Z = ζ1 + iζ2 , then ζ12 + ζ22 = 0.
Since BC is a ring (we will comment on this with more detail in the next
chapter) it is worth to single out the equivalence between (1) and (2) in Theorem
1.5.1. Indeed, in a general ring, the set of non-zero elements which are not zero-
divisors is a different set from the set of invertible elements; from this point of
view BC is a remarkable exception.
Of course the above Theorem allows us to give immediately a “dual” char-
acterization of the set of zero-divisors.
Corollary 1.5.2. Let Z = 0, then the following are equivalent.
1. Z is not invertible.
2. Z is a zero-divisor.
3. Z · Z † = 0 = Z · Z.
4. Z · Z ∗ ∈ S0 .
5. |Z|i = 0 = |Z|j .
6. |Z|k ∈ S0 .
7. If Z is given as Z = z1 + jz2 , then z12 + z22 = 0.
8. If Z is given as Z = ζ1 + iζ2 , then ζ12 + ζ22 = 0.
1.6. Idempotent representations of bicomplex numbers 15

1.6 Idempotent representations of bicomplex numbers

It turns out that there are two very special zero-divisors.
Proposition 1.6.1. The bicomplex numbers
1 + ij 1 − ij
e := and e† :=
2 2
have the properties:
e · e† = 0
(thus, each of them is a zero-divisor);

e2 = e, (e† )2 = e†

(thus, they are idempotents);

e + e† = 1, e − e† = ij .

The properties of the idempotents e and e† cause many strange phenomena.

One of them is the following
Corollary 1.6.2. There holds:

i e = −j e, i e† = j e† ,
k e = e, ke† = −e† . (1.21)

The next property has no analogs for complex numbers, and it exemplifies
one of the deepest peculiarities of the set of bicomplex numbers. For any bicomplex
number Z = z1 + jz2 ∈ BC we have:
z1 − iz2 + z1 + iz2 z2 + iz1 + z2 − iz1
Z = z1 + jz2 = +j
2 2
z1 − iz2 z1 + iz2 z1 − iz2 z1 + iz2
= + + ij − ij
2 2 2 2
1 + ij 1 − ij
= (z1 − iz2 ) + (z1 + iz2 ) ,
2 2
that is,
Z = β 1 e + β 2 e† , (1.22)
where β1 := z1 −iz2 and β2 := z1 +iz2 are complex numbers in C(i). Formula (1.22)
is called the C(i)-idempotent representation of the bicomplex number Z.
It is obvious that since β1 and β2 are both in C(i), then β1 e + β2 e† = 0 if and
only if β1 = 0 = β2 . This implies that the above idempotent representation of the
bicomplex number Z is unique: indeed, assume that Z = 0 has two idempotent
representations, say,
Z = β1 e + β2 e† = β1 e + β2 e† ,
16 Chapter 1. The Bicomplex Numbers

then 0 = (β1 − β1 ) e + (β2 − β2 ) e† and thus β1 = β1 , β2 = β2 .

The following proposition shows the advantage of using the idempotent rep-
resentation of bicomplex numbers in all algebraic operations.
Proposition 1.6.3. The addition and multiplication of bicomplex numbers can be
realized “term-by-term” in the idempotent representation (1.22). Specifically, if
Z = β1 e + β2 e† and W = ν1 e + ν2 e† are two bicomplex numbers, then

Z + W = (β1 + ν1 ) e + (β2 + ν2 ) e† ,
Z · W = (β1 ν1 ) e + (β2 ν2 ) e† ,
Z n = β1n e + β2n e† .

The proof of the formulas in the proposition above relies simply on the rather
specific properties of the numbers e and e† . For example, let us prove the second
property:
   
Z · W = β 1 e + β 2 e† · ν 1 e + ν 2 e †
= β1 e · ν1 e + β1 e · ν2 e† + β2 e† · ν1 e + β2 e† · ν2 e†
= β 1 ν 1 · e + β 1 ν 2 · 0 + β 2 ν 1 · 0 + β 2 ν 2 · e†
= β 1 ν 1 · e + β 2 ν 2 · e† .

We used the fact that e and e† are idempotents, i.e., each of them squares to itself,
and that their product is zero.
We showed after formula (1.22) that the coefficients β1 and β2 of the idem-
potent representation are uniquely defined complex numbers. But this refers to
the complex numbers in C(i), and the paradoxical nature of the idempotents e
and e† manifests itself as follows.
Take a bicomplex number Z written in the form Z = ζ1 + i ζ2 , with ζ1 , ζ2 ∈
C(j). Then a direct computation shows:

Z = α1 e + α2 e† := (ζ1 − j ζ2 )e + (ζ1 + j ζ2 )e† , (1.23)

where α1 := ζ1 − j ζ2 and α2 := ζ1 + j ζ2 are complex numbers in C(j). So,

we see that as a matter of fact every bicomplex number has two idempotent
representations with COMPLEX coefficients, one with coefficients in C(i), and
the other with coefficients in C(j):

Z = β 1 e + β 2 e † = α 1 e + α 2 e† . (1.24)

Let us find out which is the relation between them. One has that

eZ = β1 e = α1 e

and
e† Z = β 2 e† = α 2 e † ,
1.6. Idempotent representations of bicomplex numbers 17

thus the authentic uniqueness consists of the fact that not the coefficients β1 and
α1 (or β2 and α2 ) are equal, but the products β1 e and α1 e (or β2 e† and α2 e† ) are
equal respectively. What is more, β1 e = α1 e is equivalent to (β1 − α1 )e = 0, but
since e is a zero-divisor, then β1 − α1 is also a zero-divisor, that is, β1 − α1 = A· e† ,
where A can be chosen either in C(i) or in C(j). The latter is justified with the
following reasoning. Take β1 , β2 to be β1 = c1 + id1 , β2 = c2 + id2 , then

Z = β1 e + β2 e† = (c1 + i d1 ) e + (c2 + i d2 ) e†
= c 1 e − j d 1 e + c 2 e† + j d 2 e†
= e · (c1 − j d1 ) + e† · (c2 + j d2 )
= e · α 1 + e† · α 2 ,

where α1 = c1 − j d1 , α2 = c2 + j d2 ; thus

β1 − α1 = c1 + i d1 − c1 + j d1 = d1 (i + j)
= i d1 (1 − i j)
= 2 d1 i e† = 2 d1 j e† .

Example 1.6.4. Consider the bicomplex number:

Z = (1 + i) + j (3 − 2i) =: z1 + j z2 .

Then β1 = z1 − i z2 = −1 − 2i and β2 = z1 + i z2 = 3 + 4i, so in the first idempotent

representation we have:

Z = (−1 − 2i)e + (3 + 4i)e† .

Now we write the same bicomplex number as

Z = (1 + 3j) + i (1 − 2j) =: ζ1 + i ζ2 .

Then α1 = ζ1 − jζ2 = −1 + 2j and α2 = ζ1 + jζ2 = 3 + 4j. The second idempotent

representation of Z is then

Z = (−1 + 2j)e + (3 + 4j)e† .

Thus in this situation β1 = −1 − 2i = c1 + id1 , β2 = 3 + 4i = c2 − id2 , α1 =

−1 + 2j = c1 − jd1 , α2 = 3 + 4j = c2 + jd2 and as we know it should be that

β1 − α1 = d1 (i + j).

Since
β1 − α1 = −2 (i + j) = −4 i e† = −4 j e† ,
one obtains d1 = −2, which coincides with the value of d1 in this example. 
18 Chapter 1. The Bicomplex Numbers

Let us see now how the conjugations and moduli manifest themselves in
idempotent representations. Take Z = β1 e + β2 e† = α1 e + α2 e† , with β1 and β2
in C(i), α1 and α2 in C( j). Then it is immediate to see that
Z = β 2 e + β 1 e† = α 2 e + α 1 e † ;
Z † = β2 e + β1 e† = α2† e + α1† e† ;
Z ∗ = β 1 e + β 2 e† = α1† e + α2† e† .
Hence, the squares of all the three moduli become:
 2
Z  = Z · Z
j
   
= β 1 e + β 2 e† · β 2 e + β 1 e†
= β 1 β 2 e + β 1 β 2 e†
   
= α 1 e + α 2 e† · α 2 e + α 1 e †
= α1 α2 e + α1 α2 e† = α1 α2 ∈ C( j);

 2
Z  = Z · Z †
i
   
= β 1 e + β 2 e† · β 2 e + β 1 e†
= β 1 β 2 e + β 1 β 2 e† = β 1 β 2
   
= α1 e + α2 e† · α2† e + α1† e†
 †
= α1 α2† e + α1 α2† e† ∈ C(i);

 2
Z  = Z · Z ∗
k
   
= β 1 e + β 2 e† · β 1 e + β 2 e †
 2  2
= β 1 β 1 e + β 2 β 2 e† =  β 1  e +  β 2  e†
   
= α1 e + α2 e† · α1† e + α2† e†
 2  2
= α1 α1† e + α2 α2† e† = α1  e + α2  e† ∈ D+ .
Observe that in the formulas for |Z|2k the idempotent coefficients are non-
negative real numbers and we will see soon that this is a characteristic property
of non-negative hyperbolic numbers. Observe also that given Z = β1 e + β2 e† =
α1 e + α2 e† with β1 , β2 in C(i) and α1 , α2 in C( j), then
1  1 
|Z| = √ |β1 |2 + |β2 |2 = √ |α1 |2 + |α2 |2 .
2 2
We can characterize now the invertibility of bicomplex numbers in terms of
the idempotent representations.
1.6. Idempotent representations of bicomplex numbers 19

Theorem 1.6.5. Given a bicomplex number Z = 0, Z = β1 e + β2 e† = α1 e + α2 e† ,

with β1 and β2 in C(i), α1 and α2 in C( j), the following are equivalent:
1. Z is invertible;
2. β1 = 0 and β2 = 0;
3. α1 = 0 and α2 = 0.
Whenever this holds the inverse of Z is given by
Z −1 = β1−1 e + β2−1 e† = α1−1 e + α2−1 e† .
Proof. It follows using items (6) and (7) from Theorem 1.5.1 together with the
 2  2
idempotent expressions for Z i and Z j . 
Again, we have a “dual” description of zero-divisors in terms of the idempo-
tent decompositions.
Corollary 1.6.6. Given a bicomplex number Z = 0, Z = β1 e + β2 e† = α1 e + α2 e† ,
with β1 and β2 in C(i), α1 and α2 in C( j), the following are equivalent:
1. Z is a zero-divisor;
2. β1 = 0 and β2 = 0 or β1 = 0 and β2 = 0;
3. α1 = 0 and α2 = 0 or α1 = 0 and α2 = 0.
This means that any zero-divisor can be written in one of the following forms:
Z = β1 e with β1 ∈ C(i) \ {0};
†
Z = β2 e with β2 ∈ C(i) \ {0};
Z = α1 e with α1 ∈ C( j) \ {0};
†
Z = α2 e with α2 ∈ C( j) \ {0}.

One can ask if there are more idempotents in BC, not only e and e† (of course
the trivial idempotents 0 and 1 do not count). Assume that a bicomplex number
Z = β1 e + β2 e† , with β1 and β2 being complex numbers either in C(i) or C( j), is
an idempotent: Z 2 = Z. Then
β12 e + β22 e† = β1 e + β2 e†
and
β12 = β1 and β22 = β2 ,
which gives:
β1 ∈ {0, 1}, β2 ∈ {0, 1}.
Hence, combining all possible choices we have at most four candidates for idem-
potents in BC:
Z1 = 0 · e + 0 · e† = 0,
20 Chapter 1. The Bicomplex Numbers

Z2 = 1 · e + 1 · e† = 1,
Z3 = 1 · e + 0 · e† = e,
Z 4 = 0 · e + 1 · e† = e† .

Thus, one concludes that e and e† are the only non-trivial idempotents in BC.
Remark 1.6.7. The formulas

Z = β 1 e + β 2 e† and Z † = β 2 e + β 1 e† ,

with β1 and β2 in C(i), allow us to express the idempotent components of a bicom-

plex number in terms of the bicomplex number itself. Indeed:

β1 = β1 e + β1 e† = Ze + Z † e† ;
β2 = β2 e† + β2 e = Ze† + Z † e.

Writing now the number Z with coefficients in C( j), Z = γ1 e + γ2 e† , we get

a similar pair of formulas:

γ1 = γ1 e + γ1 e† = Ze + Ze† ;
γ2 = γ2 e† + γ2 e = Ze + Ze† .

1.7 Hyperbolic numbers inside bicomplex numbers

Although the hyperbolic numbers had been found long ago and although we wrote
about them at the beginning of the chapter, we believe that it would be instructive
for the reader to have an intrinsic description of the properties of hyperbolic num-
bers, and only then to show how they can be obtained by appealing to bicomplex
numbers.
For a hyperbolic number z = x + ky, its (hyperbolic) conjugate z is defined
by
z := x − ky .
The reader immediately notices that

z · z = x2 − y 2 ∈ R , (1.25)

which yields the notion of the square of the (intrinsic) modulus of z:

|z|2hyp := x2 − y 2 ,

which is a real number (it could be negative!).

If both x and y are non-zero real numbers, but x2 − y 2 = 0, then the cor-
responding hyperbolic number z = x + ky is a zero-divisor, since its conjugate is
1.7. Hyperbolic numbers inside bicomplex numbers 21

non-zero, but the product is zero: z·z = 0. All zero-divisors in D are characterized
by x2 = y 2 , i.e., x = ±y, thus they are of the form

z = λ(1 ± k)

for any λ ∈ R \ {0}.

The idempotent representation of the hyperbolic number z = x + ky ∈ D is

z = (x + y)e + (x − y)e , (1.26)

1 1
where e = (1 + k), e = (1 − k). We consciously use the same letter e that
2 2
was used for the idempotent representation in BC since, as we will soon show,
the two representations coincide in BC. Direct analogs of Proposition 1.6.1 and
Proposition 1.6.3 can be reformulated in this case.
Whenever there is no danger of confusion, we will denote the coefficients
of the idempotent representation of a hyperbolic number z by s := x + y and
t := x − y, so that we have:

z = se + te . (1.27)

Observe that

|z|2hyp = x2 − y 2 = (x + y)(x − y) = st .

Let us show now how these properties are related with their bicomplex an-
tecedents. We are interested in bicomplex numbers Z = z1 + j z2 with Im(z1 ) =
0 = Re(z2 ), that is, our hyperbolic numbers are of the form z = x1 + ij y2 and the
hyperbolic unit is k = ij. Then the -conjugation operation is consistent with the
bicomplex conjugations † and bar in the following way:
†
z = ((x1 + i0) + j(0 + iy2 )) = ((x1 + i0) + j(0 + iy2 )) = x1 − ky2 .

For this reason, from this point on we will not write the hyperbolic conjugate of
e as e anymore, but we will use the bicomplex notation e† .
For a general bicomplex number, the three moduli have been defined in Sec-
tion 1.4. Let us see what happens if they are evaluated on a generic hyperbolic
number z = x1 + ky2 . Considering it as z = z1 + jz2 := (x1 + i0) + j(0 + iy2 ) ∈ BC,
we have:

|z|2i = z12 + z22 = x21 − y22 = |z|2hyp .

Recalling that the definition of | · |i involves the †-conjugation, the definition of | · |j

involves the bar-conjugation and that on hyperbolic numbers both conjugations
coincide, we see that on hyperbolic numbers both moduli reduce to the intrinsic
modulus of hyperbolic numbers:

|z|2i = |z|2j = |z|2hyp . (1.28)

22 Chapter 1. The Bicomplex Numbers

This is not the case of the third modulus: the hyperbolic-valued modulus of Z = z
is different than the intrinsic modulus of z. Indeed, we have:

|z|2k = Z · Z ∗ = Z · Z = Z 2 = z2 . (1.29)

In (1.28) we have a relation between the squares of the three moduli |z|i , |z|j
and |z|hyp for hyperbolic numbers. The question now is how to define the modulus
|z|hyp itself, which obviously should be defined as the square root of x21 − y22 . Note
that some authors consider the non-negative values of x21 − y22 only.
It is instructive to analyze the situation more rigorously and to understand
if we have other options for choosing an appropriate value of the intrinsic modu-
lus. Although we work here with hyperbolic numbers, at the same time one can
think about bicomplex numbers also as of possible values of the square roots of a
hyperbolic number. So let us consider the solutions in BC of the equation Z 2 = R
for a given real number R. Write Z = β1 e + β2 e† , then the equation Z 2 = R is
equivalent to
β12 e + β22 e† = Re + Re†
which is equivalent to
β12 = R; β22 = R.
If R = 0, then the only solution is Z = 0. If R is positive, then
√ √
β1 = ± R; β2 = ± R,

and we get four solutions:

√ √ √ √
R; − R; k R; −k R.

These are all the solutions in BC, and they are real or hyperbolic numbers.
If R is negative, then one gets:
√ √
β1 = ±i −R, β2 = ±i −R

giving the following solutions:

√ √
i −R ; −i −R ;
√ √
ik −R = −j −R ;
√ √
−ik −R = j −R .

Thus, for R < 0 the equation Z 2 = R has four solutions none of which is a
hyperbolic number; two of them are complex numbers in C(i) and the remaining
two are complex numbers in C( j).
Returning to the intrinsic modulus |z|hyp of a hyperbolic number z we see
that in case x21 − y22 > 0 this modulus can be taken as a positive real number
1.7. Hyperbolic numbers inside bicomplex numbers 23
 
x21 − y22 or even as a hyperbolic number ±k x21 − y22 . But if x21 − y22 < 0, then
there are no solutions in D, the candidates should be taken as complex (in C(i) or
in C( j)) numbers.
2 2
 It is instructive to note that in case x1 − y2 > 0 the positive real number
2 2
x1 − y2 coincides with the equal values of |z|i and |z|j as defined in Section 1.4.
When x21 − y22 < 0, then |z|hyp can be chosen either as |z|i ∈ C(i) or as
|z|j ∈ C( j) (recall that we have agreed to take, in both cases, the value of the
square root which is in the upper half plane); as formula (1.29) shows, it cannot
be chosen as |z|k .

1.7.1 The idempotent representation of hyperbolic numbers

Recall that the “hyperbolic” idempotents e and e in (1.26) and the “bicomplex”
idempotents e and e† are the same bicomplex numbers (which are hyperbolic
numbers!). Here (x + y) and (x − y) correspond to the idempotent “coordinates”
β1 and β2 of a bicomplex number. Indeed, considering z = z1 + jz2 := (x1 + i0) +
j(0 + iy2 ) ∈ BC, its idempotent representation is

z = β1 e + β2 e† = (z1 − iz2 )e + (z1 + iz2 )e†

= (x1 − i(iy2 ))e + (x1 + i(iy2 ))e† = (x1 + y2 )e + (x1 − y2 )e† .

Recall also that we have defined the set D+ of non-negative hyperbolic numbers
as 

D+ = x + ky  x2 − y 2 ≥ 0, x ≥ 0 .
The first of the defining inequalities gives the two systems:
 
x − y ≥ 0, x − y ≤ 0,
or
x + y ≥ 0, x + y ≤ 0,

but the condition x ≥ 0 eliminates the second system; hence, the set D+ can be
described as 

D+ = x + ky x ≥ 0; |y| ≤ x ,
or as 
D+ = {νe + μe†  ν, μ ≥ 0 }.
Thus positive hyperbolic numbers are those hyperbolic numbers whose both idem-
potent components are non-negative, that somehow explains the origin of the
name.
In Fig. 1.7.1 the points (x, y) correspond to the hyperbolic numbers z =
x + ky. One sees that, geometrically, the hyperbolic positive numbers are situated
in the quarter plane denoted by D+ . The quarter plane symmetric to it with respect
to the origin corresponds to the negative hyperbolic numbers. The other points
correspond to those hyperbolic numbers which cannot be called either positive or
negative.
24 Chapter 1. The Bicomplex Numbers

2k
1

D− D+
e

O 1 x
2

− 12 k e†

Figure 1.7.1: The positive and negative hyperbolic numbers.

Analogously, the non-positive hyperbolic numbers form the set



D− = x + ky x ≤ 0; |y| ≤ |x| ,

or equivalently 
D− = {νe + μe†  ν, μ ≤ 0 }.
We will say sometimes that the hyperbolic number z = νe + μe† is semi-positive
if one of the coefficients μ and ν is positive and the other is zero.
We mentioned already that D+ plays an analogous role as non-negative real
numbers, and now we illustrate this by computing the square roots of a hyperbolic
number in D+ . Take z ∈ D+ , then z = μe + νe† with μ, ν ∈ R+ ∪ {0}, and it is
easy to see that all the four hyperbolic numbers
√ √
± μ e ± ν e†
√
√ † to z, but only one of them is a non-negative hyperbolic number: μ e +
square
νe .
We are now in a position to define the meaning of the symbol |Z|k for any
bicomplex number Z = β1 e + β2 e† . Indeed, we have obtained that |Z|2k = |β1 |2 e +
|β2 |2 e† which is a non-negative hyperbolic number, hence the modulus |Z|k can
1.8. The Euclidean norm and the product of bicomplex numbers 25

be taken as
|Z|k := |β1 |e + |β2 |e† ∈ D+ .
We will come back to this in the next chapter considering the notion of BC as a
bicomplex normed module where the norm will be D+ -valued. Meanwhile we can
complement the above reasoning solving the equation

|z|k = w

where z is an unknown hyperbolic number and w is in D+ . Writing z and w in the

idempotent form z = β1 e + β2 e† and w = γ1 e + γ2 e† we infer easily a series of
conclusions:
• If w = 0, then z = 0 is a unique solution.
• If w is a semi-positive hyperbolic number, that is, w is a positive zero-divisor:
γ1 = 0 and γ2 > 0 or γ1 > 0 and γ2 = 0, then the solutions are also zero-
divisors although not necessarily semi-positive:

z = ±γ2 e† or z = ±γ1 e ,

respectively.
• If w is positive but not semi-positive: γ1 > 0 and γ2 > 0, then all four
solutions are
z = ±γ1 e ± γ2 e† .

1.8 The Euclidean norm and the product of bicomplex

numbers
We know already that for any two bicomplex numbers Z, W one has:
√
|Z · W | ≤ 2 |Z| · |W |. (1.30)

Note that this inequality is sharp since taking Z = e, W = e, one has:

1
|e · e| = |e| = √
2
and
√ 1
2 |e| · |e| = √ .
2
But for particular bicomplex numbers we can say more.
Proposition 1.8.1. If U = u1 + j u2 ∈ BC is an arbitrary bicomplex number, but Z
is a complex number in C(i) or C(j), or Z is a hyperbolic number, then
a) if Z ∈ C(i) or C(j), then |Z · U | = |Z| · |U |;
26 Chapter 1. The Bicomplex Numbers

b) if Z = x1 + k y2 ∈ D, where x1 ∈ R and y2 ∈ R, then in general

|Z · U | =
 |Z| · |U | .

More precisely,

|Z · U |2 = |Z|2 · |U |2 + 4 x1 y2 Re(i u1 u2 ).

Proof. We prove first a). Indeed, take Z = z1 ∈ C(i) and U = u1 + ju2 =

(u1 − i u2 ) e + (u1 + i u2 ) e† , then

|Z · U |2 = | z1 (u1 + ju2 ) |2 = | (z1 u1 ) + j(z1 u2 ) |2

= |z1 u1 |2 + |z1 u2 |2 = |z1 |2 · |U |2 = |Z|2 · |U |2 ,

where we used the fact that the Euclidean norm of a complex number (both in
C(i) and in C( j)), seen as a bicomplex number, coincides with its modulus.
Take now Z = x1 + jx2 = (x1 − i x2 ) e + (x1 + i x2 ) e† ∈ C(j), then
 
|Z · U |2 = | (x1 − i x2 ) e + (x1 + i x2 ) e†
 
· (u1 − i u2 ) e + (u1 + i u2 ) e† |
= |(x1 − i x2 )(u1 − i u2 )e + (x1 + i x2 )(u1 + i u2 )e† |2
1 
= |x1 − ix2 |2 · |u1 − iu2 |2 + |x1 + ix2 |2 · |u1 + iu2 |2
2
= |Z|2 · |U |2 .

Finally, take Z = x1 + ijy2 = (x1 + y2 ) e + (x1 − y2 ) e† ∈ D, then

|Z · U |2 = |(x1 + ijy2 ) · (u1 + ju2 )|2

 2
= (x1 + y2 ) · (u1 − iu2 ) e + (x1 − y2 ) · (u1 + iu2 ) e† 
1  
= (x1 + y2 )2 · |u1 − iu2 |2 + (x1 − y2 )2 · |u1 + iu2 |2
2
= |Z|2 · |U |2 + 4 x1 y2 Re(i u1 u2 ) ,

and that is all. 

We have described some classes of factors Z for which the Euclidean norm
of the product is equal to the product of the Euclidean norms. Now let us ask the
question: can we characterize all the pairs (Z, W ) for which the Euclidean norm
is multiplicative?
Proposition 1.8.2. Let Z = β1 e + β2 e† and W = γ1 e + γ2 e† be two bicomplex
numbers, then
|Z · W | = |Z| · |W |
if and only if
|β1 | = |β2 |, or |γ1 | = |γ2 |, or both.
1.8. The Euclidean norm and the product of bicomplex numbers 27

Proof. One has:

Z · W = β 1 γ 1 e + β 2 γ 2 e† ,
1  
|Z · W |2 = |β1 γ1 |2 + |β2 γ2 |2 ,
2
1    
|Z|2 · |W |2 = |β1 |2 + |β2 |2 · |γ1 |2 + |γ2 |2 ,
4
   
|β1 |2 − |β2 |2 · |γ1 |2 − |γ2 |2 = 0,
and that is all. 
Remark 1.8.3. Since |β1 | = |β2 | implies that |Z| = |β1 | = |β2 | we may conclude
that the multiplicative property of the Euclidean norm holds if and only if the
Euclidean norm of any of the factors coincides with the modulus (as a complex
number) of its idempotent component. Of course the contents of Section 1.8.1 does
not contradict this conclusion. Note also that the pair β1 and β2 as well as the
pair γ1 and γ2 can be taken, equivalently, in C(i) or in C( j).
It turns out that the condition |β1 | = |β2 | can be usefully interpreted in terms
of the cartesian components. Take Z as Z = z1 + j z2 = β1 e + β2 e† . Assume first
that |β1 | = |β2 | where β1 = z1 − i z2 , β2 = z1 + i z2 ; then

|z1 − i z2 |2 = |z1 + i z2 |2

which is equivalent to
z1 · z 2 = λ ∈ R.
The following cases arise:
(1) if λ = 0, then Z is in C(i), or Z = j z2 ∈ j · C(i), or both; the last means that
Z = 0;
(2) if λ = 0, then
λ
z1 = · z2 ,
|z2 |2

λ
i.e., Z = z2 + j and Z becomes the product of a C(i)-complex num-
|z2 |2
ber and a C( j)-complex number.
Let us show that the reciprocal is also true. Take a = a1 + i a2 , b = b1 + j b2 ,
where a1 , a2 , b1 , b2 are real numbers, and set Z := a · b = a · b1 + j a · b2 , then

Z = (a b1 − i a b2 ) · e + (a b1 + i a b2 ) · e†
= a · (b1 − i b2 ) · e + a (b1 + i b2 ) · e† =: β1 e + β2 e†

with |β1 | = |a| · |b1 − i b2 | = |a| · |b| = |β2 |.

We summarize this reasoning in the following two statements.
28 Chapter 1. The Bicomplex Numbers

Proposition 1.8.4. A bicomplex number Z is a product of a complex number in

C(i) and of a complex number in C( j) if and only if the idempotent components
of Z have the same moduli as complex numbers.
Corollary 1.8.5. The Euclidean norm of the product of two bicomplex numbers is
equal to the product of their norms if and only if at least one of them is the product
of a complex number in C(i) and of a complex number in C( j).
The inequality √
|Z · W | ≤ 2 |Z| · |W |
says that the relation between the Euclidean norm |Z| of an invertible bicom-
plex number Z and the norm |Z −1 | of its inverse is more complicated than the
“conventional” one. Indeed,
√
1 = |Z · Z −1 | ≤ 2 · |Z| · |Z −1 |

and thus
1 √
≤ 2 · |Z −1 |.
|Z|
We ask now for which class of bicomplex numbers the “conventional” formula
1
|Z −1 | = (1.31)
|Z|

holds? The answer follows from the conditions ensuring the equality |Z · W | =
|Z| · |W |, in which we can take W = Z −1 , thus obtaining that (1.31) holds if and
only if Z is a product of a complex number in C(i) by a complex number in C( j)
or, equivalently, if and only if the Euclidean norm of Z coincides with the modulus
of any of its idempotent components.
Chapter 2

Algebraic Structures of the Set

of Bicomplex Numbers

2.1 The ring of bicomplex numbers

The operations of addition and multiplication of bicomplex numbers imply directly
Proposition 2.1.1. (BC, +, ·) is a commutative ring, i.e.,
1. The addition is associative, commutative, with identity element 0 = 0 +
j 0, and each bicomplex number has an additive inverse. This is to say that
(BC, +) is an Abelian group.
2. The multiplication is associative, commutative, with identity element 1 =
1 + j 0.
3. The multiplication is distributive with respect to the addition, i.e., for any
Z, Z1 , Z2 ∈ BC, we have:

Z (Z1 + Z2 ) = Z Z1 + Z Z2 . (2.1)

This is a very particular ring, with a number of specific features which we

will explain in what follows.
Using the language of ring theory, one says that the invertible elements are
called “units”, and generally speaking the sets of non-units and of zero-divisors
do not necessarily coincide. As we saw before, in the ring of bicomplex numbers
the non-units “almost” coincide with the zero-divisors. More exactly, if we exclude
the zero from non-units, the rest of the elements are zero-divisors. In other words,
if BC−1 denotes the set of invertible elements in BC, then we have a partition of
BC:
BC = BC−1 ∪ S ∪ {0} ,

© Springer International Publishing Switzerland 2015 29

M.E. Luna-Elizarrarás et al., Bicomplex Holomorphic Functions, F rontiers in Mathematics,
DOI 10.1007/978-3-319-24868-4_3
30 Chapter 2. Algebraic Structures of the Set of Bicomplex Numbers

where S is the set of zero-divisors.

As it happens in any non-trivial ring that is not a field, the ring of bicomplex
numbers has many ideals, but we want to single out two of them, they are BCe :=
BC · e and BCe† := BC · e† . These ideals are principal in terms of the ring theory.
The peculiarities of these ideals are:

BCe ∩ BCe† = {0} ,

BCe + BCe† = BC ,
e · BCe† = 0 and e† · BCe = 0 .

2.2 Linear spaces and modules in BC

It is known that if S is a subring of a ring R, then R is a module over the ring
S. In our situation, the sets R, C(i), C(j) and D are subrings of the ring BC, thus
BC can be seen as a module over each one of these subrings, and of course, it is
a module over itself. Now, since R, C(i) and C(j) are fields, BC is a real linear
space, a C(i)-complex linear space and a C( j)-complex linear space.
Recalling formula (1.9), we see that the mapping

BC Z = x1 + i y1 + j x2 + k y2 −→ (x1 , y1 , x2 , y2 ) ∈ R4 (2.2)

is an isomorphism of real spaces, which maps the bicomplex numbers 1, i, j, k into

the canonical basis of R4 . We will widely use this identification.
Equation (1.3) suggests the following isomorphism between BC as a C(i)-
linear space and C2 (i):

BC Z = z1 + j z2 −→ (z1 , z2 ) ∈ C2 (i). (2.3)

In this case the bicomplex numbers 1 and j are mapped into the canonical basis of
C2 (i). Composing this isomorphism and the inverse of the previous one, we have
the following isomorphism between R4 and C2 (i):

R4 (x1 , y1 , x2 , y2 ) −→ (x1 + i y1 , x2 + i y2 ) ∈ C2 (i) . (2.4)

Seeing now BC as a C( j)-linear space and using (1.4), we have the following
isomorphism:
BC Z = ζ1 + iζ2 −→ (ζ1 , ζ2 ) ∈ C2 ( j). (2.5)
This isomorphism sends the bicomplex numbers 1 and i into the canonical basis
in C2 ( j) and it induces the following isomorphism (of real linear spaces) between
R4 and C2 ( j):

R4 (x1 , y1 , x2 , y2 ) −→ (x1 + j x2 , y1 + j y2 ) ∈ C2 ( j) . (2.6)

Obviously the isomorphisms (2.4) and (2.6) are different: this shows once
again that inside BC the “complex sets” C2 (i) and C2 ( j) play distinct roles.
2.2. Linear spaces and modules in BC 31

One more difference that one notes considering BC as a C(i)- or a C( j)-linear

space is, for example, that the set {1, i} is linearly independent when BC is seen
as a C( j)-linear space, but the same set is linearly dependent in the C(i)-linear
space BC.
The reader may note that equation (1.7) suggests another C(i)-linear isomor-
phism between BC and C2 (i):
BC Z = (x1 + i y1 ) + k (y2 − i x2 ) = w1 + k w2 −→ (w1 , w2 ) ∈ C2 (i). (2.7)
The relation between the isomorphisms (2.3) and (2.7) is the following. Since under
the isomorphism (2.3) the bicomplex numbers 1, j are mapped to the canonical
basis of C2 (i), then the bicomplex number k = i j is mapped to (0, i) ∈ C2 (i). Thus
we have made a change of basis from the canonical one to the basis {(1, 0), (0, i)}.
The matrix of this change of basis is
⎡ ⎤
1 0
⎣ ⎦,
0 −i
that is, the pair (z1 , z2 ) ∈ C2 (i) is mapped to the pair (w1 , w2 ) ∈ C2 (i) by the
rule ⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
1 0 z1 z1 w1
⎣ ⎦·⎣ ⎦=⎣ ⎦ = ⎣ ⎦. (2.8)
0 −i z2 −i z2 w2
This equality gives the precise relation between (2.3) and (2.7).
A similar reasoning applies to C2 ( j), that is, inspired by equation (1.8), one
defines the isomorphism
BC Z = (x1 + j x2 ) + k (y2 − j y1 ) = ω1 + k ω2 −→ (ω1 , ω2 ) ∈ C2 ( j). (2.9)
The relation between (2.5) and (2.9) is given by a change of basis from the
canonical one to the basis {(1, 0), (0, j)}. The matrix of this change of basis is
⎡ ⎤
1 0
⎣ ⎦,
0 −j
and thus the pairs (ζ1 , ζ2 ) ∈ C2 ( j) and (ω1 , ω2 ) ∈ C2 ( j) are related as follows:
⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
1 0 ζ1 ζ1 ω1
⎣ ⎦·⎣ ⎦=⎣ ⎦ = ⎣ ⎦.
0 −j ζ2 −j ζ2 ω2
The reader may note that we are “playing” with the different linear structures
in BC. We have pointed out some differences between C(i) and C( j), although
existence of the following isomorphism of fields is evident:
ϕ : C(i) → C( j),
(2.10)
ϕ(x + i y) := x + j y.
32 Chapter 2. Algebraic Structures of the Set of Bicomplex Numbers

This isomorphism arose implicitly in the first chapter when we compared the
two idempotent representations, one with coefficients in C(i) and another with
coefficients in C( j):

Z = β1 e + β2 e† = (c1 + i d1 ) e + (c2 + i d2 ) e†
(2.11)
= α1 e + α2 e† = (c1 − j d1 ) e + (c2 + j d2 ) e† ,

†
which shows that α1 = (ϕ(β1 )) and α2 = ϕ(β2 ).
Following the line of defining isomorphisms between BC and the complex
spaces C2 (i) and C2 ( j), we will see that the idempotent representations suggest
two more complex linear spaces isomorphisms. We need first the following propo-
sition.
Proposition 2.2.1. The zero-divisors

1 + ij 1 − ij
e= and e† =
2 2
are linearly independent in BC when it is seen as a C(i)-linear space or as a
C( j)-linear space.
Proof. Using the isomorphism (2.3), the bicomplex numbers e and e† are mapped
as follows: 
1 i 1
e −→ , = (1, i) ,
2 2 2

1 i 1
e† −→ ,− = (1, −i) ,
2 2 2
and considering the equation

λ1 (1, i) + λ2 (1, −i) = 0,

with λ1 λ2 ∈ C(i), we infer immediately that λ1 = λ2 = 0.

The same for C2 ( j). 

Using now (2.11), define the isomorphisms of complex linear spaces:

BC Z = β1 e + β2 e† −→ (β1 , β2 ) ∈ C2 (i), (2.12)

† 2
BC Z = α1 e + α2 e −→ (α1 , α2 ) ∈ C ( j). (2.13)

Again, the relations between (2.3) and (2.12) as well as between (2.5) and
(2.13) are given through the change of basis from the canonical ones to the basis
  
1 i 1 −i
, , ,
2 2 2 2
2.3. Algebra structures in BC 33

in C2 (i) and to the basis

  
1 j 1 −j
, , ,
2 2 2 2
in C2 ( j). Using the matrices of change of basis we get the relations:
⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
1 −i z1 z1 − i z 2 β1
⎣ ⎦·⎣ ⎦=⎣ ⎦=⎣ ⎦ (2.14)
1 i z2 z1 + i z 2 β2

in the space C2 (i), and

⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
1 −j ζ1 ζ1 − j ζ2 α1
⎣ ⎦·⎣ ⎦=⎣ ⎦=⎣ ⎦
1 j ζ2 ζ1 + j ζ 2 α2

in the space C2 ( j).

Consider now the set D2 = D × D. With the component-wise addition in-
herited from D it is an additive abelian group. Defining also the component-wise
multiplication by the scalars from D, D2 becomes a hyperbolic, or D-, module.
Now, formulas (1.5) and (1.6) suggest the following isomorphisms of D-modules:

BC Z = (x1 + k y2 ) + i (y1 − k x2 ) = z1 + i z2 −→ (z1 , z2 ) ∈ D2 (2.15)

and

BC Z = (x1 + k y2 ) + j (x2 − k y1 ) = w1 + j w2 −→ (w1 , w2 ) ∈ D2 . (2.16)

2.3 Algebra structures in BC

We have endowed the set BC with three structures of linear spaces, namely, real
linear space, a C(i)-linear space and C( j)-linear space. Since the set BC is a ring
all these spaces generate the corresponding algebras, one is a real algebra and the
others are complex algebras.
Let us begin with the real algebra. First of all, the real spaces isomorphism
(2.2) induces the following multiplication in R4 (which is of course the “real form
of the bicomplex multiplication”):
(x1 , y1 , x2 , y2 ) · (s1 , t1 , s2 , t2 )
:= (x1 s1 − y1 t1 − x2 s2 + y2 t2 , x1 t1 + y1 s1 − x2 t2 − y2 s2 , (2.17)
x1 s2 − y1 t2 + x2 s1 − y2 t1 , x1 t2 + y1 s2 + x2 t1 + y2 s1 ).

It follows directly from the properties of the bicomplex multiplication that (2.17)
endows R4 with the structure of a commutative real algebra. Moreover, the iso-
morphism (2.2) extends up to a real algebras isomorphism.
34 Chapter 2. Algebraic Structures of the Set of Bicomplex Numbers

It is well known that there are not too many “reasonably good” multipli-
cations in R4 . The bicomplex multiplication given by (2.17) is one of these few
options and one can compare it with the multiplication generated by quaternions.
The next step is to consider both complex algebras. The isomorphisms (2.3),
(2.7) and (2.12) give us three candidates for introducing a multiplication on the
C(i)-linear space C2 (i):

(z1 , z2 ) · (p1 , p2 ) := (z1 p1 − z2 p2 , z1 p2 + z2 p1 );

(w1 , w2 ) · (θ1 , θ2 ) := (w1 θ1 + w2 θ2 , w1 θ2 + w2 θ1 );
(a1 , a2 ) · (c1 , c2 ) := (a1 c1 , a2 c2 ).

Again, the properties of the bicomplex multiplication guarantee that each of

these three formulas defines a commutative multiplication on C2 (i). Looking at
C2 (i) as just a linear space, with no basis fixed, the three formulas define indeed
2
three different multiplications.   the C 
But endowing (i) subsequently with the bases
1 i 1 i
{(1, 0), (0, 1)}, {(1, 0), (0, i)}, , , ,− , and using (2.8) and (2.14),
2 2 2 2
one can see that the three formulas define the same, unique multiplication on C2 (i)
although expressed in terms of the three bases of C2 (i).
For example, from (2.8), one has that

w 1 = z1 , θ 1 = p1 ,
and
w2 = −i z2 , θ2 = −i p2 ,

thus,

(w1 , w2 ) · (θ1 , θ2 ) = (w1 θ1 + w2 θ2 , w1 θ2 + w2 θ1 )

= (z1 p1 − z2 p2 , −i (z1 p2 + z2 p1 ))
⎡ ⎤ ⎡ ⎤
1 0 z1 p1 − z2 p2
=⎣ ⎦·⎣ ⎦
0 −i z1 p2 + z2 p1
⎡ ⎤
w 1 θ1 + w 2 θ2
=⎣ ⎦.
w 1 θ2 + w 2 θ1

The case of the complex linear space C2 ( j) is treated in exactly the same
way.
Now taking into account that BC is a module over D and over itself, and that
it is also a ring, we conclude that the set BC is a D-algebra and a BC-algebra. As
such, BC is isomorphic to the D-algebra D2 and the isomorphisms are given by
formulas (2.15) and (2.16).
2.4. Matrix representations of bicomplex numbers 35

2.4 Matrix representations of bicomplex numbers

It is well known that the field C of complex numbers is isomorphic to the set of
real 2 × 2 matrices of the form

x −y
.
y x

In other words, the mapping


x −y
φC : z = x + iy ∈ C →
y x

is an isomorphism of fields between C and

  
x −y 
 x, y ∈ R =: AC .
y x 

In particular, the matrices of this form commute under multiplication and any of
them, but zero, has its inverse. Under this isomorphism the imaginary unit i is
represented by the matrix 
0 −1
I := .
1 0
Take z = x + iy, then
  
x −y 1 0 0 −1
φC (z) = =x +y = xI2 + yI .
y x 0 1 1 0

The square of the modulus of a complex number z coincides with det φC (z).
In terms of representation theory the mapping φC is called a representation
of the field C into a subset of the set of 2 × 2 real matrices.
A similar reasoning applies to the ring BC. The mapping

z1 −z2
φC(i) : Z = z1 + jz2 ∈ BC −→ (2.18)
z2 z1

turns out to be an isomorphism (of rings) between BC and the set of matrices
  
z1 −z2 
 z1 , z2 ∈ C(i) .
z2 z1 

Note the following identifications:

 
i 0 0 −1
φC(i) (i) = = iI2 , φC(i) (j) = =: J ,
0 i 1 0
 
1 1 −i 1 1 i
φC(i) (e) = =: E , φC(i) (e† ) = =: E t .
2 i 1 2 −i 1
36 Chapter 2. Algebraic Structures of the Set of Bicomplex Numbers

Thus, for a bicomplex number Z = z1 + jz2 its image φC(i) (Z) is

φC(i) (Z) = z1 I2 + z2 J = β1 E + β2 E t .
As one can expect, there are two analogous mappings which allow us to
identify the bicomplex numbers with matrices having C(j)-complex entries, or
hyperbolic entries:

ζ1 −ζ2
φC(j) : Z = ζ1 + iζ2 ∈ BC −→ ,
ζ2 ζ1

z1 −z2
φD : Z = z1 + iz2 ∈ BC −→ .
z2 z1
It is instructive to repeat the above computations with these new mappings.
Finally, the bicomplex numbers may be identified with real 4 × 4 matrices:
⎛ ⎞
x1 −y1 −x2 y2
⎜ y1 x1 −y2 −x2 ⎟
φR : Z = x1 + iy1 + jx2 + ky2 ∈ BC −→ ⎜⎝ x2 −y2 x1 −y1 ⎠ .
⎟

y2 x 2 y1 x1
Every 4 × 4 matrix determines a linear (more exactly, a real linear) trans-
formation on R4 . Of course, not all of them remain BC-linear when R4 is seen as
BC. Those matrices which represent BC-linear mappings are of the form φR (Z).
Notice also that at this stage, we are considering again the following identi-
fication between BC and R4 :
Z = x1 + iy1 + jx2 + ky2 ←→ (x1 , y1 , x2 , y2 ) .
This “forces” the following two identifications between BC and C2 :
Z = z1 + jz2 ←→ (z1 , z2 ) = (x1 + iy1 , x2 + iy2 ) ∈ C2 (i) ←→ (x1 , y1 , x2 , y2 )
and
Z = ζ1 + iζ2 ←→ (ζ1 , ζ2 ) = (x1 + jx2 , y1 + jy2 ) ∈ C2 (j) ←→ (x1 , y1 , x2 , y2 ) .
Consider now an R-linear mapping T : R4 → R4 which represents also a C(i)-linear
mapping, then the matrix of T is of the form
⎛ ⎞
a −b c −d
⎜ b a d c ⎟
⎜ ⎟
⎝  −m u −v ⎠ ,
m  v u
meanwhile, if T represents a C(j)-linear mapping, then its matrix is of the form
⎛ ⎞
A B −E −F
⎜ C D −G −H ⎟
⎜ ⎟.
⎝ E F A B ⎠
G H C D
2.5. Bilinear forms and inner products 37

It is clear that, as one should expect, the matrix φR (Z) represents both a C(i)-
linear mapping and a C(j)-linear one.

2.5 Bilinear forms and inner products

On the real linear space R the following formula defines a bilinear form which
serves simultaneously as an inner product: if x, y ∈ R, then

BR (x, y) := x · y .

The corresponding (real) quadratic form is

Q := BR (x, x) = x2

and it defines the (square of the) Euclidean metric on R.

The set C can be seen both as a real and as a complex linear space and each
of these structures generates its own analogue of what we described above.
When C is considered as a real linear space, that is, C = R2 , then the bilinear
form is given, for z = x + iy, w = u + iv, by

BC,R (z, w) := xu + yv ,

which is exactly the canonical inner product on R2 . The corresponding quadratic

form is
QC,R (z) := BC,R (z, z) = x2 + y 2
and it defines the (square of the) Euclidean metric on C.
When C is considered as a complex linear space, then it has both a bilinear
and a sesquilinear form:
BC,1 (z, w) := z · w
and
BC,2 (z, w) := z · w ,
the second of them being the canonical complex-valued inner product on C. They
generate the quadratic forms

QC,1 (z) := BC,1 (z, z) := z 2

and
QC,2 (z) := BC,2 (z, z) := |z|2 = x2 + y 2 ,
where again QC,2 (z) = QC,R (z) is the square of the Euclidean metric on C. The
forms BC,1 and QC,1 are employed widely in different areas of mathematics, but
BC,1 is not called usually an inner product and QC,1 does not define any metric in
the classical sense.
38 Chapter 2. Algebraic Structures of the Set of Bicomplex Numbers

Let us extend these ideas onto the bicomplex context. Starting with the real
structure on BC, we see it as R4 = {(x1 , y1 , x2 , y2 ) = Z} and thus we endow it
with the (real) bilinear form

BBC,R (Z, W ) := x1 u1 + y1 v1 + x2 u2 + y2 v2 ,

which is the canonical inner product on R4 . The corresponding quadratic form is

QBC,R (Z) := BBC,R (Z, Z) := x21 + y12 + x22 + y22 ,

and it defines the (square of the) Euclidean metric on BC.

When BC is considered as a C(i)-complex linear space, BC = C2 (i), then it
has both a C(i)-bilinear and a C(i)-sesquilinear forms:
1
BBC;i,1 (Z, W ) := z1 · w1 + z2 · w2 = (ZW † + Z † W )
2
and
1
(ZW ∗ + Z † W ) ,
BBC;i,2 (Z, W ) := z1 · w1 + z2 · w2 =
2
the second of them being the canonical C(i)-valued inner product on C2 (i). They
generate the respective quadratic forms

QBC;i,1 (Z) := BBC;i,1 (Z, Z) := z12 + z22 = Z · Z † = |Z|2i

and
1
QBC;i,2 (Z) := BBC;i,2 (Z, Z) := |z1 |2 + |z2 |2 = (ZZ ∗ + Z † Z)
2
1 1
= (|Z|2k + |Z † |2k ) = (|Z|2k + |Z|2k ) ,
2 2
where, again, QBC;i,2 (Z) = QBC,R (Z) is the square of the canonical Euclidean
metric on C2 . The forms BBC;i,1 and QBC;i,1 are also widely known and used (for
instance, in the theory of complex Laplacian and its solutions called complex
harmonic functions), but BBC;i,1 is not called, usually, an inner product on BC,
and QBC;i,1 does not define any metric in the classical sense.
The other complex structure on BC, BC = C2 (j), is dealt with in the same
way. In this context, it is worth writing the corresponding quadratic forms:

QBC;j,1 (Z) := ζ12 + ζ22 = Z · Z = |Z|2j

and
1
QBC;j,2 (Z) := |ζ1 |2 + |ζ2 |2 = (ZZ ∗ + ZZ † )
2
1 1
= (|Z|2k + |Z † |2k ) = (|Z|2k + |Z|2k ) ,
2 2
2.5. Bilinear forms and inner products 39

where we notice that the last one coincides with QBC;i,2 (Z), and both are equal
to the square of the Euclidean metric.
When BC is interpreted as D2 , the situation is different. Of course, one sets:
1
BBC;D,1 (Z, W ) := z1 w1 + z2 w2 = (ZW ∗ + Z ∗ W )
2
and
1 1
BBC;D,2 (Z, W ) := z1 w1 + z2 w2 = (ZW † + Z ∗ W ) = (ZW + Z ∗ W † ) ,
2 2
imitating the previous situations, but now both forms take values in D, not in C
or R. Note also that the first of them is hyperbolic bilinear, and the second one
can be called hyperbolic sesquilinear; what is more, setting

QBC;D,1 (z) := BBC;D,1 (Z, Z) = z21 + z22 = Z · Z ∗ = |Z|2k

and

QBC;D,2 (z) := BBC;D,2 (Z, Z) = z1 z1 + z2 z2

1 1
= (ZZ † + Z ∗ Z) = (ZZ + Z ∗ Z † )
2 2
1 1
= (|Z|2i + |Z ∗ |2i ) = (|Z|2j + |Z ∗ |2j ) ,
2 2
one sees that QBC;D,1 has hyperbolic values and that QBC;D,2 takes real values,
but it is not positive definite. Thus, the geometry behind them is much more
sophisticated. In the next chapter we will elaborate on this.
Finally, BC is a bicomplex module, i.e., a module over itself, which suggests
the introduction of a bicomplex bilinear form

BBC (Z, W ) := Z · W

and of three bicomplex sesquilinear-type forms:

BBC,bar (Z, W ) := Z · W , (a bar-sesquilinear form) ,

†
BBC,† (Z, W ) := Z · W , (a †-sesquilinear form) ,
∗
BBC,∗ (Z, W ) := Z · W , (a ∗-sesquilinear form) .

The corresponding quadratic forms coincide with the three “moduli” previously
introduced, which take complex or hyperbolic values, making the geometric aspect
even more complicated than the above described case of the D-module BC = D2 .
Of course, this makes both cases even more interesting and intriguing.
Let us consider again the R-valued quadratic form QC,2 (z) = x2 + y 2 ; since
it coincides with BC,2 (z, z), then QC,2 enjoys the factorization

QC,2 = (x + iy)(x − iy) = z · z .

40 Chapter 2. Algebraic Structures of the Set of Bicomplex Numbers

This identity can be seen as one of the reasons for the necessity of introducing
complex numbers: if one wants to factorize QC,2 (z) (which is a real-valued and
positive definite quadratic form; thus, in particular, the set of its values is R-
one-dimensional) into the product of two linear forms which should be real two-
dimensional, then the imaginary unit i emerges forcedly and generates the whole
set C.
A very similar idea is related with the bicomplex numbers. Consider the
C(i)-valued quadratic form QBC,i,1 (Z) = z12 + z22 . We know that it factorizes into

QBC,i,1 = (z1 + jz2 )(z1 − jz2 ) = Z · Z † ,

where the set of the values of QBC,i,1 is C(i)-one-dimensional but the factors are
already C(i)-two-dimensional. Thus, the C(i)-algebra BC arises from a complex
quadratic form in the same way as the real algebra C arises from a real quadratic
form. Notice that the requirement for the factors to be C(i)-two-dimensional, not
one-dimensional, is crucial since without it one has an obvious factorization

z12 + z22 = (z1 + iz2 )(z1 − iz2 ) (2.19)

which does not serve our purposes.

Let us show that no other number system, but BC, can play the same role.
Assume that there exists a C(i)-two-dimensional commutative algebra such that
the four elements of it, say, a, b, c, d, ensure the identity

z12 + z22 = (az1 + bz2 )(cz1 + dz2 )

for all z1 and z2 in C(i). Hence, for all z1 and z2 it holds that

z12 + z22 = acz12 + bdz22 + (ad + bc)z1 z2 ,

which is equivalent to

ac = 1; bd = 1; ad + bc = 0.

Thus, all the coefficients are invertible elements and

c−1 d + d−1 c = 0,

i.e.,
 2
c−1 d = −1.
Therefore, denoting j := c−1 d we have that j2 = −1 and j−1 = −j; the factoriza-
tion becomes
z12 + z22 = (z1 + jz2 )(z1 − jz2 ).
Thus, the complex algebra we are looking for should be generated by 1 and by a
new element j = ±i, and we have arrived exactly at BC.
2.6. A partial order on the set of hyperbolic numbers 41

In the same way we can begin with the C( j)-valued quadratic form ζ12 + ζ22
and get the same BC which now will be seen as a C( j)-algebra.
Finally, if we begin with the D-valued quadratic form

z21 + z22

acting on the D-algebra D2 , then any of the two imaginary units, i or j, will arise
giving the factorizations into two factors each of which is D-two-dimensional.

2.6 A partial order on the set of hyperbolic numbers

2.6.1 Definition of the partial order
We have noticed already a deep similarity between the role of non-negative hyper-
bolic numbers inside D and the role of non-negative real numbers inside R. It turns
out that this similarity can be extended and the (partial) notions “greater than”
and “less than” can be introduced on hyperbolic numbers. Take two hyperbolic
numbers z1 and z2 ; if their difference z2 − z1 ∈ D+ , that is, the difference is a
non-negative hyperbolic number, then we write z2  z1 or z1  z2 and we say that
z2 is D-greater than or equal to z1 , or that z1 is D-less than or equal to z2 .
Writing these hyperbolic numbers in their idempotent form z1 = β1 e + β2 e†
and z2 = γ1 e + γ2 e† , with real numbers β1 , β2 , γ1 and γ2 , we have that

z1  z2 if and only if γ1 ≥ β1 and γ2 ≥ β2 .

