Advances in Soft and Hard Computing

Jerzy Peja■
Visit to download the full and correct content document:
https://textbookfull.com/product/advances-in-soft-and-hard-computing-jerzy-pejas/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Soft Computing Theories and Applications Proceedings of

SoCTA 2019 Advances in Intelligent Systems and
Computing Millie Pant (Editor)

https://textbookfull.com/product/soft-computing-theories-and-
applications-proceedings-of-socta-2019-advances-in-intelligent-
systems-and-computing-millie-pant-editor/

Advances in Soft Computing and Machine Learning in

Image Processing 1st Edition Aboul Ella Hassanien

https://textbookfull.com/product/advances-in-soft-computing-and-
machine-learning-in-image-processing-1st-edition-aboul-ella-
hassanien/

International Proceedings on Advances in Soft Computing

Intelligent Systems and Applications ASISA 2016 1st
Edition M. Sreenivasa Reddy

https://textbookfull.com/product/international-proceedings-on-
advances-in-soft-computing-intelligent-systems-and-applications-
asisa-2016-1st-edition-m-sreenivasa-reddy/

Recent Advances of Hybrid Intelligent Systems Based on

Soft Computing 1st Edition Patricia Melin

https://textbookfull.com/product/recent-advances-of-hybrid-
intelligent-systems-based-on-soft-computing-1st-edition-patricia-
melin/
Soft and Hard Probes of QCD Topological Structures in
Relativistic Heavy Ion Collisions Shuzhe Shi

https://textbookfull.com/product/soft-and-hard-probes-of-qcd-
topological-structures-in-relativistic-heavy-ion-collisions-
shuzhe-shi/

Soft Computing in Chemical and Physical Sciences A

Shift in Computing Paradigm 1st Edition Kanchan Sarkar

https://textbookfull.com/product/soft-computing-in-chemical-and-
physical-sciences-a-shift-in-computing-paradigm-1st-edition-
kanchan-sarkar/

Application of Soft Computing and Intelligent Methods

in Geophysics Alireza Hajian

https://textbookfull.com/product/application-of-soft-computing-
and-intelligent-methods-in-geophysics-alireza-hajian/

Soft Computing Applications Proceedings of the 8th

International Workshop Soft Computing Applications SOFA
2018 Vol I Valentina Emilia Balas

https://textbookfull.com/product/soft-computing-applications-
proceedings-of-the-8th-international-workshop-soft-computing-
applications-sofa-2018-vol-i-valentina-emilia-balas/

Soft Computing Applications Proceedings of the 8th

International Workshop Soft Computing Applications SOFA
2018 Vol II Valentina Emilia Balas

https://textbookfull.com/product/soft-computing-applications-
proceedings-of-the-8th-international-workshop-soft-computing-
applications-sofa-2018-vol-ii-valentina-emilia-balas/
Advances in Intelligent Systems and Computing 889

Jerzy Pejaś
Imed El Fray
Tomasz Hyla
Janusz Kacprzyk Editors

Advances
in Soft
and Hard
Computing
Advances in Intelligent Systems and Computing

Volume 889

Series editor
Janusz Kacprzyk, Systems Research Institute, Polish Academy of Sciences,
Warsaw, Poland
e-mail: kacprzyk@ibspan.waw.pl
The series “Advances in Intelligent Systems and Computing” contains publications on theory,
applications, and design methods of Intelligent Systems and Intelligent Computing. Virtually all
disciplines such as engineering, natural sciences, computer and information science, ICT, economics,
business, e-commerce, environment, healthcare, life science are covered. The list of topics spans all the
areas of modern intelligent systems and computing such as: computational intelligence, soft computing
including neural networks, fuzzy systems, evolutionary computing and the fusion of these paradigms,
social intelligence, ambient intelligence, computational neuroscience, artificial life, virtual worlds and
society, cognitive science and systems, Perception and Vision, DNA and immune based systems,
self-organizing and adaptive systems, e-Learning and teaching, human-centered and human-centric
computing, recommender systems, intelligent control, robotics and mechatronics including
human-machine teaming, knowledge-based paradigms, learning paradigms, machine ethics, intelligent
data analysis, knowledge management, intelligent agents, intelligent decision making and support,
intelligent network security, trust management, interactive entertainment, Web intelligence and multimedia.
The publications within “Advances in Intelligent Systems and Computing” are primarily proceedings
of important conferences, symposia and congresses. They cover significant recent developments in the
field, both of a foundational and applicable character. An important characteristic feature of the series is
the short publication time and world-wide distribution. This permits a rapid and broad dissemination of
research results.

Advisory Board
Chairman
Nikhil R. Pal, Indian Statistical Institute, Kolkata, India
e-mail: nikhil@isical.ac.in
Members
Rafael Bello Perez, Faculty of Mathematics, Physics and Computing, Universidad Central de Las Villas, Santa
Clara, Cuba
e-mail: rbellop@uclv.edu.cu
Emilio S. Corchado, University of Salamanca, Salamanca, Spain
e-mail: escorchado@usal.es
Hani Hagras, School of Computer Science & Electronic Engineering, University of Essex, Colchester, UK
e-mail: hani@essex.ac.uk
László T. Kóczy, Department of Information Technology, Faculty of Engineering Sciences, Győr, Hungary
e-mail: koczy@sze.hu
Vladik Kreinovich, Department of Computer Science, University of Texas at El Paso, El Paso, TX, USA
e-mail: vladik@utep.edu
Chin-Teng Lin, Department of Electrical Engineering, National Chiao Tung University, Hsinchu, Taiwan
e-mail: ctlin@mail.nctu.edu.tw
Jie Lu, Faculty of Engineering and Information, University of Technology Sydney, Sydney, NSW, Australia
e-mail: Jie.Lu@uts.edu.au
Patricia Melin, Graduate Program of Computer Science, Tijuana Institute of Technology, Tijuana, Mexico
e-mail: epmelin@hafsamx.org
Nadia Nedjah, Department of Electronics Engineering, University of Rio de Janeiro, Rio de Janeiro, Brazil
e-mail: nadia@eng.uerj.br
Ngoc Thanh Nguyen, Wrocław University of Technology, Wrocław, Poland
e-mail: Ngoc-Thanh.Nguyen@pwr.edu.pl
Jun Wang, Department of Mechanical and Automation, The Chinese University of Hong Kong, Shatin,
Hong Kong
e-mail: jwang@mae.cuhk.edu.hk

More information about this series at http://www.springer.com/series/11156

Jerzy Pejaś Imed El Fray
•

Tomasz Hyla Janusz Kacprzyk

•

Editors

Advances in Soft and Hard

Computing

123
Editors
Jerzy Pejaś Tomasz Hyla
West Pomeranian University of Technology West Pomeranian University of Technology
in Szczecin in Szczecin
Szczecin, Poland Szczecin, Poland

Imed El Fray Janusz Kacprzyk

West Pomeranian University of Technology Polish Academy of Sciences
in Szczecin Systems Research Institute
Szczecin, Poland Warsaw, Poland

ISSN 2194-5357 ISSN 2194-5365 (electronic)

Advances in Intelligent Systems and Computing
ISBN 978-3-030-03313-2 ISBN 978-3-030-03314-9 (eBook)
https://doi.org/10.1007/978-3-030-03314-9

Library of Congress Control Number: 2018960424

© Springer Nature Switzerland AG 2019

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

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

Advanced Computer System 2018 (ACS 2018) conference was the 21st in the
series of conferences organized by the Faculty of Computer Science and
Information Technology of the West Pomeranian University of Technology in
Szczecin, Poland. That event could not be possible without scientific cooperation
with Warsaw University of Technology, Faculty of Mathematics and Information
Science, Poland; Warsaw University of Life Sciences (SGGW), Poland; AGH
University of Science and Technology, Faculty of Physics and Applied Computer
Science, Poland; Polish Academy of Sciences (IPIPAN), Institute of Computer
Science, Poland; Kuban State University of Technology, Institute of Information
Technology and Safety, Russia; Bialystok University of Technology, Poland; and
—last but not least—Ehime University in Matsuyama, Japan. As usual, the con-
ference was held in Miȩdzyzdroje, Poland, on 24–26 September 2018.
This volume contains a collection of carefully selected, peer-reviewed papers
presented during the conference sessions. The main topics covered by the chapters
in this book are artificial intelligence, software technologies, information technol-
ogy security and multimedia systems.
It has been a tradition since the first conference that the organizers have always
invited top specialists in the fields. Many top scientists and scholars, who have
presented keynote talks over the years, have always provided an inspiration for
future research and for young and experienced participants.
The book places a great emphasis both on theory and practice. The contributions
not only reflect the invaluable experience of eminent researchers in relevant areas
but also point new methods, approaches and interesting direction for the future
researches.
In keeping with ACS mission over the last twenty years, this 21st conference,
ACS 2018, was also an event providing a comprehensive state-of-the-art summary
from keynote speakers as well as a look forward towards future research priorities.
We believe that the keynote talks provided an inspiration for all attendees. This year
authors of the keynote talks were professors: Nabendu Chaki from University of
Calcutta (India), Akira Imada from Brest State Technical University (Belarus),
Keiichi Endo and Shinya Kobayashi from Ehime University (Japan), Ryszard

v
vi Preface

Kozera from Warsaw University of Life Sciences SGGW (Poland), Jacek Pomykała
from the University of Warsaw (Poland) and Marian Srebrny from Polish Academy
of Sciences (Poland).
We would like to give a proof of appreciation to all members of the International
Programme Committee for their time and effort in reviewing the papers, helping us
to shape the scope and topics of the conference and providing us with much advice
and support. Moreover, we want to express a gratitude to all of the organizers from
the Faculty of Computer Science and Information Technology, West Pomeranian
University of Technology in Szczecin for their enthusiasm and hard work, notably
Ms. Hardej, Secretary of the Conference, and all other members of Organizing
Committee including Luiza Fabisiak, Tomasz Hyla and Witold Maćków.
We expect this book to shed new light on unresolved issues and inspire the
reader to greater challenges. We also hope that the book will provide tools or ideas
for their creation that will be more effective in solving increasingly complex
research problems and reaching common scientific goals.

September 2018 Imed El Fray

Tomasz Hyla
Janusz Kacprzyk
Jerzy Pejaś
Organization

Advanced Computer System 2018 (ACS 2018) was organized by the West
Pomeranian University of Technology in Szczecin, Faculty of Computer Science
and Information Technology (Poland), in cooperation with Warsaw University of
Technology, Faculty of Mathematics and Information Science (Poland); AGH
University of Science and Technology, Faculty of Physics and Applied Computer
Science (Poland); Ehime University (Japan); Polish Academy of Sciences IPIPAN
(Poland); Kuban State University of Technology, Institute of Information
Technology and Safety (Russia); and Bialystok University of Technology (Poland).

