Ebook Advances in Probability and Mathematical Statistics Clapem 2019 Merida Mexico 1St Edition Daniel Hernandez Hernandez Online PDF All Chapter

Advances in Probability and
Mathematical Statistics CLAPEM 2019

Mérida Mexico 1st Edition Daniel
Hernández■Hernández
Visit to download the full and correct content document:
https://ebookmeta.com/product/advances-in-probability-and-mathematical-statistics-cl
apem-2019-merida-mexico-1st-edition-daniel-hernandez%e2%80%90hernandez/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Starting Out The c3 Sicilian 1st Edition John Emms
https://ebookmeta.com/product/starting-out-the-c3-sicilian-1st-
edition-john-emms/
Advances in Probability and Mathematical Statistics

CLAPEM 2019 Mérida Mexico 1st Edition Daniel
Hernández■Hernández
https://ebookmeta.com/product/advances-in-probability-and-
mathematical-statistics-clapem-2019-merida-mexico-1st-edition-
daniel-hernandez%e2%80%90hernandez/
English Grammar Exercises with answers Part 4 Your

quest towards C2 1st Edition Daniel B. Smith
https://ebookmeta.com/product/english-grammar-exercises-with-
answers-part-4-your-quest-towards-c2-1st-edition-daniel-b-smith/
The American Revolution 1774 83 2nd Edition Daniel

Marston
https://ebookmeta.com/product/the-american-
revolution-1774-83-2nd-edition-daniel-marston/
The Paper Issue 83 1st Edition Origamiusa
https://ebookmeta.com/product/the-paper-issue-83-1st-edition-
origamiusa/
BNF 83 (British National Formulary) March 2022 Joint

Formulary Committee
https://ebookmeta.com/product/bnf-83-british-national-formulary-
march-2022-joint-formulary-committee/
Patriot vs Loyalist American Revolution 1775 83 Combat

1st Edition Si Sheppard
https://ebookmeta.com/product/patriot-vs-loyalist-american-
revolution-1775-83-combat-1st-edition-si-sheppard/
British Light Infantryman vs Patriot Rifleman American

Revolution 1775 83 1st Edition Robbie Macniven
https://ebookmeta.com/product/british-light-infantryman-vs-
patriot-rifleman-american-revolution-1775-83-1st-edition-robbie-
macniven/
Secrecy Public Relations and the British Nuclear Debate

How the UK Government Learned to Talk about the Bomb
1970 83 Cold War History 1st Edition Daniel Salisbury
https://ebookmeta.com/product/secrecy-public-relations-and-the-
british-nuclear-debate-how-the-uk-government-learned-to-talk-
about-the-bomb-1970-83-cold-war-history-1st-edition-daniel-
Progress in Probability
79
Daniel Hernández-Hernández
Florencia Leonardi
Ramsés H. Mena
Juan Carlos Pardo Millán
Editors
Advances in
Probability and
Mathematical
Statistics
CLAPEM 2019, Mérida, Mexico
Volume 79
Series Editors
Steffen Dereich, Universität Münster, Münster, Germany
Davar Khoshnevisan, The University of Utah, Salt Lake City, UT, USA
Andreas E. Kyprianou, University of Bath, Bath, UK
Sidney I. Resnick, Cornell University, Ithaca, NY, USA
Progress in Probability is designed for the publication of workshops, seminars

and conference proceedings on all aspects of probability theory and stochastic
processes, as well as their connections with and applications to other areas such as
mathematical statistics and statistical physics.
All manuscripts are peer-reviewed to meet the highest standards of scientific

literature. Interested authors may submit proposals by email to the series editors
or to the relevant Birkhäuser editor listed under “Contacts.”
More information about this series at http://www.springer.com/series/4839

Daniel Hernández-Hernández • Florencia Leonardi •
Ramsés H. Mena • Juan Carlos Pardo Millán
Editors
Advances in Probability
and Mathematical Statistics
CLAPEM 2019, Mérida, Mexico
Editors
Daniel Hernández-Hernández Florencia Leonardi
Department of Probability and Statistics Instituto de Matemática e Estatística
Centro de Investigación en Matemáticas Universidade de São Paulo
Guanajuato, Mexico São Paulo, Brazil
Ramsés H. Mena Juan Carlos Pardo Millán

Departamento de Probabilidad Department of Probability and Statistics
IIMAS-UNAM Centro de Investigación en Matemáticas
Mexico City, Mexico Guanajuato, Mexico
ISSN 1050-6977 ISSN 2297-0428 (electronic)

ISBN 978-3-030-85324-2 ISBN 978-3-030-85325-9 (eBook)
https://doi.org/10.1007/978-3-030-85325-9
Mathematics Subject Classification: 93E20, 90C40, 60J65, 60G51, 60F15, 60F10; Primary: 62C10,
60G57, 62F15, 99Z99; Secondary: 00A00, 62B10
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland
AG 2021
This work is subject to copyright. All rights are solely and exclusively licensed 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 book is published under the imprint Birkhäuser, www.birkhauser-science.com, by the registered
company Springer Nature Switzerland AG.
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
The present volume contains contributions of the XV Latin American Congress of

Probability and Mathematical Statistics (CLAPEM, by its acronym in Spanish), held
at Merida, Mexico during December 2–6, 2019.
Endorsed by the Bernoulli Society, this event is the official meeting of the
Sociedad Latinoamericana de Probabilidad y Estadística Matemática (SLAPEM)
and it is the major event in probability and statistics in the region. It gathers an
important number of researchers and students, predominantly from Latin America,
serving as an ideal forum to discuss and to disseminate recent advances in the field,
as well as to reveal the future of our profession.
Over nearly 40 years, the CLAPEMs have greatly contributed to the development
of probability and statistics by promoting collaborations in the region as well as
with the rest of the world. Previous editions were held in Caracas (1980, 1985,
2009), Montevideo (1988), Ciudad de México (1990), San Pablo (1993), Viña del
Mar (1995, 2012), Córdoba (1998), La Habana (2001), Punta del Este (2004), Lima
(2007), Cartagena (2014), and San José (2016).
On this occasion, the congress gathered scholars from over 20 countries and
included a wide set of topics on probability and statistics. The scientific program
included four plenary talks delivered by Gerard Ben Arous, Sourav Chatterjee,
Thomas Mountdford, and Judith Rousseau. The event also benefited from eight
semi-plenary talks given by Pablo Ferrari, Michele Guindani, Chris Holmes, Jean
Michel Marin, Lea Popovic, and Fernando Quintana. The program also included two
courses: “ Hierarchical Bayesian Modeling and Analysis for Spatial BIG Data ” by
Sudipto Banerjee and “ Sharpness of the phase transition in percolation ” by Vincent
Tassion, 10 thematic sessions, 21 contributed sessions, and several contributed talks
and poster presentations.
The volume begins with the chapter by Andrade, Calvillo, Manrique, and
Treviño where the authors present a probabilistic analysis of random interval graphs
associated with randomly generated instances of the data delivery on a line problem
(or DDLP). Angel and Spinka consider the infinite random geometric graph on
a circle of circumference L, which is a random graph whose vertex set is given
by a dense countable set in such circle, and find a dependency behavior on the
v
vi Preface
rationality of L. The asymptotic behavior of four binary classification methods,

when the dimension of the data increases and the sample sizes of the classes are
fixed, are studied by Bolivar-Cime. A class of transport distances based on the
Wassertein distances for random vectors of measures is considered by Catalano,
Lijoi, and Prünster. The latter leads to a new measure of dependence for completely
random vectors, and the quantification of the impact of hyperparameters in notable
models for exchangeable time-to-event data. Gil-Leyva studies the construction of
random discrete distributions, taking values in the infinite dimensional simplex, by
means of latent random subsets of the natural numbers, which are then applied
to construct Bayesian non-parametric priors. The connection between generalized
entropies based on a certain family of α-divergences and the class of some predictive
distributions is studied by Gutiérrez-Peña and Mendoza. A class of discrete-time
stochastic controlled systems composed by a large population of N interacting
individuals is considered by Higuera-Chan. The problem is studied by means of the
so-called mean field model. Kouarfate, Kouritzin, and Mackay provide an explicit
weak solution for the 3/2 stochastic volatility model which is used to develop a
simulation algorithm for option pricing purposes. Finally, León and Rouault revisit
Wschebor’s theorems on the a.s. convergence of small increments for processes with
scaling and stationarity properties and then apply such results to deduce that large
deviation principles are satisfied by occupation measures.
In summary, the high quality and variety of these chapters illustrate the rich
academic program at the XV CLAPEM. It is worth noting that all papers were
subject to a strict refereeing process with high international standards. We are very
grateful to the referees, many leading experts in their own fields, for their careful and
useful reports. Their comments were addressed by the authors, allowing to improve
the material in this volume.
We would also like to extend our gratitude to all the authors whose original
contributions appear published here as well as to all the speakers of the XV
CLAPEM for their stimulating talks and support. Their valuable contributions
encourage the interest and activity in the area of probability and statistics in Latin
America.
We hold in high regard the editors of the series Progress in Probability: Davar
Khoshnevisan, Andreas E. Kyprianou, and Sidney I. Resnick for giving us the
opportunity to publish the symposium volume in this prestigious series.
Special thanks go to the Universidad Autónoma de Yucatán and its staff for its
great hospitality and for providing excellent conference facilities. We are indebted to
Rosy Dávalos, whose outstanding organizational work permitted us to concentrate
mainly in the academic aspects of the conference.
The XV CLAPEM as well as the publication of this volume would not have been
possible without the generous support of our sponsors and the organizing institu-
tions: Bernoulli Society; Centro de Investigación en Matemáticas; Consejo Nacional
de Ciencia y Tecnología; Facultad de Ciencias–UNAM; Gobierno de Yucatán;
Google; Instituto de Investigaciones en Matemáticas Aplicadas y Sistemas–UNAM;
Instituto de Matemáticas–UNAM; Universidad Autónoma de Chapingo; Universi-
Preface vii
dad Autónoma Metropolitana; Universidad Autónoma de Yucatán; and Universidad

Juárez Autónoma de Tabasco.
Finally, we hope the reader of this volume enjoys learning about the various
topics treated as much as we did editing it.
Guanajuato, Mexico Daniel Hernández-Hernández

São Paulo, Brazil Florencia Leonardi
Mexico City, Mexico Ramsés H. Mena
Guanajuato, Mexico Juan Carlos Pardo Millán
Contents
Asymptotic Connectedness of Random Interval Graphs in a One

Dimensional Data Delivery Problem .. . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 1
Caleb Erubiel Andrade Sernas, Gilberto Calvillo Vives,
Paulo Cesar Manrique Mirón, and Erick Treviño Aguilar
Geometric Random Graphs on Circles . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 23
Omer Angel and Yinon Spinka
More About Asymptotic Properties of Some Binary Classification
Methods for High Dimensional Data .. . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 43
Addy Bolivar-Cime
Transport Distances on Random Vectors of Measures: Recent
Advances in Bayesian Nonparametrics .. . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 59
Marta Catalano, Antonio Lijoi, and Igor Prünster
Bayesian Non-parametric Priors Based on Random Sets . . . . . . . . . . . . . . . . . . . 71
María F. Gil-Leyva
Conjugate Predictive Distributions and Generalized Entropies . . . . . . . . . . . . 93
Eduardo Gutiérrez-Peña and Manuel Mendoza
Approximation and Mean Field Control of Systems of Large
Populations... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 103
Carmen G. Higuera-Chan
Explicit Solution Simulation Method for the 3/2 Model .. . . . . . . . . . . . . . . . . . . . 123
Iro René Kouarfate, Michael A. Kouritzin, and Anne MacKay
Large Deviations and Wschebor’s Theorems . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 147
José R. León and Alain Rouault
ix
Asymptotic Connectedness of Random
Interval Graphs in a One Dimensional
Data Delivery Problem
Caleb Erubiel Andrade Sernas, Gilberto Calvillo Vives,

Paulo Cesar Manrique Mirón, and Erick Treviño Aguilar
Abstract In this work we present a probabilistic analysis of random interval

graphs associated with randomly generated instances of the Data Delivery on
a Line Problem (DDLP) (Chalopin et al., Data delivery by energy-constrained
mobile agents on a line. In Automata, languages, and programming, pp. 423–434.
Springer, Berlin, 2014). Random Interval Graphs have been previously studied by
Scheinermann (Discrete Math 82:287–302, 1990). However, his model and ours
provide different ways to generate the graphs. Our model is defined by how the
agents in the DDLP may move, thus its importance goes beyond the intrinsic
interest of random graphs and has to do with the complexity of a combinatorial
optimization problem which has been proven to be NP-complete (Chalopin et
al., Data delivery by energy-constrained mobile agents on a line. In Automata,
languages, and programming, pp. 423–434. Springer, Berlin, 2014). We study the
relationship between solvability of a random instance of the DDLP with respect to
its associated interval graph connectedness. This relationship is important because
through probabilistic analysis we prove that despite the NP-completeness of DDLP,
there are classes of instances that can be solved polynomially.
Keywords Connectedness analysis · Data delivery problem · Mobile agents ·

Random interval graph
1 Introduction
The research presented in this work is in the intersection of several disciplines,

Probability Theory, Computer Science, Operations Research, and Graph Theory.
So, in this introduction we define the problem, and provide the basic concepts of
computational complexity and random graphs that are needed to make the work
C. E. Andrade Sernas · G. Calvillo Vives · P. C. Manrique Mirón · E. Treviño Aguilar ()

Unidad Cuernavaca del Instituto de Matemáticas, Universidad Nacional Autónoma de México,
Mexico City, Mexico
e-mail: erick.trevino@im.unam.mx
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 1

D. Hernández-Hernández et al. (eds.), Advances in Probability and Mathematical
Statistics, Progress in Probability 79, https://doi.org/10.1007/978-3-030-85325-9_1
2 C. E. Andrade Sernas et al.
self-contained. We also include several references for those that may want to go
deeper in the study of these subjects.
1.1 The Data Delivery on a Line Problem
The production of inexpensive, simple-built, mobile robots has become a reality

nowadays, to the point that a swarm of mobile agents can be used for different
tasks. A practical application would be, for instance, to use a swarm of drones to
explore a cave and produce a map of it by collecting and sharing geospatial data
whenever two drones meet. Or the use of a network of drones to deliver packages
to customers from retail stores or courier services. The Data Delivery Problem is
a mathematical abstraction of such scenarios and has been studied a lot lately, see
e.g., [1, 3, 4, 8, 9, 12]. Here we deal with a specific Data Delivery Linear Problem
(DDLP), namely where agents (robots) are constrained to move in a line. In this
version, a set of n energy-constrained mobile agents are placed on positions 0 ≤
xi ≤ 1, i = 1, . . . , n on the unit interval and have ranges ρi > 0 (denoting the
maximum length of a walk for each agent), the question is whether there is an order
in which the mobile agents should be deployed to pick up the data at a point s called
the source, and collectively move it to a predetermined destination called the target
t > s. The first agent, in the order found, moves to the source, picks up the data
and move it to the right according to its capacity. The second agent moves to the
point where the first agent is, takes the data and move further to the right where
a third agent comes to take over and so on until the data arrives, if possible, at its
destination t. Observe that there are two cases for the movement of an agent. If the
agent is to the left of the position d where the data is, it moves always to the right.
First to pick up the data and then to move it further to the right. If it is to the right of
d, then it has to move first to the left to reach d and then to the right as far as it can.
In both cases, the agent covers the range [xi − aρi , xi + (1 − 2a)ρi ] where a = 0
for the first case and a = (xi − d)/ρi for the second. This observation is key to
define graphs associated to the problem. Observe also that the DDLP is a decision
problem; that is to say, the answer is yes or no. In this framework now we can talk
about the computational complexity of the problem.
This problem was introduced by Chalopin et al. [9], and it was shown to be
NP-complete, although for instances where all input values are integers they gave
a quasi-pseudo-polynomial time algorithm. In Sect. 2 we show how this problem is
equivalent to a graph theoretical problem which in turn is analyzed using random
graphs in Sect. 3. So let us first briefly recall some computational complexity and
graph theoretical concepts.
Asymptotic Connectedness of Random Interval Graphs 3
1.2 Computational Complexity
This field has several roots: The seminal ideas of Jack Edmonds [13] about good
algorithms and good characterizations; the foundational work of Stephen Cook [11]
and Dick Karp [17], to name a few. The field is now represented by one of the
millennium problems: Is P=NP?. A systematic treatment of the subject can be found
in [15], [2].
A decision problem is a collection of instances (propositions) each of which is
true or false. The DDLP is one such decision problems. Each instance is of the
form: The set of agents Q = {(xi , ρi ), i = 1, . . . , n} can move the data from s to t.
This proposition is true or false. The problem is to decide for each instance which
is the correct answer. A decision problem is said to belong to the class P if there
is an algorithm (Turing Machine) that decides correctly which is the answer and
runs in polynomial time, which means that the number of steps the corresponding
Turing Machine has to perform is bounded by a polynomial in the size of the
instance (usually measured in bits). For the DDLP, the size of an instance is the
number of bits required to store Q, s and t. A famous Decision Problem in the
class P is to decide if a given set of linear inequalities has a solution. A decision
problem belongs to the class NP if for every instance with answer YES, there is a
polynomial algorithm to verify that the answer is correct. The DDLP is NP since
for an affirmative answer it suffices to show the sequence in which the agents are
used and check that effectively they move the data from s to t. This, of course can
be done very efficiently. A decision problem S belongs to the class NP-complete
if it is NP and any other NP problem can be reduced polynomially to S. Cook
provided the first NP-complete problem (satisfiability) and then Karp [17] added
a bunch of combinatorial problems to that class. Chalopin showed that DDLP is a
NP-complete problem. While question is P=NP? is open, we do not know if the
NP complete problems can be solved by polynomial time algorithms. At present,
the generalized belief is that P = NP . If this is so, NP problems will remain as
difficult problems, including the DDLP. One of the drawbacks of this theory is that
it does not recognize clearly that a large amount of an NP-complete problem can be
solved efficiently. The recognition of that fact has motivated the use of probability
to asses fringes of “easy” instances and thus isolate the really hard instances of a
problem. A good example of this methodology is in [6]. Our work follows that path.
In Sect. 5 we interpret the results presented in Sect. 3 in this sense.
1.3 Graphs
For the purpose of this work it is sufficient to deal with simple graphs and so
we omit the adjective. A graph consists of a set V whose elements are called
vertices and a collection of subsets of V of cardinality 2 which we denote by
E and call them edges. If a vertex v belongs to an edge e we say that v is an
extreme of e. Clearly every edge has two extremes. If u and v are the two extremes
of an edge we say that they are adjacent. A simple path (or path) is a sequence
v0 , e1 , v1 , e2 , v2 , . . . , vn+1 , en , vn in which all vertices vi , i = 1, . . . , n are different
and vi and vi+1 are the extremes of edge ei . Such path is said to connect vertices v0
and vn ; when v0 = vk , the graph is a cycle, denoted by Ck . A graph is connected if
for every pair of vertices u and v of the graph, there exists a path connecting them.
Given a set V of real closed intervals in the real line we can construct a graph in the
following way: The set of vertices is V and two vertices are adjacent (form an edge)
if they intersect as intervals. A graph constructed in this way is called an interval
graph. The graphs that can be constructed in this way are a small part of all possible
graphs. For example, any cycle Ck with k > 3 is not an interval graph. However
they are nice graphs to work with because they have remarkable properties. There
are several ways in which a collection of intervals can be defined. We are interested
in two of them.
Definition 1 The symmetric model of a interval graph is generated by intervals Ii
defined by its center xi and its radius ρi in the form Ii = [xi − ρi , xi + ρi ]. The
interval graph associated to the symmetric model will be denoted by GS (Q) where
Q = {(xi , ρi ), i = 1, . . . n}.
The asymmetric model is obtained from instances of DDLP considering Ii =
[xi − ai ρi , xi + (1 − 2ai )ρi ] defined in Sect. 1.1. It will be denoted by GA (Q, a),
where again Q = {(xi , ρi ), i = 1, . . . n} and a is a vector in the unit box of Rn .
A standard reference in graph theory is [7].
1.4 Random Interval Graphs
Random graphs were firstly introduced by Gilbert [16], but the work of Erdős and
Rényi [14] set the basis for the study of their evolution (a comprehensive study
on random graphs can be found in [6]). Later on, Cohen studied the asymptotic
probability that a random graph is a unit interval graph [10], and Scheinermann
replicated the ideas of Erdős and Rényi to study the evolution of random interval
graphs [18].
To continue our discussion, we introduce a probabilistic framework, so let
(Ω, F , P) be a probability space with F a σ -algebra on the set of scenarios Ω
and P a probability measure on F . All random variables and our asymmetric model
will be defined there.
Scheinerman [18] obtained random interval graphs, using the symmetric model,
by defining two random variables for each interval, the centers {xi }ni=1 and the
radii of the intervals {ρi }ni=1 , the centers with uniform distribution in [0,1] and the
radii also uniform in the interval [0, r] with r < 1, so intervals are constructed as
[xi − ρi , xi + ρi ]. We use the asymmetric model fixing the parameters {ai }ni=1 , ai ∈
[0, 1] and considering 2n independent random variables, the locations {xi }ni=1 (all
identically distributed uniformly in [0, 1]), and the ranges {ρi }ni=1 (all identically
distributed uniformly in [0, r]). The intervals are [xi − ai ρi , xi + (1 − 2ai ) ρi ],

which are substantially different to the intervals of the symmetric model. First they
are asymmetric, second they are shorter as specified by the parameter ai which
represents the percentage of a mobile agent’s energy used to move backwards to
pick up the load before going forward with it, thus it determines the behavior of
the agent. The asymptotic analysis presented in this work is analogous to the one of
Scheinermann’s, but here we consider an asymmetric model which is more general.
Moreover, it is related to a NP-complete combinatorial problem which makes the
new model interesting from the point of view of computational complexity, a fact
that we show in this work.
2 Graph Theoretical Formulation of the Data Delivery on a

