See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/349947136

Complex Numbers: Real Applications of an Imaginary Concept (CDT-56)

Preprint · February 2021

DOI: 10.13140/RG.2.2.12943.51362/3

CITATIONS READS
0 2,629

1 author:

Luciano da F. Costa
University of São Paulo
822 PUBLICATIONS   15,541 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Complex Systems View project

Pattern Recognition: Concepts and Methods View project

All content following this page was uploaded by Luciano da F. Costa on 19 March 2021.

The user has requested enhancement of the downloaded file.

 Complex Numbers:
Real Applications of an Imaginary Concept
(CDT-56)
Luciano da Fontoura Costa
[email protected]

São Carlos Institute of Physics – DFCM/USP

20th Feb. 2021

Abstract
Motivated by the need to find a solution to the square root of -1, complex numbers research steadily became one
of the most important areas in mathematics, with ample application in a large number of scientific and technological
subjects. Though related to the two-dimensional real vector space, complex numbers are substantially more general as
they provide the means for defining a large number of additional operations and functions, including multiplication,
exponentiation, logarithm and power. The present work is aimed at providing a general introduction to some of the
main basic aspects of complex numbers theory, concentrating on aspects related to signal analysis, electric engineering,
shape analysis, complex systems and fractals, as well as number theory.

“The next dimension is everywhere to be found...

Then, you’ll find another dimension.”

LdFC.

1 Introduction
The dimensionality of a mathematical space constitutes
an extremely important property from both the theoreti-
cal and applied perspectives, as it directly impacts on vir-
tually every property of the entities to be found, defined,
and modeled in these spaces, including their complexity.
Even if taken as a single feature, the dimensionality of Figure 1: The Riemann zeta function, which is a complex function of
a problem will already provide critical information about a complex variable, is directly related to the Riemann’s hypothesis,
which is considered by many as the greatest standing mathematical
the possible structural and dynamical intricacies.
problem. Shown here is the argument (or phase) of the Riemann
For instance, in the real line R, we may have sets of zeta function. Interestingly, the most complex structure of this func-
connected and/or disconnected sets of points. As such, tion is geometrically more pronounced for Re(z) < 0, but it is along
this vector space allows only very simple modeling of the its critical line 0.5 + iy that its non-trivial roots are believed to be
found (Riemann’s hypothesis).
real world, catering only to situations involving a single
variable. In order to provide subsidies for more general
and powerful models, the real line was extended into the
Cartesian plane, and then into spaces of larger dimensions terms of respective conditions and properties, these repre-
RN , with N being a positive integer. sentations are somewhat limited in the sense that they do
While such mathematical spaces can be effectively mod- not allow, among many other possibilities, the definition
eled in terms of the concept of vector space, namely of operations of product, inverse, or division of respective
a mathematical framework formalizing these spaces in vectors.