Organizing Committee

Tomasz Hyla (Chair) West Pomeranian University of Technology,

Szczecin, Poland
Sylwia Hardej (Secretary) West Pomeranian University of Technology,
Szczecin, Poland
Witold Maćków West Pomeranian University of Technology,
Szczecin, Poland
Luiza Fabisiak West Pomeranian University of Technology,
Szczecin, Poland

Programme Committee Chairs

Jerzy Pejaś West Pomeranian University of Technology,

Szczecin, Poland
Imed El Fray West Pomeranian University of Technology,
Szczecin, Poland

vii
viii Organization

Tomasz Hyla West Pomeranian University of Technology,

Szczecin, Poland

International Programming Committee

Costin Badica University of Craiova, Romania

Zbigniew Banaszak Warsaw University of Technology, Poland
Anna Bartkowiak Wroclaw University, Poland
Włodzimierz Bielecki West Pomeranian University of Technology,
Szczecin, Poland
Leon Bobrowski Bialystok Technical University, Poland
Grzegorz Bocewicz Koszalin University of Technology, Poland
Robert Burduk Wroclaw University of Technology, Poland
Andrzej Cader Academy of Humanities and Economics in Lodz,
Poland
Aleksandr Cariow West Pomeranian University of Technology,
Szczecin, Poland
Nabendu Chaki Calcutta University, India
Krzysztof Chmiel Poznan University of Technology, Poland
Ryszard S. Choraś University of Technology and Life Sciences,
Poland
Krzysztof Ciesielski Polish Academy of Sciences, Poland
Nicolas Tadeusz Courtois University College London, UK
Albert Dipanda Le Centre National de la Recherche Scientifique,
France
Bernard Dumont European Commission, Information Society
and Media Directorate General, France
Jos Dumortier KU Leuven University, Belgium
Keiichi Endo Ehime University, Japan
Özgür Ertug̃ Gazi University, Turkey
Oleg Fińko Kuban State University of Technology, Russia
Paweł Forczmański West Pomeranian University of Technology,
Szczecin, Poland
Dariusz Frejlichowski West Pomeranian University of Technology,
Szczecin, Poland
Jerzy August Gawinecki Military University of Technology, Poland
Larisa Globa National Technical University of Ukraine,
Ukraine
Janusz Górski Technical University of Gdansk, Poland
Władysław Homenda Warsaw University of Technology, Poland
Akira Imada Brest State Technical University, Belarus
Michelle Joab LIRMM, Universite Montpellier 2, France
Jason T. J. Jung Yeungnam University, Korea
Organization ix

Janusz Kacprzyk Systems Research Institute, Polish Academy

of Sciences, Poland
Andrzej Kasiński Poznan University of Technology, Poland
Shinya Kobayashi Ehime University, Japan
Marcin Korzeń West Pomeranian University of Technology,
Szczecin, Poland
Zbigniew Adam Kotulski Polish Academy of Sciences, Poland
Piotr Andrzej Kowalski AGH University of Science and Technology
and SRI Polish Academy of Sciences, Poland
Ryszard Kozera Warsaw University of Life Sciences—SGGW,
Poland
Mariusz Kubanek Częstochowa University of Technology, Poland
Mieczysław Kula University of Silesia, Poland
Eugeniusz Kuriata University of Zielona Gora, Poland
Mirosław Kurkowski Cardinal Stefan Wyszyński University
in Warsaw, Poland
Jonathan Lawry University of Bristol, UK
Javier Lopez University of Malaga, Spain
Andriy Luntovskyy BA Dresden University of Coop. Education,
Germany
Kurosh Madani Paris XII University, France
Przemysław Mazurek West Pomeranian University of Technology,
Szczecin, Poland
Andrzej Niesler Wroclaw University of Economics, Poland
Arkadiusz Orłowski Warsaw University of Life Sciences—SGGW,
Poland
Marcin Paprzycki Systems Research Institute, Polish Academy
of Sciences, Poland
Paweł Pawlewski Poznań University of Technology, Poland
Witold Pedrycz University of Alberta, Canada
Andrzej Piegat West Pomeranian University of Technology,
Szczecin, Poland
Josef Pieprzyk Macquarie University, Australia
Jacek Pomykała Warsaw University, Poland
Alexander Prokopenya Warsaw University of Life Sciences—SGGW,
Poland
Elisabeth Rakus-Andersson Blekinge Institute of Technology, School of
Engineering, Sweden
Izabela Rejer West Pomeranian University of Technology,
Szczecin, Poland
Vincent Rijmen Graz University of Technology, Austria
Valery Rogoza West Pomeranian University of Technology,
Szczecin, Poland
Leszek Rutkowski Czestochowa University of Technology, Poland
Khalid Saeed Warsaw University of Technology, Poland
x Organization

Kurt Sandkuhl University of Rostock, Germany

Albert Sangrá Universitat Oberta de Catalunya, Spain
Władysław Skarbek Warsaw University of Technology, Poland
Vaclav Snaśel Technical University of Ostrava, Czech Republic
Jerzy Sołdek West Pomeranian University of Technology,
Szczecin, Poland
Zenon Sosnowski Białystok University of Technology, Poland
Marian Srebrny Institute of Computer Science, Polish Academy
of Sciences, Poland
Peter Stavroulakis Technical University of Crete, Greece
Janusz Stokłosa Poznan University of Technology, Poland
Marcin Szpyrka AGH University of Science and Technology,
Poland
Ryszard Tadeusiewicz AGH University of Science and Technology,
Poland
Oleg Tikhonenko University of K. Wyszynski, Warsaw, Poland
Natalia Wawrzyniak Maritime University of Szczecin, Poland
Jan Węglarz Poznan University of Technology, Poland
Sławomir Wierzchoń Institute of Computer Science, Polish Academy
of Sciences, Poland
Antoni Wiliński West Pomeranian University of Technology,
Szczecin, Poland
Toru Yamaguchi Tokyo Metropolitan University, Japan

Additional Reviewers

Bilski, Adrian Landowski, Marek

Bobulski, Janusz Maleika, Wojciech
Chmielewski, Leszek Mantiuk, Radosław
Fabisiak, Luiza Maćków, Witold
Goszczyńska, Hanna Okarma, Krzysztof
Grocholewska-Czuryło, Anna Olejnik, Remigiusz
Hoser, Paweł Radliński, Lukasz
Jaroszewicz, Szymon Rozenberg, Leonard
Jodłowski, Andrzej Różewski, Przemysław
Karwański, Marek Siedlecka-Lamch, Olga
Klęsk, Przemysław Steingartner, William
Kurek, Jarosław Świderski, Bartosz
Contents

Invited Paper
Fitting Dense and Sparse Reduced Data . . . . . . . . . . . . . . . . . . . . . . . . . 3
Ryszard Kozera and Artur Wiliński

Artificial Intelligence
Survey of AI Methods for the Purpose of Geotechnical
Profile Creation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Adrian Bilski
Algorithm for Optimization of Multi-spindle Drilling Machine
Based on Evolution Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Paweł Hoser, Izabella Antoniuk, and Dariusz Strzęciwilk
Horizontal Fuzzy Numbers for Solving Quadratic Fuzzy Equation . . . . 45
Marek Landowski
Regression Technique for Electricity Load Modeling and Outlined
Data Points Explanation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Krzysztof Karpio, Piotr Łukasiewicz, and Rafik Nafkha
Correct Solution of Fuzzy Linear System Based on Interval Theory . . . 68
Andrzej Piegat and Marcin Pietrzykowski
Processing of Z þ -numbers Using the k Nearest Neighbors Method . . . . 76
Marcin Pluciński
Fingerprint Feature Extraction with Artificial Neural Network
and Image Processing Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
Maciej Szymkowski and Khalid Saeed
An Investment Strategy Using Temporary Changes in the Behavior
of the Observed Group of Investors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
Antoni Wilinski and Patryk Matuszak

xi
xii Contents

Software Technology
Measuring Gender Equality in Universities . . . . . . . . . . . . . . . . . . . . . . 109
Tindara Addabbo, Claudia Canali, Gisella Facchinetti,
and Tommaso Pirotti
Transitive Closure Based Schedule of Loop Nest
Statement Instances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
Wlodzimierz Bielecki and Marek Palkowski
Design of the BLINDS System for Processing and Analysis
of Big Data - A Pre-processing Data Analysis Module . . . . . . . . . . . . . . 132
Janusz Bobulski and Mariusz Kubanek
QoS and Energy Efficiency Improving in Virtualized Mobile Network
EPC Based on Load Balancing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
Larysa Globa, Nataliia Gvozdetska, Volodymyr Prokopets,
and Oleksandr Stryzhak
The Approach to Users Tasks Simplification on Engineering
Knowledge Portals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
Larysa Globa, Rina Novogrudska, and O. Koval
Repository Model for Didactic Resources . . . . . . . . . . . . . . . . . . . . . . . . 159
Andrzej Jodłowski, Ewa Stemposz, and Alina Stasiecka
SLMA and Novel Software Technologies for Industry 4.0 . . . . . . . . . . . 170
Andriy Luntovskyy
Applications of Multilingual Thesauri for the Texts Indexing
in the Field of Agriculture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Waldemar Karwowski, Arkadiusz Orłowski, and Marian Rusek
On Code Refactoring for Decision Making Component Combined
with the Open-Source Medical Information System . . . . . . . . . . . . . . . . 196
Vasyl Martsenyuk and Andriy Semenets
Programmable RDS Radio Receiver on ATMEGA88 Microcontroller
on the Basis of RDA5807M Chip as the Central Module in Internet
of Things Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Jakub Peksinski, Pawel Kardas, and Grzegorz Mikolajczak
Business Process Modelling with “Cognitive” EPC Diagram . . . . . . . . . 220
Olga Pilipczuk and Galina Cariowa
Algorithmic Decomposition of Tasks with a Large Amount of Data . . . 229
Walery Rogoza and Ann Ishchenko
Contents xiii

Managing the Process of Servicing Hybrid Telecommunications

Services. Quality Control and Interaction Procedure
of Service Subsystems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
Mariia A. Skulysh, Oleksandr I. Romanov, Larysa S. Globa,
and Iryna I. Husyeva

Information Technology Security

Validation of Safety-Like Properties for Entity-Based Access
Control Policies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
Sergey Afonin and Antonina Bonushkina
Randomness Evaluation of PP-1 and PP-2 Block Ciphers Round
Keys Generators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
Michał Apolinarski
New Results in Direct SAT-Based Cryptanalysis
of DES-Like Ciphers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
Michał Chowaniec, Mirosław Kurkowski, and Michał Mazur
Secure Generators of q-Valued Pseudo-random Sequences
on Arithmetic Polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
Oleg Finko, Sergey Dichenko, and Dmitry Samoylenko
A Hybrid Approach to Fault Detection in One Round
of PP-1 Cipher . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
Ewa Idzikowska
Protection of Information from Imitation on the Basis
of Crypt-Code Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
Dmitry Samoylenko, Mikhail Eremeev, Oleg Finko, and Sergey Dichenko
On a New Intangible Reward for Card-Linked Loyalty Programs . . . . 332
Albert Sitek and Zbigniew Kotulski
KaoChow Protocol Timed Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346
Sabina Szymoniak
Electronic Document Interoperability in Transactions Executions . . . . . 358
Gerard Wawrzyniak and Imed El Fray

Multimedia Systems
L-system Application to Procedural Generation of Room Shapes
for 3D Dungeon Creation in Computer Games . . . . . . . . . . . . . . . . . . . 375
Izabella Antoniuk, Paweł Hoser, and Dariusz Strzęciwilk
Hardware-Efficient Algorithm for 3D Spatial Rotation . . . . . . . . . . . . . 387
Aleksandr Cariow and Galina Cariowa
xiv Contents

