Number Theory Meets Wireless Communications 1st Ed 2020 Victor Beresnevich

motoabyer 2 views 89 slides May 24, 2025
Slide 1
Slide 1 of 89
Slide 1
1
Slide 2
2
Slide 3
3
Slide 4
4
Slide 5
5
Slide 6
6
Slide 7
7
Slide 8
8
Slide 9
9
Slide 10
10
Slide 11
11
Slide 12
12
Slide 13
13
Slide 14
14
Slide 15
15
Slide 16
16
Slide 17
17
Slide 18
18
Slide 19
19
Slide 20
20
Slide 21
21
Slide 22
22
Slide 23
23
Slide 24
24
Slide 25
25
Slide 26
26
Slide 27
27
Slide 28
28
Slide 29
29
Slide 30
30
Slide 31
31
Slide 32
32
Slide 33
33
Slide 34
34
Slide 35
35
Slide 36
36
Slide 37
37
Slide 38
38
Slide 39
39
Slide 40
40
Slide 41
41
Slide 42
42
Slide 43
43
Slide 44
44
Slide 45
45
Slide 46
46
Slide 47
47
Slide 48
48
Slide 49
49
Slide 50
50
Slide 51
51
Slide 52
52
Slide 53
53
Slide 54
54
Slide 55
55
Slide 56
56
Slide 57
57
Slide 58
58
Slide 59
59
Slide 60
60
Slide 61
61
Slide 62
62
Slide 63
63
Slide 64
64
Slide 65
65
Slide 66
66
Slide 67
67
Slide 68
68
Slide 69
69
Slide 70
70
Slide 71
71
Slide 72
72
Slide 73
73
Slide 74
74
Slide 75
75
Slide 76
76
Slide 77
77
Slide 78
78
Slide 79
79
Slide 80
80
Slide 81
81
Slide 82
82
Slide 83
83
Slide 84
84
Slide 85
85
Slide 86
86
Slide 87
87
Slide 88
88
Slide 89
89

About This Presentation

Number Theory Meets Wireless Communications 1st Ed 2020 Victor Beresnevich
Number Theory Meets Wireless Communications 1st Ed 2020 Victor Beresnevich
Number Theory Meets Wireless Communications 1st Ed 2020 Victor Beresnevich


Slide Content

Number Theory Meets Wireless Communications 1st
Ed 2020 Victor Beresnevich download
https://ebookbell.com/product/number-theory-meets-wireless-
communications-1st-ed-2020-victor-beresnevich-51992358
Explore and download more ebooks at ebookbell.com

Here are some recommended products that we believe you will be
interested in. You can click the link to download.
Number Theory Meets Wireless Communications Victorberesnevich Et Al
https://ebookbell.com/product/number-theory-meets-wireless-
communications-victorberesnevich-et-al-23665816
Noncommutative Geometry And Number Theory Where Arithmetic Meets
Geometry And Physics 1st Edition Annemarie Aubert
https://ebookbell.com/product/noncommutative-geometry-and-number-
theory-where-arithmetic-meets-geometry-and-physics-1st-edition-
annemarie-aubert-4211388
Number Theory For The Millennium Iii M A Bennett Bc Berndt N Boston H
G Diamond A J Hildebrand W Philipp
https://ebookbell.com/product/number-theory-for-the-millennium-iii-m-
a-bennett-bc-berndt-n-boston-h-g-diamond-a-j-hildebrand-w-
philipp-48176150
Number Theory Volume 1 Tools And Diophantine Equations Henri Cohen
https://ebookbell.com/product/number-theory-volume-1-tools-and-
diophantine-equations-henri-cohen-49055602

Number Theory Volume 2 Analytic And Modern Tools Henri Cohen
https://ebookbell.com/product/number-theory-volume-2-analytic-and-
modern-tools-henri-cohen-49055636
Number Theory And Combinatorics A Collection In Honor Of The
Mathematics Of Ronald Graham Bruce M Landman Editor Florian Luca
Editor Melvyn B Nathanson Editor Jaroslav Neetil Editor Aaron
Robertson Editor
https://ebookbell.com/product/number-theory-and-combinatorics-a-
collection-in-honor-of-the-mathematics-of-ronald-graham-bruce-m-
landman-editor-florian-luca-editor-melvyn-b-nathanson-editor-jaroslav-
neetil-editor-aaron-robertson-editor-50985326
Number Theory Proceedings Of The Journes Arithmtiques 2019 Xxxi Held
At Istanbul University Kaan Kurungz Editor Ayberk Zeytin Editor
https://ebookbell.com/product/number-theory-proceedings-of-the-
journes-arithmtiques-2019-xxxi-held-at-istanbul-university-kaan-
kurungz-editor-ayberk-zeytin-editor-50985344
Number Theory Proceedings Of The Turku Symposium On Number Theory In
Memory Of Kustaa Inkeri May 31june 4 1999 Reprint 2013 Matti Jutila
Editor Tauno Metsnkyl Editor
https://ebookbell.com/product/number-theory-proceedings-of-the-turku-
symposium-on-number-theory-in-memory-of-kustaa-inkeri-
may-31june-4-1999-reprint-2013-matti-jutila-editor-tauno-metsnkyl-
editor-50985360
Number Theory Analysis And Combinatorics Proceedings Of The Paul Turan
Memorial Conference Held August 2226 2011 In Budapest Jnos Pintz
Editor Andrs Bir Editor Klmn Gyry Editor Gergely Harcos Editor Mikls
Simonovits Editor Jzsef Szabados Editor
https://ebookbell.com/product/number-theory-analysis-and-
combinatorics-proceedings-of-the-paul-turan-memorial-conference-held-
august-2226-2011-in-budapest-jnos-pintz-editor-andrs-bir-editor-klmn-
gyry-editor-gergely-harcos-editor-mikls-simonovits-editor-jzsef-
szabados-editor-50985376

Mathematical Engineering
VictorBeresnevich
AlisterBurr
BobakNazer
SanjuVelani Editors
Number Theory 
Meets Wireless 
Communications

Mathematical Engineering
Series Editors
Jörg Schr¨oder, Institute of Mechanics, University of Duisburg-Essen, Essen,
Germany
Bernhard Weigand, Institute of Aerospace Thermodynamics, University of
Stuttgart, Stuttgart, Germany

Today, the development of high-tech systems is unthinkable without mathematical
modeling and analysis of system behavior. As such, many fields in the modern
engineering sciences (e.g. control engineering, communications engineering,
mechanical engineering, and robotics) call for sophisticated mathematical methods
in order to solve the tasks at hand.
The series Mathematical Engineering presents new or heretofore little-known
methods to support engineers in finding suitable answers to their questions,
presenting those methods in such manner as to make them ideally comprehensible
and applicable in practice.
Therefore, the primary focus is—without neglecting mathematical accuracy—on
comprehensibility and real-world applicability.
To submit a proposal or request further information, please use the PDF Proposal
Form or contact directly: Dr. Thomas Ditzinger ([email protected])
Indexed by SCOPUS, zbMATH, SCImago.
More information about this series athttp://www.springer.com/series/8445

Victor Beresnevich • Alister Burr • Bobak Nazer •
Sanju Velani
Editors
NumberTheoryMeets
WirelessCommunications

Editors
Victor Beresnevich
Department of Mathematics
University of York
York, UK
Alister Burr
Department of Electronic Engineering
University of York
York, UK
Bobak Nazer
Department of Electrical and Computer
Engineering
Boston University
Boston, MA, USA
Sanju Velani
Department of Mathematics
University of York
York, UK
ISSN 2192-4732 ISSN 2192-4740 (electronic)
Mathematical Engineering
ISBN 978-3-030-61302-0 ISBN 978-3-030-61303-7 (eBook)
https://doi.org/10.1007/978-3-030-61303-7
Mathematics Subject Classification:11K60, 11H06, 11J83, 11H71, 94A15, 94A40, 11J25
© Springer Nature Switzerland AG 2020
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Switzerland AG.
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Preface
Powerful techniques from various areas of Number Theory have played important
roles in breakthrough developments in areas of Wireless Communications. These
include the impact of geometry of numbers, Diophantine approximation and
algebraic number theory on lattice coding and interference alignment. This book
introduces and describes some of these developments as well as the techniques
that have made them possible. It lays particular emphasis on those that are at the
forefront of current research. The chapters are all written by leading researchers
in both areas. They present the state-of-the-art research, which illustrates the deep
interaction between number theory and wireless communications. Together, they
show that there is currently great scope to develop the mutual understanding of
methods and problems.
The book has been developed from lectures given at the international meeting
“Workshop on Interactions Between Number Theory and Wireless Communication”
held at the University of York in July 2016. Details, including list of participants,
programme and slides of talks, can be found at:
https://www.york.ac.uk/maths/events/2016/workshop-interactions-between-
number-theory-wirele/
The primary goal of the workshop was to inspire both early career and established
researchers to consolidate and build new and exciting bridges between Number
Theory and Wireless Communications. Naturally, this is also the overarching goal
of this book. With this in mind, we encouragedthe speakers to develop their written
contributions in an expository way and to provide an overview of current tools and
developments. Each chapter thus foregrounds the main concepts behind the topic
under consideration while keeping technicalities to a bare minimum. The chapters
thus offer direct and accessible information about highly exciting current research
developments to researchers in both Number Theory and Communication Theory.
Breaking down the superficial “language barrier” between the two disciplines is key
to understanding the respective central problems and is the first step towards fruitful
collaboration and progress.
v

vi Preface
To the best of our knowledge, this book is the first volume jointly edited by
individuals working in Number Theory and Communication Theory. We hope
it provides a unique insight into key concepts, cutting-edge results, and modern
techniques that play an essential role in contemporary research. Great effort has
been made to present the material in a manner that is accessible to new researchers,
including Ph.D. students. The book will also be useful for established researchers
working in Number Theory or Wireless Communications who wish to broaden their
outlook and contribute towards the deep interplay between the two.
Many communication techniques involve choosing discrete sets of points that
represent information being sent over a communication channel. These discrete sets
represent a key element of codes and are often conveniently chosen to have a linear
structure. The presence of linear structure allows for efficient and low-complexity
decoding. Furthermore, it provides powerful properties that underpin state-of-the-
art techniques for managing interference and other challenges in wireless networks.
In number theory, discrete linear structures are natural objects of study within
the geometry of numbers and Diophantine approximation. Roughly speaking, the
geometry of numbers characterizes geometric properties (such as packing and
covering radii and Minkowski minima) of linear structures known as lattices. On
the contrary, Diophantine approximation studies the properties of linear maps on
such linear structures. The basis for the theory of Diophantine approximation
is Dirichlet’s classical theorem based on the Pigeonhole principle. The opening
chapter discusses the role it plays in the world of wireless communications. The
authors present an informal discussion of aspects of wireless communications via a
series of basic examples. These allow them to introduce a variety of concepts (such
as badly approximable, singular and well approximable points) and aspects (such
as probabilistic and manifolds theories) from Diophantine approximation while
explaining their role in wireless communications. In particular, they introduce a new
concept of jointly non-singular points and use it to improve a well-known result of
Motahari et al. regarding the Degrees of Freedom (DoF) of a two-user X-channel.
An overarching goal of Chap.1is to provide an answer to the question:What is the
role of number theory in the world of wireless communications?
Consider multiple transmitters and receivers that communicate with each other
across a shared wireless channel. The two main challenges to establishing reliable
communication between users are the noise introduced by the channel and the
interference between simultaneously transmitted signals. Over the past few decades,
experts in network information theory have strived to determine the fundamental
limits of reliable communication over multi-user channels. At the same time,
they attempted to realize network architectures that, in practice, could approach
these limits. In Chap.2, the authors discuss the recent developments in network
information theory based on the use of lattice codebooks (i.e. codebooks that are
a subset of a lattice overR
n
). The inherent linearity of lattice codebooks can
be effectively used as a building block for communication strategies that operate
beyond the performance available for classical coding schemes. In general, the
performance of these lattice-based strategies is determined by how closely the
channel coefficients can be approximated by integer coefficients. In other words, the

Preface vii
performance is intertwined with the theory of Diophantine approximation. Overall,
this chapter provides a unified view of recent results that connect the performance
of the compute-and-forward strategy of recovering an integer-linear combination
of codewords to Diophantine approximation bounds. The chapter concludes by
highlighting scenarios in which novel applications of Diophantine approximation,
such as non-asymptotic approximation bounds over manifolds, have the potential to
yield new exciting results in network information theory.
With the roll out of the fifth-generation (5G) wireless systems, constructing
efficient space–time codes offering complexity reduction is crucial for many
applications including massive multiple-input multiple-output (MIMO
Traditionally, space–time codes have been developed in the context of point-to-point
MIMO communications. However, today’s wireless networks need to accommodate
numerous types of applications and devices. In view of this, the so-called distributed
space–time codes have become a prominent area of research. In practice, such
codes often exhibit a high decoding complexity. Algebraic number theory and
lattice theory provide a framework for overcoming this issue. In Chap.3,the
authors give an overview on the topic of fast decodable algebraic space–time codes.
More precisely, the chapter provides abasic introduction to space–time coding
and brings to the forefront the powerful algebraic tools needed for the design and
construction of such codes. In particular, it describes the algebraic techniques used
for reducing the decoding complexity of both single-user and multiuser space–time
codes. The key lies in utilizing the carefullychosen underlying algebraic structure.
The necessary background to both the lattice theory and algebraic number theory
is provided. The chapter concludes by describing explicit construction methods
for fast-decodable algebraic space–time codes. These are crucial for practical
implementation.
The problem of finding the densest arrangement of spheres in ann-dimensional
Euclidean space has been extensively pursued in mathematics. It is a classical and
central problem in the geometry of numbers. The celebrated Minkowski–Hlawka
theorem (dating back to 1943) states that there is a lattice inR
n
, such that the
corresponding best packing of spheres withcentres at the lattice points has density
greater thanζ(n)2
−(n−1)
—a constant dependent on the dimension. This is an
existence statement, and to date, excluding a handful of dimensions, no explicit
lattice construction achieving the Minkowski–Hlawka lower bound for the sphere
packing density is known. The proof usesprobabilistic methods to analyse a random
ensemble of lattices rather than individual instances. Recent improvements to the
Minkowski-Hlawka lower bound exploit lattices with inherited algebraic structures.
For example, Venkatesh has successfully used the structure of cyclotomic number
field lattices to obtain a super-linear improvement. The sphere packing problem
has well-established and deep connections to coding theory. Indeed, building upon
the Shannon’s seminal work from 1948, it was shown in the nineties that the same
random ensembles used to produce lattices that give rise to the Minkowski–Hlawka
lower bound can be used to construct optimal lattice codes for the basic additive
white Gaussian noise (AWGN) communication channels. In other words, for such
channels the probabalistic strategy of Minkowski and Hlawka leads to the existence

viii Preface
of capacity-achieving codes. Although the AWGN channel is a good model for
deep-space or satellite channels, which operate over a line of sight, modern wireless
communications call for more general models, which take into account the sending
of information propagated over different media (e.g. fading channels) via multiple
transmit and receive antennas (e.g. MIMO channels) and to various users (e.g. relay
networks). Such channels cannot be abstracted into a simple AWGN model and
require a different strategy. With this in mind, it is fruitful to consider lattices with
additional algebraic (multiplicative) structure, often inherited by the properties of
number fields. Indeed, the cyclotomic lattices exploited by Venkatesh play a crucial
role in some recent channel constructions. In Chap.4, the authors start by providing
a self-contained exposition of random lattices and the sphere packing problem.
This includes both the classical aspects and the recent developments. Regarding the
latter, the use of algebraic number theory to utilize the structure of algebraic lattices
is brought to the forefront. A general construction that naturally incorporates a
number of important families of algebraic lattices (such as cyclotomic, Lipschitz and
Hurwitz lattices) is described. The emphasis then switches to describing how such
lattices can be applied to build effective, reliable and secure transmission schemes
for wireless communications. The main focus on the application side is threefold:
(i) to block fading, (ii) to certain forms of MIMO channels and (iii) to improving
information security.
As alluded to in the discussion above, one of the classical problems in informa-
tion and coding theory is that of designing codes that can approach the capacity of
the AWGN channel. One promising approach is to draw codewords from a lattice
and draw upon deep results from the geometry of numbers to establish performance
bounds. However, the AWGN channel model is not sufficient for modelling the
phenomena observed in wireless communication scenarios. Recent efforts have thus
shifted towards designing codes that can approach the capacity of fading channels,
which model the wireless medium via a random matrix multiplication of the channel
input followed by the addition of Gaussian noise. Here, it is also of interest to design
good lattice codebooks that can operate near the capacity. In Chap.5, the authors
begin by reviewing the Hermite invariant approach to the design of lattice codes for
classical AWGN channels. They then propose a reduced Hermite invariant criterion
for the design of lattice codes for fading channels. Using this criterion, they are able
to translate the problem of operating within a constant gap of the fading capacity to
the problem of finding totally complex number fields with the smallest determinant.
Drawing upon powerful results from this area, they demonstrate the existence of
lattice codebooks that can operate within a constant gap of the fading capacity.
They then discuss the limitations of this approach and outline a promising direction
based on the construction of lattice codes from ideals. They conclude the chapter
with a discussion on the connection between the reduced Hermite invariant and
homogeneous forms.
A key property of the wireless medium is that a receiver’s observation can be
written as a linear superposition of all transmitted signals and Gaussian noise. By
employing codebooks based on nested lattices at each transmitter, this property
can be leveraged in order to allow the receiver to directly recover a function of

Preface ix
the codewords. The compute-and-forward framework was proposed to characterize
the achievable rates for recovering integer-linear combinations, and lattice network
coding was subsequently proposed to connect this framework to module theory from
abstract algebra, which in turns allows for a much richer set of lattice codebook
constructions. In Chap.6, the authors provide a comprehensive overview of these
frameworks, beginning with a review of necessary concepts from abstract algebra,
lattice codes and classical construction. Furthermore, methodsfor obtaining per-
formance bounds for compute-and-forward are also discussed. They then propose
multilevel lattice codes as a powerful method for reducing the decoding complexity
while maintaining the performance advantages of lattice codes. They introduce
detailed procedures for constructing such multilevel lattices, including a novel
elementary divisor construction, which captures prior methods as special cases.
From here, they generalize compute-and-forward and lattice network coding to
utilize multilevel lattices and demonstrate that this approach can yield a more
efficient method for decoding multiplemessages. They conclude by proposing
an iterative decoding procedure for multilevel lattice codes and demonstrate its
advantages via numerical simulations.
Shannon’s beautiful theorem concerning the existence of capacity-achieving
codes for an AWGN channel makes fundamental use of a random coding argument.
In short, independent identically distributed (i.i.d.) random ensembles according
to some codeword distribution are exploited to prove the existence of “optimal”
codes. In Chap.7, the final chapter, the authors begin by reviewing the main
steps in Shannon’s proof, in particular the use of the i.i.d. random ensembles in
the achievability part. They then revisit the achievability part from the viewpoint
of exploiting random structured ensembles such as random linear codes and
random lattice codes. For certain scenarios (e.g. those involving relay networks
or physical layer secrecy), random structured codes achieve better “rates” than
random i.i.d. codes. Furthermore, randomlinear codes allow for computationally
efficient encoding (since the encoding operation essentially involves simple matrix
multiplication), and random lattice codes allow for lattice decoding, which, for
example, enjoy lower complexity than maximum likelihood (ML
main goal of the chapter is to provide anaccessible account of recent developments
and simplifications in the use of random structured codes in achievability proofs.
The focus is on addressing the two questions:Can random linear codes achieve
the discrete memoryless channel (DMC) capacity?and,Can random lattice codes
achieve the additive white Gaussian noise (AWGN) channel capacity?These two
questions are discussed separately but ina parallel manner. Indeed, the introduced
framework unifies the approaches for DMC and AWGN channels into a streamlined
analysis.
York, UK Victor Beresnevich
York, UK Alister Burr
Boston, MA, USA Bobak Nazer
York, UK Sanju Velani
June 2020

x Preface
Acknowledgements
As part of the organizing committee of the 2016 workshop, we would like to thank
all the people who helped to make this meeting highly enjoyable and successful.
We are also extremely grateful to everybody who made this volume possible. In
particular, we would like to acknowledge the help of:
• Dr. Rémi Lodh at Springer for all his encouragement, patience, good humour
and fantastic support in producing this book—we have truly enjoyed working
with you;
• the distinguished speakers for their inspiring talks;
• all the contributors to this volume, fortheir hard work in producing high-quality
chapters and for having faith in us during its 3-year gestation period;
• all the reviewers of the chapters, for their work in producing informative reports,
which have helped improve the accuracy and clarity of the material presented;
• all the participants of the workshop for creating an informal and inspiring
atmosphere.
Finally, we would like to stress that the workshop was fully funded from the EPSRC
Programme Grant: EP/J018260/1. The workshop and in turn this book would not
have been possible without this generous support.