1
 The solution for the extension of the real line to spaces exponential in the Fourier transform, as well as the con-
more structured than the RN would come from a some- cept of instant frequency), electrical engineering (phase,
what unexpected issued, namely the calculation of the basic complex operations and powers), shape analysis
square root of negative numbers. Having intrigued math- (conformal mappings as means for transforming shapes),
ematicians for a long time, the solution of this problem complex systems and fractals (the Mandelbrot set), and
was largely credited to the Italian mathematician Rafael number theory (the Riemann Zeta function). These areas
Bombelli (1526–1572). His solution was as simple and can be understood as providing some important examples
creative as it could have been, proposing that the square of complex numbers applications.
root of −1 would correspond to a new number, often rep-
resented as i, called the imaginary number, giving rise to
the quintessential relationship expressed as: 2 The Square Root as an Analytic
√ Tool
i = −1
which also implies that: Any real value a > 0 has two square roots, namely b =
√ √ 2
√
−i = −1 ± a so that a = b . Observe that b = 0. Thus, strictly
(1) speaking the square root is not a function, as it takes
a number from its domain to more than one images, as
Bombelli’s insight, complemented by many other math- illustrated in Figure 2.
ematicians, paved the way to a whole new areas in math-
ematics, including complex numbers (e.g. [1, 2]) and com-
plex analysis. Remarkably, though largely based on imag-
inary concepts, these areas have found myriad applica-
tions in both theoretical and applied situations. Indeed,
complex numbers are now extensively used in theoreti-
cal, statistical and quantum physics; electric, electronic,
mechanic and civil engineering, signal processing, control
theory, complex systems, to name but a few cases, not
mentioning many mathematical areas. Though not so of-
ten realized or acknowledges, such impressive success and
popularity of complex numbers ultimately derives from Figure 2: The square root of real values taking real values is not
√
the extended mathematical structure provided by the in- defined for a < 0, and for a ≥ 0 it yields two values, ± a. For
troduction of the imaginary number, allowing the defi- these two reasons, this definition of the square root is not a function,
though it may be understood to be invertible in the sense that every
nition of multiplication and powers of vectors (as repre- image element can be associated to a single value in the domain of
sented by complex numbers). this mapping.
The advantages of using imaginary numbers was also
extended to cover larger dimensional spaces, such as in
the theory of quaternions, on which a CDT is planned to The above issue with the square root not being a func-
√
be prepared. tion can be settled by assuming only positive (+ a) or
√
Despite their name, complex numbers are actually sim- negative (− a) square root results, both of which are now
ple to be understood and used. When compared to real functions.
values, complex operations are somewhat more elaborate, Given a real number a > 0, its positive square root
√
but this is more a consequence of the fact that complex value b = + a provides a particularly interesting geo-
numbers are an expansion of a two-dimensional space metric interpretation as the length of the sides of a square
(R2 ), which is intrinsically richer than the real line R. that will have area a, i.e. a = b2 .
The present work aims at providing an intuitive and In a slightly different perspective, we can understand
hopefully accessible introduction to some of the more im- the square root as an analytical operation allowing us to
portant concepts and results in complex numbers and decompose a real value a as a product of two identical
√
complex analysis. However, given our limited space, factors b = ± a, an so on. An interesting aspect of this
the presentation is by no means exhaustive and should interpretation is that it is directly related to one of the
be eventually complemented by the related literature fundamental motivations of number theory, namely the
(e.g. [1, 2]). Though this work may be considered as a decomposition of integer numbers in terms of products of
relatively generic introduction to complex numbers, it fo- constituent parts, with emphasis on prime number fac-
cuses on aspects related to signal analysis (the complex torization. This idea can also be understood, at a more

2
general perspective, as an analytic modeling approach in
which one is interested in expressing objects in terms of
combinations of more basic parts.
So, we can say that the number 9 is composed of two
identical factors 3 (or −3). Observe that this is a very
particular example in which the square roots decompo-
sition corresponds precisely to the prime factorization of
9. Though most other real values will lead to distinct
respective decompositions, we can still imagine both the
prime factorization and the square root as an analytic
approach decomposing a value into more basic respective Figure 3: The complex conjugation of a complex number z can be
components. understood as corresponding to the mirroring of z with respect to
the x-axis.

3 The Imaginary Number

and that:
The square root is only defined for real values a ≥ 0,
not existing for a < 0. A solution to this problem was (x + i y)(x + i y)∗ = (x2 − y 2 ) ∈ R (5)
advanced by Rafael Bombelli, who imagined a new value
i as the square root of −1. As a consequence, −i also for any x, y ∈ R.
√
corresponds to −1.
This seminal idea paved the way to a whole new vector 5 The Complex Exponential
space, namely that of complex values, henceforth repre-
sented as C. The complex exponential turns out to be a central concept
Thus, it became possible to define the positive (or neg- in complex numbers. It has as power series:
ative) square root as taking complex values and yielding ∞
complex values, namely:
X sk s s2 s3
z = es = =1+ + + + ... (6)
k! 1! 2! 3!
√ k=1
f (s) = + s | s ∈ C → f (s) ∈ C (2)
converging for any complex value s, which means that
Before we discuss the square root of complex values, it every complex number can be expressed in the polar form
is interesting to get acquainted with some properties of ρ eiφ , with ρ, φ ∈ R.
real values s = x + i y, which will be our object in the The complex exponential of a purely imaginary value
following sections. i φ is a complex number z that can be expressed in terms
of the Euler’s formula as:

