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

net/publication/352005870

Harmonic Analysis on Clifford Algebras by Wavelets

Thesis · May 2021

DOI: 10.13140/RG.2.2.26889.98402

CITATIONS READS

0 466

1 author:

Hicham Banouh
Université de Bouira
10 PUBLICATIONS 26 CITATIONS

SEE PROFILE

All content following this page was uploaded by Hicham Banouh on 31 May 2021.

The user has requested enhancement of the downloaded file.

 N ◦ d’ordre : ../20..-./...
République Algérienne Démocratique et Populaire
Ministère de L’Enseignement Supérieur et de la Recherche
Scientifique
Université des Sciences et de la Technologie HOUARI BOUMEDIENE

Faculté de Mathématiques

Thèse
Présentée pour l’obtention du diplôme de DOCTORAT 3ème cycle
En : MATHEMATIQUES
Spécialité : Contrôle Optimal des EDP
Par : BANOUH Hicham
Sujet

Analyse Harmonique par Ondelettes dans

l’Algèbre de Clifford

Soutenue publiquement, le JJ/MM/AAAA, devant le jury composé

de :

Mr BAHLOUL Djilali Professeur à l’USTHB Président

Mr BEN MABROUK Anouar MCF-HDR à l’ISMAI, Kairouan Encadreur
Mr KESRI M’hamed Professeur à l’USTHB Co-Encadreur
Mr TOUZALINE Arezki Professeur à l’USTHB Examinateur
Mr MENOUNI Abdelaziz Professeur à l’Université Mostafa Ben Boulaid, Batna II Examinateur
Mr BENAISSA Abbes Professeur à l’Université Djilali Liabes, Sidi-Bel-Abbes Examinateur
Mr LAADJ Toufik MCA à l’USTHB Invité
 People’s Democratic Republic of Algeria
Ministry of Higher Education and Scientific Research
University of Science and Technology HOUARI BOUMEDIENE
Faculty of Mathematics

Thesis
rd
Presented to obtain a 3 cycle DOCTORATE degree
In : MATHEMATICS
SPECIALTY : Optimal Control of PDE
By : BANOUH Hicham
Subject

Harmonic Analysis by Wavelets on Clifford

Algebras

Publicly defended, on DD/MM/YYYY, before a jury composed of:

Mr BAHLOUL Djilali Professor at USTHB President

Mr BEN MABROUK Anouar Associate Pr. at ISMAI, Kairouan Thesis supervisor
Mr KESRI M’hamed Professor at USTHB Thesis co-supervisor
Mr TOUZALINE Arezki Professor at USTHB Examiner
Mr MENOUNI Abdelaziz Professor at Uni. Mostafa Ben Boulaid, Batna II Examiner
Mr BENAISSA Abbes Professor at Uni. Djilali Liabes, Sidi-Bel-Abbes Examiner
Mr LAADJ Toufik Associate Pr. at USTHB Invited
 University of Sciences and Technology Houari Boumediene
Faculty of Mathematics

Dissertation

Analyse Harmonique par Ondelettes dans l'Algèbre de Clifford

by
Hicham BANOUH

Thesis
Submitted in partial fulfilment of the
requirements for the degree of
Doctor of Mathematics
2020
© Copyright by
Hicham BANOUH

2020
 Approved by
First Reader: BEN MABROUK Anouar, Advisor
Associate-Professor, Higher institute of Applied Mathematics and Computer
Sciences, University of Kairaouan, Tunisia,

-------------------------------------

Second Reader: KESRI M’hamed, Co-Advisor

Professor, University of Sciences and Technology H. Boumediene,

-------------------------------------

Third Reader: BAHLOUL Djilali, Chairman

Professor, University of Sciences and Technology H. Boumediene,

-------------------------------------

Fourth Reader: TOUZALINE Arezki, Examiner

Professor, University of Sciences and Technology H. Boumediene,

-------------------------------------

Fifth Reader: MEMMOUNI Abdelaziz, Examiner

Professor, University Mostafa Ben Boulaid, Batna II

-------------------------------------

Sixth Reader: BENAISSA Abbes, Examiner

Professor, University Djilali Liabes, Sidi-Bel-Abbes

-------------------------------------

Seventh Reader: LAADJ Toufik, Invited

Associate-Professor, University of Sciences and Technology H. Boumediene,

-------------------------------------
 Acknowledgments

Firstly, I would like to thank my advisor, Dr. Anouar Ben Mabrouk, for his
unlimited patience, knowledge, guidance, support, and encouragement. You
have inspired me and helped me to grow as a mathematician. I cannot thank you
enough.
I thank my co-advisor Professor M’hamed Kesri for his great help during those
three last years and for the very usefull conversations we had.
I am very thankful to the chairman and the the members of the Thesis committee
for thier reports and thier presence.
To Mom and my late Dad, thank you for instilling in me a love of learning and
discovery. It has taken me far and served me well. Thank you for your
unconditional love, encouragement, and for putting up with my complaints.
Finally to Mina, Mazigh and Mayas : thank you for your constant love, support
and patience. Your unwavering belief in me kept me going when things were
toughest. I could not have done this without you.

Analyse Harmonique par Ondelettes dans l'Algèbre de Clifford

 (Order No. )

Hicham BANOUH

University of Sciences and Technology Houari Boumediene,

Faculty of Mathematics, 2020

Supervisors: BEN MABROUK Anouar,

Associate-Professor, Higher institute of Applied Mathematics and Computer Sciences,

University of Kairaouan, Tunisia,

KESRI M’hamed,

Professor, University of Sciences and Technology H. Boumediene,

ABSTRACT
The project falls within the general framework of harmonic analysis on Clifford
algebras. We propose more precisely to extend the famous Heisenberg uncertainty principle to
the context of Clifford algebras by applying the so-called Clifford wavelets.

In mathematical physics, Clifford analysis has been developed as an extension of the

classical harmonic analysis where concepts such as Fourier transforms and wavelets have
been extended for the case of Clifford algebras.

In the present work, our aim is to study and establish a new Heisenberg uncertainty
principle based on Clifford wavelet transform. We recall that the majority of uncertainty
principles in their different forms are based essentially on Fourier transform and the
wavelet uncertainty principles already established in the literature did not been extended
to general Clifford algebra framework.

ⴰⴳⵣⵓⵍ
 ⴰⵙⴻⵏⴼⴰⵔ ⴰⴳⵉ ⵢⴻⵜⵜⵉⵍⵉ ⴷⴻⴳ ⵙⵙⴻⴳⵔⴻⵔ ⴰⵎⴰⵜⵓ ⵏ ⵜⴰⵙⵍⴻⴹⵜ ⵜⴰⵎⵜⴰⵡⴰⵏⵜ ⴷⴻⴳ ⵓⵊⴱⴰⵔⴻⵏ ⵏ
Clifford. ⵏⵙⵙⵓⵎⴻⵔ ⴷⴻⴳ ⵓⵇⴷⵉⵛ ⴰⴳⵉ ⴰⵏⵙⵎⴻⵖⵔⴻⴷ ⴰⵣⵡⵉⵔ ⵏ ⵜⴻⴳⵍⴰⴳⵉⵍⵜ ⵏ Heisenberg ⴰⵎⵓⵍⴰⵏ
ⴷⴻⴳ ⵓⵙⵙⴻⴳⵔⴻⵔ ⵏ ⵓⵊⴱⴰⵔⴻⵏ ⵏ Clifford ⵙ ⵓⵙⵏⴻⵙⵏ ⵏ ⵜⵉⵎⴷⴻⵙⵡⴰⵍ ⵜⵉⵎⴻⵥⵢⴰⵏⵉⵏ ⵏ Clifford.

ⴷⵉ ⵜⵓⵙⵏⴰⴽⵜ ⵜⴰⵙⴻⵏⴳⴰⵎⴰⵜⵜ, ⵜⴰⵙⵍⴻⴹⵜ ⵏ Clifford ⵜⴻⵜⵜⴻⵡⵙⵏⴻⴼⵍⴻⴷ ⴰⵎ ⵓⵙⴻⵎⵖⴻⵔ ⵏ ⵜⴰⵙⵍⴻⴹⵜ

ⵜⴰⵎⵜⴰⵡⴰⵏⵜ ⴰⵏⵉⴷⴰ ⵉⵏⴻⴳⵎⴰⵏ ⴰⵎ ⵜⴱⴻⴷⴷⵉⵍⵜ ⵏ Fourier ⴰⴽⴻⴷ ⵜⵉⵎⴻⵥⵢⴰⵏⵉⵏ ⵜⵜⵡⵙⵎⴻⵖⵔⴻⵏⵜ ⵖⴻⵔ
ⵜⴰⵙⵍⴻⴹⵜ ⵏ Clifford.

ⴷⴻⴳ ⵜⵙⴻⵔⵙⵉⵜ ⴰⴳⵉ ⵏⴻⵜⵜⵏⴰⴷⵉ ⴰ ⵏⴻⵍⵎⴻⴷ ⴰⴽⴻⴷ ⴰⴷⵏⴰⴼ ⴰⵣⵡⵉⵔ ⵏ ⵜⴻⴳⵍⴰⴳⵉⵍⵜ ⵏ Heisenberg
ⴰⵎⴰⵢⵏⵓⵜ ⵢⴻⵜⵜⵡⴰⴱⵏⴰⵏ ⵖⴻⴼ ⵜⴰⴱⴻⴷⴷⵉⵍⵜ ⵏ ⵜⵉⵎⴻⵥⵢⴰⵏⵉⵏ ⵏ Clifford. ⴰⴷ ⵏⴻⵙⵎⴻⴽⵜⵉ ⴱⴻⵍⵍⵉ ⵜⵉⴳⴷⵉ
ⵏ ⵉⵣⵡⵉⵔⴻⵏ ⵏ ⵜⴻⴳⵍⴰⴳⵉⵍⵜ ⴷⵉ ⵜⴰⵍⵖⵉⵡⵉⵏ ⵏⵙⴻⵏ ⵉⵎⴳⴰⵔⴰⴷⴻⵏ ⵜⵜⵡⴰⴱⵏⴰⵏⵜ ⵖⴻⴼ ⵜⴱⴻⴷⵉⵍⵜ ⵏ Fourier
ⴻⵔⵏⵓ ⵉⵣⵡⴰⵔⴻⵏ ⵏ ⵜⴻⴳⵍⴰⴳⵉⵍⵜ ⵢⴻⵜⵜⵡⴰⴼⴻⵏ ⴷⵉ ⵜⵙⴻⴽⵍⴰ ⵓⵔ ⵜⵜⵡⵙⵎⴻⵖⵔⴻⵏ ⴰⵔⴰ ⵖⴻⵔ
ⵙⴰⵙⵙⴻⴳⵔⴻⵔ ⵏ ⵓⵊⴱⴰⵔⴻⵏ ⵏ Clifford ⴰⵎⴰⵜⵓ.

‫ملخص‬
‫ نقترح على نحو أكثر دقة توسيع‬.‫يندرج المشروع في اإلطار العام للتحليل التوافقي في جبر كليفورد‬
‫بمويجات كليفورد‬. ‫مبدأ عدم اليقين الشهير لهايزنبيرغ في سياق جبر كليفورد من خالل تطبيق ما يسمى‬

‫ تم تطوير تحليل كليفورد كامتداد للتحليل التوافقي الكالسيكي حيث تم تمديد‬،‫في الفيزياء الرياضية‬
‫مفاهيم مثل تحويالت فورييه و المويجات في حالة جبر كليفورد‬

‫ هدفنا هو دراسة وإنشاء مبدأ جديد لهايزنبيرغ لعدم اليقين يستند إلى تحويل مويجات‬,‫في العمل الحالي‬
‫ وأن مبادئ‬،‫ ونذكر أن معظم مبادئ عدم اليقين في أشكالها المختلفة تقوم أساسا على تحويل فورييه‬.‫كليفورد‬
‫عدم اليقين التي وضعت بالفعل في المؤلفات لم تمتد إلى إطار الجبر العام لكليفورد‬.

RESUME
Le projet s'inscrit dans le cadre général de l'analyse harmonique sur les algèbres de
Clifford. Nous proposons plus précisément d'étendre le fameux principe d'incertitude
d’Heisenberg au contexte des algèbres de Clifford en appliquant les ondelettes de
Clifford.

En physique mathématique, l'analyse de Clifford a été développée comme une extension

de l'analyse harmonique classique où des concepts tels que les transformations de Fourier
et en ondelettes ont été étendus pour le cas de Clifford algèbres.

Dans ce travail, notre objectif est d'étudier et d'établir un nouveau principe d'incertitude
d’Heisenberg basé sur la transformation en ondelettes de Clifford. Nous rappelons que la
majorité des principes d'incertitude sous leurs différentes formes sont basés
essentiellement sur la transformation de Fourier et que les principes d’incertitudes relatifs
aux ondelettes n’ont pas été etendus aux algèbres de Clifford générales.
Contents

1 Introduction and Main Results 1

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Control Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3 Main Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2 Wavelet Theory Revisited 12

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2 Wavelets on R . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.1 Admissibility . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.2 Rapid examples of wavelets . . . . . . . . . . . . . . . . . . 15
2.2.3 The Continuous Wavelet Transform . . . . . . . . . . . . . . 16
2.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3 Clifford Algebra/Analysis Toolkit 22

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.2 Clifford Algebras . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2.1 The original definition made by Clifford . . . . . . . . . . . 24
3.2.2 Riesz’s construction . . . . . . . . . . . . . . . . . . . . . . . 25
3.2.3 Rapid Examples in low dimensions . . . . . . . . . . . . . . 25
3.3 Gradings of a Clifford Algebra . . . . . . . . . . . . . . . . . . . . . 26
3.3.1 P in and Spin Groups . . . . . . . . . . . . . . . . . . . . . 31
3.4 Clifford Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

Thesis in Mathematics Hicham BANOUH

CONTENTS viii

3.4.1 Clifford Algebra Valued Functions . . . . . . . . . . . . . . . 33

3.4.1.1 Differential and derivative of a Clifford algebra-
valued function . . . . . . . . . . . . . . . . . . . . 35
3.4.1.2 Dirac Operator and Monogenic Functions . . . . . 36
3.4.2 Generating Monogenic Functions . . . . . . . . . . . . . . . 41
3.4.3 Cauchy-Clifford Integral Formula . . . . . . . . . . . . . . . 44
3.5 Clifford-Fourier Transform . . . . . . . . . . . . . . . . . . . . . . . 47
3.6 Clifford Wavelet Transform . . . . . . . . . . . . . . . . . . . . . . . 58
3.6.1 Proprieties of the Clifford Wavelet Transform . . . . . . . . 61
3.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

4 Clifford Wavelet Uncertainty Principle 68

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.2 Old uncertainty principle revisited . . . . . . . . . . . . . . . . . . . 68
4.3 Clifford wavelet uncertainty principle . . . . . . . . . . . . . . . . . 73
4.4 A sharper Clifford wavelet uncertainty principle . . . . . . . . . . . 76
4.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

5 Conclusion and perspectives 81

5.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
5.2 Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.2.1 Practical Applications . . . . . . . . . . . . . . . . . . . . . 82
5.2.2 Donoho-Stark uncertainty principle for the Clifford wavelet
transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.2.3 Continuous shearlet transform in Clifford algebra . . . . . . 83
5.2.4 New uncertainty principles for the Clifford wavelet transform 83
5.2.5 Controllability of Clifford algebra valued Systems . . . . . . 84

Thesis in Mathematics Hicham BANOUH

Chapter 1
Introduction and Main Results

1.1 Introduction

In the present thesis, we seek to establish a new uncertainty Heisenberg prin-

ciple applied to continuous Clifford wavelet transform in the settings of the non-
commutative Clifford algebras (named after the British mathematician and philoso-
pher William Kingdon Clifford (1845-1879). Those Clifford algebras [53, 54] are
a generalization of the real and complex numbers and also of Sir William Rowan
Hamilton’s (1805-1865) quaternions [100, 99] and Hermann Günther Grassmann’s
(1809-1877) exterior algebra [53, 91]. These algebras incorporate inside one single
structure the geometrical and algebraic properties of Euclidean space, that Clif-
ford called them geometrical algebras. They were rediscovered when Paul Adrian
Maurice Dirac (1902-1984) used what he called γ-matrices to find a linearisation
of the Klein-Gordon equation [73, 72] (those γ-matrices are just generators of a
particular Clifford algebra).
The calculus on Clifford algebra treats geometric entities depending on their
dimension such as scalars, vectors, bivectors and volume elements, etc. and also

Thesis in Mathematics Hicham BANOUH

Introduction and Main Results 2

that it encompasses all dimensions at once, as opposed to a multi-dimensional

tensorial approach with tensor products of one-dimension. The use of Clifford
algebras in harmonic analysis uses the fact that holomorphic functions in the
complex plane are in the kernel Cauchy-Riemann operator which factorizes the
Laplacian. Clifford analysis (study of Clifford algebras valued functions of vector
variable) (see for instance [37]) are generalization to higher dimension of the theory
of holomorphic functions in the complex plane, their counterpart being monogenic
functions (null solutions of a first order differential operator factorizing the multi-
dimensional Laplacian). This also means that monogenic functions are harmonic
ones. More on functional analysis of Clifford algebra valued functions can be found
in [15, 23, 27, 37, 41, 70, 74, 89, 96, 115, 119, 120, 136, 137, 138, 139, 141, 145,
168, 169, 181, 183, 186, 184, 185, 187, 193, 196, 198, 197, 199, 200, 205, 209, 210].
Mathematically and quantitatively speaking wavelet analysis of functions starts
by computing a type of transform known as wavelet transform similar to Fourier
one and which consists in a convolution product of the function with special copies
of one source analysing function called mother wavelet and which plays the role of
the exponential in Fourier analysis .
The original work on wavelet analysis has been done by Morlet in [161] to study
seismic waves. He also, with Grossman, gave a mathematical study of continuous
wavelet transform (see [93]). In [158], Meyer recognized the link between harmonic
analysis and Morlet’s theory and gave a mathematical foundation to the continuous
wavelet theory. The continuous-wavelet analysis of a square integrable function
f begins by a convolution with copies of a given mother wavelet ψ translated
and dilated respectively by b ∈ R and a > 0. Such a function ψ has to fulfil an

Thesis in Mathematics Hicham BANOUH

Introduction and Main Results 3

admissibility condition which states that

2
Z ψ(ξ)
b
Aψ = dξ < +∞,
R |ξ|
where ψb is the classical Fourier transform of ψ. More information on real wavelets
can be found in [67] and [95].
The connection between the wavelet transform and Clifford analysis was consid-
ered in [55, 130, 160]. Clifford analysis/algebra has started to take place especially
in signal and image processing (See for instance [43], [54], [66], [73], [99], [100],
[199]). In [180] a wavelet based method has been developed in the quaternion
algebra leading to quaternionic representations of face image. See also [117], [194],
[213]. In [109] real and complex Fourier transforms are extended to quaternions
and Clifford algebras, motivated by applications in nuclear magnetic resonance,
electric engineering, colour image and signal processing. In [146], the authors
introduced quaternion-valued wavelets.
The uncertainty principle also known as Heisenberg’s uncertainty principle dis-
covered in 1927 by Heisenberg in [102] and revisited next in [104] is certainly one
of the most famous and important concepts of quantum mechanics. It plays an
important role in the development and understanding of quantum physics. The
physical origin of uncertainty principle is related to quantum systems and states
that: the determination of positions by performing measurement on the system
disturbs it sufficiently to make the determination of momentum imprecise and
vice-versa. It has been described by Heisenberg as [103, Page 30]:

One can never know with perfect accuracy both of those two important
factors which determine the movement of one of the smallest particles
— its position and its velocity. It is impossible to determine accurately

Thesis in Mathematics Hicham BANOUH

Introduction and Main Results 4

both the position and the direction and speed of a particle at the same
instant.

The uncertainty principle has been extended to various transformations in different

settings. Using Fourier transform, the authors in [6] established a Stein-Weiss type
inequality for the Riesz type potential generated by a Riemann-Liouville opera-
tor. Pitt’s and Beckner logarithmic uncertainty inequalities have been also proved.
The same authors investigated in [7] a Hausdorff-Young inequality for the Fourier
transform connected with Riemann-Liouville operator. Such inequality has been
applied next to prove an entropy based uncertainty principle and a Heisenberg-
Pauli-Weyl inequality (See also [114]). In [165], [164], two types of uncertainty
principle such as Heisenberg-Pauli-Weyl and Beurling-Hörmander have been es-
tablished for the Fourier transform associated with the spherical mean operator in
some local framework.
Using real wavelet transform in [176], continuous wavelet transform associated
with the spherical mean operator has been introduced yielding a Plancherel formula
as well as its inversion. Such findings have been applied next to prove an analogue
of Heisenberg’s inequality for the introduced wavelet transform (See also [174],
[175]).
In [59] the continuous shearlet transform has been investigated to construct
mother Shearlet function applied next for an associated general uncertainty prin-
ciple. Minimizers of such uncertainty have been also developed by means of the
new wavelets.
El-Haoui et al in [79] introduced the quaternionic offset linear canonical trans-
form and derived a relationship with the quaternion Fourier transform to establish

Thesis in Mathematics Hicham BANOUH

Introduction and Main Results 5

next Plancherel like rules. These findings have been applied next to prove different
uncertainty principles including Heisenberg-Weyl’s, Hardy’s, Beurling’s and loga-
rithmic ones in the case of the new quaternionic offset linear canonical transform.
Recently El-Haoui and Fahlaoui established in [78] several uncertainty inequalities
in the real Clifford algebra Rp,q such as Hausdorf-Young inequality and qualitative
uncertainty principles of Donoho-Stark.
In [81] expansions of signals with respect to Gabor wavelets and short time
Fourier transform have been investigated. Using Heisenberg group techniques sta-
ble iterative algorithms for signal analysis and synthesis have been developed.
These algorithms have been shown to be convergent for a variety of norms and
compatibility with the time-frequency localization of signals has been proved.
While the Clifford geometric algebra Fourier Transform is global, the author
in [107] introduced the local Clifford geometric algebra wavelet concept using the
similitude group SIM (n). As an explicit example, the author introduced Clifford
Gabor wavelets. In [108] the same author derived a new directional uncertainty
principle for quaternion valued functions by means of quaternion Fourier transfor-
mation generalized to the case of Clifford geometric algebras.
In [106] basic concept multivector functions and their vector derivative in geo-
metric algebra have been introduced. Concepts of Fourier transform and Clifford
and some useful properties have been also investigated in the same framework of
geometric algebras. An uncertainty principle has been next developed for many
cases of Clifford wavelets and shown to be useful for signal processing.
In [111] a generalization of the Fourier transform in some Clifford geometric
algebras has been extended and adopted for real Clifford geometric algebra Fourier
transform. This has been applied next to to define and prove the uncertainty

Thesis in Mathematics Hicham BANOUH

Introduction and Main Results 6

principle for multivector functions in the new Clifford geometric algebras.