Contents
1 Number Theory Meets Wireless Communications: An
Introduction for Dummies Like Us........................................ 1
Victor Beresnevich and Sanju Velani
2 Characterizing the Performance of Wireless Communication
Architectures via Basic Diophantine Approximation Bounds.......... 69
Bobak Nazer and Or Ordentlich
3 On Fast-Decodable Algebraic Space–Time Codes....................... 99
Amaro Barreal and Camilla Hollanti
4 Random Algebraic Lattices and Codes for Wireless
Communications............................................................. 143
Antonio Campello and Cong Ling
5 Algebraic Lattice Codes for Linear Fading Channels................... 179
Roope Vehkalahti and Laura Luzzi
6 Multilevel Lattices for Compute-and-Forward and Lattice
Network Coding.............................................................. 201
Yi Wang, Yu-Chih Huang, Alister G. Burr, and Krishna R. Narayanan
7 Nested Linear/Lattice Codes Revisited.................................... 241
Renming Qi and Chen Feng
xi

Contributors
Amaro BarrealDepartment of Mathematics and Systems Analysis, Aalto Univer-
sity, Aalto, Finland
Victor BeresnevichDepartment of Mathematics, University of York, York, UK
Alister G. BurrDepartment of Electronic Engineering, University of York, York,
UK
Antonio CampelloWellcome Trust, London, UK
Chen FengThe School of Engineering, University of British Columbia (Okanagan
Campus), Kelowna, BC, Canada
Camilla HollantiDepartment of Mathematics and Systems Analysis, Aalto Uni-
versity, Aalto, Finland
Yu-Chih (Jerry) HuangInstitute of Communications Engineering, National
Chiao Tung University, Hsinchu City, Taiwan
Cong LingDepartment of Electrical and Electronic Engineering, Imperial College
London, London, UK
Laura LuzziLaboratoire ETIS (UMR 8051, CY Université, ENSEA, CNRS)
Cergy-Pontoise, France
Krishna R. NarayananDepartment of Electrical & Computer Engendering, Texas
A&M University, College Station, TX, USA
Bobak NazerDepartment of Electrical and Computer Engineering, Boston Uni-
versity, Boston, MA, USA
Or OrdentlichThe Rachel and Selim Benin School of Computer Science, Engi-
neering Hebrew University of Jerusalem, Jerusalem, Israel
Renming QiThe School of Engineering, University of British Columbia (Okana-
gan Campus), Kelowna, BC, Canada
xiii

xiv Contributors
Roope VehkalahtiDepartment of Communications and Networking, Aalto Uni-
versity, Helsinki, Finland
Sanju VelaniDepartment of Mathematics, University of York, York, UK
Yi WangDepartment of Electronic Engineering, University of York, York, UK

Chapter 1
Number Theory Meets Wireless
Communications: An Introduction for
Dummies Like Us
Victor Beresnevich and Sanju Velani
AbstractIn this chapter we introduce the theory of Diophantine approximation
via a series of basic examples from information theory relevant to wireless
communications. In particular, we discuss Dirichlet’s theorem, badly approximable
points, Dirichlet improvable and singular points, the metric (probabilistic) theory
of Diophantine approximation including the Khintchine-Groshev theorem and the
theory of Diophantine approximation on manifolds. We explore various number
theoretic approaches used in the analysis of communication characteristics such as
Degrees of Freedom (DoF). In particular, we improve the result of Motahari et al.
regarding the DoF of a two-user X-channel. In essence, we show that the total DoF
can be achieved for all (rather than almostall) choices of channel coefficients with
the exception of a subset of strictly smaller dimension than the ambient space. The
improvement utilises the concept of jointly non-singular points that we introduce
and a general result of Kadyrov et al. on theδ-escape of mass in the space of lattices.
We also discuss follow-up open problems that incorporate a breakthrough of Cheung
and more generally Das et al. on the dimension of the set of singular points.
1.1 Basic Examples and Fundamentals of Diophantine
Approximation
Let us start by addressing a natural question that a number theorist or more
generally a mathematician who has picked up this book may well ask:what is
the role of number theory in the world of wireless communications?We will
come clean straightaway and say that bynumber theory we essentially mean areas
such as Diophantine approximation and the geometry of numbers, and by wireless
communication we essentially mean the design and analysis of lattice/linear codes
for wireless communications which thus falls in the realm of information theory. To
V. B e r e s n ev i c h (ζ)∙S.Velani
Department of Mathematics, University of York, York, UK
e-mail:[email protected];[email protected]
© Springer Nature Switzerland AG 2020
V. Beresnevich et al. (eds.),Number Theory Meets Wireless Communications,
Mathematical Engineering,https://doi.org/10.1007/978-3-030-61303-7_1
1

2 V. Beresnevich and S. Velani
begin with, with this confession in mind, let us start by describing the role of one-
dimensional Diophantine approximation. Recall, that at the heart of Diophantine
approximation is the classical theorem of Dirichlet on rational approximations to
real numbers.
Theorem 1.1 (Dirichlet, 1842)For anyξ∈Rand anyQ∈Nthere existp, q∈Z
such that




ξ−
p
q




<
1
qQ
and 1≤q≤Q. (1.1
The proof can be found in many elementary number theory books and makes use
of the wonderfully simple yet powerful Pigeonhole Principle: ifnobjects are placed
inmboxes andn>m, then some box will contain at least two objects. See, for
example, [16, §1.1] for details. An easy consequence of the above theorem is the
following statement.
Corollary 1.1Letξ∈R\Q, that isξis a real irrational number. Then there exist
infinitely many reduced rational fractionsp/q (p, q∈Z)such that




ξ−
p
q




<
1
q
2
. (1.2
The following exposition illustrates one of the many aspects of the role of
Diophantine approximation in wireless communication. In particular, within this
section we consider a basic example of a communication channel which brings into
play the theory of Diophantine approximation. In Sect.1.2we consider a slightly
more sophisticated example which also brings into play the theory of Diophantine
approximation in higher dimensions. This naturally feeds into Sect.1.3in which
the role of the theory of Diophantine approximation of dependent variables is
discussed. The latter is also referred to as Diophantine approximation on manifolds
since the parameters of interest are confined by some functional relations. To
begin with, we consider a ‘baby’ example of a communication channel intended
to remove the language barrier for mathematicians and explicitly expose an aspect
of communications that invites the use of Diophantine approximation.
1.1.1 A ‘baby’ Example
Suppose there are twousersS 1andS 2wishing to send (transmit)theirmessages
u
1andu 2respectively along a shared (radio/wireless) communication channel to
areceiverR. For obvious reasons, users are often also referred to as transmitters.
Suppose for simplicity thatu
1,u2∈{0,1}. Typically, prior to transmission, every
message is encoded with what is called acodeword. Suppose thatx
1=x1(u1)
andx
2=x2(u2)are the codewords that correspond tou 1andu 2. In general,x 1

1 Number Theory Meets Wireless Communications: An Introduction 3
andx 2could be any functions on the set of messages. In principle, one can take
x
1=u1andx 2=u2. When the codewordsx 1andx 2are being transmitted along
a wireless communication channel, there is normally a certain degree of fading of
the transmitted signals. This for instance could be dependent on the distance of
the transmitters from the receiver and the reflection caused by obstacles such as
buildings in the path of the signal. Leth
1andh 2denote the fading factors (often
referred to aschannel gainsorchannel coefficientsorpaths loss) associated with
the transmission of signals fromS
1andS 2toRrespectively. These are strictly
positive numbers and for simplicity we will assume that their sum is one:h
1+
h
2=1. Mathematically, the meaning of thechannel coefficients is as follows: if
S
itransmits signalx i, the receiverRobservesh ixi. However, due to fundamental
physical properties of wireless medium, whenS
1andS 2simultaneously use the
same wireless communication channel,Rwill receive the superposition ofh
1x1and
h
2x2,thatis
y=h
1x1+h2x2. (1.3)
For instance, assuming thatx
1=u1andx 2=u2, the outcomes ofyare
y=











0if u
1=u2=0
h
1 ifu 1=0andu 2=1,
h
2 ifu 1=1andu 2=0,
1=h
1+h2ifu 1=u2=1.
(1.4)
A pictorial description of the above setup is given below in Fig.1.1.
The ultimate goal is for the receiverRto identify (decode) the messagesu
1and
u
2from the observation ofy. For example, with reference to (1.4), assuming the
channel coefficientsh
1andh 2are known at the receiver and are different, that is
h
1⎪=h2, the receiver is obviously able to do so. However, in real life there is
∈∈
x1
∈∈
∈∈
x2
















S1
S2
R
h1
h
y2
Fig. 1.1Two user multipleaccess channel (no noise)

4 V. Beresnevich and S. Velani
10
h
1 h2
) (()()
α≥
z
Fig. 1.2Separation of intervals of radius|z|around each possible outcome ofywhich contain the
values ofy
α
always a degree of error in the transmission process, predominantly caused by the
received signalybeing corrupted by (additive)noise. The noise can result from a
combinations of various factors including the interference of other users and natural
electromagnetic radiation. In short, ifzdenotes the noise, then instead of (1.3),R
receives the signal
y
α
=y+z=h 1x1+h2x2+z. (1.5)
Equation (1.5) represents one the simplest models of what is known as anAdditive
White Gaussian Noise Multiple Access Channel(AWGN-MAC), see Chap.2for
a formal definition. As before, the goal for the receiverRremains to decode the
messagesu
1andu 2, but now from the observation ofy
α
=y+z.Letd mindenote
theminimum distancebetween the four outcomes ofy. Then as long as the absolute
value|z|of the noise is strictly less thand
min/2, the receiver is able to recovery
and consequently the messagesu
1andu 2from the value ofy
α
. This is simply due
to the fact that the intervals of radiusd
min/2 centered at the four outcomes ofyare
disjoint andy
α
will lie in exactly one of these intervals, see Fig.1.2.Inotherwords,
Ris able to identifyyby roundingy
α
to the closest possible outcome ofy.
For example, it is easy to see that the maximum separation between the four
outcomes given by (1.4) is attained whenh
1=1/3andh 2=2/3. In this
cased
min=1/3, and we are able to recover the messagesu 1andu 2assuming
that|z|<1/6. The upshot of the above discussion is the following simple but
fundamental conclusion.
ConclusionThe greater the mutual separationd
minof the outcomes ofy, the better
the tolerance for noise we have during the transmission of the signal.
In information theory achieving goodseparation between received signals
translates into obtaining good lower bounds on the fundamental parameters of
communication channels such as Rates-of-Communications, Channel Capacity and
Degrees-of-Freedom, see Chap.2for formal definitions of these notions. Within this
chapter we will concentrate on the role of Diophantine approximation in answering
the following natural and important question:
How can a good separation of received signals be achieved and how often?
Indeed, to some extent, answering this and related questions using the tools of
Diophantine approximation, algebraic number theory and the geometry of numbers
is a reoccurring theme throughout the whole book. We will solely uselinear
encodingto achieve ‘good’ separation. In particular, within the above ‘baby’

1 Number Theory Meets Wireless Communications: An Introduction 5
example, one is able to achieve the optimal separation (d min=1/3) at the receiver
regardless of the values ofh
1andh 2by applying the following simple linear
encoding of the messagesu
1andu 2:
x
1=
1
3
h
−1
1
u1 andx 2=
2
3
h
−1
2
u2.
Indeed, before taking noise into consideration, under the above encoding the
received signals become
y=h
1x1+h2x2=
1
3
u1+
2
3
u2=











0if u
1=u2=0,
1/3if u
1=0andu 2=1,
2/3if u
1=1andu 2=0,
1if u
1=u2=1.
(1.6
To summarise, the above discussion brings to the forefront the importance
of maximizing the minimal distance/separationd
minof the received (noise-free)
signals and at the same time indicates how a linear encoding allows us to achieve
this. Nevertheless, the assumption that the messagesu
1andu 2beingsentbythe
transmittersS
1andS 2are binary in nature makes the discussion over simplistic—
especially in terms of the use of number theory to analyse the outcomes. We now
modify the ‘baby’ example to a more general situation in whichS
1andS 2wish to
send messagesu
1andu 2from the set of integers{0,...,Q}to a single receiverR.
1.1.2 Example 1 (Modified ‘baby’ Example)
Unless stated otherwise, here and throughout,Q≥2 is a fixed integer. As we shall
see, this slightly more complex setup, in whichu
1,u2∈{0,...,Q}, naturally bring
into play the rich theory of Diophantine approximation. So with this in mind, let us
assume that the codewordsx
1andx 2that are being transmitted byS 1andS 2are
simply obtained by the linear encoding of the messagesu
1andu 2as follows
x
1=αu1andx 2=βu2 (0≤u 1,u2≤Q) , (1.7
whereαandβare some positive real numbers. We emphasise that the parametersα
andβare at our disposal and this fact will be utilized later. As in the ‘baby’ example
leth
1andh 2denote the channel coefficients associated withS 1andS 2respectively.
Then, before taking noise into account,Rwill receive the signal
y=h
1x1+h2x2=h1αu1+h2βu2. (1.8

6 V. Beresnevich and S. Velani
Clearly,ytakes the values
h
1αu1+h2βu2:0≤u 1,u2≤Q. (1.9
Thus, there are potentially(Q+1)
2
distinct outcomes ofyand they lie in the interval
[0,(h
1α+h 2β)Q]. It is easily verified that if they were equally separated then their
mutual separation would be precisely
h
1α+h 2β
Q+2
. (1.10
However, this is essentially never the case. Indeed, letd
mindenote the minimal
distance between the pointsygivenby(1.9). Without loss of generality, suppose
for the sake of simplicity that
0<h
1α<h2β
and define the real number
ξ:=
h

h2β
, (1.11
which in view of the above assumption is between 0 and 1;i.e.0<ξ<1. Then, by
Dirichlet’s theorem, we have that
ζ
ζ
ζ
ζ
ξ−
p
q
ζ
ζ
ζ
ζ

1
qQ
(1.12
for an integer pair(p, q)∈Z
2
satisfying 1≤q≤Q.Since0<ξ< 1and
1≤q≤Q,wealsohavethat0≤p≤q≤Q. On multiplying (1.12)byh
2βq,we
find that
|h
1αq−h 2βp|≤
C
1
Q
(C
1:=h2β), (1.13
for some integer pair(p, q)∈Z
2
satisfying 1≤q≤Qand 0≤p≤q.Now
observe that the quantity|h
1αq−h 2βp|on the left hand side of (1.13) is exactly
the distance between the two specific values ofywithin (1.9) corresponding to
u
1=q,u2=0andu 1=0,u 2=p.Sinceqξ=0, this demonstrates that the
minimal distanced
minbetween the values ofygiven by (1.9) is always bounded
above byC
1/Q;i.e.
d
min≤
C
1
Q
. (1.14

1 Number Theory Meets Wireless Communications: An Introduction 7
For all intents and purposes, this bound on the minimal distance is smaller than the
hypothetical ‘perfect’ separation given by (1.10). In general, we have that
d
min≤min
λ
C
1
Q
,
h
1α+h 2β
Q+2
π
.
It is easily seen that we can remove the assumption that 0<h
1α<h2βif we put
C
1=max{h 1α, h2β}.
Remark 1.1On looking at (1.14), the reader may be concerned (rightly) that the
minimal distanced
minvanishes asQgrows. Luckily, this can be easily rectified
by introducing a scaling factorλ≥1 into the linear encoding of the messagesu
1
andu 2. The point of doing this is that the codewordx 1(resp.x 2)givenby(1.7)
becomesλαu
1(resp.λβ u 2) and this has no effect on the point of interestξgiven by
(1.11) but it scales up byλthe constantC
1appearing in (1.13). Thus, by choosing
λappropriately (namely, proportional toQ) we can avoid the right hand side of
(1.14) from vanishing asQgrows. In subsequent more ‘sophisticated’ examples,
the scaling factor will be relevant to the discussion and will appear at the point of
linear encoding the messages.
Now let us bring noise into the above setup. As in the ‘baby’ example, ifzdenotes
the (additive) noise, then instead of (1.8),Rreceives the signal
y
α
=y+z=h 1αu1+h2βu2+z. (1.15
Note that as long as the absolute value|z|of the noise is strictly less thand
min/2,
the receiverRis able to recoveryand consequentlyu
1andu 2from the value
ofy
α
. Commonly, the nature of noise is such thatzis a random variable having
normal distribution. Without loss of generality we will assume thatz∼N(0,1),
that is the mean value of noise is 0 and itsvariance is 1. Therefore, when taking
the randomness of noise into account, the problem of whether or not the receiver is
able to recover messages sent by the transmitters becomes probabilistic in nature.
Loosely speaking, we are interested in the probability that|z|<d
min/2—the larger
the probability the more likely the receiver is able to recover messages by rounding
y
α
to the closest possible outcome ofy. Of course, if it happens that|z|≥d min/2,
then we will have an error in the recovery ofyand thus the messagesu
1andu 2.
Whenz∼N(0,1), the probability of this error can be computed using the Gauss
error function and is explicitly equal to
1−

2/π

dmin/2
0
e
−θ
2
/2
dθ .
This gets smaller asd
mingets larger. Clearly, in view of the theoretic upper bound
ond
mingiven by (1.14) the probabilityof error is bounded above by the probability
that|z|<C
1/2Q. Thus, the closerd minis to the theoretic upper bound, the closer
we are to minimizing the probability of the error and in turn the higher the threshold
for tolerating noise. With this in mind, we now demonstrate that on appropriately

8 V. Beresnevich and S. Velani
choosing the parametersαandβassociated with the encoding procedure it is
possible to get within a constant factor of the theoretic upper bound.
1.1.3 Badly Approximable Numbers
The key is to make use of the existence of badly approximable numbers—a
fundamental class of real numbers in the theory of Diophantine approximation.
Definition 1.1 (Badly Approximable Numbers)A real numberξis said to be
badly approximableif there exists a constantκ=κ(ξ) >0 such that for all
q∈N,p∈Z




ξ−
p
q





κ
q
2
. (1.16
Note that by definition, badly approximable numbers are precisely those real
numbers for which the right hand side of inequality (1.2) associated with Dirichlet’s
corollary (Corollary1.1) cannot be ‘improved’ by an arbitrary constant factor. By
Hurwitz’s theorem [16], ifξis badly approximable then for the associated badly
approximable constantκ(ξ)we have that
0<κ(ξ)<1/

5.
It is well known that the set of badly approximable numbers can be characterized as
those real numbers whose continued fraction expansions have bounded partial quo-
tients. Moreover, an irrational number has a periodic continued fraction expansion
if and only if it is a quadratic irrational andthus every quadratic irrational is badly
approximable. In particular, it is easily verified that for any givenε>0, the golden
ratio
γ:=(

5+1)/2
satisfies inequality (1.16) withκ=1/(

5+ε)for allp∈Zandq∈Nwithq
2

1/(

5ε). This is obtained using the standard argument that involves substituting
p/qinto the minimal polynomialfofγoverZand using the obvious fact that
1≤q
2
|f(p/q)|≤q
2
|γ−p/q|·|¯γ−p/q|,where¯γ=(

5−1)/2 is the conjugate
ofγ. We leave further computational details to the reader. Observe that on taking
ε=1/

5, we find thatγis badly approximable withκ(γ)≥

5/6.
The reason for us bringing into play the notion of badly approximable numbers
is very easy to explain. By definition, on choosing the parametersαandβso that
ξ:=h
1α/h2βis badly approximable guarantees the existence of a constantκ(ξ) >
0suchthat
|h
1αq−h 2βp|≥κ(ξ)
C
1
q
∀q∈N,p∈Z.