4 Complex Numbers ei φ = z = x + i y = cos(φ) + i sin(φ)

p
x2 + y 2 = 1
Every generic complex number s = x + i y has a real part y
x = Re {s} and an imaginary part y = Im {s}, i.e.: φ = arctan
x
s = x + iy More generally, for a generic complex value s, we have
with that:
x = Re {s}
z = x + i y = es = ρ eiφ = eln(ρ) ei φ = eln(ρ)+i φ (7)
y = Im {s} p
ρ = x2 + y 2 (8)
The complex conjugate of s = x + i y is defined as: y
φ = arctan (9)
x
∗
s = s = x − iy (3)
The centrally important Euler’s formula can be directly
Observe that the complex conjugate of a complex value appreciated from graphical representations such as that in
s can be geometrically understood as mirroring s as if the Figure 4, which shows the possible values of a complex ex-
x-axis were a mirror, as illustrated in Figure 3. ponential ρ ei φ represented as a circle of radius ρ centered
We also have that: at the origin of the Argand (or complex) plane.
The polar representation of z involves two real parame-
(s∗ )∗ = s = s (4) ters: ρ and φ. It is necessary to define the ranges of these

3
 complex exponential, therefore inheriting analogous prob-
lems of function specification and non-invertibility.

6 Basic Complex Operations

Given two complex values s and v, their complex addition
is expressed as:

s + v = (xs + i ys ) + (xv + i yv ) =
(xs + xv ) + i(ys + yv ) (12)
Figure 4: Graphical representation of Euler’s formula, relating the
Cartesian (z = x + i y) and polar (z = ρ ei φ ) representations of a
The complex product of those two values is given as:
generic complex number z.

(s)(v) = (xs + i ys ) (xv + i yv ) =

= (xs xv − ys yv ) + i (xs yv + xv ys ) =
two values. While it follows immediately from Equation 8
= (ρs ei φs )(ρv ei φv ) = ρs ρv ei (φs +φv ) (13)
that ρ ∈ [0, ∞], there are several manners in which φ can
be bound. The key issue here is that eiφ is a periodic Observe that the product of a complex value s = x + i y
function with period 2π, i.e.: by its conjugate s∗ is necessarily a purely real number
corresponding to:
z = x + i y = ρ eiφ = ρ eiφ+n 2π (10)
s s∗ = x2 − y 2 (14)
where n is an integer value. This property stems di-
rectly from the periodicity of the sine and cosine functions The reciprocal of s can be expressed as:
that compose the complex exponential, i.e.:
1 1 1
= = =
z = ρe iφ+n 2π
= ρ [cos(iφ + n 2π) + i sin(iφ + n 2π)] (11) s xs + i ys (xs + i ys )(xs − i ys )
xs − i ys 1 1
= 2 2
= i φ
= ei (−φs ) (15)
This property can be readily appreciated in Figure 4, xs − ys ρs e s ρs
where each increment by 1 on n implies a full counter-
The product of the denominator by its complex con-
clockwise rotation around the circle underlying the com-
jugate s = x + i y. as means of obtaining a purely real
plex exponential.
denominator.
Though, in principle, we could have φ ∈ [0, 2π) or φ ∈
The complex division of s by v corresponds to:
(−π, π], covering exactly one full period, for generality’s
sake it is often adopted that φ ∈ R. One reason for that is s xs + i ys xs + i ys
= = =
that the latter choice allows the phase, i.e. the argument v xv + i yv (xv + i yv )(xv − i yv )
φ of the complex exponential to be accumulated to values (xs + iys )(xv − i yv )
= =
larger than a period (see, for instance, [3]). However, x2v − yv2
this also implies the complex exponential not to be well ρs ei φs ρs i (φs −φv )
defined. = i φ
= e (16)
ρv e v ρv
In particular, because n is an integer value, the complex
number z is mapped by the complex exponential into an 7 Complex Logarithm
infinite number of values, therefore implying that not only
that mapping not to be a function, but also not being Let z ∈ C, z = x + i y = ρeiφ . The complex logarithm can
invertible. In these circumstances, additional restrictions be derived from the complex exponential as:
have to be imposed so that one of these values can be  
taken as the result of the complex exponential, which then f (z) = ln(z) = ln(ρ eiφ ) = ln eln(ρ) ei φ = ln(ρ) + i φ
becomes an invertible function. The so-called principal (17)
value of φ satisfies φ ∈ (−π, π]. As observed in Section 5, the complex logarithm inher-
The multiple mapping implied by the complex exponen- its the periodicity of the complex exponential, meaning
tial without some restriction that makes it an invertible that this mapping cannot be formally understood to cor-
function has important consequences in complex number respond to a function, as it maps a complex value into
theory, because several other functions, such as the com- several other complex values. The above expression con-
plex power and logarithm, are expressed in terms of the siders the fundamental value, obtained for n = 0.