On Figure 2.6.1, z0 = x0 + ky0 is an arbitrary hyperbolic number, and one

can see that the entire plane is divided into four quarters: the quarter plane of
hyperbolic numbers which are D-greater than or equal to z0 (z  z0 ); the quarter
plane of hyperbolic numbers which are D-less than or equal to z0 (z  z0 ); and the
two quarter planes where the hyperbolic numbers are not D-comparable with z0
(neither z  z0 nor z  z0 holds).
Thus we have introduced a binary relation on D which is, obviously, reflexive:
for any z ∈ D, z  z; transitive: if z1  z2 and z2  z3 , then z1  z3 ; and
antisymmetric: if z1  z2 and z2  z1 , then z1 = z2 . Since this relation is
applicable not for any pair of elements in D, then the relation  defines a partial
order on D. As can be expected, this partial order extends the total order ≤
on R: if one takes two real numbers x1 and x2 , x1 ≤ x2 , and considers them as
hyperbolic numbers with zero imaginary parts, then x1  x2 .
In case z2 − z1 ∈ D+ \ {0} we write z2  z1 and we say that z2 is D-greater
than z1 , or we write z1 ≺ z2 and say that z1 is D-less than z2 . This implies that
z ∈ D+ is equivalent to z  0 and that z ∈ D+ \ {0} is equivalent to z  0; z ∈ D−
is equivalent to z  0 and z ∈ D− \ {0} is equivalent to z ≺ 0.
As a matter of fact, we can trace an analogy with the future and past cones in
a two-dimensional space with the Minkowski metric. Specifically, let us identify a
42 Chapter 2. Algebraic Structures of the Set of Bicomplex Numbers

k
z  z0
e z0  z
z0

O 1 1 x
2

e†

−k

Figure 2.6.1: A partial order on D

one-dimensional time with the x-axis and a one-dimensional space with the y-axis,
both axes being embedded into D and assume the speed of light to be equal to one;
then the set of zero-divisors with positive real part, x > 0 is nothing more than
the future of a ray of light which was sent from the origin towards either direction.
Analogously, the set of zero-divisors with negative real part, x < 0, represent the
past of the same ray of light.
Thus, positive hyperbolic numbers represent the future cone, i. e., they cor-
respond to the events which are in the future of the origin; the negative hyperbolic
numbers, that is, those which after multiplying by (−1) become positive, represent
the past cone, i.e., they correspond to the events in the past of the origin.
Accepting this interpretation, we see how the set of hyperbolic numbers which
are greater than a given hyperbolic number Z represents the events in the future
of the event which is represented by Z.

2.6.2 Properties of the partial order

Let us describe here some consequences of the the definition of the partial order
. We combine them into three groups. Let z, w and y be hyperbolic numbers.
2.6. A partial order on the set of hyperbolic numbers 43

(i) Consequences of the definition itself

• If z and y are comparable with respect to , then precisely one of the
following relations holds:

z ≺ y or z  y or z = y.

• The inequalities z ≺ y and y  w imply that z ≺ w.

• The inequalities z  y and y ≺ w imply that z ≺ w.
(ii) Connections between the addition and the order on D
• z ≺ w implies that z + y ≺ w + y.
• z  w implies that z + y  w + y.
• 0 ≺ z implies that −z ≺ 0.
• z1  z2 and y  w imply that z1 + y  z2 + w.
• z1  z2 and y ≺ w imply that z1 + y ≺ z2 + w.
(iii) Connections between multiplication and partial order on D
• If z and y are non-negative hyperbolic numbers, then so is their product:

z · y ∈ D+ .

• If z and y are strictly positive hyperbolic numbers, then so is their

product:
z · y ∈ D+ \ {0}.
• If z and y are strictly negative hyperbolic numbers, then their product
is strictly positive:
z · y  0.
• If one of z and y is strictly positive and another is strictly negative, then
their product is strictly negative:

z · y ≺ 0.

• If z ≺ y and w  0, then z · w ≺ y · w.
• If z ≺ y and w ≺ 0, then z · w  y · w.
• If z is a (strictly) positive hyperbolic number, then it is invertible and its
inverse is also positive: if z  0 and z ≺ y, then y−1  0 and y−1 ≺ z−1 .
Example 2.6.2.1. Let us illustrate the above properties solving for z = β1 e + β2 e†
in D the inequality
|z|k  w , (2.20)
44 Chapter 2. Algebraic Structures of the Set of Bicomplex Numbers

where w = γ1 e + γ2 e† is in D+ and | · |k is the D-valued modulus. If w = 0 the

unique solution is z = 0. Hence, consider w ∈ D+ \ {0}. The inequality (2.20) is
equivalent to
|β1 |e + |β2 |e†  γ1 e + γ2 e†
which in turn is equivalent to the system

|β1 | ≤ γ1 ,
|β2 | ≤ γ2 .

Thus the solutions of (2.20) are hyperbolic numbers z = β1 e + β2 e† with −γ1 ≤

β1 ≤ γ1 and −γ2 ≤ β2 ≤ γ2 . This means that the inequality (2.20) is equivalent
to the double hyperbolic inequality −w  z  w. 
We introduce now the notion of a hyperbolic interval (or hyperbolic segment).
Given two hyperbolic numbers a and b, a  b, we set


[a, b]D := z ∈ D  a  z  b .

Consider now two particular cases:

• Let a = k and b = 1. Since obviously k = e − e†  1 = e + e† , then the
interval [k, 1]D is well defined. The inequality

k  z = β1 e + β2 e†  1

gives:
1 ≤ β1 ≤ 1, −1 ≤ β2 ≤ 1.
It turns out that in this case the hyperbolic interval is a one-dimensional set.
See the Figure 2.6.2.
• Take now a = k and b = 2 (obviously k ≺ 2). In this case the hyperbolic
interval is given by


[k, 2]D = z = β1 e + β2 e†  1 ≤ β1 ≤ 2 and − 1 ≤ β2 ≤ 2 ,

and it is now a two-dimensional set. See Figure 2.6.3.

2.6.3 D-bounded subsets in D.

Given a subset A in D, we define as usual the notion of D-upper and D-lower
bounds, as well as the notions of a set being D-bounded from above, from below,
and finally of a D-bounded set. There are some fine points here. If A has a D-
upper or a D-lower bound α, then this means that for any a ∈ A there holds that
a is comparable with α and a  α or α  a. But this does not mean that the
elements of A are necessarily comparable between them; the same happens taking
two D-upper of D-lower bounds α and β they are not always comparable.
2.6. A partial order on the set of hyperbolic numbers 45

O 1 1 x
2

e†

Figure 2.6.2: The hyperbolic segment [k, 1]D .

If A ⊂ D is a set D-bounded from above, we define the notion of its D-

supremum, denoted by supD A, to be the least upper bound for A, and its D-
infimum inf D A to be the greatest lower bound for A. The “least” upper bound
here means that supD A  α for any D-upper bound α even if not all of the D-upper
bounds are comparable. Similarly the meaning of the “greatest” lower bound is
understood. Of course, every non-empty set of hyperbolic numbers which is D-
bounded from above has its D-supremum, and if it is D-bounded from below, then
it has its D-infimum. This can be seen immediately if one notes that there are
more convenient expressions for these notions. Given a set A ⊂ D, consider the
sets A1 := { a1 | a1 e + a2 e† ∈ A } and A2 := { a2 | a1 e + a2 e† ∈ A }. If A is
D-bounded from above, then the supD A can be computed by the formula

supD A = sup A1 · e + sup A2 · e† .

If A is D-bounded from below, then the inf D A can be computed by the formula

inf D A = inf A1 · e + inf A2 · e† .

The above formulas explain a very peculiar character of the partial order on D.
Note only that although two D-upper (or D-lower) bounds can be incomparable
nevertheless they are always comparable with supD A (or with inf D A).
46 Chapter 2. Algebraic Structures of the Set of Bicomplex Numbers

k 2e
β1 e

−e† e

O 1 1 x
z = β1 e + β2 e†
2
e†

β2 e†

2e†

Figure 2.6.3: The hyperbolic interval [k, 2]D .

Let A and B be two subsets of D, denote by −A the set of all the elements
in A multiplied by −1, and denote by A + B the set of all sums z + y with z ∈ A
and y ∈ B; define in the same fashion the set A · B. The reader is invited to prove
the following properties.
• A set A is D-bounded from above (or from below) if and only if the set −A
is D-bounded from below (or from above); for such sets it holds that

inf D (−A) = −supD A; supD (−A) = −inf D A.

• If A and B are D-bounded from below, then so is A + B and for such sets
one has:
inf D (A) + inf D (B) = inf D (A + B) .
If A and B are D-bounded from above, then so is A + B and for such sets
one has:
supD (A) + supD (B) = supD (A + B) .

• Assume that A and B are subsets of the set D+ of non-negative hyperbolic

numbers. If A and B are D-bounded from below, then so is A · B and for such
2.7. The hyperbolic norm on BC 47

sets one has:

inf D (A · B) = inf D (A) · inf D (B) .
If A and B are D-bounded from above, then so is A · B and for such sets one
has:
supD (A · B) = supD (A) · supD (B) .

2.7 The hyperbolic norm on BC

We know already, for any bicomplex Z = β1 e + β2 e† , the formula

|Z|k := |β1 |e + |β2 |e† .

That is, we have the map

| · |k : BC −→ D+
with the properties:
(i) |Z|k = 0 if and only if Z = 0;
(ii) |Z · W |k = |Z|k · |W |k for any Z, W ∈ BC;
(iii) |Z + W |k  |Z|k + |W |k .
The first two properties are clear. Let us prove (iii).

|Z + W |k = |(β1 + ν1 ) · e + (β2 + ν2 ) · e† |k
= |β1 + ν1 | · e + |β2 + ν2 | · e†
 (|β1 | + |ν1 |) · e + (|β2 | + |ν2 |) · e†
= |Z|k + |W |k .

The three properties (i)–(iii) manifest again the analogy between real positive
numbers and hyperbolic positive numbers; now we see that the hyperbolic modulus
of a bicomplex number has exactly the same properties as the real modulus of a
complex number whenever the partial order  is used instead of ≤.
Because of properties (i)–(iii) we will say that | · |k is the hyperbolic-valued
(D-valued) norm on the BC-module BC.
It is instructive to compare (ii) with (1.13) where the norm of the product
and the product of the norms are related with an inequality. We believe that one
could say that the hyperbolic norm of bicomplex numbers is better suited to the
algebraic structure of the latter although, of course, one has to allow hyperbolic
values for the norm.
Remark 2.7.1. (1) Since for any Z ∈ BC it holds that
√
|Z|k  2 · |Z| , (2.21)
48 Chapter 2. Algebraic Structures of the Set of Bicomplex Numbers

with |Z| the Euclidean norm of Z, then one has:

√
|Z · W |k  2 · |Z| · |W |k .

In contrast with property (ii) above, this inequality involves both the Euclidean
and the hyperbolic norms.
(2) Take z1 and z2 in D+ , then clearly

z1  z2 implies that |z1 | ≤ |z2 |. (2.22)

(3) Note that the definition of hyperbolic norm for a bicomplex number Z does
not depend on the choice of its idempotent representation. We have used, for
Z ∈ BC, the idempotent representation Z = β1 e + β2 e† , with β1 and β2 in
C(i). If we had started with the idempotent representation Z = γ1 e + γ2 e† ,
with γ1 and γ2 in C( j), then we would have arrived at the same definition of
the hyperbolic norm since |β1 | = |γ1 | and |β2 | = |γ2 |.
(4) The comparison of the Euclidean norm |Z| and the D-valued norm |Z|k of a
bicomplex number Z gives:
1 
||Z|k | = √ |β1 |2 + |β2 |2 = |Z| (2.23)
2
where the left-hand side is the Euclidean norm of a hyperbolic number.

2.7.1 Multiplicative groups of hyperbolic and bicomplex numbers

The set R+ of strictly positive real numbers is a multiplicative group. For the set
D+ of non-negative hyperbolic numbers the situation is more delicate although it
preserves some analogies.
Introduce the set D+ +
inv := D \ S0 of all strictly positive hyperbolic numbers.
This set has the following properties:
• if λ1 and λ2 are in D+ +
inv , then λ1 · λ2 ∈ Dinv ;

• 1 ∈ D+
inv ;

• if λ ∈ D+
inv , then λ is invertible in D and λ
−1
∈ D+
inv .

Hence, D+inv is a multiplicative group with respect to the hyperbolic multi-

plication and thus it is an exact analogue of R+ ; what is more, R+ is a subgroup
of D+inv .
The zero-divisors in D+ are either of the form z = λe or of the form z = μe†
where λ and μ are positive † real numbers;

so we will use the notations D+ e :=
+
{λe | λ > 0} and De† := μe | μ > 0 . Both sets of semi-positive hyperbolic num-
bers are closed under hyperbolic multiplication but neither of them contains the
number one.
2.7. The hyperbolic norm on BC 49

+
It turns out that, anyway, both D+e and De† can be endowed with the struc-
ture of a multiplicative group. Beginning with D+ e one observes that for any λ > 0
1
one has that λe · e = λe, and e · λe = 1 · e = e. Hence, if we endow D+ e with the
λ
multiplication which is the restriction of the hyperbolic multiplication, then on
(D+
e , ) we have that

1. if λ1 e and λ2 e are in D+ +
e , then their product λ1 e λ2 e = λ1 λ2 e is in (De , );

2. the element 1 · e = e serves as the unit 1 for the multiplication :

1 λe = e · λe = λe;

1
3. if λe ∈ (D+
e , ), then e is its -inverse:
λ
1
λe e = 1 · e = e = 1 .
λ

Thus, we conclude that (D+ e , ) is a multiplicative group. The same reasoning

applies to the set D+
e†
with obvious changes. Each of these groups is isomorphic to
the group R+ .
Obviously, the sets Dinv := D \ S0 and BCinv := BC \ S0 are multiplicative
groups with their respective multiplications, thus

D+
inv ⊂ Dinv ⊂ BCinv ,

where ⊂ means the embedding of group structures. At the same time we can use the
same arguments on De := e·D\{0} = {λe | λ ∈ R\{0} }, De† := e† ·D\{0} = {μe† |
μ ∈ R \ {0} }, BCe \ {0} := {λe | λ ∈ C \ {0} }, BCe† \ {0} := {μe† | μ ∈ C \ {0} }.
The first two of them are isomorphic to the multiplicative group of real numbers,
the last two are isomorphic to the multiplicative group of complex numbers.
We leave the details to the reader.
In addition to the references from the end of Chapter 1, such as [45, 56,
57, 58, 65, 85], where some of the algebraic structures that we developed in this
chapter have been introduced and studied, the most important contribution in
this area is the book of Alpay, Luna–Elizarrarás, Shapiro and Struppa [2]. This
study introduces for the first time the notion of the hyperbolic-valued norm on
the ring of bicomplex numbers, which we fully described in this chapter.
Chapter 3

Geometry and Trigonometric

Representations of Bicomplex
Numbers

The geometry of complex numbers coincides with the geometry of the Euclidean
space R2 , and this is because of a good compatibility between the algebraic struc-
ture of C and the geometry of R2 , which is expressed by the equality

z · z = x2 + y 2 = |z|2 . (3.1)

Moreover, the algebraic operations on complex numbers can be easily interpreted

in terms of the geometry of the plane R2 ; for instance, to multiply by a complex
number is equivalent to realizing a composition of a rotation and a homothety.
One would want to extend this idea for the Euclidean space C2 . But it turns
out that the algebraic properties of quaternions, not those of bicomplex numbers,
are compatible with the Euclidean structure of C2 in the same way as the complex
numbers are compatible with the Euclidean structure of R2 = C.
As for the bicomplex numbers, the situation is much more sophisticated since
now we have several quadratic forms to deal with. Indeed, we know already that
the analogues of (3.1) are:

Z · Z † = z12 + z22 = |Z|2i ∈ C(i), z1 , z2 ∈ C(i); (3.2)

Z · Z = ζ12 + ζ22 = |Z|2j ∈ C( j), ζ1 , ζ2 ∈ C( j); (3.3)

Z · Z ∗ = z21 + z22 = |Z|2k ∈ D, z1 , z2 ∈ D. (3.4)
2 4
In addition, the usual Euclidean structure of BC = C = R yields the correspond-
ing quadratic form
x21 + y12 + x22 + y22 .

© Springer International Publishing Switzerland 2015 51

M.E. Luna-Elizarrarás et al., Bicomplex Holomorphic Functions, F rontiers in Mathematics,
DOI 10.1007/978-3-319-24868-4_4
52 Chapter 3. Geometry and Trigonometric Representations

This means that the “authentic” geometry of bicomplex numbers is related

to two complex quadratic forms (one is C(i)-valued, the other is C( j)-valued), one
D-valued quadratic form and one real-valued quadratic form.

3.1 Drawing and thinking in R4

In this chapter we are going to analyze the geometry of the bicomplex numbers.
But we begin first with the description of some usual Euclidean geometrical objects
in R4 and we will describe them using the bicomplex numbers; this is because BC
coincides with R4 when it is seen as an R-linear space and thus we will need to
imagine geometrical objects in R4 .
In order to realize the peculiarities of the four-dimensional cube let us con-
sider first the cubes in lower dimensions. In R1 the cube with a vertex at the origin
and of length one is the segment [0, 1] (see Figure 3.1.1).

0 1 x

Figure 3.1.1: The one-dimensional cube.

Using this segment, one constructs the two-dimensional cube pasting per-
pendicularly the cube of the dimension one at each point of [0, 1], i.e., pasting the
translations of [0, 1] and thus obtaining the square with unitary sides. See Figure
3.1.2.

···

0 1 0 1

Figure 3.1.2: The two-dimensional cube.

Extending this idea, we construct next the three-dimensional cube of side

one by attaching at each point of [0, 1] perpendicularly the cube of the previous
dimension, that is, the squares of side one are attached. See Figure 3.1.3.
To construct the cube in four dimensions we repeat the procedure: the three-
dimensional cube of side one is attached now, perpendicularly, to each point of
[0, 1]. See figure 3.1.4.
In a sense, such a way of constructing contradicts our 3-dimensional intuition
but since we have now an additional dimension the arising 3-dimensional cubes
are parallel and they do not intersect at all.
3.1. Drawing and thinking in R4 53

0 0

1 1

Figure 3.1.3: The three-dimensional cube.

Figure 3.1.4: The four-dimensional cube.

The same idea allows us to visualize the whole 4-dimensional space with the
coordinate axes x0 , x1 , x2 , x3 : to each point of the axis x0 , we attach, perpendic-
ularly a copy of R3 . See Figure 3.1.5.
There is an alternative way of “seeing” the 4-dimensional world. Take, for
instance, the real 2-dimensional plane spanned by 1 and i; to each point of it we
attach perpendicularly a copy of the 2-dimensional plane spanned by j and k.
Note that whenever we mention the perpendicularity we mean, implicitly,
that the two sets pass through the origin and they are the orthogonal comple-
ments of each other; if they do not pass through the origin, then they become the
54 Chapter 3. Geometry and Trigonometric Representations

Figure 3.1.5: R4 as an infinite union of copies of R3 .

orthogonal complements of each other if they are translated to the origin. This
is reasonably clear on the level of intuition for the 3-dimensional situation, and
for the 4-dimensional case one can appeal to analogies in three dimensions; for
instance, an analogue of the two planes which are perpendicular in R4 may be two
straight lines which are perpendicular in R2 .
Traditionally, the 4-dimensional cube is presented as a solid in R4 whose 3-
dimensional surface consists of eight 3-dimensional cubes which are glued together
in a very particular way. We will see now that our approach does not contradict
this vision. The matter is that if instead of pasting a 3-dimensional cube at each
point of the interval [0, 1] on the axis x0 , the process can be repeated pasting cubes
at each point of the interval [0, 1] but now on the axis x1 , or on the axis x2 , or on
the axis x3 (we will denote for short the corresponding intervals as [0, i], [0, j] and
[0, k]), obtaining the same geometrical figure. With this in mind, let us introduce
the notations:
• Cr is the cube attached to the point r ∈ [0, 1].
• Is is the cube attached to the point s ∈ [0, i].
• Ju is the cube attached to the point u ∈ [0, j].
3.1. Drawing and thinking in R4 55

• Kv is the cube attached to the point v ∈ [0, k].

Proposition 3.1.1. The (topological) boundary of the 4-dimensional cube is the

union of the 3-dimensional cubes C0 , C1 , I0 , Ii , J0 , Jj , K0 and Kk .

Proof. Follows by noting that the cubes C0 , etc., are the “extreme cubes” when
the described process applies to each of the following segments: [0, 1], [0, i], [0, j],
[0, k]. 

In order to see more precisely how the vertices of the cubes are glued together,
we will write down the vertices explicitly.

j+k
C0 i+j+k
1+j+k k

i+k
1+k j
0 i+j
i 1+i+j+k
1+j
1+i+j
1
1+i+k
C1 1+i

Figure 3.1.6: The cubes C0 and C1 and their vertices.

The vertices of C0 are: 0, i, j, k, i + j, i + k, j + k, i + j + k. The vertices of

C1 are: 1, 1 + i, 1 + j, 1 + k, 1 + i + j, 1 + i + k, 1 + j + k, 1 + i + j + k.
The vertices of I0 are: 0, j, 1 + j, 1, k, j + k, 1 + j + k, 1 + k. The vertices of
Ii are: i, i + j, 1 + i + j, i + 1, i + k, i + j + k, 1 + i + j + k, 1 + i + k.
As one could expect, C0 and C1 are disjoint sets as well as I0 and Ii . Never-
theless, C0 and I0 have the following common vertices: 0, j, k, j + k. This means
that they are glued together along the square with these vertices. Similarly C0
and Ii are glued together along the square with vertices i, i + j, i + k, i + j + k.
The common square of the cubes C1 and I0 has the vertices 1, 1 + j, 1 +
k, 1 + j + k; the common square of the cubes C1 and Ii has the vertices 1 + i, 1 +
i + j, 1 + i + j, 1 + i + j + k.
The vertices of J0 are: 0, 1, i, 1 + i, k, 1 + k, i + k and 1 + i + k. The vertices
of Jj are: j, 1 + j, i + j, 1 + i + j, j + k, 1 + j + k, i + j + k and 1 + i + j + k.
The vertices of K0 are: 0, 1, i, j, 1 + i, i + j, 1 + i + j, 1 + j; and the vertices
of Kk are: k, 1 + k, i + k, j + k, 1 + i + k, i + j + k, 1 + i + j + k, 1 + j + k.
56 Chapter 3. Geometry and Trigonometric Representations

j+k

i+j+k
1+j+k
1+i+j+k
I0 k
j Ii
i+k
1+k i+j
1+j

0
1+i+j
i
1

i+1

1+i+k

Figure 3.1.7: The cubes I0 and Ii and their vertices.

i+k

j+k
Jj j+i+k
k

1+j+k 1+i+j+k

1+k
j
0 j+i
J0
i
j+1
1
1+i+j

1+i
1+i+k

Figure 3.1.8: The cubes J0 and Jj and their vertices.

3.2. Trigonometric representation in complex terms 57

k+j+1
k+j

k Kk
k+i k+i+j

j
1+i+j+k
0
k+1
i+j

1+j i
1+k+i K0
1
1+i+j

1+i

Figure 3.1.9: The cubes K0 and Kk and their vertices.

Table 3.1 shows:

• the vertices of the 4-dimensional cube (there are sixteen of them).
• The vertices of each of its 3-dimensional sides.
• At which vertices the 3-dimensional sides intersect.
Observe that the first column gives all the vertices and the other columns
show which cube corresponds to each vertex; besides, each row shows the cubes
with a fixed vertex.

3.2 Trigonometric representation in complex terms

Any complex number z = 0 has a trigonometric form which involves the modulus
(a real non-negative number) and the argument (also a real number). In this section
we will show that bicomplex numbers which are neither zero nor zero-divisors have
a trigonometric representation, where the modulus and the argument are complex
numbers. We will see in the next sections that there exists also a trigonometric
representation where the modulus and the argument are hyperbolic numbers: note
that such a representation is available as well for zero-divisors.
Thus, take Z ∈ BC \ S0 , that is, |Z|i = 0. Then we write:

z1 z2
Z = z1 + jz2 = |Z|i +j . (3.5)
|Z|i |Z|i
58 Chapter 3. Geometry and Trigonometric Representations

Table 3.1: The vertices of the four-dimensional cube.

0 C0 I0 J0 K0

1 C1 I0 J0 K0

i C0 Ii J0 K0

j C0 I0 Jj K0

k C0 I0 J0 Kk

1+i C1 Ii J0 K0

1+j C1 I0 Jj K0

1+k C1 I0 J0 Kk

i+j C0 Ii Jj K0

i+k C0 Ii J0 Kk

j+k C0 I0 Jj Kk

1+i+j C1 Ii Jj K0

1+i+k C1 Ii J0 Kk

1+j+k C1 I0 Jj Kk

1+i+j+k C1 Ii Jj Kk

i+j+k C0 Ii Jj Kk

Since  2  2
z1 z2
+ = 1,
|Z|i |Z|i
the system
z1 z2
cos Θ = , sin Θ = , (3.6)
|Z|i |Z|i
has at least a solution Θ ∈ C(i). Actually there are infinitely many solutions, due
3.2. Trigonometric representation in complex terms 59

to the periodicity of the complex cosine and sine functions.

Notice that if Θ0 is a solution of the system (3.6), then the equality tan Θ0 =
z2 z2
implies that = ±i (it is known that the function tan z takes any value in C
z1 z1
except ±i), but this is equivalent to saying that Z is not a zero-divisor.
In analogy with the complex case, we call any solution of the system (3.6) a
complex argument of the bicomplex number Z, and we denote by ArgC(i) (Z) the
set of all these solutions. If we denote by Θ0 = θ01 + iθ02 ∈ C(i) the particular
solution obtained by choosing its real part θ01 in the interval [0, 2π), which is
called the principal value of the complex argument, then

ArgC(i) (Z) = {Θ0 + 2mπ  m ∈ Z}.

For the principal value of the complex argument, we will use the notation argC(i) (Z).
For a fixed m ∈ Z, we denote by argC(i),m (Z) := argC(i) (Z) + 2mπ, thus
  
ArgC(i) (Z) = argC(i),m (Z)  m ∈ Z .

For an invertible bicomplex number Z, we obtain its BCi -trigonometric form:

Z = |Z|i (cos Θ + j sin Θ) , (3.7)

where Θ is an arbitrary value of the complex argument.

For complex numbers a representation of the form

z = r (cos α + i sin α) , r > 0,

implies that r = |z| and α ∈ Arg(z). The bicomplex situation is more sophisticated.
If a bicomplex number has a representation of the form

Z = C · (cos α + j sin α)

with C and α complex numbers in C(i), then C is not necessarily the complex
modulus of Z; it is so if and only if C is in the upper half-plane and in this case
α ∈ ArgC(i) (Z). If C is in the lower half-plane, then |Z|i = −C and α + π ∈
ArgC(i) (Z): this is because

Z = C (cos α + j sin α) = −C (− cos α − j sin α)

= −C (cos(α + π) + j sin(α + π)) .

We illustrate this phenomenon with the computation of the representation

(3.7) for Z = z1 ∈ C(i). By definition
 
z1 if z1 is in the upper half-plane;
2
|Z|i = |z1 |i = z1 =
−z1 if z1 is in the lower half-plane.
60 Chapter 3. Geometry and Trigonometric Representations

Since Z = z1 = z1 (cos 0 + j sin 0), then

ArgC(i) (Z) = ArgC(i) (z1 )
 

0 + 2πm = 2πm  m ∈ Z if z1 is in the upper half-plane;

= 

0 + π + 2πm = (2m + 1)π  m ∈ Z if z1 is in the lower half-plane.

Hence the trigonometric representation of any complex number in C(i) is: if
z1 belongs to the upper half-plane, then
z1 = z1 (cos(2πm) + j sin(2πm)) , m ∈ Z;
if z1 belongs to the lower half-plane, then
z1 = −z1 (cos(2m + 1)π + j sin(2m + 1)π) , m ∈ Z.
Remark 3.2.1. We used the names upper half-plane and lower half-plane as follows:
the upper half-plane for us is


Π+ := z ∈ C(i)  Im(z) > 0 or z ∈ [0, ∞) ,

and the lower half-plane is


Π− := z ∈ C(i)  Im(z) < 0 or z ∈ (−∞, 0) .

Thus z ∈ Π+ if and only if |z|i = z and z ∈ Π− if and only if |z|i = −z.
Besides if z is a real number, then |z|i = |z|, the usual real-valued modulus of real
numbers. Hence the complex modulus of a complex number extends the modulus of
real numbers, not the usual modulus of complex numbers.
Later on, we will see that the right-hand side of (3.7) is related with the
bicomplex exponential function, which will be explained in Chapter 6.
/ S0 , then the complex moduli of Z and Z † are the same,
Note that if Z ∈
which leads to

† z1 z2
Z = z1 − jz2 = |Z|i −j
|Z|i |Z|i
= |Z|i (cos Θ − j sin Θ)
with Θ from (3.7).
Given Z as in (3.7) and W = |W |i (cos Ω + j sin Ω), one has:
Z · W = |Z|i (cos Θ + j sin Θ) · |W |i (cos Ω + j sin Ω)
= |Z|i |W |i ((cos Θ cos Ω − sin Θ sin Ω) + j(cos Θ sin Ω + sin Θ cos Ω)) ,
and therefore
Z · W = |Z|i |W |i (cos(Θ + Ω) + j sin(Θ + Ω)). (3.8)
As we have previously explained, formula (3.8) does not necessarily mean
that |Z · W |i is equal to |Z|i · |W |i and that Θ + Ω belongs to ArgC(i) (Z · W ). What
we can conclude is:
3.2. Trigonometric representation in complex terms 61

(a) if the complex number |Z|i · |W |i is in the upper half-plane, then indeed

|Z · W |i = |Z|i · |W |i ;

besides, in this case

Θ + Ω ∈ ArgC(i) (Z · W )
and
ArgC(i) (Z · W ) = ArgC(i) (Z) + ArgC(i) (W );

(b) if the complex number |Z|i · |W |i is not in the upper half-plane, then

|Z · W |i = −|Z|i · |W |i ;

besides, in this case

Θ + Ω + π ∈ ArgC(i) (Z · W )

and
ArgC(i) (Z · W ) + π = ArgC(i) (Z) + ArgC(i) (W ).

The BCi -analogue of De Moivre formula is a less direct generalization of its

complex antecedent. On one hand, we can write, obviously, that

Z n = |Z|ni (cos ΘZ n + j sin ΘZ n ) .

On the other hand one can obtain by induction that

Z n = |Z n |i (cos(nΘ) + j sin(nΘ)) , (3.9)

but again not always |Z|ni and |Z n |i , as well as nΘ and ΘZ n , coincide. They do
coincide when the complex number |Z|ni is in the upper half-plane; if it is not,
then
|Z n |i = −|Z|ni and ΘZ n = nΘ + π.
We will call formula (3.9) the bicomplex De Moivre formula, although under-
standing that it does not always coincide with the trigonometric representation in
complex terms of Z n .
Since for an invertible Z there holds:

0 = Z · Z † = z12 + z22 ,

that is,
Z · Z † (z12 + z22 )−1 = Z · Z † |Z|−2
i = 1,
then using the trigonometric form of Z † we have:

Z −1 = Z † · |Z|−2
i = |Z|i (cos Θ − j sin Θ) · |Z|−2
i
62 Chapter 3. Geometry and Trigonometric Representations

and finally
Z −1 = |Z|−1
i (cos Θ − j sin Θ) . (3.10)
With this, formula (3.9) becomes true for any integer number n.
This result allows us to write the quotient of two invertible bicomplex num-
bers in trigonometric form:

Z |Z|i
= (cos(Θ − Ω) + j sin(Θ − Ω)) . (3.11)
W |W |i

In particular, if we take W = Z † we get:

Z
= cos(2Θ) + j sin(2Θ) , (3.12)
Z†
which is, obviously, a bicomplex number of complex modulus 1.
An analogous trigonometric form is obtained if we work with |Z|j . The defi-
nitions and computations are completely similar, yielding the trigonometric form

Z = |Z|j (cos Ψ0 + i sin Ψ0 ) , (3.13)

where the complex C(j)-modulus of the bicomplex number Z = ζ1 + iζ2 is


|Z|j = ζ12 + ζ22 ,

and (see the explanation above) Ψ0 is a C(j)-complex number. The notions of

argument, principal argument, and all the properties such as De Moivre formulas,
etc., are true and are in a complete analogy with the BCi case; it is clear, for
instance, what argC( j) (Z), argC( j),m (Z), ArgC( j) (Z) mean.

3.3 Trigonometric representation in hyperbolic terms

It turns out that for our goals it is convenient to work with bicomplex numbers
given in their idempotent representation

Z = β 1 e + β 2 e†

with β1 and β2 in C(i). As we know, Z is not in S0 , that is, Z is neither zero

nor a zero-divisor if and only if its hyperbolic modulus |Z|k is a positive, non
zero-divisor hyperbolic number. If this occurs, then we can write:
 
Z = |Z|k · |Z|−1
k · β1 e + β2 e
†
   
= |Z|k · |β1 |−1 · e + |β2 |−1 · e† · β1 e + β2 e†
3.3. Trigonometric representation in hyperbolic terms 63

which leads to 
β1 β2 †
Z = |Z|k · ·e+ ·e , (3.14)
|β1 | |β2 |
β1 β2
where the C(i)-complex numbers and are of (real) modulus one. Thus,
|β1 | |β2 |
they are of the form
β1 β2
= eiθ1 , = eiθ2 (3.15)
|β1 | |β2 |
{θ1 , θ2 } ⊂ [0, 2π), hence θ1 and θ2 have a well-defined and well-known geometric
meaning. Formula (3.14) leads us to consider the hyperbolic number

ΨZ := θ1 e + θ2 e†

which, we believe, deserves the names of hyperbolic argument, or hyperbolic angle,

associated to the bicomplex number Z. We will soon justify more properly both
names.
First of all, rewrite Z = x1 + iy1 + jx2 + ky2 ∈ BC \ S0 as

Z = (x1 + ky2 ) + i (y1 − kx2 ) = z1 + iz2


with z1 and z2 in D. Then |Z|2k = Z ·Z ∗ = z21 +z22 ∈ D+ \{0}, hence |Z|k = z21 + z22
is the value of the square root that belongs to D+ \ S0 . It follows that

Z = |Z|k · |Z|−1 (z1 + iz2 )

 k 
z1 z2 (3.16)
= |Z|k  2 + i 2 ,
z1 + z22 z1 + z22
z1 z2
where the hyperbolic numbers  and  2 are such that the sum of
+z21 z22 z1 + z22
their squares is equal to one.
This resembles the situation with the usual trigonometric functions cosine
and sine of real or complex variables whose squares add up to one, and indeed,
it will be shown in Chapter 6 that those trigonometric functions extend to all
bicomplex numbers, the hyperbolic numbers included, and they still satisfy the
identity
sin2 α + cos2 α = 1
for any α ∈ BC. Since this identity is true in particular for α ∈ D (this is because
bicomplex trigonometric functions of a hyperbolic variable take hyperbolic values!)
we conclude from (3.14) that for any invertible Z there exists a hyperbolic number

Ψ0 := ν1 e + ν2 e†

such that
z1 z2
cos Ψ0 = and sin Ψ0 = ; (3.17)
|Z|k |Z|k
64 Chapter 3. Geometry and Trigonometric Representations

note that because of the periodicity of trigonometric functions the choice of ν1

and ν2 is not unique.
Since Ψ0 acts somehow as a hyperbolic argument, or hyperbolic angle, of Z,
we must establish the relation between Ψ0 and ΨZ .
Proposition 3.3.1. Given an invertible bicomplex number Z, it has a trigonometric
representation in hyperbolic terms (or a BCk -trigonometric representation) given
as

Z = |Z|k · (cos ΨZ + i sin ΨZ )

 
= |Z|k · eiν1 · e + eiν2 · e†

with ΨZ = ν1 · e + ν2 · e† ∈ D+
inv being the hyperbolic angle, or hyperbolic principal
argument, of Z.
Proof. It is known (see again Chapter 6 but also [45]) that

cos Ψ0 = cos(ν1 e + ν2 e† ) = cos ν1 · e + cos ν2 · e†

and
sin Ψ0 = sin(ν1 e + ν2 e† ) = sin ν1 · e + sin ν2 · e† ;
hence

cos Ψ0 + i sin Ψ0 = (cos ν1 + i sin ν1 ) · e + (cos ν2 + i sin ν2 )e† .

Combining this equation with (3.16) and (3.14) one gets:

Z = |Z|k (cos Ψ0 + i sin Ψ0 )

 
= |Z|k (cos ν1 + i sin ν1 )e + (cos ν2 + i sin ν2 )e†

β1 β2
= |Z|k ·e+ · e† .
|β1 | |β2 |

Taking into account the periodicity of trigonometric functions and choosing the va-
lues of ν1 and ν2 in the interval [0, 2π) we obtain that Ψ0 may be taken equal to ΨZ .
We prefer the notation ΨZ emphasizing that the angle is linked to Z ∈ BC\S0 . 
In contrast with the complex argument of a bicomplex number which is de-
fined for invertible bicomplex numbers only, we can define the hyperbolic argument
for zero-divisors too. Consider, for instance, a zero-divisor Z = β1 e, then

Z · Z ∗ = |β1 |2 e

implying that |Z|k = |β1 | · e, a semi-positive hyperbolic number; thus

β1
Z = |Z|k · · e,
|β1 |
3.3. Trigonometric representation in hyperbolic terms 65

β1
and since is a complex number of (real) modulus one, it can be written as
|β1 |
iν1
e with ν1 ∈ [0, 2π). In this case the hyperbolic angle associated to Z is the
hyperbolic zero-divisor
ν := ν1 e.
Analogously, if Z = β2 e† , then |Z|k = |β2 |e† and
β2
Z = |Z|k · · e† = |Z|k eiν2 e† .
|β2 |
In this case, the hyperbolic angle associated to Z is the hyperbolic zero-divisor

ν = ν 2 e†

with ν2 ∈ [0, 2π).

3.3.1 Algebraic properties of the trigonometric representation of

bicomplex numbers in hyperbolic terms
Our next task is to analyze the algebraic and geometric consequences of this
representation. We start with the algebraic ones. First, in analogy with the (real)
argument of a complex number and with the complex arguments of a bicomplex
number, we introduce some notation. We denote by argD Z, for Z = 0, the principal
hyperbolic argument of Z, that is, argD Z = ΨZ for Z ∈ BCinv and argD Z = ν1 e
or argD Z = ν2 e† for a zero-divisor Z. Then, argm,n;D Z is

argm,n;D Z := (ν1 + 2πm)e + (ν2 + 2πn)e†

for an invertible Z and similarly for a zero-divisor. Finally, ArgD Z denotes the set
of all possible hyperbolic angles:


ArgD Z := argm,n;D Z  m, n in Z

for an invertible Z and similarly for a zero-divisor.

Example 3.3.1.1. Let us consider the trigonometric representation in hyperbolic
terms of some particular bicomplex numbers.
(a) Given a complex number z = x + iy ∈ C(i), then z = ze + ze† ; we have
mentioned before that in this case the hyperbolic modulus coincides with the
usual modulus of complex numbers, i.e.,

|z|k = |z|e + |z|e† = |z|.

Hence, the trigonometric form of z in hyperbolic terms is:


z z †  
z = |z| · e+ e = |z| eiθ e + eiθ e† = |z|eiθ ,
|z| |z|
66 Chapter 3. Geometry and Trigonometric Representations

thus, the usual trigonometric form of the complex number z and its trigono-
metric form in hyperbolic terms when seen as a bicomplex number, coincide.
In particular, the hyperbolic argument of z, Ψz = θe + θe† = θ, coincides
with the usual (real) argument of z. Hence


ArgD z = Argz = θ + 2mπ  m ∈ Z .

(b) The imaginary unit j has the idempotent form j = (−i)e + ie† , thus |j|k =
1e + 1e† = 1. Hence, its trigonometric representation in hyperbolic terms is
  3 π
j = |j|k −ie + ie† = ei 2 π e + ei 2 e† ,
3 π
and its hyperbolic argument is Ψj = πe + e† , thus
2 2
  
3 1 
ArgD j = + 2m πe + + 2n πe†  m, n ∈ Z .
2 2

(c) If Z = z = ae + be† , with a, b in R, that is, z ∈ D, |z|k = |a|e + |b|e† , then its
trigonometric representation in hyperbolic form is

a b †  
z = |z|k e + e = |z|k ±e ± e† .
|a| |b|
Hence there are four options for the principal value of the hyperbolic argu-
ment of a hyperbolic number as indicated in Figure 3.3.1:

Ψz = 0e + 0e† = 0,

or
Ψz = 0e + πe† = πe† ,
or
Ψz = πe + 0e† = πe,
or
Ψz = πe + πe† = π.

Hence the set ArgD z is, according to the quadrant the hyperbolic number
z is in, as follows:

ArgD z
⎧ 

⎪
⎪ 2π(me + ne† )  m, n ∈ Z if z is in the 1st quadrant;
⎪
⎪ 

⎪
⎨ 2πme + (2n + 1)πe  m, n ∈ Z
†
if z is in the 2nd quadrant;
= 

⎪
⎪ (2m + 1)πe + 2nπe†  m, n ∈ Z if z is in the 3rd quadrant;
⎪
⎪
⎪
⎩ 

(2m + 1)πe + (2n + 1)πe†  m, n ∈ Z if z is in the 4th quadrant. 

3.3. Trigonometric representation in hyperbolic terms 67

2nd

Ψz = πe†
3rd e 1st

Ψz = π Ψz = 0
O x

e†

Ψz = πe
4th

Figure 3.3.1: Hyperbolic arguments of hyperbolic numbers seen as bicomplex

numbers.

Coming back to the properties of the trigonometric representation in hyper-

bolic terms, take two invertible bicomplex numbers Z and W in their trigonometric
representations in hyperbolic terms, then
Z · W = |Z|k (cos ΨZ + i sin ΨZ ) · |W |k (cos ΨW + i sin ΨW )
   
= |Z|k · |W |k · eiν1 · e + eiν2 · e† · eiμ1 · e + eiμ2 · e†
= |Z · W |k · (cos(ΨZ + ΨW ) + i sin(ΨZ + ΨW )) (3.18)
 
= |Z · W |k · ei(ν1 +μ1 ) · e + ei(ν2 +μ2 ) · e† .

Hence, one concludes that

ArgD (Z · W ) = ArgD Z + ArgD W.
68 Chapter 3. Geometry and Trigonometric Representations

Similarly for the case when one of the factors, or both, are zero-divisors.
In particular, taking Z = W one gets that
 
Z 2 = |Z|2k · ei(2ν1 ) · e + ei(2ν2 ) · e†
 
= |Z|2k · (cos(2ν1 ) + i sin(2ν1 )e + (cos(2ν2 ) + i sin(2ν2 )e† .

Using induction, we obtain an analogue of the De Moivre formula: for any

n ∈ N it holds that
 
Z n = |Z|nk · ei(nν1 ) · e + ei(nν2 ) · e†
  (3.19)
= |Z|nk · (cos(nν1 ) + i sin(nν1 )e) + (cos(nν2 ) + i sin(nν2 )e† ) .

Similarly the case of zero-divisors.

For an invertible bicomplex number its hyperbolic modulus is an invertible
hyperbolic number, hence
 −iν1 
Z −1 = |Z|−1
k · e · e + e−iν2 · e†

which implies immediately that the De Moivre formula is true for any n ∈ Z.
Let us see what is the relation between the trigonometric representations
in hyperbolic terms of Z and Z ∗ . Since the ∗-conjugate of Z = β1 e + β2 e† is
Z ∗ = β 1 e + β 2 e† , we have:

|Z ∗ |k = |β1 | · e + |β2 | · e† = |Z|k ,

thus

β1 β
Z ∗ = |Z ∗ |k · · e + 2 · e†
|β1 | |β2 |
 −iν1 −iν2

= |Z|k · e ·e+e · e† ,

and we conclude that

ArgD Z ∗ = −ArgD Z,
compare again with the complex numbers case.

3.3.2 A geometric interpretation of the hyperbolic trigonometric

representation.
We will see now that, in analogy with the complex numbers situation, the hyper-
bolic trigonometric representation of a bicomplex number provides information
with clear geometrical interpretations.
Since the hyperbolic modulus of a bicomplex number Z = β1 e+β2 e† , with β1
and β2 in C(i), is the positive (not necessarily strictly positive) hyperbolic number

|Z|k = |β1 | · e + |β2 | · e† ,

3.3. Trigonometric representation in hyperbolic terms 69

one can formally define the bicomplex sphere Sγ0 centered at the origin and with
radius being a fixed positive hyperbolic number γ0 = a0 e + b0 e† ∈ D+ , i.e.,


Sγ0 := Z ∈ BC  |Z|k = γ0 .

Let us show that such a set can be visualized perfectly well.

First note that if either a0 or b0 is zero, that is, γ0 is a zero-divisor, say

γ0 = a0 · e, then
Sγ0 = {Z = β1 · e | |β1 | = a0 };

this set is a circumference in the real two-dimensional plane BCe with center at
a0
the origin and radius √ = |a0 · e|. Similarly, if γ0 = b0 · e† , then the set Sγ0 is a
2
b0
circumference in BCe† with center at the origin and radius √ = |b0 · e† |.
2
Thus, whenever the radius of a bicomplex sphere is a (positive) zero-divisor,
i.e., a semi-positive hyperbolic number, then the sphere is, in fact, a usual circum-
ference. See Figures 3.3.2 and 3.3.3.

C(i)
i
ie a0 e
e
O 1 a0
O BCe

Figure 3.3.2: The bicomplex sphere Sa0 e .

If a0 = 0 and b0 = 0, then the intersection of the hyperbolic plane D and the

sphere Sγ0 consists exactly of the four hyperbolic numbers ±a0 e ± b0 e† , that is,
in these points the plane D touches the sphere tangentially. What is more, in this
case the whole sphere Sγ0 is the surface of a three-dimensional manifold with the
shape of a torus (see Figure 3.3.4).
We emphasize the fact that although the torus is three-dimensional as a
manifold and its surface is a two-dimensional manifold, they both live in a four-
dimensional world. This is because now the sphere Sγ0 is given as


Sγ0 = Z = β1 e + β2 e†  |β1 | = a0 , |β2 | = b0

70 Chapter 3. Geometry and Trigonometric Representations

C(i)
i ie†

e† b 0 e†
O 1 b0 O

BCe†

Figure 3.3.3: The bicomplex sphere Sb0 e† .

which allows another description of this set: it is the cartesian

 product of the two
a0
circumferences, namely, one of them, denoted by Ce 0; √ , is situated in the
2
a0
plane BCe having the origin as its center and √ as its radius; the other one,
 2
b0
denoted by Ce† 0; √ , is situated in the plane BCe† having the origin as its
2
b0
center and √ as its radius. We illustrate this with Figure 3.3.4.
2
These spheres of hyperbolic radius have intrinsic interest, but for the moment
we will use them in order to present a geometric interpretation of the hyperbolic
argument of a bicomplex number Z ∈ BCinv written as
 
Z = |Z|k eiν1 · e + eiν2 · e†

with principal value hyperbolic argument argD Z = ν1 e + ν2 e† ∈ D+ inv .

Let us fix the value of |Z|k =: γ0 = a0 e + b0 e† , then all the bicomplex
numbers having it as their hyperbolic modulus belongs to the sphere Sγ0 which is
the surface of a Euclidean torus; as we mentioned before, this surface can be seen
as the cartesian product
 
a0 b0
Ce 0; √ × Ce† 0; √ .
2 2

Any point of the torus is in one-to-one correspondence with hyperbolic num-

bers of the form μ1 e + μ2 e† where μ1 and μ2 are in [0, 2π); the latter, in turn, are
in one-to-one correspondence with the pairs of points each of which belongs to the
3.3. Trigonometric representation in hyperbolic terms 71

a0
√
2

b0
√
2

Figure 3.3.4: The bicomplex sphere of hyperbolic radius a0 e + b0 e† .

corresponding circumference:

iμ1 a0
a0 e e ∈ Ce 0; √ ,
2

b0
b0 eiμ2 e† ∈ Ce† 0; √ ;
2
and moreover,
   
a0 eiμ1 e + b0 eiμ2 e† = a0 e + b0 e† · eiμ1 e + eiμ2 e†
 
= γ0 · eiμ1 e + eiμ2 e† = Z.

This shows that the value of the hyperbolic modulus of Z tells us on which sphere
Z is situated and its hyperbolic argument determines the exact place of Z on the
sphere.
Note also that the sphere Sγ0 is in a bijective correspondence with the hy-
perbolic interval


[0, 2π)D = z ∈ D  0  z ≺ 2π


= z = ν1 e + ν2 e†  ν1 , ν2 ∈ [0, 2π) ⊂ D+

given in Figure 3.3.5.

72 Chapter 3. Geometry and Trigonometric Representations

2πe
D

2π

2πe†

Figure 3.3.5: The hyperbolic interval [0, 2π)D .

The correspondence is given by

a0 eiν1 e + b0 eiν2 e† ←→ ν1 e + ν2 e† .

We will see later that this is related with the exponential functions of bicomplex
and hyperbolic variables.
Some of the geometric and trigonometric properties of hyperbolic and bicom-
plex numbers have been studied in the literature in works such as [28, 45, 56, 100].
We also mention here [87] and [11], where trigonometric representations of hy-
perbolic and hypercomplex numbers are studied, together with applications to
Special Relativity. The authors of [11] emphasize that the hyperbolic numbers are
“the mathematics of the two-dimensional Special Relativity”, and adapting this
two-dimensional hyperbolic geometry to “multidimensional commutative hyper-
complex systems” (such as bicomplex numbers) leads to a concrete application to
Special Relativity in four-dimensional Minkowski space-time. The “separation” of
the four-dimensional bicomplex space into spacelike and timelike parts (events)
is captured by the geometry of the complex lines Le and Le† , which we carefully
study in complete detail in the next chapter.
Chapter 4

Lines and curves in BC

4.1 Straight lines in BC

In this section we will investigate the complex straight lines in BC from both the
algebraic and the geometric points of view. We will start by describing the well-
known case of real straight lines in C(i), which we use as reference for the case of
complex and hyperbolic lines in BC.

4.1.1 Real lines in the complex plane

A real straight line in the plane R2 is given by the equation

a 1 x + a2 y = b , (4.1)

where a1 , a2 , b are real coefficients. Writing a = (a1 , a2 ), z = (x, y), (4.1) is equiv-
alent to
a, z (a1 , a2 ), (x, y)R2 b
= = . (4.2)
| a | | a | | a |
This formula has the following geometric description: a point (x, y) belongs to
the straight line given by (4.1) if the corresponding vector has a constant projection
b
on the straight line determined by a.
|a|
1
Setting z := x + iy and a := a1 + ia2 and recalling that x = (z + z),
2
i 1 i
y = (z − z), a1 = (a + a) and a2 = (a − a), we obtain the complex form of
2 2 2
writing equation (4.1):
az + az = 2b . (4.3)
Geometrically, a complex number z belongs to the line  given by (4.3) if
Re(az) = Re(az) is equal to the constant b (see Figure 4.1.1).

© Springer International Publishing Switzerland 2015 73

M.E. Luna-Elizarrarás et al., Bicomplex Holomorphic Functions, F rontiers in Mathematics,
DOI 10.1007/978-3-319-24868-4_5
74 Chapter 4. Lines and curves in BC

āz


āz + az̄ = 2b
z
ā

2b b

a
z̄

az̄

Figure 4.1.1: The point z belong to  if and only if Re(az) = Re(az).

If we consider a and z in their trigonometric forms, i.e., a = |a|(cos φ0 +

i sin φ0 ) and z = |z|(cos φ + i sin φ), then equation (4.3) becomes
|a||z| cos(φ − φ0 ) = b . (4.4)
The latter formula can also be obtained from (4.2).
Note also that if z1 and z2 are two points on a real line, then the vector
z2 − z1 is a vector along the line. Then, since a, z2 − z1 R2 = 0, it follows that a
is a normal vector to our line, i.e., orthogonal to it.
Thus the angle that our line makes with the positive direction of the abscissa
π
axis is φ0 + . The tangent of this angle is called the slope m of the line:
2
 π
m = − cot φ0 = tan φ0 + . (4.5)
2
a1
If the line is given by equation (4.1) and a2 = 0, then m = − .
a2
In complex form, the case we are considering, i.e., a2 = 0, implies that a = 0
and a = 0, thus if we divide by a in (4.3), we obtain:
4.1. Straight lines in BC 75

a 2b
z=− z+ . (4.6)
a a
We introduce the notation
a
M := − = −ei2φ0 = ei(2φ0 +π) .
a
a2
Since a = |a|2 a−1 , then M = − . Moreover, in order to capture the slope
|a|2
of our line, we write:

a a1 + ia2 − a1 − i tan(φ0 + π2 ) − i m−i

M =− =− = − aa21 =− = ,
a a1 − ia2 − a2 + i tan(φ0 + 2 ) + i
π
m+i

thus we get the slope:

 π 1−M a+a a1
m = tan φ0 + =i =i =− .
2 1+M a−a a2
In conclusion, M is an intrinsic characteristic complex number for our real
line, which determines its geometric slope. Equation (4.6) then becomes

z =Mz+B, (4.7)

2b
where B := .
a
We note here some particular cases. If M = −1, then a = a = a1 , a real
number, i.e., a2 = 0, which yields the vertical line a1 x = b, which has no well-
defined slope. In complex terms the equation of this vertical line is z = −z + B,
2b
where B = is a real number.
a1
If M = 1, then a = −a = ia2 , so a1 = 0, which yields the horizontal line
2b
a2 y = b. In complex terms we obtain z = z + B, where B = i is a purely
a2
imaginary number.
If it is known that the real line passes through a point z0 = x0 + iy0 and it
has slope m = 0, then the equation can be written as

y − y0 = m(x − x0 ) ,

or in complex terms as
z + z0 = M (z + z0 ) .
If the real line is determined by two points z0 = x0 + iy0 and z1 = x1 + iy1 ,
such that x0 = x1 , it is given by equation
y1 − y0
y − y0 = (x − x0 ) . (4.8)
x1 − x0
76 Chapter 4. Lines and curves in BC

The characteristic complex number M takes the form

z1 − z0
M= ,
z1 − z0
so the complex equation of the line is
z1 − z0
z + z0 = (z + z0 ) . (4.9)
z1 − z0
If in equation (4.8) we denote x1 − x0 =: ν1 and y1 − y0 =: ν2 , then, if we
assume they both are non-zero real numbers, we obtain the equation
y − y0 x − x0
= =: t ,
ν2 ν1
where t is a real parameter. We get therefore a parametrization of this line given
by
x(t) = x0 + tν1 , y(t) = y0 + tν2 .
Note that its slope is
ν2
m= ,
ν1
and its associated complex characteristic number becomes
ν1 + iν2
M= .
ν1 − iν2
Then the equation of our real line is
ν 2 x − ν 1 y = ν 2 x 0 − ν 1 y0 ,
thus, in complex notation, the complex coefficient a = ν2 − iν1 = −i(ν1 + iν2 ) and
the parametrized equation becomes
z(t) = z0 + t · ia . (4.10)
If we restrict the real variable t to be in an interval [c, d] ⊂ R, then the equation
above yields a straight line segment in R2 .
The parametric representation of straight lines is illustrated in Figure 4.1.2.
Note that every point w on the line ia seen as a complex number is of the
form w = t·ia with a fixed, and any point of the line  is obtained as the translation
t · ia + z0 of the point t · ia by the vector z0 . Thus, any straight line which passes
through the origin is the set of complex numbers

a  λ ∈ R},
 a := {w = λ"
where "a is a fixed arbitrary complex number, and if the line does not pass through
the origin but passes through a point, say z0 , then the line  is the set of complex
numbers of the form 
a + z0  λ ∈ R}
a + z0 := {λ"
for some "
a ∈ C \ {0}.
4.1. Straight lines in BC 77

z0


ia

a z0 + tia

ia

tia

Figure 4.1.2: Parametric representation of a straight line.