In [142] the quaternion ridgelet transform and curvelet transform associated to
the quaternion Fourier transform have been investigated and applied to derive an
associated reconstruction formulas, reproducing kernels and uncertainty principles.
In [153] an uncertainty principle associated with the quaternion linear canonical
transform has been proved by considering the fundamental relationship between
such transform and the quaternion Fourier transform. The new principle has
been applied to derive an inverse transform and Parseval and Plancherel formulas
associated with the quaternion linear canonical transform. (See also [147], [154]
for the same authors and similar subject).
Mawardi and Hitzer proposed in [149], [151] and [150] a construction of some
Clifford algebra valued wavelets using the similitude groups in a special case. The
new framework includes complex Gabor wavelets and extends them to multivec-
tors Clifford Gabor wavelets. A new uncertainty principle for the Clifford Gabor
wavelet transform has been proved in the new framework. Generalizations of these
results have been conducted by the same authors in [110].
In [152] the quaternionic Fourier transform has been applied to establish an un-
certainty principle for its right-sided. Such uncertainty principle has been shown to
prescribe a lower bound on the product of the effective widths of quaternion-valued
signals in the spatial and frequency domains. Furthermore, Gaussian quaternion
signals have been shown to be the only ones minimizing the uncertainty. In the
same direction in [131] the linear canonical transform has been revisited and next
generalized to quaternion-valued signals.
In [148] the continuous quaternion wavelet transform has been introduced with
admissibility condition expressed by means of the right-sided quaternion Fourier

Thesis in Mathematics Hicham BANOUH

Introduction and Main Results 7

transform. An application has been derived to establish a Heisenberg type uncer-

tainty principle for the new extended wavelets.
Recently, Mejjaoli et al considered in [157] a continuous wavelet transform asso-
ciated with the spherical mean operator relatively to some parameter h. Donoho-
Stark and Benedick-type uncertainty principles have been developed.
Yang et al [218] the authors investigated a stronger uncertainty principles in
terms of covariance and absolute covariance based on Fourier transform in both
directional and the spatial cases for real para-vector-valued signals. Conditions of
equality of the studied uncertainty principles have been discussed.
Finally, Yang and Kou in [219] applied the so-called linear canonical transforms
to extend the uncertainty principle for hypercomplex signals in the linear canonical
transform domains. Minimizers have been shown to be Gaussian signals, which
joins several works mentioned above.

Outline

This thesis is organized as follows :

• In chapter 2 we give a detailed review of the notion of continuous wavelet

transform, the definition of an admissible mother wavelet, its proprieties and
an inversion formula and a Plancherel-Parseval theorems.

• In the third chapter we give an introduction to Clifford algebras and Anal-

ysis including the different definitions of a Clifford algebra, its Z2 -grading,
the P in and Spin groups, monogenic functions and a Stokes formula and a

Thesis in Mathematics Hicham BANOUH

Introduction and Main Results 8

Cauchy representation formula and some important results and properties of

the Clifford-Fourier transform (links with the classical n-dimensional Fourier
transform) and the so called Clifford-wavelet are constructed by translation,
dilation and Spin-rotations. have been investigated.

• In the last chapter, we present a review of the uncertainty principle in dif-

ferent settings as for the Clifford-Fourier transform and we formulate and
prove the main results given in this thesis.

1.2 Control Theory

Optimal control theory is a branch of mathematical optimization that deals with

finding a control for a dynamical system over a period of time such that an ob-
jective function is optimized. It has numerous applications in both science and
engineering. For example, the dynamical system might be a spacecraft with con-
trols corresponding to rocket thrusters, and the objective might be to reach the
moon with minimum fuel expenditure. Or the dynamical system could be a na-
tion’s economy, with the objective to minimize unemployment; the controls in this
case could be fiscal and monetary policy. (For a review on the theory of control-
lability and observability see for instance [127, 126, 124, 125]).
In [207], the authors proposed an alternative feedback control in order to solve
affine control system, with quadratic cost functional based on the combination of
Haar wavelet and Generalized Hamilton-Jacobi-Bellman equation.
In [128], the authors derived the operational matrices of integration, derivative
and production of Hermite wavelets and used a direct numerical method based on
Hermite wavelet, for solving optimal control problems.

Thesis in Mathematics Hicham BANOUH

Introduction and Main Results 9

In [177], the authors gave a numerical method for solving non-linear optimal
control problems with inequality constraints by Legendre wavelet approximations.
In his PhD thesis, the author in [92] studied the interaction between problems
in control theory for partial differential equations and inequalities of the uncer-
tainty principle type. He also investigated the connection between compactness
of localization operators and uncertainty principles from an abstract harmonic
analysis perspective and gave results applied to the wavelet transform.
In [129], the authors investigated performance limitations and trade-offs in the
control design for linear time-invariant systems and showed that control specifi-
cations in time domain and in frequency domain are always mutually exclusive
determined by uncertainty relations.
For the moment, a theory of controllability and observability of Clifford algebra
valued differential systems is mainly studied and applied for the special case of the
algebra of quaternions H. For example in [216, 217] and [215] the authors used
quaternion valued equations to determine the best attitude for atmospheric entry
(see also [1, 121]).

1.3 Main Results

In this work we obtain anew results stating that the product of variances be-
tween Clifford-Fourier transform denoted by fb and the Clifford wavelet transform
Tψ [f ] (w.r.t an admissible Clifford algebra-valued mother wavelet ψ) of a square
integrable function f is lower bounded.

Thesis in Mathematics Hicham BANOUH

Introduction and Main Results 10

Theorem 1.3.1. Let ψ ∈ L2 (Rn , Rn , dV (x)) be an admissible Clifford mother

wavelet. Then for f ∈ L2 (Rn , Rn , dV (x)) the following inequality holds
Z Z  12 n
2 da (2π) 2 p
kbk Tψ [f ] (a, •, s)k2 n+1 ds ξk f ≥
b Aψ kf k22 , (1.1)
Spin(n) R + a 2 2
where k = 1, 2, · · · , n.

We even found an sharper result as given by the next theorem

Theorem 1.3.2. With the same hypothesis as in Theorem 1.3.1 we have

Z Z
da 1
q n o
( kbk Tψ [f ] (a, ·, s)k22 ds) 2 ξk fb ≥ 2n+1 π n Aψ kf k22 + 2 |hf1 , f2 i|
Spin(n) R+ an+1 2

where

1 da
R R R
ψ a,b,s (x)∂bk Tψ [f ] (a, b, s) an+1

f1 (x) =
 dV (b)ds
Aψ Spin(n) Rn R+

1 da
R R R
ψ a,b,s (x)bk Tψ [f ] (a, b, s) an+1

f2 (x) =
 dV (b)ds
Aψ Spin(n) Rn R+

These results will be detailed in Chapter 4. Moreover, we recall that they have
been partially published in the following papers [19, 18]:

• Banouh, H., Ben Mabrouk, A. and Kesri, M. Clifford-Wavelet Transform and

the Uncertainty Principle, Advances in Applied Clifford Algebras, 2019, Vol.
29, pp. 1-23. DOI:10.1007/s00006-019-1026-4.

• Banouh, H., Ben Mabrouk. A. A Sharp Clifford-Wavelet Heisenberg-type

Uncertainty Principle, J. Math. Phys., 2020, Vol. 62, Issue 9. DOI:10.1063/5.0015989.

and presented as a talk [16] in

Thesis in Mathematics Hicham BANOUH

Introduction and Main Results 11

• Banouh, H. Uncertainty Principle Associated with the Clifford Continuous

Wavelet Transform, 12th International Conference on Clifford Algebras and
Their Applications in Mathematical Physics, University of Science and Tech-
nology of China, Hefei, China, 3-7 August 2020.

We intend that this result will be extended to stronger inequalities. Some future
directions are exposed in the last chapter.

Thesis in Mathematics Hicham BANOUH

Chapter 2
Wavelet Theory Revisited

2.1 Introduction

Wavelet analysis is a time-scale representation of signals used in physics, math-

ematics, and engineering in the last few decades [57, 101, 143, 123, 214, 63, 68].
It began with the works of Jean Morlet (a French oil engineer who was analysing
geophysical data in the context of oil exploration [90]). Grossmann, Morlet, and
Paul [93, 94, 95] proved next that wavelets are simply coherent states associated
to the affine group of the real line (action of dilations and translations).
In [143], the discovery of orthonormal bases of regular wavelets has been pointed
out, and even with compact supports, as shown in [62], by changing the perspective
(of course, the orthonormal basis of the Haar wavelets was known since the be-
ginning of the century, but these are piecewise constant, discontinuous functions).
Group theory was replaced by the multiresolution analysis [144].
Most practical signals are non stationary and cover a wide range of frequencies.
In addition a direct correlation exists between the frequency of a given segment of

Thesis in Mathematics Hicham BANOUH

Wavelet Theory Revisited 13

the signal and the time duration of that segment. Low-frequency pieces tend to
last a long time, whereas high frequencies occur in general for a short moment only.
For example, vowels have low frequency and last for a long period of time, whereas
consonants short bursts of high frequency. Fourier transform (Fourier 1878) gives
us only informations about the frequency domain (symbolized by the variable ξ)
but no information on time localization (the variable x). For this reason, we need a
time-frequency representations : transform the signal from a one variable function
to a function of two variables : time and frequency as the Short Fourier Transform
or Gabor Transform (Denis Gabor in [86]) and the wavelet transform.
We obtain wavelets by starting with a function ψ of the real variable x. This
function is called a mother wavelet if it is well localized and oscillating. (It resem-
bles to a wave because it oscillates, and it is a wavelet because it is localized). The
localization condition is expressed in the usual way by saying that the function
decreases rapidly to zero as |x| −→ ∞. The second condition suggests that ψ
vibrates like a wave. Mathematically, we require that the integral of ψ be zero and
that preferably the other first N -moments of ψ also vanish.

2.2 Wavelets on R

To deal with and/or to conduct wavelet analysis of functions we usually need and
start with one source function which will be called next the mother wavelet and
which plays the role of the analysing source. Such function should satisfy several
assumptions to be able to analyse functions next.

Thesis in Mathematics Hicham BANOUH

Wavelet Theory Revisited 14

2.2.1 Admissibility

The analysing function must be square integrable But to have the CWT well
defined we add a condition on that analysis wavelet. This condition (admissibility
property) assures that the CWT can be inverted and so we can reconstruct the
signal again.

Definition 2.1. A function ψ ∈ L2 (R, C, dx) is called an admissible wavelet or

mother wavelet if it satisfy the admissibility condition ([52]):
Z b 2
|ψ(ξ)|
Aψ = 2π dξ < ∞, (2.1)
Z R |ξ|
b = √1
where ψ(ξ) e−ixξ ψ(x)dx stands for the classical Fourier transform of ψ.
2π R

The admissibility condition on ψ (2.1) can be reformulated as

2
Z ∞ ψ(tξ)
b
dt < ∞, ∀ξ ∈ R∗ . (2.2)
0 t
Also, (2.1) means that therefore
Z
ψ(0) =
b ψ(x)dx = 0.
R
This can be interpreted by the fact that the mother wavelet ψ is an oscillating func-
tion. This oscillating behaviour legitimises the use of the denomination wavelet
which has been used for decades in digital signal processing and exploration geo-
physics (one of the first occurrences of the word can be found in [178]). The
equivalent French word "ondelette" (little wave) was used by Morlet and Gross-
mann in the early 1980s in their works on oil prospecting.
In practice, we need more conditions on ψ such as a finite number N of van-
ishing moments , that is, for k = 0, 1, · · · , N "∈ N, we#have
dk ψ(ξ)
Z b
k
x ψ(x)dx = 0 ⇐⇒ = 0. (2.3)
R dξ k
ξ=0

Thesis in Mathematics Hicham BANOUH

Wavelet Theory Revisited 15

This will filter the polynomial components (at least of order ≤ N ) which is usually
the smoothest part of the signal and will show only the singularities and the sharp
parts. For example, if the analysing wavelet has vanishing moment of first order,
then the linear aspects of the signal will be ignored.

2.2.2 Rapid examples of wavelets

On R, we fortunately have many examples of explicit wavelets whom construction

as well as the proofs of their admissibility is not complicated. The simplest example
is known as the Haar wavelets [98, 97] where the mother wavelet is explicitly given
by 
1




1 ,0 ≤ x < 2


ψH (x) = −1 , 12 ≤ x < 1




, otherwise.

0


A next example is the Mexican Hat wavelet or Marr wavelet [144] obtained as the
2nd derivative of a Gaussian function
d2 − x2 x2
ψM H (x) = − 2 (e 2 ) = (1 − x2 )e− 2 .
dx
An important example is also due to Morlet [208, 93] and is based on the mother
wavelet
x2
ψM (x) = eic0 x− 2 .
This wavelet is closely related to human perception, both in the processes of au-
dition and vision.

Thesis in Mathematics Hicham BANOUH

Wavelet Theory Revisited 16

2.2.3 The Continuous Wavelet Transform

To analyse a function with wavelets and relatively to an analysing mother wavelet

we pass through the so-called Wavelet Transform of such a function. In wavelet
theory and similarly to Fourier one there is two different kinds of wavelet transform
: the continuous wavelet transform CWT (see for example [67] and [8]) and the
discrete one denoted DWT. In the present section we will introduce the CWT which
is the main topic of our work. In fact CWT is more adapted to the continuous
or time-wise signals and the DWT is used for discrete cases such as statistical
series, images, ... etc. The CWT is used in an analysis and detection and the
DWT is used for compression and signal reconstruction. The CWT is based on
the projection of a square integrable function (a physicist would say the it is a finite
energy signal) on a set of images of a chosen function by a group of symmetries.
The CWT is based on the action of two operators : The translation operator T b
defined for b ∈ R by
T b ψ(x) = ψ(x − b)
and the dilation operator Da defined for a > 0 by
1 x
Da ψ(x) = √ ψ( )
a a
We create a whole family of admissible wavelets, known sometimes as the daughter
wavelets by translating and dilating a mother wavelet ψ
1 x−b
ψ a,b (x) = T b Da ψ(x) = √ ψ( )
a a
which may be also seen as the action of the affine group ax + b on ψ.
The parameter a measures the compression or the scale and b is the translation
or the position parameter. If a < 1, the support of ψ a,b will be smaller than that
of the mother wavelet ψ and correspond to high frequencies. By the same way, if

Thesis in Mathematics Hicham BANOUH

Wavelet Theory Revisited 17

a > 1, the support of ψ a,b will be wider than that of ψ and will correspond to the
√
lower frequencial part of the signal. The factor 1/ a ensures the conservation of
the L2 -norm of both ψ and ψ a,b and plays the role of a normalization constant.
All these copies ψ a,b have the same L2 -norm as the mother wavelet ψ. Indeed,

ψ a,b L2 (R,C,dx)
= kψkL2 (R,C,dx) .

Furthermore, the frequency representation of a daughter wavelet ψ a,b satisfies

a,b (ξ) =
√ −ibξ
ψ
d ae ψ(aξ)
b

which yields that all these daughter wavelets are admissible also.

Definition 2.2. Continuous Wavelet Transform (CWT) Let ψ ∈ L2 (R, C, dx) be

an admissible wavelet and f ∈ L2 (R, C, dx). The CWT of f at the scale a and the
position b relatively to the analysing wavelet ψ is defined by the integral transform
Z
a,b
c
f (x) ψ a,b (x) dx.

Tψ [f ] (a, b) = f, ψ =
R
where c
stands for the complex conjugation.

Compared to the Fourier transform we notice that the CWT is a function of

two variables: time and frequency or position and scale. The kernel ψ a,b plays
the same role as the Fourier mode eixξ : we project a finite-energy signal f on the
space of wavelets and we calculate the correspondence between them. But on the
contrary of the Fourier analysis, the analysing function is not a single function but
a whole family generated from an admissible function called wavelet. Also, the
generated functions in addition to dilation are also translated, so the analysis is
performed on all the domain of definition of the signal. In terms of the convolution
product we have
Tψ [f ] (a, b) = (f ∗ ψ
fa )(b)

where ψ
fa (x) = √1 ψ( −x ).
a a
It is interpreted as a filter with a function of zero

Thesis in Mathematics Hicham BANOUH

Wavelet Theory Revisited 18

momentum.
We may remark that if the wavelet ψ has N -vanishing moments, the CWT
Tψ [f ] (a, b) of f will have the same order of magnitude as aN +1+1/2 (See [68, Page
102]). Indeed,
Z
Tψ [f ](a, b) = f (x)[ψ a,b (x)]c dx
R
x−b c x−b c
Z Z
1
= √ { f (b)[ψ( )] dx + f 0 (b)(x − b)[ψ( )] dx
a R a R a
Z 00 Z (N )
f (b) x−b c f (b) x−b c
+ [ψ( )] dx + · · · + (x − b)N [ψ( )] dx + · · · }
R 2 a R N! a
We see that the N first terms are zero. Consequently, putting x = at + b in the
(N + 1)-integral we get an approximation

|Tψ [f ]| ∼ aN +1+1/2 .

From the definition of the CWT we may easily notice the following properties. Let
ψ and φ be two admissible wavelets, f, g ∈ L2 (R, C, dx) and α, β ∈ C.

• Linearity rule

Tψ [αf + βg] (a, b) = αTψ [f ] (a, b) + βTψ [g] (a, b)

• The CWT is translation-invariant in the sense that

Tψ [Tb0 f ] (a, b) = Tψ [f ] (a, b − b0 )

• The CWT is dilation-invariant in the sense that

1 a b
Tψ [Da0 f ] (a, b) = √ Tψ [f ] ( , )
a0 a0 a0

• Introducing the parity operator Pf (x) = f (−x). We have

Tψ [Pf ] (a, b) = Tψ [f ](a, −b)

• Anti-linearity

Tαψ+βφ [f ](a, b) = αc Tψ [f ](a, b) + β c Tφ [f ](a, b)

Thesis in Mathematics Hicham BANOUH

Wavelet Theory Revisited 19

We may also prove that

c
• Tψ [f ] (a, b) = Tψ [f ] ( a1 , −b .

a

• TTb0 ψ [f ](a, b) = Tψ [f ](a, ab0 + b).

• TDa0 ψ [f ](a, b) = √1 Tψ [f ](a0 a, b).

a0

Besides as in Fourier analysis, we have also the possibility to reconstruct the anal-
ysed functions even non periodic one with analogues of Dirichlet, Parseval and
Plancherel rules by using the CWT.

Theorem 2.2.1. Let ψ be an admissible wavelet and f, g ∈ L2 (R, C, dx), then

Z Z +∞
da
Tψ [f ](a, b)[Tψ [g](a, b)]c 2 db = Aψ hf, giL2 (R,C,dx) . (2.4)
R 0 a

Proof. In the frequency domain we have

Tψ [f ] (a, b) = f, ψ a,b
D E
= f, ψ
b da,b , by Parseval formula for Fourier transform.
√
Z
= fb(ξ)[ ae−iξb ψ(aξ)]
b c
dξ
ZR
√ iξb c
= ae fb(ξ)[ψ(aξ)]
b dξ
R
and so
√ −iξb
Z
c
[Tψ [g](a, b)] = ae [b g (ξ)]c ψ(aξ)dξ.
b
R
This yields that
Z Z +∞ Z Z +∞ Z
c da
√ iξb c
Tψ [f ](a, b)[Tψ [g](a, b)] 2 db = { ae fb(ξ)[ψ(aξ)]
b dξ}
R 0 a R 0 R
√ −iξb
Z
da
×{ ae [b g (ξ)]c ψ(aξ)dξ}
b db (2.5)
R a2
Z Z +∞ Z Z
da
= { fb(ξ)[bg (ξ)]c [ψ(aξ)]
b cb
ψ(aξ)dξdξ} db.
R 0 R R a

Thesis in Mathematics Hicham BANOUH

Wavelet Theory Revisited 20

By interchanging the order of integration and putting aξ = z then (2.5) is equal

to
Z Z +∞ Z Z cb
[ψ(aξ)]
b ψ(aξ)
{ g (ξ)]c
fb(ξ)[b dξdξ}dadb
R 0 R R a
Z Z +∞ Z Z cb
[ψ(aξ)]
b ψ(aξ)
= { fb(ξ)[bg (ξ)]c dξdξ}dadb
R 0 R R a
Z b 2
|ψ(z)|
Z
c
= g (ξ)] ×
f (ξ)[b
b dξ
R R |z|
= Aψ hf, gi .

As a consequence of the latter result we have a Plancherel formula

Corollary 2.3. Let ψ be an admissible wavelet and f ∈ L2 (R, C, dx), then

Z Z ∞
da
|Tψ [f ](a, b)|2 2 db = Aψ kf k2L2 (R,C,dx) .
R 0 a

Also, we have an inversion formula making it possible to retrieve the analysed

function from its CWT.