Driver Drowsiness Estimation by Means of Face Depth

Map Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 396
Paweł Forczmański and Kacper Kutelski
Vehicle Passengers Detection for Onboard eCall-Compliant Devices . . . 408
Anna Lupinska-Dubicka, Marek Tabędzki, Marcin Adamski,
Mariusz Rybnik, Maciej Szymkowski, Miroslaw Omieljanowicz,
Marek Gruszewski, Adam Klimowicz, Grzegorz Rubin,
and Lukasz Zienkiewicz
An Algorithm for Computing the True Discrete Fractional
Fourier Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 420
Dorota Majorkowska-Mech and Aleksandr Cariow
Region Based Approach for Binarization of Degraded
Document Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433
Hubert Michalak and Krzysztof Okarma
Partial Face Images Classification Using Geometrical Features . . . . . . . 445
Piotr Milczarski, Zofia Stawska, and Shane Dowdall
A Method of Feature Vector Modification in Keystroke Dynamics . . . . 458
Miroslaw Omieljanowicz, Mateusz Popławski,
and Andrzej Omieljanowicz
Do-It-Yourself Multi-material 3D Printer for Rapid Manufacturing
of Complex Luminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469
Dawid Paleń and Radosław Mantiuk
Multichannel Spatial Filters for Enhancing SSVEP Detection . . . . . . . . 481
Izabela Rejer
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
Invited Paper
 Fitting Dense and Sparse Reduced Data

Ryszard Kozera1,2(B) and Artur Wiliński1

1
Faculty of Applied Informatics and Mathematics, Warsaw University of Life
Sciences - SGGW, ul. Nowoursynowska 159, 02-776 Warsaw, Poland
ryszard.kozera@gmail.com
2
Department of Computer Science and Software Engineering, The University of
Western Australia, 35 Stirling Highway, Crawley, Perth, WA 6009, Australia

Abstract. This paper addresses the topic of fitting reduced data repre-
sented by the sequence of interpolation points M = {qi }ni=0 in arbitrary
Euclidean space Em . The parametric curve γ together with its knots
T = {ti }ni=0 (for which γ(ti ) = qi ) are both assumed to be unknown.
We look at some recipes to estimate T in the context of dense versus
sparse M for various choices of interpolation schemes γ̂. For M dense,
the convergence rate to approximate γ with γ̂ is considered as a possible
criterion to force a proper choice of new knots T̂ = {t̂i }n
i=0 ≈ T . The
latter incorporates the so-called exponential parameterization “retriev-
ing” the missing knots T from the geometrical spread of M. We examine
the convergence rate in approximating γ by commonly used interpolants
γ̂ based here on M and exponential parameterization. In contrast, for
M sparse, a possible optional strategy is to select T̂ which optimizes
a certain cost function depending on the family of admissible knots T̂ .
This paper focuses on minimizing “an average acceleration” within the
family of natural splines γ̂ = γ̂ N S fitting M with T̂ admitted freely in
the ascending order. Illustrative examples and some applications listed
supplement theoretical component of this work.

Keywords: Interpolation · Reduced data

Computer vision and graphics

1 Introduction
Let γ : [0, T ] → Em be a smooth regular curve (i.e. γ̇(t) = 0) defined over
t ∈ [0, T ], for 0 < T < ∞ - see e.g. [1]. The term reduced data (denoted by
M) represents the sequence of n + 1 interpolation points {qi }ni=0 in arbitrary
Euclidean space Em . Here, each point from M satisfies the condition qi = γ(ti )
with extra constraint qi+1 = qi (i = 0, 1, . . . , n − 1). The respective knots T =
{ti }ni=0 are assumed to be unavailable. The latter stands in contrast with the
classical problem of fitting non-reduced data where both M and T are given.
Naturally, any interpolation scheme γ̂ fitting M relies on the provision of some
T̂ = {t̂i }ni=0 at best “well approximating” the unknown knots T . This paper
discusses two different approaches in selecting the substitutes T̂ of T (subject
c Springer Nature Switzerland AG 2019
J. Pejaś et al. (Eds.): ACS 2018, AISC 889, pp. 3–17, 2019.
https://doi.org/10.1007/978-3-030-03314-9_1
4 R. Kozera and A. Wiliński

to γ̂(t̂i ) = qi and t̂i < t̂i+1 ) for either dense or sparse reduced data M. The
theoretical component of this work is also complemented by several indicative
examples. The relevant discussion on the topic in question can be found e.g. in [2–
5,7,9–15,17,19,22,23,26]. The problem of interpolating reduced or non-reduced
data arises in computer graphics and vision (e.g. for trajectory modelling and
image compression or segmentation), in engineering (like robotics: path planning
or motion modelling) in physics (e.g. for trajectory modelling) and in medical
image processing (e.g. in image segmentation and area estimation) - see [27–30].
More literature on the above topic can be found among all in [2,27,31].

2 Interpolating Dense Reduced Data

For M forming dense reduced data the intrinsic assumption admits n as suffi-
ciently large. Thus upon selecting specific interpolation scheme γ̂ : [0, T̂ ] → Em
together with a particular choice of T̂ ≈ T the question of convergence rate
α in approximating γ with γ̂ (for n → ∞) arises naturally. Furthermore, an
equally intriguing matter refers to the existence of such T̂ so that the respective
convergence rates α in γ̂ ≈ γ coincide once γ̂ is taken either with T̂ or with T .
This section addresses both issues raised above. In doing so, recall first some
preliminaries (see e.g. [2,3]):
Definition 1. The sampling T = {ti }ni=0 is called admissible provided:
lim δn = 0, where δn = max {ti − ti−1 : i = 1, 2, . . . , n}. (1)
n→∞ 1≤i≤n

In addition, T represents more-or-less uniform sampling if there exist some con-

stants 0 < Kl ≤ Ku such that for sufficiently large n:
Kl Ku
≤ ti − ti−1 ≤ (2)
n n
holds, for all i = 1, 2, . . . , n. Alternatively, more-or-less uniformity requires the
existence of a constant 0 < β ≤ 1 fulfilling asymptotically βδn ≤ ti − ti−1 ≤ δn ,
for all i = 1, 2, . . . , n. Noticeably, the case of Kl = Ku = β = 1 yields T as a
uniform sampling. Lastly we call T as ε-uniformly sampled (with ε > 0) if:
iT 1
ti = φ( ) + O( 1+ε ), (3)
n n
holds for sufficiently large n and i = 1, 2, . . . , n. Here the function φ : [0, T ] →
[0, T ] is an order preserving re-parameterization (i.e. with φ̇ > 0).
Note that both (2) and (3) are genuine subfamilies of (1). We formulate now
the notion of convergence order (see again e.g. [3]):
Definition 2. Consider a family {fδn , δn > 0} of functions fδn : [0, T ] → E. We
say that fδn is of order O(δnα ) (denoted as fδn = O(δnα )), if there is a constant
K > 0 such that, for some δ̄ > 0 the inequality |fδn (t)| < Kδnα holds for all δn ∈
(0, δ̄), uniformly over [0, T ]. In case of vector-valued functions Fδn : [0, T ] → En
by Fδn = O(δnα ) we understand Fδn  = O(δnα ).
 Fitting Dense and Sparse Reduced Data 5

In case of non-reduced data represented by M and T , in Definition 2, one sets

Fδn = γ−γ̂ as both domains of γ and γ̂ coincide with [0, T ]. If only M is available
(with somehow guessed T̂ ), the domain of the interpolant γ̂ : [0, T̂ ] → Em should
be re-mapped (at best reparameterized with ψ̇ > 0) with ψ : [0, T ] → [0, T̂ ] so
that the convergence analysis of γ − γ̂ ◦ ψ can be performed. In fact here, the
function Fδn from Definition 2 reads as Fδn = γ − γ̂ ◦ ψ.
Finally, the notion of sharpness of convergence rates α is recalled:

Definition 3. For a given interpolation scheme γ̂ based on M and some T̂ ≈ T

(subject to some mapping φ : [0, T ] → [0, T̂ ]) the asymptotics γ − γ̂ ◦ φ = O(δnα )
over [0, T ] is sharp within the predefined family of curves γ ∈ J and family of
samplings T ∈ K, if for some γ ∈ J and some sampling from K, there exists t∗ ∈
[0, T ] and some positive constant K such that γ(t∗ )−(γ̂◦φ)(t∗ ) = Kδnα +O(δnρ ),
where ρ > α. A similar definition applies to non-reduced data M and T with ψ
omitted.

Suppose the unknown knots T are estimated by T̂λ with the so-called expo-
nential parameterization (see e.g. [27]):

t̂0 = 0 and t̂i = t̂i−1 + qi − qi−1 λ , (4)

for i = 1, 2, . . . , n, where λ ∈ [0, 1] is a free parameter. The technical condition

qi = qi+1 assumed in Sect. 1 guarantees t̂i < t̂i+1 . The case λ = 0 renders for T̂0
uniform knots t̂i = i which represents a “blind guess” of T . In contrast λ = 1
yields the so-called cumulative chord parameterization T̂1 (see e.g. [12,27]):

t̂i = t̂i−1 + qi − qi−1 . (5)

Visibly, the latter accounts for the geometrical layout of reduced data M. For
λ = 1 the last node T̂ from now on is denoted by T̂c = t̂n .
We pass now to different classes of splines γ̂ (see e.g. [2]) which at junction
points in M (where consecutive local interpolants are glued together) are of class
C l (for l = 0, 1, 2) and are C ∞ over sub-interval (ti , ti+1 ), with i = 0, 1, . . . , n−1.

2.1 Continuous Splines at Junction Points

To fit M with T given (i.e. for non-reduced data) one can apply piecewise-r-
degree Lagrange polynomials γL(r) (see [2]) for which if γ ∈ C r+1 then:

γL(r) = γ + O(δnr+1 ), (6)

uniformly over [0, T ]. By (6) and Definition 2 for any samplings (1) the con-
vergence order α = r + 1 prevails in γ ≈ γL(r) . Noticeably (6) is sharp (see
Definition 3).
Surprisingly, for reduced data M, if γL(r) is used with (5) (i.e. for γ̂ = γ̂L(r) )
the resulting asymptotics in γ ≈ γ̂L(r) matches (6) for r = 2, 3. At this point
recall that Newton Interpolation formula [2] (based on divided differences) yields
6 R. Kozera and A. Wiliński

i
over each consecutive sub-interval Ii = [t̂i , t̂i+2 ] the quadratic γ̂L(2) = γ̂L(2) |Ii
defined as:
i
γ̂L(2) (t̂) = γ[t̂i ] + γ[t̂i , t̂i+1 ](t̂ − t̂i ) + γ[t̂i , t̂i+1 , t̂i+2 ](t̂ − t̂i )(t̂ − t̂i+1 ) (7)

i
and also over each consecutive sub-interval I¯i = [t̂i , t̂i+3 ] the cubic γ̂L(3) =
γ̂L(3) |I¯i defined as:
i
γ̂L(3) (t̂) = γ̂L(2) (t̂) + γ[t̂i , t̂i+1 , t̂i+2 , t̂i+3 ](t̂ − t̂i )(t̂ − t̂i+1 )(t̂ − t̂i+2 ). (8)

For (7) and (8) the following result is established in [3,4]:

Theorem 1. Suppose γ is a regular C r curve in Em , where r ≥ k + 1 and k is

either 2 or 3. Let γ̂L(k) : [0, T̂ ] → Em be the cumulative chord based piecewise-
degree-k interpolant defined by M (sampled admissibly (1)) with T̂1 ≈ T defined
by (5). Then there is a piecewise reparameterization ψ : [0, T ] → [0, T̂ ] such that:

γ̂L(k) ◦ ψ = γ + O(δnk+1 ), (9)

holds uniformly over [0, T ] (i.e. here α = 3, 4). The asymptotics in (9) is sharp.