4.1.2 Real lines in BC

Take a non-zero bicomplex number Z0 , then the real straight line which passes
through Z0 and the origin (that is, its direction is determined by Z0 ) is defined to
be the set 
Z0 := {λZ0  λ ∈ R}.
Let Z1 be another bicomplex number different from zero, then the straight line in
the direction of Z0 and passing through Z1 is the set

Z0 + Z1 := {λZ0 + Z1  λ ∈ R} .

4.1.3 Complex lines in BC

In Section 4.1.1 we described real lines in C using both real and complex languages.
In what follows we are going to extend the idea onto the complex lines, seeing them
as analogues of the real lines in C but now inside BC which is seen as an analogue
of the set of complex numbers. But saying “complex” in BC may mean any of the
sets C(i) or C(j). We will start our analysis considering C(i)-complex lines.
78 Chapter 4. Lines and curves in BC

Following the analogy with the real lines, we define a complex straight line
in C2 (i) to be the set of solutions (z1 , z2 ) ∈ C2 (i) of the equation

a1 z1 + a2 z2 = b, (4.11)

where a1 , a2 , b ∈ C(i) are complex coefficients. Clearly, this equation is equivalent

to a system of two real linear equations with four real variables, and thus it defines,
generally speaking, a 2-dimensional plane in R4 ; this happens when the matrix of
coefficients has rank 2; it is clear what happens if the rank is less than 2.
If we see C2 (i) as BC in such a way that Z = z1 + jz2 , then we get: z1 =
Z + Z† Z − Z†
and z2 = −j , thus the complex line (4.11) will be considered as a
2 2
C(i)-complex line in BC with bicomplex equation

A† Z + AZ † = 2b, (4.12)

where A = a1 + ja2 = 0 is a bicomplex coefficient. Here we have to distinguish

between the point Z = z1 + jz2 = x1 + iy1 + jx2 + ky2 , etc. (using all the possible
writings for Z), and the vector Z,  which is a vector in R4 that joins the origin
and the point (x1 , y2 , x2 , y2 ). Thus, in what follows, when it will be necessary, we
will refer to the point Z as the point (z1 , z2 ) ∈ C2 (i) or (w1 , w2 ) ∈ C2 (j) (writing
Z = w1 + iw2 ) or (β1 , β2 ) ∈ C2 (i) (writing Z = β1 e + β2 e† ), or (γ1 , γ2 ) ∈ C2 (j)
(writing Z = γ1 e + γ2 e† ).
Note that if a1 = 0, a2 = 1, b = 0, equation (4.11) becomes

0 · z 1 + 1 · z2 = 0 ,

and the respective complex line is the set C(i) ⊂ BC.

If a1 = 1, a2 = 0, b = 0, then

1 · z1 + 0 · z2 = 0 ,

and the complex line is the set C(i) · j ⊂ BC.

4.1.4 Parametric representation of complex lines

Let us consider an alternative way of describing a C(i)-complex line in BC which
is equivalent to the one above. Take a fixed non-zero bicomplex number Z0 =
z10 + jz20 ∈ BC \ {0}. Consider the set

LZ0 := {λZ0  λ ∈ C(i)} ,

and let us show that it is a C(i)-complex line which we will call the C(i)-complex
line generated by Z0 and passing through the origin. Indeed, for any element
Z = z1 + jz2 in LZ0 there exists λ ∈ C(i) such that

z1 + jz2 = Z = λZ0 = λz10 + jλz20 ,

4.1. Straight lines in BC 79

which is equivalent to the system


z1 = λz10 ,
z2 = λz20 .

z2 z2
Assuming that z20 = 0, one has that λ = 0 and hence z1 = 0 z10 , or
z2 z2
equivalently:
z1 z20 − z2 z10 = 0 .
This equation is a particular case of (4.11) with a1 = z20 , a2 = −z10 and b = 0.
If z20 = 0, the set

Lz1 := {λz10  λ ∈ C(i)} = C(i)

is the complex line C(i) with equation z2 = 0.

Reciprocally, given a homogeneous equation

z1 a 1 + z 2 a 2 = 0 (4.13)

with a1 = 0, then one can take Z0 := −a2 + ja1 ; the set of solutions Z = z1 + jz2
of (4.13) coincides with LZ0 .
Example 4.1.4.1. We know already that any zero-divisor is of the form β1 e or
β2 e† , that is, Z is a zero-divisor if and only if Z ∈ Le \ {0} or Z ∈ Le† \ {0}, thus
the set of zero-divisors (together with the zero) is described by

S0 = Le ∪ Le† .

Note also that here we have two complex lines (Le and Le† ) that intersect only at
the origin. The reader should recall that in R3 a pair of two-dimensional planes
cannot intersect at just one point, but in BC ∼= R4 there is an additional dimension
which allows this phenomenon.
The corresponding equations of the complex lines Le and Le† are, respec-
tively,
z1 + iz2 = 0
and
z1 − iz2 = 0 . 
In order to define a complex line that does not necessarily pass through the
origin, take W0 = w10 +jw20 ∈ BC\{0}. Then the C(i)-complex line passing through
W0 and parallel to LZ0 (that is, a complex line with the direction determined by
Z0 ) is defined as the set

LZ0 + W0 := {λZ0 + W0  λ ∈ C(i)} .
80 Chapter 4. Lines and curves in BC

Let us show that this definition is equivalent to that of a complex line. Indeed,
given a point Z ∈ LZ0 + W0 , with Z = z1 + jz2 , there exists λ ∈ C(i) such that

z1 = λz10 + w10 ,
z2 = λz20 + w20 .

Assuming again that z20 = 0, we obtain that the complex cartesian components of
Z satisfy the equation

z1 z20 − z2 z10 = w10 z20 − w20 z10 ,

which is again a particular case of (4.11), now in its inhomogenous form.

Reciprocally, given an equation

z1 a1 + z2 a2 = b, (4.14)

with a1 = 0 and a2 = 0, let us find Z0 and W0 in BC\{0} such that any Z = z1 +jz2
whose components satisfy (4.14) belongs to LZ0 + W0 . From the above equation
we can write z1 as 
b − a2 1 − z2
z1 = + a2 .
a1 a1
z2 − 1 b − a2
Define λ := , that is, z1 = − λa2 . Substituting in (4.14), we write
a1 a1
now z2 as
z2 = 1 + λa1 .
b − a2
Let us show that Z0 := −a2 +ja1 and W0 := +j are the bicomplex numbers
a1
which we are looking for; indeed,
b − a2
λZ0 + W0 = −λa2 + jλa1 + +j
a
 1
b − a2
= − λa2 + j (λa1 + 1)
a1
= z1 + jz2 = Z,

thus Z is a point on the complex line LZ0 + W0 .

What happens if either a1 = 0 or a2 = 0? If a1 = 0, a2 = 0, then (4.14)
becomes
a2 z2 = b. (4.15)
b
We have already considered the case b = 0. If b = 0, z2 is the constant z2 =
a2
b
and z1 takes any value in C(i). Hence, (4.15) is the complex line Li + j, where

a 2
Li := λi  λ ∈ C(i) = C(i).
4.1. Straight lines in BC 81

Similarly, if a2 = 0 and a1 = 0, then (4.14) becomes

a1 z1 = b,
and if b = 0, then it is the complex line
b 

Lj + j with Lj := λj  λ ∈ C(i) .
a1
Let us come back to equation (4.12) where the coefficient A is defined to be
A = a1 + ja2 ; we know already that the bicomplex number Z0 which determines
the direction of the complex line is given as
Z0 = −a2 + ja1 = j(a1 + ja2 ) = jA ,
that is,
A = −jZ0 .
Thus (4.12) becomes
Z0 Z † − Z0† Z = 2jb . (4.16)

4.1.5 More properties of complex lines

We continue now with some other properties of complex lines.
Proposition 4.1.5.1. A non-zero bicomplex number U belongs to the complex line
LZ0 if and only if LZ0 = LU . Thus the complex line is generated by any non-zero
bicomplex number lying on it.
Proposition 4.1.5.2. The non-zero bicomplex number Z0 belongs to S if and only
if LZ0 = Le or LZ0 = Le† .
The proofs of these two propositions are direct applications of the definitions.
Corollary 4.1.5.3. Z0 ∈ S0 if and only if LZ0 ∩ Le = {0} and LZ0 ∩ Le† = {0}.
It is instructive to look at equation (4.16) in idempotent form. Write Z0 =
β10 e + β20 e† , Z = β1 e + β2 e† , then (4.16) becomes
β20 β1 − β10 β2 = 2ib . (4.17)
An immediate application of this formula is
Proposition 4.1.5.4. If Z0 ∈ S0 , then the complex line LZ0 + W0 intersects each
of Le and Le† at exactly one point.
Proof. Since Z0 ∈ S0 , then β10 = 0 and β20 = 0. Let Z be a point in the intersection
(LZ0 + W0 ) ∩ Le , then its idempotent components satisfy (4.17) and β2 = 0. Hence
2ib 2ib
we have: β1 = 0 , i.e., Z = 0 e, and there are no other points in this intersection.
β2 β2
2ib
Exactly in the same way, if Z ∈ (LZ0 + W0 ) ∩ Le† , then Z2 = − 0 e† . 
β1
82 Chapter 4. Lines and curves in BC

If we perform a similar analysis in C2 (j), then all the formulas and conclusions
above are valid in their corresponding analogues. For example, the C( j)-complex
line
α1 ζ1 + α2 ζ2 = ν,
where all numbers and coefficients are in C(j), has the bicomplex formulation
AZ + AZ = 2ν,
where A = α1 + iα2 and Z = ζ1 + iζ2 are bicomplex numbers. Nevertheless, they
are in the same set BC, so that, as a matter of fact, there are two types of complex
lines in BC, the C(i)-lines and C(j)-lines.
Given a C(i)-complex straight line, is it possible to consider it as a C(j)-
complex straight line or reciprocally? This question is related with the following
one: given a real two-dimensional plane in BC, is it always possible to consider it
as a complex line?
Here one can find some partial answers to these questions:
• What are the C(i)-complex lines containing the number 1? Assume that L is
such a C(i)-complex line of the form L = LZ0 for some Z0 ∈ BC \ {0}. Let
L1 be the complex line generated by 1, then obviously L1 = C(i). We know
also that 1 ∈ LZ0 is equivalent to LZ0 = L1 , that is, L = C(i).
Thus, among all the two-dimensional real planes which contain the axis x1 ,
there is only one which has the structure of a C(i)-complex line.
• In the same way, among all the two-dimensional real planes which contain
the axis x1 , only C(j) is a C(j)-complex line.
• Other conclusions are: C(j) is not a C(i)-complex line; C(i) is not a C(j)-
complex line.
• As we did with the C(i)-complex straight lines, it can be proved in a similar
fashion that, given Z0 ∈ BC \ {0}, the C(j)-complex line generated by Z0 ,
i.e., in the direction of Z0 , is the set

TZ0 := {μZ0  μ ∈ C(j)} .

As a consequence of these observations we can prove that:

Proposition 4.1.5.5. The C(i)- and C(j)-complex lines generated by a non-zero
bicomplex number Z0 coincide if and only if Z0 is a zero-divisor.
Proof. Assume first that Z0 ∈ S, then Z0 = βe or Z0 = γe† , with β, γ ∈ C(i). If
Z0 = βe, then
 
LZ0 = {λZ0 = λβe  λ ∈ C(i)} = {νe  ν ∈ C(i)}
 
= {(a + ib)e  a, b ∈ R} = {(a − jb)e  a, b ∈ R}

= {μe  μ ∈ C(j)} = TZ0 .
4.1. Straight lines in BC 83

Here we used the “peculiar” identity ie = −je. Analogously for Z0 = γe† .

Reciprocally, assume that TZ0 = LZ0 and suppose that Z0 ∈ S. The equality
of the sets TZ0 and LZ0 is equivalent to saying that given any λ ∈ C(i), there exists
μ ∈ C(j) such that λZ0 = μZ0 . But since Z0 is not a zero-divisor, the cancellation
law can be applied to the latter equality, concluding that C(i) λ = μ ∈ C(j), for
any λ ∈ C(i), which is impossible. Hence Z0 ∈ S. 
We now come back to equation (4.12) and we analyze some more geometrical
aspects of it. Here we are interested only in the case where A is not a zero-divisor.
We know already that on this complex line all, but at most two points, are non-
zero-divisors.
We can write A and Z in their BCi -trigonometric forms:

A = |A|i (cos Θ0 + j sin Θ0 ) , Z = |Z|i (cos Θ + j sin Θ) ,

where Θ and Θ0 are the complex arguments of A and Z, respectively. Then equa-
tion (4.12) becomes
|A|i |Z|i cos(Θ − Θ0 ) = b. (4.18)
We observe here the similarity between the equations of real lines (4.4) and
complex lines (4.18)! Moreover, inspired by (4.18), we define the C(i)-complex
projection of Z onto A, for any invertible bicomplex numbers A and Z, by

projA (Z) := |Z|i cos(Θ − Θ0 ) ,

where Θ − Θ0 can be called the complex angle between A and Z. We will come
back to this notion later.

4.1.6 Slope of complex lines

Continuing along the same path of study of real lines in R2 , we further investigate
our complex straight lines. We continue for the time being to consider the case
when A = −jZ0 is not a zero-divisor, hence also Z0 is not a zero-divisor. We study
first the subcase when a1 = 0, thus a2 = 0 and Z0 = −a2 . Hence LZ0 = L−a2 =
C(i). Our straight line is of the form

LZ0 + W0 = C(i) + W0 ,

b
with W0 = j. This is a complex line which we will call parallel to the complex
a2
line C(i). Making the convention that the complex line C(i) determines the notion
of C(i)-complex horizontality, any complex line parallel to C(i) will be called C(i)-
complex horizontal.
It follows that |A|i = ±a2 , where the sign should be chosen in accordance
with our agreement that the imaginary part of the complex number |A|i is non-
π
negative. In this case Θ0 = becomes a real number.
2
84 Chapter 4. Lines and curves in BC

The equation of this line in trigonometric form is

 π
±a2 · |Z|i cos Θ − = ±a2 |Z|i sin Θ = b .
2
Consider now the case a2 = 0, thus a1 = 0 and Z0 = ja1 . Hence
 
LZ0 = Lja1 = {λa1 j  λ ∈ C(i)} = {μj  μ ∈ C(i)} = Lj .

Notice that for any μ = a + ib ∈ C(i) the product μj is μj = aj + kb, hence the
C(i)-complex line Lj is the two-dimensional real plane generated by j and k. In
this case, our complex line is of the form

L Z0 + W 0 = L j + W 0

b
with W0 = + j.
a1
Note that for the set Lj + W0 one has:

b 
Lj + W0 = {μ · j + + j  μ ∈ C(i)}
a1
b  b 
= {(μ + 1) · j + μ ∈ C(i)} = {μ · j + μ ∈ C(i)}
a1 a1
b
= Lj + .
a1
b b
In other words, one may take W0 = and thus any point Z ∈ Lj + is of the
a1 a1
b b
form Z = μ · j + , i.e., Z = z1 + jz2 with a constant z1 = and an arbitrary
a1 a1
complex number z2 .
Making now the convention that the complex line Lj determines the notion
of C(i)-complex verticality, any C(i)-complex line Lj + W0 , that is parallel to Lj ,
will be called C(i)-vertical.
Since in this case |A|i = ±a1 and Θ0 can be taken to be zero, then the
equation of any vertical line in trigonometric form is

±a1 · |Z|i cos(Θ) = b .

Since Lj ∩ C(i) = {0}, then any horizontal line intersects any vertical line in just
one point. We have seen this phenomenon before, in the sense that R4 is “wide”
enough and a pair of two-dimensional planes may intersect at one point only.
Now some words about the case when Z0 is a zero-divisor. By Proposi-
tion 4.1.5.2, this is equivalent to LZ0 = Le or LZ0 = Le† , thus, if Z0 ∈ S,
then any C(i)-complex line LZ0 + W0 is parallel to Le or to Le† . We call these
types of complex lines zero-divisor complex lines.
4.1. Straight lines in BC 85

We will see now that it is quite appropriate to say that a complex line L :=
LZ0 + W0 has the direction along Z0 (or that it is parallel to LZ0 ) since we will
relate the complex line with the complex argument of Z0 .
Recall that Z0 = −a2 + ja1 , with

a 1 z1 + a 2 z2 = b (4.19)

being the equation of a complex line parallel to LZ0 and passing through W0 . Let
us start with the case Z0 ∈ S0 , which is equivalent to Z0 Z0† = 0; thus, we can
consider the trigonometric form of Z0 :

a2 a1
Z0 = |Z0 |i − +j = |Z0 |i (cos Ψ0 + j sin Ψ0 ) .
|Z0 |i |Z0 |i

One obtains directly that

a1
tan Ψ0 = − =: Λ ,
a2

which we will call the complex slope of the line L. Comparing with equation (4.19),
we have a deep analogy with the definition of slope of usual real lines in R2 . Thus,
the slope Λ of the complex line L is the tangent of the complex argument Ψ0 of
Z0 , with Z0 ∈ S0 being a bicomplex vector that determines the direction of the
line L.
Note that if Z0 = −a2 + ja1 is still a non-zero-divisor, but a2 = 0, we have
seen before that in this case we have a vertical line, and again the analogy with real
lines in R2 is maintained, since we can define Λ = ∞ and the complex argument
π
of Z0 is , coinciding with the usual notion of a vertical line.
2
What happens when Z0 is a zero-divisor? We know already that LZ0 = Le
or LZ0 = Le† and our line is a zero-divisor complex line. Since in this case we have
also that

1 1 1
Z0 = −a2 + ja1 = λ (1 ± ij) = λ ± iλ · j, λ ∈ C(i),
2 2 2

then
a1
Λ=− = ±i
a2
and since there are no complex numbers Ψ such that tan Ψ = ±i we can define
the slope of the zero-divisor complex lines as being equal to i or −i.

The previous reasoning motivates the question about the relationship be-
tween complex arguments of bicomplex numbers belonging to the same complex
line.
86 Chapter 4. Lines and curves in BC

4.1.7 Complex lines and complex arguments of bicomplex numbers

Proposition 4.1.7.1. Given a complex line L = LZ0 generated by Z0 ∈/ S0 and any
S ∈ LZ0 , then the complex arguments of S and Z0 are related as follows:

ArgC(i) (S) = ArgC(i) (Z0 ) or ArgC(i) (S) = ArgC(i) (Z0 ) + π.

In other words, the slope of LZ0 does not depend on the generator Z0 .
Proof. Given S ∈ LZ0 \ {0}, then there exists λ ∈ C(i) \ {0} such that
 
S = λZ0 = λ z10 + jz20 = λz10 + jλz20 ;

hence S ∈/ S0 . We know already that |S|i = ±|λ|i · |Z0 |i . More precisely we have
to consider the following cases.
(a) If λ ∈ Π+ , the
upper
 half-plane,

this is equivalent to |λ|i = λ, and thus
ArgC(i) (λ) = 2πm  m ∈ Z .
(a.1) If |λ|i · |Z0 |i = λ|Z0 |i ∈ Π+ , then |S|i = λ|Z0 |i and thus

ArgC(i) (S) = ArgC(i) (λ) + ArgC(i) (Z0 ) = ArgC(i) (Z0 );

(a.2) if |λ|i · |Z0 |i = λ|Z0 |i ∈ Π− , then |S|i = −λ|Z0 |i and thus

ArgC(i) (S) = ArgC(i) (λ) + ArgC(i) (Z0 ) + π = ArgC(i) (Z0 ) + π.

(b) Now, if λ ∈ Π − , the lower half-plane,

this is equivalent to |λ|i = −λ and thus
ArgC(i) (λ) = (2m + 1)π  m ∈ Z .
(b.1) If |λ|i · |Z0 |i = −λ|Z0 |i ∈ Π+ , then |S|i = −λ|Z0 |i and thus

ArgC(i) (S) = ArgC(i) (λ) + ArgC(i) (Z0 )



= (2m + 1)π  m ∈ Z + ArgC(i) (Z0 )

= ArgC(i) (Z0 ) + π.

(b.2) if |λ|i · |Z0 |i = −λ|Z0 |i ∈ Π− , then |S|i = λ|Z0 |i and thus

ArgC(i) (S) = ArgC(i) (λ) + ArgC(i) (Z0 ) + π = ArgC(i) (Z0 ). 

The proposition above inspires the idea of complex rays.

Definition 4.1.7.2. We define the C(i)-complex ray in the direction of Z0 = 0 to
be the set
  
RZ+
:= S = λZ 0
 λ ∈ C \ {0} and Arg C(i) (S) = Arg C(i) (Z 0 ) ∪ {0},
0

and the C(i)-complex ray in the “opposite” direction of Z0 = 0 to be the set

  
RZ−
:= S = λZ 0
 λ ∈ C \ {0} and ArgC(i) (S) = ArgC(i) (Z0 ) + π ∪ {0}.
0
4.1. Straight lines in BC 87

We are able now to provide different ways of writing the equation of a complex
line, according to the information that one has at hand. Consider first the case in
which L is not a vertical line, thus a2 = 0, and write the equation of L as

a1 b
z2 = − z1 + = Λz1 + c , (4.20)
a2 a2

b
with c = . Let Z = z1 + jz2 be such that its cartesian components satisfy (4.20).
a2
Using also the relations

1 j †
z1 = (Z + Z † ), z2 = (Z − Z) , (4.21)
2 2
one obtains:
Z = ρZ † + C , (4.22)
Λ−j 2b
with ρ := − and C := .
Λ+j a2 (Λ + j)
Assume now that we have a line L with slope Λ = Λ1 + jΛ2 = 0 and passing
through the point S0 = s01 + js02 with s01 and s02 in C(i). We know that L is of the
form LZ0 + S0 . Note also that if Z = z1 + jz2 ∈ L, then Z − S0 ∈ LZ0 ; since

Z − S0 = (z1 − s01 ) + j(z2 − s02 )

one has that the slope can be written as

z2 − s02
Λ= ,
z1 − s01

that is, in this case the equation of L is

z2 − s02 = Λ(z1 − s01 ) . (4.23)

Using again (4.21) and the recently defined ρ, the previous equation can be written
in terms of Z and Z † :
Z − S0 = ρ(Z † − S0† ) . (4.24)
Here one notes that if L passes through the origin, one may take S0 = 0, and
any non-zero Z ∈ L satisfies then

Z Λ−j
†
=ρ=− .
Z Λ+j

Following what one does for real lines in R2 , we would like now to find the
equation of a complex line passing through two given points. We need first a few
results.
88 Chapter 4. Lines and curves in BC

Proposition 4.1.7.3. A real two-dimensional plane L containing the origin (thus L

is an R-linear subspace of dimension 2) is a C(i)-complex line passing through the
origin if and only if Z ∈ L implies iZ ∈ L.
Proof. Assume first that L is a C(i)-complex line. Since 0 ∈ L, then L = LZ0 for
some Z0 ∈ BC \ {0}. Given S ∈ LZ0 , then S = λZ0 for some λ ∈ C(i), hence

iS = (iλ)Z0 ∈ LZ0 = L .

For the reciprocal statement, given any Z0 ∈ L, the set {Z0 , iZ0 } is a linearly
independent set in R4 , thus the R-linear subspace L{Z0 , iZ0 } generated by it
coincides with L. Thus L = LZ0 is a complex line. 
Proposition 4.1.7.4. If there are two points Z and W in a real two-dimensional
plane L containing the origin, such that the difference of their complex arguments
is neither zero nor π, then L is not a C(i)-complex line.
Proof. Follows directly from Proposition 4.1.7.1. 
Proposition 4.1.7.5. Given Z0 ∈ BC \ {0}, there exists a unique C(i)-complex line
L passing through Z0 and the origin.
Proof. The proof follows from the fact that L is a real two-dimensional plane
containing Z0 and iZ0 . 
Let us come back to the task of finding the equation of the complex line
passing through two given points S and W . Set Z0 := S − W and let LZ0 be the
unique C(i)-complex line that contains Z0 . If S = s01 + js02 and W = w10 + jw20 ,
s0 − w20
then Z0 = (s01 − w10 ) + j(s02 − w20 ), thus the slope of LZ0 is Λ = 20 . Having
s1 − w10
the slope and the point S that belongs to L, and using (4.23), the equation of L
is:
s0 − w20
z2 − s02 = 20 · (z1 − s02 ) . (4.25)
s1 − w10

4.2 Hyperbolic lines in BC

In this section we will follow the scheme developed for complex lines to introduce
hyperbolic lines in BC. This means that we are interested in solutions (z1 , z2 ) ∈ D2
of the equation
a1 z1 + a2 z2 = b, (4.26)
with a1 , a2 and b in D, but only in those solutions which form real two-dimensional
planes in BC; only such sets of solutions we will call hyperbolic lines. Thus, we have
to analyze equation (4.26) for different combinations of values of its coefficients.
But first we use the bicomplex language for rewriting (4.26). Writing z1 :=
x1 + ky2 and z2 := y1 + k(−x2 ), we can identify any solution (z1 , z2 ) ∈ D2 of (4.26)
4.2. Hyperbolic lines in BC 89

with the bicomplex number Z = x1 + iy1 + jx2 + ky2 = (x1 + ky2 ) + i(y1 − kx2 ) =
z1 + iz2 ∈ BC, thus we will say also that Z is a solution of (4.26).
Since Z ∗ = z1 − iz2 , hence

1 i
z1 = (Z + Z ∗ ) and z2 = (Z ∗ − Z)
2 2
and (4.26) becomes
A∗ Z + AZ ∗ = 2b, (4.27)
with the bicomplex coefficient A := a1 + ia2 .
Consider as an illustration some particular cases.
1) If a1 = 0, a2 = 1, b = 0, then z2 = 0 and z1 takes any hyperbolic value, that
is, the set of solutions of the equation

z2 = 0


is the set z1  z1 ∈ D = D.
2) If a1 = 1, a2 = 0, b = 0, we have the equation

z1 = 0,

which describes the set iD.

As it would be expected both the set D and the set iD are hyperbolic lines.
Recall that the (real) two-dimensional plane D is neither a C(i)-complex line nor
a C( j)-complex line.
In order to analyze (4.26) we write every hyperbolic number in its idempotent
form:

a1 = α11 · e + α21 · e† , a2 = α12 · e + α22 · e† , b = ν1 · e + ν2 · e† ,

z1 = η11 · e + η21 · e† , z2 = η12 · e + η22 · e† ,

where all the coefficients are real numbers, and we write

A = a1 + ia2 = (α11 e + α21 e† ) + i(α12 e + α22 e† )

= (α11 + iα12 )e + (α21 + iα22 )e†

and
Z = z1 + iz2 = (η11 + iη12 )e + (η21 + iη22 )e† .
We observe that the component (η11 + iη12 )e = η11 e + η12 ie can be seen as a point (or
vector) in the real plane BCe (which is a complex line) with coordinates (η11 , η12 );
the same comment applies to the other component of Z as well as to both com-
ponents of A.
90 Chapter 4. Lines and curves in BC

Hence, equation (4.26) in idempotent form becomes

(α11 η11 + α12 η12 )e + (α21 η21 + α22 η22 )e† = ν1 e + ν2 e† ,

which is equivalent to a couple (not a system) of independent real equations:


α11 η11 + α12 η12 = ν1 ,
(4.28)
α21 η21 + α22 η22 = ν2 .

We begin by excluding the situations in which we obtain, as the set of solu-

tions, the empty set or the whole BC:
(i) a1 = a2 = b = 0 (in this case equation (4.26) defines the whole BC), i.e.,
A = b = 0.
(ii) a1 = a2 = 0, b = 0 (in this case the set of solutions is the empty set), i.e.,
A = 0 and b = 0.
(iii) a1 = α11 e and a2 = α12 e with real coefficients α11 and α12 which are not
zero simultaneously, but b = ν1 e + ν2 e† , with ν2 = 0 (empty set), i.e.,
A = (α11 + iα12 )e and b ∈
/ BCe .
(iv) a1 = α21 e† and a2 = α22 e† again α21 and α22 real numbers not zero simulta-
neously, but b = ν1 e + ν2 e† , with ν1 = 0 (empty set), i.e., A = (α21 + iα22 )e†
and b ∈
/ BCe† .
In all these situations the solutions of (4.26) do not form a hyperbolic line.
There are more sets of solutions of (4.26) which do not form a real two-
dimensional plane; they arise if
(v) A and b are in BCe , i.e., A = (α11 + iα12 )e and b = ν1 e (note that α21 = α22 =
ν2 = 0).
(vi ) A and b are in BCe† , i.e., A = (α21 + iα22 )e† and b = ν2 e† (note that
α11 = α12 = ν1 = 0).
If condition (v) holds, then the second equation in (4.28) determines the
whole real two-dimensional plane BCe† and the first equation determines a real
line in the real plane BCe ; hence (4.28) determines the cartesian product of the
real line in BCe with the plane BCe† , that is, a three-dimensional set in BC. The
same applies to the case (vi). We do not call these sets hyperbolic lines.
Theorem 4.2.1. Equation (4.26) determines a hyperbolic line if and only if none
of the above six restrictions is valid.
Proof. The independent equations in (4.28) determine a two-dimensional set if
and only if each of them determines a one-dimensional set, which holds if and only
if none of the six conditions hold. 
4.2. Hyperbolic lines in BC 91

Corollary 4.2.2. A real two-dimensional plane P in R4 ∼ = BC is a hyperbolic line

if and only if its projections onto the complex lines BCe and BCe† are usual real
lines.
Remark 4.2.3. Let Γ1 be any real line in the plane BCe and let Γ2 be any real line
in the plane BCe† ; then Theorem 4.2.1 and Corollary 4.2.2 together say that the
set
Γ1 e + Γ2 e†
is a hyperbolic line in BC and that every hyperbolic line in BC is of this form.

4.2.1 Parametric representation of hyperbolic lines

As we did with complex lines, we will see now that any hyperbolic line that passes
through the origin can be obtained by taking “hyperbolic multiples” of a non-zero
bicomplex number. In other words, we will prove the following
Theorem 4.2.1.1. Given a bicomplex number Z0 , the set


PZ0 := μZ0  μ ∈ D

is a hyperbolic line passing through the origin if and only if Z0 ∈

/ S0 . More pre-
cisely, the set PZ0 is given by the solutions of the equation

a1 z1 + a2 z2 = 0, (4.29)

with A := a1 + ia2 = −iZ0 .

Proof. Take Z ∈ PZ0 and write Z = (u1 + iv1 )e + (u2 + iv2 )e† , then there exists
μ = m1 e + m2 e† ∈ D such that
Z = μZ0 . (4.30)
Indeed, write Z0 = β10 e+β20 e† = (u01 +iv10 )e+(u02 +iv20 )e† , then (4.30) is equivalent
to the pair of independent equations

m1 β10 = β1 , m1 ∈ R,
m2 β20 = β2 , m2 ∈ R,

or, equivalently, 
m1 (u01 + iv10 ) = u1 + iv1 , m1 ∈ R,
m2 (u02 + iv20 ) = u2 + iv2 , m2 ∈ R;
this means that the pairs (u1 , v1 ) and (u2 , v2 ) are solutions, respectively, of the
equations 
u01 u1 + (−u01 )v1 = 0,
(4.31)
u02 u2 + (−u02 )v2 = 0.
Hence Z belongs to PZ0 if and only if its components satisfy the equations (4.31).
92 Chapter 4. Lines and curves in BC

Comparing systems (4.31) and (4.28), we see that (4.31) determines a hyper-
bolic line if and only if it satisfies Corollary 4.2.2.
On the other hand the set of solutions of (4.31) admits descriptions in hy-
perbolic and bicomplex terms as in equations (4.26) and (4.27). To see this we
compute:
Z0 = z1 + iz2 = (u01 + iv10 )e + (u02 + iv20 )e†
= i(v10 + i(−u01 ))e + i(v20 + i(−u02 ))e†
= i(v10 e + v20 e† ) − (−u01 e − u02 e† )
=: ia1 − a2 = i(a1 + ia2 ) =: iA,
from where a1 = z02 and a2 = −z01 . Thus, the set PZ0 is the set of solutions of the
equation
z02 z1 − z01 z2 = 0.
Finally we know that such a set determines a hyperbolic line if and only if z02 and z01
satisfy the restrictions (i)–(vi) and this is equivalent to saying that Z0 ∈
/ S0 . 

4.2.2 More properties of hyperbolic lines

As it was in the case of complex lines, any hyperbolic line that passes through
W0 = 0 and that is parallel to PZ0 (i.e., a hyperbolic line with the direction
determined by Z0 ) is defined as the set


PZ0 + W0 := μZ0 + W0  μ ∈ D .
We leave it to the reader to show that any hyperbolic line can be described as
PZ0 + W0 for some bicomplex numbers Z0 , W0 .
We also leave it to the reader to prove the following results, which are anal-
ogous to those given in the case of complex lines.
Proposition 4.2.2.1. Given the hyperbolic line PZ0 and given U ∈ PZ0 such that
U = μ0 Z0 , with μ0 ∈ / S0 (hence U ∈ / S0 ), then PU = PZ0 . Reciprocally, if
PU = PZ0 , then U ∈ PZ0 .
Another characterization of real two-dimensional planes passing through the
origin and being a hyperbolic line is given in the following Proposition whose proof
is left to the reader.
Proposition 4.2.2.2. A real two-dimensional plane T ⊂ BC with 0 ∈ T is a hyper-
bolic line if and only if U ∈ T implies kU ∈ T .
Corollary 4.2.2.3. A hyperbolic line P through the origin, regarded as a two-
dimensional subspace of R4 ∼ = BC, is generated over R by any U0 ∈ P \ S0 and by
kU0 , i.e., it is the real span of U0 and kU0 .
Now, in analogy with complex lines, we describe the relation between the
hyperbolic arguments of the points of a hyperbolic line PZ0 .
4.2. Hyperbolic lines in BC 93

Theorem 4.2.2.4. Consider the hyperbolic line PZ0 for a point Z0 ∈ BC\S0 . Write
 0 0

Z0 = |Z0 |k eiν1 e + eiν2 e† ,

with ΨZ0 = ν10 e + ν20 e† = argD Z0 ∈ D the principal hyperbolic argument of Z0 .

Then the set ArgD U of any point U ∈ PZ0 \ S0 is one of the four sets:
    

ArgD U = ν10 + 2mπ e + ν20 + 2nπ e†  m, n ∈ Z ,

    

ArgD U = ν10 + 2mπ e + ν20 + (2n + 1)π e†  m, n ∈ Z ,

    

ArgD U = ν10 + (2m + 1)π e + ν20 + 2nπ e†  m, n ∈ Z ,

or
    

ArgD U = ν10 + (2m + 1)π e + ν20 + (2n + 1)π e†  m, n ∈ Z .

Proof. Since U ∈ PZ0 \ S0 , then U = μZ0 for some μ ∈ D \ S0 . Since the

trigonometric form of μ in hyperbolic terms is
 
μ = |μ|k ±e ± e† ,

the statement follows from the fact that

 0 0

U = μZ0 = |μZ0 |k ±eiν1 e ± eiν2 e† . 

We are now ready to define the slope of hyperbolic lines; Theorem 4.2.2.4
ensures that the slope will not depend on the bicomplex number Z0 that determines
its direction.
We now need to use a result about trigonometric functions of a hyperbolic
variable, a topic that is studied in detail in Chapter 6. As we will show there, the
tangent of a hyperbolic number ν = ν1 e + ν2 e† can be given a rigorous definition
and it turns out that
tan ν = tan ν1 e + tan ν2 e† .

Definition 4.2.2.5. The slope N of a hyperbolic line PZ0 + W0 , is defined by

N := tan ΨZ0 = tan ν10 e + tan ν20 e† ,

where ΨZ0 ∈ ArgD (Z0 ).

Example 4.2.2.6. (a) The hyperbolic line D is generated by any non zero-divisor
ae + be† whose hyperbolic argument is Ψae+be† = 0 = 0e + 0e† . Therefore the
slope of D is tan 0 = tan 0e + tan 0e† = 0. Thus, any hyperbolic line D + W0
“parallel” to D has the slope zero, and all of them together give the notion
of hyperbolic horizontality.
94 Chapter 4. Lines and curves in BC

(b) Each real line L in C(i) is exactly the intersection of a hyperbolic line P and
C(i) itself, and the slope of the hyperbolic line P coincides with the (usual
real) slope of L. Indeed, assume that 0 ∈ L, then any z ∈ L is of the form
 
z = |z| (cos θ + i sin θ) = |z|k eiθ e + eiθ e† ;

thus, the hyperbolic line Pz = P which passes through z has slope

N = tan θe + tan θe† = tan θ.

(c) Consider the hyperbolic line of the type Pie+β20 e† or Pβ10 e+ie† ; since the hy-
π π
perbolic arguments are Ψie+β20 e† = e + ν20 e† and Ψβ10 e+ie† = ν10 e + e† ,
2 2
then the slopes are not well defined but can be symbolically represented as

tan Ψie+β20 e† = ∞e + tan ν20 e† ,

tan Ψβ10 e+ie† = tan ν10 e + ∞e† .

We say in this case that these hyperbolic lines determine the notion of hy-
perbolic verticality. 
Recalling that a hyperbolic line is a real two-dimensional plane such that
its projections onto BCe and BCe† are usual real lines, we can give a precise
geometrical notion of the hyperbolic angle between two hyperbolic lines. Clearly
it is enough to define the hyperbolic angle between hyperbolic lines that pass
through the origin.
Definition 4.2.2.7. Take hyperbolic lines PZ0 and PW0 ; if argD Z0 = ν10 e + ν20 e†
and argD W0 = μ01 e + μ02 e† , then the (trigonometric) hyperbolic angle α between
PZ0 and PW0 is

α := argD Z0 − argD W0 = (ν10 − μ01 )e + (ν20 − μ02 )e† , (4.32)

and the (geometric) hyperbolic angle between PZ0 and PW0 is

|α|k := | argD Z0 − argD W0 |k

(4.33)
= |(ν10 − μ01 )e + (ν20 − μ02 )e† |k = |ν10 − μ01 |e + |ν20 − μ02 |e† .

In analogy with the complex plane, if the angle α is a positive hyperbolic

number, then we say that the angle between the lines is positively oriented; if the
angle α is a hyperbolic negative number, then we say that the angle between the
lines is negatively oriented; if the angle is neither negative nor positive, then no
orientation is assigned.
Denoting by L1Z0 and L1W0 the real lines in BCe that are the projections of
PZ0 and PW0 on BCe and by L2Z0 , L2W0 the respective projections on BCe† we
are not able to say which of PZ0 or PW0 is the initial or the final hyperbolic line
4.3. Hyperbolic and Complex Curves in BC 95

that determines the orientation of the hyperbolic angle. The reason for this is
that in the projections on BCe and BCe† the roles of L1Z0 , L1W0 and L2Z0 , L2W0
do not coincide in general; just recall that D is not a totally ordered set but a
partially ordered one. That is why we use not the difference of the arguments but
its hyperbolic modulus.

4.3 Hyperbolic and Complex Curves in BC

4.3.1 Hyperbolic curves
We have described, quite extensively, the properties of complex and hyperbolic
lines. Although their properties are rather similar, the hyperbolic lines have, in
our opinion, some clear advantages from the point of view of geometrical interpre-
tation and visualization; for example, the (hyperbolic-valued) angle between two
hyperbolic lines has been defined, and the definition is not only formal but it has
a precise geometrical description.
For this reason we will start with the study of hyperbolic curves which will be
followed by the study of complex ones. Before giving their definition, let us point
out a motivation why among all two-dimensional surfaces in R4 ∼ =R BC only very
particular surfaces will be called hyperbolic curves. First of all, they have to be
smooth two-dimensional surfaces; then the tangent (real two-dimensional) plane
has to exist at any point and we will require that it should be a hyperbolic line!
Since we know that a real two-dimensional plane is a hyperbolic line if and only
if its projections onto the complex lines BCe and BCe† are usual real lines (see
Corollary 4.2.2), this is the key idea that we will use in order to define hyperbolic
curves.

Definition 4.3.1.1. Let Γ be a two-dimensional surface with parametrization

φ : I := [a, b] × [c, d] ⊂ R2 → BC;

we say that Γ is a hyperbolic curve if the idempotent representation of φ is of the

form
φ(u, v) = φ1 (u)e + φ2 (v)e†

with φ1 : [a, b] → C(i) and φ2 : [c, d] → C(i) being parametrizations of usual curves
in C(i).

The reader may note that a hyperbolic curve is a two-dimensional mani-

fold in R4 of the simplest type. Note also that if γ1 and γ2 are two curves with
parametrizations φ1 and φ2 , then the hyperbolic curve Γ can be written as

Γ = γ 1 e + γ 2 e† .
96 Chapter 4. Lines and curves in BC

γ 1 e + γ 2 e†

γ1 BCe
γ2

BCe†

Figure 4.3.1: A hyperbolic curve.

If additionally γ1 and γ2 are two smooth arcs, then Γ is a smooth two-manifold

without self-intersections and with boundary
   
∂Γ = φ1 (a)e + γ2 e† ∪ φ1 (b)e + γ2 e†
   
∪ γ1 e + φ2 (c)e† ∪ γ1 e + φ2 (d)e† .
If γ1 and γ2 are closed Jordan curves, then the hyperbolic curve Γ is a man-
ifold without boundary. In this case we say that Γ is a closed hyperbolic curve.
As an example, recall the definition of the bicomplex sphere of hyperbolic ra-
dius ν1 e + ν2 e† ∈ D+ ; it implies obviously that such a sphere is a closed hyperbolic
curve.
If γ1 and γ2 are piece-wise smooth curves with parametrizations
φ1 : [a = a1 , a2 ) ∪ (a2 , a3 ) ∪ · · · ∪ (as−1 , as = b] −→ C(i)
and
φ2 : [c = c1 , c2 ) ∪ (c2 , c3 ) ∪ · · · ∪ (ct−1 , ct = d] −→ C(i),
then Γ is called a piece-wise smooth hyperbolic curve. The smoothness is lost on the
real one-dimensional curves φ1 (a )e+γ2 e† and γ1 e+φ2 (cr )e† with  ∈ {1, 2, . . . , s}
4.3. Hyperbolic and Complex Curves in BC 97

and r ∈ {1, 2, . . . t}; note that their union is a set of measure zero (as a subset of
the manifold Γ).
Of course, one may define more exotic types of hyperbolic curves eliminating
other restrictions that we imposed on γ1 and γ2 , but for our purposes the definition
given above is sufficient.

4.3.2 Hyperbolic tangent lines to a hyperbolic curve

We will see now that if Γ is a smooth hyperbolic curve (with boundary or closed),
then for each point Z0 ∈ Γ \ ∂Γ there exists the hyperbolic tangent line T to Γ.
Indeed, by definition of hyperbolic curve, Γ has a parametrization φ such
that any point of Γ is of the form (see Definition 4.3.1.1):
Γ Z = φ(u, v) = φ1 (u)e + φ2 (v)e†
with φ1 and φ2 being parametrizations of smooth curves γ1 and γ2 in C(i). Fix
a point Z0 ∈ Γ, then Z0 = φ1 (u0 )e + φ2 (v0 )e† , with u0 ∈ [a, b] and v0 ∈ [c, d]
such that φ1 (u0 ) = 0 and φ2 (v0 ) = 0. Hence, there exist the tangent line T1
to γ1 at φ1 (u0 ) and the tangent line T2 to γ2 at φ2 (v0 ), which means that the
set T := T1 e + T2 e† is a real two-dimensional plane tangent to Γ at Z0 and by
Corollary 4.2.2, T is a hyperbolic line. See figure 4.3.2.
The above reasoning is summarized as
Theorem 4.3.2.1. Let Γ be a smooth hyperbolic curve, then for each point Z0 which
is not on the boundary of Γ the tangent plane T to Γ at Z0 is a hyperbolic line.

4.3.3 Hyperbolic angle between hyperbolic curves

Let Γ and Λ be two smooth hyperbolic curves and let Z0 ∈ Γ ∩ Λ, Z0 ∈ ∂Γ ∪ ∂Λ;
let T be the hyperbolic tangent line to Γ at Z0 and let S be the hyperbolic tangent
line to Λ at Z0 .
Definition 4.3.3.1. The hyperbolic angle between the hyperbolic tangent lines T and
S at the point Z0 is called the hyperbolic angle between the hyperbolic curves Γ and
Λ at Z0 .
Figure 4.3.3 illustrates the hyperbolic angle between the hyperbolic curves Γ
and Λ that intersect at Z0 = β10 e + β20 e† . Since Γ and Λ are smooth hyperbolic
curves, they have the projections γ1 and λ1 on BCe and the projections γ2 and λ2
on BCe† . Similarly, since T and S are hyperbolic tangent lines, their projections
on BCe are T1 and S1 and their projections on BCe† are T2 and S2 . Thus, the
hyperbolic angle between the curves Γ and Λ is the hyperbolic angle between their
tangents: θ1 e + θ2 e† .
It is possible to introduce the oriented hyperbolic angle between two hyper-
bolic curves; since a hyperbolic angle is, in a sense, a pair of planar angles, then
there are four options for choosing an orientation of the hyperbolic angle. We will
not elaborate on this here.
98 Chapter 4. Lines and curves in BC

γ 0
β 2 e† S
T

0
β1 e

0 0
Z 0 = β 1 e + β 2 e†

W = T e + Se†

Figure 4.3.2: The hyperbolic tangent line to a hyperbolic curve.

4.3.4 Complex curves

While working with the notion of (straight) lines, we introduced the three types
of them: hyperbolic, C(i)-complex and C( j)-complex lines. Since a generalization
of the first has been given already with hyperbolic curves, then the next step is
to introduce the definition of (C(i)-) complex curves, and that is what we will
develop in this section. Analogously, as it was made with hyperbolic curves, the
definition of a complex curve will be linked with complex lines.
Definition 4.3.4.1. A smooth complex curve M is a smooth two-dimensional surface
in R4 such that at any point Z0 ∈ M the real two-dimensional tangent plane to
M is a complex line.
We admit that this definition differs significantly from its hyperbolic coun-
terpart; in particular, it does not give an intrinsic description of the object and it
4.3. Hyperbolic and Complex Curves in BC 99

BCe S2
T2
λ1
S1 T1
θ1 θ2
γ1
β10 β20

BC†e
γ2

λ2
Figure 4.3.3: The hyperbolic angle between hyperbolic curves.

is not easy to provide immediately an example of a complex curve. Thus, let us

provide a sufficient condition that ensures that M is a complex curve.
Theorem 4.3.4.2. Let Γ ⊂ R4 = C2 (i) be a real two-dimensional surface with a
parametrization ϕ : Ω ⊂ R2 −→ BC,

ϕ(u, v) = ψ1 (u, v) + jψ2 (u, v) := (ϕ1 (u, v) + iϕ2 (u, v)) + j (ϕ3 (u, v) + iϕ4 (u, v)) ,

such that ψ1 and ψ2 are holomorphic mappings. Then Γ is a complex curve.

Proof. Take Z0 ∈ Γ, then the real two-dimensional plane T tangent to Γ at Z0 is
of the form
 
∂ϕ ∂ϕ 
T = Z0 + T1 := Z0 + a (Z0 ) + b (Z0 )  a, b ∈ R
∂u ∂v

with a plane T1 passing through the origin. Thus, any point W ∈ T can be written
as

∂ϕ1 ∂ϕ1 ∂ϕ2 ∂ϕ2
W = Z0 + a (Z0 ) + b (Z0 ), a (Z0 ) + b (Z0 ),
∂u ∂v ∂u ∂v
(4.34)
∂ϕ3 ∂ϕ3 ∂ϕ4 ∂ϕ4
a (Z0 ) + b (Z0 ), a (Z0 ) + b (Z0 ) .
∂u ∂v ∂u ∂v

Recall (see Section 4.1.3) that T is a complex line if and only if T1 is a complex
line passing through the origin and that T1 is a complex line if and only if it is
invariant under the multiplication by i, that is, T1 as a subspace of dimension two
100 Chapter 4. Lines and curves in BC

in R4 is invariant under the action of the operator Mi of multiplication by i whose

matrix is ⎛ ⎞
0 −1 0 0
⎜ ⎟
⎜ 1 0 0 0 ⎟
⎜ ⎟
Mi = ⎜ ⎟.
⎜ 0 0 0 −1 ⎟
⎝ ⎠
0 0 1 0

Here the identification BC  Z = x1 + iy1 + jx2 + ky2 ←→ (x1 , y1 , x2 , y2 ) ∈ R4 is

used, under which i ∈ BC is identified with (0, 1, 0, 0) ∈ R4 . Hence
⎛ ⎞ ⎛ ⎞
x1 −y1
⎜ ⎟ ⎜ ⎟
⎜ y1 ⎟ ⎜ x1 ⎟
⎜ ⎟ ⎜ ⎟
Mi ⎜ ⎟=⎜ ⎟ (4.35)
⎜ x2 ⎟ ⎜ −y2 ⎟
⎝ ⎠ ⎝ ⎠
y2 x2

and iZ = −y1 + ix1 − jy2 + kx2 , i.e., the vector in (4.35) corresponds to iZ.
Thus, applying Mi to a point w1 ∈ T1 we get:

∂ϕ2 ∂ϕ2 ∂ϕ1 ∂ϕ1
Mi w 1 = −a (Z0 ) − b (Z0 ), a (Z0 ) + b (Z0 ),
∂u ∂v ∂u ∂v
(4.36)
∂ϕ4 ∂ϕ4 ∂ϕ3 ∂ϕ3
−a (Z0 ) − b (Z0 ), a (Z0 ) + b (Z0 ) .
∂u ∂v ∂u ∂v

Since ψ1 and ψ2 are holomorphic mappings, then the Cauchy–Riemann equations

hold:
⎧ ⎧
⎪ ∂ϕ1 (Z0 ) = ∂ϕ2 (Z0 )
⎨ ⎪
⎨
∂ϕ3
(Z0 ) =
∂ϕ4
(Z0 )
∂u ∂v and ∂u ∂v
⎪
⎩ ∂ϕ1 ∂ϕ2 ⎪
⎩ ∂ϕ3 (Z ) = − ∂ϕ4 (Z ),
(Z0 ) = − (Z0 ) 0 0
∂v ∂u ∂v ∂u
implying that Mi w1 becomes:

∂ϕ1 ∂ϕ1 ∂ϕ2 ∂ϕ2
Mi w 1 = a (Z0 ) − b (Z0 ), a (Z0 ) − b (Z0 ),
∂v ∂u ∂v ∂u

∂ϕ3 ∂ϕ3 ∂ϕ4 ∂ϕ4
a (Z0 ) − b (Z0 ), a (Z0 ) − b (Z0 )
∂v ∂u ∂v ∂u
∂ϕ ∂ϕ
= −b (Z0 ) + a (Z0 ) ∈ T1 .
∂u ∂v
Hence, T1 is invariant under multiplication by i and T is a complex line. Since Z0
was an arbitrary point of Γ, then Γ is a complex curve. 
4.4. Bicomplex spheres and balls of hyperbolic radius 101

4.4 Bicomplex spheres and balls of hyperbolic radius

We have endowed the BC-module BC with the hyperbolic-valued norm | · |k ; thus
we can apply usefully the geometric language of normed spaces although normed
spaces with hyperbolic-valued norm have practically no developed theory; such a
norm was first introduced in [2] which covers all that is actually known about it.
Given a positive hyperbolic number γ0 = a0 e + b0 e† , recall that the set

Sγ0 := {Z ∈ BC | |Z|k = γ0 }

defines “the bicomplex sphere of hyperbolic radius γ0 centered at the origin”;

similarly, the set
Bγ0 := {z ∈ BC | |Z|k ≺ γ0 }
is the (open) “bicomplex ball of hyperbolic radius γ0 centered at the origin”; it
is clear what is meant by a sphere Sγ0 ,Z0 and a ball Bγ0 ,Z0 centered at a point
Z0 ∈ BC.
As a matter of fact, we have considered already a few facts tightly related
with bicomplex spheres and balls. In Section 1.7 we have considered the equation
|z|k = R ≥ 0 in hyperbolic numbers. Such an equation defines a “hyperbolic
sphere of real radius R centered at the origin” which is the slice of the bicomplex
sphere BR by the hyperbolic plane D. It was shown that the intersection consists
of exactly four points (for R ≥ 0), and it is instructive to compare this fact with
the sphere of radius R > 0 in R: it consists of exactly two points which are the
intersection of the circumference of the same radius in C with the real axis.
In the same Section 1.7 we have considered the equation |z|k = w in hyper-
bolic numbers, with w in D+ . Thus, there we dealt with a “hyperbolic sphere of
hyperbolic radius w” and it is the intersection of the bicomplex sphere Bw and
of the hyperbolic plane D. The structure of this hyperbolic sphere depends on w:
if w is a semi-positive hyperbolic number, i.e., w is a positive zero-divisor, then
the hyperbolic sphere consists of two points which are also zero-divisors although
one of them is positive and the other is negative; if w is strictly positive, then the
hyperbolic sphere consists of four points.
Finally, in Section 2.6 we have considered the inequality |z|k w in hyper-
bolic numbers with w in D+ which is, again, about the intersection of Bw ∪ Sw
and the hyperbolic plane D, or, equivalently, about the hyperbolic ball of radius w
(that is, the solutions of the inequality |z|k ≺ w) together with the corresponding
hyperbolic sphere. It has been shown that this ball coincides with the hyperbolic
interval [−w, w]D := {w ∈ D | −w z w}, and one may compare again with R
where a “ball” is an interval which can be obtained by intersecting a “complex
ball” in C (which is a disk) with the line R.
Let us come back to the general situation, that is, let us consider the struc-
tures of arbitrary bicomplex spheres and balls.
Recall that, when the radius of the sphere is a zero-divisor of the form
γ0 = a0 e, then the sphere Sa0 e degenerates to a circumference with center at the
102 Chapter 4. Lines and curves in BC

a0
origin and radius √ , contained in the complex line BCe . Similarly, if γ0 = b0 e† ,
2
b0
the sphere Sb0 e† is the circumference with center at the origin and radius √ ,
2
contained in the complex line BCe† . Finally, if γ0 is not a zero-divisor, recall that
the sphere Sa0 e+b0 e† is the surface of a torus:


Sa0 e+b0 e† = Z = β1 · e + β2 · e†  |β1 | = a0 , , |β2 | = b0 .