1 Number Theory Meets Wireless Communications: An Introduction 9
Thus, it follows that the separation between the points given by (1.9) is at least
κ(ξ)C
1/Q. In other words, the minimal distanced minis within a constant factor of
the theoretic upper boundC
1/Qgiven by (1.14). Indeed, if we chooseαandβso
thath
1α/h2βis the golden ratioγwe obtain that
κ(γ)
C
1
Q
≤d
min≤
C
1
Q
. (1.17
The upshot is that equation (1.17) gives an explicit ‘safe’ threshold for the level
of noise that can be tolerated. Namely, the probability that|z|<d
min/2 is at least
the probability that|z|< κ(γ )C
1/Q. In principle, one can manipulate the values
ofQ∈Nandε>0 within the above argument to improve the lower bound
in (1.17). However, any such manipulation will not enable us to surpass the hard
lower bound limit ofC
1/(

5Q)imposed by the aforementioned consequence of
Hurwitz’s theorem. Therefore, we now explore a different approach in an attempt
to make improvements to (1.17) beyond this hard limit. Ideally, we would like to
replace 1/

5 by a constant arbitrarily close to one. We would also like to move
away from insisting thatξis badly approximable since this is a rare event. Indeed,
although the set of badly approximable number is of full Hausdorff dimension (a
result of Jarník from the 1920s), it is a set of Lebesgue measure zero (a result of
Borel from 1908). In other words, the (uniform) probability that a real number in
the unit interval is badly approximable is zero. We will return to this in Sects.1.2.2
and1.2.7below.
1.1.4 Probabilistic Aspects
The approach we now pursue is motivated by the following probabilistic problem:
Given0<κ

<1andQ∈N, what is the probability that a given real number
ξ∈I:=(0,1)satisfies




ξ−
p q





κ

qQ
(1.18
for all integerspand1≤q≤Q?Note that these are the real numbers for which
the right hand side of inequality (1.1) associated with Dirichlet’s theorem cannot be
improved by the factor ofκ

(Qis fixed here). It is worth mentioning at this point, in
order to avoid confusion later, that these real numbers are not the same as Dirichlet
non-improvable numbers which will be introduced below in Sect.1.1.5. To estimate
the probability in question, we consider the complementary inequality




ξ−
p
q




<
κ

qQ
. (1.19

10 V. Beresnevich and S. Velani
Let 1≤q≤Q. Then for a fixedq, the probability that a givenξ∈I:=(0,1)
satisfies (1.19)forsomep∈Zis exactly 2κ

/Q—it corresponds to the measure of
the set
E
q:=

p∈Z

p
q

κ

qQ
,
p
q
+
κ

qQ

∩I.
On summing up these probabilities overq, we conclude that the probability that
agivenξ∈Isatisfies (1.19) for some integerspand 1≤q≤Qis trivially
bounded above by 2κ

. This in turn implies that for anyκ

<1/2andanyQ∈N
the probability that (1.18) holds for all integersp, qwith 1≤q≤Qis at least
1−2κ

.
The following result shows that with a little more extra work it is possible to improve
this trivial bound.
Lemma 1.1For any0<κ

<1and anyQ∈Nthe probability that(1.18)holds
for all integersp, qwith1≤q≤Qis at least
1−
12κ

π
2
≈1−1.216κ

. (1.20
Remark 1.2Observe that when
κ


2
/12≈0.822,
the quantity 12κ


2
is strictly less than 1 and therefore the probability given by
(1.20) is greater than zero. Hence for anyQ∈N, there exist real numbersξ
satisfying (1.18) for all integerspand 1≤q≤Q.
Remark 1.3Within Lemma1.1the word ‘probability’ refers to the uniform proba-
bility over[0,1]. However, in real world applications the parameterξappearing in
(1.18) may not necessarily be a uniformly distributed random variable. For instance,
the channel coefficients could be subject to Rayleigh distribution and this will
have an obvious effect on the distribution ofξvia (1.11). Nevertheless, as long
as the distribution ofξis absolutely continuous, a version of Lemma1.1can be
established, albeit the constant that accompaniesκ

will be different. For further
details we refer the reader to [1].
ProofThe proof of Lemma1.1relies on ‘removing’ the overlaps between the
different setsE
qasqvaries. Indeed, it is easily seen that
E:=
Q

q=1
Eq=
Q

q=1

0≤p≤q
gcd(p,q)=1

p
q

κ

qQ
,
p
q
+
κ

qQ

∩I.

1 Number Theory Meets Wireless Communications: An Introduction 11
Therefore,
Prob(E)≤
Q
ρ
q=1
ρ
1≤p≤q
gcd(p,q)=1


qQ
=
Q
ρ
q=1


ϕ(q)
qQ
=


Q
Q
ρ
q=1
ϕ(q)
q
, (1.21
whereϕis the Euler function. To estimate the above sum, it is convenient to use the
Möbius inversion formula, which gives that
ϕ(q)
q
=
ρ
d|q
μ(d)
d
whereμis theMöbius function. Recall that

ρ
d=1
μ(d)
d
2
=
1
ζ(2)
=
6
π
2
.
Then
Q
ρ
q=1
ϕ(q)
q
=
Q
ρ
q=1
ρ
d|q
μ(d)
d
=
Q
ρ
q=1
ρ
dd

=q
μ(d)
d
=
ρ
dd

≤Q
μ(d)
d
=
ρ
1≤d≤Q
μ(d)
d
ρ
d

≤Q/d
1
=
ρ
1≤d≤Q
μ(d) d
[Q/d]≤Q
ρ
1≤d≤Q
μ(d)
d
2

6Q
π
2
.
Combining this with (1.21) gives the required estimate, that is a lower bound on
1−Prob(E), the probability of the complement toE. ρ
Let 0<κ


2
/12 andQ∈Nbe given. The upshot of the above discussion
is that there exist parametersαandβso that with probability greater than 1−
12κ


2
>0, the real numberξ:=h 1α/h2βsatisfies (1.18) for all integerspand
1≤q≤Q. It follows that for suchξ(or equivalently parametersαandβ)the
separation between the associated points given by (1.9) is at leastκ

C1/Qand so
the minimal distanced
minsatisfies
κ

C1
Q
≤d
min≤
C
1
Q
. (1.22

12 V. Beresnevich and S. Velani
In particular, we can chooseκ

so thatκ(γ) < κ

in which case the lower bound in
(1.22) is better than that in (1.17) obtained by making use of badly approximable
numbers. That is to say, that the lower bound involvingκ

is closer to the theoretic
upper boundC
1/Q. Moreover, the set of badly approximable numbers is a set of
measure zero whereas the set of real numbers satisfying (1.18) for all integersp
and 1≤q≤Qhas Lebesgue measure at least 1−12κ


2
. This is an important
advantage of the probabilistic approach since in reality it is often the case that the
channel coefficientsh
1andh 2are random in nature. For example, when dealing
with mobile networks one has to take into consideration the obvious fact that the
transmitters are not fixed. The upshot is that in such a scenario, we do not have
the luxury of specifying a particular choice of the parametersαandβthat leads
to the corresponding points given by (1.9) being well separated as in the sense of
(1.17). The probabilistic approach provides a way out. In short, it enables us to
ensure that the minimal distanced
minbetween the points given by (1.9) satisfies
(1.22) with good (explicitly computable) probability. See [54, Section VI.B] for a
concrete example where the above probabilistic approach is used for the analysis of
the capacity of symmetric Gaussian multi-user interference channels.
Up to this point,Qhas been a fixed integer greater than or equal to 2 and reflects
the size of the set of messages. We end our discussion revolving around Example 1
by considering the scenario in which we have complete freedom in choosingQ.In
particular, one is often interested in the effect of allowingQto tend to infinity on the
model under consideration. This is relevant to understanding the so-called Degrees
of Freedom (DoF) of communication channels, see Sect.1.2.4.
1.1.5 Dirichlet Improvable and Non-improvable Numbers
We now show that there are special values ofQfor which the minimal distanced min
satisfies (1.22) withκ

as close to one as desired. The key is to exploit the (abundant)
existence of numbers for which Dirichlet’s theorem cannot be improved. Note that in
the argument leading to (1.17) we made use of the existence of badly approximable
numbers; that is numbers for with Dirichlet’s corollary cannot be improved.
Definition 1.2 (Dirichlet Improvable and Non-improvable Numbers)Let 0<
κ

<1. A real numberξis said to beκ

-Dirichlet improvableif for all sufficiently
largeQ∈Nthere are integerspand 1≤q≤Qsuch that




ξ−
p
q




<
κ

qQ
.
A real numberξis said to beDirichlet non-improvableif for anyκ

<1 it is notκ

-
Dirichlet improvable. In other words, a real numberξisDirichlet non-improvable
if for any 0<κ

<1 there exists arbitrarily largeQ∈Nsuch that for all integersp

1 Number Theory Meets Wireless Communications: An Introduction 13
and 1≤q≤Q




ξ−
p
q





κ

qQ
.
A well know result of Davenport & Schmidt [28] states that:
a real number is Dirichlet non−improvable
σ
it is not badly approximable.
Consequently, a randomly picked real number is Dirichlet non-improvable with
probability one. The upshot of this is the following remarkable consequence: for
any random choice of channel coefficientsh
1,h2and parametersα, β,with
probability onefor anyε>0 there exist arbitrarily large integersQsuch that
the minimal distanced
minbetween the associated points given by (1.9) satisfies
(1−ε)
C
1
Q
≤d
min≤
C
1
Q
.
Clearly, this is the best possible outcome for the basic wireless communication
model considered in Example 1. We now consider a slightly more sophisticated
model which demonstrates the role of higher dimensional Diophantine approxima-
tion in wireless communication.
1.2 A ‘toddler’ Example and Diophantine Approximation in
Higher Dimensions
The discussion in this section is centred on analysing the model arising from adding
another receiver within the setup of the modified ‘baby’ example.
1.2.1 Example 2
Suppose there are two usersS 1andS 2as in Example 1 but this time there are
also two receiversR
1andR 2.LetQ≥1 be an integer and supposeS 1wishes
to simultaneously transmit independent messagesu
1,v1∈{0,...,Q},whereu 1
is intended forR 1andv 1forR 2. Similarly, supposeS 2wishes to simultaneously
transmit independent messagesu
2,v2∈{0,...,Q},whereu 2is intended for

14 V. Beresnevich and S. Velani
R1andv 2forR 2. After (linear) encoding,S 1transmitsx 1:=x 1(u1,v1)andS 2
transmitsx 2:=x2(u2,v2);thatistosay
x
1=α1u1+β1v1andx 2=α2u2+β2v2 (1.23)
whereα
1,α2,β1andβ 2are some positive real numbers. Next, fori, j=1,2, let
h
ijdenote the channel coefficients associated with the transmission of signals from
S
jtoRi. Also, lety idenote the signal received byR ibefore noise is taken into
account. Thus,
y
1=h11x1+h12x2, (1.24)
y
2=h21x1+h22x2. (1.25)
A pictorial description of the above setup is given in Fig.1.3below.
Substituting (1.23)into(1.24)and(1.25) gives that
y
1=h11α1u1+h11β1v1+h12α2u2+h12β2v2, (1.26)
y
2=h21α1u1+h21β1v1+h22α2u2+h22β2v2. (1.27)
Note that there are potentially(Q+1)
4
distinct outcomes ofy iand they lie in the
interval[0,(h
i1α1+hi1β1+hi2α2+hi2β2)Q].
Now let us bring noise into the setup. Ifz
idenotes the (additive) noise at receiver
R
i(i=1,2), then instead of (1.26)and(1.27),R 1andR 2receive the signals
y
α
1
=y1+z1 andy
α
2
=y2+z2 (1.28)
respectively. Equations (1.23)–(1.28) represent one of the simplest models of what
is known as atwo-userX-channel. The ultimate goal is for the receiverR
1to decode
∈∈
x1
∈∈
1
∈∈
x2
∈∈
2



∈∈




α
α
S1
S2
R1
R2
h11
h22
h
y
y12
h21Fig. 1.3Two-userX-channel

1 Number Theory Meets Wireless Communications: An Introduction 15
the messagesu 1andu 2from the observation ofy
α
1
and for the receiverR 2to decode
the messagesv
1andv 2from the observation ofy
α
2
. Clearly, this goal is attainable
if 2|z
1|and 2|z 2|are smaller than the minimal distance between the outcomes of
y
1given by (1.26) and the minimal distance between the outcomes ofy 2given by
(1.27) respectively.
Assume for the moment thatu
1,u2,v1,v2∈{0,1}and for the ease of discussion,
let us just concentrate on the signaly
α
1
received atR 1. Then there are generally up
to 16 different outcomes fory
1. Now there is one aspect of the above setup that
we have not yet exploited: the receiverR
1is not interested in the signalsv 1and
v
2. So if these ‘unwanted’ signals could be deliberately aligned (at the transmitters)
via encoding into a single componentv
1+v2, then there would be fewer possible
outcomes fory
1. This is merely down to the simple fact that there are 4 different
pairs(v
1,v2)as opposed to 3 different sumsv 1+v2whenv 1andv 2take on binary
values. With this in mind, suppose that
x
1=λ(h22u1+h12v1)andx 2=λ(h21u2+h11v2) (1.29)
respectively. Hereλ≥1 is simply some scaling factor. Thus, with reference to
(1.23), we have that
α
1=λh22,β1=λh12,α2=λh21,β2=λh11, (1.30)
and so (1.24)and(1.25) become
y
1=λ

(h 11h22)u1+(h21h12)u2+(h11h12)(v1+v2)

(1.31)
y
2=λ

(h 21h12)v1+(h11h22)v2+(h21h22)(u1+u2)

. (1.32)
Clearly, there are now only 12 outcomes for eithery
1ory2rather than 16. The
above discussion is a simplified version of that appearing in [52, §III: Example 3]
and constitutes the basis forreal interference alignment—a concept introduced and
developed in [48,51,52] and subsequent publications.
Remark 1.4The original idea of interference alignment exploits the availability of
‘physical’ dimensions of wireless systems such as the frequency of the signal or
the presence of multiple antennae. In short, an antenna is a device (such as an
old fashioned radio or television ariel) that is used to transmit or receive signals.
In any case, by using several antennae it is possible for a user to simultaneously
transmit several messages and these can naturally be thought of as the coordinates
of a point in a vector space, sayR
n
. Thus, when analysing such wireless systems
the transmitted signals can be treated as vectors inR
n
. The art of interference
alignment is to attempt to introduce an encoding at the transmitters (users) which
result in unwanted (interfering) signals at the receivers being forced to lie in a
subspace ofR
n
of smaller (ideally single) dimension. Such alignment is achieved

16 V. Beresnevich and S. Velani
by exploiting elementary methods from linear algebra, see for instance [37, Section
2.1] for concrete examples and a detailed overview of the process. The novel idea of
Motahari et al. involves exploiting instead the abundance of rationally independent
points in the real lineR. For instance, with reference to Example 2 above and the
transmitted signals given by (1.29), assuming thath
22/h12is irrational, the signal
x
1transmitted byS 1lies in the 2-dimensional vector subspace ofRoverQgiven by
V
1=λh22Q+λh 12Q.
Similarly, assuming thath
21/h11is irrational, the signalx 2transmitted byS 2lies in
the 2-dimensional vector subspace ofRoverQgiven by
V
2=λh21Q+λh 11Q.
In view of the alignment, the unwanted messagesv
1andv 2at receiverR 1are
forced to lie in a subspace ofRoverQof dimension one; namelyW
1=
λh
11h12Q. Similarly, the unwanted messagesu 1andu 2at receiverR 2lie in the
one-dimensionalQ-subspaceW
2=λh21h22Q.
As with the ‘baby’ example, we can easily modify the above ‘binary’ considera-
tion to the more general situation when the messagesu
1,u2,v1,v2are integers lying
in{0,...,Q}; i.e., the setup of Example 2. It is easily seen that in this more general
situation the savings coming from interference alignment are even more stark: there
are(2Q+1)(Q+1)
2
∼2Q
3
outcomes for eithery 1ory2after alignment as
opposed to(Q+1)
4
∼Q
4
outcomes before alignment. Consequently, based on the
outcomes fory
1andy 2after alignment being equally spaced, we have the following
trivial estimates for the associated minimal distances:
d
min,1≤
λ

h
11h22+h21h12+2h 11h12

Q
(2Q+1)(Q+1)
2
(1.33
and
d
min,2≤
λ

h
21h12+h11h22+2h 21h22

Q
(2Q+1)(Q+1)
2
. (1.34
We stress thatd
min,1is the minimal distance between the outcomes ofy 1given
by (1.31)andd
min,2is the minimal distance between the outcomes ofy 2given
by (1.32). As in Example 1, ‘perfect’ separation is essentially never the case and
to demonstrate this we need to bring into play the appropriate higher dimensional
version of Dirichlet’s theorem.

1 Number Theory Meets Wireless Communications: An Introduction 17
Theorem 1.2 (Minkowski’s Theorem for Systems of Linear Forms)Letβ i,j∈
R,where1≤i, j≤k, and letλ
1,...,λk>0.If
|det(β
i,j)1≤i,j≤k |≤
k
σ
i=1
λi, (1.35
then there exists a non-zero integer pointa=(a
1,...,ak)such that



|a
1βi,1+···+a kβi,k|<λi,(1≤i≤k−1)
|a
1βk,1+···+a kβk,k|≤λ k.
(1.36
The simplest proof of the theorem makes use of Minkowski’s fundamental convex
body theorem from the geometry of numbers; see, for instance [16, §1.4.1] or,
indeed, Chap.2of this book.
We now show how the minimal distanced
min,1(and similarly,d min,2) can be
estimated from above using Minkoswki’s theorem. For simplicity, consider the case
when
max{h
11h22,h21h12,h11h12}=h 11h12; (1.37
that is,h
11≥h21andh 12≥h22. Then, on applying Theorem1.2withk=3,λ 1=
(h
11h12)Q
−2
,λ2=λ3=Qand

i,j)1≤i,j≤k =


h
11h22h21h12h11h12
100
010

⎠,
we deduce the existence of integersa
1,a2anda 3, not all zero, such that









|(h
11h22)a1+(h21h12)a2+(h11h12)a3|<(h11h12)Q
−2
,
|a
1|<Q,
|a
2|≤Q.
(1.38
Remark 1.5It is worth pointing out that the argument just given above can
be appropriately adapted to establish the following generalisation of Dirichlet’s
theorem. For the details see for instance [16, Corollary 1.4.7]. Here and throughout,
given a pointx=(x
1,...,xn)∈R
n
we let|x|:=max{|x 1|,...,|x n|}.
Theorem 1.3For anyξ=(ξ
1,...,ξn)∈R
n
and anyQ∈Nthere exists(p,q)∈
Z×Z
n
such that
|q
1ξ1+···+q nξn+p|<
1
Q
n
and 1≤|q|≤Q. (1.39

18 V. Beresnevich and S. Velani
We now return to determining an upper bound ford min,1. A consequence of (1.38)
is that for any givenQ≥1 there exist integersa
1,a2,a3, not all zero, such that
ζ
ζ
ζ
ζ
h
11h22
h11h12
a1+
h
21h12
h11h12
a2+a3
ζ
ζ
ζ
ζ
<Q
−2
≤1.
This together with the triangle inequality implies that
|a
3|<
ζ
ζ
ζ
ζ
h
11h22
h11h12
a1
ζ
ζ
ζ
ζ
+
ζ
ζ
ζ
ζ
h
21h12
h11h12
a2
ζ
ζ
ζ
ζ
+1,
and so in view of our ‘maximal’ assumption (1.37), it follows that
|a
3|<|a 1|+|a 2|+1≤Q+(Q−1)+1=2Q.
Now observe that the quantity
λ×
ζ
ζ
(h
11h22)a1+(h21h12)a2+(h11h12)a3
ζ
ζ
is precisely the distance between the two specific outcomes ofy
1associated with
(1.31) given by the following choices:
Choice 1:u
1=max{0,a 1},u2=max{0,a 2},v1+v2=max{0,a 3},
Choice 2:u
1=max{0,−a 1},u2=max{0,−a 2},v1+v2=max{0,−a 3}.
We have just observed that Theorem1.2guarantees that|a
1|≤Q,|a 2|≤Qand
|a
3|≤2Qand sou 1,u2,v1,v2are integers lying in{0,...,Q}. Hence, in view of
(1.38) it follows (under the assumption (1.37)) that
d
min,1≤
λh
11h12
Q
2
=
C
2
Q
2
,whereC 2:=λh 11h12. (1.40
For all intents and purposes, this bound on the minimal distance is smaller than the
‘perfect’ separation estimate given by (1.33). A similar analysis can be carried out
when the maximum in (1.37) is attained on another term, and for estimatingd
min,2.
Obviously the parameterC
2would reflect the situation under consideration.
As mentioned earlier, the receiversR
1andR 2can decode the respective
messages provided that the respective minimal distancesd
min,1andd min,2are at
least two times larger than the noise at each receiver. Given that the nature of noise
is often a random variable with normal distribution, the overarching goal is to ensure
the probability that|z
1|<
1
2
dmin,1and|z 2|<
1
2
dmin,2is large. Indeed, as in Example
1, the larger the probability the more likely the receiversR
i(i=1,2)are able to
recover messages by roundingy
α
i
(given by (1.28)) to the closest possible outcome
ofy
i(given by (1.31)ifi=1and(1.32)ifi=2). It is therefore imperative to
understand how closed
min,1andd min,2can be to their theoretical upper bounds.