Thus for either piecewise-quadratic or piecewise-cubic Lagrange interpolants

based on reduced data M and cumulative chords (5) the missing knots T can
be well compensated by T̂1 . Indeed, to approximate γ with γ̂L(2,3) , Theorem 1
guarantees identical convergence orders as compared to those from (6). Note also
that for r = 1 the trajectories of both piecewise-linear interpolants γL(1) (based
on T ) and γ̂L(1) (based on any T̂ ) coincide as they are uniquely determined by
M. Therefore by (6), for both γ ≈ γL(1) and γ ≈ γ̂L(1) the convergence rate
α = 2.
Interestingly, raising the polynomial degree r ≥ 4 in γ̂L(r) (used with (5))
does not further accelerate α in (9) - see [3,6]. The latter stands in contrast with
(6) for which any r in γL(r) renders extra speed-up in α(r) = r + 1.
The remaining cases of exponential parameterization (4) lead to another
unexpected result (see [7–9]) which extends Theorem 1 to all λ ∈ [0, 1):

Theorem 2. Suppose γ is a regular C k+1 curve in Em sampled more-or-less

uniformly (2) (here k = 2, 3). Let M form reduced data and the unknown knots
T are estimated by T̂λ according to (4) for λ ∈ [0, 1). Then there exists a mapping
ψ : [0, T ] → [0, T̂ ] such that (see also (7) and (8)):

γ̂L(k) = γ + O(δn ), (10)

which holds uniformly over [0, T ]. The convergence rate α(λ) = 1 in (10) is
sharp. Additionally, a sharp accelerated α(λ) follows for M sampled ε-uniformly
(3), with ε > 0 and λ ∈ [0, 1):

γ̂L(2) = γ + O(δnmax{3,1+2ε} ). (11)

 Fitting Dense and Sparse Reduced Data 7

The more-or-less uniformity (2) cannot be dropped in Theorem 2. Noticeably the

mapping ψ forms a genuine reparameterization only for special λ ∈ [0, 1) - see
[11]. Both Theorems 1 and 2 underline the substantial discontinuous deceleration
effect in α(λ) dropping abruptly from α(1) = 3 for k = 2 (or from α(1) = 4 for
k = 3) to the linear one α(λ) = 1, for all λ ∈ [0, 1). A possible advantage to deal
with λ ∈ [0, 1) in (4) is to retain a certain degree of freedom (controlled by a
single parameter λ ∈ [0, 1)) at the cost of keeping much slower linear convergence
order in γ ≈ γ̂L(2,3) . Such relaxation of λ ∈ [0, 1) can be exploited if on top of
securing even a slow convergence in γ ≈ γ̂, some other extra shape-preserving
properties of γ̂L(2,3) are stipulated - see e.g. [28].

2.2 C 1 Splines at Junction Points

In order to fit reduced data with C 1 interpolant at all junction points (coinciding
here with M \ {q0 , qm }) a modified Hermite interpolation γ̂H can be applied (see
[2,3,13] or the next Sect. 3). The latter defines a piecewise-cubic γ̂H which over
each sub-interval [t̂i , t̂i+1 ] satisfies (19). It also relies on the provision of the
estimates of the missing velocities V = {γ̇(ti )}ni=0 over M (for i = 0, 1 . . . , n).
Such estimates {vi }ni=0 of V can be possibly obtained upon exploiting Lagrange
piecewise-cubic γ̂L(3) from (8) over each sub-interval I¯i = [t̂i , t̂i+3 ] with vi =
i i+1
γ̂L(3) (t̂i ). Here to compute the next vi+1 we consider γ̂L(3) defined over I¯i+1 . The
n
last four velocities {vj }j=n−3 are the derivatives of γ̂L(3) (defined over [t̂n−3 , t̂n ])
calculated at {t̂j }nj=n−3 . The following result holds (see [3,13,14]):
Theorem 3. Let γ be a regular C 4 ([0, T ]) curve in Em sampled according to
(1). Given reduced data M and knots’ estimates (5) (i.e. for λ = 1 in (4)) there
exists a piecewise-cubic C 1 reparameterization φH : [0, T ] → [0, T̂ ] such that:
γ̂H ◦ φH = γ + O(δn4 ), (12)
uniformly over [0, T ]. If additionally (1) is also more-or-less uniform (2) then
for M and (4) (with λ ∈ [0, 1)) there exists a mapping φH : [0, T ] → [0, T̂ ] such
that (uniformly over [0, T ]) we have:
γ̂H ◦ φH = γ + O(δn ). (13)
Both (12) and (13) are sharp.
Similarly to Subsect. 2.1, both (12) and (13) imply an abrupt left-hand side
discontinuity of α(λ) at λ = 1 once γ̂H is used. In addition, by (12) cumulative
chords (5) combined with M and γ̂H yield the same quartic convergence order
α(1) = 4 as established for classical case of non-reduced data M combined
with T and with exact velocities V = {γ(ti )}ni=0 , for which we also have γH =
γ + O(δn4 ) (see e.g. [2]). Here γH is a standard Hermite interpolant based on M,
T and V - see Sect. 3. Consequently fitting M with modified Hermite interpolant
γ̂H based on (5) compensates the unavailable T and V without decelerating the
asymptotic rate in trajectory estimation. For the remaining λ ∈ [0, 1) in (4),
by (13) a slow linear convergence order prevails in exchange of retaining some
flexibility (controlled by λ ∈ [0, 1) in modelling the trajectory of γ̂H .
8 R. Kozera and A. Wiliński

2.3 C 2 Splines at Junction Points

In order to fit M with some C 2 interpolant γ̂ at all junction points M \ {q0 , qn }

(and elsewhere C ∞ ) one can apply e.g. a complete spline γ̂ = γ̂CS or a natural
spline γ̂ = γ̂N S - see [2] or the next Sect. 3. The first one relies on the additional
provision of exact initial and terminal velocities v0 = γ̇(0) and vn = γ̇(tn ). The
following result holds (see [10]):
Theorem 4. Let γ be a regular C 4 ([0, T ]) curve in Em sampled according to
(1). Given reduced data M, v0 , vm and cumulative chord based knots’ estimates
(5) there exists a piecewise-cubic C 2 reparameterization φCS : [0, T ] → [0, T̂ ]
such that (uniformly over [0, T ]):

γ̂CS ◦ φCS = γ + O(δn4 ). (14)

The asymptotics in (14) is sharp.

The case of natural spline γ̂N S combined with M and (5) yields decelerated α(1)
which upon repeating the argument in [10] leads to a sharp asymptotic estimate:

γ̂N S ◦ φN S = γ + O(δn2 ). (15)

Indeed, for the natural spline γ̂N S the unknown γ̈(t0 ) and γ̈(tn ) are substituted
by ad hock taken null vectors which ultimately results in slower asymptotics
(15) over both sub-intervals [t0 , t1 ] and [tn−1 , tn ]. The latter pulls down a fast
quartic order α(1) = 4 from (14) (holding for γ̂CS ) to α(1) = 2 claimed in (15)
for γ̂N S . As previously, by (14) and (15) and [2] both C 2 interpolants γ̂CS and
γ̂N S coupled with (5) yield exactly the same asymptotics in γ approximation as
compared to γCS and γN S used with T given.
The numerical tests for γ̂CS and γ̂N S combined for λ ∈ [0, 1) in (4) indicate
the same asymptotic effects as claimed in (10) and (13). In practice, the terminal
velocities v0 and vn do not accompany reduced data M. However, they can still
 
be well estimated with w1 = γ̂L(3) (0) and wn = γ̂L(3) (t̂n ). The interpolant based
on M, w0 , wn and (4) is called modified complete spline and is denoted by γ̂CSm .
It is numerically verified in [15,19] that for M sampled more-or-less uniformly
(2), λ ∈ [0, 1) and γ ∈ C 4 the following holds:

γ̂N S ◦ φN S = γ + O(δn ) γ̂CSm ◦ φCMm = γ + O(δn ), (16)

for some C 2 mappings φN S , φCMm : [0, T ] → [0, T̂ ].

The discussion for the alternative schemes retrieving the estimates of T can
be found e.g. in [2,16,18,20,21].

3 Fitting Sparse Reduced Data

In this section a possible alternative to fit sparse reduced data M is discussed.

Since here n << ∞, an arbitrary interpolation scheme γ̂ coupled with any knots’
 Fitting Dense and Sparse Reduced Data 9

estimates T̂ ≈ T cannot result in γ̂ converging to γ. Thus for M sparse the need

for new knots’ selection criterion arises. This section addresses this issue.
Consider now a class I of admissible curves γ̂ forming piecewise C 2 curves
γ̂ : [0, T̂ ] → Em (where 0 < T̂ < ∞ is fixed) interpolating M together with the
free unknown knots T̂ in ascending order t̂i < t̂i+1 and satisfying γ̂(t̂i ) = qi . It
is assumed here that curves γ̂ ∈ I can possibly be only C 1 at M.
The task is now to find γ̂opt ∈ I which minimizes the following cost function
(with {t̂i }n−1
i=1 freed as t̂0 = 0 and T̂ is fixed):

  t̂i+1
n−1
J (γ̂) = ¨
γ̂(t)2
dt. (17)
i=0 t̂i

Note that J from (17) measures the integrated squared norm of the γ̂ acceler-
ation. The requirement for T̂ to preserve the ascending order stipulates (17) to
be optimized over an admissible zone of knots:
Ω = {(t̂0 , t̂1 , . . . , t̂n−1 , t̂n ) ∈ Rn+1 : t̂0 = 0 < t̂1 < . . . < t̂n−1 < t̂n = T̂ }. (18)
Consequently, the latter reformulates (17) into constrained optimization problem.
For further consideration we recall now the construction of a piecewise-cubic
spline interpolant γ̂C based on data points M and knots T̂ (see e.g. [2]). Here
over each sub-interval Iˆi = [t̂i , t̂i+1 ] one introduces a cubic γ̂Ci = γ̂C |[t̂i ,t̂i+1 ] :

γ̂Ci (t̂) = c1,i + c2,i (t̂ − t̂i ) + c3,i (t̂ − t̂i )2 + c4,i (t̂ − t̂i )3 , (19)
which (for i = 0, 1, 2, . . . , n − 1; cj,i ∈ Rm , where j = 1, 2, 3, 4) fulfills:

γ̂Ci (t̂i+k ) = qi+k , γ̂˙ Ci (t̂i+k ) = vi+k , k = 0, 1. (20)