4
 As an example, let’s calculate the logarithm of s = Now, let’s consider the situation in which p is a real
1 + 1 i. We have: value. Similar formulae hold as before and, in the case of
p √ Equation 22 we have that np2π will no longer be an in-
ρ = 12 + 12 = 2 teger, implying that the result ρp ei(pφ+2npπ) is no longer
 
1 π periodical. Though they will all be valid results, as per
φ = arctan =
1 4 our previous discussion, several results will be missed.
hπ i
ln(1 + 1 i) = ln(ρ) + i[φ + n2π] = 0.5 ln 2 + i + n2π Therefore, De Moivre’s relationship cannot be used for
4 calculating real powers of complex values other than that
with n integer. corresponding to the principal value.
Recall that the change of logarithmic basis can be ob- In the case of a complex exponent z, it is interesting to
tained as: consider, for a generic complex value s, that:
logb (s)
loga (s) = (18)
logb (a) s = eln(s) (23)

Thus, we have:
8 Powers of Complex and Com-
 z
plex Powers sz = eln(s) = ez ln(s)
(24)

First, it is important to remember that, given three values Observe that only the principal value is thus obtained.
a, b, c, ∈ R, in general we have that: Let’s consider some numeric examples. First, we calcu-
c late (2 + i)2 by using De Moivre’s Theorem:
(ab )c 6= a(b )
(19) p √
ρ = 22 + 12 = 5
so that it is important to define the order of the powers  
1
c
in case the expression ab is used (we suggest always to φ = arctan = 0.4636476...
2
use the brackets).
(2 + i)2 = ρ2 exp(i 2φ) = 5 ei 0.92729... = 3 + 4 i
Recall also that:
Now, we take the complex power of a complex value,
(ab )c = ab c (20)
(1 + i)2+i . Let s = 1 + i and z = s + i. First, let’s obtain
Now, let s = x + i y be a generic complex number, and the polar form of s:
p be a generic integer value. Then, we can write: p √
ρ = 12 + 12 = 2

p  
1 π
sp = (x + i y)p = ρ eiφ = ρp ep iφ = ρp ei(pφ+n2π) (21) φ = arctan =
1 4
with n being an integer value. (25)
This result, known as the De Moivre’s Theorem, pro-
vides a means for taking integer powers of complex values. Now, we apply Equation 24:
However, in the above result we considered only one
(1 + i)2+i = e(2+i) ln(1+i) = e(2+i)[ln ρ+iφ] =
of the infinite polar representations of s. Now, let’s
take all them into account (see Section 5), with n = = e2 ln ρ−φ+i[ln ρ+2φ] = −0.3097435... + 0.857658i... (26)
. . . , −2, −1, 0, 1, 2, . . .:

sp = (x + i y)p = (ρ ei(φ+n2π) )p = ρp ei(pφ+2npπ) (22) 9 Complex Roots

Now, if we compare the results in Equations 29 and 22, We are now in a better position to approach the complex
we can see that they are not identical. Though all results root of a generic complex value s, where r is also a purely
in Equation 22 will correspond to values in Equation 29, real number:
the latter misses several of the possible values in the for- √ 1 φ
s = s r = ρsr ei r (p φs ) = ρ r ei r (φ+2kπ) = ρ r ei ( r + )
1 1 1 1 1 2kπ
r n
mer (a consequence of the np sequence, which implies
(27)
gaps respectively to the sequence n). This example illus-
As√an example, let’s calculate the 7 complex roots of 1,
trates the problems of dealing with complex powers (and 1
i.e. 7 1 = 1 7 . First, we calculate ρ = 1 and φ = 0. Then,
logarithms), indicating that special care is needed because
1
(1) 7 = ρs7 ei 7 (p φs ) = 1 7 ei 7 (2kπ) = ei ( )
2kπ
some relationships between powers and logarithms fail for 1 1 1 1
7 (28)
complex numbers. To any extent, the two equations above
will agree as far as the principal value is concerned. Figure 5 depicts the so-obtained 7 complex roots of 1.

5
 Figure 6: The complex exponential as a parametric function of t is
used as the basis function for the Fourier series and transform. The
figure depicts the complex exponential ei 2πf t for f = 1, 2, 4, and 8.