Line Problem
In this section we show how the DDLP is equivalent to an existence problem

for interval graphs. The transformation from one another is completely general
and does not assume that the data have been randomly generated. The equivalent
graph theoretical decision problem is the following: Given a finite collection
Q = {(xi , ρi ), i = 1, . . . n} of points in R2 decide if there exists a vector
a = (a1 , . . . , an ) such that the interval graph GA (Q, a) defined by the asymmetric
model is connected. This problem will be called the Existential Connectedness
Problem (ECP).
We say that two decision problems are equivalent if for every instance of one
there is an instance of the other such that both have an affirmative answer or both
have a negative answer.
Theorem 1 DDLP and ECP are equivalent.
Proof First we will show that for every instance (Q, s, t) of DDLP there exists
an instance of ECP whose solution conforms with the one of the DDLP instance.
First, discard all points of Q such that either xi + ρi < s or xi − ρi > t then add
two new points (s, 0) and (t, 0). Call this new set of points Q which defines an
instance of the ECP. A positive answer to this instance comes with a vector a ∗ such
that GA (Q , a ∗ ) is a connected graph. So there exists a path T in GA (Q , a ∗ ) from
the vertex defined by (s, 0) to the vertex defined by (t, 0). The interior vertices of
T correspond to points of Q and so to agents in the DDLP instance. The order in
which the vertices of T are traversed from s to t define the order in which the agents
have to be deployed. The adjacency of consecutive vertices of T guarantee that the
corresponding agents can get in touch to pass the data from one to the next. So the
existence of an affirmative solution of ECP translates into a solution of DDLP.
It remains to be proven that if the solution to the ECP instance is negative so is
the answer to the DDLP. The contrapositive of this is that any YES answer to an
instance of DDLP translates into an affirmative answer of ECP. To prove this let
(Q, s, t) be an instance of DDLP with a true answer and a sequence ri1 , . . . , rik of
agents that carry the data from s to t. This behaviour of the agents in the sequence
define a set of parameters ai1 , . . . , aik . For the rest of the agents, which do not play
any role in the transportation of the data, let ai = 0 if xi < t and ai = 1 if xi ≥ t. We
claim that the graph GA (Q, a) is connected. The parameters ai1 , . . . , aik define a set
of intervals, a path T in GA (Q, a), that cover the segment [s, t]. We will show that
every other interval intersects T . For an agent ri such that xi < t the corresponding
interval is [xi , xi + ρi ], if xi ≥ s, then xi is in [s, t] and so it intersects some interval
of T ; if xi < s, then since irrelevant agents have been removed xi + ρi ≥ s and
therefore the interval intersects T too. The case xi ≥ t is resolved similarly.
The next result is a direct consequence to Theorem 1. We omit the details of the
proof.
Corollary 1 ECP is NP-complete
Theorem 1 shows that DDLP, which is NP-complete, is reducible to ECP.
Moreover the reduction is polynomial since to transform a DDLP given by (Q, s, t)
to a ECP given by Q it is only needed to compute which robots (xi , ρi ) satisfy
xi − ρi < s or xi + ρi > t; and to include the points (s, 0), (t, 0). This can be done
in linear time assuming arithmetic operations are performed in constant time. The
equivalence between DDLP and ECP allow us to deal with the graph theoretical
formulation and use the results and ideas of Random Graphs in order to obtain
some asymptotic results in the ECP and therefore the DDLP problems. Specifically,
parameters’ domain can be partitioned into regions, one in which the problems can
efficiently be solved. The relevance of the connectedness of an interval graph in
relation to the solvability of its associated DDLP instance cannot be overstated.
That is, if for any DDLP
instance (Q, s, t) there exists a vector a in the unit box
such that GA Q , a is connected, then the DDLP instance is solvable, the converse
is certainly not true, as shown in the counterexamples of Fig. 1. On the other hand,
given a DDLP instance (Q, s, t), if for all possible vectors a in the unit box none
of the associated interval graphs GA Q , a is connected, it is then assured that the
instance is not solvable. This last statement is summarized in the following corollary.
Corollary2 Given
a DDLP instance (Q, s, t) if its associated symmetric interval
graph GS Q is disconnected, the instance is not solvable.
Proof For every vector a in the unit box we have that
[xi − ayi , xi + (1 − 2a) yi ] ⊂ [xi − yi , xi + yi ] .

Now, recall that Q = Q ∪ {(s, 0) , (t, 0)}, so if GS Q is disconnected, it means
that for every vector a in the unit box GA Q , a is disconnected as well. This
means that ECP is not solvable, so by Theorem 1 DDLP is not solvable.

Fig. 1 For this example we take Q = 1 3 1 1
4, 4 , 2, 2 . In panel (a) we have a = 13 , 13 , in

(b) a = (0, 0), and in (c) a = 13 , 0 . In the first two cases GA Q , a is disconnected, but in the
third case it is connected and therefore solvable
3 Probabilistic Analysis of GA (Q, a)
We denote as usual the degree of a vertex vi by d(i). We write vi ∼ vj when there

is no edge joining the vertices vi and vj . We denote by vi+ = xi + (1 − 2a)ρi (resp.
vi− = xi − aρi ) the right (resp. left) boundary of vi . In this section we analyze
the connectedness of random interval graphs of the form GA (Q, a). The approach
that we follow is to fix a and analyze the asymptotic behavior of the connected
components of GA (Q, a). In order to do so, we first introduce random variables
that count the connected components of a random graph GA (Q, a). In this section
and the sequel we will specialize to the case in which a is a scalar, that is ai = a
for all i. Furthermore, we assume a ∈ [0, 1/2]. Estimations with this specification
will already yield lower bounds for the probability of solvability due to Theorem 1
which is our main goal.
For each i, i = 1, . . . , n let

1 if the right end point of vi is contained in no other interval vj , j = i,
Xi :=
0 otherwise.
The family of random variables X1 , . . . , Xn indicates where a connected component

ends. For example, assume that X1 = X2 = 1 and Xi = 0 for i > 2. Assume
without loss of generality that the interval v1 is on the left of the right end point
of v2 . Since X1 = 1 the right end point of v1 is not included in any other interval,
implying that all the intervals on the left of v1+ are disjoint from all other intervals
on its right. Thus, there are at least two components. In order to see that there are
exactly two components, assume that there is one more. For this component denote
by vj the interval such that vj+ attains the supremum over all right ends of intervals
in the components. Then, necessarily Xj = 1 with j > 2. This is a contradiction.

Then, there exists exactly two components.
We also define
n
X(n) := Xi
i=1
is the random variable that counts the number of connected components of

GA (Q, a). Note that P (X(n) = 0) = 0. The distribution of X(n) depends on
the parameter r. However, we do not write this explicitly in order not to overload
notation. In our first asymptotic result, we let the numbers of vertices n goes to
infinite while at the same time shrinking the range of the intervals (the parameter
r). The theorem gives the right trade-off for those processes in order to have
connectedness with high probability.
Theorem 2 Let β := 2
1−a and r(n) := β n1 (log(n) + c). Then,
lim inf P (X(n) = 1) ≥ 1 − e−c . (1)

n→∞
Proof Denote by vi+ = xi + (1 − 2a)ρi the right point of the interval vi . Let
B(i) := {ω | vi+ (ω) ∈ ∪k=i vk }
be the event in which the right boundary of the interval vi is included in at least one
of the other intervals. The event {X(n) = 1} satisfies
n
{X(n) = 1} = B(i).
k=1 i=k
Hence
⎛ ⎞

⎜ n n
⎟ n
P (X(n) = 1) = P ⎜
⎝ B(i)⎟
⎠≥P B(i) .
k=1 i=1 i=2
i=k
Moreover
⎛ ⎞ ⎛⎧ ⎫c ⎞
n ⎨ n ⎬ n

1− P⎝ B(j )⎠ = P ⎝ B(j ) ⎠ ≤ P B c (j ) = (n − 1)P B c (2) .
⎩ ⎭
j =2 j =2 j =2
Hence:

P (X(n) = 1) ≥ 1 − (n − 1)P B c (2) . (2)
So now we estimate P (B c (2)). The right end point vi+ is a random variable and we
denote by μ it probability distribution function under P . Note that the distribution
does not depend on i. Then,

1+(1−2a)r
P B c (2) = (1 − P (t ∈ v2 )))n−1 μ(dt),
0
due to the independency of the intervals. For a < 12 the random variable vi+ is the
sum of two independent uniform distributions and then, we can easily see that μ is
a distribution concentrated on [0, 1 + r(1 − 2a)] with density
⎧
⎪
⎪
x
[0, r(1 − 2a)]
⎪ r(1−2a)
⎪
⎨1 [r(1 − 2a), 1]
f (x) = −x+1+r(1−2a)
⎪
⎪ [1, 1 + r(1 − 2a)]
⎪
⎪ r(1−2a)
⎩
0 otherwise.
For t ∈ [0, 1 + r(1 − 2a)] let h(t, x, ρ) be the indicator function of {x − aρ ≤

t} ∩ {x + (1 − 2a)ρ ≥ t} and g(t, ρ) := (t + aρ) ∧ 1 − (t − (1 − 2a)ρ)+ ∧ 1 .
We have
r 1
1
P (t ∈ v2 ) = h(t, x, ρ)μ(dx)dρ
r 0 0
r
1
= g(t, ρ)dρ.
r 0
For t ∈ [0, (1 − 2a)r] the function g simplifies to g(t, ρ) = t + aρ − (t − (1 −

2a)ρ)1{t >(1−2a)ρ}. Hence
t
1 1 1 1−2a
P (t ∈ v2 ) = t + ar − tρ − (1 − 2a)ρ 2
2 r 2 0
1 1
= t + ar − t 2.
2 2r(1 − 2a)
Note that P (t ∈ v2 ) ≥ 12 ar. Then

(1−2a)r
1
(1 − P (t ∈ v2 )))n−1 tdt
(1 − 2a)r 0
(1−2a)r !n−1
1 1
≤ 1 − ar tdt
(1 − 2a)r 0 2
!n−1
1
= 1 − ar (1 − 2a) r.
2
For x ∈ R we will make use of the following basic inequality:

x n
1− ≤ e−x .
n
Then
!n−1
1
(n − 1) 1 − ar(n) (1 − 2a) r(n)
2
(n − 1)a
≤n exp{− r(n)}r(n)
2
! aβ n−1 " #
1 2 n aβ n − 1
= exp − c β(log(n) + c).
n 2 n
$ (1−2a)r(n)
Thus, the contribution of (n − 1) 0 (1 − P (t ∈ v2 )))n−1 μ(dt) to (n −
c
1)P (B (2)) goes to zero asymptotically. In a similar fashion we can see that
1+(1−2a)r(n)
lim (n − 1) (1 − P (t ∈ v2 )))n−1 μ(dt) = 0.
n→∞ 1−ar(n)
For t ∈ [(1 − 2a)r, 1 − ar] we have
g(t, ρ) = (t + aρ) ∧ 1 − (t − (1 − 2a)ρ)+ ∧ 1

= t + aρ − t + (1 − 2a)ρ
= (1 − a)ρ.
Hence, for t ∈ [(1 − 2a)r, 1 − ar]

r
1
P (t ∈ v2 ) = (1 − a)ρdρ
r 0
1
= (1 − a)r.
2
Moreover
1−ar 1
(n − 1) (1 − P (t ∈ v2 )))n−1 μ(dt) =(n − 1)(1 − (1 − a)r)n−1 (1 − (1 − a)r)
(1−2a)r 2
" #
1
≤(n − 1) exp − (1 − a)nr(n)
2
≤e−c .
We have proved the theorem.

Remark 1 Theorem 2 tells us that the probability that the interval graph is connected
is bounded from below as n → ∞ and r(n) converges to zero but not too fast. The
intuition behind this threshold comes from the order statistics of uniform random
variables. If we generate n random points with uniform distribution in the interval
U(k)
[0, 1], and call U(k) the k-th order statistic, then k/n → 1 in probability as k → ∞;
see e.g., [5, Thm. 1.4]. That is to say, the random points tend to be homogeneously
spattered as n grows, such that the spacing between consecutive points tends to
be 1/n. It might be reasonable to think that if r decreases as ≈ n1 the symmetric
intervals [xi − ρi , xi + ρi ] would yield a connected component. Still, we need a
little adjustment to account for the random lengths of each interval. In fact logn n is
the threshold to ensure connectivity in the symmetric model [18]. Our model is not
symmetric but rather the intervals are of the form [xi − aρi , xi + (1 − 2a) ρi ], and
in Theorem 2 we showed that the connectivity threshold must be adjusted to account
for this asymmetry.
The random variable X(n) counts the number of connected components, so
X(n) = 1 means that the interval graph is connected. If the interval graph
associated to a DDLP instance is a single connected component, with the additional
requirements that there exists intervals intersecting the boundaries, the instance is
solvable due to Theorem 1. In the next theorem we show that the parametrization
of r as function of n in Theorem 2 is optimal. For
% the statement of the theorem we
introduce some random variables. Let Y (n) := ni=1 Yi where
⎧ & '
⎨1 if d (i) = 0 and x ∈ 1 , 2
i 3 3
Yi :=
⎩0 otherwise.
Observe that Y (n) counts the number of intervals vi = [xi − ai ρi , xi + (1 − 2ai ) ρi ]

for which xi ∈ [1/3, 2/3] with isolated associated vertices in the interval graph.
1 nr(n)
Theorem 3 Let β be as in Theorem 2. Let α(n) := β log n . If α := limn→∞ α(n) <
1 then
1
lim E[Y (n)] = ∞ (3)
n→∞ log(n)
and
!
1
lim P Y (n) ≥ E[Y (n)] = 1. (4)
n→∞ log(n)
In particular, with high probability all graph instances are not connected.
Proof We start with the idea of the proof for (4). Let σ 2 (n) := variance(Y (n)). For
2 2
> 0, Chebyshev’s inequality P (|Y (n) − E[Y (n)]| ≥ ) ≤ σ (n) 2 yields 1− σ (n)
2 ≤
P (− + E[Y (n)] ≤ Y (n)). In particular, for (n) := (1 − log(n)1
)E[Y (n)]
!
σ 2 (n) 1 E[Y 2 (n)]
= (1 − )−2 − 1 ,
2 (n) log(n) E2 [Y (n)]
and we will show
E[Y 2 (n)]
lim = 1.
n→∞ E2 [Y (n)]
This establishes (4) since

1
{−(n) + E[Y (n)] ≤ Y (n)} = {Y (n) ≥ E[Y (n)]}.
log(n)
Next we compute E(Yi ). We have
E(Yi ) = P (d(i) = 0 and xi ∈ [1/3, 2/3])

= P (vi ∼ vk for all k = i and xi ∈ [1/3, 2/3])

= 1{xi ∈[1/3,2/3]} (P (vi ∼ v))n−1 dμ(vi ),
where μ is the distribution associated to the random vertex vi . Note that
P (vi ∼ v | ρ = t) = 1 − (1 − a)ρi − (1 − a)t.
Now we have

1{xi ∈[1/3,2/3]} (P (vi ∼ v))n−1 dμ(vi )
r n−1
= 1{xi ∈[1/3,2/3]} 1 − (1 − a)ρi − (1 − a) dμ(vi )
2
2/3 r &
1 r 'n−1
= dxi 1 − (1 − a)ρi − (1 − a) dρi
1/3 r 0 2
& 'n & 'n
3(1−a)
1 1− 2 r − 1− 2 r
1−a
= · . (5)
3 n(1 − a)r
From (5), we determine E[Y (n)] by

& 'n & 'n
1 β 1 − β r(n) − 1 − 3 β r(n)
1 1
E(Y (n)) =
32 r
& 'n & 'n
1 nr(n) log(n) 1 nr(n) log(n)
1 n 1 − β log(n) n − 1 − 3 β log(n) n
= β nr(n)
6 log(n)
log(n)
" #
1 1 n log(n) n log(n) n
= 1 − α(n) − 1 − 3α(n)
6 α(n) log(n) n n
n1−α − n1−3α
≈ . (6)
6α log(n)
Let us estimate E(Y 2 (n)). To this end, note first that E(Y 2 (n)) = E(Y (n)) +
n(n − 1)E(Y1 Y2 ) and E(Y1 Y2 ) = P (Y1 Y2 = 1, v1 ∼ v2 ). For this last probability,
we have
P (Y1 Y2 = 1, v1 ∼ v2 ) =

1{v1 ∼v2 } 1{x1 ,x2 ∈[1/3,2/3]}P (v1 , v2 ∼ vk for all k = 1, 2) dμ(v1 , v2 ). (7)
( )n−2
Note for k0 fixed, P (v1 , v2 ∼ vk for all k = 1, 2) = P v1 , v2 ∼ vk0 . Condi-
tioning
* *
P {v1 , v2 ∼ v} ∩ *v1± − v2∓ * ≥ (1 − a)ρ | ρ = 1 − (1 − a)ρ1 − (1 − a)ρ2 − 2(1 − a)ρ
(8)
* *
P {v1 , v2 ∼ v} ∩ *v1± − v2∓ * < (1 − a)ρ | ρ = 1 − (1 − a)ρ1 − (1 − a)ρ2 − (1 − a)ρ.
(9)
* *
Note that *v1± − v2∓ * < (1−a)ρ happens with probability O(r) as r → 0. Wrapping
all together
* *
v2 ) = (1 − O(r))P Y1 Y2 = 1, v1 ∼ v2 , *v1± − v2∓ * ≥ (1 − a)ρ
P (Y1 Y2 = 1, v1 ∼