The respective velocities v0 , v1 , v2 , . . . , vn−1 , vn ∈ Rm in (20) are treated tem-
porarily as free parameters (if unknown). The coefficients cj,i from (19) (with
Δt̂i = t̂i+1 − t̂i ) by [2] read as:
c1,i = xi , c2,i = vi ,
vi + vi+1 − 2 qi+1
Δt̂
−qi (qi+1 −qi )
Δt̂i
− vi
c4,i = i
and c3,i = − c4,i Δt̂i . (21)
(Δt̂i )2 Δt̂i
Assuming that both velocities v0 and vn are a priori given, a cubic spline γ̂C
called a complete spline (denoted as γ̂CS ) can e.g. be used to fit M with T̂ .
If in turn, the velocities v0 and vn are unavailable a possible alternative
to interpolate M with T̂ is to apply the so-called natural splines (denoted as
γ̂C = γ̂N S ) - see e.g. [2]. In fact, both missing v0 and vn can be easily expressed
in terms of the initial and terminal accelerations a0 and an . The latter as usually
not supplied are commonly substituted by ad hock admitted values a0 = an =
0 ∈ Rm . In this case the combination of nullified a0 and an with (19) and (20)
leads to the following two linear (m dimensional vector) equations:
q1 − q 0 qn − qn−1
2v0 + v1 = 3 and vn−1 + 2vn = 3 . (22)
Δt̂0 Δt̂n−1
10 R. Kozera and A. Wiliński

The remaining velocities {vi }n−1

i=1 can be computed from (19) and (20) which
leads to the tridiagonal m linear systems (strictly diagonally dominant - see [2]):

vi−1 Δt̂i + 2vi (Δt̂i−1 + Δt̂i ) + vi+1 Δt̂i−1 = bi , (23)

where
qi − qi−1 qi+1 − qi
bi = 3(Δt̂i + Δt̂i−1 ), (24)
Δt̂i−1 Δt̂i
and 1 ≤ i ≤ n − 1. Equations (22), (23) and (24) render m linear systems, each
of size (n + 1) × (n + 1), strictly diagonally dominant which can be solved e.g.
by Gauss elimination without pivoting [2]. The latter renders a unique sequence
of missing velocities {vi }ni=0 all expressed in terms of M and T̂ . Ultimately, the
substitution of computed {vi }ni=0 into (21) yields the explicit formula for the
natural cubic spline γ̂N S determined unambiguously by M and T̂ .
In case of natural splines forming the subfamily of I the cost function (17)
for γ̂N S and (19) reformulates into (see [22,23]):

  t̂i+1
n−1
J (γ̂N S ) = γ̂¨N Si (t)2 dt
i=0 tˆi
n−1

=4 (c3,i 2 Δt̂i + 3c4,i 2 (Δt̂i )3 + 3 c3,i |c4,i (Δt̂i )2 ), (25)
i=0

where ·|· is a standard dot product in Rm . Coupling (21) with (25) yields:


n−1
−1
J (γ̂N S ) = 4 3
(−3qi+1 − qi 2 + 3 vi + vi+1 |qi+1 − qi Δt̂i
i=0
(Δ t̂ i )

−(vi 2 + vi+1 2 + vi |vi+1 )(Δt̂i )2 .

For arbitrary fixed knots T̂ satisfying t̂i < t̂i+1 we have (see e.g. [2]):

Lemma 1. For a given reduced data M and fixed T̂ in arbitrary Euclidean space
Em the subclass of natural splines IT̂N S ⊂ IT̂ satisfies:

min J (γ̂) = min J (γ̂N S ). (26)

γ̂∈IT̂ γ N S ∈I N S
T̂

The last lemma reformulates the problem of minimizing (17) over IT̂ (with
T̂ temporarily fixed) to the class of natural splines IT̂N S . Hence by (26), once the
knots T̂ are permitted to vary subject to (18) the problem (17) reduces into:

min J (γ̂) = min J (γ̂N S ). (27)

γ̂∈I γ N S ∈I N S

As demonstrated above since each γ̂N S is uniquely determined by T̂ and

M the optimization of (17) reformulates into a constrained finite dimensional
 Fitting Dense and Sparse Reduced Data 11

opt
optimization task (28) over (t̂1 , t̂2 , . . . , t̂n−1 ) ∈ Ω which satisfies J (γ̂N S) =
F opt opt opt
J (t̂0 , t̂1 , t̂2 , . . . , t̂n−1 , t̂n ) (with t̂0 and t̂n fixed). Here for the optimal knots
T̂ opt = (t̂0 , t̂opt opt opt
1 , t̂2 , . . . , t̂n−1 , t̂n ) ∈ Ω (see (18)) we arrive at (see also [22,23]):


n−1
−1
J F (T̂ opt ) = 4 min 3
(−3qi+1 − qi 2 + 3 vi + vi+1 |qi+1 − qi Δt̂i
T̂ ∈Ω
i=0
(Δ t̂ i )

−(vi 2 + vi+1 2 + vi |vi+1 )(Δt̂i )2 , (28)

where the velocities {vi }ni=0 are expressed in terms of T̂ as outlined above.
The function J F is shown to be highly non-linear and non-convex [24]. The
latter makes it hard to optimize upon applying standard numerical schemes.
In addition the explicit formula (28) depends on velocities {vi }ni=0 which are
implicitly given by (22) and (23) in terms of a given M and T̂ relaxed to be
free. In fact even for n small the expression for (28) treated as a function of
T̂ (with M fixed) gets quite complicated [24] (see also Example 3 below). It
can be shown however that J F is a continuous function over T̂ ∈ Ω (see [24]).
Furthermore, though Ω is not compact, the existence of a global minimizer T̂ opt
to J F can still be justified as claimed below (see [25]):
Theorem 5. There exists at least one global minimum to (28) (and thus by (27)
to (17)) with the some optimal knots (t̂0 = 0, t̂opt opt opt
1 , t̂2 , . . . , t̂n−1 , t̂n = T̂ ) ∈ Ω.

More literature on fitting sparse reduced data can be found e.g. in [2,3,26].
In the next step we report on implementing selected schemes from Sects. 2 and 3.

4 Examples for Fitting Reduced Data

This section includes two examples to fit dense and sparse 2D and 3D reduced
data M based on computation performed in Mathematica - see [32]. For more
examples covering different n and m see [3,14,22,23]. Noticeably, all γ̂ from
Sects. 2 and 3 are applicable to multi-dimensional M. For the detailed discussion
of the numerical optimization schemes used in Subsect. 4.2 see e.g. [23,33–35].

4.1 Examples for M dense

We use two types of more-or-less uniform samplings (2) to simulate M:
⎧
i
⎪
⎨ n, if i = 2k;
i 1 i (−1)i+1
ti = n + 2n , if i = 4k + 1; (ii) ti = + . (29)
⎪
⎩ i − 1 , if i = 4k + 3, n 3n
n 2n

For a selected scheme γ̂ the numerical estimate ᾱ(λ) ≈ α(λ) for γ approxi-
mation is computed from the regression line y(x) = ᾱ(λ)x + b applied to the
collection of pairs of points {(log(n), − log(En ))}nnmax
min
(see [3]), where the error
En = supt∈[0,T ] (γ̂ ◦ ψ)(t) − γ(t) = maxt∈[0,T ] (γ̂ ◦ ψ)(t) − γ(t).
12 R. Kozera and A. Wiliński

Example 1. Take the following 3D curve γst : [0, 1] → E3 :

γst (t) = (cos(2πt), sin(2πt), (1.2)2 − (sin(2πt))2 , (30)

sampled along (29). Assume Lagrange piecewise-cubic γ̂L(3) is used to fit M -

see Subsect. 2.1. The Table 1 confirm the sharpness of (10) claimed in Theorem 2.
More experiments can be found (including piecewise-quadratics γ̂L(2) ) in [7–9].

Table 1. Computed ᾱ(λ) ≈ α(λ) for γst (30) sampled as in (29), for n ∈ {72, . . . , 162}.

λ 0.0 0.1 0.3 0.5 0.7 0.9 1.0

ᾱ(λ) for (29)(i) 1.000 1.000 1.003 1.001 1.008 1.200 4.981
ᾱ(λ) for (29)(ii) 0.993 0.994 0.996 0.998 1.009 1.186 3.961
α(λ) in Theorem 2 1.0 1.0 1.0 1.0 1.0 1.0 4.0

Example 2. Consider a planar spiral γsp : [0, 1] → E2 defined by:

γsp (t) = ((0.2 + t) cos(π(1 − t)), (0.2 + t) sin(π(1 − t))) , (31)

sampled according to (29). Assume a modified Hermite interpolant γ̂H is used to

fit M - see Subsect. 2.2. The experimental results from Table 2 confirm numeri-
cally the sharpness of Theorem 3.

Table 2. Computed ᾱ(λ) ≈ α(λ) for γsp (31) sampled as in (29), for n ∈ {96, . . . , 144}.

λ 0.0 0.1 0.3 0.5 0.7 0.9 1.0

ᾱ(λ) for (29)(i) 0.968 0.969 0.970 0.973 0.979 1.021 3.918
ᾱ(λ) for (29)(ii) 0.969 0.970 0.974 0.978 0.986 1.038 3.987
α(λ) in Theorem 3 1.0 1.0 1.0 1.0 1.0 1.0 4.0

Due to the page limitation, the examples verifying the sharpness of the
asymptotics from Subsect. 2.3 are omitted. For more see e.g. [10,15,19].

4.2 Examples for M sparse

Example 3. Consider eight 2D data points (i.e. here n = 7) in E2 with

M7 = {(0, 0), (−0.5, −4), (0.5, −4), (−0.5, 4), (0.5, 0.4), (−1, 3.8), (0.3, 0.3), (0.5, 0.5)}
 Fitting Dense and Sparse Reduced Data 13

The FindMinimum (Newton Method) [32] causes computer to hang and no result
was reported within 60 min. The respective initial guesses for the knots T̂ coin-
cide with cumulative chords (5) calculated in terms of M7 . On the other hand
FindMinimum (Secant Method) yields the corresponding optimal knots:

T opt = {0, 2.67713, 4.69731, 10.3221, 12.3943, 14.8132, 19.0316, 19.6231}.

SM
The respective execution time amounts to TM 7
= 35.708519 s. Recall that Secant
Method requires two sequences of initial knots to initialize the iteration process.
For the latter cumulative chords (5) are perturbed here by taking their left-shift
t̂ci − δi and right-shift t̂ci + δi , accordingly, to serve as two initial guesses.

Example 3 illustrates that for n ≥ 7 FindMinimum (Secant Method) offers a

feasible computational scheme to optimize (28). The next example compares the
latter with an alternative scheme called Leap-Frog (see e.g. [23]).
We pass now to the example with M representing reduced data in E3 .

Example 4. Consider for n = 6 the 3D data points (see dotted points in Fig. 1):

-2
0.5
-1 0.25
0.5 0 -1 0.5
-1 -1
-0.25
0 0 0 0
0 0
1 -0.5
.5 -0.5
.5
2

1.5
1 1.5
1 1.5
1
1.5
1

1 1 1
1
0.5
0. 0.5
0. 0.5
0.
0.5
0.
0 0
0
0
0.4
0.2
0
-0.2
-0.4

a) b) c) d)

Fig. 1. Natural splines γ̂N S interpolating M6 from (32) with (a) uniform knots T̂uni ,
opt opt2
(b) cumulative chords T̂c , (c) optimal knots T̂LF = T̂SM (d) plotted together for
(a)–(c).