1 Number Theory Meets Wireless Communications: An Introduction 19
With this in mind we now describe various tools and notions from Diophantine
approximation that can be used for this purpose. In short, they allow us to get within
a constant factor of the theoretical upper bounds. As in Example 1, we start by
attempting to manipulate the encoding process so as to exploit the existence of badly
approximable points inR
n
. Before we embark on this discussion we make a remark
concerning the scaling factorλthat first appears in (1.29).
Remark 1.6Observe that estimatingd
min,1andd min,2from below is essentially the
same as estimating from below the size of the linear forms
(h
11h22)u1+(h21h12)u2+(h11h12)(v1+v2), (1.41
(h
21h12)v1+(h11h22)v2+(h21h22)(u1+u2). (1.42
The factorλappearing in (1.31)and(1.32) only determines the scaling ofd
min,1
andd min,2and can be used to ‘adjust’ these quantities, namely, to prevent them from
vanishing asQgrows, see Remark1.1for a similar consideration within Example 1.
Indeed, the effect of multiplication byλcan be simply understood as increasing the
separation in the constellation of messages; i.e. the messagesu
1,v1,u2,v2could be
associated with{0,λ,2λ,3λ,...,Qλ}instead of{0,1,2,3,...,Q}.
1.2.2 Badly Approximable Points
We start by stating the following simple consequence of Theorem1.3. It is the higher
dimensional analogue of Corollary1.1.
Corollary 1.2For any pointξ∈R
n
there exists infinitely many(p,q)∈Z×
Z
n
\{0}such that
|q
1ξ1+···+q nξn+p|<
1
|q|
n
. (1.43
Note that in the corollary we have not imposed the condition thatξis not a
point on a rational hyperplane. This is since we do not impose, as in the one-
dimensional statement, the requirement that(p,q)isprimitive; that is, without a
non-trivial common divisor. Naturally, badly approximable points inR
n
are defined
by requiring that the right hand side of (1.43) cannot be ‘improved’ by an arbitrary
constant factor. This we now formally state.
Definition 1.3 (Badly Approximable Points)A pointξ∈R
n
is said to bebadly
approximableif there exists a constantκ=κ(ξ)>0 such that for all(p,q)∈
Z×Z
n
\{0}
|q
1ξ1+···+q nξn+p|≥
κ
|q|
n
. (1.44

20 V. Beresnevich and S. Velani
The set of badly approximable points inR
n
will be denoted byBad(n.Itis
relatively simple to verify that for any real algebraic numberξof degreen+1
the point(ξ, ξ
2
,...,ξ
n
)∈R
n
is badly approximable. Indeed, consider the norm of
the algebraic number
α
1=q1ξ+q 2ξ
2
+···+q nξ
n
+p∈Q(ξ)
which (up to sign) is the product ofα
1and its other conjugates, sayα 2,...,αn+1.
For simplicity one can assume thatξis an algebraic integer. Furthermore, we can
assume that the right hand side of (1.44) is less than one and so without loss of
generality we have that|p|⎛|q|. Then, it is easily seen that|α
j|⎛|q|for allj,
while the norm ofα
1is bounded below by 1. Here and elsewhere⎝(respectively,
⎛) is the Vinogradov symbol meaning≥(respectively≤) up to a multiplicative
constant factor. The upshot is that
|q
1ξ+q 2ξ
2
+···+q nξ
n
+p|=|α 1|⎝
n+1
σ
j=2
|αj|
−1
⎝|q|
−n
,
whence the claim that(ξ, ξ
2
,...,ξ
n
)∈Bad(nfollows. This argument can be
made explicit to obtain a specific lower bound for the badly approximable constant
κ(ξ,...,ξ
n
). Examples of badly approximable algebraic points of this ilk were first
givenbyPerron[55].
The reason for us bringing into play the notion of badly approximable numbers
is similar to that in Example 1. If the channel coefficients happen to be such that
ξ=(ξ
1,ξ2):=

h
11h22
h11h12
,
h
21h12
h11h12

=

h
22
h12
,
h
21
h11

(1.45
is a badly approximable point inR
2
, then we are guaranteed the existence of a
constantκ(ξ)>0 such that




h
11h22
h11h12
q1+
h
21h12
h11h12
q2+p





κ(ξ)
|q|
2
for all non-zero integer points(p,q)∈Z×Z
2
\{0}. Thus, it follows that for every
Q∈N:
|h
11h22q1+h21h12q2+h11h12p|≥
κ(ξ)h
11h12
Q
2
for all(q 1,q2,p)∈Z
3
with 1≤|q|≤Q, and so the separations between any two
points given by (1.31) is at least
κ(ξ)λh11h12
Q
2.Inotherworlds,
d
min,1≥
κ(ξ)C
2
Q
2
(1.46

1 Number Theory Meets Wireless Communications: An Introduction 21
which complements the upper bound (1.40). Note that instead of (1.45) one can
equivalently considerξ=(ξ
1,ξ2)to be either of the points

h
21h12
h11h22
,
h
11h12
h11h22

,

h
11h22
h21h12
,
h
11h12
h21h12

, (1.47
which will also be badly approximable if (1.45) is badly approximable. Thus, we
can in fact show that (1.46) with appropriately adjusted constantκ(ξ)holds withC
2
redefined as
C
2:=max{h 11h22,h21h12,h11h12}. (1.48
A similar lower bound to (1.46) can be established ford
min,2if

h
21h12
h21h22
,
h
11h22
h21h22

=

h
12
h22
,
h
11
h21

(1.49
or equivalently

h
11h22
h21h12
,
h
21h22
h21h12

or

h
21h12
h11h22
,
h
21h22
h11h22

(1.50
is a badly approximable point inR
2
.
Remark 1.7We end this subsection with a short discussion that brings to the
forefront the significant difference between Examples 1 & 2, in attempting to exploit
the existence of badly approximable points. In short, the encoding process (1.30)
leading to the alignment of the unwanted signals in (1.31)and(1.32) comes at a
cost. Up to a scaling factor, it fixes the parametersα
1,α2,β1,β2in terms of the
given channel coefficients. This in turn, means that our analysis of the linear forms
(1.41)and(1.42) gives rise to the points (1.45)and(1.49)inR
2
that are dependent
purely on the channel coefficients. Now either these points are inBad(2)or not. In
other words, there is no flexibility left in the encoding procedure (after alignment)
to force (1.45)or(1.49) to be badly approximable inR
n
. This is very different to
the situation in Example 1. There we had total freedom to choose the parameters
αandβin order to force the point (1.11) to be a badly approximable number. The
upshot is that in Example 2, there is no such flexibility and this exacerbates the fact
that the probability of (1.45)or(1.49) being badly approximable is already zero.
The fact thatBad(nhas measure zero can be easily deduced from Khintchine’s
theorem, which will be discussed below in Sect.1.2.4—however see Sect.1.2.7for
the actual derivation. Although of measure zero, for the sake of completeness, it is
worth mentioning thatBad(nis of full Hausdorff dimension, the same as the whole
ofR
n
. This was established by Schmidt [57,58] as an application of his remarkably
powerful theory of(α, β)-games. In fact, he proved the full dimension statement for
badly approximable sets associated with systems of linear forms (see Sect.1.2.7).

22 V. Beresnevich and S. Velani
Remark 1.8We note that ifξis any of the points (1.45)or(1.47)andξ

is any of
the points (1.49)or(1.50), then in order to simultaneously guarantee (1.46) and its
analogue ford
min,2bothξandξ

need to be badly approximable. This adds more
constraints to an already unlikely (in probabilistic terms) event, since the pointsξ
andξ

are dependent. Indeed, concerningthe latter, it is easily seen that
ξ

=f(ξ) (1.51
for one of the following choices off:R
2
→R
2
f(x, y)=

1
x
,
1
y

,

x,
x
y

,

x
y
,x

,

y,
y
x

,or

y
x
,y

. (1.52
Clearly, the set of pairs(ξ,ξ

)of badly approximable points confined by (1.51)isa
subset of the already measure zero setBad(2)×Bad(2). Nevertheless, they do exist,
as was proved by Davenport [26], and are in ample supply in the following sense: the
set of pairs(ξ,ξ

)of badly approximable points subject to (1.51) has full Hausdorff
dimension, which is two. In other words, the dimension ofBad(2)∩f(Bad(2))is
equal to the dimension ofBad(2). This follows from the results of [19].
1.2.3 Probabilistic Aspects
In this section, we consider within the higher dimensional context of Example 2,
the probabilistic approach set out in Sect.1.1.4.Given0<κ

<1andQ∈N,let
B
n(Q, κ

)be the set ofξ∈I
n
:=(0,1)
n
such that
|q
1ξ1+···+q nξn+p|≥
κ

Q
n
(1.53
for all integer points(p,q)∈Z×Z
n
such that 1≤|q|≤Q. Note thatξ∈
B
n(Q, κ

)are precisely the points inI
n
for which the right hand side of inequality
(1.39) appearing in Dirichlet’sn-dimensional theorem, cannot be improved by the
factor ofκ

(Qis fixed here). To estimate the probability ofB n(Q, κ

), we consider
the complementary inequality
|q
1ξ1+···+q nξn+p|<
κ

Q
n
. (1.54
Let 1≤|q|≤Q. Then for a fixedq, it can be verified that the probability that a
givenξ∈I
n
satisfies (1.54)forsomep∈Zis exactly 2κ

Q
−n
—this is a relatively
straightforward calculation the details of which can be found in [63,Lemma8].On
summing up these probabilities overqwithq
1≥0 (this can be assumed without

1 Number Theory Meets Wireless Communications: An Introduction 23
loss of generality), we conclude that the probability that a givenξ∈I
n
satisfies
(1.54) for some integerspand 1≤|q|≤Q, is bounded above by


Q
−n
(2Q+1)
n−1
(Q+1)∼2
n
κ

(asQ→∞).
This in turn implies the following statement.
Lemma 1.2For any0<κ

<1and anyQ∈N
Prob(B
n(Q, κ

))≥1−2
n
κ


1+
1
2Q

n−1
1+
1
Q

. (1.55
Similarly to the one-dimensional case (cf. Sect.1.1.4), the above trivial estimate can
be improved, however, we leave this task to the energetic reader. We also note that
the probability in Lemma1.2is assumed to be uniform but it is possible to obtain a
version of Lemma1.2for other (absolutely continuous) distributions as mentioned
in Remark1.3. In any case, the upshot of the above discussion is that for sufficiently
smallκ

>0 the probability that the pointξgiven by (1.45) modulo 1 belongs
toB
n(Q, κ

)is positive. Hence, it follows that for anyρ∈(0,1)there exists an
explicitly computable constantκ

>0 with the following property: with probability
greater thanρ, for a random choice of the four channel coefficientsh
ij(i, j=1,2),
the separation between the associated pointsy
1given by (1.31) is at leastκ

C2/Q
2
,
and so the minimal distanced
min,1satisfies
d
min,1≥
κ

C2
Q
2
. (1.56
Moreover, the probabilityρcan be made arbitrarily close to one. However, the cost
is that the constantκ

becomes arbitrarily small. The above analysis holds equally
well at receiverR
2and we obtain an analogous probabilistic bound for the minimal
distanced
min,2associated with the pointsy 2given by (1.32).
Remark 1.9Obviously (1.56) is a better lower bound ford
min,1 than (1.46)
wheneverκ

is greater than the badly approximable constantκ(ξ)appearing in
(1.46). However, this really is not the point—both approaches yield lower bounds
for the minimal distance that lie within a constant factor of the theoretic upper bound
(1.40). The main point is that the badly approximable approach has zero probability
of actually delivering (1.46) whereas the probabilistic approach yields (1.46) with
positive probability (wheneverκ(ξ)is sufficiently small so that the right hand side
of (1.55) withκ

=κ(ξ)is positive).
Remark 1.10In the same vein as Remark1.8, we first observe that in order to
simultaneously guarantee (1.56) and its analogue ford
min,2, both the pointsξandξ

modulo one, whereξis given by (1.45)or(1.47)andξ

is given by (1.49)or(1.50),
need to simultaneously lie inB
n(Q, κ

). Thus to obtain the desired (simultaneous)
probabilistic statement, we need to show the probability of bothξandξ

modulo one

24 V. Beresnevich and S. Velani
belonging toB n(Q, κ
α
)is positive; say 1−κ
α
in line with (1.55). This would be an
easy task if the points under consideration were independent. However, the points
ξandξ
α
are confined by (1.51) and therefore the eventsξ(mod1)∈B n(Q, κ
α
)
andξ
α
(mod1)∈B n(Q, κ
α
)are dependent. Nevertheless, it can be shown that the
probability of these two events holding simultaneously is at least 1−σ×κ
α
,where
σis an explicitly computable positive constant. We leave the details to the extremely
energetic reader.
Remark 1.11For another specific (and powerful) application of the probabilistic
approach outlined in this section we refer the reader to [53]. In short, in [53]the
probabilistic approach is used to estimate the capacity of the two-user X channel
from below and above with only a constant gap between the bounds.
Notice that the fundamental setB
n(Q, κ
α
)that underpins the probabilistic
approach is dependent onQ. Thus, asQvaries, so does the random choice of
channel coefficients that achieve (1.56). As we shall see in the next section, this
can be problematic.
1.2.4 The Khintchine-Groshev Theorem and Degrees of
Freedom
The probabilistic approach of Sect.1.2.3, relies on the pointξassociated with the
channel coefficients via (1.45) being in the setB
n(Q, κ
α
).Now,howeverlargethe
probability of the latter (a lower bound is given by (1.55)), it can be verified that
Prob(B
n(Q, κ
α
))≤1−ωκ
α
, (1.57)
whereω>0 is a constant depending only onn. The proof of this can be
obtained by utilizing the notion of ubiquity; in particular, exploiting the ideas used
in establishing Proposition 4 in [12, Section 12.1]. Moreover, for anyκ
α
>0andany
infinite subsetQ⊂Nthe probability thatξlies inB
n(Q, κ
α
)for all sufficiently large
Q∈Q(let alone all sufficiently largeQinN) is zero. This is a fairly straightforward
consequence of Theorem1.3and [9, Lemma 4]. This is an unfortunate downside
of the probabilistic approach, especially when it comes to estimating the so called
Degrees of Freedom(DoF) of communication channels. Indeed, when estimating
the DoF it is desirable to achieve, with probability one, close to optimal bounds
on the minimal distances (d
min,1andd min,2within the context of Example 2) for
all sufficiently largeQ. Of course, the badly approximable approach described in
Sect.1.2.2does this in the sense that it yields (1.56) for all largeQwhenever
ξ∈Bad(2). However, as already discussed in Remark1.9, the downside of the
badly approximable approach is that the probability of hittingBad(2)is zero. In
this section we describe another approach which overcomes the inadequacies of
both the probabilistic and badly approximable approaches. It gives an ‘ε-weaker’

1 Number Theory Meets Wireless Communications: An Introduction 25
estimate for the minimal distance but as we shall soon see it is more than adequate
for estimating the DoF. The key is to make use of the fundamental Khintchine-
Groshev theorem in metric Diophantine approximation and this is what we first
describe.
Given a functionψ:R
+→R +,whereR +denotes the set of non-negative real
numbers, let
W
n(ψ):=

ξ∈I
n
:
|q
1ξ1+···+q nξn+p|<ψ(|q|)
for i.m.(p,q)∈Z×Z
n
\{0}

. (1.58
Here and elsewhere, ‘i.m.’ is short for ‘infinitely many’ and given a subsetXinR
n
,
we will write|X|
nfor itsn–dimensional Lebesgue measure. For obvious reasons,
points inW
n(ψ)are referred to asψ-approximable. Whenn=1, it is easily seen
thatW(ψ):=W
1(ψ)is the set ofξ=ξ 1∈Isuch that




ξ−
p
q




<
ψ(q)
q
has infinitely many solutions(p, q)∈Z×N. Investigating the measure theoretic
properties ofW(ψ)was the subject of the pioneering work of Khintchine [40]
almost a century ago. The following generalisation of Khintchine’s theorem is a
special case of a result of Groshev [36] concerning systems of linear form (see
Theorem1.12in Sect.1.2.7). In the one-dimensional case, it provides a quantitative
analysis of the density of the rationals in the reals.
Theorem 1.4 (Khintchine-Groshev for One Linear Form)Letψ:R
+→R +
be a monotonic function. Then
|W
n(ψ)|n=



0if


q=1
q
n−1
ψ(q) <∞,
1if


q=1
q
n−1
ψ(q)=∞.
Remark 1.12The convergence case of Theorem1.4is a relatively simple applica-
tion of the Borel–Cantelli Lemma from probability theory and it holds for arbitrary
functionsψ. In the divergence case, the theorem was first obtained by Groshev
under the stronger assumption thatq
n
ψ(q)is monotonic. In fact, the monotonicity
assumption can be completely removed from the statement of theorem ifn≥2. This
is a consequence of Schmidt’s paper [56, Theorem 2] from the swinging sixties if
n≥3 and the relatively recent paper [10] covers then=2 case. In 1941, Duffin
& Schaeffer [29] constructed a non-monotonic approximating functionψfor which
the sum

q
ψ(q)diverges but|W(ψ)|=0. Thus, the monotonicity assumption
cannot be removed in dimension one. For completeness, we mention that in the
same paper Duffin & Schaeffer formulated an alternative statement for arbitrary
functions. This soon became known as thenotorious Duffin-Schaeffer Conjecture

26 V. Beresnevich and S. Velani
and it remained unsolved for almost eighty years until the breakthrough work of
Koukoulopoulos & Maynard [47].
An immediate consequence of the convergence case of Theorem1.4is the
following statement.
Corollary 1.3Letψ:R
+→R +be a function such that

ρ
q=1
q
n−1
ψ(q) <∞. (1.59
Then, for almost allξ∈I
n
there exists a constantκ(ξ)>0such that
|q
1ξ1+···+q nξn+p|>κ(ξ)ψ(|q|)∀(p,q)∈Z×Z
n
\{0}.(1.60
Now consider the special case whenψ:q→q
−n−ε
for someε>0. Then
Corollary1.3implies that for almost allξ∈I
n
there exists a constantκ(ξ)>0
such that
|q
1ξ1+···+q nξn+p|≥
κ(ξ)
|q|
n+ε
for all(p,q)∈Z×Z
n
\{0}. In particular, for almost allξ∈I
n
and everyQ∈Nwe
have that
|q
1ξ1+···+q nξn+p|≥
κ(ξ)
Q
n+ε
(1.61
for all(p,q)∈Z×Z
n
\{0}with 1≤|q|≤Q. Now in the same way ifξgiven
by (1.45) is badly approximable leads to the minimal distance estimate (1.46), the
upshot of (1.61) is the following statement: with probability one, for everyQ∈N
and a random choice of channel coefficientsh
ij(i, j=1,2), the separation between
the associated pointsy
1givenby(1.31) is at leastκ(ξ)C 2/Q
2+ε
and so
d
min,1≥
κ(ξ)C
2
Q
2+ε
. (1.62
Just to clarify, thatξin the above corresponds to the point given by (1.45) associated
with the choice of the channel coefficients. Note that instead of (1.45), one can
equivalently considerξto be either of the points given by (1.47) and this would lead
to (1.62) withC
2defined by (1.48). A similar lower bound statement holds for the
minimal distanced
min,2associated with the pointsy 2givenby(1.32). Of course, in
this caseξneed to be replaced byξ

given by (1.49) or equivalently (1.50).
Remark 1.13Recall thatξisgivenby(1.45)or(1.47)andξ

is given by (1.49)
or (1.50) and they are dependent via (1.51)and(1.52). Note that any of the maps

1 Number Theory Meets Wireless Communications: An Introduction 27
in (1.52) is a diffeomorphism on a sufficiently small neighborhood of almost every
point inR
2
. Therefore, ifξavoids a subset ofR
2
of measure zero, then so doesξ

.
Thus, (1.62) and an analogous bound ford
min,2are simultaneously valid for almost
all choices of the channel coefficients.
Remark 1.14Note that in the above analysis, if we had worked with the function
ψ:q→q
−n
(logq)
−1−ε
for someε>0, we would have obtained the stronger
estimate
d
min,1≥
κ(ξ)C
2
Q
2
(logQ)
1+ε
.
It will be soon be clear that (1.62) is all we need for estimating the DoF within the
context of Example 2.
A natural question arising from the above discussion is:can the constantκ(ξ)
within Corollary1.3and thus(1.62)be made independent ofξ?Unfortunately, this
is impossible to guarantee with probability one; that is, for almost allξ∈I
n
.Tosee
this, consider the set
B
n(ψ, κ):=

ξ∈I
n
:
|q
1ξ1+···+q nξn+p|>κψ(|q|)
∀(p,q)∈Z×Z
n
\{0}

. (1.63
Then for anyκandψ, observe thatB
n(ψ, κ)will not contain the region
[−κψ(|q|), κψ(|q|)]×R
n−1
whenq=(1,0,...,0)∈Z
n
. This region has positive probability; namely
2κψ(1)), and so the complement (which containsB
n(ψ, κ)) cannot haveprobability
one. Nevertheless, the following result provides not only an explicit dependence on
the probability ofB
n(ψ, κ)onκ, but shows that it can be made arbitrarily close to
one upon takingκsufficiently small.
Theorem 1.5 (Effective Convergence Khintchine-Groshev for One Linear
Form)Letψ:R
+→R +be a function such that

ρ
q=1
q
n−1
ψ(q) <∞.
Then, for anyκ>0
Prob(B
n(ψ, κ))≥1−4nκ

ρ
q=1
(2q+1)
n−1
ψ(q).