( * * )n−2
= (1 − O(r)) 1{x1 ,x2 ∈[1/3,2/3]} P v1 , v2 ∼ v, *v1± − v2∓ * ≥ (1 − a)ρ dμ(v1 , v2 )
& r 'n−2
= (1 − O(r)) 1{x1 ,x2 ∈[1/3,2/3]} 1 − (1 − a)ρ1 − (1 − a)ρ2 − 2(1 − a) dμ(v1 , v2 )
2

= (1 − O(r)) 1{x1 ,x2 ∈[1/3,2/3]} dμ(x1 , x2 ) × (10)
&
1 r r r 'n−2
1 − (1 − a)ρ1 − (1 − a)ρ2 − 2(1 − a) dρ1 dρ2
r2 0 0 2
1 [1 − (1 − a)r]n − 2 [1 − 2(1 − a)r]n + [1 − 3(1 − a)r]n
= (1 − O(r)) · . (11)
9 n(n − 1)r 2 (1 − a)2
Since α < 1 and αβ = limn→∞ nr

log n , from (5) and (10), we have
1 [1 − (1 − a)r]n − 2 [1 − 2(1 − a)r]n + [1 − 3(1 − a)r]n

n(n − 1)E(Y1 Y2 ) = (1 − O(r)) ·
9 r 2 (1 − a)2
n2 & '
≈ 2
n−2α − 2n−4α + n−6α . (12)
36α 2 log (n)
From (6) and (12), we get
E(Y 2 (n))
→ 1,
(E(Y (n)))2
when n → ∞.
Now returning to our original motivation of a DDLP, Theorem 3 does not yet
fully address the issue to determine if a connected interval graph is synonym
of solvability of the DDLP. If the interval graph associated to a DDLP instance
is a single connected component, but none of its intervals actually intersect the
boundaries, we cannot say that the instance is solvable. However, the next result
shows that the boundaries will be reached with high probability.
Theorem 4 Let r(n) be as in Theorem 2. Let G(n) be the event that the associated
interval graph is connected. Define the random variables
R(n) = max {xi } , L(n) = min {xi } (13)

i=1...,n i=1...,n
R (n) = max {xi + (1 − 2a) ρi } , L (n) = min {xi − aρi } . (14)

i=1...,n i=1...,n
Then R(n) − L(n) has Beta distribution with parameters (n − 1, 2). Furthermore,
for ∈ (0, 1)

lim P R (n) − L (n) ≤ | G(n) = 0. (15)
n→∞
Proof To simplify notation we write R, L and R , L . Recalling {xi }i=1...,n are

independent and identically distributed in [0, 1], it is straightforward to check that
P (L ≤ t) = 1 − (1 − t)n so that its density function is fL (t) = n (1 − t)n−1 .
Analogously, P (R ≤ t) = t n and its density function is fR (t) = nt n−1 . Now,
because L and R are not independent we cannot compute the density of R − L as
the convolution of R and −L, so we proceed as follows.
For d ∈ (0, 1) we want to compute P (R − L ≤ d). To this end, let do some
conditioning on L. Let t ∈ [0, 1] denote the actual minimum of {xi }i=1...,n . We
must distinguish two cases that are mutually exclusive. The first case is t < 1 − d,
so given that L = t, the probability that the remaining n − 1 points fall within the
n−1
interval [t, t + d] is given by 1−td
. The second case is t ≥ 1−d, so all n points
fall in the interval [t, 1] thus R − L < d. Therefore, by the law of total probability
1−d !n−1
d
P (R − L ≤ d) = fL (t) dt + P (L ≥ 1 − d)
0 1−t
1−d !n−1
d
= n (1 − t) n−1
dt + d n
0 1−t
= n (1 − d) d n−1 + d n .
As a consequence we see that the density function of R − L is given by fR−L (t) =

n (n − 1) t n−2 (1 − t). Recall that the density of a Beta distribution is given by
f (t) = β(r,s)
1
t r−1 (1 − t)s−1 , where β (r, s) = ΓΓ(r)Γ (s)
(r+s) . Notice that β (n − 1, 2) =
1
n(n−1) ,so fR−L (t) = β(n−1,2)
1
t n−2 (1 − t) is a Beta density function, as claimed.
Now we prove (15). Take t ∈ (0, 1). Clearly L ≤ L and
R ≤ R so that

R − L ≤ R − L , then it follows P (R − L ≥ t) ≤ P R − L ≥ t . Also,
aρi + (1 − 2a) ρi ≤ ar + (1 − 2a) r so R − L ≤ R − L + (1 − a) r, thus
P R − L ≥ t ≤ P (R − L + (1 − a) r ≥ t). Hence

P (R − L ≥ t) ≤ P R − L ≥ t ≤ P (R − L + (1 − a) r ≥ t) . (16)
The law of total probability yields

P R − L ≤ t = P R − L ≤ t | G P (G) + P R − L ≤ t | Gc P Gc

which
implies P R − L ≤ t ≤ P R − L ≤ t | G P (G) + P (Gc ) and
P R − L ≤ t ≥ P R − L ≤ t | G P (G). We then obtain the inequality

P R − L ≤ t − P (Gc )
P R − L ≤ t
≤P R −L ≤t |G ≤ (17)
P (G) P (G)
Now we have for ∈ (0, 1), sufficiently large n and for M and c as in Theorem 2
the following

P {R (n) − L (n) ≤ }
P {R (n) − L (n) ≤ } | G(n) ≤
P (G(n))

P {R (n) − L (n) ≤ }
≤
1 − Me−c
P ({R(n) − L(n) ≤ })
≤ ,
1 − Me−c
where the first inequality follows from (17), the second from Theorem 2 and the last
from R − L ≤ R − L . Now the result follows from the Beta(n − 1, 2) distribution
of R(n) − L(n).
4 Simulations
In this section we present the results of simulation experiments illustrating the main
findings of Theorems 2, 3, and 4. In Fig. 2, panel (a), we see simulations of random
interval graphs in which r follows the regime of Theorem 2 in the form r(n) =
β log(n)+c
n for a = 0.2, β = 1−a2
and c = 3.1. The number of nodes (n) changes as
indicated in each graph in the panel. What we see is that all graphs are connected
in agreement with the main claim in Theorem 2 of having connected instances with
high probability. In Fig. 2, panel (b), we see simulations of random interval graphs
in which r(·) follows the dynamic r(n) = αβ log(n) n with α = 0.5 and β as before.
What we see is that all of the simulated graphs are disconnected and with many
isolated nodes, which is to be expected by Theorem 3.
In Fig. 3 we see simulations illustrating Theorem 4. On the left side Fig. 3a
we see the histogram after 2e5 simulations of the difference R(100) − L(100),
where the random variables R(n) and L(n) are defined in Theorem 4 and the
parameters are as in the simulation exercise for Theorem 2, the only difference
is that here the number of nodes is fixed. It presents visual evidence of the
Nodes=50, r=0.351 Conn Nodes=50, r=0.098 Dis

10 4
7 13
6 12 3
9 6 9
5
3 11 2 8
5 18 7
15 16 1
1
2 4
8 14 17
10
19
23
20 45 49
21 47
48 11
22
12
24 15 13
14
16
43
18 17
25 40
39 37
41
19
38 36 33 20 21
26
27 42 30
35
32 22
31
34 46
23
28
29
30
32 37
33 41
44
31 40 42
34 49
35 38 46
36 25
39 50 29 50
44
45 27
24
48
43 26
47 28
Nodes=2000, r=0.013 Conn Nodes=2000, r=0.005 Dis