We are now ready to describe what a bicomplex ball is.

If γ0 is not a zero-divisor, i.e., a0 = 0 and b0 = 0, then

Bγ0 = Z = β1 e + β2 e† | |β1 | < a0 , |β2 | < b0 ,

i.e., if we are looking at BC as C2 (i) with the idempotent coordinates, then the
bicomplex ball Bγ0 is the bicomplex form of writing for the usual bidisk centered
at the origin and with bi-radius (a0 , b0 ).
If γ0 is a zero-divisor, then we cannot define the ball in the same way because
none of the inequalities |β1 | < 0 or |β2 | < 0 has solutions. So we define in this case
the ball Bγ0 to be one of the two disks: one is located in BCe with center at the
a0
origin and radius √ , and the other is located in BCe† with center at the origin
2
b0
and radius √ .
2
It is worth noting that for a bicomplex ball Bγ0 , with γ0 not a zero-divisor,
the respective bicomplex sphere Sγ0 is not its topological boundary but it is its
distinguished, or Shilov, boundary.

4.5 Multiplicative groups of bicomplex spheres

It is known that in the study of Euclidean spaces Rn , the cases of n = 2 and
n = 4 are peculiar for many reasons but in particular because the corresponding
unitary spheres S1 in R2 and S3 in R4 are multiplicative groups; this is thanks to
the complex numbers multiplication in R2 and the quaternionic multiplication in
R4 .
Let us consider some analogues of the above facts related to the bicomplex
multiplication and bicomplex spheres with a hyperbolic radius:

Sλ := Z = β1 e + β2 e† | |Z|k = λ

where λ ∈ D+ . The multiplicative property of the hyperbolic modulus

|Z · W |k = |Z|k · |W |k (4.37)

will be crucial for the reasoning below.

Obviously, S0 = {0}. Consider the bicomplex unitary sphere S1 ; clearly
4.5. Multiplicative groups of bicomplex spheres 103

1) 1 ∈ S1 ;
2) if Z ∈ S1 , then it is invertible and Z −1 ∈ S1 ;
moreover, formula (4.37) says that
3) if Z, W ∈ S1 , then Z · W ∈ S1 .
Thus, the bicomplex unitary sphere S1 is a multiplicative group, but this is
an exceptional case for bicomplex spheres having a non-zero-divisor as its radius,
and for other non-zero-divisor values of the parameter λ the sphere Sλ is not a
multiplicative group; the same as in R2 .
Recall what happens when λ is a positive zero-divisor, i.e., when λ = λ1 e ∈
+
D+ † +
e , with λ1 > 0, or when λ = λ2 e ∈ De† , with λ2 > 0. If λ ∈ De , then for the
†
points Z = β1 e + β2 e of Sλ we have that

|Z|k = |β1 | · e + |β2 | · e† = λ1 e,

that is,
|β1 | = λ1 , |β2 | = 0,
thus, the bicomplex sphere Sλ is

Sλ = {Z | Z = β1 e, |β1 | = λ1 > 0 },

and it can be seen as the circumference of radius λ1 centered at the origin which is
located in the real two-dimensional plane, spanned by e and ie (which is a complex
line in BC).
Take λ1 = 1 here, then Sλ becomes

Se = {Z | Z = β1 e, |β1 | = 1 } .

We are now in a situation very similar to that of Section 2.7.1, and we will use a
similar approach in order to make Se a multiplicative group. The set Se is endowed
with the multiplication  which is the restriction of the bicomplex multiplication
and which acts invariantly on Se ; then the element e is the neutral element for ,
1
every element β1 e in Se is -invertible and its -inverse is e ∈ Se .
β1
Thus, we conclude that (Se , ) is a multiplicative group which is not a sub-
group of the multiplicative group (BC \ S0 , ·). Of course, (Se , ) is isomorphic to
the multiplicative group of complex numbers of modulus one.
Note that if λ1 = 1, then Sλ1 e is not a multiplicative group. In other words,
among all the spheres Sλ1 e with λ1 > 0, only one, Se , is a multiplicative group.
The situation with λ = λ2 e† is similar.
In what follows we assume that

λ ∈ D+
inv , (4.38)
104 Chapter 4. Lines and curves in BC

that is, λ is a positive hyperbolic number; recall that D+

inv is a multiplicative group.
For any such λ the sphere Sλ enjoys the property that any Z in Sλ is invertible
and
Z ∈ Sλ if and only if Z −1 ∈ Sλ−1 .
Notice also that if Z ∈ Sλ and W ∈ Sμ , then by (4.37)

|Z · W |k = λ · μ ∈ D+
inv ,

hence
Z · W ∈ Sλμ .
These properties hint that we should consider all the spheres “together” and we
set: #
SD := Sλ
λ∈D+
inv

(i.e., SD is a disjoint union of spheres). The set SD is a multiplicative group with

respect to the bicomplex multiplication.
We discussed already some specific features of bicomplex spheres with real
radii, so consider the subset of SD defined by
#
SR := Sλ .
λ∈R+

A sphere Sλ with λ > 0 is characterized by the condition: if Z = β1 e + β2 e† ∈ Sλ ,

then
|β1 | = |β2 | = λ = |Z| = |Z|k .
It is easily seen that SR is a group also, and thus SR is a subgroup of SD . Since
S1 ⊂ SR and since S1 is a group, then S1 is a subgroup of both SR and SD .
The bicomplex spheres generate another set which can be endowed with the
structure of a multiplicative group. Indeed, introduce

" D := {Sλ }
S λ∈D+ inv

and define the multiplication “◦” of bicomplex spheres by the formula

Sλ ◦ Sμ := Sλμ .

The unitary sphere S1 will serve as the identity in S " D (i.e., the multiplicative
neutral element), and the sphere Sλ−1 is the inverse of the sphere Sλ .
The subset
" R := {Sλ }
S λ∈R+

" D is also a group and, thus, a subgroup of the group S

of S "D.
4.5. Multiplicative groups of bicomplex spheres 105

Observe that the mapping

ϕ : λ ∈ D+ "
inv −→ Sλ ∈ SD

is a group isomorphism, and its restriction ϕ|R+ is a group isomorphism as well:

"R.
ϕ|R+ : λ ∈ R+ −→ Sλ ∈ S

The sets SD and S" D have proved to be multiplicative groups since the zero-divisors
were forbidden to be radii of the spheres. Let us return to these exceptions and
let us consider the sets
# #
Se,D := Sλ ; Se† ,D := Sλ .
λ∈D+
e λ∈D+†
e

+
Note that the sets D+e and De† have become multiplicative groups when a specific
multiplication was introduced in each of them. This allows us, again, to make Se,D
and Se† ,D multiplicative groups following the pattern of Section 2.7.1. If Z ∈ Sλ
and W ∈ Sμ , i.e., λ = λ1 e and μ = μ1 e, Z = β1 e with |β1 | = λ1 > 0 and W = γ1 e
with |γ1 | = μ1 > 0; then Z  W := β1 γ1 e ∈ Sλ1 μ1 e = Sλμ . Hence, with this new
multiplication the set Se,D becomes a multiplicative group. The same for Se† ,D .
The analogues of S " D are the sets

S"e,D := {Sλ }λ∈D+

e
and S"e† ,D := {Sλ }λ∈D+ .
e†

The multiplication  of bicomplex spheres in S"e,D is defined by the formula

Sλ  Sμ := Sλμ .

Similarly for S"e† ,D . The reader is invited to complete the proofs.

Most of the material covered in this Chapter is original and does not have
counterparts in the existing literature. Our detailed study of complex and hyper-
bolic lines and curves in BC is of great importance, as mentioned at the end of
Chapter 3, e.g. for applications in the mathematics of Special Relativity, [11].
Bicomplex spheres and balls have also been studied in [49, 56] from an Eu-
clidean point of view, and, from a manifold theory point of view, in [7]. We stress
once more that our approach to this subject involves the hyperbolic norm | · |k , a
notion that was first developed in [2]. We strongly believe that a comprehensive
study of the applications of this norm in the context of Special Relativity is well
worth pursuing in the future, a realm of study that is beyond the scope of this
book.
Chapter 5

Limits and Continuity

The notion of limit for complex functions is well known and we will not redis-
cuss it here. Note that the formal proofs of its properties depend strongly on the
properties of the modulus of a complex number;
 
1
|ab| = |a| · |b| , |a + b| ≤ |a| + |b| ,   = 1 for a = 0 . (5.1)
 a  |a|

In particular, the existence of the limit of a function is equivalent to the

existence of the limits of its real and imaginary parts; moreover, the limit of a
function exists if and only if the limit of its conjugate function exists.
Our aim in this chapter is to extend the above to bicomplex functions show-
ing that there exist many similarities but differences as well. Indeed, instead of
properties (5.1) for the Euclidean modulus one has their analogs of the form
√
|Z · W | ≤ 2|Z| · |W |, |Z + W | ≤ |Z| + |W | , (5.2)

and it turns out that they allow us to repeat most of the proofs almost literally,
although one needs to take into account the fact that |Z −1 | is not always equal to
1
.
|Z|

5.1 Bicomplex sequences

Definition 5.1.1. A sequence of bicomplex numbers {Zn }n∈N is called convergent
if there exists Z0 ∈ BC such that for any  > 0 there exists N ∈ N such that for
any n ≥ N there holds:
|Zn − Z0 | <  .
In this case we say that Z0 is the limit of the sequence which we write as

lim Zn = Z0 ,
n→∞

© Springer International Publishing Switzerland 2015 107

M.E. Luna-Elizarrarás et al., Bicomplex Holomorphic Functions, F rontiers in Mathematics,
DOI 10.1007/978-3-319-24868-4_6
108 Chapter 5. Limits and Continuity

and we say that the sequence {Zn }n∈N converges to Z0 .

Recalling the different forms of writing the bicomplex numbers (1.3)-(1.9),

including the idempotent representations (1.24), we have that

|Z|2 = |z1 |2 + |z2 |2 = |ζ1 |2 + |ζ2 |2

= |z1 |2 + |z2 |2 = |w1 |2 + |w2 |2
= |w1 |2 + |w2 |2 = |ω1 |2 + |ω2 |2
= x21 + y12 + x22 + y22
1  1 
= |a1 |2 + |a2 |2 = |α1 |2 + |α2 |2 ,
2 2

which gives directly that a sequence {Zn }n∈N converges to Z0 if and only if the
corresponding coordinate sequences converge to the respective components of Z0 .
For instance, writing Zn = z1n + jz2n = a1n e + a2n e† we conclude that {Zn }n∈N
converges to Z0 = z10 + jz20 = a10 e + a2o e† if and only if {z1n }n∈N converges to
z10 and {z2n }n∈N converges to z20 , and if and only if {a1n }n∈N converges to a10
and {a2n }n∈N converges to a20 . The same happens for the other representations.
Consider now an arbitrary bicomplex convergent sequence Zn → Z0 ; if Zn =
a1n e + a2n e† , Z0 = a10 e + a20 e† and assuming that Z0 is not a zero-divisor, then
a10 = 0, a20 = 0; since a1n → a10 and a2n → a20 and these are complex sequences,
then there exists n0 ∈ N such that a1n = 0 and a2n = 0 for all n ≥ n0 . In other
words starting from n = n0 our sequence contains only invertible terms.
For the sum, product and quotient of two bicomplex sequences the usual
statements hold, and we comment on the cases of the product and of the quotient.
Let limn→∞ Zn = Z0 and limn→∞ Wn = W0 . For any n ∈ N there holds:

|Zn · Wn − Z0 · W0 | = |Wn (Zn − Z0 ) + Z0 (Wn − W0 )|

≤ |Wn (Zn − Z0 )| + |Z0 (Wn − W0 )|
√
≤ 2 (|Wn | · |Zn − Z0 | + |Z0 | · |Wn − W0 |) ,

where we have used properties (5.2). Take M > 0 such that |Zn | < M and
|Wn | < M for all n ∈ N. Now given an arbitrary  > 0 there exists N ∈ N
such that simultaneously
 
|Zn − Z0 | < √ and |Wn − Z0 | < √ , ∀n ≥ N .
2 2M 2 2M

We have just proved that the sequence {Zn · Wn }n∈N converges to Z0 · W0 .

We proceed now to the case of a quotient using here the idempotent rep-
resentations. Let {Zn }n∈N and {Wn }n∈N be convergent sequences of bicomplex
Zn
numbers: lim Zn = Z0 , lim Wn = W0 ∈ / S0 ; then the quotients are well
n→∞ n→∞ Wn
5.1. Bicomplex sequences 109

defined for sufficiently large n. We write:

Zn = β1n e + β2n e† , Wn = γ1n e + γ2n e† , (5.3)
Z0 = β10 e + β20 e† , W0 = γ10 e + γ20 e† , (5.4)
then
Zn β1n β2n †
= e+ e .
Wn γ1n γ2n
By hypothesis, the following limits exist:
lim β1n = β10 , lim β2n = β20 , lim γ1n = γ10 , lim γ2n = γ20 .
n→∞ n→∞ n→∞ n→∞
   
β1n β2n
What is more, γ10 = 0, γ20 = 0, hence the sequences and
γ1n γ2n

n∈N  n∈N
β10 β20 Zn
are well defined and converge to and respectively. Thus, is
γ10 γ20 Wn n∈N
convergent with its limit being
β10 β20 † β10 e + β20 e† Z0
e+ e = †
= .
γ10 γ20 γ10 e + γ20 e W0
Recalling the quite “good” properties of the hyperbolic modulus, we are in
position to introduce another approach to the notion of convergence of bicomplex
sequences.
Definition 5.1.2. A sequence {Zn }n∈N of bicomplex numbers D-converges (syno-
nymous: hyperbolically converges; converges with respect to the hyperbolic-valued
norm |·|k ) to the bicomplex number Z0 if for any strictly positive hyperbolic number
ε there exists N ∈ N such that for any n ≥ N there holds:
|Zn − Z0 |k ≺ ε.
Using the idempotent representations
Zn = β1n · e + β2n · e† ; Z0 = β10 · e + β20 · e† ;
ε = ε 1 · e + ε 2 · e† ,
we obtain that, equivalently,
|β1n − β10 | < ε1 and |β2n − β20 | < ε2 ;
which means that the sequence {Zn }n∈N converges to the bicomplex number Z0
with respect to the hyperbolic-valued norm if and only if it converges to Z0 with
respect to the Euclidean norm. Notice that even though the two norms cannot be
compared, as they take values in different rings, one still obtains the same sets of
convergent and divergent sequences.
Thus we will usually write “a convergent sequence” without specifying which
type of convergence is meant.
110 Chapter 5. Limits and Continuity

5.2 The Euclidean topology on BC

Since we have already endowed BC with the Euclidean norm, which is associated
with the identifications BC = R4 = C2 (i) = C2 ( j), we will consider the topological
space (BC, τeuc ) where τeuc is the Euclidean topology on R4 : its basis consists of
all open balls in R4 . Since for any bicomplex numbers Z and W there holds:
|Z + W | ≤ |Z| + |W |;
√
|Z · W | ≤ 2 |Z| · |W |,
then the operations of addition and of multiplication are continuous with respect
to τeuc , and we can speak about the respective linearity of the topological space
(BC, τeuc ); more exactly, it is a real, a C(i)-complex and a C( j)-complex linear
topological space, but also a hyperbolic and a bicomplex linear topological module.
Besides the Euclidean open balls we can consider also open bicomplex balls
with non-zero-divisor radius:

Z | |Z|k ≺ γ with γ ∈ D+ inv .

Geometrically, such a ball can be seen as a bidisk in C2 , thus all such balls form
another basis in the topology τeuc .
In Section 5.1 we introduced two formally different types of convergent se-
quences. Now we realize they are the same convergence with respect to the Eu-
clidean topology but one of the definitions deals with the Euclidean basis of it
and the other deals with the bicomplex balls with non-zero-divisor radius. Thus,
in accordance with the problem we are faced, we will use one or another basis of
the topology, depending on which one of them is more appropriate.

5.3 Bicomplex functions

Given a set Ω ⊂ BC, any mapping F : Ω → BC will be called a bicomplex function
of the bicomplex variable Z ∈ Ω. Since both Z and F (Z) are bicomplex numbers,
each of them admits any of the representations (see (1.3)–(1.9) and (1.24)), then
F can be interpreted in different ways: as a mapping from C2 (i) to C2 (i), from
C2 (j) to C2 (j), from C2 (i) generated by the idempotent representation to itself,
etc. All these mappings are different but they all coincide when the bicomplex
structure is considered.
We illustrate the above with an example. Consider the bicomplex function
F (Z) = Z 2 and several mappings generated by it. To do this we write:
F (Z) = Z 2 = (z1 + jz2 )2 = (z12 − z22 ) + j(2z1 z2 )
= (γ1 + iγ2 )2 = (γ12 − γ22 ) + i(2γ1 γ2 )
= (β1 e + β2 e† )2 = β12 e + β22 e†
= (z1 + iz2 )2 = (z21 − z22 ) + i(2z1 z2 ) .
5.3. Bicomplex functions 111

This generates the following maps:

C2 (i)  (z1 , z2 ) → (z12 − z22 , 2z1 z2 ) ∈ C2 (i);

C2 (j)  (γ1 , γ2 ) → (γ12 − γ22 , 2γ1 γ2 ) ∈ C2 (j);
C2 (i)  (β1 , β2 ) → (β12 , β22 ) ∈ C2 (i);
D2  (z1 , z2 ) → (z21 − z22 , 2z1 z2 ) ∈ D2 .

Let Z0 be a point in the closure of Ω. The function F has the limit A at Z0

if for any  > 0 there exists δ > 0 such that the condition |Z − Z0 | < δ implies
that |F (Z) − A| < . As usual in metric spaces, it is equivalent to say that for
any sequence {Zn }n∈N ⊂ Ω such that lim Zn = Z0 , the sequence {F (Zn )}n∈N
n→∞
converges to A. We obtain immediately:
(I) if the limit lim F (Z) exists, it is unique;
Z→Z0

(II) if the limit lim F (Z) exists, then the function F is bounded in a Euclidean
Z→Z0
ball with center in Z0 and it is D-bounded in a bicomplex ball with a non-
zero-divisor radius;
(III) if lim F (Z) = A ∈
/ S0 , then there exists a ball B with center in Z0 such
Z→Z0
that for all Z ∈ B, F (Z) ∈
/ S0 ;
(IV) if lim F (Z) = A, lim G(Z) = B, then the sum, the product and the
Z→Z0 Z→Z0
quotient (if B ∈
/ S0 ) have limits at Z0 and the usual formulas hold.
A bicomplex function is continuous at a point Z0 ∈ Ω ⊂ BC, if lim F (Z)
Z→Z0
exists and
lim F (Z) = F (Z0 ) .
Z→Z0

Then we say that a bicomplex function F : Ω → BC, where Ω ⊂ BC, is continuous

on Ω if and only if F is continuous at every Z0 ∈ Ω.
As in the complex case, it is easy to prove that if two functions are contin-
uous at a point, then their sum and product are also continuous at that point.
Moreover, if the second function takes at Z0 an invertible value, then the quo-
tient is continuous at this point also. Furthermore, the composition of continuous
functions is continuous.
Bicomplex sequences and notions of convergence, limits and continuity, have
also been studied in works such as [41, 45, 84]. Applications to dynamical systems,
e.g. Mandelbrot and Julia sets in the bicomplex setup, have been developed in
[29, 59, 60, 62, 91, 92, 93, 102].
Chapter 6

Elementary Bicomplex
Functions

Historically, a small collection of real functions was assigned the name of ele-
mentary functions: polynomials, rational functions, the exponential, trigonometric
functions; together with their inverses: the n-th root, logarithm, inverse trigono-
metric functions. Later, for their complex extensions the same name has been
preserved, although many other functions have emerged which could rightly be
called elementary as well.
The aim of this chapter is to show that the structure and the properties of
bicomplex numbers allow us to further extend to BC all those elementary functions
in such a way that the extensions keep having amazingly many properties and
features of their real and complex antecedents.

6.1 Polynomials of a bicomplex variable

6.1.1 Complex and real polynomials.
A complex polynomial is a function of the form
$
n
p(z) = ak z k
k=0

where the ak are complex numbers and where z is a complex variable. We assume
that the leading coefficient an = 0 so that the polynomial is said to have degree
n. In particular, a polynomial of degree zero is, by definition, a non-zero constant.
When all ak are real numbers, the polynomial p(z) is called a real polynomial of
one complex variable. Observe that p(z) is a real polynomial of a complex variable
if and only if
p(z) = p(z) (6.1)

© Springer International Publishing Switzerland 2015 113

M.E. Luna-Elizarrarás et al., Bicomplex Holomorphic Functions, F rontiers in Mathematics,
DOI 10.1007/978-3-319-24868-4_7
114 Chapter 6. Elementary Bicomplex Functions

for all z ∈ C. From this it follows that, for such a polynomial, if p(a) = 0, then
p(a) = 0; therefore either a is real or p has the conjugate pair of zeros a and a.
Rewriting (6.1) as p(z) = p(z), one concludes that the range p(C) of the real
polynomial p is symmetric with respect to the real axis.
One of the most remarkable properties of complex polynomials is captured by
Gauss’s Fundamental Theorem of Algebra which states that a complex polynomial
of degree n, n > 0, has exactly n zeros considering the multiplicities. The usual
proofs of this use methods of analysis or topology (it is not a result which follows
purely from the algebraic field property of the complex numbers). In particular,
we have then that if p(z) and q(z) are polynomials of degree not exceeding n and
if the equation
p(z) = q(z)
is satisfied at n + 1 distinct points, then p = q.
Notice that real polynomials of a complex variable obey the Fundamental
Theorem of Algebra, but the situation with real polynomials of a real variable is
totally different: such a polynomial of degree n > 0 might have any number of zeros
up to n. We will see what happens, in this sense, with bicomplex polynomials.

6.1.2 Bicomplex polynomials

Let
$
n
p(Z) = Ak Z k
k=0

be a bicomplex polynomial of degree n of a bicomplex variable Z. Let us write Z =

z1 + jz2 in its C(i)-idempotent representation: Z = β1 e + β2 e† with β1 := z1 − iz2
and β2 := z1 + jz2 . We write also the bicomplex coefficients as Ak = γk e + δk e† ,
for k = 0, . . . , n. Then Z k = β1k e + β2k e† and we rewrite our polynomial as

$
n
  $
n
 
p(Z) = γk β1k e + δk β2k e† =: φ(β1 )e + ψ(β2 )e† .
k=0 k=0

If we denote the set of distinct roots of φ and ψ by S1 and S2 , and if we

denote by S the set of distinct roots of the polynomial p, then

S = S1 e + S2 e† ,

so that the structure of the null-set of a bicomplex polynomial p(Z) of degree n is

fully described by the following three cases:
1. If both polynomials φ and ψ are of degree at least one, and if S1 = {β1,1 , . . . ,
β1,k } and S2 = {β2,1 , . . . , β2, }, then the set of distinct roots of p is given by

S = {Zs,t = β1,s e + β2,t e†  s = 1, . . . , k, t = 1, . . . , } .
6.1. Polynomials of a bicomplex variable 115

2. If φ is identically zero, then S1 = C and S2 = {β2,1 , . . . , β2, }, with  ≤ n.

Hence 
S = {Zt = λe + β2,t e†  λ ∈ C, t = 1, . . . , } .
Similarly, if ψ is identically zero, then S2 = C and S1 = {β1,1 , . . . , β1,k },
where k ≤ n. Then

S = {Zs = β1,s e + λe†  λ ∈ C, s = 1, . . . , k} .

3. If all the coefficients Ak with the exception of A0 = γ0 e + δ0 e† are complex

multiples of e (respectively of e† ), but δ0 = 0 (respectively γ0 = 0), then p
has no roots.
We now discuss a few examples, to give a flavor for computations in BC.
Example 6.1.2.1. First, consider the polynomial

1 i
p(Z) = +j Z 5 + (−(1 + 4i) + 2j(2 − i))Z 4
2 2
+ ((−11 + 6i) − j (12 + 11i)) Z 3
 
29 47
+ + 13i + j −13 + i Z2
2 2
 
13 13
+ − 17i + j 17 + i Z
2 2
 
11 11
− +i +j 1− i .
2 2

The corresponding C(i)-complex polynomials are:

φ(β1 ) = β15 − (3 + 8i)β14 + 2(−11 + 9i)β13 + 2(19 + 13i)β12

+ (13 − 34i)β1 − (11 + 2i) ,
ψ(β2 ) = β24 − 6iβ23 − 9β22 .

Their distinct roots are S1 = {i, 1 + 2i} and S2 = {0, 3i}. Then p has the following
four roots:
 
1 1 1 + 2i −2 + i 1 + 5i 1+i
S= i − j, 2i + j, +j , +j .
2 2 2 2 2 2
Example 6.1.2.2. Consider the polynomial

p(Z) = (1 + ji) Z 2 − (i − j) .

The associated complex polynomials are:

φ(β1 ) = 2(β12 − i), ψ(β2 ) ≡ 0.

116 Chapter 6. Elementary Bicomplex Functions

The null set of p is

 √ √  %
2 2 
†
S= ± +i e + λe λ ∈ C .
2 2

Example 6.1.2.3. Slightly adjusting the previous example, i.e., taking ψ(β2 ) ≡ 2,
we get the polynomial
p(Z) = (1 + ji) Z 2 + (1 − i) + j (1 − i) ,
which has no roots.
It is also important to note that a bicomplex polynomial may not have a
unique factorization into linear polynomials. For example, the polynomial p(Z) =
Z 3 − 1 has 9 zeroes. Indeed, the associated complex polynomials are
φ(β1 ) = β13 − 1, φ(β2 ) = β23 − 1.
The set of zeros of φ and ψ are, respectively:
 √ √ %
1 3 1 3
S1 = β1,1 = 1, β1,2 = − + i , β1,3 = − − i
2 2 2 2
 √ √ %
1 3 1 3
S2 = β2,1 = 1, β2,2 = − + i , β2,3 = − − i
2 2 2 2

Then the set of zeroes of p is



S = Zkl = β1,k e + β2, e†  k,  = 1 . . . 3 ,

and we have at least two distinct factorizations:
 √  √ 
3 1 3 1 3
Z − 1 = (Z − 1) Z + − i Z+ + i
2 2 2 2

and  √  √ 
3 1 3 1 3
Z − 1 = (Z − 1) Z + − j Z + +j .
2 2 2 2
It is therefore clear from what we have indicated that bicomplex polynomials
do not satisfy the Fundamental Theorem of Algebra in its original form. At the
same time, the following results are true and summarize the comments above.
Theorem 6.1.2.4 (Analogue of the Fundamental Theorem of Algebra for bicom-
$
n
plex polynomials). Consider a bicomplex polynomial p(Z) = Ak Z k . If all the
k=0
coefficients Ak with the exception of the free term A0 = γ0 e + δ0 e† are complex
multiples of e (respectively of e† ), but δ0 = 0 (respectively γ0 = 0), then p has no
roots. In all other cases, p has at least one root.
6.1. Polynomials of a bicomplex variable 117

Corollary 6.1.2.5. Assume that a bicomplex polynomial p of degree n ≥ 1 has at

least one root. Then:
1. If at least one of the coefficients Ak , for k = 1, . . . , n, is invertible, then p
has at most n2 distinct roots.
2. If all coefficients are complex multiples of e (respectively e† ), then p has
infinitely many roots.
Note that zeros of bicomplex polynomials were originally investigated in [56]
and [55].
Formula (6.1) has several analogues in the actual context. Let all the coeffi-
$n
cients of the polynomial p(Z) = Ak Z k be real numbers, then
k=0

p(Z) = p(Z) = p† (Z † ) = p∗ (Z ∗ ) (6.2)

for all Z ∈ BC. This implies that if p(Z0 ) = 0, then p(Z 0 ) = 0, p(Z0† ) = 0,
p(Z0∗ ) = 0; now if Z0 ∈ R in this case Z0 does not have “associated” roots of p. If
Z0 is not real, then the following situations arise:
• Z0 ∈ C(i) \ R, then Z 0 = Z0 is also a root of p.

• Z0 ∈ C( j) \ R, then Z0† = Z0 is also a root of p.

• Z0 ∈ D \ R, then Z 0 = Z0† = Z0 is also a root of p.

• Z0 does not belong to any of the above three sets, hence the polynomial p
has together with Z0 all the three conjugates Z 0 , Z0† , Z0∗ as its roots.
Equations (6.2) can be written as

p(Z) = p(Z); p(Z † ) = p(Z)† ; p(Z ∗ ) = p(Z)∗

meaning that the range p(BC) of the polynomial p with real coefficients possesses
all the three symmetries generated by the three conjugations.
$n
Assume now that the coefficients of the polynomial p(Z) = Ak Z k are in
k=0
C(i), then
p(Z) = p(Z † )† (6.3)

for all Z in BC. This implies that if Z0 is a root of p, p(Z0 ) = 0, then p(Z0† ) = 0; if
Z0 is a C(i)-complex number, then Z0 does not have “associated” roots of p. But
if Z0 is not in C(i), then Z0† is also a root of p. Of course, since (6.3) is equivalent
to p(Z † ) = p(Z)† , one may conclude that the range p(BC) of such a polynomial
has the symmetry determined by the †-conjugation.
118 Chapter 6. Elementary Bicomplex Functions

As the next step, we assume that the coefficients of the polynomial p are in
C( j), then
p(Z) = p(Z) (6.4)
for all Z in BC. This implies that if Z0 is a root of p, p(Z0 ) = 0, then p(Z 0 ) = 0;
again if Z0 is a C( j)-complex number, then Z0 does not have “associated” roots
of p. But if Z0 is not in C( j), then Z 0 is also a root of p. Of course, since (6.4)
is equivalent to p(Z) = p(Z), one may conclude that the range p(BC) of such a
polynomial has the symmetry determined by the bar-conjugation.
Finally, we consider the case of hyperbolic coefficients of the polynomial p.
Here
p(Z) = p(Z ∗ )∗ (6.5)
for all Z in BC. This implies that if Z0 is a root of p, p(Z0 ) = 0, then p(Z0∗ ) = 0; if
Z0 is a hyperbolic number, then Z0 does not have “associated” roots of p. But if
Z0 is not in D, then Z0∗ = Z0 is also a root of p. Of course, since (6.5) is equivalent
to p(Z)∗ = p(Z ∗ ), one may conclude that the range p(BC) of such a polynomial
has the symmetry determined by the ∗-conjugation.
Definition 6.1.2.6. We will use the name bicomplex rational functions for functions
of the form
p(Z)
q(Z)
with two bicomplex polynomials p and q.
Such a function is well-defined for those values of Z for which the polynomial
q(Z) = Bm Z m + Bm−1 Z m−1 + · · · + B1 Z + B0 takes values in BCinv := BC \ S0 .
If both polynomials are of degree one: p(Z) = A1 Z + A0 , q(Z) = B1 Z + B0 , then
we have a fractional linear transformation. If the coefficients A1 and B1 are both
invertible, then the fractional linear transform takes the form
C2
C1 +
Z + C3
with bicomplex C1 , C2 , C3 .

6.2 Exponential functions

6.2.1 The real and complex exponential functions
As Ahlfors writes in [1], if one approaches calculus exclusively from the point of
view of real numbers, there is no reason to expect any relationship between the
exponential function ex and the trigonometric functions cos x and sin x. Indeed,
these functions seem to be derived from completely different sources and with
different purposes in mind. The reader will notice, no doubt, a similarity between
the Taylor expansions of these functions, and if willing to use imaginary arguments,
6.2. Exponential functions 119

the reader will be able to derive Euler’s formula eix = cos x + i sin x as a formal
identity.
The exponential function can be derived in many different ways. If one recalls
n
1
that the Euler’s number e arises as the limit of the sequence 1 + , then one
 n
x n
is tempted to consider, for any real number x, the sequence 1 + and its limit
n
which brings us to the definition of the real exponential function as
 x n
exp(x) = ex := lim 1 + . (6.6)
n→∞ n
This approach has the advantage of using minimal mathematical tools; as a
matter of fact, only the properties of sequences and of their limits are necessary,
meanwhile the definitions as the sum of a convergent power series or as a solution
of a differential equation require much more elaborated techniques.
Defined by (6.6), the real exponential function preserves all the expected
properties; for example, if x ∈ N, then ex = e& · e'(· · · e). It has the property
x times

ex1 +x2 = ex1 · ex2 , (6.7)

thus, it is a homomorphism of the additive group of real numbers into the multi-
plicative group R \ {0}. Since the real exponential function is monotone, then it
realizes an isomorphism of these groups.
It turns out that definition (6.6) extends to the complex numbers in the sense
that for any z ∈ C
 z n
exp(z) = ez := lim 1 + . (6.8)
n→∞ n
This implies the famous Euler formula

ez = ex (cos y + i sin y)


which says, in particular that |ez | = ex and Arg ez = y + 2πk  k ∈ Z . The

analogue of (6.7) is
ez+w = ez ew .
Thus the complex exponential function is a homomorphism (but not an isomor-
phism) of the additive group of complex numbers into the multiplicative group
C \ {0}.

6.2.2 The bicomplex exponential function

In this section, we introduce the exponential function of a bicomplex variable
extending directly the ideas of the previous section.
120 Chapter 6. Elementary Bicomplex Functions

Theorem 6.2.2.1. Let Z be any bicomplex number. Then the sequence

 n
Z
Zn := 1+
n

is convergent, and if Z is written as Z = z1 + jz2 , then the limit is

ez1 (cos(z2 ) + j sin(z2 )) .

Proof. The computation below proves that the sequence is convergent. Set as
before Z = β1 e + β2 e† . Then
 n  n  n
Z β1 β2 β1 β2
1+ = 1 + e + e† = e + e † + e + e†
n n n n n
  n  n  n
β1 β2 β1 β2
= 1+ e+ 1+ e† = 1+ e+ 1+ e† .
n n n n

 Relying
n on  the fact that the corresponding sequences of complex numbers
n
β1 β2
1+ and 1 + are convergent to the complex exponentials eβ1 and
n n
eβ2 , respectively, we get that the limit of the right-hand side exists. What is more,
 n  n  n
Z β1 β2
lim 1+ = lim 1+ e+ 1+ e† = eβ1 e + eβ2 e†
n→∞ n n→∞ n n
1 β1 i
= (e + eβ2 ) + j (eβ1 − eβ2 )
2 2
1 z1 −iz2 z1 +iz2 i (6.9)
= (e +e ) + j (ez1 −iz2 − ez1 +iz2 )
2  2
1 −iz2 i
=e z1
(e + e ) + j (e−iz2 − eiz2 )
iz2
2 2
= ez1 (cos(z2 ) + j sin(z2 )) .

This concludes our proof. 

Clearly, the theorem justifies the following definition.

Definition 6.2.2.2. We set

 n
Z
eZ := lim 1+ .
n→∞ n

Hence, if Z is written as Z = z1 +jz2 , then we obtain the bicomplex Euler formula:

eZ = ez1 (cos(z2 ) + j sin(z2 )) .

6.2. Exponential functions 121

The reader may rewrite eZ in many other forms of writing of bicomplex

numbers. For instance, if Z = ζ1 + iζ2 with ζ1 and ζ2 in C( j), then

eZ = eζ1 (cos(ζ2 ) + i sin(ζ2 )) ,

which exemplifies once more the peculiarity of the different writings of bicomplex
numbers.
We pass now to the properties of this newly introduced bicomplex exponential
function.
• First we note that the bicomplex exponential is an extension to BC of the
complex exponential function: indeed, for Z = z1 + j0 ∈ C, we have that

eZ = ez1 (cos(0) + j sin(0)) = ez1 ,

which is the usual complex exponential function.

In the same way, the restriction of eZ onto C( j) coincides, obviously, with
the C( j)-complex exponential function.
The situation with the restriction onto D is more subtle; if Z = a + bk with
reals a and b, then

eZ = ea+bk = ea+bij = ea (cos(bi) + j sin(bi))

= ea (cosh(b) + ji sinh(b)) = ea (cosh(b) + k sinh(b)) ,

where cosh and sinh are the classical hyperbolic cosine and sine functions,
thus arriving at the definition of the hyperbolic (in the sense of hyperbolic
numbers) exponential function.
• Note that ez1 is the complex modulus of the bicomplex number eZ and z2 is
the complex argument of the same bicomplex number eZ . The reader may
find it instructive to compare this fact with what happens in the complex
case.
• For Z = 0 = 0e + 0e† , we have: e0 = 1e + 1e† = 1.
• For any bicomplex number Z, the exponential eZ is invertible. This is because

eZ = ez1 −iz2 e + ez1 +iz2 e†

and the exponential terms ez1 −iz2 and ez1 +iz2 are complex exponential func-
tions, so they are never zero. The inverse of eZ is

e−Z = e−(z1 −iz2 ) e + e−(z1 +iz2 ) e† = e−z1 (cos(z2 ) − j sin(z2 )) .

Thus, the range of the bicomplex exponential function does not contain either
the zero or any zero-divisors.
122 Chapter 6. Elementary Bicomplex Functions

• For e = 1 · e + 0 · e† , and e† = 0 · e + 1 · e† , we have:

  
e † 1 i i
e = e · e + 1 · e = e 2 cos + j sin
2 2
  
1 1 1
= e 2 cosh + ji sinh .
2 2
Similarly:
  
† 1 i i
ee = 1 · e + e · e† = e 2 cos − j sin
2 2
  
1 1 1 †
= e 2 cosh − ji sinh = (ee ) .
2 2
†
Notice that both numbers, ee and ee , are hyperbolic numbers. This is be-
cause so are the idempotents e and e† and because the restriction of the
bicomplex exponential is the hyperbolic exponential.
• Due to the commutativity of the multiplication in BC, we can show that for
any Z1 and Z2 in BC, the following formula holds:
eZ1 eZ2 = eZ1 +Z2 . (6.10)
Indeed, writing Z1 = z11 + jz12 and Z2 = z21 + jz22 we have:
eZ1 eZ2 = (ez11 (cos(z12 ) + j sin(z12 ))) (ez21 (cos(z22 ) + j sin(z22 )))
= ez11 ez21 ((cos(z12 ) cos(z22 ) − sin(z12 ) sin(z22 ))

+j(sin(z12 ) cos(z22 ) + sin(z22 ) cos(z12 )))

z11 +z21
=e (cos(z12 + z22 ) + j sin(z12 + z22 )) = eZ1 +Z2 .
This equality means that the exponential function is a homomorphism from
the additive group of bicomplex numbers into the multiplicative group BCinv
of invertible bicomplex numbers.
• In the case Z = 0 + jz2 , we have:
eZ = ejz2 = cos(z2 ) + j sin(z2 ).

• The complex formula eiπ + 1 = 0 remains valid for bicomplex numbers, but
it is complemented with its mirror image e jπ + 1 = 0.
• For any Z = β1 e + β2 e† ∈ BC and any invertible bicomplex number W =
γ1 e + γ2 e† , i.e., γ1 γ2 = 0, the equation eZ = W is equivalent to the system
eβ1 = γ1 and eβ2 = γ2 . Because γ1 γ2 = 0, it follows that there is always a
solution. Of course, this is the first step to talk about the bicomplex logarithm
which will be commented on below.
6.3. Trigonometric and hyperbolic functions of a bicomplex variable 123

• Recalling that the complex exponential function and the complex trigono-
metric functions are periodic, we obtain that
eZ = ez1 (cos(z2 ) + j sin(z2 ))
= ez1 +2πim (cos(z2 + 2πn) + j sin(z2 + 2πn))
= eZ+2π(mi+nj) ,
for integer numbers m and n. Thus the bicomplex exponential function is
periodic with bicomplex periods 2π(mi + nj). One can prove that these are
the only periods.

6.3 Trigonometric and hyperbolic functions of a

bicomplex variable
6.3.1 Complex Trigonometric Functions
The complex trigonometric functions cos(z) and sin(z) are defined in terms of the
complex exponential as
eiz + e−iz eiz − e−iz
cos(z) := , sin(z) := . (6.11)
2 2i
The hyperbolic functions of a complex variable, cosh(z) and sinh(z) are de-
fined in terms of the complex exponential as follows:
ez + e−z ez − e−z
cosh(z) := , sinh(z) := . (6.12)
2 2
All four of them are extensions of the respective functions of a real variable.
The inverse of the complex cosine function is obtained by solving the equation
eiz + e−iz
cos(z) = = w.
2
This is a quadratic equation in eiz with roots

eiz = w ± w2 − 1.
Because of the periodicity of the complex exponential this equation, with
unknown z, has a countable family of solutions and the formulas
 
arccos(w) := z = −i log(w ± w2 − 1) = ±i log(w + w2 − 1) (6.13)
show how to treat precisely the arccosine function of a complex variable. In a
similar fashion, the other inverse functions are dealt with.
We will now follow the same process to define bicomplex trigonometric func-
tions.
124 Chapter 6. Elementary Bicomplex Functions

6.3.2 Bicomplex Trigonometric Functions

Adding and subtracting the formulas ejz2 = cos(z2 )+j sin(z2 ) and e−jz2 = cos(z2 )−
j sin(z2 ), for any z2 ∈ C(i), we express the complex cosine and sine via the bicom-
plex exponential:

ejz2 + e−jz2
cos z2 = ,
2
ejz2 − e−jz2
sin z2 = .
2j
Thus we are in a position to introduce the bicomplex sine and cosine functions
which are direct extensions of their complex antecedents.
Definition 6.3.2.1. Let Z ∈ BC. We define the bicomplex cosine and sine functions
of a bicomplex variable as follows:

ejZ + e−jZ eiZ + e−iZ

cos Z := = ,
2 2
(6.14)
ejZ − e−jZ eiZ − e−iZ
sin Z := = .
2j 2i
Both are well defined since a direct computation gives that, indeed,

ejZ + e−jZ eiZ + e−iZ ejZ − e−jZ eiZ − e−iZ

= and = .
2 2 2j 2i

ekZ + e−kZ
Note that does not give the same cos Z but it gives the hy-
2
perbolic cosine of a bicomplex number. See Section 6.3.3 below.
Given Z = z1 + jz2 = β1 e + β2 e† ∈ BC, the properties of the bicomplex
exponential bring us immediately to the idempotent representation of cos Z and
sin Z:

cos Z = cos(β1 )e + cos(β2 )e† ,

sin Z = sin(β1 )e + sin(β)e† . (6.15)

In terms of the components of the cartesian representation, one gets:

cos Z = cos(z1 − iz2 )e + cos(z1 + iz2 )e† .

Since for a complex variable z the following formulas hold:

cosh(z) = cos(iz), sinh(z) = −i sin(iz),

we obtain that

cos Z = cosh(z2 ) cos(z1 ) − j sinh(z2 ) sin(z1 ).

6.3. Trigonometric and hyperbolic functions of a bicomplex variable 125

One may write Z in different forms to observe how the above formulas change.
We continue with a description of some basic properties of the bicomplex
trigonometric functions.

• Since the complex sine and cosine functions are periodic with principal period
2π, then taking Z = β1 e + β2 e† and setting Zk, = (β1 + 2kπ)e + (β2 + 2π)e†
for arbitrary integers k,  we have:

cos(Zk, ) = cos(Z), sin(Zk, ) = sin(Z).

Thus the real number (2π)e + (2π)e† = 2π remains the principal period of
both bicomplex sine and cosine functions.

• From (6.15), the equation cos Z = 0 is equivalent to the equations in complex

variables β1 and β2 :

cos(β1 ) = 0, cos(β2 ) = 0.
π π
The solutions are β1 = + kπ, and β2 = + π, for k,  ∈ Z. Note that β1
2 2
and β2 are never zero, so the bicomplex solutions Z to cos Z = 0 are always
hyperbolic invertible numbers. In the {1, j} basis, we get the general solution
to cos Z = 0 as
π
Z = z1 + jz2 = ((1 + k + ) + j i(k − )) , (6.16)
2
a set of hyperbolic numbers.

• Similarly, the equation sin Z = 0 is equivalent to

sin(β1 ) = 0, sin(β2 ) = 0.

The solutions are β1 = kπ, and β2 = π, for k, l ∈ Z. Note that there are
non-invertible solutions for sin Z = 0, e.g., for β1 = 0, i.e., k = 0, and β2 = 0.
In the {1, j} basis, we get the general solution for sin Z = 0 as
π
Z = z1 + jz2 = (k +  + j i(k − )) ,
2
again a set of hyperbolic numbers.

• Formulas (6.15) guarantee that the usual trigonometric identities are true,
e.g., the sums and differences of angle formulas, the double angle identities,
etc. For example:

sin2 Z + cos2 Z = (sin2 (β1 ) + cos2 (β1 ))e + (sin2 (β2 ) + cos2 (β2 ))e† = 1.
126 Chapter 6. Elementary Bicomplex Functions

• Taking the argument equal to the idempotents e and e† gives again funny
formulas. Indeed, if Z = e, i.e., β1 = 1 and β2 = 0, then
 
† cos(1) + 1 cos(1) − 1
cos e = cos(1)e + e = −j i ,
2 2

sin(1) sin(1)
sin e = sin(1)e = −j i .
2 2

Similarly, if Z = e† , i.e., β1 = 0 and β2 = 1, then

 
† † cos(1) + 1 cos(1) − 1
cos e = 1e + cos(1)e = +j i ,
2 2

sin(1) sin(1)
sin e† = sin(1)e† = +j i .
2 2

Notice that all four are hyperbolic numbers.

As in the complex case, the other bicomplex trigonometric functions are
defined in terms of the bicomplex sine and cosine functions. For example, the
tangent function is the following.
Definition 6.3.2.2. Let Z be in BC. We define the bicomplex tangent function of
a bicomplex variable:
sin Z
tan Z := (6.17)
cos Z
whenever cos Z is invertible, i.e., both complex numbers β1 and β2 are not equal
π
to plus integer multiples of π.
2
A direct computation yields:

ejZ − e−jZ
tan Z = −j = tan(β1 )e + tan(β2 )e† .
ejZ + e−jZ
In a similar fashion we have
Definition 6.3.2.3. Let Z be in BC. We define the bicomplex cotangent function
of a bicomplex variable:
cos Z
cot Z := (6.18)
sin Z
whenever sin Z is invertible, i.e., both complex numbers β1 and β2 are not integer
multiples of π.
A direct computation yields:

ejZ + e−jZ
cot Z = j = cot(β1 )e + cot(β2 )e† .
ejZ − e−jZ
6.3. Trigonometric and hyperbolic functions of a bicomplex variable 127

We have described the fundamentals of the theory of trigonometric bicomplex

functions. Of course, now the theory can be continued in many directions.
We mention now a curious fact. The real tangent function takes any real
value. It is easy to show that the complex tangent does not contain ±i in its
range. For the bicomplex tangent the excluded values are {±i, ±j}.

6.3.3 Hyperbolic functions of a bicomplex variable

We want to extend directly the notion of a hyperbolic function of a complex
variable onto a bicomplex variable. A direct way is clear.
Definition 6.3.3.1. Let Z in BC. Then we define the bicomplex hyperbolic sine and
cosine functions as

eZ + e−Z ekZ + e−kZ

cosh Z := = ,
2 2 (6.19)
eZ − e−Z ekZ − e−kZ
sinh Z := = .
2 2k
Again, both are well defined since a direct computation gives now that

eZ + e−Z ekZ + e−kZ eZ − e−Z ekZ − e−kZ

= and = .
2 2 2 2k
As above, in the idempotent representation Z = β1 e + β2 e† we get that

cosh Z = cosh(β1 )e + cosh(β2 )e† ,

sinh Z = sinh(β1 )e + sinh(β2 )e† . (6.20)

As in the previous section, these formulas would yield the usual properties analo-
gous to hyperbolic complex functions. For example,
   
cosh2 Z − sinh2 Z = cosh2 (β1 ) − sinh2 (β1 ) e + cosh2 (β2 ) − sinh2 (β2 ) e† = 1 .

The addition formulas are:

cosh(Z1 + Z2 ) = cosh Z1 cosh Z2 + sinh Z1 sinh Z2 ,

sinh(Z1 + Z2 ) = sinh Z1 cosh Z2 + cosh Z1 sinh Z2 .

Moreover, for Z = jz2 , we have:

cosh(jz2 ) = cos(z2 ), sinh(jz2 ) = j sin(z2 ).

Then for Z = z1 + jz2 we have:

sinh Z = sinh(z1 + jz2 ) = sinh(z1 ) cos(z2 ) + j cosh(z1 ) sin(z2 ).

A similar formula holds for cosh Z.

128 Chapter 6. Elementary Bicomplex Functions

6.4 Bicomplex radicals

In this and the next section we begin the study of inverse functions in BC. We
start by looking at the equation Z n = W , where the bicomplex numbers Z and
W are written as Z = z1 + jz2 = β1 e + β2 e† , and W = w1 + jw2 = γ1 e + γ2 e† .
This equation is equivalent to the following two complex equations in variables β1
and β2 :
β1n = γ1 , β2n = γ2 .
If W is invertible, i.e., γ1 γ2 = 0, then each complex equation has n distinct
complex solutions, and the equations are independent of each other. Denote these
√ √ √
solutions by γ1,k ∈ n γ1 and γ2, ∈ n γ2 , respectively, where the symbol n γ1
√
denotes the set of all solutions of the corresponding equation; the same for n γ2 .
Therefore the bicomplex equation Z n = W has n2 solutions given by the bicomplex
numbers
γ1,k + γ2, γ2, − γ1,k
Zk = γ1,k e + γ2, e† = +j
2 2i
for all k,  √ = 1 . . . n. We define the n-th root of W to be the set of all of these
solutions, n W := {Zk }.
Note that if we start with formula (3.5) for the bicomplex number W =
w1 + jw2 , i.e.,
W = |W |i (cos θ + j sin θ)

2 2
where |W |i = w1 + w2 is the complex modulus of W , and θ is the complex
argument of W , then the solutions Zk of the equation Z n = W have complex
 θ + 2π
moduli n |W |i , which is a set of n complex numbers, and arguments , for
n
 = 1, . . . , n. In conclusion, we find again that there are n2 bicomplex n − th roots,
and more precisely
√
n
 θ + 2π θ + 2π
W = { n |W |i (cos + j sin ) :  ∈ {0, 1, . . . , n − 1}}.
n n
If W = γ1 e + γ2 e† is a zero-divisor, then exactly one of the complex numbers
γ1 or γ2 is zero, so the bicomplex equation Z n = W has exactly n solutions, all of
them zero divisors. Obviously if W = 0 there is only one solution, Z = 0, to the
equation Z n = 0.

6.5 The bicomplex logarithm

6.5.1 The real and complex logarithmic functions.
Since the real exponential function is monotone, then the real logarithmic function
being its inverse does not cause much problem. The situation with its complex
extension is much more sophisticated. The complex logarithm, that is, the notion
of the logarithm of a complex number, is introduced from the study of the equation
6.5. The bicomplex logarithm 129

ez = w. Because the periodicity of the complex exponential function, this equation

has, for w = 0, a countable family of solutions of the form

ln |w| + i(arg w + 2πm)

where ln |w| is the real logarithm of the positive number |w|, m is any integer.
Hence, for any non-zero complex number there are infinitely many complex num-
bers which can be equally called its logarithm.
We will use the following notations and names:

ln(w) = ln w := ln |w| + iarg w

is called the principal value of the logarithm of w;

lnm (w) = lnm w := ln |w| + i (arg w + 2πm) ,

for a fixed m ∈ Z, is called the m-th branch of logarithm of w;



Ln (w) := lnm w  m ∈ Z

is called the complex logarithm of w. So, Ln w is a set and, thus, it does not
generate a univalued function but rather a multivalued function, which explains
the name for lnm w: for each m it determines a function, which can be called a
logarithmic function.
Note an important property of logarithm:

Ln (w1 w2 ) = Ln (w1 ) + Ln (w2 ), (6.21)

where the plus in the right-hand side among the two sets means that the resulting
set is obtained by adding all the elements of the first item and all the elements of
the second item. Observe that the various branches of the logarithmic function do
not possess this property.

6.5.2 The logarithm of a bicomplex number

In this section we define the notion of the logarithm of a bicomplex number. Take a
bicomplex number Z written as Z = z1 + jz2 , and an invertible bicomplex number
W = w1 + jw2 . We study the solutions to the bicomplex equation eZ = W . Recall
that
W = |W |i (cos θ + j sin θ),
where |W |i is the complex modulus and θ is the complex argument of W : |W |i :=

w12 + w22 , θ ∈ Argi W . From the equation

ez1 (cos z2 + j sin z2 ) = |W |i · (cos(Argi W ) + j sin(Argi W ))

it follows that
z1 ∈ Ln |W |i ,
130 Chapter 6. Elementary Bicomplex Functions

the complex logarithm of the complex number |W |i , and that

z2 ∈ Argi W = {argi W + 2πm | m ∈ Z} .