28 V. Beresnevich and S. Velani
ProofNote that
B
n(ψ, κ)=I
n
\

q∈Z
n
\{0}
Eq(ψ) ,
where
E
q:=

ξ∈I
n
:|q1ξ1+···+q nξn+p|≤κψ(|q|)for somep∈Z
ˆ
.
Now, it is not difficult to verify that|E
q|n=2κψ(|q|)-see[63, Lemma 8] for
details. Thus, it follows that
Prob(B
n(ψ, κ)):= |B n(ψ, κ)| n≥1−
ρ
q∈Z
n
\{0}
|Eq|n
=1−
ρ
q∈Z
n
\{0}
2κψ(|q|)
=1−

ρ
q=1
ρ
q∈Z
n
|q|=q
2κψ(|q|)
=1−2κ

ρ
q=1
ψ(q)
ρ
q∈Z
n
|q|=q
1
≥1−2κ

ρ
q=1
ψ(q)2n(2q+1)
n−1
,
as desired. ρ
Having set up the necessary mathematical theory, we now turn our attention
to calculating the DoF for the two-userX-channel considered in Example 2.
The advantage of utilising the Khintchine-Groshev approach rather than the badly
approximable approach, is that the value we obtain is not only sharp but it is valid
for almost every realisation of the four channel coefficientsh
ij(i, j=1,2). Here,
almost every is naturally with respect to4-dimensional Lebesgue measure. At this
point, a mathematician with little or no background in communication theory (like
us) may rightly be crying out for an explanation of what is meant by the Degrees of
Freedom of communication channels. We will attempt to provide a basic and in part
a heuristic explanation within the context of Example 2. For a more in depth and
general discussion we refer the reader to Chap.2.
The simplest example of a communication channel is one involving just one
transmitter and one receiver. For obvious reasons, such a setup is referred to as a
point to point channel. The DoF of any other communication channel model is in

1 Number Theory Meets Wireless Communications: An Introduction 29
essence a measure of its efficiency compared with using multiple point to point
channels. In making any comparison, it is paramount to compare like with like.
Thus, given that the noisez
i(i=1,2) at both receiversR iwithin Example 2 is
assumed to have normal distributionN(0,1), we assume that the noise within the
benchmark point to point channel has normal distributionN(0,1).Inthesamevein,
we assume that the messages the users transmit within both models are integers
lying in{0,...,Q}; that is to say thatQis the same in Example 2 and the point
to point channel model. The parameterQ∈Nis obviously a bound on the
message size and it provides a bound on the number of binary digits (bits)that
can be transmitted instantaneously as a single bundle. Indeed, sending the integerQ
requires transmitting a bundle oflogQ+1≈logQbits, where the logarithm is to
the base 2. Loosely speaking, the larger the message to be sent the larger the “power”
required to transmit the message (transmitting instantaneously more bits requires
more energy). Thus a bound on the message sizeQcorresponds to imposing apower
constraintPon the channel model under consideration. For physical reasons, that
are not particularly relevant to the discussion here, the power is comparable to the
square of the message size. The upshot is that a power constraintPon the channel
model places a bound on the maximal number of bits that can be reliably transmitted
as a single bundle. With this in mind, the (total) DoF of the channel characterises the
number (possibly fractional) of simple point-to-point channels, needed to reliably
transmit the same maximal number of bits as the power constraintPtends to infinity.
We now calculate the total DoF for the concrete setup of Example 2. The exposition
given below is a simplified version of that presented in [52].
In relation to Example 2, the power constraintPmeans that
|x
1|
2
≤Pand|x 2|
2
≤P, (1.64
wherex
1andx 2are the codewords transmitted byS 1andS 2as given by (1.29).
Now notice that since the messagesu
1,u2,v1,v2are integers lying in{0,...,Q},
it follows thatPis comparable to(λQ)
2
—the channel coefficientsh ijare fixed.
Recall, thatλ≥1 is a scaling factor which is at our disposal and this will be
utilized shortly. It is shown in [52], that the probability of error in transmission
within Example 2 is bounded above by
exp


d
2
min
8
φ
, (1.65
where
d
min=min{d min,1,dmin,2}.
It is a standard requirement that this probability should tend to zero asP→∞.In
essence, this is what it means for the transmission to be reliable. Then, on assuming
(1.62)—which holds for almost every realisation of the channel coefficients—it

30 V. Beresnevich and S. Velani
follows that
d
minψ
λ
Q
2+ε
, (1.66
and so the quantity (1.65) will tend to zero asQ→∞if we set
λ=Q
2+2ε
.
The upshot of this is that we will achieve reliable transmission under the power
constraint (1.64)ifwesetPto be comparable toQ
6+4ε
;thatis
Q
6+4ε
ωPωQ
6+4ε
.
Now in Example 2, we simultaneously transmit 4 messages, namelyu
1,u2,v1,v2,
which independently take values between 0 andQ. Therefore, in total we transmit
approximately 4×logQbits, which with our choice ofPis an achievable total rate
of reliable transmission; however, it maynot be maximal. We now turn our attention
to the simple point to point channel in which the noise has normal distribution
N(0,1). In his pioneering work during the forties, Shannon [62] showed that
such a channel subject to the power constraintPachieves the maximal rate of
reliable transmission
1
2
log(1+P)—for further details see Sect.2.1of Chap.2.
On comparing the above rates of reliable transmission for the two models under the same power constraint, we get that the total DoF of the two-userX-channel
described in Example 2 is at least
lim
P→∞
4logQ
1
2
log(1+P)
=lim Q→∞
4logQ
1
2
log(1+Q
6+4ε
)
=
4
3+2ε
. (1.67
Given thatε>0 is arbitrary, it follows that for almost every realisation of the
channel coefficients
DoF≥
4
3
.
Nowitwasshownin[38] that the total DoF of a two-userX-channel is upper
bounded by 4/3 for all choices of the channel coefficients, and so it follows that
for almost every realisation of the channel coefficients
DoF=
4
3
. (1.68
For ease of reference we formally state these findings, the full details of which can
be found in [52], as a theorem.

1 Number Theory Meets Wireless Communications: An Introduction 31
Theorem 1.6For almost every realisation of the four channel coefficientsh ij
(i, j=1,2), the total DoF of the two-userX-channel is
4
3
.
Remark 1.15We reiterate that by utilising the Khintchine-Groshev approach rather
than the badly approximable approach (i.e. exploiting the lower bound (1.62) instead
of (1.46) or equivalently (1.56) for the minimal distance), we obtain (1.68)for
the DoF that is valid for almost every realisation of the four channel coefficients
h
ij(i, j=1,2)rather than on a set of 4-dimensional Lebesgue measure zero.
In Sect.1.2.6, we shall go further and show that any exceptional set of channel
coefficients for which (1.68) fails is a subset arising from the notion of jointly
singular points. This subset is then shown (see Theorem1.9) not only to have
measure zero but to have dimension strictly less than 4—the dimension of the space
occupied by the channel coefficients. In short, our improvement of Theorem1.6is
given by Theorem1.10.
1.2.5 Dirichlet Improvable and Non-Improvable Points:
Achieving Optimal Separation
We now show that there are special values ofQfor which the minimal distance
d
min,1satisfies (1.56) withκ
α
as close to one as desired. Recall, the larger the
minimal distance the more tolerance we have for noise. The key is to exploit the
(abundant) existence of points for which Dirichlet’s theorem cannot be improved.
Definition 1.4 (Dirichlet Improvable and Non-Improvable Points)Let 0<
κ
α
<1. A pointξ∈R
n
is said to beκ
α
-Dirichlet improvableif for all sufficiently
largeQ∈Nthere are integer points(p,q)∈Z×Z
n
with 1≤|q|≤Qsuch that
|q
1ξ1+···+q nξn+p|<κ
α
Q
−n
. (1.69
A pointξ∈R
n
is said to beDirichlet non-improvableif for anyκ
α
<1 it is not
κ
α
-Dirichlet improvable. Thus, explicitly,ξ∈R
n
isDirichlet non-improvableif for
any 0<κ
α
<1 there exists arbitrarily largeQ∈Nsuch that for all integer points
(p,q)∈Z×Z
n
with 1≤|q|≤Q
|q
1ξ1+···+q nξn+p|≥κ
α
Q
−n
. (1.70
Remark 1.16Note that Dirichlet non-improvable points are not the same as those
considered in the probabilistic approach of Sect.1.2.3. There the emphasis is on
bothκ
α
andQbeing uniform.
In a follow-up paper [27] to their one-dimensional work cited in Sect.1.1.5,
Davenport & Schmidt showed that Dirichlet improvable points inR
n
form a set
DI(nofn-dimensional Lebesgue measure zero. Hence, a randomly picked point
inR
n
is Dirichlet non-improvable withprobability one. The upshot of this is

32 V. Beresnevich and S. Velani
the following consequence: for almost every random choice of the four channel
coefficientsh
ij(i, j=1,2)and for anyε>0 there exist arbitrarily large integers
Qsuch that the minimal distanced
min,1between the associated points given by
(1.31) satisfies
d
min,1≥
(1−ε)λh
11h12
Q
2
=(1−ε)
C
2
Q
2
. (1.71
To conclude, the Dirichlet non-improvable approach allows us to almost surely
achieve the best possible separation, within the factor(1−ε)of the theoretic upper
bound (1.40), for an infinite choice of integer parametersQ∈Q
1.
Remark 1.17Obviously, we can obtain an analogous lower bound statement for
d
min,2for an infinite choice of integer parametersQ∈Q 2. However, it is not
guaranteed that the integer setsQ
1andQ 2overlap and thus the problem of
optimisingd
min,1andd min,2simultaneously remains open.
1.2.6 Singular and Non-Singular Points: The DoF of
X-Channel Revisited
With reference to Example 2, the Khintchine-Groshev and the Dirichlet non-
improvable approaches allows us to achieve good separation for the minimal
distances (i.e., lower bounds ford
min,1andd min,2that are at most ‘ε-weaker’ than
the theoretic upper bounds) for almost all choices of the four channel coefficients
h
ij(i, j=1,2). We now turn to the question ofwhether good separation can
be achieved for a larger class of channel coefficients? For example, is it possible
that the set of exceptions not only has measure zero (as is the case with the
aforementioned approaches) but has dimension strictly less than four (the dimension
of the space occupied by the channel coefficients)?In short the answer is yes.
The key is to make use of the following weaker notion than that of Dirichlet non-
improvable points (cf. Definition1.4).
Definition 1.5 (Singular and Non-Singular Points)A pointξ∈R
n
is said to be
singularif it isκ
α
-Dirichlet improvable for anyκ
α
>0. A pointξ∈R
n
is said to
benon-singular(orregular) if it is not singular. Thus, explicitly,ξ∈R
n
isnon-
singularif there exists a constantκ
α

α
(ξ)>0 such that there exist arbitrarily
large integersQ∈Nso that for all integer points(p,q)∈Z×Z
n
with 1≤|q|≤Q
|q
1ξ1+···+q nξn+p|≥κ
α
Q
−n
. (1.72
By definition, any singular point is trivially Dirichlet improvable. Equivalently, any
Dirichlet non-improvable point is trivially non-singular.

1 Number Theory Meets Wireless Communications: An Introduction 33
We letSing(ndenote the set of singular points inR
n
. It is easily verified that
Sing(ncontains every rational hyperplane inR
n
. Therefore,
n−1≤dimSing(n≤n.
Here and throughout, dimXwill denote the Hausdorff dimension of a subsetXof
R
n
. For the sake of completeness, we provide the definition.
Definition 1.6 (Hausdorff Dimension)LetX⊂R
n
. Then the Hausdorff dimen-
sion dimXofXis defined to be the infimum ofs>0 such that for anyρ>0
and anyε>0 there exists a cover ofXby a countable familyB
iof balls of radius
r(B
i)<ρsuch that

ρ
i=1
r(Bi)
s
<ε.
Remark 1.18For most sets upper bounds for the Hausdorrf dimension can be
obtained using natural covering by small balls. Indeed, letX⊂R
n
andρ>0and
supposeXcan be covered byN
ρ(X)balls of radius at mostρ. Then, it immediately
follows for the above definition that
dimX≤lim sup
ρ→0
logN ρ(X)
−logρ
.
Note that the Hausdorff dimension of planes and more generally smooth subman-
ifolds ofR
n
is the same as their usual ‘geometric’ dimension. The middle third
Cantor setKis the standard classical example of a set with fractal dimension. Recall,
Kconsists of all real numbers in the unit interval whose base 3 expansion does not
contain the ‘digit’ 1; that is
K:= {ξ∈[0,1]:ξ=


i=1
ai3
−i
witha i=0or2}.
It is well known that
dimK=
log 2
log 3
.
For a proof of this and a lovely introduction to the mathematical world of fractals,
see the bible [30].
Now returning to singular points, in the casen=1, a nifty argument due to
Khintchine [40] dating back to the twenties shows that a real number is singular if
and only if it is rational; that is
Sing(1)=Q. (1.73

34 V. Beresnevich and S. Velani
Recently, Cheung & Chevallier [22], building on the spectacularn=2workof
Cheung [21], have proved the following dimension statement forSing(n.
Theorem 1.7 (Cheung & Chevallier)Letn≥2.Then
dimSing(n=
n
2
n+1
.
Thus,
codimSing(n=
n
n+1
.
Remark 1.19Note that since
n
2
n+1
>n−1, the theorem immediately implies that in
higher dimensionsSing(ndoes not simply correspondto rationally dependentξ∈
R
n
as in the one-dimensional case—the theory is much richer. Also observe, that
since
n
2
n+1
<n,thesetSing(nis strictly smaller thanR
n
in terms of its Hausdorff
dimension. How much smaller is measured by its codimension;i.e.n−dimSing(n.
Now if the four channel coefficientsh
ij(i, j=1,2)happen to be such that
the corresponding pointξ∈R
2
givenby(1.45) is non-singular, then there exist
arbitrarily large integersQsuch that the minimal distanced
min,1between the
associated points given by (1.31) satisfies
d
min,1≥
κ
α
(ξ)λh11h12
Q
2
=
κ
α
(ξ)C2
Q
2
. (1.74
This of course is similar to the statement in which the pointξis Dirichlet
non-improvable with the downside that we cannot replace the constantκ
α
(ξ)by
(1−ε)as in (1.71). However, the advantage is that it is valid for a much larger
set of channel coefficients; namely, the exceptional set of channel coefficients
(h
11,h12,h21,h22)∈R
4
+
for which (1.74) is not valid has dimension
10
3
,which
is strictly smaller than 4—the dimension of the ambient space occupied by
(h
11,h12,h21,h22). This result seems to be new and we state it formally.
Proposition 1.1For all choices of channel coefficients(h
11,h12,h21,h22)∈R
4
+
,
except on a subset of codimension
2
3
, there exist arbitrarily large integersQsuch
that the minimal distanced
min,1between the associated points given by(1.31)
satisfies(1.74).
The proof of the proposition will make use of the following two well known results
from fractal geometry [50].
Lemma 1.3 (Marstrand’s Slicing Lemma)For anyX⊂R
k
andl∈N, we have
that
dim(X×R

)=dimX+→.

1 Number Theory Meets Wireless Communications: An Introduction 35
Lemma 1.4LetX⊂R
k
andg:R
k
→R
k
be a locally bi-Lipschitz map. Then
dim
η
g(X)

=dimX.
Proof(Proof ofProposition1.1)Consider the following map on the channel
coefficients
g:R
4
+
→R
4
+
such thatg(h 11,h12,h21,h22)=

h 11,h12,
h
22
h12
,
h
21
h11

.
As we have already discussed, for anyξgivenby(1.45) such thatξ∈R
2
+
\Sing(2)
we have that (1.74) holds. Hence, (1.74) holds for any choice of channel coefficients
such that
(h
11,h12,h21,h22)⎪∈g
−1

R
2
+
×
η
R
2
+
∩Sing(2)


. (1.75
By Lemma1.3and Theorem1.7, it follows that
codim

R
2
+
×
η
R
2
+
∩Sing(2)


=
2
3
.
Finally, note that locally at every point ofR
4
+
the mapgis aC
1
diffeomorphism
and hence is bi-Lipschitz. Therefore, by Lemma1.4it follows thatg
−1
preserves
dimension and thus the codimension of the right hand side of (1.75)is
2
3
.This
completes the proof. ρ
Remark 1.20Just to clarify, thatξappearing in (1.74) corresponds to the point
given by (1.45) associated with the choice of the channel coefficientsh
ij(i, j=
1,2)andκ

(ξ)>0 is a constant dependent onξ. Note that instead of (1.45), one
can equivalently considerξto be either of the points given by (1.47) and this would
lead to (1.74) withC
2defined by (1.48).
Naturally, the analogue of Proposition1.1holds for the minimal distanced
min,2
between the associated points given by (1.34). However, as in the Dirichlet non-
improvable setup (cf. Remark1.17), we cannot guarantee that the arbitrary large
integersQon which the lower bounds for the minimal distances are attained,
overlap. If we could guarantee infinitely many overlaps, it would enable us to
strengthen Theorem1.6concerning the Degrees of Freedoms (DoF) of the two-
userX-channel described in Example 2. With this goal in mind, it is appropriate to
introduce the following notion of jointly singular points.
Definition 1.7 (Jointly Singular and Non-Singular Points)The pair of points

1,ξ
2)∈R
n
×R
n
is said to bejointly singularif for anyε>0 for all sufficiently
largeQ∈Nthere exists an integer point(p,q)∈Z×Z
n
with 1≤|q|≤Q
satisfying
min
1≤j≤2
|q1ξj,1+···+q nξj,n+p|<εQ
−n
,

36 V. Beresnevich and S. Velani
whereξ
j=(ξj,1,...,ξj,n), j=1,2. The pair(ξ
1,ξ
2)∈R
n
×R
n
will be called
jointly non-singularif it is not jointly singular, that is if there exists a constant
κ




1,ξ
2)>0 such that there exist arbitrarily largeQ∈Nso that for all
integer points(p,q)∈Z×Z
n
with 1≤|q|≤Q
min
1≤j≤2

∈q
1ξj,1+···+q nξj,n+p

∈≥κ

Q
−n
. (1.76
The set of jointly singular pairs inR
n
×R
n
will be denoted bySing
2
(n.This
set is not and should not be confused with the standard simultaneous singular set
corresponding to two linear forms innvariables (see Sect.1.2.7).
The above notion of jointly non-singular pairs enables us to prove the following
DoF statement.
Proposition 1.2Let(h
11,h12,h21,h22)∈R
4
+
be given and letξbe any of the
points(1.45)or(1.47),letξ

be any of the points(1.49)or(1.50). Suppose that
(ξ,ξ

)⎪∈Sing
2
(2). (1.77
Then(1.68)holds, that is the total DoF of the two-userX-channel withh
ij(i, j=
1,2)as its channel coefficients is
4
3
.
ProofTo start with, simply observe that condition (1.77) means that there exist
κ

>0 and an infinite subsetQ⊂Nsuch that for everyQ∈Qand all integer
points(p,q)∈Z×Z
2
with 1≤|q|≤Q
|q
1ξ1+q2ξ2+p|≥κ

Q
−2
and


q 1ξ

1
+q2ξ

2
+p


≥κ

Q
−2
. (1.78
Consequently, for everyQ∈Qwe can guarantee that (1.74) and its analogue for
d
min,2are simultaneously valid. This in turn implies (1.66)foreveryQ∈Q.From
this point onwards, the rest of the argument given in Sect.1.2.4leading to (1.68)
remains unchanged apart from the fact that the limit in (1.67) is now alongQ∈Q
rather than the natural numbers. ρ
Proposition1.2provides a natural pathway for strengthening Theorem1.6.This
we now describe. It is reasonable to expect that the set of(ξ,ξ

)not satisfying (1.77)
is of dimension strictly smaller than four—the dimension of the ambient space.
Indeed, this is something that we are able to prove.
Theorem 1.8Letn≥2.Then
dimSing
2
(n=2n−
n
(n+1)
. (1.79
The theorem will easily follow from a more general statement concerning systems
of linear forms proved in Sect.1.2.7below; namely, Theorem1.14. Note that
Theorem1.8is not enough for improving Theorem1.6. Within Proposition1.2,

Random documents with unrelated
content Scribd suggests to you:

– (Én láttam azt! gondolá magában Hanna és arcza sápadt lett,
mint azok a camaeák, a mik faopálba vannak vésve).
– Én igazán haragszom érte, hogy Henrik el nem jött, szólt
duzzogva Helene, mialatt háziasszonyi tevékenységgel készíté az
asztaltársak számára a reggelit, s e mellett ideges ügyetlenséggel
hol a kezét égette meg a thea-katlanon, hol a thea közé öntött tej
helyett fekete kávét. Tudom én jól, hogy ő nem szeret bennünket.
De legalább tégedet kellene szeretnie. Hogy tehette ő azt, hogy
téged egyedül bocsásson útra? Ő rossz testvér! Neki veled kellett
volna jönnie. Ez több, mint nonchalance; egy fiatal testvért egyedül
bocsátani útra. Ez shoking!
Hanna két keze összetett ujjahegyeit ajkaihoz emelé. Néma
könyörgés volt ez, hogy hagyjon fel nénje a szemrehányásokkal.
Szólni hogy tudott volna?!
Erre még ingerültebb lett Helene; a czukorfogó csiptetőt úgy
hajította le az asztalra, hogy az keresztülugrott három csészén.
– De mikor az szörnyűség! És épen az éjszakai vonatot választani
a te utazásodra, egymagadban!
Az ifjú hegedű művész igen okos védelmet vélt kigondolni, a
midőn mostohája indulatának csillapítására azt hozá fel, hogy hiszen
nappali vonat volt az, a min Hanna elindult; de útközben valami
fennakadása történt a vonatnak, azért késett el.
– Lirum, lárum! Ha a vonaton történt baleset, az már a mai
lapoknak meg lett táviratozva.
Helene hevesen megrázta a kezeügyében levő ezüst csengetyűt,
mire a komornyik bejött.
– Wenczel! A mai lapokat!
Wenczelnek pedig már megvolt hagyva, hogy a mai lapokat
mindenképen sikkaszsza el.