ll
ll
l
l
ll ll
lll
ll
l
ll
l
l lllll l
lll
l llll ll
l l l
1
l
57 54 53
l
612613 179
l l l
175174
l ll ll
3 561112
l l
178 60
l l
2
l
7910
1317
l l l ll l l
59 58 56 55 52
l
177176
l
1996
1990
l
2000 814
l
1998
1993 1887
l l ll l
1988
ll l
19991992 1886
ll
415
l
1885 180
l
1989
l l l
1991
l
1997 1985
l l
1994 161824
lll l l
1984
l l
19951986 1883 188418881889 217220
l
1987 19
l
2021
l
1981 173
l
1982
l l l
1983 23
l
252227 37
ll
34 4044 1882 1892 50 51 218222
l
1979
l
28
l
2926
l
1978 31
ll
1890 1891
l l l
172171 219
l l l
39
l l
1858 1881
l l
30
l l
1977 33
l
1980 1880
l l
1859
ll l l l
32
l l
3 6
l
221
l ll
223
l
1976
l
1854
ll
ll ll l ll
35 4345 1855
l
38 224
l l ll
4142
l
1894 170
l
46 1860
l l l
1895
l
4544
l
1974
l l
48 1856 1861 1879 1893
l l l l
1975 1853
l ll
1857 4849 169
l
47
l
43
l
1973
l lll
1896
l
l l
225 228
l ll l
1877
l l l
1972 50 495253 4746
l
1969 1852 1862
l
l
1897 4041
l
l
1851
l l l l
18761878
l l
1965
lll ll l
1971
1970 168167
ll l
1835 1841 1968 1898
l
51 57
l
37 38 42
l
163
l
l
1838 1850 1863 1874
l
226227
l
1899 229
l l
1966
l
1847
l
1848 1900 164
l
1964
llll
1967 26
l
11839
l
1864 230
l
1846
l
834 1844 1875
l l
32
l
1828 1833
l
1870
l
61 29 34 166
l l
1831
l l
1830 1963 1842 39
ll l
161
l
1822 1852 5554 18451849 1902 31
33 35 36 165
l
231 233
l l
1837 1844 1868 30
l
1826 1853 1962 1872 25 232
l
18291832 1842 1866
l
56 27 162
l l l
1823 1827 1836
1840 1849 18
ll l
1845 236
l l
18391840
18411843
l ll
1824 6069 1847
l
1871 1901
l
160
lll
234
l
1843 59 1867
l
1825 19041903 19 237
l
1865 17 3428 238
l l l l
1820 1851 1858 1961
1958 6263 136 1838 2222 235
l l
1848 58 1846
l l
1817 1856 1869
ll
1821
l
1850 1959 137
135 140 143 240239
l
1816 1819 243 241
l l ll
647071 6675 1873
l
134 1835
l
1818
ll
1957
ll l
65 15162021
l l
1854
l
1815 1834
llll
6768 127 138
l l
130 1837
l
1811 1857 126 242
l l l
14 13 1905
l l
1813 1956 7374 118119
l l
125122
l
1814 1855 72 159
l l l
1812
1808 1960 120
l
123
l l
1809 1861 8084 139 1836
l l
1906
l
1955 77 76 115 129 132 142
141
l l l ll
1954 107106 117121 133
ll
1806
1810 78 1833
l
124 128 144
llll
1859 113 131 11 216
l
1805 1862
ll
79 109 10 1907 246 245
llll
82 114
l
1807 247
ll
116
l l
93 100
l
1758
l
158
ll
89 102
l
8591
l l
1860 105110 145146
l
1801 244
l
18021803 1863
929094969799 104 151 9 12 214215
l l l ll
101
l ll l
83 149
l
1800
l l
l l
1804 8186 152153 213
l
150
l
112 147
l
103108 248250 210 212
l
1799 8788 154 1755
l
249
l ll
148
l
17571756
l l l l l
156
ll l l lll
1795 19511953 111
l ll
1798 1794
l
1925
l l l
1792 1864
l
l l l l
1865 159 1754 1908
l
1753
l
17961797 1752 1923
l
9598 157
l l
1922
l ll
1790
l l ll ll
1791 1759 1921 1909
l l
209211
l l
1790
l l
1793
l
155
l
1787 1867
l l l
1791
l l
1910
l
1789 1948 158 164 1749 1926
19241920 157
l l
1785 18661874 1952 160 1751 17871786
l l
1868 1911
l
253251
l
1788
1786 1871 8 1919
l
1750
l
161
l l
1873 1760
l l l
155
l l
1928
l
1950 1785
l
1949 1796 17921788
l
1869 1879 1748 1914
l
1875 1747 314 316 320
l
ll ll
1947
l l l l
1793 17891782 1913
l l
1870
l
1780
1784 1877 163167 1927 255252 156
l
1912 153
l
162
l
1880 1946 208
l l
311
l
1779 17831784
l
1781
1778 1872
1876 1944 1915 61 154 310313
l
1883
1881
l l ll
1882 6 149
l
1945 1794
l l ll l
1776 1744 5 71929
l
312315 317319
l l
1878 165
l
1884 1942 1746 150
l
254 152
l l l
166 1916 146
l l
1885 1943
l
1783 170171177 1742 1795 1918 144148 207
l l
1889
18861887 2 3 1931 1917 308
l l
260
l
1774 1891 322
l
1941
ll l
1888
lll l l l ll l
1775 1745 321
ll
1782 169
172 636662 147151 309 324
l
1893 168 330
l l
1777 18901892 19391940 179 1741 1930 257256 145143 318
l
327
l
1743
l
1773 1894 1896 174 204 307 325326
l l
1771 1936 173 1781 323
1799 17981739
l
1740
l
1895 19351938 1 4
l l l
182
l
1932 206
l
258
ll ll
205
l l
1770 1769 11937
934 65 64
l l
1772 1932 1737 1797 1777
l ll
1933 175 329333
l
18971898 328
l
1768 176
l l l ll
1778
l
1779
l
1776 142 331
l
1738
l l
1930 181
l
183181
l l ll
183
l l
261259
l l ll
184 180
l
19291931 203
l
1765 1933
l
1774 1780
l
332
l l l l
1763 178 1775
l
1899 1928
l
305
l l
185 1736
l l l
1767 1758 1927 141 334
l
1761 263
ll l
262
l
1801 1800 1934 336337
l
1900 1772
l l
1771
ll l l
1926
l
1759
1760 188 1735 1773
l l l l l
1764 184
l l
335
l
1734 67 137
l
1754
l
1766
l
182
l
306
l
304
l
1805 264
l
1924 186
l
1753
l l l
1918 1733 1769
l
187 1803
l
1751 185
l
1762 1752 1901 1730 200
l
1755
l
1747 1923 189 1732 1770 1935 139
l l
1802 1729 265267 140138 135
l l lll
1757 194
l l l l
190 338339
l
1914 68 70
l ll
1746 192193 188186 303
l
1731
l
1756 1749 191 1804
l l
1743 1902 340
l
1741
l l
19031906
l
1807
1808 17671768 187
l
1748 202
l
17501744
1745 1909 1925
1921 196 198 266
l
136
l
1913
1919 195 302 341
ll l
1738 1910 1806 69 7271 73 201197 342
l
17391737 197 1809 1728 189
l
1740 204
202 199 348
l
1936 1943 343
l l
1922
l l
1920
l
1917
l
201
l
200 1812
l
19041907 345
l
268
l
1726 198192 347
lll ll
17421732
1729 203
211 1810 1727 269
l l l
1911 199195 193190
l
1735 1912 207
205 206 1813 1725 1939
1937 1944 271
l
19051908
l
1731 300
l l l
1916 209 346
ll l
1733 213 208 1811 1947 270 344 350
ll
1730
1725
1723 1915 210 212 1938 273 196194 191 349
l
1941
ll l
1942
ll l ll l
1724
l
1722 214 1814 74 352351
l l
1702 1940 1945
l
1736
l l l ll
1726 216
l
1728
l
215 1723 272 133134
l l
1720 1700 1948 301
l
1946
l
1734 1721 1950
l l l l
17031701
l
217
l l l
220 218 219
l
1724
ll l
1727 1720
l l ll ll
195119521954
ll l l l
1716
1718 1719 1721 1949 299
l l l
353355
l
296
l
222 1722
l
132
l ll l
1699
ll l
1713
l l
131
l l l
237
l l ll
75
lll
227
225221
l
224 1696
l
1714 294
l
274
l l
223 1716 19531955 297298
l
l
1717
l
1704 7677 293
l l l
229
l l
1718 1719 356
l l
354
l
242
l l
239
l l
1712
l
234
l
1698
l l l l
254 246 1717 1697 129 130
l
1715 241 275
l l l
230
228
l l
236 290 292
l
235 1956
ll l
127
l l
243 1705
l l
231
l
1715 1694 1695 358357
l
253 226 1713
l
244 233
238 240 80 7879 288 291295
l
248
l l
257256 247 369370
l
266
l l
232
l
276
l l
1579 1714 1958
l
1710 251 245
l l ll
1711 1707
l
255 1616
l
125 126128 277
l
252250
l
l l l l
1613
l
267 259 1708 1615
l l
371 366 364363 360
l
1709
l l
264 260258249
262 289
ll l l
1618
l
265
l l l l
272269 1710 1957 1959 278 359
l l
1708
ll
1577
1578
l
1711 1706 818284
l l
280 1619
l
287
l l l
1573 261
l l
274 271263 1712
l
1583
l l
1620
l l
1617
llll ll
1572
1568 1832 83 279 281 284 285
l
1574 1576
l
1590 282 268
l l
1709
l l
270 9091
ll
88
l
1570 276
l
275 123124 362361
l l
1575
l ll
1571 1659 280 368367 365
l
1707 273
l
1580 15921582 290286
l
1586 1663
l
277
l
89
ll l l l
1960 85
l
285 279
l
1565 1598 1705 86
l
1581 1693
l
1584 278
l
1706 87
l
1569 1595
ll
1564 1661
l
281 282 283
l
283
l l ll
16601658
ll
1585
1587 1605 1692 1961
ll l
1588
l
1567 286
l l
1566 1603
l
1601 1614
l l
1827
l
1561 284
l
291287 1829 16641662 9293
l l
1589 16671665
l l
1562
1563 1591 121
l
1691
ll
95
l l
1560 1594 1828 1831 122
l
1593 94
l
1600 288 289 1657 1964
llll l l l l l
1558 1597
ll l l l
294
295 1826 1830 1688 1668 1655 96 118
l
1559 1596 1825
l
1609
l
1599
l
1689
l l l
1555 299 293 1690 97
l l
1704 1673 16701669 1666 1656 1962
l l l l
1552 99 119120
l
1604 303 438
l l l l
15561557 1602 307 98
l
1553
l
292
l
1549 1824
1823
l l l l l
437 440441
l
1554 1551
1550 1607 1703 1653
l l
1606 480
l
296
297 1654 1963 1 17
l
1687 1685 1676 1650 100 116
l
1608 309
311 1965
l l l
1548 1611
1612 300 1671 1651
l
1672 436
l
302 1570 1678 1641
l l
1547 1686
l
298
l
1546 1610
l
479
ll l
1822 435 439
l
511
l
1567
l l l l
310
l
1622
1614 1681 1674
1677 1652
l l l l l
478
l
301
l
434
l l
306
l l
1545 304 115
l l l
1649
1648 507
l
1615
l
1616 308 1568
l
312 1684 16801675 474 432
l
1623 1701 316 163216351637
l l
305 1640 1642 477
l l
1613 1621
1617 1625 318 1821 1647 101102 508 509
l
1618 1702 317
l
1679 433 510
l
1697
l
313
llll l l l l ll l l l
1682 1638 512
ll
1619 472 475
ll
1620 1700 1569 476
ll l
1543 320 112 513
l
1542 315
l l
431
l l
314 1633 1639 1967
l l ll
1544 1627 1695 1683
l
1626 1693 1818 16311630 1634 1636 506
l
429
l
113114
l
1624
l l
1694 111
l l l
1538 1572 1629
l l
1539 1571 1644 473
ll l
1696 1820 471
l
515
l
319
l
1628 1698 1628
ll ll ll
505
l
1541
1540 1692 1819 1645
l
430
l
1629 323 1817 503
l l ll
1699 514
ll
1816
l l
1691
l l
1630 1573 1646
l
108 110
l l
1690 103 104
l l
321
324 1575 1643 446
l l
1535
1536 1689 1625 16261627 1966 451 504 516
l l
1580 502
l
l l
1537 1631
l l l l
1534 1815 1576
l
448
l
l l
1688
l l l
325 322
l
427
l l l
1533 445 501
l l
1611
l
5517
l
518
l l
106 107
lll l l
1632
l
1684
1682 470 500
l l
157715781579 447444
l lll
1532
1530 1685 326 105
l
452 428
l l
1523 1574 450
l
497 20
l l l l l l l l
519
l
1634 1681 1623
l l l
1531
l
1679 1624 609
l l
1638 425 498499
l
327
l l
15291525 1687 449
l l
1680 1518 109
l
1635
l
1676 330
l l l
1524 1633 329 424
l l
1686 496
l l l l
1526 1639
l l
1642 1519
l
1674
l
1522 1637 1683
l
334 1622 454 610611
l l
1640 1647 328 455 458 453 426
l
1652 1766
l l l l
1636 332333 1762 15161515 1621
l
1527
l l
1519 1643 1646 16591662 1671 464 1612
l
1609
l
454
l
1677
l l
1528 1664 335 331
l l l
1670 466
l l
1648
l
1610
l l ll l l
521
l
1651 1667
l
1657
l l
1644 338 494
l
1641 445 449
l l l l
1521 1675 1764
l
460 1517 608
l
437438 1765
l l
1518
1520
l l
1678
l
1655 337 456
l l
1645
l l
1673
l l
1656
l
1658
l l
1607
l
423 606607
l l l
493
l ll
1650
l l
1654
l l l
1672 339336 1608 1605
l l l l l
1661
l l l l
447
451 453 455
l
1649 340 459
457 463 469 1763
l l l ll l
1602
l l
439 446
442
l l
1663
l l l l l
1515 462
lll
443 1761 522
l ll
1653 495
lll l l
342 452 1593
l
1600
ll l
444450
ll l
605
ll
343 465
l
1514
1512 1660 1668 433 461 472 1601 386 5526
lll l l
1511 440
l l
1594 523
lll
1669 435
ll l
160315961597
l
1665 475 1588
l
429 436 441
l l
1666 1595 1592 422
l ll
345 434 1587 603604
l
1564
l ll l l l l
1506 341 348 1606
l l
347 427428 448 467 524
l
1513
l
344 432 468 1565 25
l l l
1504 1590
l
1520 380 382 385
llll
346
l
426
l
1517 1510 474
l llllll
1501 354
350351 418425 471 477 1562 1599 15911589
1584 1582 492
l l
1507 1497 353359360
352 369 414 422 430 470 476 1563 1604 1598 11581
585 602 601
l
349 417
ll l
423
l ll
1508 1498
1496 366 3383389 409 411416 420 481 1522 1583 379378 383456388 420421
l
1516
l
1502 363 393 480 527
l
1505 1561 381 384 394 416
l
355356 405
l l
376
372 408 473 1538 1532
ll
377 412 415 431
l l
1509 14991493
1490 361 362 390 1560 1523 377 390391 415 491
l
1489 368 381 398 15251521 395
l l
357 364 370 73 396
392 395 401402
403 407 419424 397
l
1500 1494 365 379 400 406410413 421 478 1558 1539 1586 373376
l
374 386 1529
l
1491 1486 358 367371 375 388 399 457 387 389 417
l
1492 404 1559 1536 1524 375
l
1503 391 483
479 396 419
l
387 397
l
1535
l
1504
ll
1487
l
380
382 1531
1530 372 405
l
1488 392
393 528
l lll
1485 414 599
l
1999
l
1495 385 394 482 1545 1541 1502 398399 530
l
378 1557 1533
1537 1505 1968 418 529
l
1540 1528 1526
l l
484 1527 1499 374 401 413
l
2000 458 594
l
600
l
14821483 384 486
485 1534 1506 1500 1998 597
l ll l
403
ll
1501 1997
l
1479 15461543 459 4 04 531
l
1484 400
l
1510 1508 598
l
1481 1544 1542 1503 1498
1497
ll
1996 412
l
406408 410411 595
l l
14801477 487 1509 596
l
488490 1507 461 402
l
1556
l l
1475
l
1995 592591
l
1476 1478 1512 1971 1969
l
1511 1992
l
1993
ll
1513
l
1474 1549 1972 590 490
l
463 593 588
l l
493492
l
1552 1551 1547 1994
1990 1973 1970 460462 407 409
l
589 587
l
14711473 1548 1496 443
ll
1976 1974
l
1514 1989
l
1550 1991 1975 586
l
489 1553 14951494
ll l
464465 585 584
l
1984
l
491494 19871986
l l
1980
l
14691470 1472 1982
l l
1554 1979 1566
l
1978
l l l lll l
1555
l l
1977
l l
1493 19881985 572 574
l l l l ll
495
l l l
1981
l
582583
l l l
1983
l l
1492
l
579
lll l ll
l l l
466 467
l ll ll l
573 578
ll l l l
581
l l l
497 575 576577
l l l l l
1463 1467
lll l l
469 468
l l l l l
1464
l l
580
l
970
l l l l
1468
l l l ll
1465
l l
489
ll l l l l l l
1491 568567563
l l
14621466 1376
l l l
11458
l l
570
l l
1404 1373 13861390 971
ll
1490
l
1400
l
13711374
l
496 1375
l l
1453 454
1457 13971398
l l
1448
l l
498
l l l l
1399 1378 1445
l
1387 569566
ll l
1459 1402 1389
l l
972
l l
1450 1380
l l l l
1379 484
l
562559
l l
1451 1461 553
l
501 1384
l l l
499500
l
1446
l
1444 1449 1456 1372 1392 561
l l
1377
ll l
1383 1391 556555554552
l
1445 1446 14551460
1452 1396 549
ll
1401 1443
1441 571 551548
l
503 502 1385 1444 985984982 977 976 973974
l ll
558
l
1443 1448
ll l lll ll l
506 1370 1388 560 547544536 488
ll lll l
1382
l ll
1403 1381
l
1440 1447 1393
13941395
l
1442
l
504 975 641 565564
ll
546
l ll
507 636
l
509 1440 634 639
l l
627628629 557 483
l
1447
l
1432 14411442
l
1434 1437 511 505 630
l l ll
539
ll l
14381439 510 550 542 485
l
545
l
538 487
ll
1438 14881489 633 640
l ll
1369 1437 638 486
l
642 537
l
1435 508
l
l l l
622
l
514 980978
l
1425 14261436 543540
541 481
l l
512
ll
1430
l l
1423 1436 621 626631 637
l
1428 983 979
l
1449 615 617 620 625 482
l
1433 513 632
l
635
l l l l
1420 1429
ll
1422 1431 612 1407 442 535
533534
l
1419 515 1439 614618619
ll ll
1487 987988
l
1418 609 1450
l l
1421 517
l l l
625 603
l l
1435
l l
613 600
l ll
111415
417 516
l ll
518
l
1424
l
l l
602
l
620
l
1368 1367 1406
l l
1414
1416
l
616
l lll l
616
l
1405
l
14131427 981 623
l
1451
l l
1434 986
l l
611 607 624
l
1409 605
l
1410 1411 628627 621 614 1486
l
597
l l
408 610 520 1409
ll l
532
l l
1407
1412 617 530 519
l
595596 1452
l
606 604 593 1366
l
1 453
l
1406 626
l l
618 598 989
l
1408 1454
l
601 589 579
l l
522
l
1365
l
1402 1364 1485
635630 623 608599 524
l l
1404
1405 619
l l l l
527 521 1410
l l
523
l l l l
1403 615 594
l l
591 1455
1456
l
590587
l l
1401
l l
586 528 1412
l l
622
629 624 965 967
l
529525 1363
l
584
l
631 526 1457
l l
532 14111413 1421 1484
l l
1362
l
585 570
l l l l l l
1433
l l l
581
l l l
536
l l
1416
ll l
582 575 540 1361
l
541
l l l
1400 583 577 531
l l l
547
l l
1359
l l
634
l l
636 569 534 537 715
l
l l
539
543 533
l l
1399
l
573 568 563 544 538 1414 1419
1417 1482
1483
l
1398 633632 588 578
576 566567 551
557553 546 535 1357 1459
1458 966968969 722
l
542
l
716
ll l l l l
1397 640 592 561
562 550 1360 1415
1418 1422 1431
l
728
ll ll
580
l
1047 558 552 1420
l
571
l
l
1056 643 637 572 560 549 1355 1424 1425 1429 1432 1209 718
l l
1430 1481
l
639 719717 724
l
554 1480 721
l
1395 1358 1477 725
l l
1050 1045
llll
565 548 545 713714
l
1396 638 1210
l
651 574
l l
1354 1476 708
l
564 1356 1472
l
644 1423 720 732
l
1394 1054 1049 1043 641 556 14601463 1462 1478 709 723 729
ll lll lll l
1427 1464 12071208 727
l l
648 559 1465
l
642
l
1059
1052 1046
l
1353
l l l
1058 1038 1471 711
l
712
l l
1392
1393 1044 1041 650 555 1428 1461 1466 1473 726 730731
l
645
ll
1057 10531048 1206 1211 707
l l l
1039 649646 734735
ll
1042
l l
1205 710
l
1352
l l
1040
l l
1468 1204 733
l
1055 647 1426 1201
l
1202
l l
14691474 1215
l
1390
1389 653
652 1475
l
1037
1036 1467 1479 1213
l
705
l l
l
1034
l
1200
l
1212
l ll
706
l l
1203
ll
l l l
1035 1214
l l
1470
l l
1219
ll l
1386
l
1051
l
1391 1033
1031 657654655
ll l
1384
l l
1199 1226
1225
l l l l
1032 658 1217
l l
1383
l
736
l l l
1060 656 1216
ll l l
1348
13511349 1218 1223
l
11971198
l
1385
l l l
1388
l l
659 737
l l l
1029 660
l
1227
l
1030
1028 1224
l l
704
l l
1222
l l
1047
l
1382 661 1049 701 738
l
1195 1221
l l l
1381
l l
1387 703
l
1373 1351 13371325
1340 1026 1347
l
1027
l
1376 1374 662664 1341 1340 11901191 1220
ll
1378 1194 742
l l l
1375 1344
l l
10631061
l
1050
l l
13501346 1192 1193 1196
l l
1338
ll
1371
l
1023
l l
1022 663
l
1377 1046
l
1341
l
739
l l
1380 1323 1319
l l
1369 1345 1343 1048 702 740
l l
1025
l
1024
l
1346
l
1379 1364
1360 13361328
1324
1329 665
666
ll l
1339 741
l l
1372
1370 1359 1355 6667
671 1339 1188
11861189 779776
ll
1361 1356 1349
13501347 1330
1334 1062 1342 789786784
lll
1343 698700
l
1326 778780
llll l
1342 748743
l
1021
lll l
1327 70 677 1051
ll
1354 668
ll
1366 1352 1332
1335 1321 782777
lllll l
1348
l
1357 1344 1333 1322 1315 1064 669 783
l
1363 1318 1316 749 745746
ll ll
1331 1338 1337 1183
l
1020
l
1368 1314 744
l
13581353 1345
l
675676
ll
674 781
l
1362 1317
1320 1313
ll
1365 672673 682 1052 695 775
l
1312 699
l
1066 1184 785
l
681 787
l l l
678
l
1367 1187 788
l l
1311
l
686
l l
1067 679 691692 694 696 697
l
774
l
1068 683
lll l l l
1069
ll l l
747
l l
13071308 1182 1012
117711791180
l
1185
l
1310 680 684
l
1017 691 1336 1016
l
1019 687690 1017
l
1065
l
1268 1266
l
l
1309 685 688
l
1306 690 773 750
l lll
1269
lll
1013 693
l l
689 1025
l
1302
l
1304 689
l l
1303 692 1019
l l
1176 1181
l
13341335 1011
ll l
1014
l
1015 1178
l
1271
l
1305 694 697
l
1018 1015 1007 687688
l l
1071 1270 1267 1022
1026 1014 751
l
770
l
1020
l
1264
l
1300 1301 1018 1277 1274
1275 1010 1008
lll
1072 699 708
ll ll
12721265 1263 1333 1024
l
790
l lll l l
693 1273 686685
l
1073 1175
l ll
771772
ll
1029
l l
1299 695 704 1021
l
1009
l
696 701702
l l
1298 1016 1332 1023
1028
l l
1276
l l
1296 1013
l
791792
l
1297
l l l
1075
1076 709
700705 1279 1328 1174 1006 796 752 755
l
1030
l
698 684 769768
l
1074
1070 1278 1331 1027
l
1005
l
1077
l l
703 1327 1031
l l
1079 1329
ll
1294
l
794793 753754
l
1295 710711 714 12621261
l
1012 1170 1172
l
706707
l l l l
1292 767766 757
l
681680
l l
13301325
l l
1291 795797
l
1169 1171 1033
l
683
l
1173
l
1293 1011 1280 759
l
l l
1032
l l l
1034
l
716
712713 719 758
l
678675 674 763760
l l lll l
717 725
l l
720 1260 1258
l
1080 1010 1281 1259
l l
10411036 1003 756
l
1288 1045 1043 682
l
1285
1286 1083
1082 724 1326 677 764
l l l
1289 1078 1009 726
l ll ll l l
l
1008 1039 1035
1037 679
l l
715
llll
11681167
l
1282 761
l
1081 765
l l
676673
l
1084 722 1044 1042
l l
762
l ll
729
l l
1290
l
718
l
1287
l
721730
l
1007
l
1006 1004
l
731 1040 798
l
1087 1085 723 1284 672
671
l
1256
l ll
924
l
1091 1038
l
1086
l
1004 1283 670 799
l
727
l l l
1282 728
ll l
1284 1003
l
739 1166
l
997 910
l l
1005
999 735
l
1088 1001
1000 732 1001 908 666
l l
1283 1165
ll l
923922 669
l
1285
1286 12571255 1164
l l lll
998 734 902903 800
l
1280 1002 906 667
l
911 668665 662664
l l
995996
l
1089 1096
l
1090
lll ll
1288 1254
l l l
993 733 1251
l
991 740 901 905
l l lll l ll
1279 1094
l
986 741 909
l
1252 921
l
1278 994
l l
736
737 1287
l
1092 904
l l
988 1289
1291 1250 1163 907
l
992
ll
1253
l l
984
983
l l
738 999 1002
l
1277 899 659654
l
1281 1097
1093
l
982
l
919918 900
l l l
663 801
l
1249
l l ll
1275 1098 981
987 1160
l
1095 985
989 1294 661
658
l
1248 1000 897
l l
990
ll
1290
l l
742 917
l
1274 1276
l
1099 1161
l
743
l l
1272 980 1246
1247 920 656
l
898
l l
744745
l
657 653
l
1103 1100 896
l
1273 1102 1292
l
1293 1162
l
l l
914915
l l l
746 747
l l
979978 1295 1245 916 660
l l
1158 1159 995993
l
1269 748
l l l l lll
651
l l l l
655
l
997
ll
12701271
1268
l
1296 1244
l
1323 895
l ll l
1101
11081109 752 804
l ll l
975 750751
749 1157 894
ll
796
l
753
l l
892 652650 803 802 647 644
l
976 797 1297 1242 994998
l
1106
l
12661267
l
1265 1263
1262 1104 794 754 1243
l l
1264 913
l l l
1156 991 990 992
l
800 1155
l l
788 649 646 643
l
977 791 1239
l
12601261 1111
1105 792 785 779776 756 755 1298
l
1154 891893
l
1107
l
1259 803802 1299
l l l ll
1110 807 787786 761
l
795 782 759 996
l l l
757 1324
l
1258 1257
l
1112 971
973 798 789 763 758 1241 912 949
l
781
784 760 805 648 645
l
1255 771 1240 1153
l
l
1256 777
l
793 780
l l ll
951 958
ll
799
l
1113 790 766
l
810 774 770
l
804 950
l
1254
1252 11115
1114 972
974 813 801 778 772 768 762
l
1117 1300 1152 890 889
l
775 1301 1238 953
l
118 783 942
l
947
l
1249 955
l
769 1322
1321 11501151 946
l
1116
l
806 1148
l
1251 1244 805 765 1320 1149 948 956960
l
1121
l
809
l l
l l l
1126
l l
1250 811 808 1302 943945 954 962963
l
1243
l
1125
ll
767
l
1253 1120 818 764 952
ll
11191124 773
l
1147 938
ll
940
lll ll
1248 888887
l l
824
l
969
l l
1247 1241 1123 817 1146
ll l l
1317 941 959
l
823 1318
l
1122 1127 937935 957
l
812
l
1245 968 970 814 1237
l l
1319 1145 934 944
ll
931933 939 961964 806
l
1128
l
815
l
1316 1144
l
l
1240
l l l
1143
l
967
l
1246 1129
ll l l l l
828 816
l
1131 930
l
1130
l
1239 965
964 826822 1142
l l l
936
l
1242 1133 1315 885
l
886 812809808 807
l
1303 1236 1235
l
1237 821
819 932 929
l l l
966
l
820
l
1134 960 825
827 1232
l
1132 1141
l l l l
1236 928
l
1136
l l
832 835
l
829 1140
l
8 30
l l l
1231 1228
l
1232 1135 927 813
l
1137 963 831 1071
1072 884883
l
1238 1230 962 833836 1234 810
811
l
815
lll
961 1139
ll l l
1234 1139 834
l
1068
l
1229 1070 1077
l l
1141 837
ll
1138 845841 1082
l
958 9954
59 1233 1231 1073 880882
l
840 1069 926
lll
814
l
1233 957 838 1079
l
816
l l
1225 839 1066
l l l l
1235 1227 1140 130412301228 818
l
955
l
1135 1075
l
953 1138 10671074 820
l
956
l
1223 1224 842 1314 1081
1078 1087 881 817
l
1143 952 847
l
1065
l
846
844
l l l
1229 1221 843 1136 879876
1142 10631064 1076 1080
l
951 819
ll
1226
1222 925
l l
878877
l
1144 946
950 848
850 1084
l
1220
l l
849
l
1218 1145
l l l l
1219 947 1137 1085
l
851
lll l l l l
1083 1088 823
l
1216 1146 948 860 852
l
949 1060
l
1148
1147 944 874
l
1086
l l l l l
825821
ll l
1215
1217 945 1313 1134
l l
1213 940 853854 1059 826 822
l
1062
l l
942
l
875
l l
857
l
938
936 855856
l l l l
941 858 1305 1057 873
l l
943
l
831
l
1214 937 861 1133
l l l l
939 859
l
1149 1089
l l
1058
l l
1091 827 824
l
1131 1061 832
l
1150
l l
1211 862 828
l
1151 1090 836 830
l
1152
l
864
l
1212 863 872
l
1153 1056
l l
833829
l
935
l
931
932
l
1054
l l l
1155 13061311 1092
l l
1154 865
l l
1130
l l l
1156 929 866
l l
834
l
837
l
1158 875 868
873 1132 1094 871 868 835
l
1157
l ll l
1159 934 928
924 871869
867 1308 1129 1053 1093 838
l l l ll
l
870
l
1312
l
1162
l
11661161
l
1208 1160 930 1307 843
l
1207 1200 874 841
ll
1163 872 1309
l lll
927 869870 839
ll l
1196
l
1199 1168 1164 876 1310 844842
l l
1202 1170
l
1206 1197
1198 1169
1171 1165 933926 881 1096 867866 853 845 840
l
1203 1187
1194 1167 925 879 1055 1095 854851 846
1179 1174
ll
1210 1186
1184 1177 1172 882 877
878
l
1201 1191 1180 849847
l
922
ll
1195 880
l l
l
12091204 1173 855 852850848
l l
1192
l l
1119
lll
1189 865
l
1181
l
1205 923 1125 1123 1122 1120 11171097 1116 856
l
1193 11821178
1183 1118 860
l
1190 1176
1175 921916 884883 11241121 11141113 1115 863
859
861858
l
1188 886
l
1185 918915 893888885
887 11281127 1126 1099
1098 864862 857
913911
914 905
908 902
899
904 892
898889890 1100 1101 1110
919
920 917 909 897
901
906
903896
895894
900891 1102
1103 1105 1108
1104 1112
910 907
912 11061107 1111
1109
(a) (b)
Fig. 2 In panel (a) we see simulations of random interval graphs under the specifications of
Theorem 2. As expected, graphs are always connected. In panel (b) we see simulations of random
interval graphs under the specifications of Theorem 3. As expected, graphs have always more than
one component
Empirical Beta(n−1,2) distribution Conditional empirical distribution of maximal length
15
30
10
Density
Density
20
5
10
0
0
0.86 0.88 0.90 0.92 0.94 0.96 0.98 1.00 0.95 1.00 1.05 1.10 1.15
(a) (b)
Fig. 3 Empirical distributions illustrating Theorem 4. (a) Histogram of R − L. The red line
displays the density function of a Beta(n − 1, 2) distribution. (b) Histogram of R (n) − L (n)
on the event of a connected graph
convergences to a Beta(n − 1, 2) distribution with n = 100. Beyond this evidence,