Figure 5: The 7 complex roots of the real number s = 1 define

a respective 7−sides polygon in the Argand (or complex) plane.
Interestingly, though there are two square roots of real values larger for f1 6= f2 .
or equal to 0, this number can be substantially enhanced in for
complex numbers, reflecting the richer structure of the latter space.
In addition, observe that the parametric complex func-
tion sampled with N values, as used in the discrete Fourier
transform, are given by the N -square roots of 1 (e.g. [4]).
Given a complex parametric function g(t) represented
Observe that, in addition to the inherent r-multiplicity
in polar form g(t) = ρ(t) exp (i φ(t)), it is possible to de-
of the r-complex root, each of these roots will also have
fine its instant amplitude and instant frequency respec-
infinite multiples implied by the periodicity of the com-
tively as (e.g. [5]):
plex exponential function. A Therefore, restrictions need
to be imposed on both on the r and the n-multiplicity so A(t) = ρ(t) (32)
that the complex root mapping becomes a function. dφ(t)
√ f (t) = 2π (33)
The more general case v s, where z is a complex num- dt
ber, can immediately be addressed by making z = 1/v in
Equation 29: 11 Generic Complex Functions
√ 1
  v1 1
v
s = s v = eln(s) = ev ln(s)
(29) Complex functions are functions which have complex val-
ues as argument and result, so that we can write:

10 Complex Parametric Func- z = g(s) (34)

gr (s) = Re {z} (35)
tions
gi (s) = Im {z} (36)
A complex parametric function g(s) is such that its argu- where s, z ∈ C.
ment s is a function of a parameter t, i.e. g(s) = g(s(t)) = Therefore, the graphical representation of complex
g(t), with t ∈ [a, b], a, b ∈ R. functions involves two surfaces defined on the Argand
As an example, consider the function: plane. Let’s consider the example:
g(t) = ρ exp(i n2πt) = ρ [cos(n2πt) + i sin(n2πt)] (30) z = g(s) = s2 = (x + i y)2 = x2 − y 2 + i 2xy =
= Re {z} + i Im {z} = u(s) + i v(s) (37)
with t ∈ [0, 2π) and n = ±1, ±2, . . ..
2 2
This function defines m full circles with radius ρ in the u(s) = u(x, y) = Re {z} = x − y (38)
complex plane, being closely associated with the Fourier v(s) = v(x, y) = Im {z} = 2xy (39)
series and Fourier transform base functions. Figure 6
shows the visualization of the complex parametric func-
tion exp(i2πf t) for f = 1, 2, 4, and 8. 12 Analytic Functions and Con-
A particularly important property of the parametric
complex exponential is that: formal Mappings
ˆ ∞
A complex function is analytic at z0 if its Taylor series
exp(i 2πf1 t) exp(−i 2πf2 t)dt = 0 (31) converges within a sufficiently small neighborhood around
−∞

6
 (a) (b)
Figure 7: The real (a) and imaginary (b) parts of the function g(z) = z 2 , corresponding respectively to u(x, y) = x2 − y 2 and v(x, y) = 2xy.
Each of these parts can be understood as a multivariate function of the variables x and y.

z0 . Analyticity can also be inferred from the derivatives As an example, consider the function z = s2 = x2 −
2
of complex functions. y + i 2xy. We have that:
Let g(z) = u(x, y) + i v(x, y), with z = x + i y, be
a complex function. In case they exist, the Wirtinger
∂u(x, y) ∂v(x, y)
derivatives (e.g. [6]) of g() are expressed as: = 2x; = 2x
∂x ∂y
∂g(z)

1 ∂g(z) ∂g(z)
 ∂u(x, y) ∂v(x, y)
= −i = = −2y; = 2y
∂z 2 ∂x ∂y ∂y ∂x
 
1 ∂u(x, y) ∂v(x, y) ∂u(x, y) ∂v(x, y)
= +i −i + =
2 ∂x ∂x ∂y ∂y and we can conclude that z = s2 is an analytical func-
  
1 ∂u(x, y) ∂v(x, y) ∂v(x, y) ∂u(x, y) tion.
= + +i − (40)
2 ∂x ∂y ∂x ∂y The complex conjugate g(z = x + i y) = u(x, y) +
 