Theorem 2.2.2. Let ψ be an admissible wavelet and f, g ∈ L2 (R, C, dx), then

Z Z ∞
1 da
f (x) = Tψ [f ](a, b)ψ a,b (x) 2 db
Aψ R 0 a
almost everywhere.

Thesis in Mathematics Hicham BANOUH

Wavelet Theory Revisited 21

Proof. From above we have

Z Z +∞
1 da
hf, gi = Tψ [f ](a, b)[Tψ [g](a, b)]c 2 db
Aψ R 0 a
Z Z +∞ Z
1 da
= Tψ [f ](a, b)[ g(x)[ψ a,b (x)]c dx]c 2 db
Aψ R 0 R a
Z Z +∞ Z
1 da
= Tψ [f ](a, b) [g(x)]c ψ a,b (x)dx 2 db
Aψ R 0 R a
Z Z +∞ Z
1 da
= ( Tψ [f ](a, b)ψ a,b (x) 2 db)[g(x)]c dx
Aψ R 0 a
Z +∞ Z R 
1 a,b da
= Tψ [f ](a, b)ψ (•) 2 db, g
Aψ 0 R a

Now, we characterize the image of L2 (R, C, dx) by a given CWT and introduce
thus some reproducing kernels associated to the CWT. Let Hψ = L2 (R+ dadb
∗ ×R, Aψ a2 ).

It is a reproducing kernel Hilbert space with kernel

0 0 1 D a,b a0 ,b0 E
Kψ (a, b; a , b ) = ψ ,ψ
Aψ
It is the solution of the integral equation
Z Z ∞
0 0 da
F (a , b ) = Kψ (a, b; a0 , b0 )F (a, b) 2 db.
R 0 a
A function F ∈ L2 (R+ ∗ × R, Cψ a dadb) is the CWT of a signal iff
−1 −2
Z Z ∞
c da
F (a, b) = [Kψ (a, b; a0 , b0 )] F (a0 , b0 ) 2 db
R 0 a

2.3 Conclusion

In this chapter, we presented the wavelets tool used for analysing functions. We
saw some proprieties of those wavelets and proved that under certain conditions,
the continuous wavelet transform is a an invertible operator. We will use similar
characteristics in the Clifford algebra framework which will be the subject of the
next chapter.

Thesis in Mathematics Hicham BANOUH

Chapter 3
Clifford Algebra/Analysis Toolkit

3.1 Introduction

Clifford analysis may be seen as a generalization of Fourier one in signal processing

as it applies real, complex and quaternion numbers. It may be also described with
the algebra of Pauli and Dirac matrices for physical space and Minkowski space-
time and thus a unifying language for mathematics and physics (See [34], [31]).
In [116] a Clifford algebra based algorithm has been developed for segmentation
of blood vessels. In [50] some contributions for Clifford algebra based colour image
processing have been reviewed and applied to define colour alterations. Clifford
algebras has been proved to be an efficient mathematical tool to investigate the
geometry of images. See also [188]. Recently, in [21] a new model for the generation
of orientation preference maps in the primary visual cortex has been developed
on retina and based on Clifford algebra considering both orientation and scale
features.
In [47] two convolution products for Hypercomplex Fourier transforms are stud-

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 23

ied for the analysis of higher dimensional signals such as colour images based on
Clifford Hermite wavelets. In [88] a sophisticated model based on Marr wavelet
kernel has been developed and applied on samples of intensity values for each pixel
in an image to estimate the probability density function of the pixel intensity. Marr
wavelet has been also applied in [211] to detect and characterize two-dimensional
vortex for a synthetic flow and propeller wake. See also [87], [202].
In [39], the authors studied the historical development of quaternion and Clif-
ford Fourier transforms and wavelets. Basic concepts have been revisited and
mathematical formulations has been enlightened. Hypercomplex Fourier trans-
forms and wavelets has been revisited with overviews on quaternion Fourier trans-
forms, Clifford Fourier transforms, quaternion and Clifford wavelets.
In [48], [49], the authors introduced new definition for general geometric Fourier
transform covering some Clifford cases. They showed necessary constraints to
obtain linearity, scaling and shift theorem. Applications in image/signal processing
and mathematical imaging vision in general have been discussed.
In this chapter we propose to state the most relevant results on Clifford alge-
bras and the theory of functions taking value in them. In the first section, we give
some preliminary concepts on Clifford algebra as their definitions, the matrix rep-
resentation of Clifford algebras and the Spin group (which will be used to describe
rotations in Rn ). In the forth section, we give a review on monogenic functions
which are counterparts of the holomorphic functions on the complex plan. In the
two final sections, we present the generalizations of the classical Fourier transform
and the wavelet transform in the Clifford algebras framework.

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 24

3.2 Clifford Algebras

In the literature, there are different ways to introduce Clifford algebras. We will
present in this part some of them. The readers may be referred to the list of
references provided in this document for more details and other constructions.

3.2.1 The original definition made by Clifford

It starts from the Grassmann exterior algebra [91] Rn of the linear space Rn ,
V

which is an associative algebra of dimension 2n . A basis for such an algebra may

be defined from an orthonormal basis {e1 , e2 , · · · , en } of Rn by considering the
tensor products for k = 1, 2, . . . , n and (i1 , i2 , . . . , ik ) ∈ {1; 2; ...; n}k a multi-index
of length k ≥ 1,

ei1 i2 ...ik = ei1 ∧ ei2 ∧ · · · ∧ eik .

and e∅ = 1 and with the extra assumption

ei ∧ ej = −ej ∧ ei for i 6= j and ei ∧ ei = 0.

In [54] the last assumption is replaced by

ei ei = e2i = 1.

The associative 2n -dimensional algebra, defined above is the Clifford algebra Rn

associated to the euclidean space Rn . We mention also, the case where he put
ei ei = −1 in 1878 which is the Clifford algebra associated with the anti-Euclidean
space R0,n (see for instance [53]).

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 25

3.2.2 Riesz’s construction

In [179] another construction of Clifford algebras has been provided. Let (V, Q) be
a n−dimensional quadratic space and A an associative algebra with the following
rules on addition and multiplication :

x2 = Q(x)

xy + yx = 2B(x, y)

where B is the bilinear form associated to Q. If {e1 , e2 , · · · , en } is a basis of V

then the latter conditions become

ei ej + ej ei = 2B(ei , ej ).

We introduce the Clifford algebra Rn over Rn as an associative algebra gener-

ated by the basis {e1 , e2 , ·
· · , en } satisfying the rules
ei ej + ej ei = 0 if i 6= j



ei ei = e2i = 1

 ∀1 ≤ i ≤ n.
This last construction resembles to the constructions formulated in [71] and [140].

3.2.3 Rapid Examples in low dimensions

The first and simplest example is the Clifford algebra R0,1 where the elements are
written on the form x = x0 1 + x1 e1 where x0 , x1 ∈ R and e21 = −1. We have in
fact By making the identity e1 = i, the imaginary unit we have the isomorphism

R0,1 ' C.

The second example is the well known quaternions algebra [100] denoted R0,2 with
its elements q = q0 1 + q1 e1 + q2 e2 + q12 e12 where qi ∈ R for i = 0, 1, 2, q12 ∈ R and
where e2i = −1, and ei ej = −ej ei for i 6= j. Choosing e1 = i, e2 = j, e12 = k the

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 26

elements of the canonical basis of the Euclidean space R3 we obtain the Hamilton
algebra constructed in [99] known also as the quaternion algebra which satisfies

H ' R2,0 .

3.3 Gradings of a Clifford Algebra

In the later we mainly consider the Clifford algebra Rn of the euclidean space Rn .
Any element of the Clifford algebra Rn can be written as
X
a= aA eA
A
where A is an arbitrary ordered multi-indices

A = ei1 ···ik , with 1 ≤ i1 < i2 < · · · < ik ≤ n.

For example, the Clifford algebra R3 is spanned by the family

{1, (e1 , e2 , e3 ), (e12 , e23 , e13 ), e123 }

then an element a in R3 will be

X
a= aA e A

= a0 + (a1 e1 + a2 e2 + a3 e3 )

+ (a12 e12 + a23 e23 + a13 e13 )

+ a123 e123

and the a’s are real numbers. Let k be the cardinality of A (|A| = k), we have
X Xn X
a= aA eA = aA e A .
A k=0 k=|A|

The subspace Rkn = SpanR {eA | |A| = k} will be called subspace of grade k.
For example, the subspace of grade 0 is the field R whose elements are called
scalars, the one of grade 1 is the vector space Rn composed of vectors, the elements
of one of grade 2 are called bivectors and finally, the one dimensional subspace of

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 27

grade n is called the set of pseudo-scalars. We have a decomposition of Rn

Rn = R0n ⊕ R1n ⊕ .... ⊕ Rnn

So that

a = a∅ 1 + (a1 e1 + .... + an en )
scalars vectors

+ (a12 e12 + a13 e13 + ....aij eij + ..... + an−1n en−1 en )

bivectors

+ · · · + (a123...n e123....n ).
pseudo−scalar
n
X
We may write a = [a]k where [a]k is the projector of a on Rkn . The operator
k=0
[.]k : Rn −→ Rkn for k = 0, 1, · · · , n satisfy for a, b ∈ Rn and λ ∈ R

[a + b]k = [a]k + [b]k

[λa]k = λ [a]k = [a]k λ

[[a]k ]k = [a]k

We define the complexification of Rn by Cn = C ⊗ Rn which means aA ∈ C in

A aA eA ∈ Cn or λ = a + ib ∈ Cn with a, b ∈ Rn .
P

The Clifford algebra Rn is Z2 −grader, which means that it is the direct sum of
an even and odd subspaces
M M
−
Rn = R+
n ⊕ Rn = Rkn ⊕ Rkn
k even k odd
and any Clifford number a can be written as

a = [a]+ + [a]−

where [a]± ∈ R±
n . These two components
 satisfy to the
 following inclusions [163]
−
R+ + 
R+

n Rn n Rn
 
⊂ Rn and
+
⊂ R−
n.
R− R
n n
− 
 R−n Rn
+ 

Also, as Rn is the direct sum of the even and odd subspaces and being of dimension

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 28

2n we have
−
dim(R+
n ) = dim(Rn ) = 2
n−1
.

Definition 3.1. The centre of the Clifford algebra Rn is the set of elements which
commute with all the other elements of the algebra. It will be noted by Z

Z(Rn ) := {a ∈ Rn |ab = ba, ∀b ∈ Rn }

We have 
for n even


R
Z(Rn ) =
for n odd

R ⊕ Re123···n


One remarkable concept in Clifford algebra is its attempt to mix between the
Grassmann exterior algebra and the Hamilton quaternion one. Let x and y be two
vectors. We define the Clifford product by

xy = x · y + x ∧ y
Xn
x·y = − xj yj = − < x, y >Rn
j=1
X
x∧y = ei ej (xi yj − xj yi )
i<j
Observing that y ∧ x = −x ∧ y we get

yx = y · x + y ∧ x = x · y − x ∧ y.

So that
1 1
x · y = (xy + yx) and x ∧ y = (xy − yx).
2 2

All this can be generalized to the product of a vector (grade 1) with a k-

grade element Ak which can be decomposed into the sum of an inner product
(contraction) and an outer one (extension)
1
x · Ak = [xAk ]k−1 = (xAk − (−1)k Ak x) (3.1)
2

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 29

Figure 3.1: wedge product of vectors

and
1
x ∧ Ak = [xAk ]k+1 = (xAk + (−1)k Ak x). (3.2)
2
In this case the Clifford product will be

xAk = x · Ak + x ∧ Ak .

We can even expand to the Clifford product of two k−grade,Ak , and l−grade, Bl ,
elements (see [105, page 6])

Ak · Bl = [Ak Bl ]|k−l| with Ak · Bl = 0 if kl = 0

and
Ak ∧ Bl = [Ak Bl ]|k+l|
On the real Clifford algebra Rn we may also define different types of involutions.
A first type is the main involution. It extends the vectorial reflection through
the origin to the whole algebra. Sometimes called grade involution, it is denoted
e and has following proprieties : for λ ∈ R, x ∈ Rn we have

λ e = −x, eei = −ei

e = λ, x and ef |A|
A = (−1) eA .

We thus obtain for all a ∈ Rn

a=a
e
e and a = [a]+ − [a]−
e

and for all a, b ∈ Rn ,

(ab)
g =e aeb and ^
(a + b) = e
a + eb.

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 30

For a k−grade element Ak we have

fk = (−1)k Ak .
A

The second inversion type is known as the reversion denoted by ∗ . It is defined

as follow
|A|(|A|−1)
λ∗ = λ, x∗ = x, e∗j = ej , e∗A = (−1) 2 eA

and for all a, b ∈ Rn ,

∗
(ab)∗ = b∗ a∗ , (a + b)∗ = a∗ + b∗ , a∗ = [a]+ + [a]− , a∗ = a.

For a product of vectors {v i }i∈N ⊂ Rn we have

Yk
( v i )∗ = v k v k−1 · · · v 2 v 1 .
i=1
For a k−grade element Ak we have
fk = (−1) k(k−1)
A 2 Ak .

Next, the Clifford conjugation is defined as the composition of the main invo-
lution and the reversion as follows,

a∗ .
ae∗ = e

It corresponds to the complex and quaternion conjugation in the case of R0,1 ' C
and R0,2 ' H respectively. The superposition will be denoted by so we have

a∗ .
a = ae∗ = e

So that
|A|(|A|+1)
λ = λ, x = −x, ej = −ej , eA = (−1) 2 eA

and for all a, b ∈ Rn ,

(ab) = ba, (a + b) = a + b, a = a

For a k−grade element Ak we have

fk = (−1) k(k+1)
A 2 Ak .

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 31

Finally, the complex Clifford conjugation is defined on the complexified Clif-

ford algebra C⊗Rn . We speak here about another involution by taking the complex
conjugation (noted by c ) of the Clifford conjugation. An element Λ ∈ C ⊗ Rn can
be written as Λ = a + ib, a, b ∈ Rn . The complex Clifford conjugation will be
noted by †
and defined by
Λ† = a − ib. (3.3)
It satisfies the properties

(ΛΘ)† = Λ† Θ† and (λΛ + θΘ)† = λc Λ† + θc Θ†

with Λ, Θ ∈ C⊗Rn and λ, θ ∈ C. To close with the operations on Clifford algebras

we recall the concept of norms on them. We define an inner product on Rn as
 
ha, bi = ab 0 = [ba]0
X
and so the Clifford norm |a| of a multivector a = aA eA ∈ Rn satisfy
A
X
|a|2 = aa = aa = a2A . (3.4)
A
We have for a, b ∈ Rn , |a + b| ≤ |a| + |b| and generally |ab| =
6 |a| |b|. We have
instead
|ab| ≤ 2n |a| |b| .
The equality holds if at least one of them is a vector

|ax| = |a| |x| , ∀x ∈ Rn .

3.3.1 P in and Spin Groups

Let us now consider the two following groups, the formed by the products of
invertible vectors (known as the versor group)
k
Y
Γ2p,q := { vi : v1 , v2 , · · · , vk ∈ (Rp,q )× }
i=1

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 32

and
Γ1p,q := {b ∈ R× e −1 for all vector x ∈ Rp,q }.
p,q ; bxb

We define the spinor norm, denoted N , as follows (cf. 3.4)

N : Rp,q −→ Rp,q

a 7−→ aa

Definition 3.2. We define the P in(p, q) group as the subgroup of Γ1p,q of elements
b for which N (b) = ±1. It can be shown that it is also the kernel of the mapping
N : Γ1p,q −→ R∗ .
The same way we define Spin(p, q) as the subgroup of P in(p, q) of product of
an even number of elements which the spinor norm equals ±1 or

Spin(p, q) = P in(p, q) ∩ R+
p,q
2l
Y
= {s ∈ Rp,q ; s = ω j with ω 2j = ±1, 1 ≤ j ≤ 2l}. (3.5)
j=1

3.4 Clifford Analysis

Clifford analysis, in its most basic form, is a refinement of harmonic analysis in

higher dimensional Euclidean spaces. By introducing the so-called Dirac operator,
researchers introduced the notion of monogenic functions extending holomorphic
ones. In this context, different concepts of real and complex analysis have been ex-
tended to the Clifford case such as Fourier transform (extended to Clifford Fourier
transform, Derivation of functions, ....). See [3], [4], [5], [12], [11], [10], [22], [27],
[37], [29], [34], [35], [33], [28], [42], [44], [31], [47], [70], [64], [65], [82], and [159].

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 33

3.4.1 Clifford Algebra Valued Functions

We define a multivector function as a mapping f from Rn to Rn , associating a

multivector f (a) to another multivector a. For example we have

Definition 3.3 (Exponential of a multivector). Let a ∈ Rn , we define the exponential of a

(noted ea or exp a) as the series
∞
X ak
a
e = (3.6)
k=0
k!

Proposition 3.4. For two commuting multivectors a and b we have

ea eb = ea+b

In the latter, we consider only functions with values on a Clifford algebra but
where the variable is a vector in Rn taken as a part of the Clifford algebra.
Xn
We put x = (x0 , x) ∈ Rn+1 and where x = ei xi ∈ Rn so we can write
i=1
n
X Xn
x = e 0 x0 + e i xi = ei xi
i=1 i=0
The vector space Rn can be seen as the hyper-plan {x = (x0 , x) ∈ Rn+1 : x0 = 0}.
Consider functions defined on the (n + 1)-dimensional vector space Rn+1 :

f : R ⊕ Rn ' Rn+1 −→ Rn ( or Cn ).

It may be expressed as
X
f (x) = eA fA (x), (3.7)
A
where fA are real-valued functions and A ⊂ {1, 2, · · · , n}. Its conjugate f is given
by
X
f (x) = eA fA (x)
A

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 34

for a function with values in the “real” Clifford algebra Rn , and

X †
f † (x) = eA fA (x)
A
for a function with values in the complexified Clifford algebra Cn = C ⊗ Rn .
The continuity and derivability of f is to be taken component-wise. Denote for
Ω an open domain in Rn ,
X
C (r) (Ω, Rn ) = {f : Ω −→ Rn ; f = fA eA with fA ∈ C (r) (Ω, R)}
A
and
X
C (r) (Ω, Cn ) = {f : Ω −→ Cn ; f = fA eA with fA ∈ C (r) (Ω, C)}.
A
The Clifford-valued function f belongs to the Lebesgue module Lp (Ω, Rn , dV (x)) if
X Z 1/p
all the components fA ∈ Lp (Ω, R, dx), 1 ≤ p < ∞, the norm 2n p
|fA | dV (x)
A Ω
being equivalent to the norm
Z 1/p
p
kf kp = |f | dV (x)
Ω
where dV (x) stands for the Lebesgue measure on Rn . We say that f, g ∈ Lp (Rn , Rn , dV (x))
are equal if the set {x ∈ Rn ; f (x) 6= g(x)} is negligible.
The inner product on L2 (Rn , Rn , dV (x)) is given by
Z
hf, giL2 (Rn ,Rn ,dV (x)) = f (x)g(x)dV (x) (3.8)
Rn
and
Z
hf, giL2 (Rn ,Cn ,dV (x)) = [f (x)]† g(x)dV (x). (3.9)
Rn
More explicitly,
XZ
hf, giL2 (Rn ,Rn ,dV (x)) = fA (x)gB (x)eA eB .
A,B Rn

We mainly use the complex Clifford conjugation (see 3.3. The inner product (3.8)
satisfies the Cauchy-Schwartz inequality

< f, g >L2 (Rn ,Rn ,dV (x)) ≤ kf kL2 (Rn ,Rn ,dV (x)) kgkL2 (Rn ,Rn ,dV (x)) . (3.10)

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 35

3.4.1.1 Differential and derivative of a Clifford algebra-valued function

Definition 3.5 (Directional Derivative). [56, Page 29] Let f : Rn −→ Rn a

multivector-valued function of a vector variable x ∈ Rn and taking an arbitrary
direction a ∈ Rn we define the vector differential (directional derivative of f in the
direction a)
f (x + λa) − f (x)
a · ∇f (x) := lim (3.11)
λ→0 λ
provided that the limit exists well defined for all a, where the limit is taken for
scalar λ.

This is similar to the usual definition of a directional derivative but extends

it to functions that are not necessarily scalar-valued but also vector valued or
multivector valued ones. We have these proprieties

(i) (a + b) · ∇f = a · ∇f + b · ∇f for a, b ∈ Rn .

(ii) (ta) · ∇f (x) = t(a · ∇f (x)) for t ∈ R.

(iii) a · ∇(f + g)(x) = a · ∇f (x) + a · ∇g(x).

(iv) a · ∇(f g)(x) = (a · ∇f (x))g(x) + f (x)(a · ∇g(x)).

(v) a · ∇ [f (x)]k = [a · ∇f (x)]k .

(vi) Let τ : Rn −→ R and f = f (τ (x)) then

df
a · ∇f (x) = (a · ∇τ (x)) .
dτ

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 36

3.4.1.2 Dirac Operator and Monogenic Functions

Introduce the Dirac operator (compare with (3.5))

∂x : C (r) −→ C (r−1)
n
∂f X
f 7−→ ∂x f = ek (3.12)
k=1
∂ xk
and the generalized Cauchy-Riemann operator or Weyl operator as
∂
Dx = + ∂x (3.13)
∂x0
also known as the Fueter-Delanghe operator (see [84],[69]). For the sake of sim-
plicity we denote from now on
∂
= ∂i
∂xi
so the operators (3.12) and (3.13) become
Xn
∂x = ei ∂i and Dx = ∂0 + ∂x
i=1
and their conjugates will be
n
X
∂x = ei ∂i and Dx = ∂0 − ∂x .
i=1
They have an action from the left
X
∂x f (x) = ei eA ∂i fA (x)
i,A
and right
X
f ∂x (x) = eA ei ∂i fA (x).
i,A
(see the formulas (3.2)). By the same way

Dx f = ∂0 f + ∂x f and

f Dx = f ∂0 + f ∂x .