M6 = {γ(0), γ(3), γ(3.1), γ(0.4), γ(0.6), γ(3.2), γ(3.4)}, (32)

√ √ √
where γ(t) = ( t cos t, t sin t, t). The uniform interpolation knots (re-scaled
to T̂c ) coincide with:

T̂uni = {0, 1.42223, 2.84446, 4.26669, 5.68892, 7.11115, 8.53338}. (33)

The initial guess derived from cumulative chords T̂c (see (5)) is represented by:

T̂c = {0, 2.44949, 2.62868, 5.23375, 5.47857, 8.16249, 8.53338}. (34)

The natural splines γ̂ N S (based either on T̂uni or on T̂c ) render the respective
energies (28) as J F (T̂uni ) = 18.7237 > J F (T̂c ) = 13.8136. The trajectories of
both interpolants are presented in Fig. 1(a) and (b).
14 R. Kozera and A. Wiliński

The Secant Method yields (for (28)) the optimal knots (augmented by termi-
nal knots t̂0 = 0 and t̂6 = T̂c - see (5)) as:
opt1
T̂SM = {0, 1.75693, 2.33172, 4.89617, 5.49792, 8.12181, 8.53338}
opt1
with the corresponding optimal energy J F (T̂SM ) = 9.21932. The execution
SM
time equals to T1 = 33.79 s. For each free variable the Secant Method uses
here two initial numbers t̂ci ± 0.1 (i.e. perturbed cumulative chord numbers). For
other initial guesses t̂ci ± 0.2 marginally more precise knots (compatible with
Leap-Frog - see below) are generated:
opt2
T̂SM = {0, 1.76066, 2.35289, 4.90326, 5.50495, 8.12262, 8.53338} (35)
opt2
with more accurate optimal energy J F (T̂SM ) = 9.21787. Here the execution
SM
time reads as T2 = 51.88436 s and gets longer if accuracy is improved. The
resulting curve γ̂ N S is plotted in Fig. 1(c).
opt opt2
The Leap-Frog Algorithm decreases the energy to J F (T̂LF ) = J F (T̂SM )
opt opt2
(as for the Secant Method) with the iteration stopping conditions T̂LF = T̂SM
(up to 6th decimal point) upon 79 iterations. The respective execution time is
equal to T LF = 8.595979 s. < T2SM < T1SM . The 0th (i.e. J F (T̂c )), 1st, 2nd,
3rd, 10th, 20th, 40nd and 79th iterations of Leap-Frog decrease the energy to:

{13.8136, 11.3619, 10.3619, 9.88584, 9.25689, 9.21987, 9.21787, 9.21787} (36)

with only the first three iterations substantially correcting the initial guess knots
opt opt2
T̂c . Since T̂LF = T̂SM both natural splines γ̂ N S are identical - see in Fig. 1(c).
The graphical comparison between γ̂N S based on either (33) or (34) or (35) is
shown in Fig. 1(d).
Note that if the Leap-Frog iteration bound condition is adjusted e.g. to ensure
the current Leap-Frog energy to coincide with J F (T̂cSM ) (say up to 5th decimal
place) then only 40 iterations are needed which speeds-up the execution time to
TELF = 4.789121 s. < T SM with adjusted optimal knots
opt
T̂LFE
= {0, 1.76153, 2.35384, 4.90451, 5.50603, 8.12278, 8.53338}.

Evidently, at the cost of losing marginal accuracy in optimal knots’ estimation

the acceleration in Leap-Frog execution time is achieved with almost identical
opt opt2
interpolating curve as the optimal one - here T̂LF E
≈ T̂SM . Similar acceleration
follows for other a posteriori selected stopping conditions like e.g. a bound on a
relative decrease of the J F .

5 Conclusions
In this paper we discuss several methods of fitting reduced data M forming
ordered collection of n + 1 points in arbitrary Euclidean space Em . The points
in M are generated from the interpolation condition γ(ti ) = qi with the cor-
responding knots T = {ti }ni=0 assumed to be unknown. Different criteria of
 Fitting Dense and Sparse Reduced Data 15

estimating the missing knots T are discussed here in the context of sparse or
dense M fitted with various interpolation schemes.
The first part of this work addresses the problem of interpolating M when
n is large. Different interpolants γ̂ combined with exponential parameterization
(4) are discussed to determine the underlying speed of convergence in γ̂ ≈ γ.
It is also demonstrated that cumulative chords (5) yield identical convergence
orders to approximate γ as if the genuine knots T were given. The annotated
experiments conducted in Mathematica confirm the asymptotics obtained by
theoretical analysis.
The second part of this work deals with the case of fitting M, when n << ∞.
Under such scenario the convergence of γ̂ to γ with any T̂ (or T ) is obviously
excluded. Different knots’ selection strategy is therefore required. In doing so, an
infinite dimensional optimization task (17) is reformulated into a corresponding
finite dimensional minimization problem (28) - see Lemma 1. The existence of
global optimal knots T opt minimizing the “average acceleration” of γ̂ in (28)
(and given in the ascending order) justifies Theorem 5. Two numerical schemes
for optimizing (28) (i.e. Leap-Frog and Secant Method) are also implemented and
compared. The computations conducted in Mathematica indicate the superiority
of Leap-Frog over Secant Method upon accounting the execution times.

References
1. do Carmo, M.P.: Differential Geometry of Curves and Surfaces. Prentice-Hall,
Engelwood Cliffs (1976)
2. de Boor, C.: A Practical Guide to Spline. Springer, Heidelberg (1985)
3. Kozera, R.: Curve modeling via interpolation based on multidimensional reduced
data. Studia Informatica 25(4B(61)), 1–140 (2004)
4. Noakes, L., Kozera, R.: Cumulative chords piecewise-quadratics and piecewise-
cubics. In: Klette, R., Kozera, R., Noakes, L., Wieckert, J. (eds.) Geometric Proper-
ties from Incomplete Data, Computational Imaging and Vision, vol. 31, pp. 59–75.
Kluver Academic Publisher (2006)
5. Epstein, M.P.: On the influence of parameterization in parametric interpolation.
SIAM J. Numer. Anal. 13, 261–268 (1976)
6. Kozera, R., Noakes, L.: Asymptotics for length and trajectory from cumulative
chord piecewise-quartics. Fund. Inform. 61(3–4), 267–283 (2004)
7. Kozera, R., Noakes, L.: Piecewise-quadratics and exponential parameterization for
reduced data. Appl. Math. Comput. 221, 620–638 (2013)
8. Kozera, R., Wilkolazka, M.: Convergence order in trajectory estimation by piece-
wise cubics and exponential parameterization. Math. Model. Anal. (in press)
9. Kozera, R., Noakes, L.: Piecewise-quadratics and ε-uniformly sampled reduced
data. Appl. Math. Inf. Sci. 10(1), 33–48 (2016)
10. Floater, M.S.: Chordal cubic spline interpolation is fourth order accurate. IMA J.
Numer. Anal. 26(1), 25–33 (2005)
11. Kozera, R., Noakes, L.: Piecewise-quadratics and reparameterizations for interpo-
lating reduced data. In: 17th International Workshop on Computer Algebra in Sci-
entific Computing. LNCS, vol. 9301, pp. 260–274. Springer, Cham (2015). https://
doi.org/10.1007/978-3-319-24021-3 20
Another random document with
no related content on Scribd:
"But oh! Remember, if there is no room for Christ in your heart, there will be no room
for you in Christ's heaven."

"My friend, He is knocking now; it may be His last knock. He is calling now; it may be
His last call."

"'Oh, let Me in.'" He cries, "'and I will make you happy; I am bringing you forgiveness,
and peace, and joy, and rest, and all that you need. Oh, let Me in before it is too late! I
have waited so patiently and so long, and still I wait. Will you not, even this night, undo
the door?'"

When the little service was over the people went back into their houses, and Angel and
her mother went on with their work. And as Angel wiped the cups and saucers, she
sang softly to herself the chorus of the hymn—

"Oh! My soul, for such a wonder,

Wilt thou not undo the door?"

"Yes, I will!" said her mother suddenly, bursting into tears; "I will undo the door; I will
keep Him waiting no longer."

CHAPTER V

ANGEL'S BIRTHDAY

IT was a bright, sunny morning, some weeks after that little service was held in
Pleasant Place.

The sunbeams were streaming in at Mrs. Blyth's window, for the cobwebs and spiders
had some time ago received notice to quit, and the dust had all been cleared away, and
found no chance of returning.

Mrs. Blyth was a different woman. Her troubles and trials remained, and she had just as
much to do, and just as many children to look after, but she herself was quite different.
She had opened the door of her heart, and the Lord Jesus had come in. And He had
brought sunshine with Him into that dark and ignorant heart. Life, instead of being a
burden and a weariness, was now full of interest to Mrs. Blyth, because she was trying
to do every little thing to please Jesus, who had done so much for her. Whether she was
washing the children, or cleaning the house, or turning the mangle, she tried to do it all
to please Him. She remembered that He was looking at her, and that He would be
pleased if she did it well. It was wonderful how that thought helped her, and how it
made the work easy and pleasant.
So, through the bright, clean window, the morning sunbeams were streaming on little
Angel's head. Her mother was standing by her side, watching her as she lay asleep, and
waiting for her to awake.

As soon as ever Angel opened her eyes, her mother said—

"Little Angel, do you know what to-day is?"

"No, mother," said Angel, rubbing her eyes, and sitting up in bed.

"It's your birthday, Angel; it is indeed!" said her mother. "I hunted it out in your
grandmother's old Bible. It's the day you were born, just seven years ago!"

"And am I really going to have a birthday, mother?" said Angel, in a very astonished
voice.

"Yes, a real good birthday," said her mother; "so get up and come downstairs, before
any of it is gone."

Angel was not long in putting on her clothes and coming down. She found the table put
quite ready for breakfast, with a clean tablecloth, and the mugs and plates set in order
for her and her little brothers and sisters; and in a little jar in the middle of the table
was a beautiful bunch of flowers. Real country flowers they were, evidently gathered
from some pleasant cottage garden far away. There were stocks and mignonette, and
southernwood, and sweetbrier, and a number of other flowers, the names of which
neither Angel nor her mother knew.

"Oh mother, mother," said little Angel, "what a beautiful nosegay!"

"It's for you, Angel," said her mother: "for your birthday. I got it at the early market.
My father always gave me a posy on my birthday."

"Oh, mother," said little Angel, "is it really for me?"

But that was not all, for by the side of Angel's plate she found a parcel. It was tied up in
brown paper, and there was a thick piece of string round it, fastened tightly in so many
knots that it took Angel a long time to open it. Her little hands quite shook with
excitement when at last she took off the cover and looked inside. It was a little book, in
a plain black binding.

"Oh, mother," said Angel, "what is it? Is it for my birthday?"

"Yes," said her mother; "look at the writing at the beginning. I'll read it to you."

It was very uneven writing, and very much blotted, for Mrs. Blyth was only a poor
scholar; but little Angel did not notice this—it seemed very wonderful to her to be able
to write at all.