we estimated a Wasserstein distance of 0.00014 between the Beta distribution and
the empirical distribution function obtained from the simulations. To this end, we
applied R package “transport”. On the right side Fig. 3b we see the histogram of
R (100) − L (100) for the same 2e5 simulations as before but intersected with
the event that the interval graph was connected. The random variables R (n) and
L (n) are defined in Theorem 4. We see a very high probability of reaching or even
surpassing the intended length, again, as predicted.
The bifurcation of (non) connectedness for random interval graphs in our
asymmetric model has been illustrated on Fig. 2.
In Fig. 3 we started to see the behavior of empirical probabilities and now we go
deeper into characterizing by simulation the sensibility of these probabilities with
respect to parameters. In all experiments we generated a hundred trials per set of
parameters, then we checked connectedness and saved the percentage of connected
instances.
In Fig. 4, following the hypothesis of Theorem 2, we fixed two different values
for a = 0.33, 0.66 in order to numerically illustrate that our requirement of a < 1/2
is crucial. We ran simulations for different values of c, while varying the number of
agents from 0 to 10,000, to observe how the choice of c affects the convergence
speed to obtain connected interval graphs with high probability. For low values of
a we found experimentally that a value of c between 3 and 4 suffices, as shown
in Fig. 4a. As expected, high values of a yield a slow convergence speed, which is
illustrated by the low empirical probabilities of connected interval graphs observed
in Fig. 4b.
Fig. 4 Simulations illustrating the sensibility of empirical probabilities to parameters. The

dynamic of r(n) is according to Theorem 2
Fig. 5 Simulations illustrating the sensibility of empirical probabilities to parameters. The

dynamic of r(n) is according to Theorem 3
Next, following the hypothesis of Theorem 3, we considered two fixed values for
n, namely n = 10 and 100, to see how the change of the order of magnitude of n
affects the approximation of the probability that graph instances are not connected,
as stated in Theorem 3, but not only that, we also vary a to observe how fast this
result is achieved. Figure 5 shows ten different trajectories for ten values of α. As
expected, there is a monotone relationship with α. Experimentally, we found that a
value around 0.5 is a critical value where disconnected interval graphs appear with
high probability for moderate values of n. As it is to be expected, by Theorem 3, as
n grows arbitrarily large, the probability of generating disconnected interval graphs
increases for any value of α. Also, experimentally we found that large values of a
dramatically increases the probability of generating disconnected interval graphs.
In Fig. 6, following the hypothesis of Theorem 4, additionally to connectedness,
we also account for R − L ≥ 1, guaranteeing this way solvability. As before,
we compute the empirical probability by executing a hundred trials per set of
parameters. We fix the number of agents in n = 10, 100, and consider seven
different values for c. We observe from the experiments, as a variates from 0 to
1, the empirical probability raises until a peak is reached when a ≈ 3, consistently
for the two values of n. Also, we note that as the value of n increases, the empirical
Fig. 6 Simulations illustrating the sensibility of empirical probabilities to parameters. Probabil-

ities are conditioned to the event of hitting the boundaries of the interval [0, 1]. According to
Theorem 4, this probability is asymptotically high
probabilities increase for all practical purposes. A last observation is, confirming the
hypothesis that a < 12 , beyond the value a ≥ 12 it is impossible that R − L ≥ 1
and therefore the empirical probability crushes to zero.
5 Conclusion
As we have seen in Theorem 1, the DDLP has an equivalent representation as a

graph theoretical problem. Specially, solvability of a DDLP instance translates into
the existence of a connected interval graph. This interpretation is far reaching since
DDLP is NP-complete, and therefore it is likely that no efficient algorithm can solve
it. In contrast, checking for connectedness can be performed very efficiently.
The use of probability theory allowed us to explore the characteristics of a subset
of solvable DDLP instances. On the one hand, Theorem 2, showed a threshold for
the regime of r (n), under which GA (Q, a) is connected almost everywhere. Our
simulations gave a visual illustration of Theorem 2, where the selection parameter
of a and c played an important role in order to observe convergence speed,
so that connectedness shows up for reasonable values of n. On the other hand,
characterizing disconnectedness in Theorem 3, the function r(n) follows a different
regime under which disconnectedness of GA (Q, a) has high probability.
Disconnectedness in the asymmetric model does not tell us if an instance
is not solvable, but the symmetric model does guarantee that a disconnected
interval graph renders unsolvable DDLP instances. This will produce fringes for
solvability/connectedness and unsolvability/disconnectedness as we now discuss.
Indeed, Theorem 2 implies connectedness with high probability under the regime
2 log(n)+c
r(n) = 1−a n while Sheinerman shows that for r(n) ≤ log(n)/n almost all
graphs GS (Q) are disconnected, and thus the corresponding DDLP is unsolvable.
2 log(n)α(n)
Theorem 3 determines that regimes r(n) = 1−a n with limn α(n) < 1
have high probability of disconnectedness (in a severe way!). Note that there is a
gap in between both regimes in which solvability is ambiguous since it depends on

asymmetry of the model.
Our simulations for Theorem 3 showed sensitivity to the choice of parameters α
and a, in order to appreciate the asymptotic result for reasonable values of n. Lastly,
Theorem 4 provides insight into how solvable DDLP instances
arise with high
probability, as we considered the connectedness of GA Q , a . Our experimentation
showed that a ≈ 13 and a value of c > 3 are good parameters to generate solvable
DDLP instances with high probability for most values of n.
The analysis in the paper sets up for a taxonomy to generate feasible-to-
solve DDLP instances, which lead to use efficient approaches, such as checking
connectedness in an interval graph, as its inherent difficulty makes very difficult
to solve DDLP instances in general. The DDLP representation as a graph theory
problem, showed to be fruitful, because it allowed us to discover asymptotic
behavior in randomly generated instances, analogous to the one described by Erdős
and Rényi, and Scheinermann.
References
1. Anaya, J., Chalopin, J., Czyzowics, J., Labourel, A., Pelc, A., Vaxes, Y.: Collecting information
by power-aware mobile agents. In: Aguilera, M.K. (ed.) Distributed Computing, volume 7611
of Lecture Notes in Computer Science, Salvador. International Symposium on Distributed
Computing. Springer (2012)
2. Arora, S., Barak, B.: Computational Complexity. Cambridge University Press, Cambridge
(2009)
3. Bampas, E., Das, S., Derenjowski, D., Karousatou, C.: Collaborative delivery by energy-
sharing low-power mobile robots. In: Fernández, A., Jurdzinski, T., Mosteiro, M., Zhang, Y.
(eds.) Algorithms for Sensor Systems, volume 10718 of Lecture Notes in Computer Science.
ALGOSENSORS. Springer, Cham (2017)
4. Bärtschi, A., Graf, D., Mihalák, M.: Collective fast delivery by energy-efficient agents. In:
Potapov, I., Spirakis, P., Worrell, J. (eds.) 43rd International Symposium on Mathematical
Foundations of Computer Science, vol. 117, pp. 56:1–56:16. MFCS, LIPICS (2018)
5. Biau, G., Devroye, L.: Lectures on the Nearest Neighbor Method. Springer, Berlin (2015)
6. Bollobás, B.: Random Graphs. Cambridge Studies in Advanced Mathematics. Cambridge
University Press, Cambridge (2001)
7. Bondy, J.A., Murty, U.S.R.: Graph Theory. Graduate Texts in Mathematics. Springer, Berlin
(2008)
8. Chalopin, J., Das, S., Mihalák, M., Penna, P., Widmayer, P.: Data delivery by energy-
constrained mobile agents. In: 9th International Symposium on Algorithms and Experiments
for Sensor Systems, Wireless Networks and Distributed Robotics, pp. 111–122. ALGOSEN-
SORS (2013)
9. Chalopin, J., Jacob, R., Mihalák, M., Widmayer, P.: Data delivery by energy-constrained
mobile agents on a line. In: Esparza, J., Fraigniaud, P., Husfeldt, T., Koutsoupias, E. (eds.)
Automata, Languages, and Programming, volume 8573 of Lecture Notes in Computer Science,
pp. 423–434. ICALP, Springer, Berlin, Heidelberg (2014)
10. Cohen, J.E.: The asymptotic probability that a random graph is a unit interval graph,
indifference graph, or proper interval graph. Discrete Math. 40, 21–24 (1982)
11. Cook, S.: The complexity of theorem-proving procedures. In: Proceedings Third Annual ACM
Symposium (1971)
12. Czyzowics, J., Krzysztof, D., Moussi, J., Wojciech, R.: Algorithms for Communication
Problems for Mobile Agents Exchanging Energy. ArXiv:1511.05987v1 [cs.DS], November
(2015)
13. Edmonds, J.: Paths, trees and flowers. Can. J. Math. 17, 449–467 (1965)
14. Erdős, P., Rényi, A.: On the evolution of random graphs. Publ. Math. Inst. Hung. Acad. Sci.
5, 17–60 (1960)
15. Garey, M., Johnson, D.: Computers and Intractability. Bell Telephone Laboratories Incorpo-
rated, Holmdel (1979)
16. Gilbert, E.: Random graphs. Ann. Math. Stat. 30(4), 1141–1144 (1959)
17. Karp, R.: Reducibility among combinatorial problems. Complex. Comput. Comput. 40, 85–
103 (1972)
18. Scheinerman, E.: An evolution of interval graphs. Discrete Math. 82, 287–302 (1990)
Geometric Random Graphs on Circles
Omer Angel and Yinon Spinka
Abstract Given a dense countable set in a metric space, the infinite random
geometric graph is the random graph with the given vertex set and where any
two points at distance less than 1 are connected, independently, with some fixed
probability. It has been observed by Bonato and Janssen that in some, but not all,
such settings, the resulting graph does not depend on the random choices, in the
sense that it is almost surely isomorphic to a fixed graph. While this notion makes
sense in the general context of metric spaces, previous work has been restricted to
sets in Banach spaces. We study the case when the underlying metric space is a
circle of circumference L, and find a surprising dependency of behaviour on the
rationality of L.
Keywords Rado graph · Graph isomorphism · Geometric random graphs
1 Introduction and Main Results
Erdős and Rényi initiated the systematic study of the random graph on n vertices,
in which any two vertices are connected with probability p, independently. In 1964,
Erdős and Rényi [8] showed that the infinite version of this random graph, where
there are countably many vertices and any two are independently connected with
probability p, is very different from its finite counterpart. Specifically, there exists
a fixed graph R such that the infinite random graph is almost surely isomorphic
to R. Moreover, R is the same for all p ∈ (0, 1). Rado [9] gave a concrete and
concise description of R. The graph R (or, more precisely, its isomorphism type)
is therefore sometimes called the Rado graph. The Rado graph has several nice
properties. One such property, which in fact characterizes the graph, is that it is
existentially closed: for any disjoint finite sets of vertices A and B, there exists a
O. Angel () · Y. Spinka

Department of Mathematics, University of British Columbia, Vancouver, BC, Canada
e-mail: angel@math.ubc.ca; yinon@math.ubc.ca
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 23

D. Hernández-Hernández et al. (eds.), Advances in Probability and Mathematical
Statistics, Progress in Probability 79, https://doi.org/10.1007/978-3-030-85325-9_2
24 O. Angel and Y. Spinka
vertex which is adjacent to every vertex in A and to no vertex in B. We refer the

reader to [7] for more information on the Rado graph and its properties.
The Erdős–Rényi random graph, both its finite version and its infinite version, are
non-geometric models—they are random subgraphs of a complete graph. Random
geometric graphs have been studied extensively. In these graphs, the vertices are
embedded in some previously defined metric space (X, d), and the probability of
a connection depends on the distance between the vertices. If the set of vertices is
locally finite, the structure of the resulting graph can be expected to mirror that of the
underlying metric space X. However, if the set is dense in X, a very different story
unfolds. Bonato and Janssen [2] initiated the study of random geometric graphs in
which any two points of a countable metric space that are at distance less than one
from each other are independently connected with probability p. They introduced
a property of these graphs called geometric existentially closed (g.e.c.), analogous
to the existentially closed property of the Rado graph. A graph G whose vertex set
is a metric space (V , d) is said to satisfy g.e.c. if, for any vertex s ∈ V , any disjoint
finite sets of vertices A, B ⊂ V which are contained in the open unit-ball around s,
and any > 0, there exists a vertex v ∈ V \(A∪B) which is adjacent to every vertex
in A, is not adjacent to any vertex in B, and satisfies that d(v, s) < . They then
showed that, for any countable metric space in which every point is an accumulation
point, the corresponding geometric random graph almost surely satisfies the g.e.c.
property.
A countable metric space is said to be Rado if the infinite random geometric
graph has a unique isomorphism type, i.e., if two independent samples of the
geometric random graph, possibly with distinct p, are almost surely isomorphic.
Such a space is called strongly non-Rado if two such samples are almost surely
non-isomorphic. When referring to these terms in the context of a countable subset
of a metric space, we are actually referring to the metric space induced on that set.
Thus, if S is a countable subset of a metric space (V , d), then we say that S is Rado
(strongly non-Rado) if the metric space induced on S, namely (S, d|S×S ), is Rado
(strongly non-Rado). We informally say that a metric space has the Rado property
if a typical (e.g., generic or random) dense countable subset of it is a Rado set.
To make this precise, if the metric space has some probability measure, one can
consider the set S given by an infinite i.i.d. sequence of samples from the measure.
The basic question then arises: which metric spaces have the Rado property? For
example, if the space has diameter less than 1, then any two points are connected
with probability p and the geometric random graph is nothing but the original Rado
graph.
In the case of the metric space (R, | · |), Bonato and Janssen [2] prove that the
Rado property holds: there exists a fixed graph, denoted GR(R), such that for a
generic dense countable subset of R (more precisely, for any dense countable set
having no two points at an integer distance apart) and for any p, the random graph is
almost surely isomorphic to GR(R). They also extend this to the case of the metric
space (Rd , ∞ ). Here too, there is a fixed graph GR(Rd ) for each d ≥ 1. The
graphs R, GR(R), GR(R2 ), . . . are all non-isomorphic to one another. In contrast,
they show that this is not true for the Euclidean metric space (Rd , 2 ), where every
dense countable set is strongly non-Rado [2], nor for the hexagonal norm on R2 ,
Geometric Random Graphs on Circles 25
where a randomly chosen dense countable set is almost surely strongly non-Rado
[3]. They later showed that many normed spaces fail to have the Rado property [4],
including (Rd , p ) for any 1 < p < ∞ (and also for p = 1 when d ≥ 3). Balister,
Bollobás, Gunderson, Leader and Walters [1] subsequently showed that (Rd , ∞ )
are the unique finite-dimensional normed spaces for which the Rado property holds.
In fact, in any other normed space, a generic dense countable subset is strongly
non-Rado.
Certain infinite-dimensional normed spaces (all of which are Banach spaces)
have also been considered. Bonato, Janssen and Quas [5] studied the space c of
real convergent sequences with the sup norm and the subspace c0 of sequences
converging to zero, and showed that both have the Rado property. They also showed
that Banach spaces can be recovered from the random graph in the sense that,
if two Banach spaces yield isomorphic random graphs, then the two spaces are
isometrically isomorphic. In a subsequent paper [6], the same authors considered
the space C[0, 1] of continuous functions with sup norm, and proved that the Rado
property holds for certain subspaces, including the spaces of piece-wise linear paths,
polynomials, and Brownian motion paths.
Though the notion of an infinite random geometric graph is defined in the general
context of (countable) metric spaces, we are unaware of any previous works which
deal with metric spaces other than normed spaces. One of the goals of this paper is
to investigate the random graph in such spaces, and we do so through the example
of the cycle.
Let L > 0 and consider SL := R/LZ, the circle of circumference L with its
intrinsic metric (so that, for example, the diameter of the metric space is L/2). Let
S be a dense countable subset of SL . Let GL,S be the unit-distance graph on S,
i.e., the graph whose vertex set is S and whose edge set consists of all pairs of
points in S whose distance in SL is less than 1. Given p ∈ (0, 1), let GL,S,p be a
random subgraph of GL,S obtained by retaining each edge of GL,S with probability
p, independently for different edges. See Fig. 1 for an example with a finite set S.
As usual, we say that two graphs G and G are isomorphic if there exists a
bijection ϕ from the vertex set of G to that of G such that ϕ(u) and ϕ(v) are
adjacent in G if and only if u and v are adjacent in G. In this case, we write G ∼
= G .
If L ≤ 2, then GL,S,p is easily seen to be isomorphic to the Rado graph R. Thus,
we henceforth always assume that L ≥ 2.
Our first result is concerned with distinguishing between the different metric
spaces, showing that different values of L produce non-isomorphic graphs, so that
one can recover the length L of the circle from (the isomorphism type of) the random
graph. When L = ∞, it is natural to interpret S∞ as the metric space (R, | · |).
Theorem 1.1 For any L ∈ [2, ∞], any dense countable S ⊂ SL and any p ∈
(0, 1), the cycle length L can be recovered almost surely as a measurable function
of the graph GL,S,p which is invariant to graph isomorphisms.
Remark 1.2 There is a minor delicate issue with the definition of recoverability of
L. For a graph on some countable set of vertices, which may be enumerated by N,
we have the usual product σ -algebra generated by the presence of edges. The claim
is that there exists a measurable function f from the set of graphs on vertex set N
Fig. 1 A random geometric graph in S3 with 32 equally spaced vertices
to R+ such that, for any L ∈ [2, ∞], any p ∈ (0, 1), any dense countable S ⊂ SL
and any enumeration of S, we almost surely have f (GL,S,p ) = L. Moreover, f is
invariant to relabeling the vertices of the graph. Crucially, this invariance is complete
and not just probabilistic. That is, f (G) is invariant to any permutation of the vertex
labels of G, even if the permutation is allowed to depend on which edges are present
in G. To clarify the strength of this property, consider i.i.d. sequences (Xi ) of
Bernoulli random variables. The expectation
% p can be recovered almost surely from
the law of large numbers as lim n1 i≤n Xi . However, this function is not invariant
to arbitrary permutations of the sequence if the permutation is allowed to depend
on the sequence. The reason our function is strongly invariant to relabeling is that,
for any , the set of graphs with f (G) ≤ is described as those graphs satisfying a
certain second-order proposition, which does not involve the labels.
Our second and main result is concerned with self-distinguishability, showing
that SL has the Rado property if and only if L is rational. For rational L, we say
that a set S ⊂ SL is integer-distance-free if no two points in S are at a distance
which is a multiple of m1 , where L = m is irreducible. This terminology may seem
mysterious, so we remark that a set S is integer-distance-free if and only if, starting
at any point of S and moving an integer distance along the cycle, possibly winding
multiple times around the cycle, one can never terminate at another point of S.
Theorem 1.3 Let L ≥ 2, let S, S be dense countable subsets of SL and let p, p ∈
(0, 1). Let G = GL,S,p and G = GL,S ,p be independent. Then, almost surely,
1. G ∼
G if L ∈
= / Q.
∼
2. G = G if L ∈ Q and S and S are integer-distance-free.
The theorem implies that, for rational L, a generic dense countable set is a Rado
set, whereas, for irrational L, every such set is strongly non-Rado. In the rational
case, there exist dense countable sets which are non-Rado (see Remark 5.4). In the
irrational case, we can show more—namely, that up to isometries of SL , one may
recover S from (the isomorphism type of) the random graph.
Theorem 1.4 Let L > 2 be irrational, let S, S be dense countable subsets of SL
and let p, p ∈ (0, 1). Suppose that G = GL,S,p and G = GL,S ,p can be coupled
so that G ∼
= G with positive probability. Then S and S differ by an isometry of SL .
2 Definitions and Notation
We view elements of SL as real numbers modulo L, and we sometimes identify SL