In the special case where f is vector-valued, then we have ∂x f = div f + curl f

where the divergence and curl of f can be computed as the scalar and bivector

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 37

parts
1
div f = ∂x · f = (∂x f + f ∂x )
2
and
1
curl f = ∂x ∧ f = (∂x f − f ∂x ).
2
We consider the action of the group Spin(n) on a Clifford-valued function f (see
[89, Thm 3.6]) given by

s ∈ Spin(n) → Ls : f (x) → sf (s̄xs)s.

Definition 3.6. A partial differential operator with constant coefficients is called

Spin-invariant if it commutes with Ls .

Proposition 3.7. The Dirac operator is Spin-invariant i.e

∂x Ls = Ls ∂x .

Proof. We have
n
X
∂x Ls f (x) = ei ∂i {sf (sxs)s}
i=1
Xn
= ei s∂i {f (sxs)} s
i=1
n
X
= s( ei ∂i {f (sxs)})s
i=1

= s∂x f (sxs)s

= Ls ∂x f (x).

Remark 3.8. We remark (see [168, pp. 139]) that the Dirac operator ∂x maps even
parts to odd parts and odd parts to even parts (see the splitting in (3.3))

∂x [f (x)]+ = [∂x f (x)]−

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 38

and
∂x [f (x)]− = [∂x f (x)]+
As such, f is monogenic if and only if ∂x [f (x)]+ = 0 and ∂x [f (x)]− = 0. More
generally, the equation
∂x f = g
where f and g are two Clifford algebra-valued functions defined on Rn , can be
splitted into a system of two equations

 ∂x [f ]+ = [g]−


 ∂x [f ]− = [g]+


where [·]± is the projection onto even and odd sub-algebras R±

n.

Definition 3.9. A function f ∈ C 1 (Ω, Rn ) is called left-monogenic (resp. right-

monogenic) on Ω iff

Dx f = (∂x0 + ∂x )f (x0 + x) = 0, (3.14)

resp.
f Dx = 0.

In the special case where the function f takes values only on the vector space
Rn taken as a component of the real Clifford algebra Rn , so it can be written as
X n
f (x) = ei fi (x)
i=0
where, by analogy with (3.7), the (n + 1) functions fi are real valued. In this case,
the monogenicity condition (3.14) becomes
X
ei ej ∂i fj (x) = 0.
i,j

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 39

Or as f is vector-valued ∂x f = ∂x · f + ∂x ∧ f = 0 so we have
n

 X
∂x · f = − ∂xj fj = −div(f )



j=1
 X

∂x ∧ f = eij ∂xi fj − ∂xj fi = rot(f )



i<j
known as the Riesz system which describes an irrotational flow [203]. One can
show that

∂x2 = −∆n and ∆m+1 = Dx Dx ,

where ∆n and ∆n+1 are the Laplacian in Rn and Rn+1 respectively. This means
that a monogenic function is also a harmonic one and hence infinitely differentiable;
even more, it is an multivector-valued analytic function in Ω ⊂ Rn and so each
of its components fA is real-analytic in its domain. Now, we will try to write the
previous operators in terms of spherical coordinates. For that, we recall that we
can write x = rη with η = ni=0 ei ηi and ηi = xri for i = 0, 1, · · · , n. More explicitly
P

(see [24, Page 65])

   
x0 cos θ0
   
   

 x1 


 sin θ0 cos θ1 

   

 x2 

 sin θ sin θ cos θ
 0 1 2


.. ..
   
. .
   
   
  = r 
   Qi−1 

 xi 


 j=1 sin θj cos θi


 ..   .. 

 . 


 . 

   Q 
 n−1
xn−1  j=1 sin θj cos θn
  
  
   
Qn
xn j=1 sin θj
where 0 < θ1 , . . . , θn−1 < π and 0 < θn < 2π. Using the spherical coordinates, the
Dirac operator and its conjugate can be written (see [37, Page 49]) as
1 1
∂x = η(∂r + ∂η ) and ∂x = η(∂r + ∂η )
r r

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 40

where ∂η spherical Dirac operator acts on the sphere S n . We have [37, Page 49]
n ∂η
X ∂θi
· ∂θi
∂η = 2 and ∂η = x ∧ ∂x .
∂ηi
i=1
∂θi
Putting

Γη = η∂η and Γ?η = η∂η (3.15)

called the spherical Dirac operators, we also define their adjoints as

fη = ∂η η and Γ
Γ f? = ∂η η.
η

Consequently,
 
1 1 f?
∂x = η ∂r + Γη = (∂r + Γ )η
r r η
and
 
1f
∂x = ∂r + Γη η
r
Observing that
∆n = ∂x ∂x = ∂x ∂x (3.16)
and that ηη = ηη = 1 we obtain
1 fη )∂r + 1 (Γ
∆n = ∂r2 + (Γη + Γ fη Γη − Γη ).
r r
By the same way we get
1 f?η )∂r + 1 (Γ
∆n = ∂r2 + (Γ?η + Γ f? Γ? − Γ?η ).
r r η η
This means that
1 fη )∂r + 1 (Γ
fη Γη − Γη ) = ∂ 2 + 1 (Γ? + Γ f? )∂r + 1 (Γ
∂r2 + (Γη + Γ r η η
f? Γ? − Γ? )
r r r r η η η

Next as we have ([37, Page 50])

fη = Γ? + Γ
Γη + Γ f? = n
η η

then (3.16) becomes

n
∆n = ∂r2 +∂r + ∆η
r
where ∆η is the Laplace-Beltrami operator. Denoting next I the identity applica-

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 41

tions we get
 
∆η = (n − 1) I − Γ?η Γ?η .
n
X
Putting E = xi ∂i called the Euler operator, we obtain E = r∂r . As a result,
i=0
we get the following identities

• xDx = E − x ∧ ∂x + (x0 ∂x − x∂0 ) ,

• xDx = E + Γη ,

• x∂x + ∂x x = −2E − n,

• Γη = −
P
i<j eij (xi ∂j − xj ∂i ).

Γη is called the angular Dirac operator and Lij = xi ∂j − xj ∂i for i, j = 1, 2, · · · , n

are the angular momentum operators.

3.4.2 Generating Monogenic Functions

We know that a harmonic function is monogenic so we seek the inverse : given a

harmonic function f : Ω ⊂ Rn −→ R, is there a Clifford algebra-valued function
g, defined on the same domain Ω such that the restriction of g to its real values is
f or
f (x) = [g(x)]0 , ∀x ∈ Ω ?

Definition 3.10. [25, Def. 1] An open subset Ω of Rn is said to be star-shaped

with respect to some x0 ∈ Ω, if for all x in the interior of Ω, the subset Ω contains
the segment {(1 − t)x0 + tx; 0 ≤ t < 1}.

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 42

Proposition 3.11. [37, Page 48] If Ω is open and star-shaped with respect to the
origin and u : Ω → R is harmonic in Ω then the function
Z 1 Z 1 
n−1 n−1
g(x) = f (x)e0 + t ∂x f (tx)xdt − t ∂x f (tx)dt
0 0 0
is left monogenic in Ω and its scalar part is precisely the function f .

As for the previous result, we seek a monogenic function starting with an ana-
lytic one. For this, we use the Cauchy-Kowalevski extension (CK-extension). The
idea behind the CK-extension is to characterize solutions of a system of partial dif-
ferential equations by the restriction of some of their derivatives to a sub-manifold
of co-dimension one.
Let f be an analytic function on an open set U of Rn , then we can always
monogenically extend it to an open set Ω of Rn+1 where U = Rn ∩ Ω (known as
the Cauchy-Kowalevski extension (see for instance the original works [51, 133] and
more recently [38, 40, 199, 51][37, Page 111] and [71, Page 151] ), then the solution
F of the system 
 ∂x F (x0 , x) = −∂x F (x0 , x) in Rm+1

0

 F (0, x) = f (x)


is monogenic on Rn+1 and its restriction to Rn is f .

The extension can be realized by the operation
∞
X (−x0 )k k
F (x0 , x) = e−x0 ∂x
f (x) = ∂x f (x). (3.17)
k=0
k!
understood in the symbolic way. In fact, formally we have
Dx e−x0 ∂x f (x) = (∂0 + ∂x ) e−x0 ∂x f (x)

= ∂0 e−x0 ∂x f (x) + ∂x e−x0 ∂x f (x)

= −∂x e−x0 ∂x f (x) + ∂x e−x0 ∂x f (x)

=0

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 43

Definition 3.12. [58] A monogenic function f is called axial monogenic if it takes

the form
f (x0 , x) = A (x0 , r) + ηB (x0 , r)
where r > 0 and η are the spherical coordinate given in section (3.4.1.2) and A, B
are two scalar valued function. The monogenicity of f leads to the equations
n−1
∂x0 A − ∂r B = B and
r
∂x0 B + ∂r A = 0
called the Vekua system.

Using the following result we can extend a holomorphic complex function into
a monogenic one.

Theorem 3.4.1. [189, 84] Let n be odd and let z = x + iy 7−→ f (x + iy) be
holomorphic function. Then the function
n−1
(∆n+1 ) 2 f (x0 + x)

is monogenic.

The authors in [171] proved a similar result for n even. Now, we present another
method of extending a function on Rn into a monogenic one. This is due to
[132, 172]

Theorem 3.4.2. Let f ∈ L2 (Rn , Cn , dV (x) . Then f can be monogenically ex-

tended to a monogenic function F on Rn+1 with the estimate

|F (x0 , x)| ≤ ceR|(x0 ,x)|

if and only if
Supp(fb) ⊂ B(0, R)

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 44

where
B(0, R) = {x ∈ Rn : |x| < R}
In this case we have
Z
1
F (x0 , x) = e(x0 , x, ξ)fb(ξ)dξ,
(2π)n Rn
where

e(x0 , x, ξ) = e+ (x0 , x, ξ) + e− (x0 , x, ξ), e± (x0 , x, ξ) = eix·ξ e∓x0 |ξ| χ± (ξ)

and
ξ
 
1
χ± (ξ) = 1+i
2 |ξ|

Proof. [173, Page 118].

Now, we give some results about integration of Clifford-valued function which are
similar to those that are well known in the theory of complex-valued functions as
a Stokes formula and a Cauchy representation formula.

3.4.3 Cauchy-Clifford Integral Formula

In the sequel, Ω is an open subset of Rn , U ⊂ Ω a compact and orientable piecewise

differentiable bounded domain in Rn with lipschitzian boundary ∂U . The surface
element on ∂U will be
n
X
dσx = (−1)j ej dx[j]
j=1
where dx[j] = dx1 ∧ dx2 ∧ · · · dxj−1 ∧ dxj+1 ∧ · · · ∧ dxn (we omit the jth term). If
we call η(x) the unit outward pointing normal-vector at x ∈ ∂U , then

dσx = η(x)dS(x)

with dS(x) being the surface element on ∂U .

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 45

Theorem 3.4.3 – Clifford-Stokes Theorem . Let f, g ∈ C 1 (Ω) then for U ⊂ Ω

Z Z
f (x)dσx g(x) = [(f ∂x )g + f (∂x g)]dV (x) (3.18)
∂U U
In particular, if f ≡ 1 then
Z Z
dσx g(x) = ∂x g(x)dV (x)
∂U U
which relates the values of a function in a domain to its values on its boundary of
the given domain. Let f be left-monogenic and g right-monogenic on an open set
Ω ⊂Rn then
Z
g(x)dσx f (x) = 0
∂U
for any U ⊂ Ω.

As a consequence we have.

Corollary 3.13. Suppose that g is a right monogenic function on Ω ⊂ Rn then

foe every subset U of Ω we have
Z
g(x)dσx = 0
∂U

Theorem 3.4.4. Let f left-monogenic and g right-monogenic on an open U ⊂ Ω

(as in theorem 3.4.3) and let y ∈ U then
Z
1
f (y) = E(x − y)η(x)f (x)dσx
an ∂U
and
Z
1
g(y) = g(x)η(x)E(x − y)dσx .
an ∂U

Proof. We only prove the result for the left-monogenic function f . We consider the
sphere S n−1 (y, r) for r > 0 chosen small enough such that the disc whose boundary

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 46

is S n−1 (y, r) is included in U . Applying formula (3.18) on the kernel E and f we

get
Z Z
E(x − y)η(x)f (x)dσx = E(x − y)η(x)f (x)dσx .
∂U S n−1 (y,r)
The sphere S n−1 (y, r) being of curvature 1, the outward pointing normal-vector
η at x ∈ S n−1 (y, r) is given by (see [185, Page 231])
x y−x y−x
η(x) = = =
|x| x−y r
so on the surface of S n−1 (y, r)
x−yy−x r2 1
E(x − y)η(x) = n
= n+1
= n−1
r r r r
this gives us
Z Z
1
E(x − y)η(x)f (x)dσx = n−1
f (x)dσx
S n−1 (y,r) S n−1 (y,r) r

f (x) − f (y) + f (y)

Z
= dσx
S n−1 (y,r) rn−1
f (x) − f (y)
Z Z
= dσx + f (y)dσx
S n−1 (y,r) rn−1 S n−1 (y,r)

f (x) − f (y)
Z Z
= n−1 dσx + f (y) dσx
S n−1 (y,r) x − y S n−1 (y,r)

f (x) − f (y) f (y)

Z Z
1
= dσx + dσx
an S n−1 (y,r) x − y n−1 an S n−1
f (x) − f (y)
Z
= n−1 dσx + an f (y).
S n−1 (y,r) x − y
By continuity we have
f (x) − f (y)
Z
lim dσx = 0.
r→0 S n−1 (y,r) x − y n−1

Finally, we have f (y) = a1n ∂U E(x − y)η(x)f (x)dσx . By a similar procedure we

R

prove the result for g.

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 47

Theorem 3.4.5 – Mean Value Theorem. Let y ∈ U and R > 0 and consider
D(y, R), the disk of centre y and radius R such that D(y, R) ⊂ U then
Z
1
f (y) = n f (x)dV (x).
R an D(y,R)

Proof. Applying theorem (3.4.4) on D(y, R) we obtain

Z
1
f (y) = E(x − y)η(x)f (x)dσx
an ∂D(y,R)
Z
1
= E(x − y)η(x)f (x)dσx
an ∂D(y,R)
Z
1 1
= n−1
f (x)dσx .
an ∂D(y,R) r
Integrating by respect to r will give us
Z
n 1
f (y)R = f (x)dV (x).
an D(y,R)
So that
Z
1
f (y) = f (x)dV (x).
an Rn D(y,R)

3.5 Clifford-Fourier Transform

In this section we propose to review some basic concepts of the Clifford-Fourier

transform. Fore more details we may refer to [83] and [118]. Recall that the
classical Fourier transom can be seen as the operator exponential (see [162, 9, 26,
195, 75, 182, 204, 76])
∞
 π  X 1  π k k
F = exp −i H = −i H (3.19)
2 k=0
k! 2
where H is the scalar-valued operator
−1
∆n + x2 + n (3.20)


H=
2

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 48

called Hermite operator.

An eigenfunction for the n-dimensional Fourier transform is given by the Gaus-
sian function x 7−→ G(x) = e−|x| which satisfies
2 /2

G(ξ)
b = G(ξ).

Proposition 3.14. The two operators H and exp(−i π2 H) are Fourier invariant
in the sense that
[) = H(fb)
H(f
and that
\ π π
exp(−i H)(f ) = exp(−i H)(fb).
2 2

Proof. We have
−1
Z
1
[)(ξ) =
H(f n (∆n f (x) + x2 f (x) + nf (x))e−ix·ξ dV (x)
(2π) Rn 2
2

−1 1
Z Z
−ix·ξ 1
= [ n
2
∂ f (x)e dV (x) + n x2 f (x)e−ix·ξ dV (x)
2 (2π) 2 Rn x (2π) 2 Rn
Z
1
+n n f (x)e−ix·ξ dV (x)]
(2π) 2 Rn
−1 h 2 b i
= ξ f (ξ) + ∂ξ2 fb(ξ) + nfb(ξ)
2
= H(fb)(ξ).

For the second assertion we have

Z
\ π 1 π
exp(−i H)(f )(ξ) = n exp(−i H)(f )(x)e−ix·ξ dV (x)
2 (2π) 2 Rn 2
Z (X ∞
k )
1 −i π2
= n Hk (f )(x) e−ix·ξ dV (x)
(2π) Rn k=0
2 k!
∞
k 
−i π2
Z 
X 1 k −ix·ξ
= n H (f )(x)e dV (x)
k=0
k! (2π) 2 Rn
∞
k
X −i π2 \
= Hk (f )(ξ).
k=0
k!

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 49

Lets calculate H
\ k (f ). We have from the first assertion :

\
H \
k (f ) = H {H k−1 (f )}

= HH\
k−1 (f )

\
= HH {H k−2 (f )}

= H2 H\
k−2 (f )

..
.

= Hk H
\ 0 (f )

= Hk fb.

This gives us
∞
k
\ π X −i π
exp(−i H)(f )(ξ) = 2
Hk fb(ξ)
2 k=0
k!
π
= exp(−i Hfb).
2

The idea behind the extension of the classical Fourier transform to Clifford
algebra-valued one resides in the generalization of the Hermite operator into a
multivector-valued one. We use the factorization of the Laplace operator by the
Dirac operator to obtain two new Clifford-valued operators that factorize the Her-
mite operator. This method has been developed in [35, 66, 30, Ch. 12]. For that
purpose we introduce the following operators (see [32])
1
O1 = (∂x − x) (∂x + x)
2
1
O2 = (∂x + x) (∂x − x)
2
This operators has the following properties

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 50

Proposition 3.15.
O1 = H + Γx
O2 = H − Γx + n
= 2 H + n2


O1 + O2
= 2 Γx − n2


O1 − O2

Proof. We have that Γx = −1

2
(x∂x − ∂x x − n). So
1
O1 = (∂x − x) (∂x + x)
2
1 2
∂x + ∂x x − x∂x − x2


=
2
1 2
1
= ∂x − x2 − n − (x∂x − ∂x x − n)
2 2
= H + Γx .

And
1
O2 = (∂x + x) (∂x − x)
2
1 2
∂x − x2 − n + x∂x − ∂x x + n


=
2
1
= H + (x∂x − ∂x x + 2n − n)
2
= H − Γx + n.

We have also

O1 + O2 = H + Γx + H − Γx + n
 n
=2 H+
2
O1 − O2 = H + Γx − H + Γx − n
 n
= 2 Γx − .
2

For the two operators O1 and O2 to be used in the definition of the new Fourier

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 51

transform in the Clifford algebra’s setting, they have to be Fourier-invariant. We

have from proposition (3.14)
[) = H(fb).
H(f
So we have just to prove that
\
Γ x (f ) = Γξ (f ).
b

Indeed, formally we can write

\ −1 \
Γ x (f ) = (x∂x − ∂x x − n)
2
−1  [ [ 
= x∂x f − ∂x xf − nfb
2
−1  d 
= ∂ξ ∂x f − ξ xf − nf
c b
2
−1  
= ∂ξ ξ − ξ∂ξ − n fb
2
= Γξ (fb).

As 

O

1 = H + Γx

O2
 = H − Γx + n
then 

O

1f
d = Hf
c +Γ
dx f = O1 f
b

O
 c2 = H{c −Γ dx f + nf = O2 f .
b b
Now, we are able to define the Clifford-Fourier transform

Definition 3.16. The Clifford-Fourier transform is the pair of exponential oper-

ators
π π
F+ = exp(−i H+ ) and F− = exp(−i H− ).
2 2
If we want that the classical Fourier transform F to be the harmonic average
of the couple F+ and F−
F 2 = F+ F−

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 52

and as F = exp(−i π2 H) we have to choose H+ and H− to satisfy

1
H = (H+ + H− ).
2
But having O1 + O2 = 2H + n we get

H+ + H− = O1 + O2 − n.

So we set 
n

H

+ = O1 − 2

n

H−
 = O2 − 2
hence 
= H + Γx − n2
  
H

+
(3.21)
= H − Γx − n2 .
  
H−


One alternative for the operators H+ and H− is given in [66, Sec. 12.3.2] and in
[35, Def. 4.2] 

H+
 = O1

H−
 = O2 − n
and in this case we have 

H

+ = H + Γx

H−
 = H − Γx .
Keeping our operators as in (3.21) we obtain in terms of the exponential operator
π
F+ = exp(−i H+ )
2
 π h n i
= exp −i H + Γx −
2 2 (3.22)
 π n   π 
= exp −i Γx − exp −i H
2 2 2
 π n 
= exp −i Γx − F
2 2
and in the same way
 π n 
F− = exp i Γx − F. (3.23)
2 2

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 53

We obtain an integral representation for the new Clifford-Fourier transform:

Z
1  π n  −ix·ξ
F+ [f ](ξ) = n exp −i Γ ξ − e f (x)dV (x)
(2π) 2 Rn 2 2

and
Z
1  π n  −ix·ξ
F− [f ](ξ) = n exp i Γξ − e f (x)dV (x). (3.24)
(2π) 2 Rn 2 2
We have defined the Clifford-Fourier transform as a pair of operators satisfying

F 2 = F+ F−

so we can write
1 1
F = F+2 F−2
where the square root is the Fractional Fourier Transform (see [170] and [66,
Ch.11]) given by
1  π 
F+ = exp −i H+
2

4
1  π 
F−2 = exp −i H− .
4
We can factorize the classical Fourier transform as the products
 π   π 
F = exp −i H+ exp −i H−
4 4
 π   π 
= exp −i H− exp −i H+
4 4
This transform can be inverted. Using (3.22) and (3.23) we have
 π 
−1
F+ = exp i H+
2
and
 π 
F−−1 = exp i H−
2

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 54

which gives
 π n  −1
F+−1 = exp i Γx − F
2 2
and
 π n  −1
F−−1 = exp −i Γx − F
2 2
The Clifford-Fourier transform has properties similar to those of the classical
Fourier transform.