In analogy with the pattern of complex logarithms we introduce the following

definitions for an invertible bicomplex number W : the number
 
ln(W ) = lnW := ln|W |i  + i arg|W |i + j argi W

is called the principal value of the (bicomplex) logarithm of W ; then the number

lnm,n (W ) := lnm |W |i + j (argi (W ) + 2πn) ,

with two arbitrary integers m and n, is called the (m, n)-th branch of the bicomplex
logarithm; and finally the set


Ln(W ) := lnm,n (W )  m, n ∈ Z

is called the bicomplex logarithm of W . Again, Ln(W ) is a set, not a (univalued)

function.
If the idempotent representation W = γ1 e + γ2 e† is used, then it turns out
that
Ln(W ) = Ln(γ1 )e + Ln(γ2 )e† .
This is because if Z = β1 e + β2 e† , then the equation eZ = W is equivalent to the
two complex equations
eβ1 = γ 1 , eβ 2 = γ 2
which have as solutions the complex logarithms Ln(γ1 ) and Ln(γ2 ) respectively;
*
1 1 γ1 √ γ1
Ln(W ) = Ln(γ1 γ2 ) + ji Ln = Ln γ1 γ2 + ji Ln
2 2 γ2 γ2
= Ln|W |i + j Argi (W ),

where we used the fact that *

γ1
θ = i ln . (6.22)
γ2
We state below some properties of the bicomplex logarithm.
• The bicomplex logarithm is not defined for zero-divisors, as the bicomplex
exponential W = eZ is always invertible.
• If Z = z1 + jz2 is an invertible bicomplex number and if m, n ∈ Z, then

elnm,n (Z) = elnm |Z|i +j argi (Z)+2nπj = elnm |Z|i e j argi (Z)
= |Z|i (cos(argi (Z)) + j sin(argi (Z))) = Z.
6.6. On bicomplex inverse trigonometric functions 131

• For Z = 1 = 1 + j0, we have:

lnm,n (1) = 0 + 2mπi + 2nπj

for all m, n ∈ Z.
• For Z1 and Z2 two invertible bicomplex numbers, the following formula holds:

Ln(Z1 Z2 ) = Ln(Z1 ) + Ln(Z2 ). (6.23)

6.6 On bicomplex inverse trigonometric functions

The inverses of the bicomplex trigonometric functions are defined in complete
analogy with the complex case, as we have already properly defined the notions
of bicomplex exponential, logarithm, and square root.
For example, the inverse of the bicomplex cosine function is obtained by
solving the equation
ejZ + e−jZ
cos(Z) = = W.
2
This is a quadratic equation in ejZ with roots

ejZ = W ± W 2 − 1.

Therefore, the set Arccos(W ) should be defined as

    
Arccos(W ) := −j lnm,n W ± W 2 − 1  m, n ∈ Z
    
= ± j lnm,n W + W 2 − 1  m, n ∈ Z .

It is clear how to introduce the principal value, the (m, n)-th branch, etc., as
well as how to deal with other inverse functions. We leave this to the interested
reader, who is recommended also to consider what happens if we solve the equation

eiZ + e−iZ
cos(Z) = = W.
2

6.7 The exponential representations of bicomplex

numbers
We will show here that the trigonometric representations of bicomplex numbers
can be written with the help of the bicomplex exponential function. Indeed, recall
that an invertible bicomplex number Z has a C(i)-trigonometric representation
and a C( j)-one:
Z = |Z|i (cos θ + j sin θ)
132 Chapter 6. Elementary Bicomplex Functions

where θ is a C(i)-argument of Z, i.e., any element in ArgC(i) (Z), see (3.7); and

Z = |Z|j (cos ψ + i sin ψ)

where ψ is a C( j)-argument of Z, i.e., any element in ArgC( j) (Z), see (3.13). But

cos θ + j sin θ = e jθ and cos ψ + i sin ψ = eiψ ,

hence
Z = |Z|i · e jθ = |Z|i · e j argC(i),m (Z) (6.24)
and
Z = |Z|j · eiψ = |Z|j · ei argC( j),n (Z) (6.25)
for any integers m and n. We will refer, sometimes, to (6.24) and (6.25) as the
(complex) exponential representations of Z. Note also that the properties of the
exponential function imply that

Z † = |Z|i · e−jθ

and
Z = |Z|j · e−iψ .
The trigonometric representation of a bicomplex number in hyperbolic terms
is allowed both for invertible numbers and for zero-divisors. Beginning with an
invertible Z we should recall that
 
Z = |Z|k (cos ΨZ + i sin ΨZ ) = |Z|k eiν1 · e + eiν2 · e†

with ΨZ = ν1 e + ν2 e† ∈ D+inv being any element in ArgD Z, i.e., a hyperbolic

argument of Z. Appealing again to the properties of the exponential function we
see that

cos ΨZ + i sin ΨZ = eiν1 · e + eiν2 · e† = eiΨZ = ei argm,n;D (Z)

for any integers m and n. Thus, the equality

Z = |Z|k · ei argm,n;D (Z) (6.26)

is the exponential representation of an invertible bicomplex number in hyperbolic

terms.
If Z is a zero-divisor, then Z = |Z|k · eiν1 e or Z = |Z|k · eiν2 e† with real
numbers ν1 and ν2 ; the corresponding hyperbolic arguments are argD (Z) = ν1 e
and argD (Z) = ν2 e† respectively. But eiν1 e is equal to eiν1 e + e† , not to eiν1 e,
hence the analogue of the exponential representation in hyperbolic terms for this
case is  
Z = |Z|k ei argm,D (Z) − e† . (6.27)
6.7. The exponential representations of bicomplex numbers 133

Similarly we get for Z = |Z|k · eiν2 e†

 
Z = |Z|k ei argn,D (Z) − e . (6.28)

So although the formulas (6.27)–(6.28) have a different structure than (6.24),

(6.25) and (6.26), we believe that the name “exponential representation” is appli-
cable also.
Our study mimics the approach done in usual complex analysis (e.g. [1])
and develops the most important properties of the elementary functions in the
bicomplex setup. These are basically the main functions of the bicomplex analysis,
on which we “test” the notions of derivability, differentiability, holomorphy, and
integration in the following Chapters.
Aspects of the theory of bicomplex elementary functions have been also stud-
ied in [45, 55, 57, 99].
Chapter 7

Bicomplex Derivability and

Differentiability

7.1 Different kinds of partial derivatives

In this chapter we will study bicomplex-valued functions of a bicomplex varia-
ble, and we will examine the notions of derivability and of holomorphy for such
functions (see [46]). We will consider BC as a topological space endowed with
the Euclidean topology of R4 and Ω an open set. Let F be a bicomplex function
F : Ω → BC of a bicomplex variable

Z = x1 + iy1 + jx2 + ky2 .

Due to the various ways in which bicomplex numbers can be written, the function
F inherits analogous representations, specifically:

F = f1 + jf2 = ρ1 + iρ2 = g1 + kg2 = γ1 + kγ2 = f1 + if2 = g1 + jg2

= f11 + if22 + jf21 + kf22 , (7.1)

where f1 , f2 , g1 , g2 are C(i)-valued functions, ρ1 , ρ2 , γ1 , γ2 are C(j)-valued func-

tions, f1 , f2 , g1 , g2 are hyperbolic-valued functions, and fk are real-valued func-
tions, all of a bicomplex variable Z. We will, at different times in this section, use
all of these representations.
Take now a point Z0 ∈ Ω and let H = h11 + ih12 + jh21 + kh22 be the
increment. The partial derivatives of F with respect to the variables x1 , y1 , x2 , y2
are defined as usual (when they exist) and we give their formulas in order to fix
the notation:

© Springer International Publishing Switzerland 2015 135

M.E. Luna-Elizarrarás et al., Bicomplex Holomorphic Functions, F rontiers in Mathematics,
DOI 10.1007/978-3-319-24868-4_8
136 Chapter 7. Bicomplex Derivability and Differentiability

∂F F (Z0 + h11 ) − F (Z0 )

(Z0 ) := lim ,
∂x1 h11 →0 h11
∂F F (Z0 + ih12 ) − F (Z0 )
(Z0 ) := lim ,
∂y1 h12 →0 h12
∂F F (Z0 + jh21 ) − F (Z0 )
(Z0 ) := lim ,
∂x2 h21 →0 h21
∂F F (Z0 + kh22 ) − F (Z0 )
(Z0 ) := lim . (7.2)
∂y2 h 22 →0 h22

Recalling formulas (1.3) and (1.4) we can write the increments as

H = h1 + jh2 := (h11 + ih12 ) + j(h21 + ih22 )

= κ1 + iκ2 := (h11 + jh21 ) + i(h12 + jh22 ) .

This leads us to
Definition 7.1.1. The complex partial derivatives of the bicomplex function F are
defined as the following limits (if they exist):

F (Z0 + h1 ) − F (Z0 )
Fz1 (Z0 ) := lim , (7.3)
h1 →0 h1
F (Z0 + jh2 ) − F (Z0 )
Fz2 (Z0 ) := lim , (7.4)
h2 →0 h2
F (Z0 + κ1 ) − F (Z0 )
Fζ1 (Z0 ) := lim , (7.5)
κ1 →0 κ1
F (Z0 + iκ2 ) − F (Z0 )
Fζ2 (Z0 ) := lim . (7.6)
κ2 →0 κ2
Similarly we introduce the hyperbolic partial derivatives: if we recall for-
mula (1.5) and write the increment as

H = h1 + ih2 := (h11 + kh22 ) + i(h12 + k(−h21 )) ,

then the hyperbolic partial derivatives are

F (Z0 + h1 ) − F (Z0 )
Fz1 (Z0 ) := lim ,
h1 ∈S
/ 0 (D), h1 →0 h1
F (Z0 + i h2 ) − F (Z0 )
Fz2 (Z0 ) := lim , (7.7)
h2 ∈S
/ 0 (D), h2 →0 h2

where with the symbol S0 (D) := S0 ∩ D we indicate the set of the hyperbolic
zero-divisors together with 0 ∈ D.
Example 7.1.2. From the point of view of the classic theory of functions of two
complex variables, one may think, because of the “symmetry” of the imaginary
7.2. The bicomplex derivative and the bicomplex derivability 137

units i and j, and thus the analogy between the corresponding complex variables,
that there is no reason to define all the complex partial derivatives (7.3)–(7.6). The
following simple example illustrates that inside the bicomplex realm the differences
are relevant. Consider to this purpose the function F : BC → BC, F (Z) = Z † . It
is immediate to see that

Fz1 (Z0 ) = 1, Fz2 (Z0 ) = −j

and that neither Fζ1 (Z0 ) nor Fζ2 (Z0 ) exist for any Z0 ∈ BC.
Remark 7.1.3. Formulas (1.6)–(1.8) suggest the introduction of six more partial
derivatives: four complex and two hyperbolic. But since there exists a direct relation
between the corresponding variables, namely,

w 1 = z1 , w2 = −iz2 , ω1 = ζ1 , ω2 = −jζ2 , w1 = z1 , w2 = −kz2 ,

it is easy to see that we do not get anything essentially new. The situation with
partial derivatives with respect to variables arising from the idempotent represen-
tations of bicomplex numbers is on the other hand much more interesting, and will
be discussed in detail later.

7.2 The bicomplex derivative and the bicomplex

derivability
We continue now with the definition of the derivative of a bicomplex function
F : Ω ⊂ BC → BC of one bicomplex variable Z as follows:
Definition 7.2.1. The derivative F  (Z0 ) of the function F at a point Z0 ∈ Ω is the
limit, if it exists,

F (Z) − F (Z0 ) F (Z0 + H) − F (Z0 )

F  (Z0 ) := lim = lim , (7.8)
Z→Z0 Z − Z0 S0  H→0 H
for Z in the domain of F such that H = Z − Z0 is an invertible bicomplex number.
In this case, the function F is called derivable at Z0 .
Corollary 7.2.2. The function F is derivable at Z0 if and only if there exists a
function αF,Z0 such that
lim αF,Z0 (H) = 0
S0  H→0

and

F (Z0 + H) − F (Z0 ) = F  (Z0 ) · H + αF,Z0 (H)H for all H ∈ S0 . (7.9)

Remark 7.2.3. It is necessary to make a comment here. Traditionally, see e.g. [103],
p. 138 and p. 432, if h is either a real or a complex increment, the symbol o(h)
138 Chapter 7. Bicomplex Derivability and Differentiability

is used to indicate any expression of the form α(h)|h| with lim α(h) = 0. Since,
h→0
|h|
both in the real and in the complex case, the expression remains bounded when
h
h → 0, it is clear that one could replace α(h)|h| by α(h)h in the expression of o.
However, the situation is quite different in the bicomplex case. Here, the expression
|H|
is not bounded when H → 0, and therefore we need to carefully distinguish the
H
two expressions. In accordance with the usual notation, we will always use o(H)
to denote a function of the form α(H)|H|, and therefore the expression in the
previous corollary αF,Z0 (H)H is not, in general, o(H). This distinction is at the
basis of the notions of weak and strong Stoltz conditions for bicomplex functions,
which are used by G.B.Price in [56].
Remark 7.2.4. A bicomplex function F derivable at Z0 enjoys the following pro-
perty:
lim (F (Z0 + H) − F (Z0 )) = 0 . (7.10)
H ∈S
/ 0 , H→0

In other words, a function F , which is derivable at a point Z0 , enjoys some sort

of “weakened” continuity at that point, in the sense that F (Z) converges to F (Z0 )
as long as Z converges to Z0 in such a way that Z − Z0 is invertible. We will see
later on that this restriction can be removed under reasonable assumptions.
It turns out that the arithmetic operations on derivable functions follow the
usual rules of their real and complex antecedents.
Theorem 7.2.5. (Derivability and arithmetic operations) Let F and G be two
bicomplex functions defined on Ω ⊂ BC and derivable at Z0 ∈ Ω. Then:
1. The sum and difference of F and G are derivable at Z0 and

(F ± G) (Z0 ) = F  (Z0 ) ± G (Z0 ).

2. For any bicomplex number C the function C · F is derivable at Z0 and

(C · F ) (Z0 ) = C · F  (Z0 ).

3. The product of F and G is derivable at Z0 and

(F · G) (Z0 ) = F  (Z0 ) · G(Z0 ) + F (Z0 ) · G (Z0 ).

F
4. If G is continuous at Z0 and G(Z0 ) ∈ S0 , then the quotient is derivable
G
at Z0 and
 
F F  (Z0 ) · G(Z0 ) − F (Z0 ) · G (Z0 )
(Z0 ) = .
G (G(Z0 ))2
7.2. The bicomplex derivative and the bicomplex derivability 139

Proof. From Corollary 7.2.2, there exist two bicomplex functions αF,Z0 and αG,Z0
such that
lim αF,Z0 (H) = lim αG,Z0 (H) = 0
S0  H→0 S0  H→0

and

F (Z0 + H) − F (Z0 ) = F  (Z0 ) · H + αF,Z0 (H)H , (7.11)


G(Z0 + H) − G(Z0 ) = G (Z0 ) · H + αG,Z0 (H)H , (7.12)

for all H ∈ S0 . The first statement about the sum or difference F ± G is ob-
tained easily by adding (or subtracting) formulas (7.11) and (7.12), and noting
that the functions αF ±G,Z0 := αF,Z0 ± αG,Z0 respect the properties from Corol-
lary 7.2.2 (which is an “if and only if” statement). Similarly, if we multiply (7.11)
by the number C, we obtain statement (2). Note that C needs not necessarily be
an invertible bicomplex number. For example, if C = c e, where c ∈ C(i), then
from (7.11) (multiplied by C) we obtain that

c e · F  (Z0 ) = (c F  (Z0 ))e ∈ BCe

is the derivative of ce · F at Z0 , which is a zero-divisor. The non-invertibility of C

does not affect any arguments, as H is always invertible.
If we multiply the terms F (Z0 + H) and G(Z0 + H) obtained from the left-
hand side of formulas (7.11) and (7.12) we obtain:

F (Z0 + H) · G(Z0 + H) = F (Z0 )G(Z0 ) + (F  (Z0 ) · G(Z0 ) + F (Z0 ) · G (Z0 )) · H

+ (F (Z0 ) αG,Z0 (H) + G(Z0 ) αF,Z0 (H)) · H
+ (F  (Z0 )G (Z0 ) + F  (Z0 ) αG,Z0 (H) + G (Z0 ) αF,Z0 (H)
+ αF,Z0 (H) αG,Z0 (H)) · H 2 .

Now, writing

αF G,Z0 (H) := F (Z0 )αG,Z0 (H) + G(Z0 )αF,Z0 (H)

+ (F  (Z0 )G (Z0 ) + F  (Z0 )αG,Z0 (H)+
+ G (Z0 )αF,Z0 (H) + αF,Z0 (H)αG,Z0 (H)) · H,

we obtain:
lim αF G,Z0 (H) = 0
S0  H→0

and

F (Z0 + H) · G(Z0 + H) − F (Z0 )G(Z0 )

= (F  (Z0 ) · G(Z0 ) + F (Z0 ) · G (Z0 )) · H + αF G,Z0 (H)H ,

which proves that the product F · G is derivable at Z0 and the usual product
formula holds.
140 Chapter 7. Bicomplex Derivability and Differentiability

We assume now that G(Z0 ) ∈ S0 and that G is continuous at Z0 , which

implies that G(Z) ∈ S0 in a neighborhood of Z0 . All the computations below are
performed in this neighborhood. Then we write:
F (Z0 + H) F (Z0 ) F (Z0 + H)G(Z0 ) − G(Z0 + H)F (Z0 )
− =
G(Z0 + H) G(Z0 ) G(Z0 + H)G(Z0 )
(F (Z0 + H) − F (Z0 )) G(Z0 ) − (G(Z0 + H) − G(Z0 )) F (Z0 )
=
G(Z0 + H)G(Z0 )
 
(F (Z0 )·G(Z0 ) − G (Z0 )·F (Z0 ))·H + (αF,Z0 (H)G(Z0 ) + αG,Z0 (H)F (Z0 ))·H
= .
G(Z0 + H)G(Z0 )
(7.13)
Now note that if we write

1 1 G (Z0 ) + αG,Z0 (H)
− = − · H =: α1 (H) ,
G(Z0 + H) G(Z0 ) G(Z0 + H)G(Z0 )
then the function α1 (H) has the property
lim α1 (H) = 0 .
S0  H→0

Thus using the formula:

1 1
= + α1 (H) ,
G(Z0 + H) G(Z0 )
rewriting (7.13) leads to the expression:
F (Z0 + H) F (Z0 ) F  (Z0 ) · G(Z0 ) − G (Z0 ) · F (Z0 )
− = · H + α F ,Z0 (H) · H ,
G(Z0 + H) G(Z0 ) (G(Z0 ))2 G

where α F ,Z0 (H) is the bicomplex function containing all the remaining terms and
G
which enjoys the property
lim α F ,Z0 (H) = 0 .
S0  H→0 G

Using again Corollary 7.2.2, the proof of the derivability of the quotient and of
the corresponding formula is finished. 
Theorem 7.2.6. Let F and G be two bicomplex functions, F is defined on an open
set Ω and G is defined on F (Ω) ⊂ BC. Assume that there is a point Z0 ∈ Ω such
that W0 = F (Z0 ) is an interior point of F (Ω) and such that
F (Z) − F (Z0 ) ∈ S0 , implies Z − Z0 ∈ S0 , Z ∈ Ω.
Assume now that F is derivable at Z0 and G is derivable at W0 . Then the com-
position G ◦ F is derivable at Z0 and
(G ◦ F ) (Z0 ) = G (F (Z0 )) · F  (Z0 ).
7.2. The bicomplex derivative and the bicomplex derivability 141

Moreover, if F  (Z0 ) ∈ S0 and if F is a bijective function around Z0 , then its

inverse F −1 is derivable at W0 = F (Z0 ) and
(F −1 ) (W0 ) · F  (Z0 ) = 1 .
Proof. Via Corollary 7.2.2, there exist two bicomplex functions αF,Z0 and αG,W0
such that
lim αF,Z0 (H) = lim αG,W0 (K) = 0
S0  H→0 S0  K→0

and
F (Z0 + H) − F (Z0 ) = F  (Z0 ) · H + αF,Z0 (H)H , (7.14)
G(W0 + K) − G(W0 ) = G (W0 ) · K + αG,W0 (K)K , (7.15)
for all H, K ∈ S0 .
Denote by VW0 a neighborhood of W0 contained in F (Ω) which exists by
∗
hypothesis. Let then VW 0
be the set obtained from VW0 by eliminating the points
W for which W − W0 are zero-divisors. Let UZ∗ 0 be the set of all Z ∈ UZ0 :=
F −1 (VW0 ) such that Z − Z0 is not a zero-divisor.
Now we note that formula (7.15) is true for those invertible K such that
∗
WK := W0 + K ∈ VW0 ; such K exist by taking WK ∈ VW 0
. Then by the properties
∗
of F , for any such K there exists a ZK ∈ UZ0 such that WK = W0 + K = F (ZK ).
Now we write:
ZK = Z0 + H := Z0 + (ZK − Z0 ) ,
where H is invertible by the choice of ZK . Therefore we replace W0 +K = F (ZK ) =
F (Z0 + H), W0 = F (Z0 ) and K = F (Z0 + H) − F (Z0 ) in formula (7.15), and we
obtain:
G(F (Z0 + H)) − G(F (Z0 )) = G (F (Z0 )) · (F (Z0 + H) − F (Z0 ))
+α"G,F (Z0 ) (H) · (F (Z0 + H) − F (Z0 )) ,
where α "G,F (Z0 ) (H) := αG,F (Z0 ) (F (Z0 + H) − F (Z0 )) is a function having the
property from Corollary 7.2.2, as when K → 0 so does H → 0, in both cases along
invertible bicomplex numbers. Now we use formula (7.14) and we get:
G(F (Z0 + H)) − G(F (Z0 )) = G (F (Z0 )) · (F  (Z0 ) · H + αF,Z0 (H)H)
+α"G,F (Z0 ) (H) · (F  (Z0 ) · H + αF,Z0 (H)H)
=: G (F (Z0 )) · F  (Z0 ) · H + αG◦F (H)H ,
where the function αG◦F (H) has obviously the desired property from Corollary 7.2.2.
The last statement of the theorem can be proved quite analogously to the
cases of real and complex functions. If one assumes beforehand the existence of the
derivative of the inverse function at Wo , then the corresponding formula becomes
a particular case of the chain rule above:
1 = (F −1 ◦ F ) (Z0 ) = (F −1 ) (F (Z0 )) · F  (Z0 ) ,
142 Chapter 7. Bicomplex Derivability and Differentiability

where we used that the derivative of the identity function is one which is obvious.

Theorem 7.2.7 (The derivability of bicomplex elementary functions). All bicom-
plex elementary functions introduced in Chapter 6 are derivable at any point
Z = Z0 where they are defined. In more detail:
1. Any constant function is derivable with the derivative zero.
2. Bicomplex polynomials are derivable for any Z ∈ BC. In particular, we have:

(Z n ) = nZ n−1 .

3. The derivative of the bicomplex exponential function eZ is eZ .

4. (sin Z) = cos Z and (cos Z) = − sin Z.
1
5. (tan Z) = for any Z such that cos Z is an invertible bicomplex
(cos Z)2
number, i.e., using the idempotent representation Z = β1 e+β2 e† , the complex
π
numbers β1 and β2 are not equal to plus integer multiples of π.
2
−1
Similarly, (cot Z) = for any Z such that sin Z is an invertible
(sin Z)2
bicomplex number, i.e., whenever β1 and β2 are not integer multiples of π.
6. (cosh Z) = sinh Z and (sinh Z) = cosh Z.
7. The (m, n)-branch logarithmic function F (Z) = lnm,n (Z), which is defined
for all invertible bicomplex numbers Z, is derivable for all such Z, and
1
(lnm,n ) (Z) = .
Z

Proof. Item (1) is obviously true.

Note that the formula

An − B n = (A − B)(An−1 + An−2 B + · · · + B n−1 )

holds for any A, B ∈ BC, due to the algebraic properties of addition and multi-
plication of bicomplex numbers. Applying this for A = Z + H and B = Z, we
obtain:
(Z + H)n − Z n
lim
S0  H→0 H
 
H (Z + H)n−1 + (Z + H)n−2 Z + · · · + Z n−1
= lim
S0  H→0 H
 
= lim (Z + H) n−1
+ (Z + H)n−2 Z + · · · + Z n−1 = nZ n−1 .
S0  H→0
7.2. The bicomplex derivative and the bicomplex derivability 143

$
n
It follows that any bicomplex polynomial p(Z) = Ak Z k is derivable at any
k=0
Z ∈ BC, and its derivative is a polynomial of degree one less, given by

$
n
p (Z) = kAk Z k−1 .
k=1

Let us consider the bicomplex exponential function eZ . For an arbitrary

Z ∈ BC and an invertible bicomplex number H, we have:

eZ+H − eZ eH − 1
= eZ · .
H H
We show now that
eH − 1
lim = 1.
S0  H→0 H
For this, let us write H = he + ke† in the idempotent representation, where
h, k ∈ C(i) are such that hk = 0. Then eH = eh e + ek e† , thus

eH − 1 e h e + e k e† eh − 1 ek − 1 †
= = e + e .
H he + ke† h k
Therefore
 
eH − 1 eh − 1 ek − 1
lim = lim e + lim e† = e + e† = 1 .
S0  H→0 H h→0 h k→0 k

It follows that

eZ+H − eZ eH − 1
lim = eZ · lim = eZ ,
S0  H→0 H S0  H→0 H

so (eZ ) = eZ , for all Z ∈ BC.

Recall that the bicomplex trigonometric functions are defined by:

ejZ + e−jZ ejZ − e−jZ

cos Z = , sin Z = .
2 2j

Since the exponential function is derivable for all Z ∈ BC, it follows that cos Z is
derivable for all Z and

jejZ − je−jZ ejZ − e−jZ

(cos Z) = =− = sin Z .
2 2j

Similarly the function sin Z is derivable for all Z and (sin Z) = cos Z.
144 Chapter 7. Bicomplex Derivability and Differentiability

sin Z
The bicomplex tangent function is defined as the quotient , whenever
cos Z
cos Z is invertible. Using the quotient rule, we obtain:

cos2 Z + sin2 Z 1
(tan Z) = = .
cos2 Z cos2 Z
A similar argument applies for the function cot Z.
The hyperbolic functions of a bicomplex variable are defined as

eZ + e−Z eZ − e−Z
cosh Z = , sin Z = .
2 2
Again, using that eZ is derivable for any Z, it follows that cosh Z and sinh Z are
derivable for any Z and

eZ − e−Z eZ + e−Z
(cosh Z) = = sinh Z, (sinh Z) = = cosh Z .
2 2
Recall that the (m, n)-branch bicomplex logarithmic function lnm,n , defined
for all invertible bicomplex numbers, is the inverse function of eZ (at least on a
subset of its domain of definition). Then we have:
 
1 = elnm,n (Z) = elnm,n (Z) · (lnm,n ) (Z) = Z · (lnm,n ) (Z) .

Since Z runs over all invertible bicomplex numbers, we obtain

1
(lnm,n ) (Z) = ,
Z
which concludes our proof. 

7.3 Partial derivatives of bicomplex derivable functions

Since bicomplex numbers admit many representations, in this section we inves-
tigate what the existence of the derivative implies for different types of partial
derivatives.
Theorem 7.3.1. Consider a bicomplex function F : Ω ⊂ BC → BC derivable at
Z0 ∈ Ω. Then the following statements are true:
∂F ∂F
1. The real partial derivatives (Z0 ) and (Z0 ) exist, for  = 1, 2.
∂x ∂y
2. The real partial derivatives satisfy the identities:
∂F ∂F ∂F ∂F
F  (Z0 ) = (Z0 ) = −i (Z0 ) = −j (Z0 ) = k (Z0 ) . (7.16)
∂x1 ∂y1 ∂x2 ∂y2
7.3. Partial derivatives of bicomplex derivable functions 145

Proof. Because F is derivable at Z0 , limit (7.8) exists no matter how H converges

to zero , as long as H is invertible. Consider first H to be real, i.e., of the form
H = x1 = x1 + i0 + j0 + k0 → 0, x1 = 0, which is always invertible. Then, since
there exists F  (Z0 ), the following limit also exists:

F (Z0 + x1 ) − F (Z0 )
lim ,
x1 →0 x1
∂F
and it coincides with (Z0 ).
∂x1
The rest of the proof follows by considering the three specific forms of the
increment H along the units i, j, k, i.e., H = iy1 , with y1 = 0; H = jx2 , with
x2 = 0; H = ky2 , with y2 = 0, and noting that all are invertible bicomplex
numbers. 
Let us write the bicomplex function as F = f11 + if12 + jf21 + kf22 in terms
of its real components, which are all real functions of a bicomplex variable. An
immediate consequence of the theorem above is
Corollary 7.3.2. If F is derivable at Z0 , then the the real Jacobi matrix of F at
Z0 is of the form
⎛ ⎞
a −b −c d
⎜ b a −d −c ⎟
JZ0 [F ] := ⎜
⎝ c −d a
⎟, (7.17)
−b ⎠
d c b a

where
∂f11 ∂f12 ∂f21 ∂f22
a := = = = ,
∂x1 ∂y1 ∂x2 ∂y2
∂f11 ∂f12 ∂f21 ∂f22
b := − = =− = ,
∂y1 ∂x1 ∂y2 ∂x2
∂f11 ∂f12 ∂f21 ∂f22
c := − =− = = ,
∂x2 ∂y2 ∂x1 ∂y1
∂f11 ∂f12 ∂f21 ∂f22
d := =− =− = , (7.18)
∂y2 ∂x2 ∂y1 ∂x1
and where all the partial derivatives are evaluated at the point Z0 .
Proof. The proof relies on a direct computation using equalities (7.16) written in
terms of fk . 

Remark 7.3.3. The reader should notice that the special form of the real Jacobi
matrix above encodes several Cauchy-Riemann type conditions on (certain pairs
of ) the real functions fk , a fact which we will exploit in detail below.
146 Chapter 7. Bicomplex Derivability and Differentiability

Remark 7.3.4. Every 4 × 4 matrix with real entries determines a linear transfor-
mation on R4 . Of course, not all of them remain linear when R4 is seen as BC,
i.e., not all of them are BC-linear. Those matrices which are BC-linear are of the
form (7.17). Taking into account that the entries of (7.17) are the values of the
partial derivatives at Z0 , one may conclude that (7.17) determines a BC-linear
operator acting on the tangential BC-linear module at the point Z0 . Moreover, in
the matrix (7.17) there are hidden the two linearities with respect to C(i) and C(j),
in the following sense: fixing the identifications

(x1 + iy1 , x2 + iy2 ) = (z1 , z2 ) ←→ (x1 , y1 , x2 , y2 )

and
(x1 + jx2 , y1 + jy2 ) = (ζ1 , ζ2 ) ←→ (x1 , y1 , x2 , y2 )
we have two different identifications C2 (i) ↔ R4 and C2 (j) ↔ R4 . It follows that
a 4 × 4 matrix with real entries determines not only a real transformation but a
C(i)-linear one, if and only if it is of the form
⎛ ⎞
l −m u −v
⎜ m l v u ⎟
⎜ ⎟
⎝ t −s g −h ⎠ .
s t h g

Similarly, a 4 × 4 matrix with real entries represents a C(j)-linear transformation

if and only if it is of the form
⎛ ⎞
A B −E −F
⎜ C D −G −H ⎟
⎜ ⎟.
⎝ E F A B ⎠
G H C D

Thus the structure of the matrix (7.17) includes both complex structures.
Remark 7.3.5. An easy computation shows that the real Jacobian, i.e., the deter-
minant of the matrix (7.17), is given by
   
det(JZ0 ) = (b + c)2 + (a − d)2 · (b − c)2 + (a + d)2 ∈ R+ .

Moreover, a direct computation yields that the expression above is nothing but
  
|F (Z0 )|2i 2 ,

i.e., the square of the usual modulus of the square of the complex modulus of
F  (Z0 ).
A simple computation shows that det(JZ0 ) = 0 if and only if F  (Z0 ) ∈ S0 ,
which is equivalent to

b = −c and a = d, or b = c and a = −d . (7.19)

7.3. Partial derivatives of bicomplex derivable functions 147

Note that these relations are between the real partial derivatives of the real com-
ponents of F at Z0 . Recall that the derivative of F at Z0 is given by

F  (Z0 ) = a + b i + c j + d k .

Indeed, if b = −c and a = d, then

F  (Z0 ) = a(1 + k) + b(i − j) = a(1 + k) + i b(1 + k) = 2(a + i b) · e ,

so F  (Z0 ) ∈ BCe ⊂ S0 is a zero-divisor or zero. Similarly, if b = c and a = −d,

then

F  (Z0 ) = a(1 − k) + b(i + j) = a(1 − k) + i b(1 − k) = 2(a + i b) · e† ,

so F  (Z0 ) ∈ BCe† ⊂ S0 . Conversely, we note that F  (Z0 ) = 0 if and only if

a = b = c = d = 0, and F  (Z0 ) ∈ S0 if and only if relations (7.19) hold.
For example, if all four of the real partial derivatives of any real function fk
are zero at Z0 , it follows that F  (Z0 ) = 0. In particular, if one of fk is constant
around Z0 , then F  (Z0 ) = 0.
As a next step, we investigate the consequence of the existence of F  (Z0 )
in terms of the complex variables z1 , z2 ∈ C(i), where we write the bicomplex
variable as Z = z1 + jz2 . We prove the following
Theorem 7.3.6. Consider a bicomplex function F = f1 + jf2 derivable at Z0 . Then
we have:
1. The C(i)-complex partial derivatives Fz (Z0 ) exist, for  = 1, 2.
2. The complex partial derivatives above verify the identity:

F  (Z0 ) = Fz1 (Z0 ) = −jFz2 (Z0 ) , (7.20)

which is equivalent to the C(i)-complex Cauchy-Riemann system for F (at

Z0 ), also called the generalized Cauchy-Riemann system in [68] and [69]:
   
f1,z 1
(Z0 ) = f2,z 2
(Z0 ) , f1,z 2
(Z0 ) = −f2,z 1
(Z0 ) . (7.21)

Proof. As in the proof of Theorem 7.3.1, the limit of the difference quotient must
be the same regardless of the path on which H approaches zero, as long as H is
invertible. Let us choose H = h1 = h1 + j 0 → 0, and note that if Z0 = z01 + jz02 ,
then Z0 + H = (z01 + h1 ) + jz02 . Then, since F  (Z0 ) exists, the following limit also
exists:
F (Z0 + h1 ) − F (Z0 )
lim = Fz1 (Z0 ).
h1 →0 h1
Similarly, taking H = j h2 → 0 we get:
F (Z0 + jh2 ) − F (Z0 )
F  (Z0 ) = lim = −jFz2 (Z0 ) .
h2 →0 jh2
148 Chapter 7. Bicomplex Derivability and Differentiability

In conclusion, the complex partial derivatives of F with respect to z1 and z2 exist

(at Z0 ), and they verify the equality

F  (Z0 ) = Fz1 (Z0 ) = −jFz2 (Z0 ) . (7.22)

If we write F = f1 + jf2 , the complex partial derivatives of F at Z0 are therefore

given by

Fz1 (Z0 ) = f1,z


1

(Z0 ) + jf2,z 1
(Z0 ), Fz2 (Z0 ) = f1,z

2

(Z0 ) + jf2,z 2
(Z0 ) .

Now it is immediate to see that equality (7.22) is equivalent to the C(i)-complex

Cauchy-Riemann conditions for F at Z0 . 

Notice that the symbol f1,z 1
(and the other similarly defined symbols) denotes
in our situation the “authentic” complex partial derivative at a point, not the
formal operation

∂f 1 ∂f ∂f
:= −i
∂z1 2 ∂x1 ∂y1
defined on C 1 -functions. At the same time, these results indicate that bicomplex
functions which are derivable in a domain are related to holomorphic mappings of
two complex variables. We will come back in more detail to this issue at the end
of this section.

Corollary 7.3.7. Let F = f1 + jf2 be a bicomplex function derivable at Z0 , then

the real components of the functions f1 = f11 + if12 and f2 = f21 + if22 verify the
usual real Cauchy-Riemann system (at Z0 ) associated with each complex variable
z1 = x1 + iy1 and z2 = x2 + iy2 ; in complex notation this is equivalent to

∂F ∂F
(Z0 ) = (Z0 ) = 0 , (7.23)
∂z 1 ∂z 2

i.e.,
∂f1 ∂f1 ∂f2 ∂f2
(Z0 ) = (Z0 ) = (Z0 ) = (Z0 ) = 0 , (7.24)
∂z 1 ∂z 2 ∂z 1 ∂z 2
∂ ∂
where the symbols and are the commonly used formal operations on C(i)-
∂z 1 ∂z 2
valued functions of z1 and z2 , having partial derivatives at Z0 , namely,

∂f 1 ∂f ∂f
:= +i
∂z 1 2 ∂x1 ∂y1

and 
∂f 1 ∂f ∂f
:= +i .
∂z 2 2 ∂x2 ∂y2
7.3. Partial derivatives of bicomplex derivable functions 149

Proof. Let us employ equalities (7.16) involving the real partial derivatives of
F = f1 + jf2 at Z0 . The second equality in (7.16) is
∂F ∂F
(Z0 ) = −i (Z0 ) , (7.25)
∂x1 ∂y1
which is equivalent to (for simplicity we eliminate the explicit reference to Z0 )

∂f11 ∂f12 ∂f21 ∂f22 ∂f11 ∂f12 ∂f21 ∂f22
+i +j + ij = −i +i +j + ij .
∂x1 ∂x1 ∂x1 ∂x1 ∂y1 ∂y1 ∂y1 ∂y1
Because the functions fk are real, this is equivalent to the system
∂f11 ∂f12 ∂f11 ∂f12
= , =− ,
∂x1 ∂y1 ∂y1 ∂x1
∂f21 ∂f22 ∂f21 ∂f22
= , =− . (7.26)
∂x1 ∂y1 ∂y1 ∂x1
These are the real Cauchy-Riemann conditions for the complex functions f1 =
f11 + if12 and f2 = f21 + if22 , with respect to the complex variable z1 = x1 + iy1 .
In complex notation, if we write
equality (7.25) in terms of the complex differential
1 ∂ ∂ ∂
operator = +i , it becomes equivalent to the first part of (7.23).
2 ∂z 1 ∂x1 ∂y1
We repeat this reasoning for the equality
∂F ∂F
−j (Z0 ) = ij (Z0 ), (7.27)
∂x2 ∂y2
which, after division by −j, is equivalent to the second equality in (7.23). This
leads to the conclusion that f1 and f2 verify the real Cauchy-Riemann system
with respect to the variable z2 at Z0 . 
Of course, this corollary shows the relation between bicomplex derivability
and classical holomorphy in two complex variables.
Let us now express the bicomplex variable as Z = ζ1 +i ζ2 , and the bicomplex
function as F = ρ1 + iρ2 , where ρ1 = f11 + jf21 and ρ2 = f12 + jf22 are C(j)-valued
functions. We prove the following
Theorem 7.3.8. Consider a bicomplex function F derivable at Z0 . Then
1. The C(j)-complex partial derivatives Fζ (Z0 ) exist, for  = 1, 2.
2. The complex partial derivatives above verify the equality:

F  (Z0 ) = Fζ1 (Z0 ) = −iFζ2 (Z0 ) , (7.28)

which is equivalent to the C(j)-complex Cauchy-Riemann system (at Z0 ):

ρ1,ζ1 (Z0 ) = ρ2,ζ2 (Z0 ) , ρ1,ζ2 (Z0 ) = −ρ2,ζ1 (Z0 ) . (7.29)

150 Chapter 7. Bicomplex Derivability and Differentiability

Proof. In the definition of F  (Z0 ) we consider the limit of the difference quotient
as H → 0 first through invertible values of the form H = κ1 + i0, and then
through values of the form H = 0 + iκ2 , where κ1 , κ2 ∈ C(j). A computation as
in Theorem 7.3.6 leads to the existence of the complex partial derivatives of F at
Z0 with respect to the variables ζ1 , ζ2 ∈ C(j), and one can show that they verify
the equation:
F  (Z0 ) = Fζ1 (Z0 ) = −iFζ2 (Z0 ) , (7.30)
where

Fζ1 (Z0 ) = ρ1,ζ1 (Z0 ) + iρ2,ζ1 (Z0 ), Fζ2 (Z0 ) = ρ1,ζ2 (Z0 ) + iρ2,ζ2 (Z0 ) .

Now it is immediate to see that the equality (7.30) is equivalent to the sys-
tem (7.29). 
Corollary 7.3.9. If F = ρ1 + iρ2 is bicomplex derivable at Z0 , then the real com-
ponents of the functions ρ1 = f11 + jf21 and ρ2 = f12 + jf22 verify the usual real
Cauchy-Riemann system (at Z0 ) associated to each complex variable ζ1 = x1 + jx2
and ζ2 = y1 + jy2 ; in complex notation this is equivalent to
∂F ∂F
(Z0 ) = ∗ (Z0 ) = 0 , (7.31)
∂ζ1∗ ∂ζ2
i.e.,
∂ρ1 ∂ρ1 ∂ρ2 ∂ρ2
(Z0 ) = ∗ (Z0 ) = ∗ (Z0 ) = ∗ (Z0 ) = 0 . (7.32)
∂ζ1∗ ∂ζ2 ∂ζ1 ∂ζ2
Here
 
∂ 1 ∂ ∂ ∂ 1 ∂ ∂
∗ := +j , ∗ := +j .
∂ζ1 2 ∂x1 ∂x2 ∂ζ2 2 ∂y1 ∂y2

Proof. The following equalities from (7.16) among the real partial derivatives of
F
∂F ∂F ∂F ∂F
(Z0 ) = −j (Z0 ), −i (Z0 ) = ij (Z0 ) (7.33)
∂x1 ∂x2 ∂y1 ∂y2
are equivalent to (7.31). 
Remark 7.3.10. We showed in Example 7.1.2 that it is possible for a bicomplex
function to have partial derivatives with respect to z1 , z2 , and not to have partial
derivatives with respect to ζ1 , ζ2 . Now we see that if the function is bicomplex
derivable, then such a situation is not possible: a bicomplex derivable function has
complex partial derivatives with respect to the C(i)-complex variables z1 and z2
as well as with respect to the C( j)-variables ζ1 and ζ2 . Moreover, formulas (7.20)
and (7.28) show that such derivatives are related by

Fz1 (Z0 ) = Fζ1 (Z0 ) = −jFz2 (Z0 ) = −iFζ2 (Z0 ) .

7.3. Partial derivatives of bicomplex derivable functions 151

We express now the bicomplex variable as Z = z1 + iz2 , where z1 = x1 + ky2

and z2 = y1 + k(−x2 ) are hyperbolic numbers, and the function as F = f1 + if2 ,
where f1 = f11 + kf22 and f2 = f12 + k(−f21 ). We prove the following
Theorem 7.3.11. Consider a bicomplex function F = f1 + if2 derivable at a point
Z0 . Then
1. The hyperbolic partial derivatives Fz exist for  = 1, 2.
2. The partial derivatives above verify the equality

F  (Z0 ) = Fz1 (Z0 ) = −iFz2 (Z0 ) ,

which is equivalent to the following Cauchy-Riemann type system for the

hyperbolic components of a bicomplex derivable function:

f1,z1 (Z0 ) = f2,z2 (Z0 ) , f1,z2 (Z0 ) = −f2,z1 (Z0 ) . (7.34)

Proof. The existence of the hyperbolic partial derivatives of F = F (z1 , z2 ) is ob-

tained in a similar fashion as above, letting H → 0 through invertible hyperbolic
values H = h1 + i0 and then on paths of the form H = ih2 , where h1 and h2 are
invertible hyperbolic numbers. We obtain also:

F  (Z0 ) = Fz1 (Z0 ) = −iFz2 (Z0 ) ,

which is equivalent to system (7.34). 

Corollary 7.3.12. If F = f1 + if2 is derivable at Z0 , then the real components of
the hyperbolic functions f1 = f11 + kf22 and f2 = f12 + k(−f21 ) verify the Cauchy-
Riemann type systems with respect to both variables z1 , z1 ∈ D; in hyperbolic terms
this is equivalent to
∂F ∂F
(Z0 ) =  (Z0 ) = 0 , (7.35)
∂z1 ∂z2
i.e.,
∂f1 ∂f1 ∂f2 ∂f2
(Z0 ) =  (Z0 ) =  (Z0 ) =  (Z0 ) = 0 , (7.36)
∂z1 ∂z2 ∂z1 ∂z2
where
 
∂ 1 ∂ ∂ ∂ 1 ∂ ∂
= −k , = +k . (7.37)
∂z1 2 ∂x1 ∂y2 ∂z2 2 ∂y1 ∂x2

Proof. We use once again the following equalities from (7.16):

∂F ∂F ∂F ∂F
(Z0 ) = k (Z0 ), (Z0 ) = −k (Z0 ) , (7.38)
∂x1 ∂y2 ∂y1 ∂x2

where k = ij. Recalling formulas (7.37) we obtain equality (7.35). 

152 Chapter 7. Bicomplex Derivability and Differentiability

Remark 7.3.13. Appealing again to formulas (1.6)–(1.8) which deal with the com-
plex variables w1 , w2 , ω1 , ω2 and the hyperbolic variables w1 , w2 , one may wonder:
what about the Cauchy-Riemann conditions with respect to the corresponding par-
tial derivatives? Remark 7.1.3 explains how they can be obtained directly from the
previous statements. We omit the details.

Definition 7.3.14. Let F be a bicomplex function defined on a non-empty open set

Ω ⊂ BC. If F has bicomplex derivative at each point of Ω, we will say that F is a
bicomplex holomorphic, or BC-holomorphic, function.

Thus for a BC-holomorphic function F all the conclusions made in this section
hold in the whole domain. Theorem 7.3.6 says that F is holomorphic with respect
to z1 for any z2 fixed and F is holomorphic with respect to z2 for any z1 fixed.
Thus, see for instance [39, pages 4-5], F is holomorphic in the classical sense of two
complex variables. This implies immediately many quite useful properties of F , in
particular, it is of class C ∞ (Ω), and the reader may compare this with Remark 7.2.4
where we were able, working with just one point, not with a domain, to state a
weakened continuity at the point.

7.4 Interplay between real differentiability and

derivability of bicomplex functions
7.4.1 Real differentiability in complex and hyperbolic terms.
We begin now assuming that Ω is an open set in BC and F : Ω ⊂ BC → BC
is a bicomplex function of class C 1 (Ω) with respect to the canonical coordinates
x1 , y1 , x2 , y2 . We are going to work with BC-holomorphic functions and we want
to determine the place that BC-holomorphic functions occupy among the bicom-
plex C 1 -functions. The condition F ∈ C 1 (Ω, BC) ensures that F is real differen-
tiable for any Z ∈ Ω, i.e., that

∂F ∂F
F (Z + H) − F (Z) = (Z)h11 + (Z)h12
∂x1 ∂y1
∂F ∂F
+ (Z)h21 + (Z)h22 + o(H) , (7.39)
∂x2 ∂y2

where Z = x1 + iy1 + jx2 + ky2 , H = h11 + ih12 + jh21 + kh22 . As a matter

of fact, this formula does not depend on how the function F and the variable Z
are written, and therefore it will be quite helpful to analyze the structure of such
functions.
First of all, let us write (7.39) in terms of the C(i)-complex variables h1 :=
h11 + ih12 and h2 := h21 + ih22 , so that
7.4. Real differentiability and derivability 153

h1 + h1 h1 − h 1
h11 = , h12 = ,
2 2i
h2 + h2 h2 − h 2
h21 = , h22 = .
2 2i
Using this in (7.39) and grouping adequately, we obtain:

1 ∂F ∂F
F (Z + H) − F (Z) = (Z) − i (Z) · h1
2 ∂x1 ∂y1
 
1 ∂F ∂F 1 ∂F ∂F
+ (Z) + i (Z) · h1 + (Z) − i (Z) · h2
2 ∂x1 ∂y1 2 ∂x2 ∂y2

1 ∂F ∂F
+ (Z) + i (Z) · h2 + o(H) .
2 ∂x2 ∂y2
If we employ the C(i)-complex variables z1 = x1 + iy1 and z2 = x2 + iy2 , and the
usual complex differential operators
 
∂ 1 ∂ ∂ ∂ 1 ∂ ∂
:= −i , := +i ,
∂z1 2 ∂x1 ∂y1 ∂z 1 2 ∂x1 ∂y1
 
∂ 1 ∂ ∂ ∂ 1 ∂ ∂
:= −i , := +i , (7.40)
∂z2 2 ∂x2 ∂y2 ∂z 2 2 ∂x2 ∂y2
we obtain the formula
∂F ∂F
F (Z + H) − F (Z) = (Z) · h1 + (Z) · h1
∂z1 ∂z 1
∂F ∂F
+ (Z) · h2 + (Z) · h2 + o(H) . (7.41)
∂z2 ∂z 2
We emphasize that (7.41) does not express a new notion; indeed, this is simply
the condition of real differentiability for a C 1 -bicomplex function, although it is
now expressed in C(i)-complex terms.
Note that in Section 7 we introduced the symbols Fz1 (Z) and Fz2 (Z) in-
∂F ∂F
stead of the symbols (Z) and (Z) because the former are complex partial
∂z1 ∂z2
derivatives, defined, as usual, as limits of suitable difference quotients, meanwhile
the latter indicates well known operators acting on C 1 -functions. The relationship
between these two notions is clarified by the following definition and theorem.
Definition 7.4.1. A bicomplex C 1 -function F is called C(i)-complex differentiable
if
F (Z + H) − F (Z) = Fz1 (Z) · h1 + Fz2 (Z) · h2 + o(H) .
Theorem 7.4.2. A C 1 -bicomplex function F is C(i)-complex differentiable if and
only if both its components f1 , f2 are holomorphic functions in the sense of two
complex variables.
154 Chapter 7. Bicomplex Derivability and Differentiability

∂
Proof. The partial derivative Fz1 (Z) exists in Ω if and only if the operator
∂z 1
∂
(which one can think of as a dual to the operator ) annihilates the function
∂z1
F ; that is, if and only if F is holomorphic as a function of z1 ; this can be proved
by taking h2 = 0 and h1 = 0 in (7.41). What remains is

∂F ∂F
F (Z + H) − F (Z) = F (Z + h1 ) − F (Z) = (Z) · h1 + (Z) · h1 + o(H) ,
∂z1 ∂z 1

h1
and since has no limit when h1 → 0, then we conclude that Fz1 (Z) exists if
h1
∂F ∂F
and only if (Z) = 0. In this case, Fz1 (Z) = (Z).
∂z 1 ∂z1

Similarly, the partial derivative Fz2 (Z) exists in Ω if and only if the operator
∂
annihilates the function F ; this is because we can take now h1 = 0, h2 = 0
∂z 2
∂F
in (7.41). Therefore Fz2 (Z) = (Z).
∂z2
Finally, we can assume both conditions

∂F ∂F
(Z) = (Z) = 0
∂z 1 ∂z 2

to be fulfilled in Ω, with (7.41) becoming

F (Z + H) − F (Z) = Fz1 (Z) · h1 + Fz2 (Z) · h2 + o(H) . (7.42)

This concludes the proof. 

Note that for an arbitrary C(i)-complex differentiable bicomplex function, in
general, there is no relation between its complex partial derivatives.
Very similar calculations can be made if we write the bicomplex number as
Z = ζ1 + iζ2 and the C( j)-complex increments as κ1 := h11 + jh21 and κ2 :=
h12 + jh22 . If one follows the steps indicated above, one eventually obtains the
analogue of (7.41):

∂F ∂F
F (Z + H) − F (Z) = (Z)κ1 + (Z)κ1
∂ζ1 ∂ζ 1
∂F ∂F
+ (Z)κ2 + (Z)κ2 + o(H) , (7.43)
∂ζ2 ∂ζ 2

∂
where the expressions (and similar) are the usual complex differential op-
∂ζ1
erators in C(j). Again, this simply represents the real differentiability of a C 1 -
bicomplex function in C(j)-complex terms.
7.4. Real differentiability and derivability 155

The same analysis as above applies to equation (7.43). In particular, the

definition of the C(j) complex differentiability of a bicomplex function is
F (Z + H) − F (Z) = Fζ1 (Z) · κ1 + Fζ2 (Z) · κ2 + o(H) .
Notice that Remark 7.1.3 explains why the other complex variables w1 , w2 , ω1 , ω2
do not present any interest in the analysis of the increment of the function.
Our last step consists in writing Z = z1 + iz2 , where z1 := x1 + k y2 and
z2 := y1 + k(−x2 ) are hyperbolic variables. The hyperbolic increments are h1 :=
h11 + kh22 and h2 := h12 + k(−h21 ). Then, using the formulas
h1 + h1 h1 − h1
h11 = , h22 = ,
2 2k

h2 + h2 h2 − h2
h12 = , h21 =− ,
2 2k
we regroup the right-hand side of (7.39) as follows:
F (Z + H) − F (Z)
∂F h1 + h1 ∂F h2 + h2
= (Z) · + (Z) ·
∂x1 2 ∂y1 2
∂F h2 − h2 ∂F h1 − h1
− (Z) · + (Z) · + o(H)
∂x 2k ∂y2 2k
 2 
1 ∂F ∂F 1 ∂F ∂F
= (Z) + k (Z) h1 + (Z) − k (Z) h1
2 ∂x1 ∂y2 2 ∂x1 ∂y2
 
1 ∂F ∂F 1 ∂F ∂F
+ (Z) − k (Z) h2 + (Z) + k (Z) h2
2 ∂y1 ∂x2 2 ∂y1 ∂x2
+ o(H) .
The “hyperbolic” differential operators appearing above are consistent with the
ones in the context of hyperbolic analysis. For the hyperbolic variable z = x + k y,
the formal hyperbolic partial derivatives are given by the formulas:
 
∂ 1 ∂ ∂ ∂ 1 ∂ ∂
= +k , = − k ,
∂z 2 ∂x ∂y ∂z 2 ∂x ∂y
where z = x − k y is the hyperbolic conjugate of z. Therefore, in terms of the
hyperbolic variables z1 = x1 + k y2 and z2 = y1 + k(−x2 ) and the corresponding
hyperbolic differential operators, we obtain:
F (Z + H) − F (Z)
∂F ∂F ∂F ∂F
= (Z)h1 +  (Z)h1 + (Z)h2 +  (Z)h2 + o(H) , (7.44)
∂z1 ∂z1 ∂z2 ∂z2
which is a hyperbolic reformulation of the real differentiability of a C 1 -bicomplex
function.
156 Chapter 7. Bicomplex Derivability and Differentiability

Again, we can apply to the equation (7.44) a similar reasoning as made

above. In particular, the definition of the hyperbolic differentiability of a bicomplex
function is
∂F ∂F
F (Z + H) − F (Z) = (Z)h1 + (Z)h2 + o(H) .
∂z1 ∂z2
We are not aware of any other work that studies what could be called hyperbolic
holomorphic mappings from D2 to D2 , that is, pairs of holomorphic functions of
two hyperbolic variables.