with the interval [0, L). An open arc is the image of an open interval in R modulo
L. For a, b ∈ SL , we write (a, b)SL for the positive/anti-clockwise open arc from
a to b, i.e., for (a, b) when 0 ≤ a ≤ b < L and for (a, b + L) modulo L when
0 ≤ b < a < L. Thus, (a, b)SL and (b, a)SL partition SL \ {a, b}. The length of
an open arc (a, b)SL is b − a if b ≥ a and is b − a + L if a > b. When a and b
are real numbers (not necessarily in SL ) with a − b < L, we may unambiguously
define (a, b)SL by interpreting a and b modulo L. With a slight abuse of notation,
we simply write (a, b) for (a, b)SL . Closed arcs and half-open-half-closed arcs are
similarly defined. The distance between two points u, v ∈ SL , denoted u − v, is
the length of the shorter of the two arcs between u to v and can be written as
u − v = min{|u − v|, L − |u − v|}.
Given a graph G, we write N(v) for the neighbourhood of a vertex v in G, and we

write distG (u, v) for the graph-distance between vertices u and v in G. The length
of a path in G is the number of edges in the path, so that distG (u, v) is the length of
a shortest path between u and v.
A graph G whose vertex set is a subset S of SL is called g.e.c. (geometrically
existentially closed) if, for any vertex s ∈ S, any disjoint finite sets A, B ⊂ S which
are contained in (s − 1, s + 1), and any > 0, there exists a vertex v ∈ S \ (A ∪ B)
which is adjacent to every vertex in A, is not adjacent to any vertex in B, and satisfies
that v − s < . The graph G is said to have unit threshold if any two adjacent
vertices u, v ∈ S satisfy that u − v < 1. The notions of g.e.c. and unit threshold
exist in the general context of a graph whose vertex set is a metric space.
For a dense countable subset S of SL , let GL,S denote the collection of all g.e.c.
graphs on S having unit threshold. Let GL denote the union of GL,S over all such S.
As it turns out, once we establish the following simple property that GL,S,p belongs
to GL,S , almost all our arguments become completely deterministic.
Lemma 2.1 Let L ∈ [2, ∞], let S be a dense countable subset of SL and let p ∈
(0, 1). Then, almost surely, GL,S,p ∈ GL,S .
Proof It is trivial from the definition that GL,S,p almost surely has unit threshold.
Fix a vertex s ∈ S, disjoint finite sets A, B ⊂ S which are contained in the open
unit-ball around s, and a rational 0 < < 1 − maxu∈A∪B u − s. Since S is
countable, it suffices to show that, almost surely, there exists a vertex v ∈ / A∪B
which is adjacent to every vertex in A, is not adjacent to any vertex in B, and satisfies
that v − s < . Since the open ball of radius around s contains infinitely many
points v which are not in A ∪ B, and since the sets of edges incident to these v are
independent, it suffices to show that each such v has a positive (fixed) probability to
be adjacent to all of A and to none of B. Indeed, since any such v has u − v < 1
for all u ∈ A ∪ B, this event has probability p|A| (1 − p)|B| .
3 Distinguishing Graphs Arising from Different L
In this section, we prove Theorem 1.1. Our strategy is to introduce a graph-theoretic

quantity which allows to differentiate between graphs arising from different L.
For a graph G, define λ(G) to be the supremum over λ ≥ 0 such that for
every finite set of vertices U ⊂ G, there exists a vertex in G having at least λ|U |
neighbours in U . Thus,
|N(v) ∩ U |
λ(G) := inf sup .
U ⊂G v∈G |U |
0<|U |<∞
Consider a graph G ∈ GL,S . It is easy to check that λ(G) = 1 if L ≤ 2, since

G is just the Rado graph. If L = ∞, then λ(G) = 0, since S contains arbitrarily
large finite sets U such that all vertices of U are at distance more than 2 from each
other, so that |N(v) ∩ U | ≤ 1 for all v. In fact, as we now show, λ(G) depends
on G only through L, and moreover, is equal to 2/L. Theorem 1.1 is an immediate
consequence of Lemma 2.1 and the following.
Proposition 3.1 Let L ∈ [2, ∞] and G ∈ GL . Then λ(G) = 2/L.
Proof Let L ∈ [2, ∞] and G ∈ GL . By definition of GL , the vertex set of G is a
dense countable subset S of SL . For a finite U ⊂ S and an arc A ⊂ SL , we call
|U ∩ A|/|U | the density of U in A.
We begin by proving the upper bound λ(G) ≤ 2/L. Since G has unit threshold,
for any v, N(v) is contained in an arc of length 2. Thus it suffices to exhibit, for
any > 0, a finite set U ⊂ S whose density is no more than 2/L + in any
arc of length 2. Any set that is close to evenly distributed on the cycle will do.
For completeness, here is one construction: Let n be a large integer and consider
the set V consisting of points v0 , . . . , vn−1 ∈ SL , where vi := iL/n. Since S is
dense, there exist a finite set U consisting of points u0 , . . . , un−1 ∈ S such that
ui − vi < 1/n. It is straightforward to verify that, for any arc A of length r, we
have that |U ∩ A| ≤ |V ∩ A| + 2 ≤ rn/L + 3. Thus, U has density at most
2/L + 3/n in any arc of length 2.
We now turn to the lower bound λ(G) ≥ 2/L. To show this, we show that the
situation described above is essentially the worst case. Precisely, given a finite U ⊂
S, we claim that there exists an open arc A of length 2 in which U has density at least
2/L. This is easy to verify, since if x is uniform in SL , then the expected number of
points of U in the arc (x − 1, x + 1) is (2/L)|U |, so for some x it is at least that
large. Since S is dense, and G is g.e.c., the arc A contains a vertex v ∈ S which is
adjacent to all vertices of U in A. This proves the lower bound λ(G) ≥ 2/L.
4 Recovering Distances and Non-isomorphism for

Irrational L
In this section, we prove part (1) of Theorem 1.3, namely that for irrational L > 2,
the independent graphs GL,S,p and GL,S ,p are almost surely non-isomorphic. The
key step is to show that by looking at the graph-structure of G alone, it is possible
to determine the distance in SL between any two vertices.
Throughout this section, we fix L ∈ [2, ∞] and assume G ∈ GL,S for some
dense countable S ⊂ SL . It is easy to check that, for any two vertices u, v ∈ S and
any integer k ≥ 2, we have
u − v < k ⇐⇒ distG (u, v) ≤ k.
Indeed, if u − v < k, then for some ε > 0, there is a path u = x0 , x1 , . . . , xk = v

with xi −xi−1 < 1−ε. By g.e.c. there is a perturbation xi of xi for i = 1, . . . , k−1
so that (xi , xi−1
) are edges of G for all i = 1, . . . , k. Conversely, the unit-threshold
property shows that a path of length at most k in G from u to v implies that the cycle
distance is less than k. Note that for k = 1 this equivalence fails, since {u, v} may
or may not be an edge of G. See [2, Theorem 2.4] for a similar statement.
We may rewrite the above as

u − v + 1 if u − v ≥ 1
distG (u, v) = . (1)
1 or 2 if u − v < 1
Thus, distances in G are predetermined for points of S which are at distance at

least 1 in SL . However, we are more interested in the other direction of implication:
forgetting that the vertices of G are labelled by elements of SL and looking only at
the graph structure of G in relation to (u, v), one may recover u − v, unless it
is 0 or 1.
To formalize these types of ideas, we require some definitions. A graph with k
distinguished vertices is a graph G, together with an ordered k-tuple (v1 , . . . , vk ) of
distinct vertices of G. Let GL,•,• denote the collection of all graphs in GL with
two distinguished vertices. Let π denote the projection from GL,•,• to the class
G•,• of isomorphism classes of graphs with two distinguished vertices. The above
may be restated as saying that the function (G, u, v) → u − v1{u−v≥2} from
GL,•,• to Z can be written as a composition of a function from G•,• to Z with π.
Indeed, (1) gives a concrete such description, since the graph-distance between the
distinguished vertices is invariant to isomorphisms of the graph. In this case, we say
that u − v1{u−v≥2} can be recovered from the graph structure of (G, u, v).
More generally, we say that a function f : GL,•,• → can be recovered from
the graph structure of (G, u, v) if f = F ◦π for some F : G•,• → . For brevity, we
say that f (G, u, v) can be recovered from π(G, u, v). We extend these definitions to
graphs with k distinguished points, writing π also for the projection from GL,•,...,•
to the corresponding set of isomorphism classes of graphs with k distinguished
vertices. We shall also talk about sets of vertices being recoverable from the graph.
For example, Lemma 4.4 below says that the set of vertices in the shorter arc
between u and v along the cycle is recoverable. Formally, a set A = A(G, u, v)
of vertices of G can be recovered from π(G, u, v), if the function 1{x∈A} can be
recovered from π(G, u, v, x) for any x ∈ G.
The main ingredient in this section is the following proposition, which shows
that we can recover plenty of information on the distances in SL from the graph
structure (for both rational and irrational L). Given this, part (1) of Theorem 1.3 is
easily deduced.
Proposition 4.1 Let L > 2, let G ∈ GL and let u, v ∈ G be adjacent. Then the
sequence
(u − v + kL)k≥1
can be recovered from π(G, u, v).

The values u − v + kL can be thought of as the distance from u to v, moving
k additional times around the cycle instead of directly. The assumption that u, v are
adjacent in G can be removed from this proposition, but it simplifies the proof and
does not significantly impact the application for the proof of Theorem 1.3. In the
case of irrational L, this gives the following, stronger corollary, which immediately
implies the more general result.
Corollary 4.2 Let L > 2 be irrational, let G ∈ GL and let u, v ∈ G. Then u − v
can be recovered from π(G, u, v).
Proof Consider first the case when u and v are adjacent in G, so that Proposition 4.1
applies. It suffices to see that the mapping x → (x +kL)k≥1 is injective on [0, L).
It is a well known fact that for irrational L the fractional parts (kL−kL) are dense
in [0, 1]. Let 0 ≤ x < y < L. Since the fractional parts are dense, it follows that for
some k we have x + kL = y + kL, and the sequences differ.
For any path in G, we can therefore recover from G the total length of the
edges along SL . If u and v are not adjacent, then there is a path in G from u to
v which moves around SL in the shorter direction without backtracking. Since we
can recover the cycle distance in each edge of the path, the sum is the distance from
u to v. Any other path in G must give a larger sum. Thus we can recover u − v as
the minimal sum of cycle distances along paths from u to v in the graph.
Since the cycle distance between any two vertices can be recovered from the
graph, we have the following.
Corollary 4.3 Let L > 2 be irrational, let S, S be dense countable subsets of SL
and G ∈ GL,S , G ∈ GL,S . If f : S → S is a graph isomorphism between G and
G , then it is an isometry between S and S .
Corollary 4.3 immediately implies Theorem 1.4. We now prove part (1) of
Theorem 1.3.
Proof of Theorem 1.3(1) Let G = GL,S,p and G = GL,S ,p be independent, as in
the statement of the theorem. Consider a bijection f : S → S . By Corollary 4.3,
if f is not an isometry between S and S , then it is not an isomorphism between
G and G . Thus it suffices to consider isometries f . There are at most countably
many isometries between S and S (for an arbitrary v0 ∈ S, there are at most two
isometries for any given choice of f (v0 )). Since any fixed isometry f is almost
surely not an isomorphism between G and G , we conclude that there almost surely
does not exist an isomorphism between G and G .
4.1 Proof of Proposition 4.1
The overall strategy for the proof of Proposition 4.1 is as follows: we define a
graph-theoretic notion of a cyclic ordering of vertices. This notion, though defined
completely in terms of the graph G, will be such that it guarantees that the
corresponding points in SL are cyclically ordered as well. This will allow to define
another graph-theoretic notion of a uni-directional path in G, which will correspond
to a path in SL that winds around the circle in a fixed direction. We then show
that any uni-directional path in G has a well-defined (again, in terms of the graph)
winding number which counts the number of times the corresponding path in SL
winds around the circle. Finally, using this we deduce that from the graph G, for any
two adjacent vertices u, v ∈ G, we may recover the sequence (u − v + kL)k≥1 ,
which is Proposition 4.1.
Fix L > 2, a dense countable subset S ⊂ SL and a graph G ∈ GL,S . For x, y ∈ S

having x − y < 1, let Ax,y denote the set of points of S in the shorter arc of [x, y]
and [y, x]. It is convenient to include the endpoints x, y in the arc.
The starting point of our argument is the following lemma which shows that
the shortest arc between two adjacent vertices can in fact be described as a graph-
theoretic property. Its proof is postponed to the end of the section.
Lemma 4.4 Let a, b ∈ G be adjacent. Then Aa,b can be recovered from π(G, a, b).
We say that a triplet (a, b, c) of distinct points in G is cyclically ordered in SL
if Aa,b ∩ Ab,c = {b}. This means that when moving from a to b to c along the cycle
(always via the shorter arc), the direction of movement is maintained. We say that
a path p = (v0 , . . . , vn ) in G consisting of distinct vertices is a uni-directional if
the triplet (vi−1 , vi , vi+1 ) is cyclically ordered in SL for each 1 ≤ i ≤ n − 1. Thus,
by going along the shorter arc between consecutive vertices, a uni-directional path
in G may be thought to correspond to a continuous path in SL that always winds
around the circle in a single direction. In light of Lemma 4.4, this property can be
determined from the graph structure of G, so that we may talk about the path being
uni-directional in G. The winding number of a uni-directional path p is defined
as the number of complete revolutions its continuous counterpart makes around the
cycle—if its total cycle-length is , this is /L. The winding number can also
be calculated as the number of indices 1 ≤ i ≤ n − 1 such that v0 ∈ Avi ,vi+1 .
Consequently, the winding number of a uni-directional path p can be recovered
from π(G, v0 , . . . , vn ).
It will also be useful to be able to identify the direction in which a uni-directional
path winds around the circle; by this we do not mean the absolute direction
(clockwise/anticlockwise), but rather whether it goes from the start point to the
end point by starting through the short/long arc between them. This can be done
in one of several ways. We choose here a simple definition, which comes at the cost
of requiring the start and end points to be at distance less than 1. For u, v ∈ S, a
good path from u to v is a uni-directional path p = (x0 , x1 , . . . , xn ) in G such that
x0 = u, xn = v and v ∈ Au,x1 . Thus, a good path is required to go towards v in the
shorter direction, and overshoot v in its first step. In particular, its winding number
is at least 1. Of course, this is only possible if u − v < 1. The following shows
that good paths exist.
Lemma 4.5 Let k ≥ 1 and u, v ∈ G be such that u − v < 1. Then G contains a
good path from u to v with winding number k and length n = u − v + kL + 1.
Moreover, there is no good path from u to v with winding number k and length less
than n.
Proof For concreteness, we assume that the short arc from u to v is in the positive
direction, so that v = u + u − v. Set := u − v, so that we seek a path of length
n = + kL + 1. We start by specifying approximate locations for the points of the
path. These approximate locations, denoted xi , are points of the circle and need not
be vertices of the graph. Let x1 = u + t for some t ∈ ( , 1), so that v ∈ Au,x1 and
x1 − u < 1. The total cycle-length of the path will be + kL, and the remaining
points (xi )n−1
i=2 will be equally spaced with gap
+ kL − t
= .
n−1
Thus, the approximate points are xi = u + t + (i − 1) for 1 ≤ i ≤ n − 1. Note

that for t close enough to 1, we have < 1, since + kL < n.
Fix ε > 0 such that max{t, } < 1 − 2ε, and set U0 := {u}, Un := {v} and
Ui := S ∩ (xi − ε, xi + ε) for 1 ≤ i ≤ n − 1. This choice guarantees that any
point in Ui is at distance less than 1 from any point in Ui−1 , and the shorter arc
between them is positively oriented. Since G is g.e.c., there exists u1 ∈ U1 such that
u1 is adjacent to u0 := u in G. Continuing by induction, we see that there exists
a sequence (ui )0≤i≤n−2 such that ui ∈ Ui \ {v, u0 , . . . , ui−1 } and ui is adjacent to
ui−1 in G. Finally, there exists un−1 ∈ Un−1 \ {v, u0 , . . . , un−2 } which is adjacent
to both un−2 and un := v. By construction, (u0 , . . . , un ) is a good path in G from u
to v of length n and winding number k.
Finally, there can be no uni-directional path (good or otherwise) from u to v with
winding number k and length at most + kL, since the total length of the arcs in
such a path is smaller than + kL.
We are now ready to prove Proposition 4.1.
Proof of Proposition 4.1 Let u, v ∈ G be adjacent and fix k ≥ 1. Our goal is to
recover k := u − v + kL from π(G, u, v). By Lemma 4.5, k + 1 is the
shortest length of a good path from u to v with winding number k. Since whether a
path (x0 , . . . , xn ) is good can be determined from π(G, x0 , . . . , xn ), this shows that
k can be recovered from π(G, u, v).
4.2 Proof of Lemma 4.4
Let a, b ∈ G be adjacent. Recall that Aa,b is the set of points of S in the shorter arc
of [a, b] or [b, a]. As a warm-up, the reader may find it instructive to note that, when
L ≥ 5, one may easily recover Aa,b as the intersection over v ∈ S of all intervals
(v − 2, v + 2) ∩ S which contain {a, b}. By (1), each such interval can be recovered
as the set of vertices at graph-distance at most 2 from v.
When L ≥ 3, one may try a similar approach, recovering Aa,b as the intersection
over v ∈ S of all intervals (v − 1, v + 1) ∩ S which contain {a, b}. Indeed, it is not
hard to check that this produces the correct set, however, as (1) does not allow to
recover intervals of the form (v − 1, v + 1) ∩ S, we will need to slightly modify the
approach.
The proof is split into two cases according to whether L ≥ 3 or 2 < L < 3. The
following lemma will be useful in both cases. Say that a set U ⊂ G is small if it is
finite and some vertex of G is adjacent to every vertex in U . Say that U is large if
it is finite and not small, i.e., no vertex of G is adjacent to every vertex in U . The
following simple lemma shows that a finite set is small if and only if it is contained
in some open arc of length 2. Equivalently, a finite set is large if and only if it leaves
no gap of length greater than L − 2 in its complement.
Lemma 4.6 Let U ⊂ G be finite. Then U is small if and only if U ⊂ (v − 1, v + 1)
for some v ∈ G.
Proof Suppose first that U is small. By definition, there exists a vertex w ∈ G that
is adjacent to every vertex in U . Since G has unit threshold, u − w < 1 for all
u ∈ U . Thus, U ⊂ (w − 1, w + 1), as required.
Suppose now that U ⊂ (v − 1, v + 1) for some v ∈ G. Since G is g.e.c., there
exists a vertex w ∈ G which is adjacent to all of U . Thus, U is small, as required.