∂g(z) 1 ∂g(z) ∂g(z) i v(x, y) = x − i y, with u(x, y) = x and v(x, y) = −y
= +i =
∂z 2 ∂x ∂y is not an analytic function:
 
1 ∂u(x, y) ∂v(x, y) ∂u(x, y) ∂v(x, y)
= +i +i − =
2 ∂x ∂x ∂y ∂y ∂u(x, y) ∂v(x, y)

1 ∂u(x, y) ∂v(x, y)

∂u(x, y) ∂v(x, y)
 = 1; = −1
= − +i + (41) ∂x ∂y
2 ∂x ∂y ∂y ∂x ∂u(x, y) ∂v(x, y)
= 0; =0
∂y ∂x
The Cauchy-Riemann, a necessary and sufficient con-
dition for a complex function g(z) to be analytic, can be
expressed as: All analytic complex functions have the important
∂g(z) property of being conformal, in the sense of preserving the
=0 (42)
∂z angles between any two parametric curves, as illustrated
in Figure 8. Informally speaking, this property can also
being verified for every z in the considered domain, im-
understood as preserving shape locally around each point.
plying the existence of the Wirtinger derivative as well as
Though preservation of orientation is also often required,
it assuming null value.
it is possible to consider conformality without orientation
meaning that:
preservation, therefore emphasizing local shape preserva-
∂u(x, y) ∂v(x, y) tion.
= (43) Not being an analytic function, the complex conjugate,
∂x ∂y
∂u(x, y) ∂v(x, y) does not preserve angle orientation, but it does preserve
=− (44) angles and local shape.
∂y ∂x

7
 for (j in seq(1,N)){
x <- (i-1)*dx + x_min
y <- (j-1)*dy + y_min
c <- x + 1i*y
z <- 0; p <- 1
while ((Mod(z) < 2)&(p < 50)){
z <- z^2 + c
p <- p+1
}
if (Mod(z) < 2) img[i,j] <- 0
else img[i,j] <- p
}
}

Observe that, here, M od(z) calculates the magnitude

Figure 8: An analytical mapping is said to be conformal as it pre-
of a complex value z.
serves the angles (and possibly also angle orientation) between in-
tersections of parametric curves before (a) and after (b) transforma- The matrix img obtained in the above code can be
tion. Observe that the involved angles are given by the derivatives visualized as shown in Figure 9.
of the respective parametric curves.

13 The Mandelbrot Set

Fractals are amazing mathematical structures presenting
infinite self-affinity (e.g. [7]). Remarkably, fractals are
also related to attractors of chaotic dynamic systems.
One of the most disseminated fractals is the Mandelbrot
set (e.g. []). Interestingly, this intricate structure can be
obtained by considering the convergence of the complex
recurrence relationship:

z [n+1] = z [n] + c (45)

with n = 1, 2, . . ., z [1] = 0 (other configurations are Figure 9: The Mandelbrot set generated by the described R code.
possible), and c is each of the considered points in the
Argand plane.
Points c that lead to the convergence of the above re- As one zooms into this structure, it will prove to be
lationship after a sufficiently large number of interactions self-similar at any possible scale, which is a characteris-
(100 is often adopted when plotting the Mandelbrot set) tic of a fractal. As a suggestion, the reader is motivated
are understood to belong to the Mandelbrot convergence to adapt the above code in order to zoom into parts of
set. Otherwise, the number of interactions taken until the obtained structure, but it should be noticed that this
the magnitude of x[n+1] exceeds a given threshold (e.g. is limited by the intrinsic precision for representing real
2), is often represented as a respective intensity of the values in programming languages. Specially designed pro-
respective point when represented as an image. grams capable of enhanced zooming capabilities have been
The following R code illustrates a possible implemen- made available in the Internet.
tation allowing the obtention of an overall picture of the
Mandelbrot set. 14 The Riemann Zeta Function
N <- 500; img <- matrix(0,N,N)
Consider the real series, known as the harmonic series:
x_min <- -2.5; x_max <- 1.5
∞
y_min <- -1.5; y_max <- 1.5 1 1 1 X1
1+ + + + ... = (46)
2 3 4 k
k=1
dx <- (x_max-x_min)/(N-1)
dy <- (y_max-y_min)/(N-1) Though each new term 1/k is progressively smaller and
smaller, this series can surprisingly be shown not to con-
for (i in seq(1,N)){ verge.