Wenczel azt mondta, hogy azok még nem érkeztek meg. Hát
menjen a szerkesztőségbe, az expeditiókba s szidja le őket, miért ily
rendetlenek!
Ez egy kis derült időközt hozott. A míg Wenczel odajár, mindenki
csendes lélekkel elköltheti a maga kávéját, theáját és pumpernikkel-
szeleteit sonkával; s ez idő alatt a háziasszony a női
szeretetreméltóság eszményképévé alakul át, mindenkit elhalmoz
figyelmével; ő maga theát iszik, de férje kávéjába czukordarabkát
mártogat, s beletorkoskodik a kaviárba, a mit az zsemlyeszeletére
mázolt; Inczének elmondja az egész napra való programmot, hová
fognak együtt elmenni, minden nevezetességét Prágának
megtekinteni; még másnap is ott kell időznie, mert arra egy nap
nem elég. Ne törődjék a rendőrséggel: az ebben a palotában nem
parancsol. Ha akarjuk, magát a rendőrfőnököt is magunkkal viszszük
s olyan jezsuita-spitzet itatunk vele, hogy maga kezdi rá a «Shuselka
nam posnyét».
Egy szép szelíd papagály azalatt odarepült a vállára s gyönyörű
smaragd fejét füléhez hajtva, gyöngéden suttogott neki titkos
madárbeszédet, a mi oly ideges gyönyört idézett elő nála.
E jókedv, e női dévajság közepett egyszerre elképedt a a hölgy:
– Jézus, Mária! Mi az?
– No hát mi az?
Pedig csak Wenczel jött be.
Ez az ember engem halálra ijeszt, valahányszor belép! Micsoda
ijedt képpel jön! Hát mi történt? Tűz van? Kiáradt a Moldva? Valami
attentatum? Szólj: mi baj van?
Wenczel szabódott, hogy ő nincs megijedve, nem történt semmi;
csakhogy már az Isten úgy teremtette az arczát, mintha mindig meg
volna ijedve.

– Aztán mindig olyan kisértetes hangon beszél, mint egy halott-
bemondó, vagy egy koldus.
Wenczelnek csak annyi mondanivalója volt, hogy a hírlapokat az
éjjel mind lefoglalta a rendőrség, azért nem hordatnak ki.
– Miért foglalta volna le, te tolpats?
Wenczel annak tudta okát találni.
– Pan Rieger beszéde volt benne.
– Hát a hivatalos lapot is lefoglalták? abban is benne volt a pan
Rieger beszéde?
Már ebből a kelepczéből nem segítette ki Wenczelt a tudománya;
azt felelte, hogy «azt is».
Ez iszonyú dühbe hozta Helenát.
– Marsch ki előlem! Takarodjál rögtön. Ez az ember pusztúljon el
rögtön a háztól! Szemem elé ne kerüljön többet! Ily arczátlanul
hazudni a szemembe! Ostobának tart ez az ember engem? Leó,
nekem rögtön más komornyikot fogadj! De még ma délig! Engem ez
az ember nem szolgál ki többet!
Senki sem szólt ellene. Annyit már tudott mindenki, hogy haragos
asszonyt kérlelni annyit tesz, mint fujtatni a tüzet.
Azután ideges szelídséggel fordult a hegedűművészhez s azt
interpellálta fenyegetően lágy és rettentően hizelgő hangon:
– Ugyan édes Vilmos, nézz utána te, hová lettek a hírlapok mind?
Vilmos, művészek szokása szerint nem sokat válogatott a
fillentések között; a melyiket legközelebb találta, előrántotta azt.
– Hiszen ünnep volt tegnap, mama, hát nem jelenhettek meg a
lapok!
Ekkor aztán egész búbánatos érzelgéssel kezdé rá Helén:

– Ugy-e kedves Vilmos, rossz mostohád voltam én neked mindig?
Kegyetlen voltam hozzád? Szívtelenül bántam veled? Elrontottam
örömeidet? Mikor beteg voltál, cselédre bíztam a gondviselésedet?
Szidtalak szembe, rágalmaztalak az apád előtt? Azzal érdemeltem
meg, hogy engem nevetség tárgyává tégy az egész világ előtt,
idegen vendégek előtt!?
A szegény hegedűvirtuoz még jobban odaszorította a
vállpereczéhez az állával a képzelmi violint, mutatva, hogy nem érti,
mivel követte el ezt a nagy bűnt?
Helene aztán kitört a commentárral.
– Mintha nem tudná azt egész Prága, hogy én a legbuzgóbb
templomjáró vagyok! Hogy nincs olyan ünnep, a melyen én szokott
helyemet a Teyn templomban el ne foglalnám, s Istennek tartozó
imádatom adóját le ne rónám. S te mégis azt mondod nekem, hogy
tegnap ünnep volt. Ünnepnap, a melyen én nem voltam
templomban. E szerint én tehát vagy bolond vagyok, vagy atheista?
No hát melyik vagyok? A hét bolond leányok egyike, a ki elaluszsza
az ünnepnapot? vagy manichæus? sabbathianus? iconomachus? Mi?
Soha ilyen kelepczébe nem volt még szorulva a szegény művész.
Leo megszánta a fiát s ki akarta segíteni a veremből okosan, s
utána szállt maga is.
– Engedjen meg kedvesem. Fogadalmi ünnep volt tegnap, így
értette Vilmos: Ziska halálának évnapja. A betűszedők mind
összebeszéltek, hogy ezen a napon nem fognak dolgozni; még a
kormánylapot sem szedik ki. Tudod, a betűszedők az egész világon
mindenütt a democrata párthoz tartoznak: azok az ultrák
primipilusai. Tegnap mind kinn voltak a «Fehérhegyen».
Helene erre a históriai magyarázatra összecsapta a kezeit és
elkezdett kaczagni; de a fogait csikorgatta nevetés közben.
– Ez igazán mulatságos! Csak azért, hogy nekem ne legyen
igazam, hanem Hannának; csak azért, hogy Henrik szívtelensége ki

legyen mentve: összeesküszik ellenem mindenki a háznál, s
hazudnak először háromemeleteset, azután toronymagasságnyit,
utoljára olyan nagyot, mint a Fehér-hegy! Idehozzák nekem pan
Riegert, s az egész politzeyt; új ünnepet csinálnak a naptárban;
összeesküvést forraltatnak a betűszedők közőtt s még a szegény vak
Ziskát is előhurczolják a koporsójából. Igazán kaczagni való tréfa.
Incze parázson ült e jelenet alatt, a minek indokait nagyon jól
érté s azt mondá latinul a vele szemben ülő bankárnak, hogy
«medicina pejor morbo».
– Ne beszéljetek én előttem magyarul! fordult most ő ellene az
úrnő haragja, mely azonban egyszerre átváltozott még a haragnál is
veszélyesebb furfanggá. De hisz igaz! Mit zaklatódom én itt a
hírlapok után? Hisz itt van ön, uram. Ön egy vonaton jött Hannával;
leghitelesebben megmondja, történt-e valami baleset a vaspályán
vagy sem? A többiek mind hazudnak. Jól áll nekik. Hanna nő:
szabadalma van rá. Wenczel cseléd természetében van. Vilmos
művész: ráragadt az. Leó bankár: neki ez mestersége. De ön uram,
katona, a ki az igazságért harczolt, a ki becsületét, csak büszkeségét
növelte; önnek nem lehet hazudnia. Mondja meg az igazat.
Incze számára nem volt kitérés!
– Megmondom, bárónő. A vasúton csakugyan történt baleset. Két
vonat egymásba ütközött. Henrik gróf kisérte Hanna grófnőt, s e
balesetnél oly sérüléseket szenvedett, a mik miatt vissza kellett
térnie Königinhofba. Ez az oka, a miért nem jöhetett ide. Ezt titkolta
ön előtt egész környezete.
Helene nem ijedt meg, nem kapott szívgörcsöket; nyugodtan
mondá: ezt én sejtettem, ezt sugta valami.
S aztán könybelábadt szemekkel tekinte végig a család tagjain,
elmélázva, míg egyszer csak megint odafordult Inczéhez.
– De ugyebár, nem lesz szükséges nála semmi amputatio?
– Nem lesz szükséges, bárónő.

– Az rettenetes volna, folytatá kerekre felnyílt szemekkel, ha
egyszer csak úgy jönne elém, hogy fél keze, vagy fél lába
hiányzanék, vagy arcza nem volna az, a mi volt. Itt eltakarta
tenyerével szemeit. S megint hirtelen Incze karjára tette kezét.
«Ugyebár, uram, a viszontlátáskor olyannak fogom őt látni, a milyen
mindig volt?»
– A viszontlátáskor, úrnőm, bizonnyal olyannak fogja ön Henrik
grófot látni, a milyen mindig volt.
Helene szemei tudtak olvasni az arczvonalmakban, s Incze
vonásai áruló nyitott könyvek voltak mindig. Helene megragadá
mindkét kezével Incze kezét s ezt susogá:
– A viszontlátáskor olyan lesz ő, a milyen mindig volt? úgy-e?
mert meghalt?
S forrón megszorítá Incze kezét.
– Ezt köszönöm önnek.
És aztán nem tört ki belőle sem köny, sem ijedtség, sem
kaczagással vegyült zokogás. Felkelt helyéről nyugodtan; odament
Hannához, azt átölelte, megcsókolta, s meleg, nyugodt hangon
suttogá neki:
– Bocsáss meg, én szentem, én martyrom. A ki eltitkoltad
előttem keservedet, hogy engem kimélj; nem sírtál, nevető arczot
mutattál előttem s tűrted, hogy czivódom veled. Ugy-e, milyen
kegyetlen zsarnok egy idegbeteg asszony! Nagyobb zsarnok a
híveihez, mint a dahomeyi szultán; mert ti ezt a szörnyű rabigát
azért viselitek: mivel szerettek. No de nem soká tart ez már neked.
Csitt!
A további beszédet Hanna lecsókolta ajkáról.
Helene aztán sorban bocsánatot kért férjétől, annak homlokát
érintve csókjával; aztán Vilmostól, fehér kezét nyujtva neki csókra, s
megsimogatva hosszú hajfürteit, végre Wenczelt is behivatta s azt

mondta neki kegyesen: «maradhatsz továbbra is» s egy selyem
nyakkendőt ajándékozott neki engesztelés jelképeül, melyet
komornája már készen tartott a számára.
(Ez már tíz év óta a nyolczadik selyemkendő, a mit Wenczel kap:
sugá Leó Inczének, de oly hangon, hogy Helene is meghallhatta).
– Ez azt bizonyítja, szólt az úrnő, hogy igen sokszor
megharagszom, de ugyanannyiszor kibékülök.
Azzal kezét Incze karjába fűzve, átvezetteté magát vele a
salonba, menetközben közel hajtott fővel mondva neki:
– Lássa ön, az a modor, a melylyel ön e balesetet tudtomra adta,
egészen elvette e rémhír fulánkját. Most már meg vagyok nyugodva,
s holnap alapítványt teszek a kapuczinusok kolostorában, örök
misére, szegény Henrikünk lelki üdveért. Isten nyugtassa meg
szegényt!
A salon asztalait albumok foglalták el művészi aquarellekből,
fényképekből összeállítva; a tájak közt több volt olyan, melyet Leo is,
Incze is meglátogatott egykor, s a fényképezett celebritások közt
szinte akadtak olyanokra, a kikkel mind a ketten ismerősök voltak. Az
ismeretlenek közt a bárónő volt Incze cicerónéja.
Egy díszalbumnak mindjárt a harmadik lapján (a két első a
császár és császárné helye volt) egy érdekes hölgy fényképe volt:
gömbölyű, életvidám arcz, nagy sötét szemekkel, termetén sötét
hamuszín habos selyem öltöny, széles vállszalaggal, keblén csillag-
kereszt és fején diadémmal körített bársonysüveg.
– Ez császárnénk testvére, mondá Helene bárónő.
Incze beérte annyival.
A családi albumban aztán voltak a családtagoknak külömböző
időkből felvett fényképei, rokonok, jó barátok. Helene bárónő
szeretetreméltó készséggel ajánlá fel Inczének az ő családtagjainak
arczképeit, cserébe kivánva az övét, melyet ugyanazon lapra

helyezett el, a hol legközelebbi családtagjai voltak; úgy, hogy mikor
az album összecsukódott, Incze fényképezett arczának Hannáét
kellett érinteni.
Ominosus véletlen!
Az asszonyok szeretnek összeboronálni.
Mikor Wenczel, a komornyik, jött jelenteni, hogy a hintó előjárt,
az uraságokat sétakocsizásra elviendő, Walterné a felülésnél úgy
intézé a rendet, hogy Incze ismét szemközt kerüljön Hannával.
Azután az egész sétakocsizás alatt Hanna grófnő volt Incze
cicerónéja.
Prága tele van történelmi nevezetességekkel. Azokat mind leírni
Bædeckeren elkövetett plagium volna. A messze középkorból
leszármazó emlékek azok: századok hamvától befeketítve; hajdani
nemzetnagyság kővévált tanui.
– Vannak-e önöknek is ily műemlékeik, fővárosukban,
Budapesten? kérdé Hanna grófnő.
– Nincsenek, felelt Incze. «A mi királyi székvárosunk nem
Budapest, hanem Székesfejérvár, s ott csak multunk romjai vannak:
három korszak omladékain épült fel a negyedik».
Végigjárták a nevezetes egyházakat, Prága büszkeségeit. Az
ezüstből kivert koporsó, ezüst őrangyalaival, a márvány sírboltok,
királyok ravatalaival, nem temették el: élve tartják a multak iránti
kegyeletet.
– Vannak-e önöknek ilyen pompás egyházaik? Ilyen királyi
sírboltjaik?
– Nincsenek. A mi királyaink hamvait kiszórta koporsóikból a
barbar hadviselés, s a mi ereklyénk megmaradt, az egyetlen egy
királyi jobb kéz, a mit minden esztendőben egy napon, augusztus

20-ikán, megmutatnak a népnek, annak bizonyságául, hogy a
királyoknak jobb kezük is szokott lenni.
– Csitt! Hátha valaki meghallja!
A séta-út a Hradzsinba visz föl. Itt a hintót hátra kellett hagyni. A
két úr a két hölgyet karjára véve vezette végig a palotákból alkotott
városrészen. A hajdani hussziták, az újabbkori declaransok nem
jönnek itt szembe rájuk: minden ismerős, a ki velük találkozik,
délutáni egy órakor «wünsch wohl gespeist zu haben»-nel köszönti
őket. Csak a kőszobrok protestálnak még itt.
A legszebb kilátás Prágára a Therezia-rend hölgyeinek palota-
erkélyéről nyílik, melyre beléphet mindenki, a magyarázó
házmesternek fizetett ötven krajczár szabott díj mellett.
Az ide vezető termek egyikében ismét találkozott Incze azzal a
mosolygó arczképpel, mely Walterné albumában a harmadik helyet
foglalta el. Az eredeti olajfestmény itt volt. A házfelügyelő újólag
megerősíté abban az ismeretben, hogy ez a ragyogó szépségű hölgy,
azzal a koronás bársonysüveggel, a mi császárnénk testvére.
(Mindenki tudja, hogy Prágának saját császárja is van, ki a
Hradzsinban lakik).
Inczét megragadta a gyönyörű kilátás ez erkélyről az ódon cseh
fővárosra. Milyen magasztos érzet, ilyen szép kriptába eltemetve
lenni!
Alig veszi észre, hogy a két hölgy közül az egyiknek a szemébe
köny tolúl, midőn az erkélyről széttekint. Az Hanna grófnő. Miért
keresné ő e könyek indokát? úgy sem találná meg!
Csak akkor lett figyelmes e könnyekre, mikor Walterné
felsóhajtott: «szegény Hanna!» s aztán átkarolva rokonát, gyöngéd
erőszakkal vonta el az erkélyről. Hanna ajkaiba harap, hogy
zokogását elfojtsa, s szemei égnek a könyek visszatartásától. Igy
nem testvérét siratja az ember, hanem saját magát.
Vajjon mi baja lehet?

Ha ezt ki tudná találni Incze, saját életének talányát oldaná meg
vele!
– Milyen hideg van e termekben! suttogá Walterné, ideges
borzongással, mikor elhagyták a Therezia-rend palotáját.
A Hradzsin tulsó oldalán megmutatták Inczének azt a kis
egyszerű tanácstermet, a melynek ablakából a cseh tanácsurakat a
mélységbe levetették s azok épen maradtak.
Hanna grófnő is felszökött ez ablak mellvédére s tekintetében az
a gondolat volt olvasható: ne kisértenők meg, hogy lehet az?!
Walterné erősen fogta a kezét. Nőknek, a kik szenvednek, nem jó
magasból alánézni. A repülés vágya támad fel bennük.
Inczének eszébe jutott a dudai hegyi út sziklameredélye. No de a
sors nem ismételi egy ember életében ugyanazt a történetet kétszer.
– «Ez a rejtély nem te neked van feladva megoldásra.»
Pedig mégis mindig azon törte a fejét.
A sétakocsizásból hazatérve, a hölgyeket felkisérték
lakosztályukba, s aztán Walter Leo Inczével ismét visszatért a
fogatához, s rendre eljárta vele a hivatalos helyeket, a hol Inczének
okvetlen be kellett magát jelenteni. Tapasztalá, hogy Walter
kiséretében sokkal gyorsabban elvégezték a dolgát s előzékenyebbek
voltak iránta. Hanem azt tudtára adták, hogy a huszonnégy órája,
mely a prágai időzésre kiszabatott, ma éjjel letelik; tehát még az
éjjeli vonattal tessék odább utazni. Egy rendőrbiztos kisérni fogja a
magyar határig.
Útközben aztán prózai ügyekről beszélt vele a bankár.
– Minő sikerrel működött ön Kaliforniában?
– Én nem jártam a digginsekben aranyat ásni.
– Azt gondoltam is.

– Kertészkedtünk szegény jó angyalommal együtt. Mikor
eljöttünk, eladtuk kertünket, házunkat; még van belőle hatezer
dollárom.
– Az tizennégyezer négyszáz osztrák forint az agióval. S mit
szándékozik ön azzal kezdeni?
– Beteszem a pesti takarékpénztárba, ott biztos helyen lesz.
– Az igaz, hogy a pesti takarékpénztárban az ön pénze biztos
helyen lesz; de nem lesz a pesti takarékpénztár betéti könyve biztos
helyen, ha az ön zsebében lesz. Ön egy esztendő alatt el fogja azt a
pénzt vesztegetni, maga sem veszi észre: hogyan? De én bizonyosan
tudom. Három dolgot ismerek: önnek a jellemét, a magyar ember
általánostermészetét, meg azt a világot, a mibe ön most belemegy.
Jobb lesz, hagyja ön itt a pénzét nálam. Én először is tíz perczenttel
fogom azt kamatoztatni minden koczkáztatás nélkül az ön részéről;
másodszor, biztosítom önt róla, hogy önnek a pénze nem fog
vendégszerepekre kéredzkedni.
– Köszönöm. Elfogadom az ajánlatot. De nagyobb perczentet
nem fogadok el a törvényesnél.
– Azon majd megegyezünk.
Ebéd előtt még e prózai ügyet egészen el lehetett intézni.
Ebédnél ismét összejött az egész család. Helene bárónőnek ma
az a különös szeszélye volt, hogy szüntelen sírásban akarta a
társaságot tartani. Minduntalan olyan tárgyakra vitte a társalgást, a
mikből nem lehetett könyezés nélkül kiszabadulni. Ha Inczéhez szólt,
annak megholt nejét hozta elő; elmondatta magának, hogy éltek,
hogy küzdöttek az élettel oly hosszú ideig együtt? Hogy hozta el kis
fiát magával a koporsóban? Ha pedig Hanna grófnőhöz fordult,
annak szegény Henrikről beszélt, ki most a ravatalon fekszik.
– De kedvesem, inté Helenet Walter Leo, te már egész
siralomházat csinálsz a lakománkból!