4.2.1 Proof of Lemma 4.4 when L ≥ 3
Step 1 Let (a, b, c, d) be a path in G and suppose that {a, b, c, d} is large. Then
Ab,c can be recovered from π(G, a, b, c, d).
We first show that such a path (a, b, c, d) must be uni-directional. Indeed, if the
arcs (a, b) and (b, c) are in opposite directions in SL , then {a, b, c, d} ⊂ (c − 1, c +
1), and so the set is small. Similarly, if (b, c) and (c, d) are in opposite directions,
then {a, b, c, d} ⊂ (b − 1, b + 1). This shows that {a, b, c, d} are distinct vertices
and that Aa,b ∩ Ab,c = {b} and Ab,c ∩ Ac,d = {c}.
For a finite U ⊂ G, we denote
C(U ) := {w ∈ G : U ∪ {w} is small}.
We will show that Aa,b ∪ Ab,c is precisely the set
W := {w ∈ G : C({a, b, c}) = C({a, b, c, w})}.
Since Ab,c ∪ Ac,d is similarly obtained, this will show that we can determine Ab,c
as
Ab,c = (Aa,b ∪ Ab,c ) ∩ (Ab,c ∪ Ac,d ).
Before showing that Aa,b ∪ Ab,c = W , we first observe that any interval (v −
1, v+1) containing {a, b, c}, contains Aa,b ∪Ab,c and does not contain d. Indeed, no
such interval contains {a, b, c, d} since {a, b, c, d} is large. Suppose that (a, b, c, d)
winds in the positive direction (the other case being similar). Since L ≥ 3, the circle
is a disjoint union of the arcs [a, b), [b, c), [c, d), and [d, a). Thus any interval in
the circle that contains a, b, c but not d also contains [a, b] ∪ [b, c]. Since [a, b] is
the short arc between a, b, we have Aa,b = [a, b] ∩ S, and similarly for Ab,c . Thus
any such interval (v − 1, v + 1) must contain Aa,b ∪ Ab,c .
To see that Aa,b ∪ Ab,c ⊂ W , fix w ∈ Aa,b ∪ Ab,c and let us show that
C({a, b, c}) = C({a, b, c, w}). The containment C({a, b, c, w}) ⊂ C({a, b, c})
is clear, and the opposite containment follows from the fact that any interval
(v − 1, v + 1) containing {a, b, c} also contains Aa,b ∪ Ab,c .
To see that W ⊂ Aa,b ∪Ab,c , let w ∈ W and suppose towards a contradiction that
w∈ / Aa,b ∪ Ab,c . In order to reach a contradiction with the fact that C({a, b, c}) =
C({a, b, c, w}), it suffices to find a vertex of G which belongs to an interval (v −
1, v + 1) containing {a, b, c}, but does not belong to any interval (v − 1, v + 1)
containing {a, b, c, w}.
Recall that (a, b, c, d) is a uni-directional path, and note that one of
(a, b, c, d, w) or (a, b, c, w, d) is also a uni-directional path. Suppose for
concreteness that it is the latter, and that moreover, (a, b, c, w, d) winds around the
cycle in the positive direction. In particular, Aa,b = [a, b] ∩ S and Ab,c = [b, c] ∩ S.
Moreover, A := [c, w] ∩ S is the arc between c and w, which does not contain a,
b or d (it is not necessarily the short arc between c and w). Note that, since any
interval (v − 1, v + 1) containing {a, b, c} must contain Aa,b ∪ Ab,c and cannot
contain d, it follows that any interval (v − 1, v + 1) containing {a, b, c, w} must
contain Aa,b ∪ Ab,c ∪ A = [a, w] ∩ S. Observe also that any such interval is
contained in (w − 2, a + 2). However, if v ∈ (c − 1, w − 1), then the interval
(v − 1, v + 1) contains {a, b, c}, but is not contained in (w − 2, a + 2) (note that
the latter is not all of SL since (a, w) is longer than (a, c), which in turn has length
more than 1). We have thus reached a contradiction.
Step 2 Let a, b ∈ G be adjacent. Then Aa,b can be recovered from π(G, a, b).
In light of the previous step, it suffices to show that there exists a path (x, a, b, y)
so that {x, a, b, y} is large. Denote := a − b and suppose they are oriented so
that b = a + . There exists x ∈ (a − 1, a − 1 + /2) adjacent to a, and similarly,
there exists y ∈ (b + 1 − /2, b + 1) adjacent to b. Since L ≥ 3, the only way
{x, a, b, y} is contained in an open arc of length 2 is if y − x < 2, which is not the
case by our choice of x and y.
4.2.2 Proof of Lemma 4.4 when 2 < L < 3
We write L = 2 + δ for some δ ∈ (0, 1). The reader may keep in mind that small δ
is the more difficult case.
Step 1 Let u, v ∈ G. Then 1{u−v<δ} can be recovered from π(G, u, v).
This will follow if we show that
u − v < δ ⇐⇒ U ∪ {u} and U ∪ {v} are large for some small U.
Suppose first that U ∪ {u} and U ∪ {v} are large for some small U . Let us show
that u − v < δ. Let {u− , u+ } be the two vertices of U nearest to u from either
side. Recall that a finite set is large if and only if it leaves no gap of length greater
than δ = L − 2 in its complement. Therefore, since U ∪ {u} is large, (u− , u) and
(u, u+ ) each has arc-length at most δ. Since U ∪ {v} is large, but U is small, it must
be that v ∈ (u− , u+ ) so that u − v < δ, as required.
Suppose next that u − v < δ. Let us show that U ∪ {u} and U ∪ {v} are large
for some small U . Assume without loss of generality that u ∈ (v, v + δ), and let
0 < ε < (δ − u − v)/3. Let V be the arc (v + δ − ε, v − 2ε). Note that v ∈ / V and
|V | = L − (δ + ) = 2 − ε < 2. Recall that a finite set is small if and only if it is
contained in some open arc of length 2. Thus, any finite subset of V is small. Since
(v − 2ε, v) and (v, v + δ − ε) each has arc-length less than δ, it follows that U1 ∪ {v}
is large for some finite U1 ⊂ V . Similarly, since (v − 2ε, u) and (u, v + δ − ε) each
has arc-length less than δ, we have that U2 ∪ {u} is large for some finite U2 ⊂ V .
Thus, U := U1 ∪ U2 is a small set such that both U ∪ {v} and U ∪ {u} are large, as
required.
Step 2 Let a, b ∈ G satisfy a − b < δ. Then Aa,b can be recovered from
π(G, a, b).
By the first step, Av := (v − δ, v + δ) ∩ S can be recovered from π(G, v). Let
W be the intersection of all Av containing {a, b}. We claim that Aa,b = W .
To see that Aa,b ⊂ W , we must show that Av contains Aa,b whenever it contains
{a, b}. Indeed, if {a, b} ⊂ Av then, since Av has arc-length 2δ, a − b < δ and
3δ < L, it follows that Aa,b ⊂ Av .
To see that W ⊂ Aa,b , we must show that for any u ∈ G\Aa,b there exists v ∈ G
such that {a, b} ⊂ Av and u ∈ / Av . Since 2δ < L, it is straightforward that such a v
exists.
Step 3 Let a, b ∈ G be adjacent. Then Aa,b can be recovered from π(G, a, b).
Set n := 1/δ, so that 1 ≤ δn < 1 + δ. Since a − b < 1 ≤ δn, g.e.c. implies
that there exists a uni-directional path P = (u0 , . . . , un ) in G such that u0 = a,
un = b, and ui − ui−1 < δ for all i. Since the long arc from a to b has length
larger than 1 + δ, and since the total cycle-length of P is at most δn < 1 + δ, it must
be that Aa,b = Au0 ,u1 ∪ · · · ∪ Aun−1 ,un . Thus, by the previous steps, one can recover
Aa,b from π(G, a, b).
5 Constructing an Isomorphism for Rational L
In this section, we prove part (2) of Theorem 1.3, namely, that GL,S,p and GL,S ,p
are almost surely isomorphic when L is rational and S and S are integer-distance-
free. In light of Lemma 2.1, this is an immediate consequence of the following. Let
GL,idf ⊂ GL consist of those graphs whose vertex sets are integer-distance-free.
Proposition 5.1 Let L > 2 be rational and let G, G ∈ GL,idf. Then G ∼
= G .
Bonato and Janssen proved the analogous statement for the case L = ∞
(corresponding to the real line) [2, Theorem 3.3]. Our construction follows similar
lines.
Let G ∈ GL,S and G ∈ GL,S , where S and S are integer-distance-free. Our goal
is to construct an isomorphism f between G and G . In fact, we have the flexibility
to map one vertex of G to an arbitrary vertex of G . Thus, by translating S and S if
necessary, we may assume without loss of generality that both S and S contain 0,
and we aim to construct an isomorphism f : S → S such that f (0) = 0.
For a point v ∈ SL , consider the sequence (v + kL)k≥0 . We have seen in the
previous section that a similar sequence can be recovered from π(G, v, 0). Thus,
when trying to construct an isomorphism between G and G , we shall want to
preserve this sequence. That is, we will always map a vertex v ∈ S to a vertex v ∈ S
in such a way that (v + kL)k≥0 = (v + kL)k≥0 . A key observation is that, since
L = /m is rational, (v + kL)k≥0 is determined by its first m elements. More
precisely, the sequence is periodic after subtracting the fixed sequence (kL)k≥0 .
Since there are only finitely many possibilities for the sequence, there are many
candidates for such a v ∈ S .
To be more precise, let
L=
m
be irreducible. Then there are precisely possibilities for the sequence (v +
kL)k≥0 , according to the value of mv. We thus partition SL into the arcs
[ mi , i+1
m ) for 0 ≤ i ≤ − 1.
j j +1
Note that two points u ∈ [ mi , i+1
m ) and v ∈ [ m , m ) satisfy (u + kL)k≥0 =
(v + kL)k≥0 if and only if i = j . We henceforth let qv ∈ {0, . . . , − 1} and
rv ∈ [0, m1 ) be the unique numbers such that
qv
v= + rv .
m
Note that qv = mv and v ∈ [ qmv , qvm+1 ). Moreover, since S is integer-distance-free,

rv = ru for distinct u, v ∈ S, and in particular rv = 0 for v = 0 in S, and similarly
for S .
Let S̄ ⊂ S and S̄ ⊂ S , and suppose that both contain 0. A bijection f : S̄ → S̄
is called an extended step-isometry if f (0) = 0 and, for every u, v ∈ S̄, we have
qu = qf (u) and ru < rv ⇐⇒ rf (u) < rf (v).
Though we will not require it, we note that such an extended step-isometry f
satisfies that
m · u − v = m · f (u) − f (v) for all u, v ∈ S̄.

Another random document with
no related content on Scribd:
The Project Gutenberg eBook of Then luck came
in
This ebook is for the use of anyone anywhere in the United States
and most other parts of the world at no cost and with almost no
restrictions whatsoever. You may copy it, give it away or re-use it
under the terms of the Project Gutenberg License included with this
ebook or online at www.gutenberg.org. If you are not located in the
United States, you will have to check the laws of the country where
you are located before using this eBook.
Title: Then luck came in
Author: Andrew A. Caffrey
Release date: May 3, 2024 [eBook #73525]
Language: English
Original publication: New York, NY: The Butterick Publishing

Company, 1928
Credits: Roger Frank and Sue Clark
*** START OF THE PROJECT GUTENBERG EBOOK THEN LUCK

CAME IN ***
THEN LUCK CAME IN
An Aviation Sergeant Who Yearned To Fly
By Andrew A. Caffrey
The sergeant was a much abused man. Wartime flying had not used
him any too well; nor had after the war aviation done any better. Now
he was nearing the end of his Army career.
The sergeant had wanted to fly. He wanted to go solo and do his
own birding. It had always been his one ambition. And it was through
no fault of his own that the big desire had never been fully realized.
Fact is, along those lines the much abused sergeant was without
fault. He had always done his share.
The sergeant was too willing in 1917. Later—too late—he realized
this. Had he held off, as the other millions did, and waited for the war
to get at good speed, he would have made his way into a ground
school and started right. But the sergeant did not know that there
were to be such schools. None knew this. So the sergeant enlisted.
Willingly the aviation branch of the Signal Corps took him. Oh, yes,
of course, they said he would fly.
But the sergeant turned out to be a handy mechanic. Good
mechanics were few—and are still—so the sergeant, though he
didn’t guess it, was never going to get to fly.
On the other side of the pond his bad luck continued. That was
when they made him a sergeant, made him a sergeant, chief
airplane rigger, while they made flying cadets of the goldbricks in his
squadron. That hurt—hurt like—well, it hurt.
“But look here, Sergeant,” his commanding officer said in rebuttal,
“now let’s be reasonable; it takes years to make a good mechanic.
And only hours to lache a full fledged pilot; and the stuff of which
airmen are made need not know anything—or much. See the point?
You’re important on this field; these other birds going out as cadets
are, as a rule, culls we’re glad to be rid of. Now get back to your
hangar and feel satisfied that you are doing your bit, and a hell of a
big bit, Sergeant!”
That line of official chatter did not help the sergeant at all.
“I’ve heard it before,” he told his rigging crews. “Doing my bit! Bit
be damned! The effect of my first patriotic drunk has worn off. What I
want to do is fly and I’m going to!”
The sergeant did learn to fly; but he “stole” the flying time, begged
all the dual control instruction he could mooch and waxed mighty
handy on rudder bar and stick. And he learned quickly. You see, like
many other mechanics, he really knew how to fly before he ever had
a ship in his hands. Once in the air he merely had to gain the feel of
the thing. And he got it too. He made a takeoff on the third hop,
landed on his fifth.
His job was on a pursuit field—all single seater planes. The ship
on which he had learned—a Nieuport 23—was a two place visitor.
He was all set to fly alone. Then, that same day, they took the 23
away. The sergeant saw red, and spoke in the same color.
“Cheated again!” he said. “I’m going into town, get all drunked up
and take an M.P. apart! Wait and see!”
You can not get the sergeant’s point of view unless you have loved
air and wanted to fly. But if you had loved air and wanted to fly, you
would have gone to town with him and helped take a flock of M.P’s
apart.
Unofficially grabbing flying time wherever and whenever he could
get any, the sergeant lived in hopeless hope, if such a thing exists.
But our war lasted only a day; and once gone it was gone forever.
The sergeant’s field did not go directly out of business, with the
coming of the Armistice, but his interest in things did. For him it was
the end of everything—and nothing.
Then, with the idea of training more pilots for future wars,
headquarters sent the sergeant’s squadron on to an Avro, two place,
training field. The sergeant’s interest came back. He stole lots of
time, loved Avros and added acrobatics to his straight flying. The war
after the war was treating him better.
New made flying cadets came to that field. Lord! Where did they
get such dubs? The sergeant wondered. From every orderly room at
the center was the answer. It was a dog robbers’ holiday.
“I’ll get the C.O.’s permission to turn you loose, Sergeant,” an
instructor said. “You can fly rings round any bird in this group. I’ll get
papers through for you too; no reason why you shouldn’t get a
brevet. I understand that they’ve handed commissions to a few 31st
men.”
The sergeant said that they had.
For a night, life couldn’t be improved upon.
Next morning, February 12, headquarters “washed out” all flying
and called in the Avros. They say that the sergeant took a lieutenant
of M.P’s apart at high noon of the same day in the public square at
Issoudun. After that, for him, the world fused.
The sergeant’s outfit came back to the States. Air Service wanted
to hold some of its best mechanics. At Mitchel Field they promised
the sergeant and some of his gang that, were they to reenlist for
another stretch, flying would be their dish for sure.
The sergeant took his discharge. Then he was tempted—and fell.
He put up his hand for another hitch. And headquarters shipped him
to Carlstrom Field, Florida.
New classes of cadets came to that field. Even one of the cooks
from the sergeant’s overseas squadron was among them. They were
the worst cadets the sergeant ever saw. But he worked planes for
them; and in turn, headquarters never did put the sergeant on flying
status. But the much abused one continued to mooch some
unofficial airwork. So the months of his one year enlistment dragged
by and he came toward the happy end, the end which was going to
be so welcome because he did not give a good, bad or indifferent
damn. And he told his C.O. as much when that worthy asked him
whether he intended to sign up for a third cruise.
“You’re not talking to me, Lieutenant,” the sergeant said. “For three
years I’ve lived on hope. When I took on this reenlistment, they
promised me, on a stack of Bibles, that I’d fly. And have I?”
Any number of ex-overseas men could answer this.
“But this time you will,” the lieutenant said. “This school has the
ships and men now, and I’ll promise you—”
“Tie that outside, Lieutenant,” the sergeant answered, “I’ve heard it
all before.
“By this time next Monday afternoon, America will have one more
civilian on her hands. And she’s going to collect a mean problem,
too. I’m sore, Lieutenant. I’ve been cheated too often to smile and
turn the other cheek. This deal I’ve had handed me by Air Service
smells like a eucalyptus kitty— See that guy climbing into that rear
cockpit—” the sergeant pointed to a plane at the deadline—“well,
that same jaybird used to be a bum cook in my outfit overseas.
Shane’s his name. All that feller ever did for American honor was lap
up French booze and make trouble. He was our ace of aces at it,
too. Shane and me, Lieutenant, have been two different kinds of
soldiers, but today he’s getting in official flying time and I’m still
begging rides like a raw John Recruit. Where’s your damn’ justice in
that? I’ll answer—out for lunch with two rags around her eyes! Me,
reenlist? In a pig’s eye! Wonder what’s wrong with that plane.”
The plane into which they had watched Cadet Shane climb had
started for a takeoff, bounced into the air, fluttered a few rods and
dropped again for a hasty landing. It taxied back to where they were
standing. It was one of the sergeant’s ships. At the deadline the
instructor, Lieutenant Black, swung from his front cockpit, removed
his goggles and said:
“Wish you’d look this ship over, Sergeant. The controls jam in the
air. Bob Watts was flying it this morning and he had the same
trouble.”
“I’ll work her over,” the sergeant promised. He looked at his watch.
“Four o’clock now,” he said. “You won’t want to fly any more today,
Lieutenant. She’ll be jake in the morning.”
“That’s O.K. with me, Sergeant,” Lieutenant Black agreed, and
walked away with the sergeant’s C.O.
Cadet Shane was sore. He had been robbed of his afternoon
period and did not care who knew that he was burned up.
“Damn’ funny you guys can’t keep ships in condition,” he said. “I
haven’t had two hours’ airwork outa this hangar in two weeks.”
“Too damn’ bad about you, Shane,” was all the sympathy the
sergeant extended. “If you’re as rotten a flyer as you were a cook,
the field will be the winner if you never fly.”
For the next hour the sergeant, with a helper, worked the ship that
went wrong in the air. At the end of said time he had located nothing
wrong with the controls. Bob Watts came along during operations
and told his story. Then, just to be on the safe side, the sergeant
sent for the field inspector, Blackie Milander. He came along and
demanded—
“Wot’s eatin’ you, kid?”
“This crate, Blackie, was turned in because her controls froze in
the air,” the sergeant said. “I’ve looked her over, and my fair haired
helper here has looked her over, and Lieutenant Watts was on hand
and had his say and look, and we find nothing wrong. The control
cables, all of ’em, are O.K. Not a fray on any of them. The ball socket
joint is jake; and the pulleys are free. Now, you give her the expert
eye, Blackie, and say what’s to be done. Gladly we pass the buck to
you and, if failing, you muff the torch thus thrown, well you’ll get
burnt.”
Blackie, working till long after retreat, scratched his head finally
and announced:
“Damned if she ain’t got me stopped! On the ground here,
everything’s free. D’you know what I think, Sergeant?”
“If a thought there be, Blackie, shoot before it burns you out. What
do you guess?”
“I think that Watts and Black are full of hop! There’s nothing wrong
with this pile of wreckage, and I’ll give her a clear bill. Let me O.K.
that flying sheet.”
When the hangars opened in the morning the sergeant’s C.O. was
at hand.
“What did you learn about that plane of Black’s?” he wanted to
know. “Anything haywire?”
“Not a thing, Lieutenant,” the sergeant admitted. “What say if you
and I give it a hop right now? See if we can locate any ‘bugs’ in the
air.”
“We’ll do that little thing,” the C.O. agreed. “Got a helmet and
goggles I can use?”
While the C.O. waited, and the men started the plane’s motor, the
squadron clerk came to the hangar for the C.O. They talked for a few
minutes, then the C.O. told the sergeant:
“I’ll have to call this flight off for now. There’re some papers for me
to sign. I’ll see you later.”
Fifteen minutes before the first cadet class reported for the nine
o’clock period, Lieutenant Black came to the line. The sergeant told
the lieutenant all that he had not learned.
“But I don’t want to pass the buck too crudely,” the sergeant
concluded. “What’s the matter with us two going up in the thing and
learning what’s to be learned?”
What the sergeant wanted was more airwork. He would have
taken his flying on the tail end of a rocket were no other means
offered. The fact that a ship’s action was in question meant nothing
to him. More than likely the sergeant was glad that nobody had been
able to locate the kink; test flying is always to the liking of a real lover
of air. The betting’s even that the sergeant had planned this moment
during the previous night. As he talked, he talked Black toward the
waiting plane. The instructor was adjusting helmet and goggles, and
his silence gave consent.
“It’s funny,” he finally said, as they waited for the motor man to
warm the engine, “but those controls did jam. I don’t want any of my
cadets to get in dutch through mechanical faults. They’re bad
enough without that. The Lord only knows when I’ll be able to turn
any of them loose. Such an iron fisted bunch of shovel apprentices
I’ve never met. They wouldn’t’ve made good K.P’s. for the wartime
kadets.
“And these damn’ Jennies have got to be right, Sergeant. As right
as they can be, and if they were twice as right as that, they’d still be
all wrong. Climb in and we’ll take a turn of the field.”
While they were adjusting the safety belts, Cadet Shane came
running along the line of hangars. He scrambled aboard Black’s
lower wing and talked into the instructor’s left ear. Black throttled his
motor low, pushed back his goggles, thought for half a minute,
studied his instrument board dials, shook and kicked his controls,
then turned to the man in the rear seat and said:
“Sergeant, I’m going to give the cadet his hop. These controls
seem to be O.K. Chances are, there was nothing wrong with them.
“Jump out, Sergeant, and I’ll let you know how they act. Watch my
first turn of the field and see how I’m getting along. Climb in, Shane!
Let’s get going!”
The sergeant went back to the hangar. He wasn’t talking to
anybody, for the time being, but he hurled an open can of red paint
the length of the big building and said to a few idle privates—
“Clean that up!”
Then, where a group of flying cadets were busily rolling two small
cubes on a work bench, the sergeant came down in hot wrath, threw
the harmless squares through the skylight and yelled—
“Get to hell out of this hangar and stay out!”
After that the sergeant went out, retrieved the dice and
reestablished the game. He told the cadets that he was sore about
something but could not recall just what. After sending the privates
off to goldbrick in the post exchange, the sergeant mopped up the
paint.
Master Sergeant Sciples, in charge of the hangar, came along to
start the day. Sciples was spending this enlistment on the
construction of certain souvenirs. And at no time did he allow hangar
work to cut in on his program. He was an easy boss. Sciples looked
at his sergeant rigger and came out in language that lay people
erroneously suppose is solely characteristic of the Marine Corps.
Here and there, without half trying, Master Sergeant Sciples could
extemporize in a manner that would make the Marine Corps’
glossary look like a first reader for morons. Sciples’ language, to say
the least, was able.
“Sergeant,” he said, “one look at you, you tells me that you haven’t
had your morning flight. When will you forget this flying stuff and put
your mind on next week’s debut into the outer world? Why, you
— Snap into it and get wise!”
“But, Sciples,” the sergeant said. “It’s the same old story. The
same thing that I’ve been up against for three years. And it makes
me mad, Sciples. Hell, if I live to be a hundred, I’ll never lose this
desire to fly. It’s different with you, you old decrepit”—the sergeant
was never entirely tongue tied himself—“You don’t care about flying.
The bug’s never grazed upon you. You don’t know the hell and pain
and longing that an egg like me faces, Sciples. Why, Sciples, this
thing of giving a right arm for something is nothing. I’d do another
stretch in this damn’ Army if I really thought that I’d aviate. And that is
what I call bravery.”
“Crazy as a loon!” Sciples exclaimed. “Why you—you don’t know
enough to—”
“And this was the most cruel thrust of all, Sciples,” the sergeant
went on, “this thing that came off half an hour ago, why—” The
hangar’s telephone rang, and Sciples, with the sergeant still talking,
strolled toward the instrument—“why, there I was all set to take off
with Black. Had myself nicely planted in the rear seat, and who
comes out and robs me but my ex-cook, that rotten cook, Shane,
and—” There were tears in the thick voice.
For a minute Sciples talked over the line. In the end he said, “Well
that’s hell,” and hung up.
“What’s hell?” the sergeant forgot his own troubles long enough to
ask.
“Cadet Shane,” Master Sergeant Sciples said, “Shane, the man
who unseated you, Shane and Black spun into the ground ten miles
from here. They both burned to death.”
THE END
Transcriber’s Note: This story appeared in the November 15,