7.4.2 Real differentiability in bicomplex terms

Formula (7.39) as well as any of the formulas (7.41), (7.43) and (7.44) expresses
the real differentiability of a bicomplex function, although written in different
languages: the first of them is in real language, the next two in complex (C(i)-
and C(j)-) language and the last is given in the hyperbolic one. Now we are going
to see what the bicomplex language will give.
Let us first consider the bicomplex increment H = h1 + jh2 , for which we
have:
H + H† H − H†
h1 = , h2 = ,
2 2j
H + H∗ H − H∗
h1 = , h2 = . (7.45)
2 2j
In this setup, formula (7.41) becomes:
 
1 ∂F ∂F 1 ∂F ∂F
F (Z + H) − F (Z) = −j (Z) · H + +j (Z) · H †
2 ∂z1 ∂z2 2 ∂z1 ∂z2
 
1 ∂F ∂F 1 ∂F ∂F
+ −j (Z) · H + −j (Z) · H ∗
2 ∂z 1 ∂z 2 2 ∂z 1 ∂z 2
+ o(H) . (7.46)
We introduce the following bicomplex differential operators:
 
∂ 1 ∂ ∂ ∂ 1 ∂ ∂
:= −j , := + j ,
∂Z 2 ∂z1 ∂z2 ∂Z † 2 ∂z1 ∂z2
 
∂ 1 ∂ ∂ ∂ 1 ∂ ∂
:= −j , := + j . (7.47)
∂Z 2 ∂z 1 ∂z 2 ∂Z ∗ 2 ∂z 1 ∂z 2
We obtain the following intrinsic expression of the real differentiability of the
bicomplex function F in terms of bicomplex differential operators and variables:
F (Z + H) − F (Z)
∂F ∂F ∂F ∂F
= (Z)H + (Z)H † + (Z)H + (Z)H ∗ + o(H) . (7.48)
∂Z ∂Z † ∂Z ∂Z ∗
7.4. Real differentiability and derivability 157

While real differentiability uniquely defines the coefficients in (7.48), one may think
that if we had begun with another writing of Z and H, then formula (7.48) would
be different, that is, other operators would have appeared in it. Direct computa-
tions however confirm that no matter in which form we write the functions and the
variables (recall that the operators in (7.47) are given in terms of C(i)-complex dif-
∂ ∂ ∂ ∂
ferential operators) the operators , †
, , and in the right-hand side
∂Z ∂Z ∂Z ∂Z ∗
of (7.48) are uniquely defined. But this requires us to clarify what is meant by this
“uniqueness”. These are the same operators but only when acting on bicomplex
functions, without taking into account any concrete intrinsic substructure in BC.
For instance, if the functions are considered C2 (i)-valued or C2 (j)-valued, then the
operators are of course different; this is just because they act on objects of differ-
ent nature. For this reason, one should be careful when working with bicomplex
functions and operators, having in mind all the time what is exactly the structure
of BC which is of interest for a specific goal.
We write here each operator in the various possible ways:
 
∂ 1 ∂ ∂ 1 ∂ ∂
= −j = −i
∂Z 2 ∂z1 ∂z2 2 ∂ζ1 ∂ζ2
 
1 ∂ ∂ 1 ∂ ∂
= −i = +k
2 ∂z1 ∂z2 2 ∂w1 ∂w2
 
1 ∂ ∂ 1 ∂ ∂
= +k = +j
2 ∂ω1 ∂ω2 2 ∂w1 ∂w2

1 ∂ ∂ ∂ ∂
= −i −j +k , (7.49)
4 ∂x1 ∂y1 ∂x2 ∂y2

 
∂ 1 ∂ ∂ 1 ∂ ∂
= + j = − i
∂Z † 2 ∂z1 ∂z2 2 ∂ζ1∗ ∂ζ2∗
 
1 ∂ ∂ 1 ∂ ∂
= − i = − k
2 ∂z1 ∂z2 2 ∂w1 ∂w2
 
1 ∂ ∂ 1 ∂ ∂
= − k = + j
2 ∂ω1∗ ∂ω2∗ 2 ∂w1 ∂w2

1 ∂ ∂ ∂ ∂
= −i +j −k , (7.50)
4 ∂x1 ∂y1 ∂x2 ∂y2
 
∂ 1 ∂ ∂ 1 ∂ ∂
= −j = +i
∂Z 2 ∂z 1 ∂z 2 2 ∂ζ1 ∂ζ2
 
1 ∂ ∂ 1 ∂ ∂
= + i = − k
2 ∂z1 ∂z2 2 ∂w1 ∂w2
 
1 ∂ ∂ 1 ∂ ∂
= −k = − j
2 ∂ω1 ∂ω2 2 ∂w1 ∂w2
158 Chapter 7. Bicomplex Derivability and Differentiability

1 ∂ ∂ ∂ ∂
= +i −j −k , (7.51)
4 ∂x1 ∂y1 ∂x2 ∂y2
and
 
∂ 1 ∂ ∂ 1 ∂ ∂
= +j = +i ∗
∂Z ∗ 2 ∂z 1 ∂z 2 2 ∂ζ1∗ ∂ζ2
 
1 ∂ ∂ 1 ∂ ∂
= +i = +k
2 ∂z1 ∂z2 2 ∂w1 ∂w2
 
1 ∂ ∂ 1 ∂ ∂
= + k = − j
2 ∂ω1∗ ∂ω2∗ 2 ∂w1 ∂w2

1 ∂ ∂ ∂ ∂
= +i +j +k . (7.52)
4 ∂x1 ∂y1 ∂x2 ∂y2
∂
Again, take a bicomplex function F and apply to it, say, the operator ; the
∂Z ∗
∂F
resulting function is a bicomplex function. But if we omit the bicomplex
∂Z ∗
structure and consider F to be a mapping from C2 (i) to C2 (i), then such an F
does understand already what the action of the operator
⎛ ⎞
∂ ∂
 −
1 ∂ ∂ 1 ⎜ ∂z 1 ∂z 2 ⎟
+j = ⎜ ⎝
⎟
2 ∂z 1 ∂z 2 2 ∂ ∂ ⎠
∂z 2 ∂z 1

1 ∂ ∂
means, but it does not understand the action of the operator + i .
2 ∂ζ1∗ ∂ζ2∗
As a consequence of the previous discussion, we obtain the following result:
Theorem 7.4.3. Let F ∈ C 1 (Ω, BC): if F is BC-holomorphic, then
∂F ∂F ∂F
(Z) = (Z) = (Z) = 0 . (7.53)
∂Z † ∂Z ∂Z ∗

holds on Ω.
Proof. Since F is BC-holomorphic, formula (7.9) holds for all H ∈ / S0 . But F is
a C 1 -function, hence (7.48) holds as well for any H = 0, thus both formulas hold
for non-zero-divisors. Then (7.53) follows directly by recalling that both (7.9)
and (7.48) are unique representations for a given function F , and by comparing
them. 
Remark 7.4.4. As we will show later the converse of this result is true as well, but
we need some additional steps before we can prove it.
In order to have more consistency with the previous reasonings of this section
and in analogy with the cases of functions of real or complex variables, we introduce
the following definition.
7.4. Real differentiability and derivability 159

Definition 7.4.5. A bicomplex function F ∈ C 1 (Ω, BC) is called bicomplex- (BC-)

differentiable on Ω if

F (Z + H) − F (Z) = AZ · H + α(H)H (7.54)

with α(H) → 0 when H → 0 and AZ a bicomplex constant.

Note that in this definition H is allowed to be a zero-divisor but taking H
in (7.54) to be any non-zero-divisor, we see from (7.9) that BC-differentiability
implies BC-derivability. The reciprocal statement is more delicate and will not be
treated immediately.
It turns out that Theorem 7.4.3 has many deep and far-reaching consequences
which we will discuss in the next section.

7.5 Bicomplex holomorphy versus holomorphy in two

(complex or hyperbolic) variables
Take a BC-holomorphic function F , which we write as F = f1 + jf2 on a domain
Ω ⊂ BC with the independent variable written as Z = z1 + jz2 . By Theorem 7.4.3,
and Remark 7.4.4, this is equivalent to saying that F verifies in Ω the system

∂F ∂F ∂F
= = = 0.
∂Z † ∂Z ∂Z ∗
For the operators involved we use the appropriate representations from the table
∂
in the previous section. Using such a representation for the operators and
∂Z
∂
leads to the system
∂Z ∗
∂F ∂F ∂F ∂F
−j = 0, +j = 0,
∂z 1 ∂z 2 ∂z 1 ∂z 2
∂F ∂F
implying that = 0 = . The latter is equivalent to the holomorphy, in
∂z 1 ∂z 2
the sense of complex functions of two C(i)-complex variables, of the components
f1 , f2 of the function F . Thus F can be seen as a holomorphic mapping from
∂F
Ω ⊂ C2 (i) → C2 (i). But we still have more information. Since = 0, then
∂Z †
F = f1 + jf2 verifies

∂F 1 ∂ ∂
(Z) = +j (f1 + jf2 )(Z)
∂Z † 2 ∂z1 ∂z2
 
1 ∂f1 ∂f2 ∂f2 ∂f1
= − (Z) + j (Z) + (Z) = 0.
2 ∂z1 ∂z2 ∂z1 ∂z2
160 Chapter 7. Bicomplex Derivability and Differentiability

Thus
∂f1 ∂f2 ∂f1 ∂f2
= , =− , (7.55)
∂z1 ∂z2 ∂z2 ∂z1
that is, the complex partial derivatives of the holomorphic functions f1 , f2 are not
independent; they are tied by the Cauchy-Riemann type conditions (7.55). We
reformulate the reasoning as
Proposition 7.5.1. A function F = f1 + jf2 : Ω ⊂ BC → BC is BC-holomorphic if
and only if, seen as a mapping from Ω ⊂ C2 (i) → C2 (i), it is a holomorphic map-
ping with its components related by the Cauchy-Riemann type conditions (7.55).
In other words, the theory of bicomplex holomorphic functions can be seen
as a theory of a proper subset of holomorphic mappings in two complex variables.
Each equation in Theorem 7.4.3 plays a different role: two of them together guar-
antee the holomorphy of the C(i)-complex components and the third one provides
the relation between them.
∂
It is worth noting that the operator arises in the works of J. Ryan [69] on
∂Z †
complex Clifford analysis as a Cauchy-Riemann operator which is defined directly
on holomorphic mappings with values in a complex Clifford algebra.
Next, take a bicomplex holomorphic function F in the form F = g1 + ig2 ,
where g1 and g2 take values in C(j) and we write now Z = ζ1 + iζ2 , then the
corresponding differential operators are:

∂ 1 ∂ ∂
= − i ,
∂Z † 2 ∂ζ1∗ ∂ζ2∗

∂ 1 ∂ ∂
= +i ,
∂Z 2 ∂ζ1 ∂ζ2

∂ 1 ∂ ∂
= + i .
∂Z ∗ 2 ∂ζ1∗ ∂ζ2∗
Using again Theorem 7.4.3 and Remark 7.4.4 we have that to be a bicomplex
holomorphic function means for F that

∂F 1 ∂F ∂F
= − i ∗ = 0,
∂Z † 2 ∂ζ1∗ ∂ζ2

∂F 1 ∂F ∂F
= +i = 0,
∂Z 2 ∂ζ1 ∂ζ2

∂F 1 ∂F ∂F
∗
= ∗ + i ∗ = 0.
∂Z 2 ∂ζ1 ∂ζ2
The first and the third equations together give, again, that the components g1 and
g2 of F are holomorphic functions of the C(j)-complex variables ζ1 and ζ2 , while
the second equation gives:
∂g1 ∂g2 ∂g1 ∂g2
= =− . (7.56)
∂ζ1 ∂ζ2 ∂ζ2 ∂ζ1
7.5. Bicomplex holomorphy versus holomorphy in two variables 161

Proposition 7.5.2. A function F = g1 + ig2 : Ω ⊂ BC → BC is BC-holomorphic if

and only if, seen as a mapping from Ω ⊂ C2 (j) → C2 (j), it is a holomorphic map-
ping with its components related by the Cauchy-Riemann type conditions (7.56).

Finally, take a BC-holomorphic function F : Ω ⊂ BC → BC in the form

F = u1 + iu2 , where u1 , u2 take values in D and write Z = z1 + iz2 , then the
corresponding operators are:


∂ 1 ∂ ∂
= − i ,
∂Z † 2 ∂z1 ∂z2

∂ 1 ∂ ∂
= + i ,
∂Z 2 ∂z1 ∂z2

∂ 1 ∂ ∂
= + i .
∂Z ∗ 2 ∂z1 ∂z2

Using again Theorem 7.4.3 and Remark 7.4.4 we have that to be a BC-holomorphic
function means for F that

∂F 1 ∂F ∂F
= − i = 0,
∂Z † 2 ∂z1 ∂z2

∂F 1 ∂F ∂F
= + i = 0,
∂Z 2 ∂z1 ∂z2

∂F 1 ∂F ∂F
= + i = 0.
∂Z ∗ 2 ∂z1 ∂z2

The first and second equations together imply that the components u1 and u2 of
F are holomorphic functions of hyperbolic variables z1 and z2 , meanwhile the last
equation gives the Cauchy-Riemann type conditions:

∂u1 ∂u2 ∂u1 ∂u2

= =− . (7.57)
∂z1 ∂z2 ∂z2 ∂z1

They look exactly as their antecedent in one complex variable, but this is a totally
different thing: we deal here with D-valued functions of two hyperbolic variables
and with hyperbolic partial derivatives.

Proposition 7.5.3. A function F = u1 + iu2 : Ω ⊂ BC → BC is BC-holomorphic if

and only if, seen as a mapping from Ω ⊂ D2 → D2 , it is a holomorphic mapping
with its components related by the Cauchy-Riemann type conditions (7.57).
162 Chapter 7. Bicomplex Derivability and Differentiability

7.6 Bicomplex holomorphy: the idempotent

representation
Take a bicomplex function F : Ω ⊂ BC → BC on a domain Ω. We write all the
bicomplex numbers involved in C(i)-idempotent form, for instance,
Z = β1 e + β2 e† = (1 + im1 )e + (2 + im2 )e† ,
F (Z) = G1 (Z)e + G2 (Z)e† ,
H = η1 e + η2 e† = (u1 + iv1 )e + (u2 + iv2 )e† .
Let us introduce the sets

Ω1 := {β1  β1 e + β2 e† ∈ Ω} ⊂ C(i) (7.58)
and 
Ω2 := {β2  β1 e + β2 e† ∈ Ω} ⊂ C(i). (7.59)
It is easy to prove that Ω1 and Ω2 are domains in C(i), see book [56].
We assume that F ∈ C 1 (Ω) where the real partial derivatives are taken
with respect to the “idempotent real variables”: 1 , m1 , 2 , m2 ; how this is related
with the cartesian variables x1 , y1 , x2 , y2 will be discussed later. The condition
F ∈ C 1 (Ω) ensures the real differentiability of F in Ω:
∂F ∂F
F (Z + H) − F (Z) = (Z) · u1 + (Z) · v1
∂1 ∂m1
∂F ∂F
+ (Z) · u2 + (Z) · v2 + o(H) (7.60)
∂2 ∂m2
for H → 0. We are going to follow Section 7.4 so we omit many details. First of
all, let us translate formula (7.60) in C(i)-complex language: since
1 1
u1 = (η1 + η 1 ); u2 = (η2 + η 2 );
2 2
i i
v1 = (η 1 − η1 ); v2 = (η 2 − η2 ) ,
2 2
then
∂F 1 ∂F i
F (Z + H) − F (Z) = (Z) · (η1 + η 1 ) + (Z) · (η 1 − η1 )
∂1 2 ∂m1 2
∂F 1 ∂F i
+ (Z) · (η2 + η 2 ) + (Z) · (η 2 − η2 ) + o(H)
∂2 2 ∂m2 2
 
1 ∂F ∂F 1 ∂F ∂F
= η1 (Z) − i (Z) + η 1 (Z) + i (Z)
2 ∂1 ∂m1 2 ∂1 ∂m1
 
1 ∂F ∂F 1 ∂F ∂F
+ η2 (Z) − i (Z) + η 2 (Z) + i (Z) + o(H)
2 ∂2 ∂m2 2 ∂2 ∂m2
∂F ∂F ∂F ∂F
=: η1 (Z) + η 1 (Z) + η2 (Z) + η 2 (Z) + o(H) .
∂β1 ∂β 1 ∂β2 ∂β 2
7.6. Bicomplex holomorphy: the idempotent representation 163

Remark 7.6.1. Note that the above calculations show that, as is well known, the
bicomplex function F of class C 1 , seen as a mapping from C2 (i) → C2 (i), is holo-
∂F
morphic with respect to βq (q = 1, 2) if and only if (Z) = 0 in Ω. Note that if
∂β q
F is a bicomplex function, and we express it in cartesian coordinates, it turns out
that F is BC-holomorphic if and only if its components are holomorphic as func-
tions of two complex variables and satisfy a Cauchy-Riemann type relation between
them. As we will show later, this is definitely not the case when we express F in
the idempotent representation. In this case, BC-holomorphy will be equivalent to
the requirement that each component is a holomorphic function of a single complex
variable and there are no relations between the components.
We emphasize that now we are considering the identification between BC
and C2 (i),
Z = β1 e + β2 e† ←→ (β1 , β2 ) ∈ C2 (i) ,
where however the basis in C2 (i) is not the canonical basis {1, j}, but rather the
idempotent basis {e, e† }.
For the next step recall the formulas
H = η 1 e + η 2 e† , H † = η 2 e + η 1 e† ,
H = η 2 e + η 1 e† , H ∗ = η 1 e + η 2 e† ,
which imply that
η1 = He + H † e† , η 1 = H ∗ e + He† ,
η2 = H † e + He† , η 2 = He + H ∗ e† .
The condition of real differentiability after substitutions becomes:
F (Z + H) − F (Z)
 
∂F ∂F ∂F ∂F
=H (Z)e + (Z)e† + H † (Z)e + (Z)e†
∂β1 ∂β2 ∂β2 ∂β1 (7.61)
 
∂F ∂F ∂F ∂F
+H (Z)e + (Z)e† + H ∗ (Z)e + (Z)e† + o(H) .
∂β 2 ∂β 1 ∂β 1 ∂β 2
Note that the expressions in the parentheses are not, yet, the idempotent forms
of anything, since the coefficients of e and e† are bicomplex numbers, not C(i)-
complex numbers. Thus we are required to make one more step. Using the formula
F = G1 e + G2 e† we arrive at
F (Z + H) − F (Z)
 
∂G1 ∂G2 † † ∂G1 ∂G2 †
=H (Z)e + (Z)e + H (Z)e + (Z)e
∂β1 ∂β2 ∂β2 ∂β1
 
∂G1 ∂G2 † ∗ ∂G1 ∂G2 †
+H (Z)e + (Z)e + H (Z)e + (Z)e + o(H) .
∂β 2 ∂β 1 ∂β 1 ∂β 2
(7.62)
164 Chapter 7. Bicomplex Derivability and Differentiability

This formula is valid for any F in C 1 (Ω), so let us analyze how BC-holomorphic
functions are singled out among those of class C 1 .
Theorem 7.6.2. The C 1 -function F is BC-holomorphic if and only if the three
bicomplex coefficients of H † , H and H ∗ in (7.62) are all zero for any Z in Ω.
Proof. The if direction follows as in Theorem 7.4.3. Specifically, since F is BC-
holomorphic, formula (7.9) holds for all H ∈ / S0 . But F is a C 1 -function, hence
(7.62) holds as well for any H = 0, thus both formulas hold for non-zero-divisors.
Then the result follows directly by recalling that both (7.9) and (7.62) are unique
representations for a given function F , and by comparing them.
In order to prove the only if, it is helpful to write explicitly the meaning of
the vanishing of these coefficients, namely:

∂G1 ∂G2
(Z)e + (Z)e† = 0 ,
∂β2 ∂β1
∂G1 ∂G2
(Z)e + (Z)e† = 0 ,
∂β 2 ∂β 1
∂G1 ∂G2
(Z)e + (Z)e† = 0 . (7.63)
∂β 1 ∂β 2

Now note that the second and the third equations, because of the independence
of e and e† , impose that G1 and G2 are C(i)-valued holomorphic functions of the
complex variables β1 , β2 and thus they have authentic complex partial derivatives.
What is more, the first equation in (7.63) says that one of the partial derivatives
∂G1 ∂G2
of each G1 and G2 is identically zero: (Z) = 0, (Z) = 0 for any Z ∈ Ω.
∂β2 ∂β1
Hence, using (7.58) and (7.59), G1 is a holomorphic function of the single variable
β1 ∈ Ω1 and G2 is a holomorphic function of the single variable β2 ∈ Ω2 . We now
want to show that these equations imply that F is BC-holomorphic. But in fact,
because of these equations, we have that for any invertible H there holds:

F (Z + H) − F (Z) G1 (β1 + η1 ) − G1 (β1 ) G2 (β2 + η2 ) − G2 (β2 ) †

= e+ e ,
H η1 η2

where Z is an arbitrary point in Ω.

Now, by the properties of G1 and G2 we deduce that the right-hand side has,
for S0  H → 0, the limit G1 (β1 )e + G2 (β2 )e† , which concludes the proof: the
limit in the left-hand side exists also for any Z ∈ Ω with S0  H → 0 and it
coincides with the derivative F  (Z) making F BC-holomorphic in Ω. 
As a matter of fact, the proof allows us to make a more precise characteri-
zation of C 1 -functions which are BC-holomorphic.
Theorem 7.6.3. A bicomplex function F = G1 e + G2 e† : Ω ⊂ BC → BC of class
C 1 is BC-holomorphic if and only if the following two conditions hold:
7.6. Bicomplex holomorphy: the idempotent representation 165

(I) The component G1 , seen as a C(i)-valued function of two complex variables

(β1 , β2 ) is holomorphic; moreover, it does not depend on the variable β2 and
thus G1 is a holomorphic function of the variable β1 .
(II) The component G2 , seen as a C(i)-valued function of two complex variables
(β1 , β2 ) is holomorphic; moreover, it does not depend on the variable β1 and
thus G2 is a holomorphic function of the variable β2 .

Remark 7.6.4. The functions G1 and G2 are independent in the sense that there
are no Cauchy-Riemann type conditions relating them.

We are in a position now to prove that the converse to Theorem 7.4.3 is true
as well.

Theorem 7.6.5. Given F ∈ C 1 (Ω, BC), then condition (7.53) implies that F is
BC-holomorphic.

Proof. If (7.53) holds, then a direct computation shows that all the three formulas
in (7.63) are true, and by Theorem 7.6.2 F is BC-holomorphic. 

The direct computation mentioned above is quite useful and instructive and
we will perform it later.

Corollary 7.6.6. Let F be a BC-holomorphic function in Ω, then F is of the form

F (Z) = G1 (β1 )e + G2 (β2 )e† with Z = β1 e + β2 e† ∈ Ω and its derivative is given
by
F  (Z) = G1 (β1 )e + G2 (β2 )e† .

Taking into account the relations between β1 , β2 and the cartesian compo-
nents z1 , z2 , we have also that

F  (z1 + jz2 ) = G1 (z1 − iz2 )e + G2 (z1 + iz2 )e† ;

F  (Z) = G1 (Ze + Z † e† )e + G2 (Z † e + Ze† )e† .

This implies that a BC-holomorphic function has derivatives of any order and
(n) (n)
F (n) (Z) = G1 (β1 ) e + G2 (β2 ) e†
(n) (n)
= G1 (Ze + Z † e† ) e + G2 (Z † e + Ze† ) e† .

Remark 7.6.7. Although formula (7.62) is quite similar to formula (7.48) its con-
sequences for the function F are paradoxically different: while formula (7.48) has
allowed us to conclude that the cartesian components f1 , f2 are holomorphic func-
tions of two complex variables which are not independent, formula (7.62) explains
to us that the idempotent components G1 , G2 are usual holomorphic functions of
one complex variable which are, besides, independent.
166 Chapter 7. Bicomplex Derivability and Differentiability

Remark 7.6.8. We have proved that if F is BC-holomorphic, then for any Z =

β1 e + β2 e† ∈ Ω it is of the form

F (Z) = G1 (β1 )e + G2 (β2 )e† .

" := Ω1 · e +
But the right-hand side of the latter is well-defined on the wider set Ω
†
Ω2 · e ⊃ Ω (in general, this inclusion is proper), with the notations as in (7.58)
and (7.59). Moreover, by Theorem 7.6.3 the function F" defined by

F" (Z) := G1 (β1 )e + G2 (β2 )e† , ",

Z∈Ω

is BC-holomorphic in Ω. " Since F"  ≡ F we see that, unlike what happens in
Ω
the complex case, not every domain in BC is a domain of BC-holomorphy: every
function which is BC-holomorphic in a domain Ω extends BC-holomorphically up
to the minimal set of the form X1 · e + X2 · e† containing Ω. One can compare this
with [81].
Remark 7.6.9. We recall that if Ω ⊂ C(i) is an open set bounded by a simple closed
curve, then there exists a holomorphic function f on Ω with the following proper-
" is any open set which strictly contains Ω, then there is no holomorphic
ty: if Ω
function f" on Ω
" such that f" restricted to Ω equals f . Usually Ω is called a domain
of holomorphy for f .
Consider now a bicomplex holomorphic function F on a domain Ω. Suppose
that Ω1 and Ω2 are domains of holomorphy for G1 and G2 , respectively, i.e., G1
and G2 cannot be holomorphically extended to any bigger open set in C(i). Then
the bicomplex function F = G1 e + G2 e† , defined on Ω " := Ω1 · e + Ω2 · e† cannot be
BC-holomorphically extended to any open set containing Ω. " Combining this fact
with the comments of the previous Remark 7.6.8, we can say that Ω " is a domain
of bicomplex holomorphy for F .
Remark 7.6.10. The same analysis can be done for the idempotent representation
with C(j) coefficients.
Remark 7.6.11. Recall that at the beginning of this chapter we worked with the
cartesian representation of bicomplex numbers and we investigated many proper-
ties of derivable bicomplex functions, in particular, such functions proved to have
complex partial derivatives with respect to z1 and z2 . This approach fails immedi-
ately when one tries to apply it to the case of the idempotent representation: this is
because the definition of the derivative excludes precisely the values of H which are
necessary for the complex partial derivatives with respect to β1 , β2 . But in the proof
of Theorem 7.4.3 we have shown, as a matter of fact, that such partial derivatives
∂F
of a BC-holomorphic functions do exist and, moreover, (Z) = G1 (β1 ) · e and
∂β1
∂F
(Z) = G2 (β2 ) · e† .
∂β2
7.7. Cartesian versus idempotent representations in BC-holomorphy 167

Remark 7.6.12. Using the results of this section, we can give the proofs of the
statements of Theorem 7.2.7 in terms of the idempotent writing of the bicomplex
elementary functions. For example, to prove that the bicomplex exponential func-
tion F (Z) = eZ is BC-holomorphic, we write:

F (Z) = eZ = eβ1 e + eβ2 e† = G1 (β1 )e + G2 (β2 )e† ,

for all Z = β1 e + β2 e† ∈ BC. But G1 is a complex holomorphic function depend-

ing on β1 , similar for G2 , thus Theorem 7.6.3 guarantees that F is a bicomplex
holomorphic function. Besides,

(eZ ) = (eβ1 ) e + (eβ2 ) e† = eβ1 e + eβ2 e† = eZ .

7.7 Cartesian versus idempotent representations in BC-

holomorphy
In the previous sections we worked with the real linear space R4 which was en-
dowed sometimes with the standard coordinates x := (x1 , y1 , x2 , y2 ) and thus
we denote it by R4x , and sometimes endowed with the idempotent coordinates
 := (1 , m1 , 2 , m2 ) thus denoted by R4 . The relation between both is given by


 1 = x 1 + y2 , m1 = y1 − x2 ,  2 = x 1 − y2 , m2 = y1 + x2 , (7.64)

or, equivalently, by
⎛ ⎞
1 0 0 1
⎜ 0 1 −1 0 ⎟
(1 , m1 , 2 , m2 )t := ⎜
⎝ 1 0
⎟ · (x1 , y1 , x2 , y2 )t . (7.65)
0 −1 ⎠
0 1 1 0

The 4 × 4 matrix from the right-hand side of (7.65) has determinant equal 4, so
it is invertible, hence
⎛ ⎞
1 0 1 0
1⎜ 0 1 0 1 ⎟
(x1 , y1 , x2 , y2 )t := ⎜ ⎟ · (1 , m1 , 2 , m2 )t , (7.66)
2 ⎝ 0 −1 0 1 ⎠
1 0 −1 0

that is,
1 + 2 m1 + m2 m2 − m1 1 − 2
x1 = , y1 = , x2 = , y2 = . (7.67)
2 2 2 2
Altogether we have an isomorphism of linear spaces φ : R4x → R4 defined by

φ
(x1 , y1 , x2 , y2 ) −→ (1 , m1 , 2 , m2 ) (7.68)
168 Chapter 7. Bicomplex Derivability and Differentiability

with the inverse linear map φ−1 : R4 → R4x given by (7.66). Because φ is a linear
isomorphism, the matrix is the Jacobi matrix of φ, so we denote it by Jx [φ]. Note
that considering the matrix transpose of (7.65) we can write:

 = x · Jx [φ]t .

Similarly, we denote the 4 × 4 matrix from the right-hand side of (7.66) by

J [φ−1 ], which is nothing but the inverse matrix of Jx [φ], having determinant equal
1
to . Hence, we can simply write (7.66) as
4

x =  · J [φ−1 ]t .

Remark 7.7.1. Because the determinant of Jx [φ] is 4, then φ does not preserve
Euclidean distances, i.e., it is not an isometry between R4x and R4 : if the Euclidean
norm of x is 1, then the Euclidean norm of  is
 √  √ √
|| = 21 + m21 + 22 + m22 = 2 x21 + y12 + x22 + y22 = 2|x| = 2 .

Let us denote by F(V ) any bicomplex module of BC-valued functions defined

on V ⊂ R4 , i.e., F(V ) is a bicomplex module under the usual operations of addi-
tion and multiplication by bicomplex constant. We define the change of variables
operator Wφ by

Wφ : F(U ) → F(V ), Wφ [g](x) := (g ◦ φ)(x) = g(φ(x)) , (7.69)

where U := φ−1 (V ) ⊂ R4x .

Since φ is a linear isomorphism, Wφ is a well defined linear operator, i.e., for
all λ, μ ∈ BC and for all g1 , g2 ∈ F(U ) we have:

Wφ [λg1 + μg2 ] = λWφ [g1 ] + μWφ [g2 ] .

The operator Wφ has an inverse defined by

Wφ−1 : F(U ) → F(V ), Wφ−1 [f ]() := (f ◦ φ−1 )() = f (φ−1 ()) , (7.70)

since (Wφ ◦ Wφ−1 )[f ] = f and (Wφ−1 ◦ Wφ )[g] = g. Therefore Wφ is a linear

isomorphism between F(V ) and F(U ).
Having established an isomorphism between the two BC-modules we can
establish now an isomorphism between the corresponding BC-algebras of linear
operators acting on these BC-modules. Indeed, given an operator B acting on
F(V ), we define the mapping

B −→ A := Wφ ◦ B ◦ Wφ−1 ,
7.7. Cartesian versus idempotent representations in BC-holomorphy 169

where A is an operator acting on F(U ). Then any linear combination or compo-

sition of operators is preserved under the mapping above: if B1 and B2 are two
operators acting on F(V ) and Λ1 and Λ2 are two bicomplex numbers, then

Wφ ◦ (Λ1 B1 + Λ2 B2 ) ◦ Wφ−1 = Λ1 (Wφ ◦ B1 ◦ Wφ−1 ) + Λ2 (Wφ ◦ B2 ◦ Wφ−1 )

=: Λ1 A1 + Λ2 A2

and

Wφ ◦ (B1 ◦ B2 ) ◦ Wφ−1 = (Wφ ◦ B1 ◦ Wφ−1 ) ◦ (Wφ ◦ B2 ◦ Wφ−1 )

=: A1 ◦ A2 .

The above general reasoning is specified now for the case of our interest,
namely, for F(V ) = C ∞ (V ) and F(U ) = C ∞ (U ) where the operators of partial
derivatives act with respect to the corresponding variables.
Let g be a function in C ∞ (V ), then g(φ(x)) is in C ∞ (U ) and the chain rule
gives:
∂(Wφ [g]) ∂
(x) = [g(φ(x))]
∂x1 ∂x1
∂g ∂1 ∂g ∂m1
= (φ(x)) · + (φ(x)) ·
∂1 ∂x1 ∂m1 ∂x1
∂g ∂2 ∂g ∂m2
+ (φ(x)) · + (φ(x)) ·
∂2 ∂x1 ∂m2 ∂x1
+ , + ,
∂g ∂g
= Wφ (x) + Wφ (x)
∂1 ∂2
+ ,
∂g ∂g
= Wφ + (x) .
∂1 ∂2
Since both the function g and the variable x are arbitrary, we obtain:

∂ ∂ ∂
◦ Wφ = Wφ ◦ + .
∂x1 ∂1 ∂2
Similar computations are made for the other variables. Summarizing, we get:

∂ ∂ ∂
= Wφ ◦ + ◦ Wφ−1 ,
∂x1 ∂1 ∂2

∂ ∂ ∂
= Wφ ◦ + ◦ Wφ−1 ,
∂y1 ∂m1 ∂m2

∂ ∂ ∂
= Wφ ◦ − ◦ Wφ−1 ,
∂x2 ∂m2 ∂m1

∂ ∂ ∂
= Wφ ◦ − ◦ Wφ−1 . (7.71)
∂y2 ∂1 ∂2
170 Chapter 7. Bicomplex Derivability and Differentiability

Briefly, we have established what the partial derivatives with respect to the canon-
ical cartesian coordinates turn out to be when one makes the change of variables
passing to the idempotent coordinates.
Considering the usual gradient operators in R4x and R4 ,
 
∂ ∂ ∂ ∂ ∂ ∂ ∂ ∂
∇x = , , , and ∇ = , , ,
∂x1 ∂y1 ∂x2 ∂y2 ∂1 ∂m1 ∂2 ∂m2

respectively, we rewrite the formulas above in the following compressed form:

 
∇x = Wφ ◦ ∇ · Jx [φ] ◦ Wφ−1 . (7.72)

If F is a bicomplex function, seen as a mapping F = (f11 , f12 , f21 , f22 ) from R4x to
R4x , then we get a relation between the Jacobi matrices of F in the two coordinates
systems considered: - .
Jx [F ] = J Wφ−1 [F ] · Jx [φ] . (7.73)

It is useful to have explicitly the reciprocal relations. Let us note that from (7.71)
we obtain:
∂ ∂ ∂
Wφ−1 ◦ ◦ Wφ = + ,
∂x1 ∂1 ∂2
∂ ∂ ∂
Wφ−1 ◦ ◦ Wφ = − ,
∂y2 ∂1 ∂2
which leads to

∂ 1 ∂ ∂
= Wφ−1 ◦ + ◦ Wφ , (7.74)
∂1 2 ∂x1 ∂y2

∂ 1 ∂ ∂
= Wφ−1 ◦ − ◦ Wφ . (7.75)
∂2 2 ∂x1 ∂y2

Similarly, we have:

∂ 1 ∂ ∂
= Wφ−1 ◦ − ◦ Wφ ,
∂m1 2 ∂y1 ∂x2

∂ 1 ∂ ∂
= Wφ−1 ◦ + ◦ Wφ . (7.76)
∂m2 2 ∂y1 ∂x2

We rewrite the formulas above at the level of gradients:

 
∇ = Wφ−1 ◦ ∇x · J [φ−1 ] ◦ Wφ . (7.77)

Now for a bicomplex function G, seen again as a mapping from R4 to R4 , we get:

J [G] = Jx [Wφ [G]] · J [φ−1 ] . (7.78)

7.7. Cartesian versus idempotent representations in BC-holomorphy 171

The reader should immediately notice that the formula above is consistent with (7.73)
for F = Wφ [G] and using the fact that Jx [φ] and J [φ−1 ] are matrices inverse to
each other.
In conclusion: if we want to translate a differential expression written in the
cartesian coordinates (x1 , y1 , x2 , y2 ) into an expression with the idempotent coor-
dinates (1 , m1 , 2 , m2 ), then we use formulas (7.71); if we start with an expression
given in the idempotent coordinates, then (7.74), (7.75) and (7.76) give its equiva-
lent in the cartesian coordinates. In other words, we have obtained direct relations
between the differential operators acting on two copies of R4 , one copy with the
cartesian coordinates and another with the idempotent coordinates.
The next step is to extend the ideas above onto the C(i)-complex differential
operators (7.40) in the complex variables z1 and z2 . Writing now the C(i)-complex
idempotent coordinates β1 := 1 + im1 and β2 := 2 + im2 , the usual associated
C(i)-complex differential operators are:
 
∂ 1 ∂ ∂ ∂ 1 ∂ ∂
:= −i , := −i ,
∂β1 2 ∂1 ∂m1 ∂β2 2 ∂2 ∂m2
 
∂ 1 ∂ ∂ ∂ 1 ∂ ∂
:= +i , := +i . (7.79)
∂β 1 2 ∂1 ∂m1 ∂β 2 2 ∂2 ∂m2

Introducing the C(i)-complex gradient operator


∂ ∂ ∂ ∂
∇z := , , , ,
∂z1 ∂z 1 ∂z2 ∂z 2

and similarly ∇β , we summarize these formulas as follows:

∇x = ∇z · M, ∇z = ∇x · M −1 ,

∇ = ∇β · M, ∇β = ∇ · M −1 , (7.80)

where
⎛ ⎞ ⎛ ⎞
1 i 0 0 1 1 0 0
⎜ 1 −i 0 0 ⎟ 1⎜ −i i 0 0 ⎟
M := ⎜
⎝ 0 0
⎟, M −1 := ⎜ ⎟.
1 i ⎠ 2⎝ 0 0 1 1 ⎠
0 0 1 −i 0 0 −i i

Now, if we want to derive the formulas relating, say, the differential operators
in z with the ones in , we just have to combine the relations (7.80) above with
the formulas (7.72) and (7.77):
   
∇z = ∇x · M −1 = Wφ ◦ ∇ · Jx [φ] ◦ Wφ−1 · M −1
 
= Wφ ◦ ∇ · Jx [φ] · M −1 ◦ Wφ−1 . (7.81)
172 Chapter 7. Bicomplex Derivability and Differentiability

A simple matrix multiplication yields the matrix Jx [φ]t · M −1 to be equal to

⎛ ⎞
1 1 −i i
1⎜⎜ −i i −1 −1 ⎟ .
⎟
2⎝ 1 1 i −i ⎠
−i i 1 1

From this we can deduce, for example, the formula:


∂ 1 ∂ ∂ ∂ ∂
= Wφ ◦ −i + −i ◦ Wφ−1 .
∂z1 2 ∂1 ∂m1 ∂2 ∂m2

 we incor-
If we want now to express the operators in z in terms of the ones in β,
porate in (7.81) the relation between the gradient in  and the one in β:

 
∇z = Wφ ◦ ∇ · Jx [φ] · M −1 ◦ Wφ−1
 
= Wφ ◦ ∇β · M · Jx [φ] · M −1 ◦ Wφ−1 .

To get a hands-on experience with the computations above, we could also

write directly from (7.71):

∂ 1 ∂ ∂
= −i
∂z1 2 ∂x1 ∂y1

1 ∂ ∂ ∂ ∂
= Wφ ◦ + −i −i ◦ Wφ−1
2 ∂1 ∂2 ∂m1 ∂m2

∂ ∂
= Wφ ◦ + ◦ Wφ−1 . (7.82)
∂β1 ∂β2

Similarly, we obtain the following formulas:


∂ ∂ ∂
= Wφ ◦ (−i) − ◦ Wφ−1 ,
∂z2 ∂β1 ∂β2

∂ ∂ ∂
= Wφ ◦ + ◦ Wφ−1 ,
∂z 1 ∂β 1 ∂β 2

∂ ∂ ∂
= Wφ ◦ i − ◦ Wφ−1 . (7.83)
∂z 2 ∂β 1 ∂β 2

A similar computation leads to the reciprocal transformations:


∂ 1 ∂ ∂
= Wφ−1 ◦ +i ◦ Wφ ,
∂β1 2 ∂z1 ∂z2

∂ 1 ∂ ∂
= Wφ−1 ◦ −i ◦ Wφ ,
∂β2 2 ∂z1 ∂z2
7.7. Cartesian versus idempotent representations in BC-holomorphy 173

∂ 1 ∂ ∂
= Wφ−1 ◦ −i ◦ Wφ ,
∂β 1 2 ∂z 1 ∂z 2

∂ 1 ∂ ∂
= Wφ−1 ◦ +i ◦ Wφ . (7.84)
∂β 2 2 ∂z 1 ∂z 2

We can write the formulas above in terms of matrix multiplication:

⎛ ⎞
 1 0 −i 0 
∂ ∂ ∂ ∂ ⎜ 0 1 0 i ⎟ ∂ ∂ ∂ ∂
, , , =⎜⎝ 1 0 i
⎟·
⎠ , , , ,
∂z1 ∂z 1 ∂z2 ∂z 2 0 ∂β1 ∂β 1 ∂β2 ∂β 2
0 1 0 −i

where the 4 × 4 matrix above is M · Jx [φ] · M −1 , as expected. The reciprocal

transformation is given by
⎛ ⎞
 1 0 1 0 
∂ ∂ ∂ ∂ 1⎜ 0 1 0 1 ⎟ ∂ ∂ ∂ ∂
, , , = ⎜ ⎟· , , , ,
∂β1 ∂β 1 ∂β2 ∂β 2 2⎝ i 0 −i 0 ⎠ ∂z1 ∂z 1 ∂z2 ∂z 2
0 −i 0 i

where the 4 × 4 matrix above is M · J [φ−1 ] · M −1 .

In conclusion: we have obtained direct relations between the complex differ-
ential operators acting on two copies of C2 (i), one is with cartesian coordinates
(z1 , z2 ) and another is with idempotent coordinates (β1 , β2 ).
Remark 7.1. Note that we could have started these calculations from the C(j)-
idempotent representation of bicomplex numbers, Z = γ1 e + γ2 e† , where γ1 , γ2 ∈
C(j) are given by

γ1 = 1 + j(−m1 ), γ2 = 2 + jm2 .

Notice that the computations will not yield the same formulas, since one has to
start with a different linear isomorphism φ, " for which the defining matrix is ob-
tained from (7.65) by multiplying the second column by (−1). For example, the
formulas for the bicomplex differential operators in terms of the C(j)-complex
idempotent operators are:

∂ ∂ ∂ †
= Wφ ◦ e+ e ◦ Wφ−1 ,
∂Z ∂γ1 ∂γ2

∂ ∂ ∂ †
= W φ ◦ e + e ◦ Wφ−1 ,
∂Z † ∂γ1∗ ∂γ2∗

∂ ∂ ∂ †
= Wφ ◦ e+ e ◦ Wφ−1 ,
∂Z ∂γ2 ∂γ1

∂ ∂ ∂ †
= W φ ◦ e + e ◦ Wφ−1 .
∂Z ∗ ∂γ2∗ ∂γ1∗
174 Chapter 7. Bicomplex Derivability and Differentiability

We leave it as an exercise to the reader to develop in full detail the above

computations in the language of C( j)-complex differential operators.
Let us turn now to the bicomplex differential operators (bicomplex ana-
logues of the complex conjugate Cauchy–Riemann operators). They were defined
by (7.47) and we have noted several times that they do not depend on the repre-
sentation of the bicomplex variable Z. Let us comment on this in more detail.
Introduce the bicomplex gradient operator

∂ ∂ ∂ ∂
∇Z := , , , .
∂Z ∂Z † ∂Z ∂Z ∗
In standard real coordinates x, we have: ∇Z = ∇x · T , where
⎛ ⎞
1 1 1 1
1⎜ −i −i i i ⎟
T := ⎜ ⎟. (7.85)
4 ⎝ −j j −j j ⎠
k −k −k k
If we want to write ∇Z in terms of z differential operators, we simply combine
the formulas for the gradients in question:
∇Z = ∇x · T = ∇z · M · T ,
where ⎛ ⎞
1 1 0 0
1⎜ 0 0 1 1 ⎟
M ·T = ⎜⎝
⎟, (7.86)
2 −j j 0 0 ⎠
0 0 −j j
i.e., we get exactly formulas (7.47) which express the bicomplex differential oper-
ators in terms of the complex differential operators in z1 and z2 .
 we compute:
If we want to obtain ∇Z in the idempotent coordinates β,
   
∇Z = ∇x · T = Wφ ◦ ∇ · Jx [φ] ◦ Wφ−1 · T
 
= Wφ ◦ ∇ · Jx [φ] · T ◦ Wφ−1 ,
where the matrix Jx [φ] · T is given by
⎛ ⎞
1+k 1−k 1−k 1+k
1⎜ −i +j −i − j i+j i−j ⎟
Jx [φ] · T = ⎜ ⎟. (7.87)
4⎝ 1−k 1+k 1+k 1−k ⎠
−i − j −i + j i−j i+j
Since
1+k 1−k
e= , e† = ,
2 2
i + j = 2ie† , i − j = 2ie ,
7.7. Cartesian versus idempotent representations in BC-holomorphy 175

we obtain: ⎛ ⎞
e e† e† e
1⎜ −ie −ie† ie† ie ⎟
Jx [φ] · T = ⎜ ⎟. (7.88)
2 ⎝ e† e e e† ⎠
−ie† −ie ie ie†
∂
For example, if we want to write in the real idempotent coordinates, we get
∂Z
from above:

∂ 1 ∂ ∂ ∂ † ∂ †
= Wφ ◦ e−i e+ e −i e ◦ Wφ−1 .
∂Z 2 ∂1 ∂m1 ∂2 ∂m2

Furthermore, writing ∇Z in β coordinates we obtain

 
∇Z = Wφ ◦ ∇β · M · Jx [φ] · T ◦ Wφ−1 ,

where the matrix M · Jx [φ] · T is given by

⎛ ⎞
e e† 0 0
⎜ 0 0 e† e ⎟
⎜ † ⎟. (7.89)
⎝ e e 0 0 ⎠
0 0 e e†

Consider a bicomplex holomorphic function F , then


∂F
∇Z [F ] = , 0, 0, 0 .
∂Z

Writing F in the idempotent form F = G1 e + G2 e† , we get:

∂F ∂F ∂F † ∂G1 ∂G2 †
= e+ e = e+ e ,
∂Z ∂β1 ∂β2 ∂β1 ∂β2
∂F † ∂F ∂G1 ∂G2 †
0= e + e= e+ e ,
∂β1 ∂β2 ∂β2 ∂β1
∂F † ∂F ∂G1 ∂G2 †
0= e + e= e+ e ,
∂β 1 ∂β 2 ∂β 2 ∂β 1
∂F ∂F † ∂G1 ∂G2 †
0= e+ e = e+ e ,
∂β 1 ∂β 2 ∂β 1 ∂β 2
which yields once more the conclusion of Theorem 7.6.3.
Remark 7.7.2. Under the same hypothesis that F = f11 + if12 + jf21 + kf22 is a bi-
complex holomorphic function written in the standard coordinates x, the derivative
of F is given by
∂F ∂F
= = a + bi + cj + dk,
∂Z ∂x1
176 Chapter 7. Bicomplex Derivability and Differentiability

where a, b, c, d are real partial derivatives of the functions fk related by (7.18).

Written in the idempotent coordinates l, the expression above becomes
+ ,  
∂F 1 ∂ F̂ ∂ F̂ ∂ F̂ ∂ F̂
Wφ−1 = e−i e+ e† − i e† ,
∂Z 2 ∂1 ∂m1 ∂2 ∂m2

where F̂ = Wφ−1 [F ].
Remark 7.7.3. Let us analyze in more detail the crucial difference between the
complex differential operators in the z and β variables. For example, if we look
side-by-side at the formulas:

∂ ∂ ∂
= + ,
∂z1 ∂Z ∂Z †

∂ −1 ∂ ∂ †
= Wφ ◦ ·e+ · e ◦ Wφ ,
∂β1 ∂Z ∂Z †

we notice immediately a huge difference between them: the first one mixes the one-
dimensional bicomplex operators in Z and Z † , while the second one keeps them
separate (!). For example, consider the bicomplex function F (Z) = (Z −Z † )2 ; then

∂F ∂F
(Z) = 0, but (Z) = 2(β1 − β2 ) ,
∂z1 ∂β1

for all Z ∈ BC, thus F is constant with respect to z1 , but not with respect to β1 .
This is tightly related, of course, with the fact that a BC-holomorphic func-
tion when seen as a mapping from C2 to C2 with the cartesian coordinates is a
pair of holomorphic functions which depend on two complex variables and which
have a Cauchy–Riemann type relation, meanwhile the same function written in the
idempotent coordinates becomes a pair of holomorphic functions in one variable
which moreover are independent of each other.
Remark 7.7.4. Let us see now the relation between bicomplex differential operators
and the idempotent representation of hyperbolic numbers. Recalling the formula
∂
for , we regroup the terms in a different way:
∂Z
 
∂ 1 ∂ ∂ ∂ ∂
= Wφ ◦ −i e+ −i e† ◦ Wφ−1
∂Z 2 ∂1 ∂m1 ∂2 ∂m2
 
1 ∂ ∂ † ∂ ∂ †
= Wφ ◦ e+ e −i e+ e ◦ Wφ−1 . (7.90)
2 ∂1 ∂2 ∂m1 ∂m2

But if we write Z = z1 + iz2 , where

z1 = x1 + ky2 , z2 = y1 − kx2 ,
7.7. Cartesian versus idempotent representations in BC-holomorphy 177

and if we use the intrinsic hyperbolic idempotent representations

z1 = (x1 + y2 )e + (x1 − y2 )e† = 1 e + 2 e† ,

z2 = (y1 − x2 )e + (y1 + x2 )e† = m1 e + m2 e† ,

then the hyperbolic differential operators have the expressions:

 
∂ 1 ∂ ∂ ∂ ∂ †
= +k = Wφ ◦ e+ e ◦ Wφ−1 ,
∂z1 2 ∂x1 ∂y2 ∂1 ∂2
 
∂ 1 ∂ ∂ ∂ ∂ †
= −k = Wφ ◦ e+ e ◦ Wφ−1 .
∂z2 2 ∂y1 ∂x2 ∂m1 ∂m2

Going back to formula (7.90), we obtain the following formulation of the bicomplex
differential operator with respect to Z, in terms of the hyperbolic derivatives:

∂ 1 ∂ ∂
= Wφ ◦ −i ◦ Wφ−1 ,
∂Z 2 ∂z1 ∂z2

and similarly for the other operators.

Remark 7.7.5. Analogous computations can be made for all bicomplex differential
operators, starting with either one of the other writings in terms of the C(j) or
hyperbolic operators. We leave the details for the reader, with the note that all for-
mulas are consistent from all points of view. The plethora of formulas in different
writings, coordinates or change of bases of BC, all of which agree perfectly with
each other, is again a phenomenon very specific to the bicomplex setup. Think of
having multiple pairs of glasses which allow you to see a picture (or a movie) from
different points of view, in different dimensions, distinguishing different colorings,
etc.
Chapter 8

Some Properties of Bicomplex

Holomorphic Functions

In what follows, we investigate several properties of bicomplex holomorphic func-

tions defined on a set Ω in BC. Also we assume that Ω can be written in the
form


Ω = Ω1 e + Ω2 e† := β1 e + β2 e†  β1 ∈ Ω1 , β2 ∈ Ω2
with Ω1 and Ω2 domains, that is, open and connected sets, in C(i). We will call such
Ω “a product-type domain” in BC. Note that such an Ω should
be seen
 as the carte-

sian product Ω1 ×Ω2 in the “idempotent C2 (i)” (which is (β1 , β2 )  β1 , β2 ∈ C(i) )

where the elements are written as bicomplex numbers in their idempotent form.
The topological boundary ∂Ω is the union of the three sets: ∂Ω1 × Ω2 , Ω1 × ∂Ω2
and ∂Ω1 × ∂Ω2 . The latter is called, sometimes, the distinguished boundary, or
Shilov boundary, of Ω.

8.1 Zeros of bicomplex holomorphic functions

We will study the zeros of bicomplex holomorphic functions. For this, let F be
such a function defined on a domain Ω = Ω1 e + Ω2 e† , so in its idempotent form

F (Z) = G1 (β1 )e + G2 (β2 )e† ,

for all Z = β1 e + β2 e† ∈ Ω. Assume first that neither G1 nor G2 are identically

zero. Because G1 and G2 are complex holomorphic functions in β1 and β2 , each
has isolated zeros. This means that for  = 1, 2, if G (β ,0 ) = 0, then there exists a
disk B of center β ,0 and radius r such that G (β ) = 0, for all β ∈ B \ {β ,0 }.
Then on the open set B := B1 e+B2 e† , the only zero of F is at Z0 := β1,0 e+β2,0 e† .
Therefore F has isolated zeros in the set Ω := Ω1 e + Ω2 e† .