8
 The Riemann zeta function, introduced in 1859, is de- be a generic vector in R2 . We immediately have that:
fined in terms of the complex variable s = x + i y as:

1 1 1 X 1 ∞ s = x+i y ⇐⇒ ~v = [x, y]T for the same values of x and y.

ζ(s) = 1 + s + s + s + . . . = (47)
2 3 4 ks
k=1
In other words, we can associate the x-axis in R2 to the
where s ∈ C.
real axis in C, while the y-axis corresponds to the imag-
Observe that the real series in Equation 46 can be un-
inary axis of C, so that each complex number becomes
derstood as a specific case of the Riemann zeta function
directly associated to the respective vector for the same
when s = 1, corresponding to a situation in which that
values of x and y, and vice-versa.
function does not converge. Indeed, it can be verified that
Motivated by this direct association, an important
the Riemann zeta function converges provided x > 1.
question arises: are R2 and C distinct representations of
This function can also be continued analytically other-
the same space, a kind of ‘isomorphism’ ? To answer this
wise, allowing its estimation in the critical strip 0 < x < 1
question, it is important to consider not only the basic el-
(e.g. [8]).
ements of those two spaces, which are in direct correspon-
The Riemann zeta function is a complex function of a
dence, but also other properties of those spaces, especially
complex variable. As such, it can be visualized in terms
in which regarding operations between the respective ele-
of its real and imaginary parts expressed in terms of a
ments.
subregion of the Argand space. Figure 10 illustrates the
Though the two spaces above share the addition oper-
real and imaginary parts of the Riemann zeta function as
ation between their elements, many other operations and
visualized in terms of two images corresponding to its real
functions can be defined in the complex space, includ-
(a) and imaginary (b) parts.
ing product, exponentiation, logarithm and power. These
important features allows a vast number of possibilities
of both theoretical and applied nature. In particular, the
possibility to define functions on the complex values paves
the way to an entire area, known as complex analysis, with
many applications in mathematics, physics and engineer-
ing.
Strictly speaking, it is also possible to consider sev-
eral types of products between the vectors in R2 , such as
the inner and external products, but these are typically
not included in the formal characterization of that vector
space, which incorporates only addition and multiplica-
(a) (b)
tion by scalar. A similar situation holds for functions,
Figure 10: The real (a) and imaginary (b) parts of the Riemann in the sense that it is also possible to define scalar and
zeta function. vector fields in R2 . However, there are several intrinsic
advantages provided by C, such as the fact that both the
aforementioned fields can be represented in unified terms
The Riemann zeta function plays an important role in
of a complex function of a complex variable. There are
theory of numbers, as it is intrinsically related to prime
many other advantages derived from the extended struc-
numbers (e.g. [9, 10]). It is also the main subject of the
ture of Cm when compared to R2 . That is the key point
Riemann’s hypothesis, seen by many as the most challeng-
in defining mathematical structures, i.e. each of these for-
ing standing mathematical problem, which states that all
malize what can be represented and operated. A space
non-trivial roots of the Riemann zeta function are to be
with more operations and properties is, in principle, paves
found along the critical line 0.5 + iy (e.g. [8]).
the way to achieving more effective and powerful resources
for respective applications.
15 R2 = C? Thus, going back to our initial question, though R2 and
C share several characteristics, including direct relation-
One interesting aspect that is common to both the Argand ships between their elements, the addition operation and
(or complex) space C and the vector space R2 is that product by scalar, the complex space provides enhanced
a direct correspondence can be established between the mathematical structure, yielding operations and functions
elements of each of these spaces. More specifically, let that substantially extends its possible theoretical and ap-
s = x + i y be a generic complex number and ~v = [x, y]T plied utilization.