– Kedveseink miatt sírni gyönyör!
S ő úszni akart e gyönyörben; kéjelegni, duskálkodni a
fájdalmakban; a megittasodásig élvezni a könyek mámorító
cseppjeit. Poppæának nem voltak kegyetlenebb gyönyörtulzásai.
A bankár házánál szokás volt étkezés előtt és után asztaláldást
imádkozni. A hölgyek ájtatos katholikusok voltak még hazulról.
– Hála Isten; volt mit ennünk – és sirnunk! Amen. Imádkozék
prózai humorral az ebéd végeztével Walter Leo. Igazán mondom,
náthát kaptam a sok sírástól.
Ebéd után egy kis családi hangverseny következett.
Vilmos előhozta hegedüjét s mostohája zongorakisérete mellett
eljátszott valami hosszú «étude»-öt, a mi bizonyosan igen szép
lehetett, de Incze nem értett hozzá. Neki a zenéhez csak «érzése»
volt, de nem «érzéke». Kedélyére hatott a dallam, de nem a
művészet. Neki mindegy volt: Miska czigány, vagy Ole Bull, csak a
nóta legyen szép.
Akkor aztán Vilmos Hanna grófnőt kérte fel, hogy üljön a
zongorához.
Hanna vonakodott. Hivatkozott összetört kedélyére.
Azután Helene is unszolni kezdé: «hát valami ünnepélyes komoly
melódiát: a mi vendégünk kedvéért!»
Incze tisztelgő fejbólintással is csatlakozott e kérelemhez.
Arra aztán Hanna grófnő minden tétovázás nélkül helyet foglalt a
zongora előtti tabouretten s csodálatos szép hajlékony ujjait az
elefántcsont-billentyűkön végigfuttatva, elkezdett egy szép
ünnepélyes komoly melódiát játszani, – a kedves vendég tiszteletére.
A melódia valóban igen szép, klasszikus mű: Haydn remeke. Egy
hymnus, melynek magasztos dallamát egy túlboldog népnek kell

énekelni az angyalok zenekísérete mellett. A szövege így kezdődik:
«Gott erhalte unsern Kaiser.»
Csakhogy «ez a mi kedves vendégünk» egy magyar forradalmi
ezredes, a ki ezt a gyönyörű néphymnust akkor hallotta legutoljára,
mikor a temesvári vesztett csata után a bástya fokáról harsogta azt
alá az osztrák tábori zenekar; hirdetve égnek és földnek, hogy
Magyarország elesett! s ez a «mi kedves vendégünk» most is az
útfélen elmaradó sebesülteket, a futó zászlókat, a leszerelt ágyúkat
látja maga előtt… midőn ezt a melódiát újra hallja.
Hanna újra kezdi a dallamot és változatokat játszik belőle; míg a
mi kedves vendégünk egyszer csak azon veszi észre magát, hogy
sebesültek, ágyúk, zászlók helyett csak két mélabús szemet lát és
tündéri ujjakat, a mik nem is a zongorahúrokon, hanem emberi
szívek idegein játszanak már.
Akkor egyszerre felvillan az a fátyolozott fényű szempár s eddig
ismeretlen tűz sugárzik ki belőle: a tündérujjak rácsapnak a
zongorabillentyűkre teljes erővel, szilajon, s a mi felhangzik a lázadó
érczsodronyokból – az a Rákóczy-induló.
A mi kedves vendégünk aztán látott most már villámló szemeket
is, meg villámló ágyúkat is; lobogó zászlókat és lobogó hajfürtöket
is, egyszerre maga előtt.
Ekkor egy gondolatja támadt. Azt hitte, hogy megtalálta a rejtély
kulcsát. A titkos irásjegyeket, melyek nyomán e bámulatos hölgy
felfoghatatlan jelleme olvasható.
Az utazás a testvér kiséretében… Az út tovább folytatása a
szeretett bátya borzasztó halála után is… A büszke visszautasító dacz
s a rögtön rákövetkező előzékenység, midőn útitársában a magyar
forradalmi vezért felismeri;… a könyek, miket nem bírt visszatartani,
midőn Prágát maga előtt látja;… a néphymnus és a magyar nemzeti
induló egybefűzve;… ez egy politikai combinatio gépezetének
összeilleszthető alkatrésze! Hanna grófnő egyike tán azoknak a

magasztos női alakoknak, a kik egy nemzet helyett éreznek,
gondolkoznak és cselekszenek?
Incze meg volt elégedve ezzel a megoldással. Így egészen
nemes, földfelettien magas régiókban marad a rokonszenvező
szellemek találkozása. A magyar társadalomban és a lengyelben is s
általában minden szerencsétlen nemzet társadalmában van ilyen nő
sok. Ez azokhoz lesz hasonló, a kiknek sorsuk az, hogy imádva
legyenek; de nem szeretve.
Ez a gondolat lecsillapítá lelkét. Most már félelem nélkül kezet
mert szorítani Hanna grófnéval. Azt hitte, hogy megérté őt. Nem
látta benne többé a szép hölgyet, csak a cseh honleányt.
Ezzel a gondolattal vett bucsút tőle.
S ez a gondolat nagy áldás volt ránézve.
Vannak babonás emberek, a kik azt hiszik, hogy kedves halottaik
láthatlan szellem alakban is környezik őket, s nagy szorongattatások,
kétségek, töprengések idején ha hirtelen egy szabadító ötletük
támad, azt ők sugalták titokteljesen.
Talán neki is Serena súgta ez eszmét?
Nem volt az igaz; de Inczének jó volt azt hinni, hogy úgy van.
Ez a csalódás volt az ő menedéke; mert azokban a szép villámló
szemekben az ő végzete lakik…
ISMÉT ITTHON.
Mikor Áldorfai Incze Orsovánál a csónakba lépett, mely az utolsó
menekülteket szállítá a túlpartra, kihajolt a csónak párkányáról s egy
kavicsot vett fel a Duna fenekéről. Ez még hazájának egy darabja
volt. Ezt a kis gömbölyű követ azóta mindig magával hordozta. Ezzel
a kis kővel körüljárta a földtekét. Most ismét visszakerültek együtt: a

kő is, meg ő is. Vajjon érzi-e azt a haza, hogy hiányzottak belőle:
egy kavics, meg egy ember?
Egy napot időzött Bécsben: volt kihallgatáson a legmagasabb
úrnál, hódolatát kifejezni a hazatértet megelőző amnestiáért s aztán
sietett Pestre.
A határnál elmaradt rendőri kisérete s Pozsonynál már új alakok
foglalták el mellette a coupéeban az ülőhelyeket: minőket azelőtt
még sohasem látott. Urak népies viseletben.
Azt a 48–9-iki szabadságharczot úgy vítták végig, hogy az
európai divatot megtartották mellette, s a magyar nemzeti viselet
csak mint egyenruha, díszöltöny, vagy népies gúnya szerepelt a
maga helyén és emberén. Most pedig minden ember sarkantyút
visel: mintha senki sem akarna infanterista lenni többé.
Hogy itt most már teljes szabadság van, azt útitársai beszédéből
kiveheté. Őt nem ismerte senki s nem igen törlészkedett hozzá, még
egészen yankeenek volt öltözve; bő zsákkabátban, magas
czilinderben, kaucsuk felczipőkben, kezében esernyővel.
Este későn, a mint a vonat megérkezett, a vendéglőből, a hova
megszállt, egyenesen sietett a képviselői klubbot fölkeresni.
Gondolta, hogy ott régi ismerősökre talál.
A mint a népesebb utczákból egy mellékutczába betér, három
férfival találkozik szemközt, kik széltiben elfoglalták a járdát.
Az egyik egy magas, szikár alak, veresrépa szinű orral, s
villahegyesen kifent fenyegető bajuszszal. Túri süvege fél szemére
nyomva; azon hatalmas sastoll; czifrán kihányt szűre félvállára vetve,
s a szabadon maradt kezében egy hosszú csáti bot.
A másik férfiúnak termetéhez a legnevezetesebb contingenst
szolgáltatta a has; a jóllét túlságától piros az arcza, s kövér tokája
eltakarja a selyem-nyakravalót. Bajusza fel van kunkorítva, mint a
gácsér farka, nagy mesterséggel. Melege lévén, a süvegét a kezében
hordja, s agyarára fogott pipájának olyan kurta szára van, hogy fél

szemét behúnyva kell tartania, hogy ki ne süsse a pipa-kupakkal.
Fegyvere nincs más, csak egy nagy sallangos bőrdohányzacskó, de
abban van három font dohány beletömve; ha azzal valakit fejbecsap,
hát az mehet magát beiratni a temetkezési egyletbe.
A harmadik pedig egy kis tömzsi ember, termetéhez nem illő
nagy fejjel, melynek rengeteg szakálla a három sor mellénygombot
eltakarva tartja; s aztán erre a nagy fejre fel van téve valami olyan
kicsiny karimájú kalap, mint egy palacsintasütő, s ehhez a kicsiny
emberhez jár egy olyan hosszú és vastag csibukszár, mely lehetne
kozák dárda, meszelő, és más egyéb, csak pipaszár nem.
Mind a három polgártársnak igen széles kedve van: most jönnek
ki a «kis pipából».
Incze veszi azt észre s szépen a falhoz lapulva igyekezik mellettük
elsuhanni. De veszedelmére szolgál felötlő idegen viselete, a mit már
e délkör alatt senki sem hord. Belébotlanak.
– Szervus czilinder! dörg eléje oroszlánhangon a kövér termetű
polgártárs.
– Üsd a fejébe! kiált fel onnan alulról, borízű hangon a kis petten
ember.
A hórihorgas nem szól semmit, de megcselekszi: olyat huzván
Incze figarójának a tetejére a csáti bottal, hogy az menten
összehorpadt.
«No ez szép üdvözlet, első belépésre!» gondolá magában Incze;
azonban ő sem volt rest a «fogadj Istennel»: ott kinn jól megtanulta
a boxolást. A jobb öklével úgy orronütötte a hórihorgast, hogy az
rögtön elvesztette a szép világot maga elől; a bal öklével felcsapta a
kis seprüszakállunak az állát, hogy az azt hitte, hogy most a feje
elrepül nála nélkül, s egy harmadik lökéssel a kerekhasúnak tisztelé
meg teste legkiválóbb alkatrészét, a mitől az rögtön leült a kőre, s
egy miatyánk hosszáig nem tudott lélekzetet venni.
Azzal ott hagyta őket és sietett tovább.

Siettében meggondolta a dolgot.
Bizony Isten mégis annak a három jó embernek volt igaza. Ez a
nemzeti viselet a mi egyedüli fegyverünk az idegen invasio ellen. A
hová a sarkantyús lábunkat leteszszük, odáig van Magyarország.
Nem akarja a világ meghallani, a mi igazságunk van; tehát lássa
meg rajtunk gomb, zsinór és toll-alakban. És végre a nép tudja meg
a ruhánkról, hogy vele egyenlők vagyunk, hogy nem akarunk urak
lenni vele szemközt: Amerikában a napszámos is városi divat szerint
öltözködik; nálunk a mágnás is paraszt viseletet hord: egyre megy ki
s így is jól van.
Most már Incze nem a klubba sietett elébb, hanem egy ruhaárus
boltba, hogy átöltözzék igaz hazafinak.
Egy boltban lakott szabó és csizmadia. Mind a kettő derék
szabadelvű honpolgár volt. Olyan szépen felöltöztették Áldorfai
Inczét, hogy maga is gyönyörűséggel nézhetett végig magán, mikor
a tükör előtt megállt.
Mert szó a mi szó, a magyar viselet nagyon szép.
Most azután, hogy a váczi-utczán végigment, egyszerre minden
ember ismerőse lett. Tízen, huszan ölelkeztek össze vele, agyba-főbe
csókolták; hét közül hatra nem emlékezett; de őt jól ismerte minden
ember; utolja felé már tolongás kezdett támadni, a merre járt, s
messziről kiabálták, hogy «éljen Áldorfai Incze!»
Ő pedig mondá magában: «köszönöm neked ezt, óh sarkantyús
csizmám!»
Régi honvéd bajtársak, új képviselő kollégák körülfogták, vitték
magukkal vacsorálni. Tele volt minden ember jó kedvvel,
dicsőséggel, arany reményekkel; áldomás, adoma folyt a poharazás
közben; mindannyinak központja Áldorfai Incze volt. Ő maga pedig,
az ünnepelt hős, hallgatag arczczal ült a zajos társaság közepett.
Nem értette, hogy minek örülnek olyan nagyon s mit remélnek olyan
hévvel? Nem értett belőle semmit. Idegen volt idehaza. Különösen

nyomasztólag hatott rá, hogy őt ilyen nagyon ünnepelik. Hát már
annyira megfogyatkoztak a nagy emberek Magyarországon, hogy
egy harmadrendű celebritás is epochát alkot a megjelenésével? De
talán mindez csak a viszontlátás örömeinek róható fel?
Erre a kétségre nagyhamar megkapta a választ Incze.
Éjfél körül lehetett, midőn hazakerült a távolabbi vendéglői
szállásra. Készült levetkőzni.
Hanem az nem megy olyan könnyen. Azt a nyalka, szűktorkú
sarkantyús csizmát lehúzni nem olyan tréfadolog ám, mint a
mehádiai útszoroson át egy menekülő dandárt keresztülhúzni.
Sehogysem boldogult vele. Hozzá volt az nőve a lábához. Már a
famacskát is összetörte a hasztalan küzdelemben a csizmájával.
Fogva volt benne. A csizma maradt a győztes.
Azonban a csizmának igaza volt. S hogy azt le nem tudta húzni a
lábáról, ez ismét a providentiális esetek közé tartozik. Mert a midőn
már megadni készült magát a leküzdhetlen végzetnek, künn az
utczán nagy zajt hall közelíteni, melyből nemsokára felharsan a
lelkesítő Rákóczy-induló zenéje, el-elnyomva ezernyi éljenkiáltástól s
a házfalakra veres tűzfény vetődik, mely bevilágítja az ő szobáját is.
Fáklyás zene közelít.
«Most van a szüretje», gondolá magában Áldorfai Incze; s ő is
felölté ismét dolmányát, hogy megnézze – kiváncsiságból – e szép
jelenetet. Az utcza megtelt néppel, fő főt ért, volt ott háromszáz
fáklya is, délczeg ifjak kezében; a tömegben félkört tágítottak a
fáklyákkal a szónokok számára, épen a vendéglővel szemközt.
Itt bizonyosan valami nagy celebritás lakik alattam az első
emeleten, gondolá magában Áldorfai Incze s találgatta magában, ki
lehet az? A mint a nagy éljenriadalt még hangosabb «halljuk»
kiáltásoknak elvégre sikerült elcsillapítani, kiállt a félkör közepére egy
délczeg fiatal polgár, s mondott egy hatalmas szónoklatot a
megtisztelt egyéniséghez, ki nevét napsugarakkal irta fel hős

tetteiben a haza egére; s ki a mellett szintoly bölcs, mint a milyen
vitéz; s egyrészt hivatva van, alkotmányunk hajóját egy kezével a
békés kikötőbe kormányozni, másrészt a másik kezével zászlóinkat
lobogtatva, tűzön és viharon keresztül diadalra vezetni, kinek
jelszavát az egész nemzet várja, s a kit követni fognak ifjak és vének
életbe és halálba.
«Ugyan ki lehet az a minden tekintetben utólérhetetlen nagyságú
férfiú, a ki a bölcseséget és a hősi erényeket ennyire egyesíteni tudta
magában?» gondolá Áldorfai Incze s maga is kiváncsi volt
megismerni ezt az urat, s várta, hogy mikor jön elő valahonnan a
hozzá intézett üdvözlő beszédre megfelelni? De az üdvözlött csak
nem állt elő, pedig odalenn ugyancsak kiáltozták már az «éljennel»
kevert «halljukot».
Egyszer siető léptek közelednek a folyosón végig sarkantyút
pengetve, Incze ajtaja felé; felnyílik az, s nagy robajjal tör be rajta
egy tizenkét tagú küldöttség, előrebocsátva a beköszöntő éljent.
Incze szemközt találja magát az esteli rencontre hőseivel. Legelől
a hórihorgas, kinek még most is félreáll az orra s fel van dagadva,
mint egy uborka, a kapott ökölütéstől. A másik kettő is itt van.
Incze azt hiszi, ezek most verekedni jönnek. De nem ám. Ragyog
azoknak az arcza az örömtől.
– Hozott az Isten, szeretett barátom, nagy hazámfia! Kiált, karját
nem birkozásra, de ölelésre terjesztve szét a hórihorgas. Tán nem is
ismersz már?
(Hogyne ismernélek? gondolta magában Incze, hiszen ezelőtt
három órával ütöttem így félre az orrodat.)
– Én vagyok Sámsoni Lenczi, őrnagy, hajdani bajtársad, a kivel
annyi csatában harczoltál együtt.
Incze csakugyan kezdett rá emlékezni, hogy látott valaha
ilyenforma nemzetőri őrnagyot valami ütközet után tartott «lakoma
alkalmával», ki «ott» igen vitézül viselte magát.

Ezzel összeölelte, csókolta Inczét, s bemutatta neki a másik két
urat is: a kis tömzsi «Csángó Náczi», a potrohos pedig «Ordasy
Miczu». Az előbbinek fel van kötve az álla fehér keszkenővel: tán a
foga fáj?
Mind «régi» jó ismerősök.
– Azért nem ismertél tán rám, mert az orrom így megdagadt.
Hübnere van a dolognak. Az este egy német spitzli kötött belénk az
utczán, csúfolta a kaczagányunkat. Én aztán jót húztam neki a csáti
bottal a czilinderére: s arra a spitzli vaktában úgy hozzáütött az
orromhoz, hogy ilyen lett belőle.
Lenczi bácsi ezt olyan humorral adta, hogy mindenkinek nevetni
kellett rajta.
– Hát aztán a czilinderes spitzliből mi lett? kérdé Incze
malicziával.
– Az meghótt! felelt rá komolyan Lenczi. De csak ne lett volna a
félkezem a guba alá dugva!
– Akkor kétszer halt volna meg! tódítá Incze.
– Hanem hát már most gyere le a tisztelőidhez. Azt hiszik, hogy
alszol.
– Én? Hát nekem szólt a megtiszteltetés?
– Hát ki a hét-világ csodájának? Nem hallottad azt a szép
dikcziót? Most gyere le szaporán! Azért jöttünk deputáczióba, hogy
levigyünk: aztán gyujts rá valami szép beszédre.
– Hagyjatok nekem békét, szabódék Incze. Nincs én nekem
semmiféle érdemem a világon, a mivel erre az ováczióra
rászolgáltam volna. Nem tudok én nektek semmit mondani.
– Csak te gyere le szaporán. Majd sugom én neked, hogy mit
mondj! Majd a szádba adom én a szót szépen.

– Ejh, én nekem most nincs semmi kedvem a szónokláshoz,
monda Incze boszusan. Mondjátok a tisztelt polgártársaknak, hogy
lefeküdtem, el vagyok rekedve, izzadok: nem mehetek!
Könnyű azt mondani! de mikor az embert egyszerre csak
megrohanja orozva két markos legény kétfelől, átnyalábolják a
lábszárait, felkapják a levegőbe, akkor aztán már négy lába van s
menni kell, a merre azok viszik. Azok pedig leviszik az utczára s
addig le nem eresztik a földre, míg a megtisztelő közönséget egy
szép szónoklattal ki nem elégíti. Incze nem is maradt azzal adós; s
habár nem is azt mondta el, a mit Sámsoni Lenczi sugott volna neki
(pedig az sokkal nagyobb hatású lett volna), de még is beszélt annyi
mindenfélét, a mi eszébe jutott, hogy alig tudta végét szakítani.
Mikor aztán letették a földre, még egy ember nem ereszté el a
gallérját a markából. Az egy ujságiró volt. Arra ösztönözte, hogy adja
oda neki a holnapi lap számára azt a szép beszédet – irásban.
Nem szabadulhatott meg tőle, fel kellett őt vinnie a szobájába s
lediktálni, a mit beszélt.
Egy nyeresége mégis volt ebből is. A becsületes riporter, látva,
hogy a nagy hazafi nem tudja a csizmáját lehúzni a lábáról, segített
neki a keserves munkában, melyet végre még is csak a kivánt siker
koronázott.
Inczének nem jutott akkor eszébe, mikor a hadjárat alatt lerohadt
a lábáról a csizma, hogy az a hazáért történik; hanem most, hogy
ezt a csizmát le kellett húzni a lábáról, kezdé megismerni, hogy mi a
hazaszeretet?
Úgy hajnal felé aztán hozzájutott, hogy lefeküdjék. De még egy
meglepetés várt rá, mielőtt a gyertyáját elfujja. Egyszer csak csörren
az ablak, s a két üveglapon keresztül berepül a szobájába egy darab
kő s odaesik az ágyára.
Ez az ellenvélemény!