© Springer International Publishing Switzerland 2015 179

M.E. Luna-Elizarrarás et al., Bicomplex Holomorphic Functions, F rontiers in Mathematics,
DOI 10.1007/978-3-319-24868-4_9
180 Chapter 8. Some Properties of Bicomplex Holomorphic Functions

Now if, say, G2 ≡ 0 but G1 is not identically zero, then F (Z) = G1 (β1 )e, so
F (Z) = 0 if and only if Z = β1,0 e + λe† , where β1,0 is a zero of G1 and λ is an
arbitrary C(i)-complex number. Therefore the set of zeros of F is a countable (or
finite) union of portions of complex lines parallel to the zero-divisor line Le† =
BCe† ; in this case the complex line that passes through β1,0 e and which is parallel
to BCe† is of the form β1,0 e + BCe† . A similar statement is true for G1 ≡ 0.
Summarizing, we have just proved the following
Theorem 8.1.1. Consider a bicomplex holomorphic function F : Ω → BC which is
not identically zero. Then
1. if F takes at least one invertible value, then its zero set is either the empty
set or a set of isolated points in Ω;
2. if F (Z) ∈ BCe for all Z ∈ Ω, then its zero set is a countable (or finite) union
of portions of complex lines parallel to the zero-divisor line Le† ;
3. if F (Z) ∈ BCe† for all Z ∈ Ω, then its zero set is a countable (or finite)
union of portions of complex lines parallel to the zero-divisor line Le .
Example 8.1.2. Consider the bicomplex holomorphic function F (Z) = Z 2 =
f1 (Z) + jf2 (Z), with f1 (z1 , z2 ) = 2z1 z2 and f2 (z1 , z2 ) = z22 − z12 , which are defined
and holomorphic on C2 (i). Their zero sets are:
 
V1 = {(z1 , z2 ) ∈ C2 (i)  z1 = 0} ∪ {(z1 , z2 ) ∈ C2 (i)  z2 = 0} ,
 
V2 = {(z1 , z2 ) ∈ C2 (i)  z1 = z2 } ∪ {(z1 , z2 ) ∈ C2 (i)  z1 = −z2 } ,
respectively. Each of them is a union of two complex lines in C2 (i), respectively.
Note that the intersection V1 ∩ V2 = {(0, 0)}, which is just a point and which is
the only zero of F .
A direct computation shows that using the idempotent variables β1 = z1 −iz2
and β2 = z1 + iz2 we get:
F (Z) = β12 e + β22 e† =: G1 (β1 )e + G2 (β2 )e† ,
thus the functions G1 and G2 , which are holomorphic on C(i) with respect to their
corresponding variable, have only one zero: β1 = 0 and β2 = 0, respectively.
Example 8.1.3. Consider now the bicomplex exponential function F (Z) = eZ ,
which is bicomplex holomorphic on BC and takes only invertible values for all
Z ∈ BC. The functions f1 (z1 , z2 ) = ez1 cos z2 and f2 (z1 , z2 ) = ez1 sin z2 have their
zero sets
 
 (2n + 1)π 
2 
V1 = {(z1 , z2 ) ∈ C (i) cos z2 = 0} = z1 , 2 
∈ C (i) n ∈ Z ,
2
 
V2 = {(z1 , z2 ) ∈ C2 (i)  sin z2 = 0} = {(z1 , nπ) ∈ C2 (i)  n ∈ Z} ,
which are obviously non-compact sets in C2 (i). But V1 ∩ V2 = ∅, so F has no zeros,
as we already know.
8.2. When bicomplex holomorphic functions reduce to constants 181

8.2 When bicomplex holomorphic functions reduce to

constants
In this section we investigate some conditions under which a bicomplex holomor-
phic function reduces to a constant function.
Theorem 8.2.1. Let F : Ω → BC be a bicomplex holomorphic function on a product-
type domain Ω. Then F  (Z) = 0 for all Z ∈ Ω if and only if F is a constant
function.
Proof. If F is a constant function, its derivative is zero on Ω. Conversely, using
the notation and result of Corollary 7.6.6, we are assured that

0 = F  (Z) = G1 (β1 )e + G2 (β2 )e† ,

for all Z = β1 e+β2 e† ∈ Ω. This is equivalent to G1 (β1 ) = 0 on Ω1 and G2 (β2 ) = 0
on Ω2 . Because Ω1 and Ω2 are domains in the complex plane, this implies that
G1 , G2 are constant functions with respect to the variables β1 , β2 , respectively,
i.e., G1 (β1 ) = a1 ∈ C(i) for all β1 ∈ Ω1 and G2 (β2 ) = a2 ∈ C(i) for all β2 ∈ Ω2 ,
therefore F (Z) = a1 e + a2 e† , a bicomplex constant for all Z ∈ Ω. 
Remark 8.2.2. The proof of the theorem above could have been obtained without
using the idempotent representation of F . For example, writing F = f1 + jf2 on
Ω, the existence of F  on Ω implies the existence of the complex partial derivatives
of f1 and f2 on Ω with respect to z1 and z2 , related by the Cauchy-Riemann type
∂F
conditions (7.55). Using the fact F  (Z) = = 0, we can derive easily that the
∂Z
partial derivatives of f1 and f2 with respect to both z1 and z2 are zero, which
implies that f1 and f2 are constant on the domain Ω, so F is constant on Ω.
Now we are interested in examining under what conditions a bicomplex holo-
morphic function F has zero derivative on a domain Ω.
Theorem 8.2.3. Let F be a bicomplex holomorphic function on a product-type sim-
ply connected domain Ω in BC. Then F is a constant function if and only if any
of the following equivalent conditions holds:
1. Writing F = f11 +if12 +jf21 +kf22 , one of the real functions fk is constant.
In particular, F can be “missing” one (or more) of the real components fk ,
i.e., fk = 0.
2. Writing F = f1 +jf2 , either f1 or f2 is a constant C(i)-function. Similarly for
F = g1 + kg2 . In particular, F can be either C(i)-valued (i.e., f2 = g2 = 0),
or j · C(i)-valued (i.e. f1 = 0), or k · C(i)-valued (i.e., g1 = 0).
3. Writing F = ρ1 + iρ2 , either ρ1 or ρ2 is constant. Similarly for F = γ1 + kγ2 .
In particular, F can be either C(j)-valued (i.e., ρ2 = γ2 = 0), or i·C(j)-valued
(i.e., ρ1 = 0), or k · C(j)-valued (i.e., γ1 = 0).
182 Chapter 8. Some Properties of Bicomplex Holomorphic Functions

4. Writing F = f1 + if2 , either f1 or f2 is constant. Similarly, for F = g1 + jg2 .

In particular, F can be either D-valued (i.e., f2 = g2 = 0), or i · D-valued
(i.e., f1 = 0), or j · D-valued (i.e., g1 = 0).
5. F (Z) = A · f (Z), where A ∈ BC is an invertible bicomplex number and f is
either a C(i), C(j), D, jC(i), kC(i), iC(j), kC(j), iD or jD-valued function.
Observe that if A ∈ S, then both F (Z) and F  (Z) are in S.
6. The function F is of the form F = f + jλf with a C(i)-valued f and λ ∈
C(i)\{±i}. Similarly F can be of the form F = ρ+i(μρ) with μ ∈ C( j)\{±j}
and a C( j)-valued ρ; or F = f + i(af) with a ∈ D. Observe that if λ = ±i,
then the values of F (Z) and of F  (Z) are in S0 .
7. Either one of the complex moduli |F (Z)|i or |F (Z)|j equals a non-zero con-
stant (in C(i) or C(j)) for all Z ∈ Ω. Observe that either one of the complex
moduli is zero for all Z if and only if F (Z) and so F  (Z) are in S0 .
8. The hyperbolic modulus |F (Z)|k equals an invertible hyperbolic constant for
all Z ∈ Ω. Observe that |F (Z)|k is a hyperbolic zero-divisor if and only if
F (Z) and so F  (Z) are in S0 .
9. F (Z) is invertible for all Z ∈ Ω and either one of the complex arguments
argi (F (Z)) or argj (F (Z)) are constant for all Z.
Proof. Assume the function f11 is constant on Ω. Using relations (7.18) it follows
that a = b = c = d = 0, i.e., all the other real functions fk have real partial
derivatives zero with respect to all four real variables x1 , y1 , x2 and y2 . Since Ω
is a domain, it follows that all fk are constant functions, thus F is a constant
bicomplex function which provides a proof of (1).
The proofs of the next items (2)–(4) are easily reduced to (1). For example
if f2 is constant, then f21 and f22 are constant, so F is constant, etc. But let us
further investigate the case f2 (z1 , z2 ) = λ ∈ C(i) in the idempotent representation
F = G1 e + G2 e† . We obtain for any Z ∈ Ω:
G1 (Z) = f1 (Z) − iλ, G2 (Z) = f1 (Z) + iλ .
Since F  (Z) exists on all Ω, then G1 depends only on β1 = Ze and G2 depends
only on β2 = Ze† , i.e.,
∂G1 ∂G2
0 = (Z), 0 = (Z).
∂β2 ∂β1
Recalling the relation between the partial derivatives with respect to z1 and z2
and those with respect to β1 and β2 we have:
+  ,
∂ 1 ∂ ∂
= Wφ ◦ −i ◦ Wφ−1 ,
∂β2 2 ∂z1 ∂z2
+  ,
∂ 1 ∂ ∂
= Wφ ◦ +i ◦ Wφ−1 ,
∂β1 2 ∂z1 ∂z2
8.2. When bicomplex holomorphic functions reduce to constants 183

which implies that


1 ∂f1 ∂f1
−i = 0,
2 ∂z1 ∂z2

1 ∂f1 ∂f1
+i = 0,
2 ∂z1 ∂z2
hence the partial derivatives of f1 with respect to z1 and z2 are zero, so f1 is a
constant. Similar arguments hold for the other cases (2)–(4).
We use the same type of calculations for the proof of the next assertion: let
F (Z) = A · f (Z), where f is either a C(i)-, C(j)-, D-, etc. valued function, and
A = a1 e + a2 e† , say, with a1 , a2 ∈ C(i), a1 a2 = 0. Then

F = (a1 f )e + (a2 f )e† =: G1 e + G2 e† .

∂G1 ∂G2
Again, because F  (Z) exists in Ω, then (Z) = (Z) = 0, which will force
∂β2 ∂β1
the partial derivatives of f with respect to the C(i)-, C(j)- or hyperbolic variables
to be zero, i.e., f is a constant. Therefore F is constant.
Note that (5) can also be proved directly, using the fact that if F is deriv-
able on Ω, then multiplying it by any bicomplex constant (in our case A−1 ) will
yield a bicomplex derivable function G(Z). Therefore, if F (Z) = A · f (Z), then
G(Z) = f (Z) is a complex, hyperbolic, etc. valued BC-holomorphic function, so
it is constant.
The proofs of the case when, for example, F = f + j(λf ) is easily obtained
from the previous one: we write F = (1 + jλ)f , and note that A := 1 + jλ is
invertible if and only if λ = ±i. We note also that a direct proof of this fact is
obtained using the Cauchy-Riemann conditions for f1 = f and f2 = λf : at any Z
we obtain
∂f ∂f1 ∂f2 ∂f
= = =λ ,
∂z1 ∂z1 ∂z2 ∂z2
∂f ∂f1 ∂f2 ∂f
= =− = −λ ,
∂z2 ∂z2 ∂z1 ∂z1
from which we get:
∂f ∂f
(1 + λ2 ) = (1 + λ2 ) = 0,
∂z1 ∂z2
so if λ = ±i, then f is constant, and so on.
Assume now that |F (Z)|i = a ∈ C(i) for any Z ∈ Ω. In the idempotent
representation, we have:

a2 = |F (Z)|2i = G1 (β1 ) · G2 (β2 ) ,

for all Z ∈ Ω. If a = 0, then we get that G1 (β1 ) = 0 or G2 (β2 ) = 0 for all

Z = β1 e + β2 e† . Since G1 and G2 are complex holomorphic functions on Ω1 ,
184 Chapter 8. Some Properties of Bicomplex Holomorphic Functions

respectively Ω2 , domains in C(i), their zeros are isolated, unless they are identically
zero. Thus, it follows that either G1 ≡ 0 on Ω1 or G2 ≡ 0 on Ω2 . In either case,
the result is that F (Z) ∈ S0 for all Z ∈ Ω.
If a = 0, then the following argument applies: the partial derivatives with
respect to both β1 and β2 of the product G1 · G2 are zero:
∂(a2 ) ∂(G1 (β1 ) · G2 (β2 ))
0= = = G1 (β1 ) · G2 (β2 ) ,
∂β1 ∂β1
∂(a2 ) ∂(G1 (β1 ) · G2 (β2 ))
0= = = G1 (β1 ) · G2 (β2 ) ,
∂β2 ∂β2
for all β1 ∈ Ω1 and β2 ∈ Ω2 , so in the case that neither G1 nor G2 are identically
zero, then both G1 and G2 are constant functions on their domains. Therefore, F
is constant. Similarly in the C(j) case.
A special computation is necessary when the hyperbolic modulus of F equals
a hyperbolic constant a = a1 + ka2 , where now a1 , a2 ∈ R. Let F = f1 + if2 , and
write each hyperbolic function f1 = s1 e + t1 e† and f2 = s2 e + t2 e† in idempotent
components (in the intrinsic hyperbolic writing), where s1 , s2 , t1 , t2 are real-valued
functions of the bicomplex variable Z. Then

F = (s1 e + t1 e† ) + i(s2 e + t2 e† ) = (s1 + is2 )e + (t1 + it2 )e† =: G1 e + G2 e† .

Then the square of the hyperbolic modulus of F is given by

a21 e + a22 e† = a2 = |F (Z)|2k = (s21 (Z) + s22 (Z))e + (t21 (Z) + t22 (Z))e† ,

therefore
a21 = s21 (Z) + s22 (Z) = G1 (Z) · G1 (Z)
and
a22 = t21 (Z) + t22 (Z) = G2 (Z) · G2 (Z)
for all Z. Because s1 , s2 , t1 and t2 are real-valued functions, then a1 = 0 or a2 = 0
is equivalent to s1 = s2 ≡ 0 or t1 = t2 ≡ 0, respectively, which is equivalent to
F (Z) being a zero-divisor for all Z.
The function G1 (Z) is a complex holomorphic function which depends on β1
only: G1 = G1 (β1 ), thus the function G1 is anti-holomorphic in β1 , so that in the
identity a21 = G1 (β1 ) · G1 (β1 ) we cannot take the derivatives of both sides but we
∂
can apply the Cauchy–Riemann operator which gives:
∂β1

∂(G1 · G1 ) ∂G1
0= (β1 ) = (β1 ) · G1 (β1 ) = G1 (β1 ) · G1 (β1 ) ,
∂β1 ∂β1

∂G1 ∂G1
where we used the fact that = = 0. Since this happens for all Z, so for
∂β1 ∂β 1
all β1 ∈ Ω1 , it follows that either G1 ≡ 0 so G1 ≡ 0, or that G1 is constant. An
8.3. Relations among bicomplex, complex and hyperbolic holomorphies 185

analogous argument holds for G2 . Note that G1 ≡ 0 if and only if s1 = s2 = 0, so

F (Z) ∈ S0 for all Z, and similarly for G2 . Otherwise, it follows that F is constant.
Consider now F (Z) to be an invertible bicomplex number for all Z ∈ Ω.
Assume that, in its C(i)-complex trigonometric form

F (Z) = |F (Z)|i ejΘ = |F (Z)|i (cos Θ + j sin Θ),

the principal argument Θ = argi (F (Z)) is constant for all Z. Then the expression
ejΘ is an invertible bicomplex number, say, A. Therefore F (Z) = A · f (Z), where
f (Z) = |F (Z)|i is a C(i)-valued function. Because F is bicomplex holomorphic on
Ω, it follows that F is constant. Similarly for the C(j) case. 
Corollary 8.2.4. Let F be a bicomplex holomorphic function on a product-type
domain Ω. Each of the following conditions implies that F  (Z) ∈ S for all Z ∈ Ω:
1. F (Z) ∈ S for all Z ∈ Ω.
2. If we write F = G1 e + G2 e† , then either G1 or G2 is a constant function.
This happens if and only if F  (Z) ∈ S for all Z.
3. If F (Z) = A · f (Z), where A ∈ S and f is either a C(i)-, C(j)-, D-, jC(i)-,
kC(i)-, iC(j)-, kC(j)-, iD- or jD-valued function.
4. |F (Z)|i = 0 for all Z ∈ Ω.
5. |F (Z)|j = 0 for all Z ∈ Ω.
6. |F (Z)|k ∈ S for all Z ∈ Ω.

8.3 Relations among bicomplex, complex and

hyperbolic holomorphies
We have proved that if F is a bicomplex holomorphic function on a bicomplex
domain Ω, then its components in all bicomplex writings have very specific and
rigid properties.
For example, if F = f1 + jf2 , then f1 and f2 are C(i)-complex holomorphic
functions in the two complex variables z1 and z2 and they are related by the
Cauchy-Riemann type equation (7.21). Similarly, if F = ρ1 + iρ2 , then ρ1 and ρ2
are C(j)-complex holomorphic functions in the two complex variables ζ1 and ζ2 and
they are related by the Cauchy-Riemann type equation (7.29). Next, if F = f1 +if2 ,
then f1 and f2 are hyperbolic holomorphic functions in the two hyperbolic variables
z1 and z2 and they are related by the Cauchy-Riemann type equation (7.34).
Finally, in both idempotent representations the idempotent components of F are
complex (in C(i) or C(j)) holomorphic functions each of them of its corresponding
complex variable.
Also we have seen that if we start with a C 1 -bicomplex function F , then all
the holomorphy properties above are neatly obtained from the point of view of
186 Chapter 8. Some Properties of Bicomplex Holomorphic Functions

bicomplex, complex, and hyperbolic differential operators. In summary, if we use

the different linear isomorphisms between R4 , C2 (in i or j), D2 and BC, then all
the differential operators whose annihilations lead to the notions of holomorphy
in each case appear in the formulation of the bicomplex differential operators. So
all four operators
∂ ∂ ∂ ∂
, †
, and
∂Z ∂Z ∂Z ∂Z ∗
encode multiple notions of complex and hyperbolic holomorphy when applied to
the corresponding components of the function F .
In this respect, the differences between the theory of several complex variables
and bicomplex analysis lie in the fact that in the latter case we work with pairs of
complex holomorphic functions which are not independent: they are related by a
Cauchy-Riemann type system. The same fact happens in the usual one complex
variable theory, when the two real functions of two real variables are connected by
the usual Cauchy-Riemann equations. By this reason and the numerous arguments
we have already discovered so far, bicomplex analysis is in fact a one-variable
theory inside the theory of mappings in two complex variables.
Therefore, having an example of a bicomplex holomorphic function F pro-
vides examples of two complex holomorphic functions and two hyperbolic holo-
morphic functions. If F is such a function, then writing F = f1 + jf2 , we have
that f1 and f2 are complex holomorphic functions in two variables, which are re-
lated though, so f2 is somehow almost determined by f1 . We will elaborate on
this fact later on. Moreover, if we write F = f1 + if2 , then f1 and f2 are hyperbolic
holomorphic functions of two hyperbolic variables.
Example 8.3.1. Consider the bicomplex exponential function F (Z) = eZ . Then
f1 (z1 , z2 ) = ez1 cos z2 and f2 (z1 , z2 ) = ez1 sin z2 are holomorphic in z1 and z2 .
Also f1 (z1 , z2 ) = ez1 cos z2 and f2 (z1 , z2 ) = ez1 sin z2 are hyperbolic holomorphic
functions in two hyperbolic variables. Moreover, writing Z = w1 + kw2 , then also
the functions f1 (w1 , w2 ) = ew1 cosh w2 and f2 (w1 , w2 ) = ew1 sinh w2 are holomor-
phic in w1 and w2 .
Of course, this example is simple but there is a fine point here: the pairs
of functions that arise possess many additional properties which in no way are
immediately seen from inside the theory of complex-valued holomorphic functions
in C2 . An illustration of this is the fact that the arising functions are null solutions
to the complex Laplacian; we will comment on this later.

8.4 Bicomplex anti-holomorphies

We can define three notions of anti-holomorphy in the bicomplex context. Let us
start with a bicomplex function F of class C 1 on a domain Ω. We have proved that
∂ ∂
F is bicomplex holomorphic if and only if the three differential operators ,
∂Z † ∂Z
8.4. Bicomplex anti-holomorphies 187

∂
and anihilate the function F for all Z ∈ Ω, i.e.,
∂Z ∗
∂F ∂F ∂F
(Z) = (Z) = (Z) = 0 . (8.1)
∂Z † ∂Z ∂Z ∗
In this case the bicomplex derivative of F exists for all Z, and it is related to the
fourth differential operator:

∂F
F  (Z) = (Z) .
∂Z
∂ ∂
We have seen that the operators and are intimately related, and the
∂Z ∂Z †
∂ ∂
same is true for and .
∂Z ∂Z ∗
One may wonder what happens if instead of having relation (8.1) we impose
one of the following conditions for F for all Z in its domain:

∂F ∂F ∂F
(Z) = (Z) = (Z) = 0 , (8.2)
∂Z ∂Z ∂Z ∗
∂F ∂F ∂F
(Z) = †
(Z) = (Z) = 0 , (8.3)
∂Z ∂Z ∂Z ∗
∂F ∂F ∂F
(Z) = †
(Z) = (Z) = 0 . (8.4)
∂Z ∂Z ∂Z
Equalities (8.2) define what can be called †-anti-holomorphy while equali-
ties (8.3) are about bar-anti-holomorphy and equalities (8.4) are about ∗-anti-
holomorphy. It is clear how to introduce the corresponding derivatives.
We can interpret each of these conditions using all the formulas we developed
so far, in all the possible bicomplex writings. Note that if we compose F with a
change of bicomplex variable W → Z † , then (8.2) is equivalent to

∂F † ∂F † ∂F †
(W ) = (W ) = (W ) = 0 ,
∂W † ∂W ∂W ∗
†
where F † (W ) = (F (W )) . This is equivalent to the fact that the function F † is a
bicomplex holomorphic function of a bicomplex variable W = Z † ; more exactly,
the function F † (Z † ) is simply BC-holomorphic. In the same fashion, F is bicomplex
bar-anti-holomorphic if and only if the function F is bicomplex holomorphic with
respect to Z; F is bicomplex ∗-anti-holomorphic if and only if the function F ∗ is
bicomplex holomorphic with respect to Z ∗ . We leave the details to the reader.
Let us study in more detail the notion of bicomplex †-anti-holomorphy. If we
choose to write F = ρ1 + iρ2 , then (8.2) leads to ρ1 and ρ2 being anti-holomorphic
functions of the two C(j)-complex variables ζ1 and ζ2 , i.e., they are holomorphic
188 Chapter 8. Some Properties of Bicomplex Holomorphic Functions

with respect to the j-conjugate variables ζ1∗ and ζ2∗ , related by the Cauchy-Riemann
type system:
∂ρ1 ∂ρ2 ∂ρ1 ∂ρ2
∗ = ∗, ∗ =− ∗.
∂ζ1 ∂ζ2 ∂ζ2 ∂ζ1
For the same condition (8.2), using the idempotent representation F = Γ1 e+Γ2 e† ,
where Z = γ1 e + γ2 e† with γ1 , γ2 ∈ C(j), we obtain that Γ1 and Γ2 are complex
anti-holomorphic functions depending on only one variable: γ1∗ for Γ1 and γ2∗ for
Γ2 , respectively. Similar statements are true for the other two types of bicomplex
anti-holomorphy.
In the C(i)-idempotent representation F = G1 e+G2 e† , where Z = β1 e+β2 e†
with β1 , β2 ∈ C(i), equation (8.2) leads to the conclusion that G1 and G2 are
complex holomorphic in variables β2 and β1 , respectively: note here the reversed
role of the complex variables β1 and β2 .

8.5 Geometric interpretation of the derivative

We have described previously some properties of smooth hyperbolic curves which
make them similar to the usual, real, curves in the complex plane. It turns out
that this similarity extends quite far and it allows us to obtain a geometric inter-
pretation of the derivative at a point.
Definition 8.5.1. Let Ω be a domain in BC, let Z0 be a point in Ω and let f : Ω ⊂
BC −→ BC be a function. The function f is called a hyperbolic angle preserving
mapping at Z0 , or a D-conformal mapping at Z0 , whenever the following holds: if
Γ1 and Γ2 are two hyperbolic smooth curves passing through Z0 and if their images
f (Γ1 ) and f (Γ2 ) are also hyperbolic smooth curves passing through W0 = f (Z0 ),
then the hyperbolic angle between Γ1 and Γ2 at Z0 is the same as the hyperbolic
angle between f (Γ1 ) and f (Γ2 ) at W0 .
We begin with an auxiliary lemma which is of interest by itself.
Lemma 8.5.2. If f is bicomplex holomorphic in Ω and Γ is a smooth hyperbolic
curve in Ω, then f (Γ) is also a smooth hyperbolic curve.
Proof. Since f is bicomplex holomorphic, then it is of the form f (Z) = F1 (β1 )e +
F2 (β2 )e† for Z = β1 e + β2 e† ∈ Ω. Since Γ is a smooth hyperbolic curve, then
it has a parametrization of the form ϕ(u, v) = ϕ1 (u)e + ϕ2 (v)e† where ϕ1 and ϕ2
are parametrizations of smooth curves in C(i); thus, ϕ1 (u) = 0 and ϕ2 (v) = 0 for
any u and v.
It is clear that f ◦ ϕ is a parametrization of f (Γ):
 
f ◦ ϕ(u, v) = f ϕ1 (u)e + ϕ2 (v)e†
= F1 (ϕ1 (u)) e + F2 (ϕ2 (v)) e† .
Since both F1 ◦ ϕ1 and F2 ◦ ϕ2 are parametrizations of smooth curves in C(i), it
follows that f ◦ ϕ is a parametrization of a smooth hyperbolic curve. 
8.5. Geometric interpretation of the derivative 189

Theorem 8.5.3. If f is bicomplex holomorphic at Z0 ∈ Ω and f  (Z0 ) is neither

zero nor a zero-divisor, then f is a D-conformal mapping at Z0 .
Proof. Since f is bicomplex holomorphic we can write f (Z) = f (β1 e + β2 e† ) =
F1 (β1 )e + F2 (β2 )e† ; moreover

f  (Z0 ) = F1 (β10 )e + F2 (β20 )e†

and
argD f  (Z0 ) = θ1 e + θ2 e† ∈ D.
Next, take a smooth hyperbolic curve Γ passing through Z0 and let T be the
hyperbolic tangent line to Γ at Z0 . By the previous Lemma, f (Γ) = Λ = λ1 e+λ2 e†
is also a smooth hyperbolic curve with hyperbolic tangent line P at W0 = f (Z0 ).
Since Γ = γ1 e + γ2 e† has a parametrization

ϕ(u, v) = ϕ1 (u)e + ϕ2 (v)e† ,

where ϕ1 and ϕ2 are, respectively, the parametrizations of the smooth, real curves
γ1 and γ2 in C(i), then one can take the projections of T and P on BCe and BCe†
and write:
T = t 1 e + t 2 e† , P = p 1 e + p 2 e† .
Writing also Z0 = β10 e + β2 e† = ϕ1 (u0 )e + ϕ2 (v0 )e† , we note that t1 is the (real)
tangent line to γ1 at ϕ1 (u0 ) as well as t2 , p1 , p2 are the (real) tangent lines to γ2 ,
λ1 λ2 at ϕ2 (v0 ), F1 (ϕ1 (u0 )), F2 (ϕ2 (v0 )) respectively.
Denote now by μ1 and μ2 the real numbers such that tan μ1 and tan μ2 are
the (real) slopes of t1 and t2 respectively. Similarly, the real numbers κ1 and κ2
are such that tan κ1 and tan κ2 are the (real) slopes of p1 and p2 . Hence, the
hyperbolic slopes of the hyperbolic lines T and P are, respectively,

tan μ = tan μ1 e + tan μ2 e† ,

tan κ = tan κ1 e + tan κ2 e† .

But θ1 = arg F1 (β10 ) and θ2 = arg F2 (β2 (β20 ) hence

κ1 = μ 1 + θ 1 and κ2 = μ 2 + θ 2

which means that

κ = μ + argD f  (Z0 ). (8.5)
In other words, the hyperbolic argument argD f  (Z0 ) can be seen as an “increment”
which the hyperbolic angle μ, characterizing the tangent line T , gets when we pass
to the tangent line P . As in the complex analysis case this increment depends on
Z0 only, and not on the curve Γ.
190 Chapter 8. Some Properties of Bicomplex Holomorphic Functions

" passing
We are ready to complete the proof. Take now two curves Γ and Γ
through Z0 , then for each of the (8.5) it holds that:
κ = μ + argD f  (Z0 ),
" + argD f  (Z0 ).
"=μ
κ
Substracting them gives:
κ−κ
"=μ−μ
"
which means that f is a D-conformal mapping ar Z0 . 

Remark 8.5.4. Consider also an interpretation of the hyperbolic modulus of f (Z0 ).
By the properties of such moduli, one has:
  
f (Z0 ) = lim |f (Z) − f (Z0 )|D
D Z→Z0 |Z − Z0 |D
for Z − Z0 ∈ S0 . Hence, for any Z close to Z0 , Z − Z0 ∈ S0 , with respect to the
hyperbolic modulus there holds:
     
f (Z) − f (Z0 ) ≈ f  (Z0 ) · Z − Z0  . (8.6)
D D D

In analogy with the case of complex holomorphic functions, we say that a bicom-
plex holomorphic function with the derivative at Z0 different from zero and zero-
divisors,
 realizes locally, at Z0 , a hyperbolic homothety, with the hyperbolic coeffi-
cient f  (Z0 )D , in all directions except those of the cone of zero-divisors generated
by Z0 .

8.6 Bicomplex Riemann Mapping Theorem

In one complex variable the Riemann mapping theorem has different formulations.
We will use the one from [47].
Theorem 8.6.1. Let A be a simply connected domain such that A = C. Then there
exists a bijective conformal map f : A −→ B(0, 1) where B(0, 1) is the unit disk
in the complex plane. Furthermore, for any fixed z0 ∈ A, we can find a function f
such that f (z0 ) = 0 and f  (z0 ) > 0. With such a specification f is unique.
In the classical setting of holomorphic functions of two complex variables,
there are no analogues of this theorem for domains in C2 . It turns out that there
exists an analogue of it in the context of bicomplex holomorphic functions.
Theorem 8.6.2 (Bicomplex Riemann mapping theorem). Let Ω ⊂ BC be a product-
type domain Ω = Ω1 e+Ω2 e† such that Ω1 and Ω2 are simply connected domains in
C(i) and Ω1 = C(i), Ω2 = C(i). Then
 there
exists a bijective D-conformal mapping
f : Ω → B1 , with B1 := Z ∈ BC  |Z|k ≺ 1 the bicomplex ball of hyperbolic radius
1 centered at the origin. Furthermore for any fixed Z0 ∈ Ω, we can find an f such
that f (Z0 ) = 0 and f  (Z0 ) is a strictly positive hyperbolic number. With such a
specification f is unique.
8.6. Bicomplex Riemann Mapping Theorem 191

Proof. Since Ω = Ω1 e+Ω2 e† we can apply the complex Riemann mapping theorem
to Ω1 and Ω2 obtaining the bijective complex conformal mappings F1 and F2
onto the unit disk B(0, 1) ⊂ C(i), which are complex holomorphic functions with
derivatives different from zero. Set

F := F1 e + F2 e† .

Clearly F is a bicomplex holomorphic function on Ω such that the derivative

F  (Z) = F1 (β1 )e + F2 (β2 )e†

is not in S0 for any Z ∈ Ω. Moreover, F realizes a bijective correspondence

between Ω and B1 . Thus, the first part of the theorem is proved.
Take now Z0 ∈ Ω hence Z0 = β10 e + β20 e† , and let F1 and F2 be such that

F1 (β10 ) = 0, F2 (β20 ) = 0 and F1 (β10 ) > 0, F2 (β20 ) > 0.

Hence F (Z0 ) = 0 and F  (Z0 ) is a strictly positive hyperbolic number. 

Definition 8.6.3. Given two domains Ω and Ξ in BC, they are called D-conformally
equivalent if there exists a bijective D-conformal mapping from Ω onto Ξ.
Corollary 8.6.4. Any simply-connected product-type domain in C2 (i) such that
none of its idempotent components is the whole C(i) is D-conformally equivalent
to the unitary bidisk.
The analytic theory of bicomplex functions has also been carried out in Price
[56]. Using computational algebra techniques, properties of bicomplex holomorphy
such as removability of singularities, Cauchy-Kowalevsky extensions, and bicom-
plex hyperfunctions have been developed in [19, 20, 96].
There are numerous applications of bicomplex holomorphic functions in both
mathematics and physics, from the study of bicomplex manifolds [7, 21, 22, 49],
bicomplex pseudoanalytic functions [8], bicomplex fractal theory [12, 13, 29, 48,
59, 60, 61, 62, 102], bicomplex zeta functions [43, 42, 63], functional analysis with
bicomplex scalars [2], all the way to applications in electromagnetism [3], homo-
thetic motions [4, 5, 6, 38], and quantum mechanics [32, 33, 34, 64, 66, 67]. For
example, the algebraic structure and the geometry of bicomplex numbers are a
natural setup for the space of quantum mechanics projectors, as the bicomplex
holomorphic functions are for the complex fields in relativistic quantum field the-
ory.
Chapter 9

Second Order Complex and

Hyperbolic Differential
Operators

9.1 Holomorphic functions in C and harmonic functions

in R2
It is well known that complex holomorphic functions are tightly related with har-
monic functions of two real variables, a fact that proved to be of crucial importance
for the theories of both classes of functions. On the general level, the same occurs
with hyperholomorphic (synonymously - monogenic, regular) functions of (real)
Clifford analysis and the harmonic functions of the respective number of (real)
variables. By this reason, both one complex variable theory and Clifford analysis
are considered as refinements of the corresponding harmonic function theories.
This relation is due to the following factorizations of the respective Laplace oper-
ators. If
 
∂2 ∂2 ∂ 1 ∂ ∂ ∂ 1 ∂ ∂
ΔR2 := + 2; := −i , := +i ,
∂x2 ∂y ∂z 2 ∂x ∂y ∂z 2 ∂x ∂y

then
∂ ∂ ∂ ∂ 1
◦ = ◦ = ΔR2 . (9.1)
∂z ∂z ∂z ∂z 4

Similarly, if ΔRk denotes the Laplace operator in Rk , DCR denotes the Cauchy-
Riemann operator of Clifford analysis in Rn+1 , and DDir denotes the Dirac oper-
ator in Rn (see [23] and [9]), then

© Springer International Publishing Switzerland 2015 193

M.E. Luna-Elizarrarás et al., Bicomplex Holomorphic Functions, F rontiers in Mathematics,
DOI 10.1007/978-3-319-24868-4_10
194 Chapter 9. Second Order Complex and Hyperbolic Differential Operators

DCR ◦ DCR = DCR ◦ DCR = ΔRn+1 ,

2
DDir = −ΔRn .

Factorization (9.1) manifests the essence of the relation between complex holomor-
phic functions and harmonic functions. First of all, note that the Laplace operator
ΔR2 acts initially on real-valued functions but its action extends onto complex
valued functions component-wise: if f = u + iv, then

ΔR2 [f ] := ΔR2 [u] + iΔR2 [v] .

Thus equality (9.1) holds on complex-valued functions (of class C 2 , not just C 1 ).
∂f
Let f = u + iv be a holomorphic function, that is, = 0, hence by (9.1)
 ∂z
∂ ∂f
ΔR2 [f ] = 4 = 0 and thus f is a harmonic function. Reciprocally, taking
∂z ∂z
∂u
a real-valued harmonic function u, consider (which is a formal derivative,
∂z
that is, the result of the action of an operator, not the “honest” derivative of a
holomorphic function), one has:
+ ,
∂ ∂u 1
= ΔR2 [u] = 0 ,
∂z ∂z 4

∂u
which means that the complex-valued function generated by the harmonic
∂z
function u is holomorphic.

9.2 Complex and hyperbolic Laplacians

Although we have noted already certain asymmetries in the properties of complex
and hyperbolic numbers inside BC, for the purposes of this chapter each of the
sets C(i), C(j) and D plays for bicomplex numbers a role similar to that played
by R for complex numbers. Thus, one can consider the three candidates for the
corresponding Laplacians:

∂2 ∂2
ΔC2 (i) := + ; (9.2)
∂z12 ∂z22
∂2 ∂2
ΔC2 (j) := 2 + 2 ; (9.3)
∂ζ1 ∂ζ2
∂2 ∂2
ΔD := 2 + 2 . (9.4)
∂z1 ∂z2

Operators (9.2) and (9.3) are called complex (C(i) and C(j) respectively) Lapla-
cians and (9.4) is the hyperbolic Laplacian. The first of them acts on C(i)-valued
9.2. Complex and hyperbolic Laplacians 195

holomorphic functions of two complex variables z1 and z2 ; the second acts on C(j)-
valued holomorphic functions of the complex variables ζ1 and ζ2 ; and the third
acts on D-valued holomorphic functions of the hyperbolic variables z1 and z2 .
∂
Although we wrote in (9.2)–(9.4) the formal partial derivatives , etc.,
∂z1
in fact we mean authentic partial derivatives with respect to the corresponding
variables z1 , z2 , ζ1 , ζ2 , z1 , z2 . Of course in this situation the formal and the authen-
tic derivatives coincide but at the same time a confusion may arise, so that we
rewrite (9.2)–(9.4) as
ΔC2 (i) = ∂z2 + ∂z2 ; (9.5)
1 2

ΔC2 (j) = ∂ζ2 + ∂ζ2 ; (9.6)

1 2

Δ D2 = ∂z2 + ∂z2 . (9.7)

1 2

This is in keeping with the notation we adopted where we wrote complex partial
derivatives as fz 1 , etc. Hence ∂z2 f = fz1 , etc.
1
The next step consists in extending the operators (9.5)–(9.7) onto bicom-
plex-valued functions. It turns out that how we write bicomplex numbers becomes
important and it depends on the operators in (9.5)–(9.7). Indeed, for the operator
ΔC2 (i) the bicomplex function F should be holomorphic in the sense of two complex
variables and if we want the same for its components, then we are forced to consider
F as F = f1 + jf2 . Now we set:
ΔC2 (i) [F ] := ΔC2 (i) [f1 ] + jΔC2 (i) [f2 ] .
A similar reasoning holds for the other two operators: for ΔC2 (j) we take F =
ρ1 + iρ2 and we set:
ΔC2 (j) [F ] := ΔC2 (j) [ρ1 ] + iΔC2 (j) [ρ2 ] ;
for ΔD2 we take F = f1 + if2 and we set:
ΔD2 [F ] := ΔD2 [f1 ] + iΔD2 [f2 ] .
The analogues of the formula (9.1) arise if one uses the corresponding opera-
tors in formulas (7.49), (7.50), (7.51) and (7.52); more exactly the operators should
∂
be taken in an appropriate form. For instance, if we want to use the operators
∂Z †
∂
and in the form
∂Z
 
∂ 1 ∂ ∂ ∂ 1 ∂ ∂
= + j , = − j ,
∂Z † 2 ∂z1 ∂z2 ∂Z 2 ∂z1 ∂z2
then we are forced to write the variable as Z = z1 + jz2 and the function F as
F = f1 + jf2 . With this understanding we have that
∂2 ∂2 ∂ ∂
ΔC2 (i) = 2 + 2 =4 , (9.8)
∂z1 ∂z2 ∂Z ∂Z †
196 Chapter 9. Second Order Complex and Hyperbolic Differential Operators

where the operators act on BC-valued functions holomorphic in the sense of

the complex variables z1 , z2 . Hence, the theory of BC-holomorphic functions can
(should?) be seen as a function theory for the C(i)-complex Laplacian.
Analogously to what was done above, the factorization allows us to establish
direct relations between BC-holomorphic functions and complex harmonic func-
tions, that is, null solutions to the operator ΔC2 (i) . Indeed, let F = f1 + jf2 be a
BC-holomorphic function, that is, it is holomorphic in the sense of two complex
∂F
variables, but also = 0. Then
∂Z †

∂ ∂F
ΔC2 (i) [F ] = 4 = 0,
∂Z ∂Z †

and thus F is a BC-valued complex harmonic function. Reciprocally, taking a C(i)-

∂
valued complex harmonic function f1 , consider the BC-valued function [f1 ] =
∂Z
∂f1
(which is a formal operation on a holomorphic function of two complex vari-
∂Z
ables, not the bicomplex derivative of a BC-holomorphic function). One has:
+ ,
∂ ∂f1 1
†
= ΔC2 (i) [f1 ] = 0 ,
∂Z ∂Z 4

∂f1
which means that the BC-valued function generated by the complex harmonic
∂Z
function f1 is BC-holomorphic.
The operators in (9.2) and (9.3) are dealt with in exactly the same way,
although now other first-order operators enter into the game. In the case of Z =
ζ1 + jζ2 we take
 
∂ 1 ∂ ∂ ∂ 1 ∂ ∂
= −i , = +i ,
∂Z 2 ∂ζ1 ∂ζ2 ∂Z 2 ∂ζ1 ∂ζ2

arriving at

∂2 ∂2 ∂ ∂
ΔC2 (j) = 2 + 2 =4 ; (9.9)
∂ζ1 ∂ζ2 ∂Z ∂Z

in the case of Z = z1 + iz2 we take

 
∂ 1 ∂ ∂ ∂ 1 ∂ ∂
= −i , = +i ,
∂Z 2 ∂z1 ∂z2 ∂Z ∗ 2 ∂z1 ∂z2

arriving at

∂2 ∂2 ∂ ∂
ΔD = 2 + 2 =4 . (9.10)
∂z1 ∂z2 ∂Z ∂Z ∗
9.3. Complex and hyperbolic wave operators 197

So, we have provided each of the operators (9.2)-(9.4) with an adequate func-
tion theory. There is a fine point here: it is, in a sense, one and the same theory of
BC-holomorphic functions, but where different aspects of the latter are taken into
account. This “common” function theory allows us to realize the similarities and
the differences between the three operators (9.2)-(9.4). Indeed, (9.2) and (9.3) look
identical from the viewpoint of classical complex analysis, but viewing them from
the bicomplex perspective reveals subtle differences between them. At the same
time, (9.4) seems to be quite different with any of (9.2) and (9.3) but bicomplex
functions, again, show that there exists a deep underlying unity between all three
of them.
The direct relations between the null solutions of any of the operators ΔC2 (j)
and ΔD2 and the BC-holomorphic function theory is established in the same way
as we did for the operator ΔC2 (i) .

9.3 Complex and hyperbolic wave operators

Another well-known second-order operator, in the real case, is the wave operator,
and one can be tempted to look at some of its complex analogues:
∂2 ∂2
C2 (i) := 2 − , (9.11)
∂w1 ∂w22
∂2 ∂2
C2 (j) := − . (9.12)
∂ω12 ∂ω22
They do not produce great novelties, the same BC-holomorphic function theory
provides all the necessary information. This can be obtained in two ways. First of
all, the simple holomorphic changes of variables

(w1 , w2 ) → (z1 , −iz2 ), (ω1 , ω2 ) → (ζ1 , −jζ2 )

turn the operator C2 (i) into ΔC2 (i) and the operator C2 (j) into ΔC2 (j) . But they
can be factorized directly:
 
∂ ∂ ∂ ∂
C2 (i) = +k · −k
∂w1 ∂w2 ∂w1 ∂w2
∂ ∂
=4 ;
∂Z ∂Z †
compare with (9.8). Although formally they are the same operators, they are
employed in other forms taken in the table in Section 4. In the same fashion,
 
∂ ∂ ∂ ∂
C2 (j) = +k · −k
∂ω1 ∂ω2 ∂ω1 ∂ω2
∂ ∂
=4 ,
∂Z ∂Z
198 Chapter 9. Second Order Complex and Hyperbolic Differential Operators

where we take, again, another form of writing the operators involved. Somewhat
paradoxically, the same bicomplex “tricks” do not work for the “hyperbolic wave
operator”
∂2 ∂2
2 −
∂w1 ∂w22
which can neither be factorized directly nor reduced to (9.4); this is because in
the hyperbolic world there is only one hyperbolic-type imaginary unit and there
are no complex-type imaginary units.
One can consider another hyperbolic Laplacian:

∂2 ∂2
+
∂w21 ∂w22

but the change of variables

(w1 , w2 ) → (z1 , −kz2 )

reduces it to (9.4).

9.4 Conjugate (complex and hyperbolic) harmonic

functions
Let F be a bicomplex holomorphic function in a domain Ω ⊂ BC which can be
written as F (Z) = f1 (z1 , z2 ) + jf2 (z1 , z2 ) = ρ1 (ζ1 , ζ2 ) + iρ2 (ζ1 , ζ2 ) = f1 (z1 , z2 ) +
if2 (z1 , z2 ). It has been shown above that:

• f1 and f2 are C(i)-complex harmonic functions in Ω ⊂ C2 (i), and they are

related by the Cauchy–Riemann conditions

∂f1 ∂f2 ∂f2 ∂f1

= ; =− .
∂z1 ∂z2 ∂z1 ∂z2

Such a pair of functions f1 and f2 will be called a pair of conjugate C(i)-

complex harmonic functions.
• ρ1 and ρ2 are C( j)-complex harmonic functions in Ω ⊂ C2 ( j), and they are
related by the Cauchy–Riemann conditions

∂ρ1 ∂ρ2 ∂ρ2 ∂ρ1

= ; =− .
∂ζ1 ∂ζ2 ∂ζ1 ∂ζ2

Such a pair of functions ρ1 and ρ2 will be called a pair of conjugate C( j)-

complex harmonic functions.
9.4. Conjugate (complex and hyperbolic) harmonic functions 199

• f1 and f2 are D-harmonic functions in Ω ⊂ D2 , and they are related by the

Cauchy–Riemann conditions
∂f1 ∂f2 ∂f2 ∂f1
= ; =− .
∂z1 ∂z2 ∂z1 ∂z2
Such a pair of functions f1 and f2 will be called a pair of conjugate D-harmonic
functions.
Bicomplex holomorphic functions play an important role in understanding
constant coefficient second-order complex and hyperbolic differential operators, as
emphasized also in e.g. [19, 20, 96], where computational algebra techniques are
used to study such objects.
Chapter 10

Sequences and Series of

Bicomplex Functions

10.1 Series of bicomplex numbers

We consider here briefly the series of the form
∞
$
Zn (10.1)
n=1

with bicomplex numbers Zn as general terms. Writing Zn as Zn = β1,n e + β2,n e†

equation (10.1) becomes
∞
$ $∞
β1,n e + β2,n e† , (10.2)
n=1 n=1

thus the series of bicomplex numbers (10.1) is convergent if and only if both
complex series in (10.2) are convergent.
∞
$
A typical example is the bicomplex geometric series Zn = 1 + Z + Z2 +
n=1
· · · + Z n + · · · which converges for Z in the bicomplex unitary ball, that is, the
bicomplex ball of hyperbolic radius 1, and which diverges outside this ball; this
is because the bicomplex geometric series is equivalent to the pair of complex
geometric series:
$∞ ∞
$ ∞
$
Zn = β1n e + β2n e† .
n=1 n=1 n=1
∞
$
Theorem 10.1.1 (Cauchy criteria). The series Zn converges if and only if the
n=1
sequence of its partial sums is a D-Cauchy sequence, i.e., for any positive hyperbolic

© Springer International Publishing Switzerland 2015 201

M.E. Luna-Elizarrarás et al., Bicomplex Holomorphic Functions, F rontiers in Mathematics,
DOI 10.1007/978-3-319-24868-4_11
202 Chapter 10. Sequences and Series of Bicomplex Functions

number ε there exists a natural number N = N (ε) such that

m 
$ 
 
 ZN +k  ≺ ε
 
k=1 D

for any natural m.

Proof. Follows by applying the Cauchy criteria to the complex series in (10.2). 
This theorem gives a necessary condition of convergence of the series (10.1):
the sequence {Zn } D-converges to zero. Another consequence of the theorem is:
∞
if {An }n=1 is a sequence of non-negative hyperbolic numbers: An ∈ D+ , and if
$∞
 
Zn  ≺ An for n ≥ N , then the convergence of An implies the convergence of
D
n=1
∞
$
Zn .
n=1
∞
$
A bicomplex series Zn is called D-absolutely convergent if the series
n=1
∞
$  
Z n  of hyperbolic numbers is convergent. Of course, a D-absolutely con-
D
n=1
vergent series is convergent.
It is easy to prove that:
• If the bicomplex series (10.1) is D-absolutely convergent, then any series
composed of its members is also convergent and has the same sum as the
initial series.
∞
$
• If the series Zn converges D-absolutely and has the sum Z and if the
n=1
∞
$
series Wn converges to W , then the Cauchy product of the series, that is,
n=1
∞
$ $
n
the series Un with Un = Zk Wn−k , is convergent and its sum equals
n=1 k=0
ZW .

10.2 General properties of sequences and series of

functions
Consider now a sequence of functions fn : Ω ⊂ BC −→ BC. We say that the
sequence converges on Ω to a function f (called the limit function) if for any
∞
Z ∈ Ω the sequence of bicomplex numbers {fn (Z)}n=1 converges to f (Z), that
10.2. General properties of sequences and series of functions 203
 
is, for any ε  0 there exists N = N (ε, Z) such that fn (Z) − f (Z)D ≺ ε for
n ≥ N (ε, Z). Since the functions fn are not necessarily bicomplex holomorphic
their idempotent components F1,n and F2,n depend, in general, on β1 and β2 , not
∞
on one of them. But the convergence of {fn }n=1 is equivalent to the convergence
∞ ∞
of the two complex sequences {F1,n }n=1 and {F2,n }n=1 on Ω.
∞
A sequence {fn }n=1 is D-uniformly convergent to f if for any ε  0 the
number
 N depends
 on ε only: N = N (ε), and thus for n ≥ N the inequality
fn (Z) − f (Z) ≺ ε holds for all Z. The D-uniform convergence holds if and only
D
 for any ε  0 there exists N = N (ε) such that for m ≥ N , n ≥ N one has:
if
fn (Z)−fm (Z) ≺ ε. Of course, a D-uniformly convergent sequence is convergent.
D

Theorem 10.2.1. If all the functions fn are continuous on Ω and the sequence
converges D-uniformly on Ω, then the limit function f is continuous on Ω.
Proof. Mimics the complex functions case using the properties of hyperbolic pos-
itive numbers instead of real positive numbers. 

Theorem 10.2.2. Let Γ be a rectifiable curve in BC without self-intersections and

∞
let a sequence {fn }n=1 converge D-uniformly on Γ, then
/ /
lim fn (t) dt = lim fn (t) dt.
n→∞ Γ Γ n→∞

Proof. Follows the same line as the previous one. 

Theorem 10.2.3 (Bicomplex holomorphy of the limit functions; an analogue of

Weierstrass’ theorem). If every fn is bicomplex holomorphic in Ω = Ω1 e + Ω2 e†
∞
and if the sequence {fn }n=1 converges D-uniformly on every compact set in Ω to a
∞f , then f is bicomplex holomorphic in Ω and for every k ∈ N the sequence
function

(k)
fn of the k-th derivatives converges D-uniformly on compacts to f (k) .
n=1

Proof. Since each fn is bicomplex holomorphic, then fn (Z) = F1,n (β1 )e+F2,n (β2 )e†
where for any n the functions F1,n are holomorphic in Ω1 and all F2,n are holo-
morphic in Ω2 . If K is a compact subset in Ω and K = K1 e + K2 e† , then K1 is
∞
a compact subset in Ω1 and K2 is a compact subset in Ω2 . Moreovoer, {fn }n=1
∞
converges D-uniformly on K if and only if {F1,n }n=1 converges uniformly on K1
∞
and {F2,n }n=1 converges uniformly on K2 .
Note that a priori the limit function f does not need to be bicomplex holo-
morphic, in general, thus, in its idempotent representation

f (Z) = f (β1 e + β2 e† ) = F1 (β1 e + β2 e† )e + F2 (β1 e + β2 e† )e†

the components F1 and F2 would depend on both variables β1 and β2 . But the
∞
complex Weierstrass theorem implies that the sequence {F1,n }n=1 converges to
∞
F1 uniformly on compacts and {F2,n }n=1 converges to F2 uniformly on compacts,
204 Chapter 10. Sequences and Series of Bicomplex Functions

thus, F1 is holomorphic in Ω1 and F2 is holomorphic in Ω2 . Because of the unique-

ness of the limit function, F1 depends on β1 only and F2 depends on β2 only.
Recalling that f = F1 e + F2 e† we conclude that
 f is bicomplex
∞  holomorphic
∞ in Ω.
(k) (k)
Moreover, for any k ∈ N the sequences F1,n and F2,n converge
n=1 n=1
(k) (k) (k)
uniformly on compacts to F1 and F2 respectively. Hence, the sequence fn =
(k) (k) (k) (k)
F1,n e+F2,n e† converges D-uniformly on compacts to the function F1 e+F2 e† =
f (k) which completes the proof. 

10.3 Convergent series of bicomplex functions

We now consider series of the form
∞
$
fn (Z) = f1 (Z) + f2 (Z) + · · · + fn (Z) + · · · (10.3)
n=1

where all the functions fn are defined on the set Ω in BC. Their behavior is char-
$n
acterized by the sequence Sn (Z) = fk (Z) of partial sums, hence the properties
k=1
of sequences (convergence at a point, D-uniform convergence on the set, etc.) ap-
∞
ply to the sequence {Sn (Z)}n=1 and become the respective properties of series
of functions. For instance, in order that the series of functions (10.3) converges
D-uniformly on Ω it is necessary and sufficient  that for any positive hyperbolic

number ε there exists N = N (ε) such that fN +1 (Z) + · · · + fN +m (Z)D ≺ ε for
any Z ∈ Ω and any m ∈ N (Cauchy’s criteria).
∞
$ ∞
$  
A series fn (Z) is called D-absolutely convergent if the series fn (Z)
D
n=1 n=1
converges on Ω.
Theorem 10.3.1 (Weierstrass test for D-uniform convergence). Given a series of
$∞ $∞
functions fn (Z), if there exists a convergent series An of hyperbolic non-
n=1   n=1
negative numbers such that the inequality fn (Z) An holds for n = n∗ , n∗ +
$∞
1, . . . , for some n∗ ∈ N and for all Z in Ω, then the series fn (Z) converges
n=1
D-uniformly and D-absolutely on Ω.
Proof. Consists in applying Cauchy’s criteria twice. 
The properties of sequences of functions give rise to the following statements
about series of functions.
• If the functions fn are continuous on Ω and the series converges D-uniformly
on Ω, then the sum S(Z) is a continuous function.
10.4. Bicomplex power series 205

• Let Γ be a rectifiable curve without self-intersections, let every function fn

$∞
be continuous and assume that the series fn converges D-uniformly on
n=1
Γ, then
/ $
∞ ∞ /
$
fn (t) dt = fn (t) dt.
Γ n=1 n=1 Γ

∞
$
• (Weierstrass theorem.) If every fn is bicomplex holomorphic in Ω and fn (Z)
n=1
converges D-uniformly on compact subsets of Ω, then the sum S is bicom-
∞
$
plex holomorphic in Ω and the series fn(k) (Z) converges D-uniformly on
n=1
compact subsets in Ω to S (k) (Z) for any k ∈ N.

10.4 Bicomplex power series

Here we deal with the series of the form
∞
$
An (Z − Z0 )n (10.4)
n=0

with bicomplex coefficients An . Its members An (Z − Z0 )n are defined for any

Z ∈ BC but, of course, it is a separate question to determine where such a series
is convergent.
Equivalently, one can consider the series
∞
$
An Z n (10.5)
n=0

which is, obviously, convergent at the origin Z = 0.

Theorem 10.4.1
 (Analogue

of Abel’s theorem). Consider a series (10.5) and let
S = Sρ = Z  |Z|D = ρ be a bicomplex sphere of a hyperbolic radius ρ  0 which
is assumed to be an invertible hyperbolic number. If there is a point Z" in S such
$∞
that the series An Z n converges at Z," then it converges D-absolutely in the
n=0
open bicomplex ball B = Bρ of the same radius, and it converges D-uniformly on
compact subsets of this ball.
Proof. First of all, note that if ρ is a zero-divisor, then any point of S is a zero-
divisor, while if ρ is invertible, then all the points of S are invertible bicomplex
numbers.
206 Chapter 10. Sequences and Series of Bicomplex Functions

Let Z" ∈ S be a 
point of convergence.
 Take K to be a closed subset of Bρ
Z  
and consider the set    Z ∈ K . It is D-bounded and, hence, it has a D-
Z" D   
Z  
  
supremum, supD ; what is more, supD   Z ∈ K =: q ≺ 1. Since the se-
Z" D
∞
$
ries An Z"n converges, then lim An Z" n = 0, thus the sequence An Z" n is D-
n→∞
n=0  
bounded: there exists a positive hyperbolic number M such that An Z"n D M
for n ∈ N. For Z ∈ K we get now an estimate
  n 
   
An Z n  = An Z" n · Z  M q n .
 