Proposition 3.17. For two Clifford algebra-valued functions f and g and a, b ∈ Cn

we have

F+ [f a + gb] = F+ [f ] a + F+ [g] b

and

F− [f a + gb] = F− [f ] a + F− [g] b.

Proof. This results from (3.24).

Proposition 3.18. For λ > 0 we have

1 ξ
F+ [f (λ•)] (ξ) = F + [f ] ( )
λn λ
and
1 ξ
F− [f (λ•)] (ξ) = F − [f ] ( ).
λn λ

Proof. First, let us prove that

1 ξ
F [f (λ•)] (ξ) = F [f ] ( ).
λn λ
We have
Z
1
F [f (λ•)] (ξ) = n f (λx)e−ix·ξ dV (x).
(2π) 2 Rn

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 55

If we put y = λx then dV (x) = 1

λn
(y), so
dV
Z
1 1 ξ
F [f (λ•)] (ξ) = n n f (y)e−ix· λ dV (y)
λ (2π) 2 Rn
1 ξ
= n fb( ).
λ λ
Then from (3.22) and (3.23)
 π n  1 b ξ
F+ [f (λ•)] (ξ) = exp −i Γξ − f( )
2 2 λn λ
and
 π n  1 b ξ
F− [f (λ•)] (ξ) = exp i Γξ − f ( ).
2 2 λn λ
Next, we prove that Γx = Γ λx . We have
x
Γ λx = − ∧ ∂ λx ,
λ
if we put y = λ ⇐⇒ xi = λyi and so
x

n n
X ∂ X ∂
∂y = ei = ei λ = λ∂x .
i=1
∂yi i=1
∂xi
So finally,
x x
Γ λx = − ∧ ∂ λx = − ∧ λ∂x = Γx .
λ λ

We obtain the formulas

 π n  1 b ξ 1 ξ
F+ [f (λ•)] (ξ) = exp −i Γξ − f ( ) = F + [f ] ( )
2 λ 2 λn λ λn λ
and
 π n  1 b ξ 1 ξ
F− [f (λ•)] (ξ) = exp i Γξ − n
f ( ) = n F− [f ] ( ).
2 λ 2 λ λ λ λ

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 56

Proposition 3.19. For all Clifford-valued function f we have

F+ [•f (•)] (ξ) = −(−i)n ∂ξ F+ [f ] (ξ)

and

F− [•f (•)] (ξ) = (i)n ∂ξ F− [f ] (ξ)

Proof. We know from the proprieties of the classical Fourier transform that

F [•f (•)] (ξ) = i∂ξ F [f ] (ξ).

Then
 π n 
F+ [•f (•)] (ξ) = exp −i Γξ − F [•f (•)] (ξ)
2 2
 π n 
= i exp −i Γξ − ∂ξ F [f ] (ξ)
(∞ 2 2 )
X (−i π )k  n k
=i 2
Γξ − ∂ξ F [f ] (ξ). (3.25)
k=0
k! 2
We know from [162, Thm 2.4] that

∂ξ Γξ + Γξ ∂ξ = (n − 1)∂ξ .

Then
h i
Γξ ∂ξ = ∂ξ n − 1 − Γξ ,

and
n n
∂ξ = ∂ξ .
2 2
So
 
 n n−2
Γξ − ∂ξ = ∂ξ − Γξ .
2 2
 
Applying Γξ − n
2
gives us
 
 n 
 n  n n−2
Γξ − Γξ − ∂ξ = Γξ − ∂ξ − Γξ
2 2 2 2
  
n−2 n−2
= ∂ξ − Γξ − Γξ .
2 2

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 57

Repeating this k times we obtain

 k
 n k n−2
Γξ − ∂ξ = ∂ξ − Γξ .
2 2
Then (3.25) becomes
(∞ )
X (−i π )k  n k
2
F+ [•f (•)] (ξ) = i Γξ − ∂ξ F [f ] (ξ)
k! 2
( k=0
∞
)
X (−i π2 )k  n k
=i Γξ − ∂ξ F [f ] (ξ)
k! 2
( k=0
∞ k )
(−i π2 )k

X n−2
=i ∂ξ − Γξ F [f ] (ξ)
k=0
k! 2
 π n 
= i∂ξ exp i Γξ − + 1 F [f ] (ξ)
2 2
 π  π n 
= i exp i ∂ξ exp i Γξ − F [f ] (ξ)
2 2 2
= −∂ξ F− [f ] (ξ).

The same yields for F− and we have

F− [•f (•)] (ξ) = ∂ξ F+ [f ] (ξ).

These results may be generalized to

F+ x2k f (ξ) = (−1)k ∂ξ2k F+ [f ] (ξ)

 

and

F+ x2k+1 f (ξ) = −(−1)k ∂ξ2k+1 F− [f ] (ξ)

 

By applying the recurrence rule.

The explicit form of the kernel of (3.24) is a difficult problem. In the case

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 58

where n = 2 it has been given in [34, Sec. 4]

Z
1
F+ [f ](ξ) = exp(ξ ∧ x)f (x)dV (x)
2π R2
and
Z
1
F− [f ](ξ) = exp(x ∧ ξ)f (x)dV (x).
2π R2
For even dimensions, a first attempt was done in [36], for example for n = 4 we
have in the term of the Bessel function [9, Sec. 2.2.2]
(ξ ∧ x)
r  
π −1/2
K+ (x, ξ) = |x ∧ ξ| (1 + x · ξ)J1/2 (|x ∧ ξ|) + J3/2 (|x ∧ ξ|)(x · ξ)
2 |x ∧ ξ|

and
(ξ ∧ x)
r  
π −1/2
K− (x, ξ) = |x ∧ ξ| (1 − x · ξ)J1/2 (|x ∧ ξ|) + J3/2 (|x ∧ ξ|)(x · ξ)
2 |x ∧ ξ|
and for n = 6 we have
 
pπ −1/2 (x·ξ)2
K+ (x, ξ) = 2 {|x ∧ ξ| J1/2 (|x ∧ ξ|) + |x∧ξ| J3/2 (|x ∧ ξ|)
 2

+|x ∧ ξ|−3/2 2(x · ξ)J3/2 (|x ∧ ξ|) + (ξ ∧ x)J3/2 (|x ∧ ξ|) + (x·ξ)
|x∧ξ|
(ξ ∧ x)J 5/2 (|x ∧ ξ|) }

and
 
pπ −1/2 (x·ξ)2
K− (x, ξ) = 2
{|x ∧ ξ| J1/2 (|x ∧ ξ|) + J (|x
|x∧ξ| 3/2
∧ ξ|)
 
(x·ξ)2
−|x ∧ ξ|−3/2 2(x · ξ)J3/2 (|x ∧ ξ|) + (ξ ∧ x)J3/2 (|x ∧ ξ|) + |x∧ξ|
(ξ ∧ x)J5/2 (|x ∧ ξ|) }.
In [65], the authors found a general expression for the kernels for all even dimen-
sions.

3.6 Clifford Wavelet Transform

We introduce the concept of the Clifford-wavelet transform and some of its impor-
tant properties to be used later. Besides of the translation and dilation, we will use
rotation using the action of the spin group which is a double-cover of the special
orthogonal group in Rn . In this context, a function ψ ∈ L1 ∩L2 (Rn , Rn , dV (x)) will

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 59

be considered as a Clifford mother wavelet. To join the admissibility assumptions

in the case of wavelets on R, here-also we define

Definition 3.20 (Clifford Wavelet). Let ψ ∈ L1 ∩ L2 (Rn , Rn , dV (x)) such that

h i†
• ψ(ξ)
b ψ(ξ)
b is scalar.
h i†
Z ψ(ξ) ψ(ξ)
b b
• The admissibility condition : Aψ = (2π)n dV (ξ) < ∞.
Rn |ξ|n
The function ψ is called an admissible Clifford mother wavelet and Aψ is its ad-
missibility constant. We can see that this condition implies that
Z
ψ(0) = 0 ⇐⇒
b ψ(x)dV (x) = 0.
Rn
Starting with an admissible wavelet, we create a whole set of daughter wavelets by
translating, dilating and Spin-rotating the mother wavelet.

Example 3.21. In [44], the authors defined a generalization of the n-dimensional

Mexican Hat wavelet [144] as the CK-extension of
1
ψ(x) = exp( x2 )Hn (x)
2
x2
= (−1)n ∂x exp( )
2
where Hn are the radial Hermite polynomials given in [42, 44, 13, 14]. Its Fourier
transform is
ξ2
b = (2π) n2 (−i)n ξ n exp(
ψ(ξ) )
2
and so it is an admissible Clifford algebra-valued mother wavelet since
2
Z ψ(ξ)
b
Aψ = (2π)n n dV (ξ)
Rn ξ
2
ξ2
Z
2n
= (2π) exp( ) dV (ξ)
Rn 2

< ∞.

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 60

Example 3.22. From [9] we know that for 0 < t < 1−n−2α
2
we have
Z

α
xk Gn,t,α+t (x) 1 + |x|2 dV (x) = 0
Rn
where Gn,t,α+t are the Clifford-Gegenbauer polynomials and specially
Z Z
2 α

α+t
(−1)t ∂xt 1 + |x|2


Gn,t,α+t (x) 1 + |x| dV (x) = dV (x)
Rn Rn

=0

and this is just the admissibility condition 3.6. So the functions x 7−→ ψn,t,α (x) =
α+t
(−1)t ∂xt (1 + |x|2 ) can be taken as mother wavelets. We call them Clifford-Gegenbauer Wavelets.

Example 3.23. In [28] the authors defined the so called Clifford-Laguerre Wavelets
as
(−1)k ∂xk exp(−|x|)|x|α+2k P +

for α > −n, l > 0 and P + (x) = 21 (1 + i |x|

x
) is the Clifford-Heaviside function.

Definition 3.24. For (a, b, s) ∈ R+ × Rn × Spin(n), we denote

1 s(x − b)s
ψ a,b,s (x) = n sψ( )s.
a2 a
It holds in fact that these copies are also admissible and that
an/2
Aψa,b,s = Aψ < ∞.
(2π)n
Now we will see that indeed the family of wavelets ψ a,b,s can be used to analyse
or decompose square integrable Clifford-valued functions, for that we have

Proposition 3.25. The set ψ a,b,s : a > 0, b ∈ Rn , s ∈ Spin(n) is dense in L2 (Rn , Rn , dV (x)).

Proof. Let f be an analysed function such that

< ψ a,b,s , f >L2 (Rn ,Rn ,dV (x)) = 0, ∀a > 0, b ∈ Rn and s ∈ Spin(n).

We shall prove that f = 0. Using the Parseval identity of the Clifford-Fourier

transform : for f, g ∈ L1 ∩ L2 (Rn , C, dV (x)
D E
hf, giL2 (Rn ,C,dV (x) = fb, gb . (3.26)
L2 (Rn ,C,dV (ξ)

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 61

we obtain

< ψ a,b,s , f >L2 (Rn ,Rn ,dV (x)) =< ψ

[ a,b,s , fb > 2 n
L (R ,Rn ,dV (x)) = 0.

Since
Z h i†
n
[
<ψ a,b,s , fb > 2 n
L (R ,Rn ,dV (x)) = a 2 eib·ξ s ψ(asξs)
b sfb(ξ)dV (ξ) = 0,
Rn
then necessarily
h i†
s ψ(asξs) sfb(ξ) = 0, ∀ξ ∈ Rn .
b

Recall now that for a fixed ξ 6= 0 in Rn (see [66], pages 48 and 49)

asξs, a > 0 and s ∈ Spin(n) = Rn .



It results that
fb = 0 and so f = 0.

As for the real case, we define the Clifford Wavelet Transform as the projection
of the signal f on the set of admissible Clifford wavelets

Definition 3.26. The Clifford-wavelet transform of a function f ∈ L2 (Rn , Rn , dV (x))

with respect to an admissible mother wavelet ψ is 1

Tψ [f ] (a, b, s) =< ψ a,b,s , f >L2 (Rn ,Rn ,dV (x))

Z
 a,b,s †
= ψ (x) f (x)dV (x) (3.27)
Rn
  †
s(x − b)s
Z
1
= n s ψ sf (x)dV (x). (3.28)
a 2 Rn a

3.6.1 Proprieties of the Clifford Wavelet Transform

It has the following covariance proprieties

1 †
We used the complex Clifford conjugation (see (3.3)) but we could have used (3.3) too.

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 62

• Covariance by translation

Tψ [f (• − c] (a, b, s) = Tψ [f ] (a, b − c, s).

• Covariance by Dilation
 
1 • a b
Tψ n f( ) (a, b, s) = Tψ [f ] ( , , s).
λ2 λ λ λ

• Covariance by Spin rotation

Tψ [Lt f ] (a, b, s) = tTψ [f ] (a, tbt, ts)t.

Proof. We have
  †
s(x − b)s
Z
1
Tψ [f (• − c] (a, b, s) = n s ψ sf (x − c)dV (x).
a 2 Rn a
Put y = x − c =⇒ x − b = y − (b − c) and dV (x) = dV (y) so
s(y − (b − c))s †
Z   
1
Tψ [f (• − c] (a, b, s) = n s ψ sf (y)dV (y)
a 2 Rn a
= Tψ [f ] (a, b − c, s).

For the covariance by dilation

    †
• s(x − b)s
Z
1 1 1 x
Tψ n f( ) (a, b, s) = n s ψ s n f ( )dV (x),
λ2 λ a 2 Rn a λ2 λ
y
put y = λ =⇒ x = λy, x − b = λy − b and dV (x) = λ dV (y) so
n

s(λy − b)s †
    
•
Z
1 λ n
Tψ n f( ) (a, b, s) = ( ) 2 s ψ sf (y)dV (y)
λ2 λ a Rn a
" !#†
s(y − λb )s
Z
1
= a n s ψ a sf (y)dV (y)
( λ ) 2 Rn λ
a b
= Tψ [f ] ( , , s).
λ λ
Finally, for the case of the Spin rotation first recall the action of the Spin
group (3.4.1.2)
Ls : f (x) → sf (s̄xs)s

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 63

Then we have
  †
s(x − b)s
Z
1
Tψ [Lt f ] (a, b, s) = n s ψ stf (txt)tdV (x)
a 2 Rn a
put y = txt =⇒ x = tyt and dV (x) = tdV (y)t. So
" !#†
s(tyt − b)s
Z
1
Tψ [Lt f ] (a, b, s) = n s ψ stf (tyt)tdV (y),
a 2 Rn a
having tt = tt = 1, then
" !#†
s(tyt − b)s
Z
1
Tψ [Lt f ] (a, b, s) = n s ψ stf (tyt)tdV (y)
a2Rn a
" !#†
st(y − tbt)ts
Z
1 
=t n ts ψ {st} f (tyt)tdV (y)t
a 2 Rn a

= tTψ [f ] (a, tbt, ts)t

Definition 3.27. (Inner product relation) Let Hψ = {Tψ [f ] , f ∈ L2 (Rn , Rn , dV (x))}

be the image of L2 (Rn , Rn , dV (x)) relatively to the operator Tψ . We define the
inner product for two square integrable functions f and g by
Z Z Z
1 da
[Tψ [f ] , Tψ [g]] = (Tψ [f ] (a, b, s))† Tψ [g] (a, b, s) n+1 dV (b)ds,
Aψ a
Spin(n) Rn R+
where ds stands for the Haar measure on Spin(n).

Proposition 3.28. The range of an isometry T : H −→ H0 is a closed subspace

of H0 .

Knowing all that, we can now introduce a result that permits us to invert the
Clifford wavelet transform

Proposition 3.29. Tψ : L2 (Rn , Rn , dV (x)) −→ Hψ is an isometry.

Proof. We have to show that

[Tψ [f ] , Tψ [g]] =< f, g >L2 (Rn ,Rn ,dV (x)) . (3.29)

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 64

Put
h i† 
Φψ (a, s, ξ) [f ] (−b) = ψ(asξs) sf (ξ) (−b)
b b

and similarly
h i† 
Φψ (a, s, ξ) [g] (−b) = ψ(asξs) sb
b g (ξ) (−b).
We obtain
n n
Tψ [f ] (a, b, s) = a 2 s(2π) 2 Φψ\
(a, ξ, s) [f ] (−b)
and
n n
Tψ [g] (a, b, s) = a 2 s(2π) 2 Φψ\
(a, ξ, s) [g] (−b).
Applying Parseval formula we get
D E
\
Φψ (a, \
•, s) [f ], Φψ (a, •, s) [g] = hΦψ (a, •, s) [f ] , Φψ (a, •, s) [g]i .

And so
Z Z Z 
1 † da
[Tψ [f ] , Tψ [g]] = (Φψ (a, ξ, s) [f ] (ξ)) Φψ (a, ξ, s) [g] (ξ)dV (b) ds
(2π)n Aψ Rn a
Spin(n) R+
Z Z (Z  h †
† h
†
)
1 i i da
= n
( ψ(asξs)
b sfb(ξ) ψ(asξs)
b sb
g (ξ)dV (ξ) ds
(2π) Aψ Rn a
Spin(n) R+
Z Z Z h i† i† 
1 h da
= n
f (ξ) sψ(asξs) ψ(asξs) sb
b b b g (ξ)dV (ξ) ds
(2π) Aψ Spin(n) Rn a
R+
 
Z h †
Z Z † da
1 i  h i 
= f
b(ξ) s ψ(asξs)
b ψ(asξs)
b s ds gb(ξ)dV (ξ).
(2π)n Aψ Rn  Spin(n) a 
R+
Observing now that
Aψ
Z Z h i† da
sψ(asξs)
b ψ(asξs)
b s ds = , (3.30)
Spin(n) a (2π)n
R+

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 65

we get immediately
Z h i†
[Tψ [f ] , Tψ [g]] = fb(ξ) gb(ξ)dV (ξ)
Rn

=< fb, gb >

=< f, g > . (3.31)

We can also say that it is an isometry between the two spaces of integrable
functions L2 (Rn , Rn , dV (x)) and L2 (R+ × Rn × Spin(n), A−1
ψ a
−(n+1)
dadV (b)ds).
This is an analogue to Parseval’s formula and as a result we have also a Plancherel’s
formula
Z Z Z
da
|Tψ [f ] (a, b, s)|2 n+1
dV (b)ds = Aψ kf k22 . (3.32)
a
Spin(n) Rn R+
As a result of the last Proposition and as in the real case, we have here a Clifford-
wavelet reconstruction formula.

Proposition 3.30. For all f ∈ L2 (Rn , Rn , dV (x)) we have

Z Z Z
1 da
f (x) = ψ a,b,s (x)Tψ [f ] (a, b, s) n+1 dV (b)ds
Aψ a
Spin(n) Rn R+
which holds weakly in L2 (Rn , Rn , dV (x)) .

In other words, the Clifford wavelet transform decomposes the signal f in terms
of the analysing wavelets ψ a,b,s with coefficients Tψ [f ].

Proof. Let f and g two square integrable Clifford-valued functions with Clifford
wavelet transforms (with respect to a mother wavelet ψ) Tψ [f ] and Tψ [g] respec-

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 66

tively. We have from proposition (3.29) and using (3.28)

Z Z Z
1 da
< f, g >L2 = (Tψ [f ] (a, b, s))† Tψ [g] (a, b, s) n+1 dV (b)ds
Aψ a
Spin(n) Rn R+
Z Z Z
1 da
= [Tψ [f ] (a, b, s)] † Tψ [g] (a, b, s) n+1 dV (b)ds
Aψ a
Spin(n) Rn R+
Z Z Z
1 da
= < ψ a,b,s , f >†L2 Tψ [g](a, b, s) n+1 dV (b)ds
Aψ Spin(n) a
Rn R+
Z Z Z
1 da
= < f, ψ a,b,s >L2 Tψ [g](a, b, s) n+1 dV (b)ds
Aψ Spin(n) a
Rn R+
Z Z Z
1 da
=< f, ψ a,b,s (x)Tψ [g](a, b, s) n+1 dV (b)ds >L2
Aψ Spin(n) a
Rn R+
Then
Z Z Z
1 da
g(x) = ψ a,b,s (x)Tψ [g](a, b, s) dV (b)ds
Aψ Spin(n) an+1
Rn R+
where the equality is understood in the L2 -sense.

Now, we present a result similar to the reproducing kernel given by (2.2.3).

Theorem 3.6.1. A function F ∈ L2 R+ × Rn × Spin(n), A−1 −(n+1)

ψ a dadV (b)ds
is the Clifford wavelet transform of a square integrable function f iff
Z Z Z +∞  †
1 dã
F (a, b, s) = Kψ (a, b, s; ã, b̃, s̃) F (ã, b̃, s̃) n+1 dV (b̃)ds̃
Cψ Spin(n) Rn 0 ã
where Kψ (a, b, s; ã, b̃, s̃) = Tψ ψ (ã, b̃, s̃) =< ψ ã,b̃,s̃ , ψ a,b,s > is the reproducing
 a,b,s 

kernel.

Thesis in Mathematics Hicham BANOUH

Clifford Algebra/Analysis Toolkit 67

3.7 Conclusion

In this part, we gave an introduction to the theory of Clifford algebras and we

saw that modulo a condition similar to the Cauchy-Riemann equations in the
complex plan, we generalized the notion of holomorphic and harmonic functions
to n-dimensions. Also, we presented the extension of the well known Fourier
transform to Clifford algebra settings and presented the main topic of this thesis
namely the Clifford wavelet transform.