9
16 Concluding Remarks Sine_Cosine_Periodicity_Phase_Sine_CDT-33.
[Online; accessed 1-March-2020.].
Mainly motivated as a solution to the square root of −1,
complex numbers have proven to be critically important
for extending the two-dimensional Euclidean space into a
[4] L. da F. Costa. Signals: From analog
complex space endowed with several additional important
to digital, and back. Researchgate, 2020.
operations and functions. These concepts have found es-
https://www.researchgate.net/publication/
sential applications not only in mathematics, but in virtu-
344595846_Signals_From_Analog_to_Digital_
ally every theoretical and applied scientific areas, includ-
and_Back_CDT-39. [Online; accessed 09-March-
ing but by no means limited to: signal analysis, electri-
2020.].
cal and mechanical engineering, acoustics, number theory,
control theory, scientific visualization, shape analysis and
complex systems.
The present work aimed at introducing some of the [5] L. da F. Costa. Instantaneous signal analysis.
most important basic concepts related to complex num- https://www.researchgate.net/publication/
bers, from the idea of an imaginary number to its appli- 344750267_Instantaneous_Signal_Analysis_
cations in interesting subjects as fractals and number the- CDT-40, 2020. [Online; accessed 02-Feb-2021].
ory. Among the several concepts that have been briefly
covered we have: the complex exponential, basic com-
plex operations, complex powers of real and complex val- [6] Wikipedia. Wirtinger derivatives. https://en.
ues, complex roots, complex parametric functions, generic wikipedia.org/wiki/Wirtinger_derivatives,
complex functions, the concept of analyticity and confor- 2021. [Online; accessed 02-Feb-2021].
mality, the Mandelbrot set, the Riemann zeta function,
and a consideration on the similarities and distinctions
between the two important spaces R2 and C.
Despite the general perspective adopted by the present [7] H.-O. Peitgen, H. Jürgens, and D. Saupe. Chaos and
work, special attention has been given to concepts related Fractals: New Frontiers of Science. Springer, 2004.
to signal analysis, electrical and mechanical engineering,
acoustics, image and shape analysis, dynamical systems,
complex systems, fractals, and number theory. These [8] A. Ledoan. Zeros of partial sums of the
constitute but a small sample of the wide range of ap- Riemann zeta-function. Report, 2021.
plications of complex numbers. The prospective reader https://web.williams.edu/Mathematics/
is encouraged to probe further into related literature and sjmiller/public_html/ntandrmt/talks/
applications (e.g. []). OremTalk-2009GraduateWorkshop_Ledoan.pdf.
[Online; accessed 5-March-2021.].
Acknowledgments.
Luciano da F. Costa thanks CNPq (grant
no. 307085/2018-0) and FAPESP (grant 15/22308- [9] L. da F. Costa. A first glance at prime numbers.
2). https://www.researchgate.net/publication/
349466484_A_First_Glance_at_Prime_Numbers_
CDT-55, 2021. [Online; accessed 02-Feb-2021].

References [10] Wikipedia. Proof of the Euler product formula for

the Riemann zeta function. Wikipedia, The Free
[1] J. Brown and R. Churchill. Complex Variables and
Encyclopaedia, 2021. https://en.wikipedia.org/
Applications. Pearson, 3rd edition, 2018.
wiki/Proof_of_the_Euler_product_formula_
for_the_Riemann_zeta_function. [Online; ac-
[2] S. D. Fisher. Complex Variables. Dover, 1999.
cessed 10-Feb-2021.].
[3] L. da F. Costa. Sine, cosine, periodicity, phase,
sine, ... Researchgate, 2020. https://www.
researchgate.net/publication/341722757_

10
 Costa’s Didactic Texts – CDTs
CDTs intend to be a halfway point between a
formal scientific article and a dissemination text
in the sense that they: (i) explain and illustrate
concepts in a more informal, graphical and acces-
sible way than the typical scientific article; and
(ii) provide more in-depth mathematical develop-
ments than a more traditional dissemination work.

It is hoped that CDTs can also incorporate new

insights and analogies concerning the reported
concepts and methods. We hope these character-
istics will contribute to making CDTs interesting
both to beginners as well as to more senior
researchers.

Each CDT focuses on a limited set of interrelated

concepts. Though attempting to be relatively
self-contained, CDTs also aim at being relatively
short. Links to related material are provided in
order to provide some complementation of the
covered subjects.

Observe that CDTs, which come with absolutely

no warranty, are non distributable and for non-
commercial use only.

Please check for new versions of CDTs, as they can

be revised. Also, CDTs can and have been cited,
e.g. by including the respective DOI. Please cite
this CDT in case you use it, so that it may also be
useful to other people. The complete set of CDTs
can be found at: https://www.researchgate.
net/project/Costas-Didactic-Texts-CDTs,
and a respective guide at: https://www.
researchgate.net/publication/348193269_A_
Guide_to_the_CDTs_CDT-0

11
View publication stats