Szép gömbölyű kavics volt, a milyen a dunaparton terem. Incze
eltette azt emlékül a másikhoz. Van már kettő!
Alkalmasint valami hivatalából kiküszöbölt Beamtertől jöhetett ez
a merénylet!
Mi pedig azt hiszszük, hogy maga a buzgalmas reporter hajította
be azzal a kavicscsal az ablakát, azért, hogy másnap megirhassa,
milyen orsini-bombát vetettek be a reactio zsoldosai a nagy férfiu
hálószobájába; a mi annak a hirét még jobban fogja növelni s
személyének becsét még magasabbra emelendi a közvélemény előtt.
Kitelik tőle!
A HIRESSÉG ÁTKA.
Áldorfai Incze napról-napra jobban tapasztalá, hogy őt idehaza
nagy embernek tartják. El nem tudta gondolni, hogy mi okot
adhatott rá?
Egyik megtisztelés a másik után érte. Fényképészek állították
camera obscuráik elé, minden elképzelhető állásban; rajzoló
művészek ültették le, életnagyságú arczképének lemásolása végett.
Egy faragó genie kifaragta a szobrát körtefából; egy lelkes honleány
domborműben idomította azt ki viaszból s egy reményteljes
szabólegény kivarrta azt egész alakjában posztóból. Egy esztergályos
botokat készített Incze fejével bunkó helyett; egy kartonfestő
zsebkendőkre nyomatta a képét; kilencz ujságíró kiadta azt a
hetilapjában. Irtak róla életirást, novellát, költeményt, soha meg
nem történt kalandokkal hozva őt kapcsolatba; még szinpadra is
hozták s elfoglaltatták vele Buda várát lóháton. Kalapot, nyakkendőt
készítettek a nevére, rostélyos sültet kereszteltek el róla. Meghítták
keresztapának újszületett polgártársakhoz, s elvitték körutakat tenni
veszélyeztetett választókerületekbe. Kapott ajándékokat is:
díszkardot, dohányzacskót, hímzett szőnyeget, antik mentekötő

lánczot, paripát és agarat. Megválasztották keletkező humanitási
egyletek elnökükké; utoljára még egy hirlapot is kapott ajándékba,
mely szerencséjének tartotta, ha őt szellemi vezérének
nyilváníthatja.
Ez mind igen szép volna, ha az embert nagy férfiunak tartják;
bárha saját maga meg van is felőle győződve, hogy nem az; hanem
nagyobb baj az annál, hogy az embert egyúttal gazdag uraságnak is
tartják, mikor ő maga nagyon jól tudja, hogy abból ugyan semmi
sincs.
A nagy férfiu szerepe igen sok elkerülhetlen kiadással is jár.
Mindazok a megtiszteltetések okvetlenül vonnak maguk után
valami viszonzást is. Költségbe kerül a klub, a körutazás, az egyleti
elnökség, a hírlapi szellemi vezérség, a komaság, az arczképek, még
a kapott ajándékok is. Ha a nagy férfiunak valaki egy pisztolyt
ajándékoz, legalább is egy ágyút vár tőle vissza.
Még azok a legirgalmasabb tisztelői, a kik olyan nagy dolgokat
kivánnak tőle, a miket lehetetlen teljesíteni.
Rokkant vitézek jönnek hozzá követelni, hogy fizettesse ki nekik
az országgal a 49 óta elmaradt évdíjaikat. Csizmadiák,
nyereggyártók bíznak benne, hogy a magyar hadseregnek
kiszolgáltatott holmiaik fizetetlen árjegyzékét ki fogja egyenlíteni.
Síránkozó özvegyasszonyok hozzák eléje dugaszban tartogatott
magyar bankjegyeiket, könyörögve, hogy váltassa be azokat más
forgalomban levő pénznemekre. Nagyérdemű hazafiak vezetik be
hozzá jól készült fiaikat, hogy szerezze be őket valami hivatalba.
Pörlekedő volt úrbéresek, és nemes atyafiak mutogatják be neki
zsíros pörirataikat, azt várva tőle, hogy a sérelmes itéleteket majd
megsemmisíti. Deputatiók jönnek hozzá, hogy védelmezze meg a
népeket a falun az adóexecutió ellen. Ezeknek mind könnyü felelni.
(A «kérvényi bizottsághoz» utasittatnak.)
Hanem aztán vannak igények, a mik elől becsülettel nem lehet
meghátrálni s azok mind rátalálnak Inczére, mintha az ő szállásán

adtak volna egymásnak közös találkozót. A kötelességszerű áldozat
mellett aztán, melyhez tisztességes ember szívesen szokott járulni,
jön egész Sisera hada az orczátlan kéregetésnek s az
emberismeretlen nagy férfiu nem tud különbséget tenni közöttük.
Hajh, mindez nem így volna, ha a jó gondos feleség megvolna, a
zsaroló had nem meri oda betenni a lábát, a hol asszonyt érez a
háznál. Incze kész préda e hadnak. A képviselői napidíj nem elég a
százszájú polypnak.
Az első negyedév végével már kénytelen Incze Walter Leonak
irni, hogy küldje meg a tőkéje kamatját; nem bánja már, ha tiz
perczent lesz is, Walter aztán küldött neki még a jövő évnegyedre
valót is.
Incze arra a gondolatra jött, hogy az ő polgártársai talán annak a
fejében, hogy ő most jött Kaliforniából, azt hiszik, hogy teleszedte
magát aranynyal?
De bizony még annál is furcsábbat hittek felőle.
Ki terjeszthette el felőle? azt nem tudni; de annyi bizonyos, hogy
általánosan beszélt adat volt Áldorfairól, hogy ő Napoleon herczeg
legbensőbb megbízottja, a kinek a herczeg százezer forintot adott át,
arra a czélra, hogy Magyarországon propagandát csináljon. Azt is
tudták, hogy minő furfanggal csempészte be Incze azt a százezer
forintot a határon? ki árulta el azt az osztrák rendőrségnek s hogy
lett ez rászedve hiresen?
Csak Incze nem tudott felőle semmit. Ő a krimi hadjárat óta nem
is látta a herczeget. Pedig mennyi helye lett volna annak a gyönge
százezer forintnak!
Veszedelmes állapot valakinek egy ilyen potya százezer forinttal
hirbe keveredni!
Azokkal még kevesebb baj van, a kik arra személyes igényeket
támasztanak; de sokkal nagyobb a veszedelem arról az oldalról, a

hol a czélt magát veszik komolyan s várnak belőle országdöntő
eredményeket.
Incze meg nem birta magának magyarázni azt a helyzetet, hogy
ő vele egészen komoly, szilárd jelemű, derék hazafiak elkezdenek
messzeható politikai és strategiai tervekről értekezni; mintha
azokhoz az ő beleegyezése kivántatnék. Hallgatását, naiv
tudatlanságát kiszámított titkolózásnak veszik, míg a vendéglői
közvélemény fényes nappal népes utczán, tele torokkal kiáltja őt
leendő vezérének s üldözi őt – kitüntetéseivel.
S ez a jó hír a magánéletre is kihat.
Incze feltünő szép, délczeg férfialak. Nem több harminczhat
évesnél. Most vesztette el a nejét. Akad, a ki a megszomorodott
özvegy vigasztalására gondol.
Egy délután (a délelőtt az országgyülésé) két hölgy látogatta meg
Áldorfai Inczét a lakásán. Nem első eset volt ez.
A beküldött látogatójegyeken e két nevet olvasá Incze:
– Stomfay Gídeonné, Caesarine, honvéd őrnagy özvegye.
– Stomfay Fatime Serena.
Nagyot dobbant rá a szíve. Sejtelme volt.
Mindig üldözte az a gondolat, hogy most, mikor börtönök, sírok,
és hivatalbureauk kiokádják halottaikat, egyszer csak Gideon is előtte
terem: s ha az előkerül, bizonyosan országgyülési képviselő lesz;
sokkal szabadelvűbb, mint ő! Hogy fogja az őt majd a hazafiuság és
szabadelvűség magaslatából lemennydörögni, – a hátramaradottat!
Azonban hát Caesarine «özvegy»; ha ugyan a látógatójegyeknek
hinni lehet.
Sietett őket elfogadni.

Caesarine és leánya a legutolsó magyar divat szerint voltak
öltözve: az anya rókatorkos, sokpitykés gyöngyösi mentében,
aranycsipkés, veres bársonyszalagos főkötővel; a leánya kifűzött
bársony vállderékkal, dudoros csipkeujjakkal, hímzett előkötővel,
leeresztett hajfonatokkal, s fején gyöngyös pártával.
Caesarinenek még mindig voltak igényei a világ irányában. Arcza
még mindig eltűrte a szépítő szereket; bárha azon szerencsétlen
arczok közé tartozott is, melyek olyanok, mintha két különböző
kölcsönkért arczból volnának összeállítva: két egymáshoz nem illő
fél. A leánya egy felserdült növendék, se gyermek se leány még;
kifejletlen termet és vonások; semmit sem mondó tekintet; csupán a
két nagy fekete szem, hosszú selyem pilláktól árnyazva, sejteti az
egykor kifejlendő bűbájt. Különben a tizenhárom éves gyermekek
minden tulajdonával felruházva; ügyetlen és szemérmetes, vagy
bölcselkedő és merész, érzékeny vagy durczás, hallgatag vagy
közbekottyanó.
Incze nem óvhatja meg magát attól, hogy keresztleánya kezet ne
csókoljon neki. Hiszen tartozik vele.
Leülteti őket s némi kiváncsisággal vizsgálja keresztleánya
arczvonásait.
– Egészen az apja képe! sipogja Caesarine. Szegény jó
Gideonomé. Ilyen korán jutott árvaságra!
Incze egy veres plajbászt tartott a kezében, azzal a szándékkal,
hogy valahányszor Caesarine egyet hazudik, mindig egy vonást
csináljon vele az előtte heverő papirra.
Tehát – Nro 1.
– S Gideon csakugyan meghalt? Kérdezé Caesarinetól.
– Elvérzett a hazáért. A temesvári ütközetben lehelte ki nemes
lelkét a dühös kozákok dárdáitól átszúrva.
Nro 2, jegyzé Incze.

– S azóta kegyed egészen magára van hagyva?
– Elhagyva az egész világtól. Napszámból tartom szegény
gyermekemet.
Nro 3, szólt a veres iral.
– Mit tehetek önök sorsának enyhítésére?
– Egyedül ön tehet és mindent. Leányomnak gyönyörű sopran
hangja van. Egy Malibran rejlik benne. De a világ nem tekinti a
tehetséget, hanem a szépséget. Én pedig úgy őrzöm gyermekem
tiszta erényét, mint a szemem fenyét, s készebb volnék neki vitriollal
ragyákat festeni az arczára, hogy irtózzanak tőle, hogysem elbukni
engedjem.
Nro 4.
– Én a mi főurainkhoz fordulni nem tudok; nem akarok. Én ugyan
személyes tapasztalásból nem ismerem őket, de a világ annyi rosszat
beszél felőlök.
Nro 5.
– Önbe van vetve egyedüli reménységünk. Ön «nagy ember»;
egy szava mindent tehet. Ha ön egy szót szólna gyermekem mellett
a nemzeti szinház igazgatóságának, hogy vegyék fel, bárha eleinte
csak az énnekkarba is, az kétségtelen sikert aratna számunkra.
Incze végre megúnta a rovást jegyezni s megigérte, hogy majd
beszélni fog a «comitée» tagjaival.
De Fatimének nem engedte többé megcsókolni a kezét.
– Csókolja meg tehát a homlokát, szólt Caesarine, hiszen
keresztleánya!
Azt hát megtette neki.

Még aznap este a casinoban elő is hozta e tárgyat az ismerős
«comitée»-tagoknak s másnap Fatime szerződtetve volt a
szinházhoz.
Incze ugyan legkomolyabb képpel esküdött rá, hogy neki
legkisebb köze sincs az egész leányhoz, csupán szép tehetsége
végett szólal fel, hanem azt bizony nem hitte neki senki. Azontúl
köztudomású dolog lett, hogy neki is van valakije a szinháznál, a kit
protegál. Skandalum is, hogy eddig nem volt.
Ő pedig úgy elfelejtette az egész leányt anyjával együtt, hogy
többet nem is kérdezősködött felőlük.
Volt neki egyéb dolga, mint szinházba járni.
A mint a halomra tornyosult országos teendők néhány napi
szabad időt engedtek neki, ezt arra használta fel, hogy leránduljon
választókerületébe, magát választóinak megmutatni.
Itt Pesten azt sugdosták lerándulásáról (tele torokkal), hogy a
guerilla-harczot megy organizálni a Bakonyba.
Ismét a kolostor volt szállomáshelye. Ott találta még mind az
öregeket, s többeket hajdani ifju társai közül, kik együtt harczoltak
vele a csaták végeig s mikor elmult a harczok ideje, visszatértek
ismét a kolostorba, felvették a skapulárét, s miséztek és
praelegáltak, mint azelőtt.
Incze még egyszer látta a kis szobát, melyben hajdan lakott, a
lefüggönyözött madonnaképet, melyhez annyiszor imádkozott, még
egyszer bejárta az emlékezetes helyeket, mikhez annyi nehéz
gondolatja volt fűzve. Ez eszmék mind a tenger fenekén feküsznek
már. Jobb lett volna tán azokat soha ki sem gondolni!
A jámbor földnépével érintkezve, meggyőződött felőle, hogy az
még most is a régi: – önző és áldozatkész, panaszteljes és
békülékeny, ragaszkodó és ingatag. Minden faluban tartott egy
szónoklatot, néhol kettőt is, a hol németek és magyarok laktak
vegyest. Szekere tele volt koszorúkkal, mire visszatért.

Mikor visszakerült Budapestre, a legelső volt Sámsoni Lenczi, a ki
üdvözletére sietett.
Elhozta neki az összegyüjtött hirlapokat mind, a mikben *-i
körútjának diadalai le voltak irva.
Az ilyen előzmény nagy következményeket szokott maga után
vonni.
A ki ennyira birja a nép kegyét, arra nagy kötelességek várnak.
– Neked most rögtön el kell menned Mike-Bogyára, a
követválasztást megejtetni. A régi leköszönt, új választás lesz.
– De hát mi közöm nekem Mike-Bogyával? Azt sem tudom, hol
van?
– Majd megtudod. Együtt megyünk. Én is veled megyek. Ezt igy
határozta a nemzeti comitée.
Sámsoni Lenczi maga nem volt képviselő, de a választásoknál
döntő befolyású egyéniségnek tartatott. Itélt élők és halottak felett.
– S kit kell megválasztani Mike-Bogyán?
– Nem kisebb embert, mint magát Napoleon herczeget.
– Eredj már! ne csinálj belőlem bolondot! Mi köze Jeremiás
herczegnek a magyar alkotmányhoz?
– Csak lassan a testtel! Tudom én, hogy semmi köze sincs hozzá,
hanem hát Mike-Bogyán a conservativek nagy erőben vannak. Az
oláhság mind velük tart. Az ő jelöltjük báró Schafskopf nagybirtokos
azon a vidéken. Ezzel mi nem mérkőzhetünk. Hanem ha egy olyan
nagy celebritást állíthatunk fel ellenében, mint Napoleon herczeg,
azzal az oláhokat megnyerhetjük, s Schafskopf megbukik egészen.
– De hát mi mit nyerünk vele? hiszen Napoleon herczeg csak
nem jön mi nekünk ide magyar képviselői széket elfoglalni.
Á

– Azt bölcsen tudom én is. Ámde ha egyszer meg van törve a
jég, a többi magától jön; – ha egyszer Schafskopf báró fiascot csinál,
akkor aztán a leköszönt Napoleon herczeg helyébe bizton
felállíthatjuk a mi jelöltünket, Kondorossy Mukit; egyedül maradunk a
téren.
Incze még szabadkozni kezdett; hanem aztán Sámsoni Lenczi
annyira komolyan fogta, hogy át kellett látnia, miszerint e megbízást
elfogadnia a legszentebb hazafiúi kötelesség. El is ment Mike-
Bogyára s nagy küzdelmek árán megválasztatta elébb Napoleon
herczeget magyar képviselőnek s a mint annak távirati lemondása
megérkezett, kitűzte új követjelöltnek Kondorossy Mukit és újra
végigjárta ennek a zászlóival a kerület falvait.
Nehéz munka volt ez! Napoleon herczeget csak tudta még, a
hogy tudta, ajánlani a választóknak, de Kondorossy Muki érdemei
merőben ismeretlenek voltak előtte. Egyszer találkozott vele
életében s akkor azt kérdezte tőle az érdemes képviselőjelölt, hogy
«igaz-e, hogy a pesti követek a buffetben az ülés alatt nyers halikrát
esznek?» s miután ezt Incze igenlőleg bizonyította, öntudatosan
mondá rá: «de már mi csak maradjunk a jó kassai sonkánál».
Többet nem tudott felőle.
Hanem azért az ő ajánlatára megválasztották Kondorossy Mukit
képviselőnek.
A diadal ime tökéletes volt.
Hanem van aztán egy árnyoldala a diadalnak. Néhány nap mulva
hoznak Inczéhez egy hosszú árjegyzéket a választási mulhatatlan
alkotmányos kiadásokról, a mik mennek valami négy ezer forintra.
Ki fog ezeknek helyt állani?
– Bizony, kedves czimborám, mond Sámsoni Lenczi, ezt nekünk
kettőnknek kell megfizetnünk.
Incze nagyon bámult erre a szóra. Először is nem értette, hogy
mire lehetett annyi tenger-pénzt kivetni, a midőn senkit sem

vesztegettek meg a szavazatáért? másodszor pedig sehogy sem
ment a fejébe, hogy miért kelljen ő neki két ezer forintot fizetni a
végett, hogy ne Schafskopf báró egye a buffetben a káviárt
zsemlyével, hanem Kondorossy Muki a kassai sonkát, eczettel,
paprikával?
Azonban már fizetni kellett. Incze nem tagadhatta el, hogy ő volt
a pártelnök Mike-Bogyán, s ha nem tudta eddig, hogy ez a hivatal
együttjár a költségek előteremtésével, hát már most megtanulta.
Még szerencsésnek nevezhette magát, hogy a megtérítendő
kiadások halaszthatóbb felét választási primipilus társa, Sámsoni
Lenczi elvállalta magára.
Kénytelen volt Walter Leonak írni, hogy küldjön a számára a
tőkepénzéből két ezer forintot.
Igazán fájt a lelke, mikor ezt a pénzt kiadta a kezéből. Mily
sanyarogva szerezte ezt megbodogult kedvesével együtt; mennyit
kellett a szegény asszonynak dolgozni, hogy ilyen két ezer forintot
összegyüjtsön s mennyi élvezetet megtagadni magától, hogy a
szerzett pénz megmaradjon; s ezt ő most három nap alatt betöltötte
a szomjú – homokba!
Felfogadta magában, hogy ez volt az utolsó leczke.
Dehogy volt utolsó.
Egy hét mulva betoppan hozzá Sámsoni Lenczi s minden
bevezetés nélkül előhúz a zsebéből egy váltót. (Nem is szól többről,
csak két ezer forintról.)
– Itt van egy váltó, kedves pajtásom; én vagyok az elfogadó,
Kondorossy Muki czimboránk a kibocsátó: te lészsz rajta a forgató.
Egész simán, mint a parancsolat.
Incze, mint aféle tudatlan yankee, a ki soha életében nem látott
másforma váltókat, mint üzletieket, a mik adósra és hitelezőre
vonatkoznak, a maga naiv észjárásával azt gondolta, hogy ime, ez a

két derék ember most ő neki a kifizetett két ezer forintjáért egy
váltóval akar kárpótlásul kedveskedni. Pontos, lelkiismeretes
czimborák ezek. Ezt nem szabad visszautasítani. Tétovázás nélkül
ráirta a nevét a váltó hátára forgatmányosnak s készült azt a
fiókjába eltenni.
– De ne csukd el a fiókodba, pajtás! kiálta rá Sámsoni Lenczi.
Mert arra nekem van szükségem!
– Neked? kérdé elbámulva Incze.
– Nekem hát. Én akarok erre kétezer forintot kapni a
takarékpénztárból.
– De hát akkor minek irattad rá az én nevemet is? hisszen én
nem akarok pénzt kérni a takarékpénztárból!
– Hiszen nem is neked adják azt, hanem nekem.
– Neked? Az elfogadónak? Ezt én nem értem.
– Hja, pajtás, nálunk egészen másforma váltó-törvények
regnálnak, mint arra ki Amerikában. Ez nálunk így megy.
– De már most mégis az lesz belőle, hogy tartozom a
takarékpénztárnak; Muki tartozik én nekem; és se én nem láttam a
takarékpénztár pénzét, se Muki az enyimet.
– Csak te nyugodjál meg benne. Ennek így kell lenni.
Incze meg volt lepve a váratlan támadás által, lefegyverezve és
fogolylyá téve.
Mert a pisztoly elé kiállani mindennapi bátorság dolga; olyan vitéz
ember is van elég, a ki az ágyútelepet megostromolja: de ahoz, hogy
valaki meg merje mondani a «jó pajtás»-nak, hogy «én biz a te
váltód hátára nem irom a nevemet», oly hősi elszántság kivántatik, a
minőnek példái csak nagy ritkán vannak feljegyezve Magyarország
történetében.

Incze nem birt e rendkivüli bátorsággal. Ott hagyta a nevét Muki
váltóján.
Lejáratkor azután természetesen neki praesentálták a váltót. Ezt
tudtuk előre.
Inczét már ez egészen kihozta a sodrából.
Azt hitte, egészen igazságos utat követ, a midőn azt kivánja,
hogy vegyék meg az összeget az elfogadón.
Az ügyvéd azt kérdezte tőle, hogy mit exequálnak Sámsoni
Lenczin: az attilája gombjait-e, vagy a tajték pipáját? mert annak
egyéb vagyona nincs.
Ennek Amerikában nagyon határozott nevet adnának: nálunk úgy
hijják, hogy gavalléros tréfa.
Azonban Inczének ezúttal nem volt kedve gavallért játszani. Jól
van; ha az elfogadó nem tud fizetni, álljon elő a kibocsátó: be kell
perelni Kondorossy Mukit!
Hallatlan vakmerőség pedig ez. Beperelni egy képviselőnek a
másikat váltóügyben! Hisz ez parlament elleni merénylet! Hisz ehhez
képest Guy Facques puskaporos összeesküvése még csak «kisebb
hatalmaskodás».
Az itélet elmarasztaló volt, hanem a végrehajtásnál kiderült, hogy
Kondorossy Muki e földi birtoka annyira be van már páczolva birói
foglalásokkal, hogy annak már semmi újabb törvényszéki
elmarasztalás meg nem árthat.
Áldorfai pedig már akkor dühbe volt hozva s nem akarta magát
kinevettetni. Lefoglaltatá a collegájának a képviselői napidijait.
Ez pedig már olyan merénylet, a mire nincs elég halál, hogy azt
kiengesztelje.
Ez árulással rokon cselekvény után menten tapasztalhatá Áldorfai
Incze, hogy a vendéglői közvélemény hangulata egyszerre hogy

Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.
More than just a book-buying platform, we strive to be a bridge
connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.
Join us on a journey of knowledge exploration, passion nurturing, and
personal growth every day!
ebookbell.com