Thesis in Mathematics Hicham BANOUH

Chapter 4
Clifford Wavelet Uncertainty Principle

4.1 Introduction

The uncertainty principle is central for information processing and quantum physics.
In the Clifford algebras framework, the uncertainty principle provides us data
about the way a multivector valued function and its Clifford-Fourier transform are
related. We will see that the uncertainty principle for the Clifford wavelet trans-
form establishes a lower bound of the product of the variances of Clifford wavelet
transform of a square integrable multivector-valued function and its Clifford-Fourier
transform.

4.2 Old uncertainty principle revisited

First, we recall some results concerning the classical Heisenberg uncertainty prin-
ciple. For more backgrounds on the uncertainty principle, its variants, Fourier and
wavelet transforms on the Euclidean space Rn the readers may be referred also to
[122], [166], [191] and [201]. Mathematically, the uncertainty principle states that a

Thesis in Mathematics Hicham BANOUH

Clifford Wavelet Uncertainty Principle 69

non-zero function and its Fourier transform cannot both be sharply localized. The
next theorem formally summarizes the Heisenberg’s Uncertainty Principle.

Theorem 4.2.1. Uncertainty Principle [212]

Let A and B be two self-adjoint operators on a Hilbert space X with domains
D(A) and D(B) respectively and [A, B] = AB − BA their commutator. Then
1
kAf k2 kBf k2 ≥ |< [A, B] f, f >| , ∀f ∈ D([A, B]). (4.1)
2

Proof. Let f ∈ D([A, B]). Then f ∈ D(A) ∩ D(B). Hence ,

| h[A, B] f, f i | = | hABf, f i − hBAf, f i |.

As A and B are two self-adjoints operators, we get

| h[A, B] f, f i | = | hAf, Bf i − hBf, Af i | = 2|Im {hAf, Bf i} |.

Applying next the Cauchy-Schwartz inequality, we obtain

| h[A, B] f, f i | ≤ 2kAf k2 kBf k2 .

Which reads as
1
kAf k2 kBf k2 ≥ |h[A, B] f, f i| . (4.2)
2
which completes the proof.

Lets apply theorem (4.2.1) to the case of the Fourier transform. We mainly review
the results of [83] and [118]. Let f ∈ L1 ∩ L2 (Rn , Rn , dV (x)) We define for k ∈
{1, 2, · · · , n} the two families of operators

Ak f (x) = xk f (x)

Bk f (x) = ∂xk f (x).

Thesis in Mathematics Hicham BANOUH

Clifford Wavelet Uncertainty Principle 70

Using the fact that

 Z 
1 ix·ξ b
∂xk f (x) = ∂xk n e f (ξ)dV (ξ)
(2π) 2 Rn
Z
1 n
ix·ξ b
o
= n ∂x e f (ξ) dV (ξ)
(2π) 2 Rn k
Z
1 ix·ξ
n o
= n e iξk f (ξ) dV (ξ)
b
(2π) 2 Rn
h i
= F −1 iξk fb(•) (x)

whence

xk f (ξ) = iξk f (ξ).

∂d b

and so we have the norm equality

∂d
xk f = ξk fb .
2 2

In a similar way we can see that

Z
1
x\
k f (•)(ξ) = n xk f (x)e−ix·ξ dV (x)
(2π) Rn
2
Z
1
f (x)i∂ξk e−ix·ξ dV (x)

= n
(2π) 2 Rn
= i∂ξk F [f ] (ξ).

We have Ak f (x) = xk f (x) and Bk f (x) = ∂xk f (x). So

kAk f k2 = kxk f k2

and
kBk f k2 = ξk fˆ
2
Applying theorem (4.2.1) to both operators Ak and Bk yields :
1
kAk f k2 kBk f k2 ≥ |< [Ak , Bk ] f, f >| .
2

Thesis in Mathematics Hicham BANOUH

Clifford Wavelet Uncertainty Principle 71

As the commutator

[Ak , Bk ] f = Ak Bk f − Bk Ak f

= xk ∂xk f (x) − ∂xk {xk f (x)}

= xk ∂xk f (x) − f (x) − xk ∂xk f (x)

= −f (x)

we obtain finally
1
kxk f k2 ξk fb ≥ kf k22 . (4.3)
2 2

Theorem 4.2.2. For A and B two symmetric operators on a Hilbert space H and
for f ∈ L2 (H, dx) we have
1
q
2
kAf k2 kBf k2 ≥ |h[A, B] f, f i|2 + [A, B]+ f, f (4.4)
2
where [A, B]+ = AB + BA is the anti-commutator.

Proof. See [190] and [80].

As for Theorem 4.3.1, the first step is to apply this result to the Fourier transform.
So we have

Theorem 4.2.3. [80] Let f ∈ L1 ∩ L2 (Rn , Rn , dV (x)) and define Ak f (x) = xk f (x)
and Bk f (x) = 1i ∂xk f (x) then
√
kxk f k2 ξk fb ≥ 2(kf k22 + |2 hxk ∂xk f, f i|).
2

Proof. We have by Theorem 4.2.2

1
q
2
kAk f k2 kBk f k2 ≥ |h[Ak , Bk ] f, f i|2 + [Ak , Bk ]+ f, f . (4.5)
2

Thesis in Mathematics Hicham BANOUH

Clifford Wavelet Uncertainty Principle 72

Observe next that

[Ak , Bk ] f (x) = Ak Bk f (x) − Bk Ak f (x)

1 1
= xk ∂xk f (x) − ∂x xk f (x)
i i k
1 1 1
= xk ∂xk f (x) − f (x) − xk ∂xk f (x)
i i i
1
= − f (x).
i
By the same way
1
[Ak , Bk ]+ f (x) = (2xk ∂xk f (x) + f (x)).
i
Substituting in (4.5) we get
1
q
kxk f k2 k∂xk f k2 ≥ kf k42 + |h2xk ∂xk f + f, f i|2 .
2
We know that

xk f (ξ) = iξk f (ξ).

∂d b

So
k∂xk f k2 = ∂d
xk f = ξk fb .
2 2
Then (4.5) becomes
1
q
kxk f k2 ξk fb ≥ kf k42 + |h2xk ∂xk f + f, f i|2
2 2q
1 2
= kf k42 + kf k22 + 2 hxk ∂xk f, f i
2q
1
≥ kf k42 + 4 |hxk ∂xk f, f i|2 .
2
q √ √
Since for a, b > 0, a +b then 2(a + b). So
2 2 a+b 1
≥ a 2 + b2 ≥
2 2 2
1
q
kxk f k2 ξk fb ≥ kf k42 + 4 |hxk ∂xk f, f i|2
2 2
√
≥ 2(kf k22 + |2 hxk ∂xk f, f i|).

Thesis in Mathematics Hicham BANOUH

Clifford Wavelet Uncertainty Principle 73

4.3 Clifford wavelet uncertainty principle

In this section, we establish the main result of the thesis : a new Heisenberg
uncertainty principle for the Clifford wavelet transform. This result have been
published in [19].

Theorem 4.3.1. Let ψ ∈ L2 (Rn , Rn , dV (x)) be an admissible Clifford mother

wavelet. Then for f ∈ L2 (Rn , Rn , dV (x)) the following inequality holds
Z Z  12 n
2 da (2π) 2 p
kbk Tψ [f ] (a, •, s)k2 n+1 ds ξk f ≥
b Aψ kf k22 ,
Spin(n) R + a 2 2
where k = 1, 2, · · · , n.

To prove this result we need the following lemma.

Lemma 4.1. Given ψ and f as in Theorem (4.3.1) then

Z Z Z
\ 2 da Aψ 2
|ξk Tψ [f ](a, ξ, s)| dV (ξ) ds = ξk f
b .
R+ Rn an+1 (2π)n 2
Spin(n)

Proof. As the daughter wavelet ψ a,b,s has the following Fourier expression
n
[
ψ a,b,s (ξ) = a 2 e−i<b,ξ> sψ(asξs)s,
b

we get
 h i† 
n
−1
Tψ [f ] (a, b, s) = a F
2 s ψ(as
b • s) sfb(•) (b)
and in the frequency domain
n
h i†
\
Tψ [f ](a, ξ, s) = a s ψ(asξs) sfb(ξ).
2 b (4.6)

Thesis in Mathematics Hicham BANOUH

Clifford Wavelet Uncertainty Principle 74

Therefore by (3.8)
Z
\ 2
|ξk Tψ [f ](a, ξ, s)| dV (ξ)

Rn
Z h i†
n
= |ξk a 2 s ψ(asξs)
b sfb(ξ)|2 dV (ξ)
Rn
Z  h i† † h i†
n n
= ξk a 2 s ψ(asξs) sf (ξ) ξk a 2 s ψ(asξs)
b b b sfb(ξ)dV (ξ)
Rn
Z h i†  h i† †
= ξk2 an s ψ(asξs)
b sfb(ξ) s ψ(asξs)
b sfb(ξ) dV (ξ)
Rn
Z h i† h i†
= ξk2 an s ψ(asξs) sf (ξ) f (ξ) sψ(asξs)sdV
b b b b (ξ)
Rn
Z h i† h i†
= ξk2 an ψ(asξs) f (ξ) f (ξ) ψ(asξs)dV
b b b b (ξ)
Rn
Z h i†  h i† 
= ξk2 an ψ(asξs)
b ψ(asξs)
b fb(ξ) fb(ξ) dV (ξ). (4.7)
Rn
Using (4.7) we obtain
Z Z Z
\ 2 da
|ξk T ψ [f ](a, ξ, s)| dV (ξ) n+1 ds
a
Spin(n) R+ Rn
Z Z Z h i†  i† 
2 n
h da
= ξk a ψ(asξs)
b ψ(asξs)
b fb(ξ) fb(ξ) dV (ξ) n+1 ds
a
Spin(n) R+ Rn
Z Z Z h i†  i† 
2
h da
= ξk ψ(asξs) ψ(asξs)
b b f (ξ) f (ξ)
b b dV (ξ) ds
a
Spin(n) R+ Rn
 h i† 
Z   Z Z ψ(asξs)
b ψ(asξs)
b 
  h i† 
2
= dads ξk fb(ξ) fb(ξ) dV (ξ).

 a 

Rn +
Spin(n) R
According to (3.30), we get finally
Z Z Z
\ 2 da Aψ 2
|ξk Tψ [f ](a, ξ, s)| dV (ξ) n+1 ds = ξk f
b .
a (2π)n 2
Spin(n) R+ Rn

Thesis in Mathematics Hicham BANOUH

Clifford Wavelet Uncertainty Principle 75

Proof. of Theorem (4.3.1). Using the inequality (4.3) and setting x = b ∈ Rn , we

obtain
1
\
kbk Tψ [f ] (a, •, s)k2 ξk Tψ [f ](a, •, s) ≥ kTψ [f ] (a, •, s)k22 .
2 2
Therefore
Z Z
\ da
kbk Tψ [f ] (a, •, s)k2 ξk T ψ [f ](a, •, s) ds
Spin(n) R+ 2 an+1
Z Z
1 da
≥ kTψ [f ] (a, •, s)k22 n+1 ds
2 Spin(n) R+ a
According to the Cauchy-Schwartz inequality (3.10), it follows that

Z Z
da 2 da
kbk Tψ [f ] (a, •, s)k22 ds × \
ξk Tψ [f ](a, •, s) ds
Spin(n)×R+ an+1 Spin(n)×R+ 2 an+1

Z !2
1 da
≥ |Tψ [f ] (a, b, s)|2 dV (b) n+1 ds . (4.8)
2 Spin(n)×R+ ×Rn a

Now, using Lemma 4.1 and the fact that the Clifford wavelet transform is an isometry,

we get by (3.32)

Z Z Z
da
|Tψ [f ] (a, b, s)|2 dV (b)ds = Aψ kf k22 .
an+1
Spin(n) Rn R+

The inequality (4.8) becomes

 1
2
Z Z  1
da Aψ 2 2 1
kbk Tψ [f ] (a, •, s)k22 ds ξk fb ≥ Aψ kf k22 . (4.9)
 

a n+1 (2π)n 2 2
Spin(n) R+

Hence, we obtain

 1
2
Z Z n
da (2π) 2 p
kbk Tψ [f ] (a, •, s)k22 n+1 ds ξk fb ≥ Aψ kf k22 .
 
a 2

2
Spin(n) R+

Thesis in Mathematics Hicham BANOUH

Clifford Wavelet Uncertainty Principle 76

4.4 A sharper Clifford wavelet uncertainty princi-

ple

In the present section we state and prove the second main result which concerns
a sharper formulation of the Clifford-wavelet uncertainty principle.

Theorem 4.4.1. [17, 18]Let ψ ∈ L2 (Rn , Rn , dV (x)) be an admissible Clifford

mother wavelet. Then for f ∈ L2 (Rn , Rn , dV (x)) the following inequality holds

Z Z
da 1
q n o
( kbk Tψ [f ] (a, •, s)k22 n+1
ds) 2 ξk fb ≥ 2n+1 π n Aψ kf k22 + 2 |hf1 , f2 i|
Spin(n) R+ a 2

where

1 da
R R R
ψ a,b,s (x)∂bk Tψ [f ] (a, b, s) an+1

f1 (x) =
 dV (b)ds
Aψ Spin(n) Rn R+

1 da
R R R
ψ a,b,s (x)bk Tψ [f ] (a, b, s) an+1

f2 (x) =
 dV (b)ds
Aψ Spin(n) Rn R+

Proof. From Theorem 4.2.3 we have

√
kxk f k2 ξk fb ≥ 2(kf k22 + |2 hxk ∂xk f, f i|).
2

We substitute f (•) by Tψ [f ] (a, •, s), then

√
kbk Tψ [f ] (a, •, s)k2 ξk Tψ [f\
] (a, •, s) ≥ 2(kTψ [f ] (a, •, s)k22
2

+ |2 hbk ∂bk Tψ [f ] (a, •, s), Tψ [f ] (a, •, s)i|).

Hence
Z Z
da
kbk Tψ [f ] (a, •, s)k2 ξk Tψ [f\
] (a, •, s) ds.
Spin(n) R+ 2 an+1

Thesis in Mathematics Hicham BANOUH

Clifford Wavelet Uncertainty Principle 77

√ Z Z
da
≥ 2 (kTψ [f ] (a, •, s)k22 +|2 hbk ∂bk Tψ [f ] (a, •, s), Tψ [f ] (a, •, s)i|) ds.
Spin(n) R+ an+1
By the inequality of Cauchy-Schwartz
Z Z
da
kbk Tψ [f ] (a, •, s)k2 ξk Tψ [f\ ] (a, •, s) ds
Spin(n) R+ 2 an+1
Z Z
da 1
≤( kbk Tψ [f ] (a, •, s)k22 n+1 ds) 2
Z Spin(n)Z R+ a
2 da 1
×( ξk Tψ [f\
] (a, •, s) n+1
ds) 2 .
Spin(n) R+ 2 a
Then

Z Z Z Z
da 1 2 da 1
( kbk Tψ [f ] (a, •, s)k22 ds) × (
2 ξk Tψ [f\
] (a, •, s) ds) 2
Spin(n) R+ an+1 Spin(n) R+ 2 a n+1

√ Z Z
≥ 2 (kTψ [f ] (a, •, s)k22
Spin(n) R+

da
+ |2 hbk ∂bk Tψ [f ] (a, •, s), Tψ [f ] (a, •, s)i|) ds
an+1
√ Z Z
da
= 2 (kTψ [f ] (a, •, s)k22 ds
Spin(n) R+ an+1
√ Z Z
da
+2 2 |hbk ∂bk Tψ [f ] (a, •, s), Tψ [f ] (a, •, s)i|) ds.
Spin(n) R+ an+1

Knowing that


2 Aψ 2
da
R R
 \
ξk Tψ [f ](a, •, s) n+1 ds = ξk fb
Spin(n) R+ (2π)n

2 a 2

kTψ [f ] (a, •, s)k22 da

= Aψ kf k22 .
R R
ds


Spin(n) R+ an+1

Then we have

Z Z
da 1 Aψ 2 √
( kbk Tψ [f ] (a, •, s)k22 ds) 2 × ( ξ k
b ) 12 ≥ 2Aψ kf k2 .
f 2
Spin(n) R+ an+1 (2π)n 2

√ Z Z
da
+2 2 |hbk ∂bk Tψ [f ] (a, •, s), Tψ [f ] (a, •, s)i|) ds.
Spin(n) R+ an+1

Thesis in Mathematics Hicham BANOUH

Clifford Wavelet Uncertainty Principle 78

So

Z Z
da 1
q
( kbk Tψ [f ] (a, •, s)k22 n+1
ds) 2 ξk fb ≥ 2n+1 π n Aψ kf k22
Spin(n) R+ a 2

s
2n+3 π n
Z Z
da
+ |hbk ∂bk Tψ [f ] (a, •, s), Tψ [f ] (a, •, s)i|) ds
Aψ Spin(n) R+ an+1
q
= 2n+1 π n Aψ kf k22
s
2n+3 π n
Z Z
da
+ |hbk ∂bk Tψ [f ] (a, •, s), Tψ [f ] (a, •, s)i|) ds.
Aψ Spin(n) R+ an+1

We know that

Z Z
da
|hbk ∂bk Tψ [f ] (a, •, s), Tψ [f ] (a, •, s)i|) ds
Spin(n) R+ an+1

Z Z Z
da
= [∂bk Tψ [f ] (a, b, s)]† bk Tψ [f ] (a, b, s)dV (b) ds
Spin(n) R+ Rn an+1

Z Z Z
da
≥ [∂bk Tψ [f ] (a, b, s)]† bk Tψ [f ] (a, b, s)dV (b) ds
Spin(n) R+ Rn an+1

= |Aψ [∂bk Tψ [f ] , bk Tψ [f ]]| .

Then we have

Z Z
da 1
q
( kbk Tψ [f ] (a, •, s)k22 ds) 2 ξk fb ≥ 2n+1 π n Aψ kf k22
Spin(n) R+ an+1 2

q
+ 2n+3 π n Aψ |[∂bk Tψ [f ] , bk Tψ [f ]]| .

Thesis in Mathematics Hicham BANOUH

Clifford Wavelet Uncertainty Principle 79

Since for f, g ∈ L2 (Rn , dV (x) (see for instance [66])

[Tψ [f ] , Tψ [g]] = hf, gi

we may write

Z Z
da 1
q n o
( kbk Tψ [f ] (a, •, s)k22 ds) 2 ξk fb ≥ 2n+1 π n Aψ kf k22 + 2 |hf1 , f2 i|
Spin(n) R+ an+1 2

where

1 da
R R R
ψ a,b,s (x)∂bk Tψ [f ] (a, b, s) an+1

f1 (x) =
 dV (b)ds
Aψ Spin(n) Rn R+

1 da
R R R
ψ a,b,s (x)bk Tψ [f ] (a, b, s) an+1

f2 (x) =
 dV (b)ds
Aψ Spin(n) Rn R+

This result is evidently more powerful than the one given in Theorem (4.3.1) (see
[20, 19]). For that, we notice that there is a slit increase in the lower bound :

q n o √ np
n o
2n+1 π n Aψ kf k22 + 2 |hf1 , f2 i| = 2(2π) 2 Aψ kf k22 + 2 |hf1 , f2 i|
np
(2π) 2 Aψ
≥ kf k22
2

where f1 and f2 are as given above.

Those result may be seen as an improvement of those given in [148, 154, 77]
in the special case of the continuous wavelet transform defined on the quaternions
algebra H as for the Clifford wavelet transform defined on the geometric algebra
Cln,0 for n = 2 ≡ [4] established in [156] and for n = 2, 3 ≡ [4] given in [112, 107]
and of the similar results obtained by E. Hitzer and M. Bahri for the Cl3,0 in

Thesis in Mathematics Hicham BANOUH

Clifford Wavelet Uncertainty Principle 80

[149, 151] .

4.5 Conclusion

In this chapter, we could formulate and prove a new result on Clifford wavelet
uncertainty principle stating that we can’t know simultaneously the values of the
Clifford-Fourier transform and the Clifford wavelet transform of a square integrable
multi-vector valued function. The results are based on the generalizations of the
uncertainty principle to Clifford-Fourier transform.

Thesis in Mathematics Hicham BANOUH

Chapter 5
Conclusion and perspectives

5.1 Conclusion

In this thesis, we answered the main problematic which is that the Clifford wavelet
transform and the Clifford-Fourier transform of multivector valued function can’t
be both sharp an uncertainty principle associated with the continuous wavelet
transform in the Clifford algebra’s settings has been formulated and proved. Start-
ing from the definition of real Clifford algebra and the real continuous wavelet
transform, we have presented a continuous Clifford wavelet transform, displayed its
properties and formulated an associated uncertainty principle. This research aimed
to state a new uncertainty principle for the Clifford wavelet transform. Based
on proprieties of Clifford algebra-valued monogenic admissible mother wavelets
and harmonic analysis of the Clifford-Fourier transform, we concluded on the im-
possibility for a Clifford wavelet transform of a function and its Clifford-Fourier
transform to be simultaneously sharply concentrated. which expresses the limita-
tions on the simultaneous concentration of Tψ [f ], and fb. This results have been

Thesis in Mathematics Hicham BANOUH

Conclusion and perspectives 82

published in

• Banouh, H., Ben Mabrouk, A. and Kesri, M. Clifford-Wavelet Transform and

the Uncertainty Principle, Advances in Applied Clifford Algebras, 2019, Vol.
29, pp. 1-23. DOI:10.1007/s00006-019-1026-4.

• Banouh, H., Ben Mabrouk. A. A Sharp Clifford-Wavelet Heisenberg-type

Uncertainty Principle, Journal of Mathematical Physics, 2020, Vol. 62, Issue
9. DOI:10.1063/5.0015989.