1928 issue of Adventure magazine.
*** END OF THE PROJECT GUTENBERG EBOOK THEN LUCK
CAME IN ***
Updated editions will replace the previous one—the old editions will
be renamed.
Creating the works from print editions not protected by U.S.

copyright law means that no one owns a United States copyright in
these works, so the Foundation (and you!) can copy and distribute it
in the United States without permission and without paying copyright
royalties. Special rules, set forth in the General Terms of Use part of
this license, apply to copying and distributing Project Gutenberg™
electronic works to protect the PROJECT GUTENBERG™ concept
and trademark. Project Gutenberg is a registered trademark, and
may not be used if you charge for an eBook, except by following the
terms of the trademark license, including paying royalties for use of
the Project Gutenberg trademark. If you do not charge anything for
copies of this eBook, complying with the trademark license is very
easy. You may use this eBook for nearly any purpose such as
creation of derivative works, reports, performances and research.
Project Gutenberg eBooks may be modified and printed and given
away—you may do practically ANYTHING in the United States with
eBooks not protected by U.S. copyright law. Redistribution is subject
to the trademark license, especially commercial redistribution.
START: FULL LICENSE

THE FULL PROJECT GUTENBERG LICENSE
PLEASE READ THIS BEFORE YOU DISTRIBUTE OR USE THIS WORK
To protect the Project Gutenberg™ mission of promoting the free

distribution of electronic works, by using or distributing this work (or
any other work associated in any way with the phrase “Project
Gutenberg”), you agree to comply with all the terms of the Full
Project Gutenberg™ License available with this file or online at
www.gutenberg.org/license.
Section 1. General Terms of Use and

Redistributing Project Gutenberg™
electronic works
1.A. By reading or using any part of this Project Gutenberg™
electronic work, you indicate that you have read, understand, agree
to and accept all the terms of this license and intellectual property
(trademark/copyright) agreement. If you do not agree to abide by all
the terms of this agreement, you must cease using and return or
destroy all copies of Project Gutenberg™ electronic works in your
possession. If you paid a fee for obtaining a copy of or access to a
Project Gutenberg™ electronic work and you do not agree to be
bound by the terms of this agreement, you may obtain a refund from
the person or entity to whom you paid the fee as set forth in
paragraph 1.E.8.
1.B. “Project Gutenberg” is a registered trademark. It may only be

used on or associated in any way with an electronic work by people
who agree to be bound by the terms of this agreement. There are a
few things that you can do with most Project Gutenberg™ electronic
works even without complying with the full terms of this agreement.
See paragraph 1.C below. There are a lot of things you can do with
Project Gutenberg™ electronic works if you follow the terms of this
agreement and help preserve free future access to Project
Gutenberg™ electronic works. See paragraph 1.E below.
1.C. The Project Gutenberg Literary Archive Foundation (“the
Foundation” or PGLAF), owns a compilation copyright in the
collection of Project Gutenberg™ electronic works. Nearly all the
individual works in the collection are in the public domain in the
United States. If an individual work is unprotected by copyright law in
the United States and you are located in the United States, we do
not claim a right to prevent you from copying, distributing,
performing, displaying or creating derivative works based on the
work as long as all references to Project Gutenberg are removed. Of
course, we hope that you will support the Project Gutenberg™
mission of promoting free access to electronic works by freely
sharing Project Gutenberg™ works in compliance with the terms of
this agreement for keeping the Project Gutenberg™ name
associated with the work. You can easily comply with the terms of
this agreement by keeping this work in the same format with its
attached full Project Gutenberg™ License when you share it without
charge with others.
1.D. The copyright laws of the place where you are located also
govern what you can do with this work. Copyright laws in most
countries are in a constant state of change. If you are outside the
United States, check the laws of your country in addition to the terms
of this agreement before downloading, copying, displaying,
performing, distributing or creating derivative works based on this
work or any other Project Gutenberg™ work. The Foundation makes
no representations concerning the copyright status of any work in
any country other than the United States.
1.E. Unless you have removed all references to Project Gutenberg:
1.E.1. The following sentence, with active links to, or other

immediate access to, the full Project Gutenberg™ License must
appear prominently whenever any copy of a Project Gutenberg™
work (any work on which the phrase “Project Gutenberg” appears, or
with which the phrase “Project Gutenberg” is associated) is
accessed, displayed, performed, viewed, copied or distributed:
This eBook is for the use of anyone anywhere in the United
States and most other parts of the world at no cost and with
almost no restrictions whatsoever. You may copy it, give it away
or re-use it under the terms of the Project Gutenberg License
included with this eBook or online at www.gutenberg.org. If you
are not located in the United States, you will have to check the
laws of the country where you are located before using this
eBook.
1.E.2. If an individual Project Gutenberg™ electronic work is derived

from texts not protected by U.S. copyright law (does not contain a
notice indicating that it is posted with permission of the copyright
holder), the work can be copied and distributed to anyone in the
United States without paying any fees or charges. If you are
redistributing or providing access to a work with the phrase “Project
Gutenberg” associated with or appearing on the work, you must
comply either with the requirements of paragraphs 1.E.1 through
1.E.7 or obtain permission for the use of the work and the Project
Gutenberg™ trademark as set forth in paragraphs 1.E.8 or 1.E.9.
1.E.3. If an individual Project Gutenberg™ electronic work is posted

with the permission of the copyright holder, your use and distribution
must comply with both paragraphs 1.E.1 through 1.E.7 and any
additional terms imposed by the copyright holder. Additional terms
will be linked to the Project Gutenberg™ License for all works posted
with the permission of the copyright holder found at the beginning of
this work.
1.E.4. Do not unlink or detach or remove the full Project

Gutenberg™ License terms from this work, or any files containing a
part of this work or any other work associated with Project
Gutenberg™.
1.E.5. Do not copy, display, perform, distribute or redistribute this

electronic work, or any part of this electronic work, without
prominently displaying the sentence set forth in paragraph 1.E.1 with
active links or immediate access to the full terms of the Project
Gutenberg™ License.
1.E.6. You may convert to and distribute this work in any binary,
compressed, marked up, nonproprietary or proprietary form,
including any word processing or hypertext form. However, if you
provide access to or distribute copies of a Project Gutenberg™ work
in a format other than “Plain Vanilla ASCII” or other format used in
the official version posted on the official Project Gutenberg™ website
(www.gutenberg.org), you must, at no additional cost, fee or expense
to the user, provide a copy, a means of exporting a copy, or a means
of obtaining a copy upon request, of the work in its original “Plain
Vanilla ASCII” or other form. Any alternate format must include the
full Project Gutenberg™ License as specified in paragraph 1.E.1.
1.E.7. Do not charge a fee for access to, viewing, displaying,

performing, copying or distributing any Project Gutenberg™ works
unless you comply with paragraph 1.E.8 or 1.E.9.
1.E.8. You may charge a reasonable fee for copies of or providing

access to or distributing Project Gutenberg™ electronic works
provided that:
• You pay a royalty fee of 20% of the gross profits you derive from
the use of Project Gutenberg™ works calculated using the
method you already use to calculate your applicable taxes. The
fee is owed to the owner of the Project Gutenberg™ trademark,
but he has agreed to donate royalties under this paragraph to
the Project Gutenberg Literary Archive Foundation. Royalty
payments must be paid within 60 days following each date on
which you prepare (or are legally required to prepare) your
periodic tax returns. Royalty payments should be clearly marked
as such and sent to the Project Gutenberg Literary Archive
Foundation at the address specified in Section 4, “Information
about donations to the Project Gutenberg Literary Archive
Foundation.”
• You provide a full refund of any money paid by a user who

notifies you in writing (or by e-mail) within 30 days of receipt that
s/he does not agree to the terms of the full Project Gutenberg™
License. You must require such a user to return or destroy all
copies of the works possessed in a physical medium and
discontinue all use of and all access to other copies of Project
Gutenberg™ works.
• You provide, in accordance with paragraph 1.F.3, a full refund of

any money paid for a work or a replacement copy, if a defect in
the electronic work is discovered and reported to you within 90
days of receipt of the work.
• You comply with all other terms of this agreement for free
distribution of Project Gutenberg™ works.
1.E.9. If you wish to charge a fee or distribute a Project Gutenberg™

electronic work or group of works on different terms than are set
forth in this agreement, you must obtain permission in writing from
the Project Gutenberg Literary Archive Foundation, the manager of
the Project Gutenberg™ trademark. Contact the Foundation as set
forth in Section 3 below.
1.F.
1.F.1. Project Gutenberg volunteers and employees expend

considerable effort to identify, do copyright research on, transcribe
and proofread works not protected by U.S. copyright law in creating
the Project Gutenberg™ collection. Despite these efforts, Project
Gutenberg™ electronic works, and the medium on which they may
be stored, may contain “Defects,” such as, but not limited to,
incomplete, inaccurate or corrupt data, transcription errors, a
copyright or other intellectual property infringement, a defective or
damaged disk or other medium, a computer virus, or computer
codes that damage or cannot be read by your equipment.
1.F.2. LIMITED WARRANTY, DISCLAIMER OF DAMAGES - Except

for the “Right of Replacement or Refund” described in paragraph
1.F.3, the Project Gutenberg Literary Archive Foundation, the owner
of the Project Gutenberg™ trademark, and any other party
distributing a Project Gutenberg™ electronic work under this
agreement, disclaim all liability to you for damages, costs and

Ebook Advances in Probability and Mathematical Statistics Clapem 2019 Merida Mexico 1St Edition Daniel Hernandez Hernandez Online PDF All Chapter

Uploaded by

Copyright:

Available Formats

You might also like

Ebook Advances in Probability and Mathematical Statistics Clapem 2019 Merida Mexico 1St Edition Daniel Hernandez Hernandez Online PDF All Chapter

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ebook Advances in Probability and Mathematical Statistics Clapem 2019 Merida Mexico 1St Edition Daniel Hernandez Hernandez Online PDF All Chapter

Uploaded by

Copyright:

Available Formats

Advances in Probability and

Mathematical Statistics CLAPEM 2019

Starting Out The c3 Sicilian 1st Edition John Emms

Advances in Probability and Mathematical Statistics

English Grammar Exercises with answers Part 4 Your

The American Revolution 1774 83 2nd Edition Daniel

BNF 83 (British National Formulary) March 2022 Joint

Patriot vs Loyalist American Revolution 1775 83 Combat

British Light Infantryman vs Patriot Rifleman American

Secrecy Public Relations and the British Nuclear Debate

Progress in Probability is designed for the publication of workshops, seminars

All manuscripts are peer-reviewed to meet the highest standards of scientific

More information about this series at http://www.springer.com/series/4839

Ramsés H. Mena Juan Carlos Pardo Millán

ISSN 1050-6977 ISSN 2297-0428 (electronic)

The present volume contains contributions of the XV Latin American Congress of

rationality of L. The asymptotic behavior of four binary classification methods,

dad Autónoma Metropolitana; Universidad Autónoma de Yucatán; and Universidad

Guanajuato, Mexico Daniel Hernández-Hernández

Asymptotic Connectedness of Random Interval Graphs in a One

Caleb Erubiel Andrade Sernas, Gilberto Calvillo Vives,

Abstract In this work we present a probabilistic analysis of random interval

Keywords Connectedness analysis · Data delivery problem · Mobile agents ·

The research presented in this work is in the intersection of several disciplines,

C. E. Andrade Sernas · G. Calvillo Vives · P. C. Manrique Mirón · E. Treviño Aguilar ()

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 1

1.1 The Data Delivery on a Line Problem

The production of inexpensive, simple-built, mobile robots has become a reality

1.2 Computational Complexity

1.4 Random Interval Graphs

distributed uniformly in [0, r]). The intervals are [xi − ai ρi , xi + (1 − 2ai ) ρi ],

2 Graph Theoretical Formulation of the Data Delivery on a

In this section we show how the DDLP is equivalent to an existence problem

[xi − ayi , xi + (1 − 2a) yi ] ⊂ [xi − yi , xi + yi ] .

3 Probabilistic Analysis of GA (Q, a)

We denote as usual the degree of a vertex vi by d(i). We write vi ∼ vj when there

The family of random variables X1 , . . . , Xn indicates where a connected component

in the components. Then, necessarily Xj = 1 with j > 2. This is a contradiction.

is the random variable that counts the number of connected components of

lim inf P (X(n) = 1) ≥ 1 − e−c . (1)

B(i) := {ω | vi+ (ω) ∈ ∪k=i vk }

For t ∈ [0, (1 − 2a)r] the function g simplifies to g(t, ρ) = t + aρ − (t − (1 −

Note that P (t ∈ v2 ) ≥ 12 ar. Then

For x ∈ R we will make use of the following basic inequality:

For t ∈ [(1 − 2a)r, 1 − ar] we have

g(t, ρ) = (t + aρ) ∧ 1 − (t − (1 − 2a)ρ)+ ∧ 1

Hence, for t ∈ [(1 − 2a)r, 1 − ar]

We have proved the theorem. 

Observe that Y (n) counts the number of intervals vi = [xi − ai ρi , xi + (1 − 2ai ) ρi ]

This establishes (4) since

E(Yi ) = P (d(i) = 0 and xi ∈ [1/3, 2/3])

where μ is the distribution associated to the random vertex vi . Note that

P (vi ∼ v | ρ = t) = 1 − (1 − a)ρi − (1 − a)t.

From (5), we determine E[Y (n)] by

Since α < 1 and αβ = limn→∞ nr

1 [1 − (1 − a)r]n − 2 [1 − 2(1 − a)r]n + [1 − 3(1 − a)r]n

From (6) and (12), we get

R(n) = max {xi } , L(n) = min {xi } (13)

We have proved the theorem.

R (n) = max {xi + (1 − 2a) ρi } , L (n) = min {xi − aρi } . (14)

Proof To simplify notation we write R, L and R , L . Recalling {xi }i=1...,n are

u − v = min{|u − v|, L − |u − v|}.

u − v < k ⇐⇒ distG (u, v) ≤ k.

Indeed, if u − v < k, then for some ε > 0, there is a path u = x0 , x1 , . . . , xk = v

Thus, the approximate points are xi = u + t + (i − 1) for 1 ≤ i ≤ n − 1. Note

u − v < δ ⇐⇒ U ∪ {u} and U ∪ {v} are large for some small U.

Note that qv = mv and v ∈ [ qmv , qvm+1 ). Moreover, since S is integer-distance-free,

m · u − v = m · f (u) − f (v) for all u, v ∈ S̄.