Now, what was written in the little book was this:

"Given to little Angel by her dear mother; and she hopes she will promise to read it, and
will keep her promise better than I did."

"But I can't read, mother," said Angel.

"No; but you must learn," said her mother. "I mean that you shall go to school regular
now, Angel. Why, you're seven years old to-day!"

Poor little Angel's head was nearly turned; it was such a wonderful thing to have a
birthday.

But the wonders of the day were not over yet; for when, after breakfast, Angel asked
for the clothes to mangle, her mother said: "They're all done Angel; I'm just going to
take them home. I've done a lot these three nights when you was in bed, that we might
have a bit of a holiday to-day."

"A holiday, mother!" said Angel. "Oh, how nice! No mangling all day!"

"No mangling all day," repeated the mother, as if the thought were as pleasant to her as
to Angel.

But the wonders of the day were not yet over.

"Angel," said her mother, as they were washing the children, "did you ever see the
sea?"

"No, mother," said Angel; "but Tim has; he went last Easter Monday with his uncle."

"Well," said her mother, "if it doesn't rain, you shall see it to-day."

"Oh, mother!" was all that little Angel could say. And who do you think is going to take
you, child? "I don't know, mother."

"Why, Angel, your father is. He came in last night as soon as you'd gone to bed. He sat
down in that arm-chair by the fire, and he said, 'Dear me! how comfortable things is
just now at home! If they was always like this, I wouldn't stop out of an evening.'"

"So I said, 'If God helps me, John, they always shall be like this, and a deal better, too,
when the children gets a bit bigger.' And your father stopped at home and read his
newspaper, Angel, and then we had a bit of supper together. It was like when we was
first married, child; and as we ate our supper, Angel, I said, 'It's Angel's birthday to-
morrow, John.' And your father said, 'Is it? Why, to-morrow's Saturday. Let's all go to
the sea together;' and he took quite a handful of shillings out of his pocket. 'Here's
enough to pay,' he said. 'Have them all ready at dinner-time, and we'll go by the one-
o'clock train.'"

"Oh, mother," said little Angel, "it is so nice to have a birthday!"

True to his promise, John Blyth came home at dinner-time, with the shillings still in his
pocket. His mates had tried hard to persuade him to turn into the Blue Dragon on his
way home, but he told them he had an engagement, and had no time to stay.

What a happy afternoon that was!

Angel had never been in a train before, and her father took her on his knee, pointing
out to her the houses, and trees, and fields, and sheep, and cows, and horses, as they
went by. And then they arrived at the sea, and oh! What a great, wonderful sea it
seemed to Angel! She and her little brothers and sisters made houses in the sand, and
took off their shoes and stockings and waded in the water, and picked up quite a
basketful of all kinds of beautiful shells; whilst her father and mother sat, with the baby,
under the shadow of the cliffs and watched them.

And then they all came home together to tea, and her father never went out again that
night, but sat with them by the fire, and told Angel stories till it was time to go to bed.

"Oh, mother," said Angel again, a sleepy head on the pillow, "it is nice to have a
birthday!"

CHAPTER VI

THE GREAT BIRTHDAY

THE bells were ringing merrily from the tower of the old church close to Pleasant Place.

The street near the church was full of people bustling to and fro, going in and out of the
different shops, and hurrying along as if none of them had any time to lose. The shops
were unusually gay and tempting, for it was Christmas Eve. Even Pleasant Place looked
a little less dull than usual. There were sprigs of holly in some of the windows, and most
of the houses were a little cleaner and brighter than usual.

Angel and her mother had been very busy all day. They had just finished their
mangling, and had put all the clothes out of the way for Christmas Day, when they
heard a knock at the door, and Angel went to open it.

"It's a basket, mother," she said. "It can't be for us."

The man who had brought the basket laughed.

"It's for an Angel!" he said. "Have you got any of that article in here? Here's the
direction I was to bring it to—'Little Angel, No. 9, Pleasant Place.'"

"Then, please, it's for me," said Angel.

"For you!" said the man. "Well, to be sure! So you are the angel, are you? All right,
here's your basket!" And he was gone before they could ask more.

The basket was opened with some difficulty, for it was tightly tied up, and then Angel
and her mother put out the contents on the table amidst many exclamations.

There was first a plum-pudding, then a number of oranges and apples, then a large
cake, and then a pretty Christmas card, with a picture of a robin hopping about in the
snow, and these words printed on it, "A Happy Christmas to you all."

"Where can they all have come from?" said little Angel, as one good thing after another
came out of the basket. At the very bottom of the basket they found a tiny note.
"This will tell us about it," said Mrs. Blyth. "Why, it's directed to you, Angel!"

So Angel's mother sat down, stirred the fire, spelt it carefully out, and read it aloud by
the firelight.

"MY DEAR LITTLE ANGEL,"

"I send you a few little things for Christmas
Day. I hope you will have a very happy day. Do not
forget whose Birthday it is. Your friend,"
"MABEL DOUGLAS."

"Whose birthday is it, mother?" asked little Angel.

"The Lord Jesus Christ's," said her mother reverently. "Did I never tell you that, little
Angel? It's the day we think about Him being born a little baby at Bethlehem."

"SO YOU ARE THE ANGEL, ARE YOU? HERE'S YOUR BASKET."

Angel was sitting on her stool in front of the fire thinking, and it was some time before
she spoke again. Then she said suddenly, "What are you going to give Him, mother?"

"Give who, Angel?"

"What are you going to give the Lord Jesus for His birthday?"
"Oh, I don't know," said her mother. "I don't see how we can give Him anything."

"No," said little Angel sadly; "I've only got one penny,—that wouldn't buy anything good
enough. I would have liked to give Him something on His birthday; He did such a lot for
us."

"We can try to please Him, Angel," said her mother, "and do everything that we think
He would like."

"Yes," said little Angel, "we must try all day long."

That was a very happy Christmas Day for Angel and for her mother.

"This is the Lord Jesus' birthday," was Angel's first thought when she awoke in the
morning; and all through the day she was asking herself this question, "What would
Jesus like?" And whatever she thought He would like that she tried to do.

Angel's father was at home to dinner, and was very kind to her all day. He had not been
seen inside a public-house since Angel's birthday. It was a very good little Christmas
dinner. As they were eating it, Mr. Blyth said:

"Emily, have you seen those bills on the wall at the top of the court?"

Angel's mother said, "No; I have not been out to-day."

"There's to be a meeting to-night in that little schoolroom just a bit of way down the
street. That new young minister's going to speak; and it says on the bills it will all be
over in half an hour. I've a good mind to go and hear what he's got to say. Will you
come with me?"

"Yes, that I will," said Mrs. Blyth, with tears in her eyes. She had not been inside a
place of worship with her husband since the first year they were married.

"Can't Angel come too?" said her father, as he looked at her earnest little face.

"Not very well," said Mrs. Blyth; "we can't all go. Some one must stop with baby and
the children."

When Angel's large plum-pudding was put on the table, a sudden thought seized her.
"Mother," she whispered, "don't you think Jesus would like poor old Mrs. Sawyer to have
a bit of it?"

"Yes," said Mrs. Blyth, "I'll cut her a slice, and one for Annie too, poor girl. Will you take
them in?"

So Angel went next door with her two slices of plum-pudding. She found Mrs. Sawyer
and her niece Annie just beginning their dinner. There was nothing on the table but
some tea, and a loaf of bread with a few currants in it, so Angel felt very glad she had
brought the pudding. She was sure Jesus would be pleased they should have it; and she
thought it would make Him glad on His birthday to see how Mrs. Sawyer and Annie
smiled when they saw what she had brought them.

"Are you going to this meeting to-night?" said Annie, as Angel turned to go.

"No, I'm not going," said Angel; "but father and mother are. I must mind the children."
"I'll tell you what," said Annie; "if you'll bring them in here, I'll mind them. I can't leave
aunt, and they'll be a bit of company for her."

And so it came to pass that Pleasant Place beheld the wonderful sight of Mr. and Mrs.
Blyth and Angel all going together to the little meeting in the schoolroom.

A good many Pleasant Place people were there; and they looked round in astonishment
as Mr. Blyth came in, for they thought him about the most unlikely man in the whole
court to be there. And his wife and little Angel, as they sat beside him, prayed very
earnestly that he might get a blessing.

Mr. Douglas's text was a very strange one for Christmas Day—at least, so many of the
people thought when he gave it out. It had only four words, so that even little Angel
could remember it quite well—

"GIVE ME THINE HEART."

"Suppose," said the minister, "it was my birthday, and every one in my house was
keeping it. They all had a holiday and went out into the country, and there was a very
good dinner, which they all very much enjoyed, and altogether it was a very pleasant
day to them indeed."

"But suppose that I, whose birthday it was, was quite left out of it. No one gave me a
single present; no one even spoke to me; no one took the slightest notice of me. In
fact, all day long I was quite forgotten; I never once came into their thoughts."

"Nay, more. Not only did they do nothing whatever to give me pleasure, but they
seemed all day long to take a delight in doing the very things which they knew grieved
me and pained me, and were distressing to me."

"Surely, my friends, that would be a strange way of keeping my birthday; surely I

should feel very hurt by such conduct; surely it would be a perfect sham to pretend to
be keeping my birthday, and yet not take the slightest notice of me, except to annoy
and wound me! My friends," said the minister, "this afternoon I took a walk. In the
course of my walk I saw a number of people who pretended to be keeping a birthday.
And yet what were a great many of them doing? They were eating and drinking and
enjoying themselves, and having a merry time of it."

"But I noticed that the One whose birthday it was, was quite forgotten: they had not
given Him one single present all day long they had never once spoken to Him; all day
long He had never been in their thoughts; all day long He had been completely and
entirely passed by and forgotten."

"Nor was this all. I saw some who seemed to be taking a pleasure in doing the very
things He does not like, the very things which offend and grieve Him—drinking and
quarrelling, and taking His holy name in vain."

"And yet all these, my friends, pretended to be keeping the Lord Jesus Christ's
birthday!"

"But, I trust, by seeing you here to-night, that you have not been amongst their
number. I would therefore only put to you this one question—"
"The Lord Jesus Christ's birthday! Have you made Him a present to-day?"

"A present!" you say. "What can I give Him? He is the King of kings and Lord of lords.
What have I that is fit for a present to a king?"

"Give Him what He asks for, my friends. He says to you to-night, 'Give Me thine heart.'"

"That is the birthday present He is looking for. Will you hold it back?"

"Oh, think of what we are commemorating to-day. Think how He left His glory, and
came to be a poor, helpless babe for you; think, my friends, of all His wonderful love to
you. And then I would ask you, Can you refuse Him what He asks? Can you say—"

"Lord, I cannot give Thee my heart. I will give it to the world, to pleasure, to sin, to
Satan, but not to Thee,—no, not to Thee. I have no birthday present for Thee to-night?"

"Oh, will you not rather say—"

"'Lord, here is my heart; I bring it to Thee; take it for Thine own.

Cleanse it in Thy blood; make it fit to be Thine'"?

"Will you not this night lay at your King's feet the only birthday present you can give
Him—the only one He asks for—your heart?"

"Mother," said little Angel, as they walked home, "we can give Him a present, after all."

It was her father who answered her.

"Yes, Angel," he said, in a husky voice; "and we mustn't let Christmas Day pass before
we have done it."