5.2 Perspectives

5.2.1 Practical Applications

We attend to apply the results of this thesis to some concrete Clifford wavelets such
as Clifford-Hermite, Clifford-Bessel, Clifford-Laguerre and Clifford-Gegenbauer
wavelets [28, 45, 42, 12, 11, 9, 14, 29] and other.

5.2.2 Donoho-Stark uncertainty principle for the Clifford

wavelet transform

Following [78], [2] and [113] we may try to extend the Dohono-Stark uncertainty
principle for ε-concentrated Clifford wavelet transforms. We recall that for Ω ⊂ Rn
the function f : Ω −→ Rn is ε-concentrated in the Lp norm on Ω if there exists
Ω > 0 such
Z  p1
p
|f (x)| dV (x) ≤ εΩ k f k p .
Rn \Ω

Thesis in Mathematics Hicham BANOUH

Conclusion and perspectives 83

5.2.3 Continuous shearlet transform in Clifford algebra

As for the continuous wavelet transform, the shearlet transform [135, 60, 134, 61]
can be derived from a square-integrable group representation of a specific group
namely the shear group R∗ × Rn−1 × Rn with the operators defined the following
way : we set for a ∈ R∗ and s = (s1 , s2 , · · · , sn−1 ) ∈ Rn−1 the dilation matrix
 
 a ... 0 
 . .
.. . . ... 

Aa = 
 
 
0 ... a
and the shear matrix  
1 s1 s2 . . . sn−1
..
 
.
 

 0 1 s1 s2 

 .. .. .. .. .. 
Ss = 
 . . . . . .

.. ..
 
. . 1 s1
 
 
 
0 ... ... 0 1
We may try to extend this transform into the Clifford algebra Rn and deduce a new
uncertainty principle (see for example [192, 59, 206, 46, 167, 155]) and formulate
new uncertainty principles .

5.2.4 New uncertainty principles for the Clifford wavelet

transform

We may find a generalised result for the Lp -variance instead of the square integrable
one and a logarithmic uncertainty principle based on [155] and [154].

Thesis in Mathematics Hicham BANOUH

Conclusion and perspectives 84

5.2.5 Controllability of Clifford algebra valued Systems

One future perspective is to introduce the theory of controllability and observabil-

ity in the settings of non-commutative Clifford algebra Rn and apply it to some
practical problems as three dimensional movements of planes and space shuttles
[216, 85, 215, 217, 121]

Thesis in Mathematics Hicham BANOUH

Bibliography

[1] A Geometric-algebraic View of Linear Systems. In John L. Casti, editor,

Mathematics in Science and Engineering, chapter 9, pages 246–316. Elsevier,
1977.
[2] A. Abouelaz, A. Achak, R. Daher, and N. Safouane. Donoho-Stark’s Uncer-
tainty Principle for the Quaternion Fourier Transform. Boletín de la Sociedad
Matemática Mexicana, jul 2019.
[3] M. Abramowitz and I. A. Stegun. Handbook of Mathematical Functions with
Formulas, Graphs, and Mathematical Tables, volume 55 of Applied Mathe-
matics Series. National Bureau of Standards, 1964.
[4] R. Abreu-Blaya, J. Bory-Reyes, and P. Bosch. Extension Theorem for Com-
plex Clifford Algebras-Valued Functions on Fractal domains. Boundary Value
Problems, 2010.
[5] J. B. Almeida. Can physics laws be derived from monogenic functions ?
ArXiv: physics/ 0601194 v1, 2006.
[6] B. Amri and L. T. Rachdi. Beckner logarithmic uncertainty principle for the
Riemann-Liouville operator. International Journal of Mathematics, 24(09),

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 86

2013.
[7] B. Amri and L. T. Rachdi. Uncertainty Principle in Terms of Entropy for
the Riemann–Liouville Operator. Bulletin of the Malaysian Mathematical
Sciences Society, 39(1):457–481, 2015.
[8] J.-P. Antoine, R. Murenzi, P. Vandergheynst, and S. T. Ali. Two-
Dimensional Wavelets and their Relatives. Cambridge University Press,
2004.
[9] S. Arfaoui. Some Monogenic Gegenbauer-Jacobi Polynomials and Associ-
ated Spheroidal Wavelets. PhD thesis, Faculté Des Sciences De Monastir,
Universite De Monastir, 2017.
[10] S. Arfaoui and A. Ben Mabrouk. Some Generalized Clifford-Jacobi Polyno-
mials and Associated Spheroidal Wavelets. arXiv preprint arXiv:1704.03513,
April 2017.
[11] S. Arfaoui and A. Ben Mabrouk. Some Old Orthogonal Polynomials Revis-
ited and Associated Wavelets : Two-Parameters Clifford-Jacobi Polynomi-
als and Associated Spheroidal Wavelets. Acta Applicandae Mathematicae,
155(1):177–195, December 2017.
[12] S. Arfaoui and A. Ben Mabrouk. Some Ultraspheroidal Monogenic Clifford-
Gegenbauer-Jacobi Polynomials and Associated Wavelets. Advances in Ap-
plied Clifford Algebras, 27:2287–2306, May 2017.
[13] S. Arfaoui, A. Ben Mabrouk, and C. Cattani. New Type of Gegenbauer-
Hermite Monogenic Polynomials and Associated Clifford Wavelets. Journal
of Mathematical Imaging and Vision, October 2019.
[14] S. Arfaoui, A. Ben Mabrouk, and C. Cattani. New Type of Gegenbauer-
Jacobi-Hermite Monogenic Polynomials and Associated Continuous Clifford

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 87

Wavelet Transform. Acta Applicandae Mathematicae, mar 2020.

[15] A. Axelsson, R. Grognard, J. Hogan, and A. McIntosh. Harmonic Analy-
sis of Dirac Operators on Lipschitz Domains. In Clifford Analysis and Its
Applications, pages 231–246. Springer Netherlands, 2001.
[16] H. Banouh. Uncertainty Principle Associated with the Clifford Continuous
Wavelet Transform. University of Science and Technology of China, Hefei,
China, August 2020.
[17] H. Banouh and A. Ben Mabrouk. A sharp Clifford-wavelet Heisenberg-type
uncertainty principle.
[18] H. Banouh and A. Ben Mabrouk. A sharp Clifford wavelet Heisenberg-type
uncertainty principle. J. Math. Phys., 61(9):093502, sep 2020.
[19] H. Banouh, A. Ben Mabrouk, and M. Kesri. Clifford-Wavelet Transform and
the Uncertainty Principle. Advances in Applied Clifford Algebras, 29(5):1–23,
October 2019.
[20] H. Banouh, A. Ben Mabrouk, and M. Kesri. Clifford-wavelet Transform and
the uncertainty principle. arXiv preprint arXiv:1905.10169v1, 2019.
[21] E. Baspinar, G. Citti, and A. Sarti. A Geometric Model of Multi-scale
Orientation Preference Maps via Gabor Functions. Journal of Mathematical
Imaging and Vision, 2018.
[22] A. Ben Mabrouk, S. Arfaoui, and I. Rezgui. Wavelet analysis on the sphere:
Spheroidal wavelets. De Gruyter, March 2017.
[23] S. Bernstein, editor. Topics in Clifford Analysis. Springer International
Publishing, 2019.
[24] L. E. Blumenson. A Derivation of n-Dimensional Spherical Coordinates. The
American Mathematical Monthly, 67(1):63, jan 1960.

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 88

[25] A. Boulkhemair, A. Chakib, and A Sadik. On a Shape Derivative Formula

for a Family of Star-Shaped Domains. working paper or preprint, January
2020.
[26] R. N. Bracewell. The Fourier Transform and Its Applications. Electrical
engineering series. McGraw Hill, 2000.
[27] F. Brackx, J. S. R. Chisholm, and V. Souček, editors. Clifford Analysis and
Its Applications. Springer Netherlands, 2001.
[28] F. Brackx, N. De Schepper, and F. Sommen. The Clifford-Laguerre continu-
ous wavelet transform. Bulletin of The Belgian Mathematical Society-Simon
Stevin, 11(2):201–215, 2004.
[29] F. Brackx, N. De Schepper, and F. Sommen. The Clifford-Gegenbauer Poly-
nomials and the Associated Continuous Wavelet Transform. Integral Trans-
forms and Special Functions, 15(5):387–404, October 2004.
[30] F. Brackx, N. De Schepper, and F. Sommen. The Clifford-Fourier transform.
Journal of Fourier Analysis and Applications, 11(6):669–681, 2005.
[31] F. Brackx, N. De Schepper, and F. Sommen. Clifford-Hermite and Two-
dimensional Clifford-Gabor Filters for Early Vision. In K. Gürlebeck and
C. Könke, editors, Proceedings 17th International Conference on the Ap-
plication of Computer Science and Mathematics in Architecture and Civil
Engineering, Bauhaus-Universität Weimar, 2006.
[32] F. Brackx, N. De Schepper, and F. Sommen. Clifford-Hermite Monogenic
Operators. Czechoslovak Mathematical Journal, 56(4):1301–1322, dec 2006.
[33] F. Brackx, N. De Schepper, and F. Sommen. Clifford-Jacobi polynomials and
the associated continuous wavelet transform in euclidean space. In Wavelet
Analysis and Applications, pages 185–198. Springer, 2006.

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 89

[34] F. Brackx, N. De Schepper, and F. Sommen. The Two-dimensional Clifford-

Fourier Transform. J. Math. Imaging, 26:5–18, 2006.
[35] F. Brackx, N. De Schepper, and F. Sommen. The Fourier Transform in
Clifford analysis. In Advances in Imaging and Electron Physics, volume 156,
pages 55–201. Elsevier, 2009.
[36] F. Brackx, N. De Schepper, and F. Sommen. The Clifford-Fourier Inte-
gral Kernel in Even Dimensional Euclidean Space. Journal of Mathematical
Analysis and Applications, 365(2):718–728, may 2010.
[37] F. Brackx, R. Delanghe, and F. Sommen. Clifford Analysis, volume 76 of
Research Notes in Mathematics. Pitman Advanced Publishing Program,
1982.
[38] F. Brackx, R. Delanghe, and F. Sommen. Cauchy-Kowalewski theorems in
Clifford analysis: a survey. Proceedings of the 11th Winter School on Abstract
Analysis, pages 55–70, 1984.
[39] F. Brackx, E. Hitzer, and S. J. Sangwine. History of Quaternion and Clifford-
Fourier Transforms and Wavelets. In Quaternion and Clifford Fourier Trans-
forms and Wavelets, pages 12–28. Springer Basel, 2013.
[40] F. Brackx, H. De Schepper, R. Lávička, and V. Souček. The Cauchy-
Kovalevskaya Extension Theorem in Hermitian Clifford Analysis. Journal of
Mathematical Analysis and Applications, 381(2):649–660, September 2011.
[41] F. Brackx, T. E. Simos, G. Psihoyios, and C. Tsitouras. Two Powerful
Theorems in Clifford Analysis. In AIP Conference Proceedings, volume 1281,
pages 3–7. American Institute of Physics, AIP, 2010.
[42] F. Brackx and F. Sommen. Clifford-Hermite wavelets in Euclidean space. J.
Fourier Anal. Appl., 6(3):299–310, 2000.

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 90

[43] F. Brackx and F. Sommen. The Continuous Wavelet Transform in Clifford

Analysis. In Clifford Analysis and Its Applications, pages 9–26. Springer
Netherlands, 2001.
[44] F. Brackx and F. Sommen. The Generalized Clifford-Hermite Continuous
Wavelet Transform. Advances in Applied Clifford Algebras, 11(1):219–231,
2001.
[45] F. Brackx and F. Sommen. Clifford–Bessel wavelets in Euclidean space.
Mathematical methods in the applied sciences, 25(16-18):1479–1491, 2002.
[46] K. Brahim, B. Nefzi, and E. Tefjeni. Uncertainty Principles For The Contin-
uous Quaternion Shearlet Transform. Advances in Applied Clifford Algebras,
29(3), April 2019.
[47] R. Bujack, H. De Bie, N. De Schepper, and G. Scheuermann. Convolu-
tion Products for Hypercomplex Fourier Transforms. J Math Imaging Vis,
48(3):606–624, mar 2013.
[48] R. Bujack, G. Scheuermann, and E. Hitzer. A General Geometric Fourier
Transform. In E. Hitzer and S. J. Sangwine, editors, Quaternion and Clifford
Fourier transforms and wavelets, Trends in Mathematics 27, pages 155–176.
Birkhauser, Basel, 2013.
[49] R. Bujack, G. Scheuermann, and E. Hitzer. A General Geometric Fourier
Transform Convolution Theorem. Adv. Appl. Clifford Alg., 23(1):15–38,
2013.
[50] P. Carré and M. Berthier. Advanced Color Image Processing and Analy-
sis, chapter Chapter 6, Color Representation and Processes with Clifford
Algebra, pages 147–179. Springer, New York, 2013.
[51] A.-L. Cauchy. Oeuvres Complètes. Série 1. Tome 7. Gauthier-Villars (Paris,

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 91

1882-1974), 2018.
[52] C. K. Chui. An Introduction to Wavelets. Academic Press Professional, Inc.,
San Diego, CA, USA, 1992.
[53] W. K. Clifford. Applications of Grassmann’s Extensive Algebra. American
Journal of Mathematics, 1(4):350, 1878.
[54] W. K. Clifford. On the Classification of Geometric Algebras. Mathematical
Papers, pages 397–401, 1882.
[55] J. Cnops. The Wavelet Transform in Clifford Analysis. Computational Meth-
ods and Function Theory, 1(2):353–374, November 2001.
[56] J. Cnops. An Introduction to Dirac Operators on Manifolds. Birkhäuser
Boston, 2002.
[57] J. M. Combes, A. Grossmann, Ph. Tchamitchian, and Allan D. Pierce.
Wavelets Time-Frequency Methods and Phase Space. Springer Berlin Hei-
delberg, 1989.
[58] A. K. Common and F. Sommen. Axial monogenic functions from holo-
morphic functions. Journal of Mathematical Analysis and Applications,
179(2):610–629, 1993.
[59] S. Dahlke, G. Kutyniok, P. Maass, C. Sagiv, H.-G. Stark, and G. Teschke.
The Uncertainty Principle Associated with the Continuous Shearlet Trans-
form. International Journal of Wavelets, Multiresolution and Information
Processing, 6(02):157–181, 2008.
[60] S. Dahlke, G. Steidl, and G. Teschke. The continuous shearlet transform in
arbitrary space dimensions. Journal of Fourier Analysis and Applications,
16(3):340–364, October 2009.
[61] S. Dahlke and G. Teschke. The continuous shearlet transform in higher

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 92

dimensions variations of a theme. Citeseer, March 2009.

[62] I. Daubechies. Orthonormal bases of compactly supported wavelets. Com-
munications on Pure and Applied Mathematics, 41(7):909–996, oct 1988.
[63] I. Daubechies and Bruce J. Bates. Ten Lectures on Wavelets, volume 61.
SIAM, March 1993.
[64] H. De Bie. Clifford algebras, Fourier transforms, and quantum mechanics.
Mathematical Methods in the Applied Sciences, 35(18):2198–2228, September
2012.
[65] H. De Bie and Y. Xu. On the Clifford-Fourier Transform. International
Mathematics Research Notices, 2011(22):5123–5163, jan 2011.
[66] N. De Schepper. Multi-dimensional Continuous Wavelet Transforms and
Generalized Fourier Transforms in Clifford Analysis. PhD Thesis, Ghent
University, 2006.
[67] L. Debnath and F. A. Shah. Wavelet Transforms and Their Applications.
Birkhauser Boston, 2015.
[68] L. Debnath and F. A. Shah. Lecture Notes on Wavelet Transforms. Springer
International Publishing, 2017.
[69] R. Delanghe. On Regular-Analytic Functions with Values in a Clifford Al-
gebra. Math. Ann., 185:91–111, 1970.
[70] R. Delanghe. Clifford Analysis: History and Perspective. Computational
Methods and Function Theory, 1(1):107–153, September 2001.
[71] R. Delanghe, F. Sommen, and V. Souček. Clifford Algebra and Spinor-valued
Functions. Springer Netherlands, 1992.
[72] P. A. M. Dirac. The quantum theory of the electron part ii. Proceedings of
the Royal Society of London, A118:351–361, 1928.

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 93

[73] P. A. M. Dirac. The Quantum Theory of the Electron Part I. Proceedings of

the Royal Society of London. Series A, Containing Papers of a Mathematical
and Physical Character, A117(778):610–624, 1928.
[74] J. Dubinskii and M. Reissig. Variational problems in Clifford analysis. Math-
ematical Methods in the Applied Sciences, 25(14):1161–1176, 2002.
[75] H. Dym and H. P. McKean. Fourier Series and Integrals. Academic Press
New York, 1972.
[76] D. Eelbode and E. Hitzer. Operator exponentials for the Clifford Fourier
transform on multivector fields in detail. Advances in Applied Clifford Alge-
bras, 26(3):953–968, October 2016.
[77] Y. El Haoui and S. Fahlaoui. The continuous quaternion algebra-valued
wavelet transform and the associated uncertainty principle.
[78] Y. El Haoui and S. Fahlaoui. Donoho-stark’s uncertainty principles in real
clifford algebras. Advances in Applied Clifford Algebras, 29(5):94, September
2019.
[79] Y. El Haoui, S. Fahlaoui, and E. Hitzer. Generalized uncertainty prin-
ciples associated with the quaternionic offset linear canonical transform.
ArXiv:1807.04068v2 [math.CA], 2019.
[80] M. Fei, Y. Pan, and Y. Xu. Some Shaper Uncertainty Principles for
Multivector-Valued Functions. International Journal of Wavelets, Multires-
olution and Information Processing, 14(06):1650043–1650043, 2016.
[81] H. G. Feichtinger and K. Gröchenig. Gabor Wavelets and the Heisenberg
Group: Gabor Expansions and Short Time Fourier Transform from the
Group Theoretical Point of View. Wavelets, pages 359–397, 1992.
[82] M. Felsberg and G. Sommer. The monogenic signal. IEEE Transactions on

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 94

Signal Processing, 49(12):3136–3144, 2001.

[83] Y. Fu and L. Li. Uncertainty Principle for Multivector-Valued Functions.
International Journal of Wavelets, Multiresolution and Information Process-
ing, 13(01):1550005, January 2015.
[84] R. Fueter. Zur Theorie Der Regularen Funktionen Einer Quaternionenvari-
ablen. Monat. fur Math. und Phys., 43:69–74, 1935.
[85] K. Fujimoto, T. Takeuchi, and Y. Matsumoto. On Port-Hamiltonian Model-
ing and Control of Quaternion Systems. IFAC-PapersOnLine, 48(13):39–44,
2015.
[86] D. Gabor. Theory of communication. Part 1: The analysis of information.
Journal of the Institution of Electrical Engineers-Part III: Radio and Com-
munication Engineering, 93(26):429–441, 1946.
[87] T. Gao, Z.-G. Liu, W.-Ch. Gao, and J. Zhang. A robust technique for
background subtraction in traffic video. In Mario Köppen Nikola and
Kasabov George Coghill, editors, Advances in Neuro-Information Process-
ing, 15th International Conference, ICONIP 2008, pages 736–744, Auckland,
New Zealand, 2008.
[88] T. Gao, Z.-G. Liu, S.-H. Yue, J.-Q. Mei, and J. Zhang. Traffic video-based
moving vehicle detection and tracking in the complex environment. Cyber-
netics and Systems: An International Journal, 40(7):569–588, 2009.
[89] J. E. Gilbert and M. A. M. Murray. Clifford algebras and Dirac operators in
harmonic analysis, volume 26 of Cambridge studies in advanced mathematics.
Press Syndicate of the University of Cambridge, cambridge university press
edition, 1991.
[90] P. Goupillaud, A. Grossmann, and J. Morlet. Cycle-octave and related trans-

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 95

forms in seismic signal analysis. Geoexploration, 23(1):85–102, oct 1984.

[91] H. Grassmann. Die lineale Ausdehnungslehre ein neuer Zweig der Math-
ematik. His Die Wissenschaft der extensiven Grösse, oder die Aus-
dehnungslehre. O. Wigand, 1844.
[92] A. W. Green. The Uncertainty Principle in Control Theory for Partial Dif-
ferential Equations. PhD thesis, Graduate School of Clemson Univesity,
2020.
[93] A. Grossmann and J. Morlet. Decomposition of Hardy Functions into Square
Integrable Wavelets of Constant Shape. SIAM Journal on Mathematical
Analysis, 15(4):723–736, July 1984.
[94] A. Grossmann, J. Morlet, and T. Paul. Transforms Associated to Square In-
tegrable Group Representations. I. General Results. Journal of Mathematical
Physics, 26(10):2473–2479, October 1985.
[95] A. Grossmann, J. Morlet, and T. Paul. Transforms Associated to Square
Integrable Group Representations. II. Examples. Ann. Inst. Henri Poincaré,
45(3):293–309, 1986.
[96] K. Gürlebeck, U. Kähler, J. Ryan, and W. Sprößig. Clifford Analysis Over
Unbounded Domains. Advances in Applied Mathematics, 19(2):216–239,
1997.
[97] A. Haar. Zur theorie der orthogonalen funktionen-systeme. Mathematische
Annalen, 69:331–371, 1910.
[98] A. Haar. Der Massbegriff in der Theorie der kontinuierlichen Gruppen. An-
nals of Mathematics, 34(1):147–169, 1933.
[99] W. R. Hamilton. On a New Species of Imaginary Quantities Connected
with a Theory of Quaternions. In Proceedings of the Royal Irish Academy,

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 96

volume 2, pages 424–234, 1844.