D
Z" D

The series with general term M q n converges and it may serve as a majorant for
∞
$
the series An Z n on K; the Weierstrass theorem allows us to conclude that
n=0
∞
$
An Z n
converges D-absolutely and D-uniformly on K. Since any point of Bρ
n=0
belongs to some subset K, the theorem is proved. 
As was noted above, if ρ is a zero-divisor, i.e., ρ = ρ1 e or ρ = ρ2 e† , then
all the points of Sρ are zero-divisors of the same form. What can be said if
$∞
Z" ∈ Sρ and the series An Z"n converges? Writing An = an e + bn e† we get:
n=0 ∞  ∞ 
$∞ $∞
    $ $
An Z"n = an e + bn e† β"1 e + β"2 e† = an β"n e +
1 bn β"n e† .
2
n=0 n=0 n=0 n=0
Since Z" is of one of the forms: Z" = β"1 e or Z" = β"2 e† , then we can make
$∞
a conclusion for one of the bicomplex components of the series An Z n =
∞  ∞  n=0
∞
$ $ $
an β1n e + bn β2n e† . Indeed, if ρ = ρ1 e, then the series an β1n e
n=0 n=0 n=0
∞
$
converges for |β1 | < ρ1 but we have no information for the series bn β2n e† ;
n=0
similarly for ρ = ρ2 e† . Of course, this does not mean that there is no information
in principle; we have just taken an “unlucky” point Z. "

Proposition 10.4.2 (Sets of convergence of bicomplex power series). Given a series

$∞
An Z n , one of the following options holds:
n=0

(1) the series converges at the origin only;

10.4. Bicomplex power series 207

(2) the series converges on the whole BC;

(3) the series converges on one of the idempotent complex lines BCe or BCe† and
diverges on the other;
(4) the series converges on the disk centered at the origin and of (real) radius r
situated on BCe or BCe† and diverges on the complement of its closure;
(5) the series converges on B(0, r)e + BCe† or on BCe + B(0, r)e† with r > 0 and
diverges on the complement of its closure;
(6) the series converges on the bicomplex ball centered at the origin of hyperbolic
radius R = r1 e + r2 e†  0 and diverges on the complement of its closure.
$∞ ∞
$Zn
Proof. The first two options are illustrated with the series .n! Z n and
n=0 n=0
n!
Assume next that the first two cases are excluded and consider the idempo-
tent representation
∞
∞  ∞ 
$ $ $
An Z n = an β1n e + bn β2n e† .
n=0 n=0 n=0

Several possibilities arise now. Let, first, one of the idempotent series converge at
the origin only, then the other may converge in a disk of finite radius which gives
the case (4) or in the whole C(i) which gives the case (3); illustrations for (3) are
∞ 
$ $∞ 
1 n † 1 n n †
n! β1n e + β e or β e + n! β2 e
n=0
n! 2 n=0
n! 1

and illustrations for (4) are

∞
$ ∞
$
   
γn β1n e + n! β2n e† or n! β1n e + γn β2n e†
n=0 n=0

∞
$
where γn z n is a complex series with a (real) radius of convergence r > 0. Sec-
n=0
ond, both idempotent series converge on a disk which may be of a finite or infinite
∞
$
radius thus justifying the cases (5) and (6); for the same complex series γn z n
∞  ∞ 
n=0
$ 1 n † $ 1 n n †
any of the bicomplex series γn β1n e + β e or β e + γn β2 e
n=0
n! 2 n=0
n! 1
illustrates the former case. 
Analogously to the case of complex analysis, we may speak in all the cases
about bicomplex balls BR of hyperbolic radii R where:
208 Chapter 10. Sequences and Series of Bicomplex Functions

(1) R = 0 = 0e + 0e† , and the ball reduces to a point;

(2) R = ∞ = ∞e + ∞e† , and the ball means the whole BC;
(3) R = 0e + ∞e† or R = ∞e + 0e† , and the ball means one of the complex lines
BCe or BCe† ;
(4) R = re or R = re† , and the ball means one of the “labeled” disks B(0, r)e
or B(0, r)e† ;
(5) R = re + ∞e† or R = ∞e + re† ;
(6) R = r1 e + r2 e† , and this is an “authentic” bicomplex ball of a hyperbolic
radius.
This R will be called the hyperbolic radius of convergence of a bicomplex
power series.
Proposition 10.4.3. If R is strictly positive (i.e., R = r1 e + r2 e† , or R = ∞, or
R = re + ∞e† or R = ∞e + re† ), then the sum S(Z) of a power series is a
bicomplex holomorphic function and
∞
$
S  (Z) = n An Z n−1 .
n=1

Proof. Follows directly from the theorems of Abel and Weierstrass. 

S (n) (0)
Proposition 10.4.4. If R is strictly positive, then An = .
n!
Proof. It follows from the previous proposition that
∞
$
S (k) (Z) = n(n − 1) · · · (n − k + 1) An Z (n−k)
n=k

for k ∈ N. In particular, S (k) (0) = k(k − 1) · · · 1 Ak = k! Ak . 

10.5 Bicomplex Taylor Series

Definition 10.5.1. Let Ω be a domain in BC, and assume that a function f : Ω −→
C is such that it has derivatives of any order at Z0 ∈ Ω. Then the power series
$∞
f (n) (Z0 ) n
(Z − Z0 ) is called the Taylor series for f at Z0 .
n=0
n!
∞
$ n
Theorem 10.5.2. If a power series An (Z − Z0 ) has a strictly positive radius
n=0
of convergence R, then it is the Taylor series for its sum at Z0 .
10.5. Bicomplex Taylor Series 209

S (n) (Z0 )
Proof. By Proposition 10.4.4, An = , thus
n!
∞
$ $∞
n S (n) (Z0 ) n
S(Z) = An (Z − Z0 ) = . (Z − Z0 ) 
n=0 n=0
n!

Theorem 10.5.3. Let Ω be a product-type domain in BC, Ω = Ω1 e + Ω2 e† , let f be

bicomplex holomorphic in Ω. For a fixed Z0 in Ω consider the Taylor series of f at
Z0 . Denote by d the hyperbolic distance between Z0 and the distinguished boundary
of Ω, that is,
 

d := inf D Z − Z0 D | Z ∈ ∂Ω1 e + ∂Ω2 e† ∈ D.

Then the Taylor series converges in the bicomplex ball Bd (Z0 ) where its sum co-
incides with f (Z):
$∞
f (n) (Z0 ) n
f (Z) = (Z − Z0 )
n=0
n!

for any Z ∈ Bd (Z0 ).

Proof. As f is bicomplex holomorphic we can write

f (Z) = f (β1 e + β2 e† ) = F1 (β1 )e + F2 (β2 )e† ,

where F1 and F2 are (complex) holomorphic, respectively, in Ω1 and Ω2 , and

Z0 = β10 e + β20 e† ,

where β10 ∈ Ω1 , β20 ∈ Ω2 . Denote by d1 the Euclidean distance between β10 and
∂Ω1 , and by d2 the Euclidean distance between β20 and ∂Ω2 , then by the complex
Taylor theorem one has:

$∞ (n)
F1 (β10 )  n
F1 (β1 ) = β1 − β10
n=0
n!

 
for β1 ∈ B β10 , d1 , and

$∞ (n)
F2 (β20 )  n
F2 (β2 ) = β2 − β20
n=0
n!

 
for β2 ∈ B β20 , d2 . Hence
210 Chapter 10. Sequences and Series of Bicomplex Functions

f (Z) = F1 (β1 )e + F2 (β2 )e†

$∞ (n) $∞ (n)
F1 (β10 )  n F2 (β20 )  n
= β1 − β10 e + β2 − β20 e†
n=0
n! n=0
n!
1  (n) 0   n 
$∞
(n) n 
= F1 (β1 ) e + F2 (β20 ) e† β1 − β10 e + β2 − β20 e†
n=0
n!
$∞
f (n) (Z0 ) n
= (Z − Z0 )
n=0
n!
   
for any Z in B β10 , d1 e + B β20 , d2 e† = Bd (Z0 ). The proof is completed. 
Example 10.5.4. Recalling the formulas for the derivatives of elementary functions
we obtain the following expansions into bicomplex Taylor series:
$∞ ∞
$ ∞
$
Zn n Z 2n+1 n Z 2n
eZ = ; sin Z = (−1) ; cos Z = (−1) .
n=0
n! n=0
(2n + 1)! n=0
(2n)!

In addition to the book of Price [56], we refer the reader also to publications
such as [40, 41, 84], where certain properties of bicomplex sequences and series are
studied.
Chapter 11

Integral Formulas and

Theorems

In this chapter we establish bicomplex analogues of the main integral theorems

and formulas of one-dimensional complex analysis. We are not going to reach
the highest level of generality for curves and surfaces involved since our aim is
to present some basic ideas and structures for those formulas; the more general
setting will be presented elsewhere.

11.1 Stokes’ formula compatible with the bicomplex

Cauchy–Riemann operators
Let Ω be a domain in BC and consider a function F of class C 1 (Ω, BC). It is thus
real differentiable and, recalling (7.48), we have that its real differential dF is

∂F ∂F ∂F ∂F
dF = dZ + dZ + dZ † + dZ ∗ .
∂Z ∂Z ∂Z † ∂Z ∗

Note that this is just a “bicomplex combination” of the real differentials of the
components of F ; hence the function itself and the variable Z can be written in
any form as well as all the differentials and the differential operators; for instance,
dZ = dx1 + idy1 + jdx2 + kdy2 = dz1 + jdz2 = dζ1 + idζ2 = . . ., dZ = dx1 −
idy1 + jdx2 − kdy2 = dz 1 + jdz 2 = . . ..
Next, consider the action of the exterior differentiation operator on the dif-
ferential form F dZ; we have that

© Springer International Publishing Switzerland 2015 211

M.E. Luna-Elizarrarás et al., Bicomplex Holomorphic Functions, F rontiers in Mathematics,
DOI 10.1007/978-3-319-24868-4_12
212 Chapter 11. Integral Formulas and Theorems

d(F dZ) = dF ∧ dZ

∂F ∂F ∂F † ∂F ∗
= dZ + dZ + dZ + dZ ∧ dZ
∂Z ∂Z ∂Z † ∂Z ∗ (11.1)
∂F ∂F ∂F
= dZ ∧ dZ + †
dZ † ∧ dZ + dZ ∗ ∧ dZ.
∂Z ∂Z ∂Z ∗
The same can be done with the three other differential forms, that is, F dZ, F dZ † ,
F dZ ∗ , arriving at
∂F ∂F ∂F
d(F dZ) = dZ ∧ dZ + dZ † ∧ dZ + dZ ∗ ∧ dZ; (11.2)
∂Z ∂Z † ∂Z ∗
∂F ∂F ∂F
d(F dZ † ) = dZ ∧ dZ † + dZ ∧ dZ † + ∗
dZ ∗ ∧ dZ † ; (11.3)
∂Z ∂Z ∂Z
∂F ∂F ∂F
d(F dZ ∗ ) = dZ ∧ dZ ∗ + dZ ∧ dZ ∗ + dZ † ∧ dZ ∗ . (11.4)
∂Z ∂Z ∂Z †
Of course, each of the formulas (11.2), (11.3) and (11.4) can be obtained also by
making the corresponding change of variable.
Next, take Γ to be a two-dimensional, piecewise smooth, oriented surface in
Ω whose boundary γ = ∂Γ is a piecewise smooth curve.
Then integrating both parts of (11.1) and applying Stokes theorem we get:
/ /
F dZ = d(F dZ)
γ Γ
/  (11.5)
∂F ∂F † ∂F ∗
= dZ ∧ dZ + dZ ∧ dZ + dZ ∧ dZ .
Γ ∂Z ∂Z † ∂Z ∗
Similar formulas arise if one uses (11.2), (11.3) and (11.4).
Formula (11.5) contains several special cases which are of interest by them-
selves. In particular, we can take F as
F (z1 , z2 ) = f1 (z1 , z2 ) + jf2 (z1 , z2 )
where f2 (z1 , z2 ) = 0 in Ω and f1 is holomorphic in the sense of two complex
variables z1 and z2 which leads to
/ / 
∂f1 ∂f1
f1 (z1 , z2 )(dz1 + jdz2 ) = +j j dz1 ∧ dz2
γ Γ ∂z1 ∂z2
∂f1 ∂f1
since = = 0 in Ω and
∂Z ∂Z ∗
dZ † ∧ dZ = 2 j dz1 ∧ dz2 .
Separation of the complex components results in the formulas
/ /
∂f1
f1 (z1 , z2 )dz1 = − dz1 ∧ dz2 ;
γ Γ ∂z2
11.1. Stokes’ formula compatible with the bicomplex CR operators 213
/ /
∂f1
f1 (z1 , z2 )dz2 = dz1 ∧ dz2 .
γ Γ ∂z1
The formulas remain true if, in addition, f does not depend on z2 :
/
f1 (z1 )dz1 = 0,
γ
/ /
f1 (z1 )dz2 = f1 (z1 )dz1 ∧ dz2 .
γ Γ
Analogously, if f1 does not depend on z1 , then
/ /
f1 (z2 )dz1 = − f1 (z2 )dz1 ∧ dz2 ,
γ Γ
/
f1 (z2 )dz2 = 0.
γ
But F can be taken also as
F (ζ1 , ζ2 ) = ρ1 (ζ1 , ζ2 ) + iρ2 (ζ1 , ζ2 )
where ρ2 (ζ1 , ζ2 ) = 0 in Ω and ρ1 is holomorphic in the sense of two complex
variables ζ1 and ζ2 ; now we can repeat the reasoning and arrive at very similar
formulas and conclusions.
Finally, F can be taken as
F (z1 , z2 ) = f1 (z1 , z2 ) + if2 (z1 , z2 )
where z1 and z2 are hyperbolic variables, f1 and f2 are D-valued functions; what is
more, we assume that f2 (z1 , z2 ) = 0 in Ω and that f1 is holomorphic in the sense
∂f1 ∂f1
of two hyperbolic variables z1 and z2 . Then = = 0 and dZ ∗ ∧ dZ =
∂Z ∂Z †
2 idz1 ∧ dz2 ; thus (11.5) gives the equality
/ / 
∂f1 ∂f1
f1 (z1 , z2 )(dz1 + idz2 ) = +i i dz1 ∧ dz2 .
γ Γ ∂z1 ∂z2
Separation of the hyperbolic components gives:
/ /
∂f1
f1 (z1 , z2 )dz1 = − dz1 ∧ dz2 ,
γ Γ ∂z 2
/ /
∂f1
f1 (z1 , z2 )dz2 = dz1 ∧ dz2 .
γ Γ ∂z1

The same comments as above can be made although the situation is rather different
because holomorphic functions of two hyperbolic variables do not have yet their
theory.
But, of course, for us the most important consequence of (11.5) is the bicom-
plex Cauchy integral theorem.
214 Chapter 11. Integral Formulas and Theorems

Theorem 11.1.1 (Bicomplex Cauchy integral theorem). Let F be a bicomplex holo-

morphic function in a product-type domain Ω ⊂ BC. If γ is any piecewise smooth
curve which is the boundary of a two-dimensional, piecewise smooth surface Γ ⊂ Ω,
then
/
f (Z) dZ = 0.
γ

∂F ∂F ∂F
Proof. Since F is bicomplex holomorphic, then = = = 0 in Ω, and
∂Z ∂Z † ∂Z ∗
the result follows from (11.5). 

11.2 Bicomplex Borel–Pompeiu formula

Having in mind further developments and applications, we are not looking for a
high level of generality; our aim is to obtain a bicomplex analogue of the com-
plex Borel–Pompeiu formula which is based on the idempotent representation of
bicomplex numbers. This requires us to work with curves and surfaces in BC of
particular shape and with BC-valued C 1 -functions of a particular structure. It is
explained more precisely immediately after this brief introduction.
Let Ω be a domain in BC, and consider a two-dimensional, simply connected,
piecewise smooth surface Γ ⊂ Ω with boundary γ = ∂Γ ⊂ Ω which has the
following properties: Γ has a parametrization ψ = ψ(u, v) such that ψ = ψ1 e+ψ2 e†
where ψ1 and ψ2 are the parametrizations, respectively, of domains Γ1 and Γ2 in
C(i) which are simply connected; γ has a parametrization ϕ = ϕ(t) which is the
restriction of ψ onto ∂Γ and is such that ϕ = ϕ1 e + ϕ2 e† where ϕ1 and ϕ2 are
the parametrizations, respectively, of γ1 := ∂Γ1 and of γ2 := ∂Γ2 , γ1 and γ2 being
piecewise smooth, closed, Jordan curves in C(i).
We illustrate now with an example that the restrictions on Γ and γ are not
contradictory. Let Γ1 be a domain in C(i) with the above described properties and
let Γ2 be another domain in C(i) which is conformally equivalent to Γ1 via the
mapping ψ2 : Γ1 → Γ2 ; if we denote as ψ1 : u + iv ∈ Γ1 → u + iv ∈ Γ1 the identity
mapping, then the parametrization ψ := ψ1 (u, v)e + ψ2 (u, v)e† determines the
surface Γ in BC with the necessary properties.
As the next step we consider the integrals over γ and Γ. If g is a continuous
function on γ, then we assume additionally that g is of the form g(Z) = g1 (β1 )e +
g2 (β2 )e† for Z = β1 e + β2 e† . Note that this property is true automatically for
bicomplex holomorphic functions, but for continuous functions their idempotent
coefficients depend, in general, on both variables β1 and β2 . The requirement above
is caused by the method which we use below. Now, using the definition and the
properties of the integral of the differential form g(Z) dZ we have:
11.2. Bicomplex Borel–Pompeiu formula 215

/ / b
g(Z) dZ = g(ϕ(t)) ϕ (t) dt
γ a
/ b   
= g1 (ϕ1 (t))e + g2 (ϕ2 (t))e† ϕ1 (t)e + ϕ2 (t)e† dt
a
/ b  
= g1 (ϕ1 (t)) ϕ1 (t)e + g2 (ϕ2 (t)) ϕ2 (t)e† dt
a
/ /
†
=e g1 (β1 ) dβ1 + e g2 (β2 ) dβ2 ,
γ1 γ2

that is, / / /
g(Z) dZ = e g1 (β1 ) dβ1 + e† g2 (β2 ) dβ2 . (11.6)
γ γ1 γ2

As a matter of fact, the first equality is the definition of the integral of the
bicomplex differential form g(Z) dZ along γ, and it is easy to see that this definition
is well posed.
A similar reasoning applies to the surface Γ and the differential form g(Z) dZ∧
dZ ∗ . Indeed,
/ /
∗
   
g(Z) dZ ∧ dZ = g(Z) dβ1 e + dβ2 e† ∧ dβ 1 e + dβ 2 e†
Γ
/Γ
  
= g1 (β1 )e + g2 (β2 )e† dβ1 ∧ dβ 1 e + dβ2 ∧ dβ 2 e†
/Γ
 
= g1 (β1 ) dβ1 ∧ dβ 1 e + g2 (β2 ) dβ2 ∧ dβ 2 e†
Γ
/  
∂ψ1 ∂ψ 1 ∂ψ1 ∂ψ 1
= g1 (ψ1 (u, v)) · − · e
(u,v) ∂u ∂v ∂v ∂u

∂ψ2 ∂ψ 2 ∂ψ2 ∂ψ 2
+ g2 (ψ2 (u, v)) · − · e† du dv
∂u ∂v ∂v ∂u
/ /
= g1 (β1 ) dβ1 ∧ dβ 1 e + g2 (β2 ) dβ2 ∧ dβ 2 e† ,
Γ1 Γ2

that is,
/ / /
∗
g(Z) dZ ∧dZ = g1 (β1 ) dβ1 ∧ dβ 1 e+ g2 (β2 ) dβ2 ∧ dβ 2 e† . (11.7)
Γ Γ1 Γ2

Remark 11.2.1. Under the same hypotheses, we say that the function g is integrable
in the improper sense along Γ if the function g is integrable in the improper sense
along the domain Γ for  = 1 and  = 2.
Theorem 11.2.2 (Bicomplex Borel–Pompeiu formula). Let g ∈ C 1 (Ω) be such that
g(Z) = g1 (β1 )e + g2 (β2 )e† , Z = β1 e + β2 e† , and let γ and Γ be as described above.
216 Chapter 11. Integral Formulas and Theorems

Then for any Z ∈ Γ \ γ,

/ / ∂g
1 g(t) dt 1 ∂t∗ dt ∧ dt∗ ,
g(Z) = + (11.8)
2πi γ t−Z 2πi Γ t−Z

∂ ∂ ∂
where t = t1 e + t2 e† and =e + e† .
∂t∗ ∂t1 ∂t2
Proof. One has:
g(Z) = g1 (β1 )e + g2 (β2 )e† .
Using the complex Borel–Pompeiu formula gives:

∂g1
/ /
1 g1 (t1 ) dt1 1 ∂t1
g(Z) = e+ dt1 ∧ dt1 e
2πi γ1 t1 − β 1 2πi Γ1 1 − β 1
t
∂g2
/ /
1 g2 (t2 ) dt2 † 1 ∂t2
+ e + dt2 ∧ dt2 e†
2πi γ2 t2 − β 2 2πi Γ2 t2 − β2

which, by regrouping the terms, equals

/    
1 g1 (t1 ) e + g2 (t2 ) e† · dt1 e + dt2 e†
2πi γ (t1 e + t2 e† ) − (β1 e + β2 e† )

∂ † ∂
/ e +e [g1 e + g2 e† ]
1 ∂t1 ∂t2  
+ † †
dt1 ∧ dt1 e + dt2 ∧ dt2 e†
2πi Γ (t1 e + t2 e ) − (β1 e + β2 e )
/ / ∂g (t)
1 g(t) dt 1 ∂t∗
= + dt ∧ dt∗ .
2πi γ t − Z 2πi Γ t − Z

This completes the proof. 

Note that we used in the proof formulas (11.6) and (11.7) and Remark 11.2.1.
Theorem 11.2.3 (The bicomplex Cauchy integral representation). Let Ω be a
product-type domain in BC, let f be a bicomplex holomorphic function in Ω, and
let Z be an arbitrary point in Ω. Then, for any surface Γ ⊂ Ω passing through Z
and with the above described properties the Cauchy representation formula holds:
/
f (t)
F (Z) = dt,
γ t−Z

where γ = ∂Γ.
11.2. Bicomplex Borel–Pompeiu formula 217

Proof. Since a bicomplex holomorphic function is of the form f (Z) = f1 (β1 )e +

∂f ∂f ∂f
f2 (β2 )e† , then we can apply (11.8). Moreover, as we know, = †
= =0
∂Z ∂Z ∂Z ∗
and thus the surface integral in (11.8) vanishes. 
Note that if a C 1 -function g is of the form g(Z) = g1 (β1 )e + g2 (β2 )e† , then
∂g ∂g
the conditions = †
= 0 are valid. Hence the bicomplex holomorphic func-
∂Z ∂Z
tions are singled out among the C 1 -functions of this form by the unique condition
∂g
= 0, which explains why in the Borel–Pompeiu formula just one operator
∂Z ∗
∂
remains.
∂Z ∗
We have shown how Stokes’ formula can be used to obtain the bicomplex
Cauchy Integral Theorem, the bicomplex Borel-Pompeiu formula, and the bicom-
plex Cauchy Integral Representation formula.
Integration theory in the context of complexified Clifford Analysis, in which
bicomplex analysis constitutes a first step, have been extensively studied by Ryan
[68]–[81].
Bibliography

[1] L.V. Ahlfors. Complex Analysis. McGraw–Hill Book Co. (1966).

[2] D. Alpay, M.E. Luna, M. Shapiro, D.C. Struppa. Basics of func-
tional analysis with bicomplex scalars, and bicomplex Schur analysis. Series
SpringerBriefs in Mathematics (2014).
[3] H.T. Anastassiu, P.E. Atlamazoglou, D.I. Kaklamani. Application
of bicomplex (quaternion) algebra to fundamental electromagnetics: a lower
order alternative to the Helmholtz equation. IEEE Trans. Antennas and
Propagation, v. 51, No. 8 (2003), 2130–2136.
[4] F. Babadag, Y. Yayli, N. Ekmekci. Homothetic motions at (E 8 ) with
bicomplex numbers (C3 ). Int. J. Contemp. Math. Sci. v. 4, No. 33–36 (2009),
1619–1626.
[5] F. Babadag, Y. Yayli, N. Ekmekci. Homothetic motions and bicomplex
numbers. Commun. Fac. Sci. Univ. Ank. Ser A1 Math. Stat. v. 58, No. 1
(2009), 23–28.
[6] F. Babadag. Homothetic motions and bicomplex numbers. Algebras Groups
Geom. v. 26, No. 2 (2009), 193–201.
[7] P. Baird, J.C. Wood. Harmonic morphisms and bicomplex manifolds. J.
Geom. Phys. v. 61 (2011), 46–61.
[8] P. Berglez. On some classes of bicomplex pseudoanalytic functions.
Progress in analysis and its applications, Hackensack, NJ, World Sci. Publ.
(2010), 81–88.
[9] J. Bory-Reyes, M. Shapiro. Clifford analysis versus its quaternionic
counterparts. Math. Methods Appl. Sciences, v. 33, issue 9, (2010), 1089-
1101.
[10] F. Brackx, H. De Schepper, V. Soucek. On the Structure of Complex
Clifford Algebra. Adv. Appl. Clifford Algebras v. 21 (2011), 477–492.
[11] F. Catoni, D. Boccaletti, R. Cannata, V. Catoni, E. Nichelatti,
P. Zampetti. The Mathematics of Minkowski Space-Time. Birkhäuser-
Basel, 2008.

© Springer International Publishing Switzerland 2015 219

M.E. Luna-Elizarrarás et al., Bicomplex Holomorphic Functions, F rontiers in Mathematics,
DOI 10.1007/978-3-319-24868-4
220 Bibliography

[12] K.S. Charak, D. Rochon, N. Sharma. Normal families of bicomplex

holomorphic functions. Fractals v. 17 No. 3 (2009), 257–268.
[13] K.S. Charak, D. Rochon, N. Sharma. Normal families of bicomplex
meromorphic functions. Ann. Polon. Math. v. 103 No. 3 (2012), 303–317.
[14] K.S. Charak, D. Rochon. On factorization of bicomplex meromorphic
functions. Hypercomplex Analysis, Series Trends Math., Birkhäuser, Basel
(2009), 55–68.
[15] J. Cockle. On Certain Functions Resembling Quaternions and on a New
Imaginary in Algebra. London-Dublin-Edinburgh Philosophical Magazine,
series 3, v. 33 (1848), 43–59.
[16] J. Cockle. On a New Imaginary in Algebra. London-Dublin-Edinburgh
Philosophical Magazine, series 3, v. 34 (1849), 37–47.
[17] J. Cockle. On the Symbols of Algebra and on the Theory of Tessarines.
London-Dublin-Edinburgh Philosophical Magazine, series 3, v. 34 (1849),
40610.
[18] J. Cockle. On Impossible Equations, on Impossible Quantities and on Tes-
sarines. London-Dublin-Edinburgh Philosophical Magazine, series 3, v. 37
(1850), 2813.
[19] F. Colombo, I. Sabadini, D.C. Struppa, A. Vajiac, M.B. Vajiac.
Bicomplex hyperfunctions. Ann. Mat. Pura Appl. (4) v. 190, No. 2 (2011),
247–261.
[20] F. Colombo, I. Sabadini, D.C. Struppa, A. Vajiac, M.B. Vajiac.
Singularities of functions of one and several bicomplex variables. Ark. Mat.
v. 49, No. 2 (2011), 277–294.
[21] V. Cruceanu. Almost product bicomplex structures on manifolds. An. Sti-
int. Univ. Al. I. Cuza Iasi. Mat. (N.S.) v. 51, No. 1 (2005), 99–118.
[22] V. Cruceanu. A product bicomplex structure on the total space of a vector
bundle. An. Stiint. Univ. Al. I. Cuza Iasi. Mat. (N.S.) v. 53, No. 2 (2007),
315–324.
[23] R. Delanghe, F. Sommen, V. Souček. Clifford Algebra and Spinor–
Valued Functions. Kluwer Academic Publishers, (1992).
[24] S. Dimiev, R. Lazov, S. Slavova. Remarks on bicomplex variables and
other similar variables. Topics in contemporary differential geometry, com-
plex analysis and mathematical physics, World Sci. Publ. Hackensack, NJ.
(2007), 50–56.
[25] M.A. Dzavadov, N.T. Abbasov. Bicomplex and biquaternionic hyperbolic
spaces. Azerbaidzan. Gos. Univ. Ucen. Zap. Ser. Fiz.-Mat. i Him. Nauk, No.
1 (1964), 9–15.
Bibliography 221

[26] L.D. Egorova, L.I. Krjuckova, L.B. Lobanova. Bicomplex and bidual
spaces. Moskov. Oblast. Ped. Inst. Ucen. Zap. v. 262 (1969), 76–103.
[27] R. Fueter. Analytische Funktionen einer Quaternionen Variablen. Comm.
Math. Helv. 4 (1932), 9-20.
[28] S.G. Gal. Introduction to geometric function theory of hypercomplex vari-
ables. Nova Science Publishers, Inc. Chapter 4 (2004), xvi + 319.
[29] V. Garant-Pelletier, D. Rochon. On a generalized Fatou-Julia theorem
in multicomplex spaces. Fractals v. 17, No. 3 (2009), 241–255.
[30] G. Gentili, D.C. Struppa. A new theory of regular functions of a quater-
nionic variable. Adv. Math. v. 216, No. 1 (2007), 279-301.
[31] G. Gentili, C. Stoppato, D.C. Struppa. Regular Functions of a Quater-
nionic Variable. Springer Verlag (2013).
[32] R. Gervais-Lavoie, L. Marchildon, D. Rochon. Infinite-dimensional
bicomplex Hilbert spaces. Ann. Funct. Anal. v. 1 No. 2 (2010), 75–91.
[33] R. Gervais-Lavoie, L. Marchildon, D. Rochon. The bicomplex quan-
tum harmonic oscillator. Nuovo Cimento Soc. Ital. Fis. B, v. 125, No. 10
(2010), 1173–1192.
[34] R. Gervais-Lavoie, L. Marchildon, D. Rochon. Finite-dimensional
bicomplex Hilbert spaces. Adv. Appl. Clifford Algebr. v. 21, No. 3 (2011),
561–581.
[35] K. Gürlebeck, F. Kippig. Complex Clifford-Analysis and Elliptic Bound-
ary Problems. Adv. Appl. Clifford Algebras, v. 5, No. 1 (1995), 51–62.
[36] W.R. Hamilton. On quaternions, or on a new system of imaginaries in
algebra. Philosophical Magazine. Vol. 25, no. 3 (1844), 489–495.
[37] W.R. Hamilton. Lectures on Quaternions: Containing a Systematic State-
ment of a New Mathematical Method. Dublin: Hodges and Smith (1853).
[38] H. Kabadayi, Y. Yayli. Homothetic Motions at E4 with Bicomplex Num-
bers. Adv. Appl. Clifford Algebras, v. 21, No. 2 (2002), 541–546.
[39] Krantz. Several Complex Variables. Second Edition, AMS Chelsea Pub-
lishing (2001).
[40] R.S. Krausshar. Eisenstein Series in Complexified Clifford Analysis.
Comp. Methods and Func. Theory, v. 2, No. 1 (2002), 29–65.
[41] J. Kumar, R.K. Srivastava. On a class of entire bicomplex sequences.
South East Asian J. Math. Math. Sci. v. 5, No. 3 (2007), 47–67.
[42] J. Kumar, R.K. Srivastava. A note on poles of the bicomplex Riemann
zeta function. South East Asian J. Math. Math. Sci. v. 9, No. 1 (2010),
65–75.
222 Bibliography

[43] J. Kumar, R.K. Srivastava. On entireness of bicomplex Dirichlet series.

Int. J. Math. Sci. Eng. Appl. v. 5, No. 2 (2011), 221–228.
[44] D.A. Lakew, J. Ryan. The Intrinsic π-Operator on Domain Manifolds in
Cn+1 . Complex Anal. Oper. Theory, (2010), 271–280.
[45] M.E. Luna–Elizarrarás, M. Shapiro, D.C. Struppa, A. Vajiac. Bi-
complex numbers and their elementary functions. Cubo A Mathematical
Journal v. 14. No. 2 (2012), 61-80.
[46] M.E. Luna–Elizarrarás, M. Shapiro, D.C. Struppa, A. Vajiac.
Complex Laplacian and derivatives of bicomplex functions. Complex Analy-
sis and Operator Theory, v. 7, No. 5, (2013), 1675-1711.
[47] J.E. Marsden, M.J. Hoffman. Basic Complex Analysis. Third edition,
1999, W. H. Freeman and Company.
[48] E. Martineau, D. Rochon. On a bicomplex distance estimation for the
Tetrabrot. Internat. J. Bifur. Chaos Appl. Sci. Engrg. v. 15, No. 9 (2005),
3039–3050.
[49] A.I. Melent’ev. A normalized double sphere and a real model for a con-
nection over the algebra of bicomplex numbers. Proceedings of the Seminar
of the Department of Geometry, Izdat. Kazan. Univ., Kazan, No. VI (1971),
57–69.
[50] G. Moisil. Sur les quaternions monogenes. Bull. Sci. Math. Paris v. 55 (2)
(1931), 169-194.
[51] G. Moisil, N. Teodorescu. Fonctions holomorphes dans l’espace. Math-
ematica (Cluj) 5 (1931), 142-159.
[52] I.A. Morev. A generalisation of the Cauchy-Riemann equations and the
harmonicity of monogenic hypercomplex functions. Izv. Vyss. Ucebn. Zaved.
Matematika, v. 3, No. 4 (1958), 176–182
[53] I.A. Morev. A class of monogenic functions. Mat. Sb. (N.S.), v. 50, No. 92
(1960), 233–240.
[54] S. Olariu. Complex numbers in n Dimensions. North-Holland, Series Math-
ematics Studies First edition, v. 290 (2002), chapter 3.
[55] A.A. Pogorui, R.M. Rodriguez-Dagnino. On the set of zeros of bicom-
plex polynomials. Complex Variables and Elliptic Equations, v. 51, No. 7
(2006), 725–730.
[56] G.B. Price. An Introduction to Multicomplex Spaces and Functions. Mono-
graphs and Textbooks in Pure and Applied Mathematics, 140, Marcel
Dekker, Inc., New York, 1991.
[57] J.D. Riley. Contributions to the theory of functions of bicomplex variable.
Tohoku Math. J. v. 2 (1953), 132–165.
Bibliography 223

[58] D. Rochon. Sur une généralisation des nombres complexes: les

tétranombres. Master Thesis, Université de Montréal (1997), 66pp.
[59] D. Rochon. A generalized Mandelbrot set for bicomplex numbers. Fractals,
v. 8, No. 4 (2000), 355–368.
[60] D. Rochon. Dynamique bicomplexe et théorème de Bloch pour fonctions
hyperholomorphes. PhD Thesis, Université de Montréal (Canada), (2001),
77pp.
[61] D. Rochon. A Bloch constant for hyperholomorphic functions. Complex
Variables Theory Appl. v. 44, No. 2 (2001), 85–101.
[62] D. Rochon. On a generalized Fatou-Julia theorem. Fractals, v. 11, No. 3
(2003), 213–219.
[63] D. Rochon. A bicomplex Riemann zeta function. Tokyo J. Math. v. 27, No.
2 (2004), 357-369.
[64] D. Rochon. On a relation of bicomplex pseudoanalytic function theory to
the complexified stationary Schrödinger equation. Complex Var. Elliptic Equ.
v. 53, No. 6 (2008), 501–521
[65] D. Rochon, M. Shapiro. On algebraic properties of bicomplex and hyper-
bolic numbers. An. Univ. Oradea Fasc. Mat. v. 11 (2004), 71–110.
[66] D. Rochon, S. Tremblay. Bicomplex quantum mechanics. I. The gener-
alized Schrödinger equation. Adv. Appl. Clifford Algebr. v. 14, No. 2 (2004),
231–248.
[67] D. Rochon, S. Tremblay. Bicomplex quantum mechanics. II. The Hilbert
space. Adv. Appl. Clifford Algebr. v. 16, No. 2 (2006), 135–157.
[68] J. Ryan. Topics in Hypercomplex Analysis. Doctoral Thesis, University of
York, Britain (1982).
[69] J. Ryan. Complexified Clifford Analysis. Complex Variables, v. 1 (1982),
119–149.
[70] J. Ryan. Singularities and Laurent expansions in complex Clifford Analysis.
Appl. Anal. v. 16 (1983), 33–49.
[71] J. Ryan. Special functions and relations within complex Clifford Analysis,
I. Complex Variables Theory Applications, v. 2 (1983), 177–198.
[72] J. Ryan. Hilbert modules with reproducing kernels within complex Clifford
analysis. Mathematical Structures - Computational Mathematics- Mathe-
matical Modelling, v. 2 (1984) 273–278.
[73] J. Ryan. Conformal Clifford Manifolds arising in Clifford Analysis. Pro-
ceedings of the Royal Irish Academy, v. 85A, No. 1 (1985), 1–23.
224 Bibliography

[74] J. Ryan. Duality in Complex Clifford Analysis. Journal of Functional Anal-

ysis, v. 61, No. 2 (1985), 117–135.
[75] J. Ryan. Cells of harmonicity and generalized Cauchy integral formulae.
Proc. Lond. Math. Soc. III Ser. v. 60, No. 2 (1990), 295–318.
[76] J. Ryan. Complex Clifford analysis and domains of holomorphy. J. Aust.
Math. Soc. Ser. A, v. 48, No. 3 (1990), 413–433.
[77] J. Ryan. Plemelj formula and Transformations Associated to Plane Wave
Decompositions in Complex Clifford Analysis. Proc. London Math. Soc. V.
64, No. 3 (1992), 70–94.
[78] J. Ryan. Intertwining operators for iterated Dirac operators over Minkowski-
type spaces. J. Math. Anal. Appl. v. 177 No. 1 (1993), 1–23.
[79] J. Ryan. Intrinsic Dirac Operators in Cn . Advances in Mathematics, v. 118
(1996), 99–133.
[80] J. Ryan. Basic Clifford Analysis. Cubo Matemática Educacional, v. 2
(2000), 226–256.
[81] J. Ryan. C2 Extensions of Analytic Functions Defined in the Complex Plane.
Adv. Appl. Clifford Algebras, v. 11 S1 (2001), 137–145.
[82] C. Segre. Le rappresentazioni reali delle forme complesse e gli enti iperal-
gebrici. Math. Ann. v. 40 (1892), 413–467.
[83] G. Scorza Dragoni. Sulle funzioni olomorfe di una variabile bicomplessa.
Reale Accad. d’Italia, Mem. Classe Sci. Nat. Fis. Mat. v. 5 (1934), 597–665.
[84] K.D. Shoinbekov, B.I. Maukeev. Sequences of bicomplex numbers and
the notion of a function of a bicomplex variable. Investigations in the theory
of functions and differential equations, v. 202, Kazakh. Gos. Univ., Alma
Ata (1985), 173–180.
[85] K.D. Shoinbekov, B.I. Maukeev. Algebra of bicomplex numbers. Investi-
gations in the theory of functions and differential equations, v. 203, Kazakh.
Gos. Univ., Alma Ata (1985), 180–190.
[86] G.L. Shpilker. Some differential properties of a commutative hypercomplex
potential. (Russian). Dokl. Akad. Nauk SSSR, v. 293, No. 3 (1987), 578–583.
[87] G. Sobczyk. The Hyperbolic Number Plane. The College Mathematics Jour-
nal, v. 26, No. 4 (1995), 268–280.
[88] N. Spampinato. Estensione nel campo bicomplesso di due teoremi, del Levi-
Civita e del Severi, per le funzioni olomorfe di due variabili bicomplesse I,
II. Reale Accad. Naz. Lincei, v. 22 No. 38–43 (1935), 96–102.
[89] N. Spampinato. Sulla rappresentazione di funzioni di variabile bicomplessa
totalmente derivabili. Ann. Mat. Pura Appli. v. 14 (1936), 305–325.
Bibliography 225

[90] R.K. Srivastava. Certain points in the theory of bicomplex numbers. Math.
Student, v. 70, No. 1–4 (2001), 153–160.
[91] R.K. Srivastava. Bicomplex numbers: analysis and applications. Math.
Student, v. 72, No. 1–4 (2003), 69–87.
[92] R.K. Srivastava. Certain topological aspects of bicomplex space. Bull. Pure
Appl. Math.v. 2, No. 2 (2008), 222–234.
[93] R.K Srivastava, S. Singh. Certain bicomplex dictionary order topologies.
Int. J. Math. Sci. Eng. Appl. v. 4, No. 3 (2010), 245–258.
[94] N.T. Stelmasuk. On certain linear partial differential equations in dual
and bicomplex algebras. An. Sti. Univ. ”Al. I. Cuza” Iasi Sect. I (N.S.), v. 9
(1963), 63–72.
[95] N.T. Stelmasuk. Some linear partial differential equations in dual and
bicomplex algebras. Izv. Vyss. Ucebn. Zaved. Matematika, v. 40, No. 2 (1964),
136–142.
[96] D.C. Struppa, A. Vajiac, M.B. Vajiac. Remarks on Holomorphicity
in Three Settings: Complex, Quaternionic, and Bicomplex. Hypercomplex
Analysis and Applications, Trends in Mathematics, Springer, I. Sabadini, F.
Sommen editors (2011), 261–274.
[97] A. Sudbery Quaternionic Analysis. Mathematical Proceedings of the Cam-
bridge Philosophical Society v. 85 (1979), 199?225.
[98] V.A. Tret’jakov. Some properties of mappings in the space accompaning
the algebra of bicomplex numbers in modulus (Russian). Application of func-
tional analysis in the approximation theory (Russian), Kalinin. Gos. Univ.,
Kalinin (1979), 129–136.
[99] V.A. Tret’jakov. On the properties of some elementary functions that
are defined on the algebra of bicomplex numbers (Russian). Mathematical
analysis and the theory of functions (Russian), Moskov. Oblast. Ped. Inst.,
Moscow (1980), 99–106.
[100] I.M. Yaglom. Complex Numbers in Geometry. Academic Press, New York-
London (1968), 243 pp.
[101] K. Yoneda. An integration theory in the general bicomplex function theory.
Yokohama Math. J. v. 1 (1953), 225–262.
[102] A. Zireh. A generalized Mandelbrot set of polynomials of type Ed for bi-
complex numbers. Georgian Math. J. v. 15, No. 1 (2008), 189–194.
[103] V.A. Zorich. Mathematical Analysis, Volumes I and II. Springer–Verlag
Berlin Heidelberg (2004).
Index

4 × 4 real matrices and BC-linearity, Properties, 60

146 BCk -trigonometric representation
BC of invertible bicomplex numbers,
as a C( j)-linear space, 30 64
as a C(i)-linear space, 30 C( j)-complex Cauchy–Riemann sys-
as a D-module, 33 tem, 149
as an R-linear space, 30 C( j)-complex differentiable bicomplex
BC-differentiable functions, 159 functions, 155
BC-holomorphic functions C( j)-complex partial derivatives of a
Characterization for being a con- bicomplex function
stant function, 181 Its relation with bicomplex deriv-
Characterization with respect to ability, 149, 150
idempotent components, 164 C(i)-complex Cauchy–Riemann sys-
Definition, 152 tem, 147
Inside C 1 -functions, 152 C(i)-complex differentiable bicomplex
relation with D-conformality, 189 functions, 153
Relation with holomorphic func- C(i)-complex partial derivatives of a
tions of two complex vari- bicomplex function
ables, 160, 161 Its relation with bicomplex deriv-
Relation with holomorphic func- ability, 147, 150
tions of two hyperbolic vari-
D-bounded subsets in D, 44
ables, 161
D-conformal mapping, 188
relation with hyperbolic angle pre-
D-conformally equivalent, 191
serving, 189
D-convergence, 109
relation with the hyperbolic mod-
ulus of the derivative, 190 D-infimum, 44
Relations with complex and hy- D-supremum, 44
perbolic holomorphies, 185 ∗-anti-holomorphy
BC-valued functions and complex har- Definition, 187
monic functions, 196 Properties, 188
BCi -analogue of De Moivre formula, ∗-conjugation, 8
61 †-anti-holomorphy
BCi -trigonometric form of invertible Definition, 187
bicomplex numbers Properties, 187
Definition, 59 †-conjugation, 8

© Springer International Publishing Switzerland 2015 226

M.E. Luna-Elizarrarás et al., Bicomplex Holomorphic Functions, F rontiers in Mathematics,
DOI 10.1007/978-3-319-24868-4
Index 227

bar-anti-holomorphy BC-differentiable, 159

Definition, 187 C(i)-complex differentiable, 153
Properties, 188 D-conformal, 188
Bar-conjugation, 8 Bicomplex derivative
Bicomplex anti-holomorphies, 187 At a point, 137
Bicomplex arccos function, 131 Bicomplex-differentiable, 159
Bicomplex ball of hyperbolic radius, Complex partial derivatives, 136
101 Definition, 110
Bicomplex Borel–Pompeiu formula, 215 Different representations, 135
Bicomplex Cauchy integral represen- Hyperbolic angle preserving, 188
tation, 216 Hyperbolic partial derivatives, 136
Bicomplex Cauchy integral theorem, Real partial derivatives, 136
214 Bicomplex holomorphic functions
Bicomplex cosine function Characterization for being a con-
Definition, 124 stant function, 181
Idempotent representation, 124
Characterization with respect to
Properties, 125
idempotent components, 164
Bicomplex cotangent function
Definition, 152
Definition, 126
Inside C 1 -functions, 152
Idempotent representation, 126
relation with D-conformality, 189
Bicomplex derivability
Relation with holomorphic func-
Chain rule, 140
tions of two complex vari-
Definition, 137
ables, 160, 161
Derivative of the inverse func-
Relation with holomorphic func-
tion, 140
tions of two hyperbolic vari-
Elementary functions, 142
ables, 161
Relation with C( j)-complex par-
tial derivatives, 149 relation with hyperbolic angle pre-
Relation with hyperbolic partial serving, 189
derivatives, 151 relation with the hyperbolic mod-
Relation with real partial deriva- ulus of the derivative, 190
tives, 144 Relations with complex and hy-
Relations with arithmetic oper- perbolic holomorphies, 185
ations, 138 Bicomplex hyperbolic cosine function
Bicomplex derivative Definition, 127
Relation with C(i)-complex par- Idempotent representation, 127
tial derivatives, 147 Properties, 127
Bicomplex derivative of a bicomplex Bicomplex hyperbolic sine function
function Definition, 127
At a point, 137 Idempotent representation, 127
Bicomplex exponential function Properties, 127
Definition, 120 Bicomplex increment
Properties, 121 Definition, 135
Bicomplex functions Different representations, 136
228 Index

Bicomplex logarithm of a bicomplex Definition, 126

number Idempotent representation, 126
(m, n)-branch, 130 Bicomplex Taylor series, 208
Definition, 130 Bicomplex-differentiable functions, 159
Principal value, 130 Bilinear forms in BC
Properties, 130 BC-valued, 39
Bicomplex numbers C(i)-valued, 38
addition of , 5 hyperbolic-valued, 39
definition, 5 real-valued, 37
Different representations, 7 Boundary of a hyperbolic curve, 96
Euclidean norm, 11
Invertibility, 12 Cauchy–Riemann operator of Clifford
Matrix representation analysis, 193
with entries in C( j), 36 Cauchy–Riemann type conditions, 146
with entries in C(i), 35 Closed hyperbolic curves, 96
with entries in D, 36 Complex C( j) Laplacian
with entries in R, 36 Factorization, 196
multiplication of , 5 Complex C(i) Laplacian
Bicomplex polynomials Definition, 194
Analogue of Fundamental Theo- Factorization, 195
rem of Algebra, 116 Complex C( j) Laplacian
Associated roots, 117 Definition, 194
Definition, 114 Complex analogues of the wave oper-
Set of roots, 114 ator
Some properties, 116 Definition, 197
Bicomplex power series Factorization, 197
Abel’s Theorem, 205 Complex argument of bicomplex num-
Definition, 205 bers
Hyperbolic radius of convergence, Definition, 59
208 principal value, 59
Sets of convergence, 206 Complex curves
Bicomplex radicals Among two-dimensional surfaces,
Definition, 128 99
Bicomplex rational functions, 118 Definition, 98
Bicomplex sequences, 107 Complex exponential function, 119
Bicomplex sine function Complex partial derivatives of a bi-
Definition, 124 complex function, 136
Idempotent representation, 124 Complex polynomials
Properties, 125 Definition, 113
Bicomplex sphere of hyperbolic ra- Fundamental Theorem of Alge-
dius, 69, 101 bra, 114
Bicomplex spheres as multiplicative Complex ray in a given direction, 86
groups, 102 Complex straight lines
Bicomplex tangent function in BC
Index 229

Characterization among real two- exponential representation in hyper-

dimensional planes, 87 bolic terms
Definition, 78 of an invertible bicomplex num-
Parametric representation, 78 ber, 132
Properties, 81 of zero-divisors, 132
Relation with complex argu-
ments, 86 Geometric interpretation of the bi-
Slope, 83 complex derivative, 189, 190
in C2 (i), 77
Conjugate Hyperbolic imaginary unit , 6
C( j)-complex harmonic functions, Hyperbolic angle between hyperbolic
198 curves, 97
C(i)-complex harmonic functions, Hyperbolic angle of bicomplex num-
198 bers
D-harmonic functions, 199 Definition, 63
Conjugations in BC Hyperbolic angle preserving mapping,
Idempotent representation, 18 188
properties, 9 Hyperbolic argument of bicomplex num-
bers
Definition, 63
Derivability of bicomplex functions Hyperbolic curves
Chain rule, 140 Closed, 96
Elementary functions, 142 Definition, 95
Jacobi matrix, 145 Piece-wise smooth, 96
Relation with C( j)-complex par- Without topological boundary, 96
tial derivatives, 149 Hyperbolic differentiable bicomplex func-
Relation with C(i)-complex par- tions, 156
tial derivatives, 147 Hyperbolic Laplacian
Relation with hyperbolic-complex Definition, 194
partial derivatives, 151 Factorization, 196
Relation with real partial deriva- Hyperbolic lines in BC
tives, 144 Characterization
Derivative of a BC-holomorphic func- in accordance with its projec-
tions in terms of the idem- tions, 90
potent components, 165 Definition, 88
Distinguished boundary, 179 Hyperbolic angle between, 94
in bicomplex language, 89
Euclidean norm More properties, 92
Its relation with the product in Parametric representation, 91
BC, 25 Slope, 93
Multiplicative properties, 27, 28 Hyperbolic numbers
Euclidean topology in BC, 110 Definition, 6
Euler formula, 119 Idempotent representation, 23
Euler’s number, 119 Inside bicomplex numbers, 20
230 Index

Hyperbolic partial derivatives of a bi- Positive hyperbolic numbers

complex function, 136 cartesian representation, 7
Hyperbolic partial derivatives of a bi- Product-type domains, 179
complex function
Its relation with bicomplex deriv- Quadratic forms in BC
ability, 151 C( j)-valued, 38
Hyperbolic version of Cauchy–Riemann C(i)-valued, 38
type system, 151 hyperbolic-valued, 39
Hyperbolic-valued norm in BC, 47 real-valued, 38

Real differentiability
Idempotent
In C( j)-complex terms, 154
Elements, 15
In C(i)-complex terms, 153
Representation, 15
In bicomplex terms, 156
Idempotent real variables, 162
In hyperbolic terms, 155
Invertible bicomplex numbers
Real exponential function, 119
Characterization, 14
Real idempotent components and real
Idempotent representation, 19
cartesian components
Jacobi matrix Matrix of change of variable, 167
Relation with bicomplex deriv- Relation between them, 167
ability, 145, 146 Real Laplace operator
Definition, 193
Moduli of bicomplex numbers Factorization, 193
Definitions, 9 in n variables, 193
Idempotent representation, 18 Real lines in BC, 77
Properties, 11 Real partial derivatives of a bicom-
plex function
Negative hyperbolic numbers, 7 Definition, 136
Non-negative hyperbolic numbers Its relation with bicomplex deriv-
cartesian representation, 7 ability, 144
Idempotent representation, 23 Real polynomials
Non-positive hyperbolic numbers, 7 Definition, 113
Idempotent representation, 23 Properties, 114
Real straight lines
Operator of the change of variable Complex form, 73
and BC-modules, 168 in R2 , 73
and gradient, 170 Parametric representation, 76
and operators acting on BC-modules, Riemann Mapping Theorem
168 Bicomplex case, 190
and real partial derivatives, 169 Complex case, 190
Definition, 168
Sequences of bicomplex functions
Partial order on D D-uniformly convergent
Definition, 41 Bicomplex analogue of Weier-
Properties, 42 strass’ Theorem, 203
Index 231

Definition, 203 The bicomplex differential operators

Properties, 203 ∂ ∂ ∂ ∂
, †
, , and
Convergence, 202 ∂Z ∂Z ∂Z ∂Z ∗
Definition, 202 BC-holomorphy, 158
Series of bicomplex functions The four-dimensional cube, 52
Trigonometric representation in hy-
D-absolutely convergent
perbolic terms
Definition, 204
algebraic properties, 65
Properties, 204
De Moivre formula, 68
Weierstrass test, 204
geometric properties, 68
Definition, 204
of invertible bicomplex numbers,
Series of bicomplex numbers
64
D-absolutely convergent, 202 of zero-divisors, 64
Cauchy criteria, 201
Definition, 201 Weak Stoltz condition for bicomplex
In idempotent form, 201 functions, 138
Sesquilinear forms in BC
BC-valued Zero-divisors, 12
∗-sesquilinear, 39 Characterization, 14
†-sesquilinear, 39 Idempotent representation, 19
bar-sesquilinear, 39 Zeros of bicomplex holomorphic func-
C(i)-valued, 38 tions, 179, 180
hyperbolic-valued, 39
Shilov boundary, 179
slope of complex lines, 83
Stokes formula and bicomplex Cauchy–
Riemann operators, 211
Strong Stoltz condition for bicomplex
functions, 138

Tangent to a smooth hyperbolic curve,

97
The n-th root of a bicomplex num-
ber, 128
The bicomplex differential operator
∂
, 157
∂Z
∂
, 156
∂Z
∂
, 156, 158
∂Z ∗
∂
, 156, 157
∂Z †
∂
, 156, 157
∂Z