[100] W. R. Hamilton. Elements of quaternions. Longmans, Green, & Company,
1866.
[101] C. Heil and D. Walnut. Continuous and discrete wavelet transforms. SIAM
Review, 31(4):628–666, December 1989.
[102] W. Heisenberg. Uber Den Anschaulichen Inhalt Der Quantentheoretischen
Kinematik Und Mechanik. Zeitschrift fur Physik, 43:172–198, 1927.
[103] W. Heisenberg. Die Physik der Atomkerne. Vieweg+Teubner Verlag, Wies-
baden, 1949.
[104] W. Heisenberg. Uber Den Anschaulichen Inhalt Der Quantentheoretischen
Kinematik Und Mechanik, pages 478–504. Springer, Berlin, 1985.
[105] D. Hestenes and G. Sobczyk. Clifford Algebra to Geometric Calculus.
Springer Netherlands, 1984.
[106] E. Hitzer. Tutorial on fourier transformations and wavelet transformations
in clifford geometric algebra. In Lecture notes of the International Workshop
for ‘Computational Science with Geometric Algebra’(FCSGA2007), pages
65–87, 2007.
[107] E. Hitzer. Clifford (Geometric) Algebra Wavelet Transform. volume 2, pages
94–101. IEEE, 2009.
[108] E. Hitzer. Directional Uncertainty Principle for Quaternion Fourier Trans-
form. Advances in Applied Clifford Algebras, 20(2):271–284, July 2009.
[109] E. Hitzer. New Developments in Clifford Fourier Transforms. Advances in
Applied and Pure Mathematics, pages 19–25, 2014.
[110] E. Hitzer and B. Mawardi. Clifford Fourier Transform on Multivector Fields
and Uncertainty Principles for Dimensions n = 2(mod4) and n = 3(mod4).

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 97

Advances in Applied Clifford Algebras, 18(3-4):715–736, May 2008.

[111] E. Hitzer and B. Mawardi. Uncertainty Principle for the Clifford-Geometric
Algebra Cl3,0 based on Clifford Fourier Transform. ArXiv:1306.2089v1
[math.RA], June 2013.
[112] E. Hitzer, T. E. Simos, G. Psihoyios, and C. Tsitouras. Real Clifford Algebra
Cln,0 , n = 2, 3(mod4) Wavelet Transform. In AIP Conference Proceedings.
AIP, 2009.
[113] K. Hleili. Continuous wavelet transform and uncertainty principle related
to the Weinstein operator. Integral Transforms and Special Functions,
29(4):252–268, jan 2018.
[114] K. Hleili, S. Omri, and L. T. Rachdi. Uncertainty principle for the Riemann-
Liouville operator. CUBO A Mathematical Journal, 13(03):91–115, October
2011.
[115] S. Huang, Y. Y. Qiao, and G. Wen. Real and Complex Clifford Analysis.
Kluwer Academic Publishers, 2006.
[116] Z. J. Huang, G. H. Huang, and L. l. Cheng. Medical Image Segmentation of
Blood Vessels Based on Clifford Algebra and Voronoi Diagram. Journal of
Software, 13(6):361–373, 2018.
[117] M. H. J. Janssen, A. J. E. M. Janssen, E. J. Bekkers, J. O. Bescos, and
R. Duits. Design and processing of invertible orientation scores of 3d images.
Journal of Mathematical Imaging and Vision, 60:1427–1458, 2018.
[118] R. Jday. Heisenberg and Hardy Uncertainty Principles in Real Clifford Alge-
bras. Integral Transforms and Special Functions, 29(8):663–677, June 2018.
[119] B. Jefferies. Clifford Analysis. In Lecture Notes in Mathematics, pages 27–38.
Springer Berlin Heidelberg, may 2004.

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 98

[120] B. Jefferies and A. Mcintosh. The Weyl Calculus and Clifford Analysis.
Bulletin of the Australian Mathematical Society, 57(2):329–341, April 1991.
[121] B. Jiang, Y. Liu, K.-I. Kou, and Z. Wang. Controllability and Observability
of Linear Quaternion-valued Systems. Acta Mathematica Sinica, 06 2020.
[122] P. Jorgensen and F. Tian. Non-commutative analysis. World Scientific, 2017.
[123] G. Kaiser. A Friendly Guide to Wavelets. Birkhauser Boston Inc., Cam-
bridge, MA, USA, 1994.
[124] M. Kesri. Solutions globales du système intégro-différentiel non linéaire de
Liouville–Boltzmann–Poisson et l'équation spectrale. Colloquium Mathe-
maticum, 106(2):265–281, 2006.
[125] M. Kesri. An Explicit Method for Optimal Control Problems. Applied Math-
ematics and Computation, 219(15):8213–8221, apr 2013.
[126] M. Kesri. On the Class of Infinite Horizon Optimal Control Problems Affine,
Which Are Structurally Stable. Mathematical Methods in the Applied Sci-
ences, 39(11):2799–2807, oct 2015.
[127] M. Kesri. Continuous Dependence of the Value Function on a Class of One-
Dimensional Autonomous Optimal Control Problems and Existence of Solu-
tions. Indian Journal of Pure and Applied Mathematics, 50(1):269–281, mar
2019.
[128] A. Kheirabadi, A. M. Vaziri, and S. Effati. Solving Optimal Control Prob-
lem Using Hermite Wavelet. Numerical Algebra, Control & Optimization,
9(1):101–112, 2019.
[129] J. King. Uncertainty Principle in Control Theory, Part I: Analysis of Per-
formance Limitations. ArXiv pre-print : 1402.5468v1.
[130] V. Kisil. Wavelets in Banach Spaces. Acta Applicandae Mathematicae,

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 99

59(1):79–109, 1999.
[131] K. I. Kou, J.-Y. Ou, and J. Morais. On uncertainty principle for quaternionic
linear canonical transform. Abstract and Applied Analysis, 2013:1–15, 2013.
[132] K.-I. Kou and T. Qian. The Paley-Wiener Theorem in Rn with the Clifford
Analysis Setting. Journal of Functional Analysis, 189(1):227–241, feb 2002.
[133] S. V. Kovalevskaja. Zur Theorie Der Partiellen Differentialgleichungen. J.
Fur Die Reine Und Angew. Mathem., 80:1–32, 1875.
[134] D. Kumar. Multi dimensional continuous shearlet transform and multi scale
wavelet shrinkage with integro differential equations. American Review of
Mathematics and Statistics, 4(2):39–47, December 2016.
[135] G. Kutyniok and D. Labate, editors. Shearlets. Birkhäuser Boston, 2012.
[136] P. Van Lancker. Clifford analysis on the sphere. In Clifford Algebras and
Their Application in Mathematical Physics, pages 201–215. Springer Nether-
lands, 1998.
[137] G. Laville. Some Topics in Clifford Analysis. Advances in Applied Clifford
Algebras, 19(3-4):721–775, nov 2009.
[138] J.-W. Lee, L.-W. Liu, H.-K. Hong, and J.-T. Chen. Applications of the Clif-
ford Algebra Valued Boundary Element Method to Electromagnetic Scatter-
ing Problems. Engineering Analysis with Boundary Elements, 71:140–150,
2016.
[139] Z. Li, H. Yang, and Y. Qiao. A New Cauchy Integral Formula in the Complex
Clifford Analysis. Advances in Applied Clifford Algebras, 28(4), jul 2018.
[140] P. Lounesto. Clifford Algebras and Spinors. Cambridge University Press,
2001.
[141] W. Luo and J. Du. Generalized Cauchy Theorem in Clifford Analysis and

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 100

Boundary Value Problems for Regular Functions. Advances in Applied Clif-

ford Algebras, 27(3):2531–2583, may 2017.
[142] G. Ma and J. Zhao. Quaternion Ridgelet Transform and Curvelet Transform.
Adv. Appl. Clifford Algebras, 28(80), 2018.
[143] S. Mallat. A Theory for Multiresolution Signal Decomposition: The Wavelet
Representation. IEEE Transactions on Pattern Analysis and Machine Intel-
ligence, 11(7):674–693, jul 1989.
[144] S. Mallat. A Wavelet Tour of Signal Processing, Third Edition: The Sparse
Way. Academic Press, Inc., Orlando, Fl, Usa, 2008.
[145] M. Martin. Function Spaces in Quaternionic and Clifford Analysis. In Op-
erator Theory, pages 1–26. Springer Basel, 2014.
[146] B. Mawardi. Construction of quaternion-valued wavelets. MATEMATIKA,
26(1):107–114, 2010.
[147] B. Mawardi and R. Ashino. Logarithmic Uncertainty Principle for Quater-
nion Linear Canonical Transform. In Proceedings of the 2016 International
Conference on Wavelet Analysis and Pattern Recognition. IEEE, July 2016.
[148] B. Mawardi, R. Ashino, and R. Vaillancourt. Two-dimensional quaternion
wavelet transform. Applied Mathematics and Computation, 218(1):10–21,
2011.
[149] B. Mawardi and E. Hitzer. Clifford Algebra Cl( 3, 0) valued Wavelets and
Uncertainty Inequality for Clifford Gabor Wavelet Transformation. Inter-
national Journal of Wavelets, Multiresolution and Information Processing,
05(06):997–1019, November 2006.
[150] B. Mawardi and E. Hitzer. Clifford Fourier Transformation and Uncertainty
Principle for the Clifford Geometric Algebra Cl3,0 . Advances in Applied

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 101

Clifford Algebras, 16(1):41–61, February 2006.

[151] B. Mawardi and E. Hitzer. Clifford Algebra Cl(3,0)-valued Wavelet Trans-
formation, Clifford Wavelet Uncertainty Inequality and Clifford Gabor
Wavelets. International Journal of Wavelets, Multiresolution and Informa-
tion Processing, 5(6):997–1019, 2007.
[152] B. Mawardi, E. Hitzer, A. Hayashi, and R. Ashino. An Uncertainty Prin-
ciple for Quaternion Fourier Transform. Computers & Mathematics with
Applications, 56(9):2398–2410, November 2008.
[153] B. Mawardi and A. Ryuichi. A simplified proof of uncertainty principle for
quaternion linear canonical transform. Abstract and Applied Analysis, 2016,
2016.
[154] B. Mawardi and A. Ryuichi. A Variation on Uncertainty Principle and Log-
arithmic Uncertainty Principle for Continuous Quaternion Wavelet Trans-
forms. Abstract and Applied Analysis, 2017:1–11, 2017.
[155] B. Mawardi, Firdous A. Shah, and Azhar Y. Tantary. Uncertainty principles
for the continuous shearlet transforms in arbitrary space dimensions. Integral
Transforms and Special Functions, pages 1–18, jan 2020.
[156] B. Mawardi, A. Sriwulan, and J. Zhao. Clifford Algebra-Valued Wavelet
Transform on Multivector Fields. Advances in Applied Clifford Algebras,
21(1):13–30, July 2010.
[157] H. Mejjaoli, N. Ben Hamadi, and S. Omri. Localization operators, time
frequency concentration and quantitative-type uncertainty for the contin-
uous wavelet transform associated with spherical mean operator. Inter-
national Journal of Wavelets, Multiresolution and Information Processing,
17(4), 2019.

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 102

[158] Y. Meyer. Ondelettes, fonctions splines et analyses graduées. Lectures given

at the University of Torino, Italy, 9, 1986.
[159] V. Michel. Lectures on constructive approximation. Birkhäuser Boston, 2013.
[160] M. Mitrea. Clifford Wavelets Singular Integrals, and Hardy Spaces. Springer-
Verlag, 1994.
[161] J. Morlet, G. Arens, E. Fourgeau, and D. Giard. Wave propagation and
sampling theory Part II: Sampling theory and complex waves. Geophysics,
47(2):222–236, 1982.
[162] A. J. Morris. Fourier and Wavelet Analysis of Clifford-Valued Functions.
PhD Thesis, University of Newcastle, Australia, Callaghan, NSW, 2308,
Australia, February 2014.
[163] R. A. Mosna, D. Miralles, and J. Vaz. Z2 -gradings of Clifford Algebras and
Multivector Structures. Journal of Physics A: Mathematical and General,
36(15):4395–4405, April 2003.
[164] N. Msehli and L. T. Rachdi. Beurling-Hörmander uncertainty principle for
the spherical mean Operator. J. Inequal. Pure Appl. Math., 10(2), 2009.
[165] N. Msehli and L. T. Rachdi. Heisenberg-Pauli-Weyl Uncertainty Principle
for the Spherical Mean Operator. Mediterranean Journal of Mathematics,
7(2):169–194, 2010.
[166] K. Nagata and T. Nakamura. Violation of Heisenberg’s Uncertainty Princi-
ple. Open Access Library Journal, 2015.
[167] B. Nefzi, K. Brahim, and A. Fitouhi. Uncertainty principles for the multivari-
ate continuous shearlet transform. Journal of Pseudo-Differential Operators
and Applications, April 2019.
[168] C. A. Nolder. Conjugate Harmonic Functions and Clifford Algebras. Journal

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 103

of Mathematical Analysis and Applications, 302(1):137–142, feb 2005.

[169] E. Obolashvili. Some partial differential equations in Clifford analysis. Ba-
nach Center Publications, 37(1):173–179, 1996.
[170] H. M. Ozaktas, M. A. Kutay, and D. Mendlovic. Introduction to the Frac-
tional Fourier Transform and Its Applications. In Advances in Imaging and
Electron Physics, pages 239–291. Elsevier, 1999.
[171] T. Qian. Generalization of Fueter’s result to Rn+1 . Atti della Accademia
Nazionale dei Lincei. Classe di Scienze Fisiche, Matematiche e Naturali.
Rendiconti Lincei. Matematica e Applicazioni,Serie 9, 8(2):111–117, 1997.
[172] T. Qian. Derivation of monogenic functions and applications. In Interna-
tional Conference on Harmonic Analysis and Related Topics, pages 118–127.
Centre for Mathematics and its Applications, Mathematical Sciences Insti-
tute, 2003.
[173] T. Qian. Paley-Wiener Theorems and Shannon Sampling in the Clifford
Analysis Setting. In Clifford Algebras, pages 115–124. Birkhauser Boston,
2004.
[174] L. T. Rachdi, B. Amri, and A. Hammami. Uncertainty principles and
time frequency analysis related to the Riemann–Liouville operator. Annali
Dell’universita’ Di Ferrara, 2018.
[175] L. T. Rachdi and H. Herch. Uncertainty Principles for Continuous Wavelet
Transforms Related to the Riemann–Liouville Operator. Ricerche Di Matem-
atica, 66(2):553–578, feb 2017.
[176] L. T. Rachdi and F. Meherzi. Continuous wavelet transform and uncertainty
principle related to the spherical mean operator. Mediterranean Journal of
Mathematics, 14(1), dec 2016.

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 104

[177] M. Razzaghi and S. Yousefi. Legendre Wavelets Method for Constrained

Optimal Control Problems. Mathematical Methods in the Applied Sciences,
25(7):529–539, 2002.
[178] N. Ricker. Wavelet Contraction, Wavelet Expansion, and the Control of
Seismic Resolution. GEOPHYSICS, 18(4):769–792, oct 1953.
[179] M. Riesz. Clifford Numbers and Spinors. Springer Netherlands, 1993.
[180] D. Rizo-Rodríguez, H. Mendez-Vazquez, and E. Garcia-Reyes. Illumination
invariant face recognition using quaternion-based correlation filters. J. Math.
Imaging Vis., 45:164–175, 2013.
[181] S. Rolf and J. Ryan. Clifford and harmonic analysis on cylinders and tori.
Revista Matemática Iberoamericana, 21(1):87–110, 2005.
[182] W. Rudin. Functional analysis. International Series in Pure and Applied
Mathematics. McGraw-Hill Education - Europe, New York, 1991.
[183] J. Ryan. Cauchy-Green type formulae in Clifford analysis. Transactions of
the American Mathematical Society, 347(4):1331–1341, 1995.
[184] J. Ryan. A decomposition theorem in Clifford analysis. Journal of Operator
Theory, 39(2):297–308, 1998.
[185] J. Ryan. Basic Clifford Analysis. CUBO Math. Educ., 2:226–256, 2000.
[186] J. Ryan. Clifford Algebras in Analysis and Related Topics. CRC Press,
March 2018.
[187] I. Sabadini and F. Sommen. Clifford Analysis on the Space of Vectors,
Bivectors and l-vectors. In Advances in Analysis and Geometry, pages 161–
185. Birkhäuser Basel, 2004.
[188] K. Sau, R. K. Basaka, and A. Chanda. Image Compression based on Block
Truncation Coding Using Clifford Algebra. Procedia Technology, 10:699–706,

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 105

2013.
[189] M. Sce. Osservazioni Sulle Serie Di Potenze Nei Moduli Quadratici. Atti
Accad. Naz.Lincei. Rend. Cl. Sci. Fis. Mat. Nat., 8(23):220–225, 1957.
[190] K. K. Selig. Uncertainty Principles Revisited. Electron. Trans. Numer. Anal.,
14:165–177, 2002.
[191] D. Sen. The Uncertainty Relations in Quantum Mechanics. Current Science,
107(2):203–2018, July 2014.
[192] J. Sharma and S. K. Singh. Clifford Valued Shearlet Transform. Advances
in Applied Clifford Algebras, 30(3), jun 2020.
[193] S. Shirrell and R. Walter. Hermitian Clifford Analysis and Its Connec-
tions with Representation Theory. Complex Variables and Elliptic Equations,
62(9):1329–1342, March 2017.
[194] H. Skibbe and M. Reisert. Spherical tensor algebra: A toolkit for 3d image
processing. J. Math. Imaging Vis., 58(3):349–381, 2017.
[195] I. N. Sneddon. Fourier Transforms. Dover Publications Inc., 2003.
[196] F. Sommen. Special Functions in Clifford Analysis and Axial Symmetry.
Journal of Mathematical Analysis and Applications, 130(1):110–133, Febru-
ary 1988.
[197] F. Sommen. Clifford analysis and integral geometry. In Clifford Algebras
and their Applications in Mathematical Physics, pages 293–311. Springer
Netherlands, 1992.
[198] F. Sommen. Clifford Analysis in Two and Several Vector Variables. Appli-
cable Analysis, 73(1-2):225–253, oct 1999.
[199] F. Sommen and H. De Schepper. Introductory Clifford Analysis. Operator
Theory, pages 1–27, 2015.

Thesis in Mathematics Hicham BANOUH

BIBLIOGRAPHY 106

[200] F. Sommen and W. Sprößig. Introduction to clifford analysis. Mathematical

Methods in the Applied Sciences, 25(16-18):1337–1342, nov 2002.
[201] P. A. Stabnikov. Geometric interpretation of the uncertainty principle. Nat-
ural Science, 11(5):146–148, 2019.
[202] S. Stankovic, L. Stankovic, and I. Orovic. Compressive Sensing Approach
in the Hermite Transform Domain. Mathematical Problems in Engineering,
2015.
[203] E. M. Stein. Singular Integrals and Differentiability Properties of Functions.
Monographs in harmonic analysis. Princeton University Press, 1970.
[204] E. M. Stein and G. Weiss. Introduction to Fourier Analysis on Euclidean
Spaces (PMS-32). Princeton University Press, dec 1972.
[205] D. C. Struppa. Quaternionic and Clifford Analysis in Several Variables. In
Operator Theory, pages 1549–1578. Springer Basel, 2015.
[206] Y. Su. Heisenberg type uncertainty principle for continuous shearlet trans-
form. Journal of Nonlinear Science and Applications, 9(3):778–786, 2016.
[207] W. Swaidan and A. Hussin. Feedback Control Method Using Haar Wavelet
Operational Matrices for Solving Optimal Control Problems. Abstract and
Applied Analysis, 2013:1–8, 2013.
[208] A. Teolis. Computational Signal Processing with Wavelets. Birkhauser
Boston, 1998.
[209] J. G. Vargas and D. G. Torr. Clifford-Valued Clifforms: A Geometric Lan-
guage for Dirac Equations. In Clifford Algebras and their Applications in
Mathematical Physics, volume 1, pages 135–154. Birkhäuser Boston, 2000.
[210] N. Vieira. Some results in fractional Clifford analysis. In International
Conference on the Applications of Computer Science and Mathematics in

Thesis in Mathematics Hicham BANOUH

 BIBLIOGRAPHY 107

Architecture and Civil Enginering, 2015. Event-place: Bauhaus-University

Weimar.
[211] L. Wang, R. Hu, J. Zhang, and Y. Ma. On the vortex detection method using
continuous wavelet transform with application to propeller wake analysis.
Mathematical Problems in Engineering, 2015.
[212] H. Weyl. The Theory of Groups and Quantum Mechanics. New York, second
edition, 1950.
[213] L. Wietzke and G. Sommer. The signal multi-vector. Journal of Mathemat-
ical Imaging and Vision, 37(2):132–150, mar 2010.
[214] P. Wojtaszczyk. A mathematical introduction to wavelets, volume 37. Cam-
bridge University Press, 1997.
[215] D. Xu, J. He, A. Sheng, Z. Chen, and D. Wang. Robust Attitude Tracking
Control of a Rigid Spacecraft Based on Nonlinearly Controlled Quaternions.
In 2017 11th Asian Control Conference (ASCC). IEEE, dec 2017.
[216] Y. Yang. Spacecraft Attitude Determination and Control: Quaternion Based
Method. Annual Reviews in Control, 36(2):198–219, dec 2012.
[217] Y. Yang. Spacecraft Modeling, Attitude Determination, and Control
Quaternion-based Approach. CRC Press, feb 2019.
[218] Y. Yang, P. Dang, and T. Qian. Stronger Uncertainty Principles for Hyper-
complex Signals. Complex Variables and Elliptic Equations, 60(12):1696–
1711, May 2015.
[219] Y. Yang and K. I. Kou. Uncertainty principles for hypercomplex signals in
the linear canonical transform domains. Signal Processing, 95:67–75, Febru-
ary 2014.

Thesis in Mathematics Hicham BANOUH

View publication stats