And that night amongst the angels in heaven there was joy—joy over one sinner who
repented of the evil of his way, and laid at his Lord's feet a birthday present, even his
heart.

There was joy amongst the angels in heaven; and a little Angel on earth shared in their
joy.
 "PLEASE, MR. SOLEMN, WHEN YOU DIE,
WHO'LL HAVE TO DIG YOUR GRAVE?"

LITTLE DOT

CHAPTER I

OLD SOLOMON'S VISITOR

IT was a bright morning in spring, and the cemetery on the outskirts of the town looked
more peaceful, if possible, than it usually did. The dew was still on the grass, for it was
not yet nine o'clock. The violets and snowdrops on little children's graves were peeping
above the soil, and speaking of the resurrection. The robins were singing their sweetest
songs on the top of mossy gravestones—happy in the stillness of the place. And the
sunbeams were busy everywhere, sunning the flowers, lighting up the dewdrops, and
making everything glad and pleasant. Some of them even found their way into the deep
grave in which Solomon Whitaker, the old grave-digger, was working, and they made it
a little less dismal, and not quite so dark.
Not that old Whitaker thought it either dismal or dark. He had been a grave-digger
nearly all his life, so he looked upon grave-digging as his vocation, and thought it, on
the whole, more pleasant employment than that of most of his neighbours.

It was very quiet in the cemetery at all times, but especially in the early morning; and
the old man was not a little startled by hearing a very small voice speaking to him from
the top of the grave.

"What are you doing down there, old man?" said the little voice.

The grave-digger looked up quickly, and there, far above him, and peeping cautiously
into the grave, was a child in a clean white pinafore, and with a quantity of dark brown
hair hanging over her shoulders.

"Whoever in the world are you?" was his first question.

His voice sounded very awful, coming as it did out of the deep grave, and the child ran
away, and disappeared as suddenly as she had come.

Solomon looked up several times afterwards as he threw up fresh spadefuls of earth,

but for some time he saw no more of his little visitor. But she was not far away; she
was hiding behind a high tombstone, and in a few minutes she took courage, and went
again to the top of the grave. This time she did not speak, but stood with her finger in
her mouth, looking shyly down upon him, as her long brown hair blew wildly about in
the breeze.

Solomon thought he had never seen such a pretty little thing. He had had a little girl
once, and though she had been dead more than thirty years, he had not quite forgotten
her.

"What do they call you, my little dear?" said he, as gently as his husky old voice would
let him say it.

"Dot," said the child, nodding her head at him from the top of the grave.

"That's a very funny name," said Solomon. "I can't think on that I ever heard it afore."

"Dot isn't my real name; they call me Ruth in my father's big Bible on our parlour
table."

"That's got nothing to do with Dot as I can see," said the grave-digger musingly.

"No," she said, shaking her long brown hair out of her eyes; "it's 'cause I'm such a little
dot of a thing that they call me Dot."

"Oh, that's it, is it?" said Solomon; and then he went into a deep meditation on names,
and called to mind some strange ones which he read on the old churchyard
gravestones.

When Solomon was in one of his "reverdies," as his old wife used to call them when she
was alive, he seldom took much notice of what was going on around him, and he had
almost forgotten the little girl, when she said suddenly, in a half-frightened voice—

"I wonder what they call you, old man?"

"Solomon," said the grave-digger; "Mr. Solomon Whitaker—that's my name."

"Then, please, Mr. Solemn, what are you doing down there?"

"I'm digging a grave," said Solomon.

"What's it for, please, Mr. Solemn?" asked the child.

"Why, to bury folks in, of course," said the old man.

Little Dot retreated several steps when she heard this, as if she were afraid Mr. Solomon
might want to bury her. When he looked up again there was only a corner of her white
pinafore in sight. But as he went on quietly with his work, and took no notice of her, Dot
thought she might venture near again, for she wanted to ask Mr. Solomon another
question.

"Please," she began, "who are you going to put in that there hole?"

"It's a man as fell down dead last week. He was a hard-working fellow, that he was,"
said the grave-digger; for he always liked to give people a good word when digging
their graves.

Dot now seemed satisfied; and, on her side, told the old man that she had come to live
in one of the small cottages near the cemetery gates, and that they used to be "ever so
far off" in the country.

Then she ran away to another part of the cemetery, and old Solomon shaded his eyes
with his hand to watch her out of sight.

CHAPTER II

DOT'S DAISIES

DOT'S mother had lived all her life in a remote part of Yorkshire, far away from church
or chapel or any kind of school. But her husband had been born and brought up in a
town, and country life did not suit him. And so, when Dot was about five years old, he
returned to his native place, and took one of the cottages close to the cemetery, in
order that his little girl might still have some green grass on which to run about, and
might still see a few spring flowers.

The cemetery was some way out of the town; and Dot's mother, having had but little
education herself, did not think it at all necessary that Dot, at her tender age, should go
to school, and therefore the little girl was allowed to spend most of her time in the
cemetery, with which she was very well pleased. She liked to run round the
gravestones, and climb over the grassy mounds, and watch the robins hopping from
tree to tree.
But Dot's favourite place was by old Solomon's side. She went about with him from one
part of the cemetery to another, and he liked to feel her tiny hand in his. She took a
great interest, too, in the graves he was digging. She watched him shaping them neatly
and making them tidy, as he called it, until she began, as she fancied, to understand
grave-digging nearly as well as he did. But she sometimes puzzled the old man by her
questions, for Dot always wanted to know everything about what she saw.

"Mr. Solemn," she said one day, "shall you make me a little grave when I die?"

"Yes," he said, "I suppose I shall, little woman."

Dot thought this over for a long time.

"I don't want to go into a grave," she said; "it doesn't look nice."

"No," said the grave-digger, "you needn't be frightened; you won't have to go just yet.
Why, you're ever such a little mite of a thing!"

"Please, Mr. Solemn, when you die, who'll have to dig your grave, please?"

"I don't know," said Solomon uneasily; "they'll have to get a new digger, I suppose."

"Maybe you'd better dig one ready when you've a bit of time, Mr. Solemn."

But though Solomon was very fond of digging other people's graves—for he was so
much used to it that it had become quite a pleasure to him—he had no wish to dig his
own, nor did he like thinking about it, though Dot seemed as if she would not let him
forget it.

Another day, when he was working in a distant part of the cemetery, she asked him—

"Whereabouts will they bury you, Mr. Solemn?"

And when they were standing over a newly made grave, and Solomon was admiring his
work, she said—

"I hope they will make your grave neat, Mr. Solemn."

But though these questions and remarks made old Whitaker very uneasy—for he had a
sort of uncomfortable feeling in his heart when he thought of the day when his grave-
digging would come to an end—still, for all that, he liked little Dot, and he would have
missed the child much if anything had kept her from his side. She took such an interest
in his graves, too, and watched them growing deeper and deeper with as much pleasure
as he did himself. And, whether we be rich or poor, high or low, interest in our work
generally wins our hearts. And by and by Dot found herself a way, as she thought, of
helping old Solomon to make his graves look nice.

He was working one day at the bottom of a grave, and Dot was sitting on the grass at a
little distance. He thought she was busy with her doll, for she had not been talking to
him for a long time, and he gave a jump as he suddenly felt something patting on his
head, and heard Dot's merry little laugh at the top of the grave. She had filled her
pinafore with daisies, and thrown them upon him in the deep grave.
"Whatever in the world is that for?" said the old man, good-naturedly, as he shook the
flowers off his head.

"It's to make it pretty," said Dot. "It'll make it white and soft, you know, Mr. Solemn."

Solomon submitted very patiently; and from that time the child always gathered daisies
to scatter at the bottom of Solomon's graves, till he began to look upon it as a
necessary finish to his work. He often thought Dot was like a daisy herself, so fresh and
bright she was. He wondered at himself when he reckoned how much he loved her. For
his own little girl had been dead so many years; and it was so long now since he had
dug his old wife's grave, that Solomon had almost forgotten how to love. He had had no
one since to care for him, and he had cared for no one.

But little Dot had crept into his old heart unawares.

CHAPTER III

THE LITTLE GRAVE

OLD Solomon was digging a grave one day in a very quiet corner of the cemetery. Dot
was with him, as usual, prattling away in her pretty childish way.

"It's a tidy grave, is this," remarked the old man, as he smoothed the sides with his
spade; "nice and dry too; it'll do me credit."

"It's a very little one," said Dot.

"Yes, it's like to be little when it's for a little girl; you wouldn't want a very big grave,
Dot."

"No," said Dot; "but you would want a good big one, wouldn't you, Mr. Solemn?"

The mention of his own grave always made Solomon go into one of his "reverdies." But
he was recalled by Dot's asking quickly—

"Mr. Solemn, is she a very little girl?"

"Yes," said the old man; "maybe about your size, Dot. Her pa came about the grave. I
was in the office when he called, 'and,' said he, 'I want a nice quiet little corner, for it is
for my little girl.'"

"Did he look sorry?" said Dot.

"Yes," he said; "folks mostly do look sorry when they come about graves."
Dot had never watched the digging of a grave with so much interest as she did that of
this little girl. She never left Solomon's side, not even to play with her doll. She was
very quiet, too, as she stood with her large eyes wide open, watching all his
movements. He wondered what had come over her, and he looked up several times
rather anxiously as he threw up the spadefuls of earth.

"Mr. Solemn," she said, when he had finished, "when will they put the little girl in?"

"To-morrow morning," said the old man, "somewhere about eleven."

Dot nodded her head, and made up her mind she would be in this corner of the
cemetery at eleven o'clock.

When Solomon came back from his dinner, and went to take a last look at the little
grave, he found the bottom of it covered with white daisies which Dot had thrown in.

"She has made it pretty, bless her!" he murmured.

Dot crept behind the bushes near the chapel the next day, to watch the little girl's
funeral arrive. She saw the small coffin taken from the hearse, and carried on in front.
Then she watched the people get out of the carriages, and a lady and gentleman, whom
she felt sure were the little girl's father and mother, walked on first. The lady had her
handkerchief to her eyes, and Dot could see that she was crying. After her walked two
little girls, and they were crying also.

There were a few other people at the funeral, but Dot did not care to look at them; she
wanted to see what became of the little girl's coffin, which had just been carried into
the chapel. She waited patiently till they brought it out, and then she followed the
mournful procession at a little distance, till they reached the corner of the cemetery
where Solomon had dug the grave.

Solomon was there, standing by the grave, when the bearers came up with the coffin.
Dot could see him quite well, and she could see the minister standing at the end of the
grave, and all the people in a circle round it. She did not like to go very near, but she
could hear the minister reading something in a very solemn voice, and then the coffin
was let down into the grave. The little girl's mamma cried very much, and Dot cried too,
she felt so sorry for her.

When the service was over, they all looked into the grave, and then they walked away.
Dot ran up as soon as they were gone, and, taking hold of Solomon's hand, she peeped
into the grave. The little coffin was at the bottom, and some of Dot's daisies were lying
round it.

"Is the little girl inside there?" said Dot in an awestruck voice.

"Yes," said Solomon, "she's in there, poor thing. I'll have to fill it up now."

"Isn't it very dark?" said Dot.

"Isn't what dark?"

"In there," said Dot. "Isn't it very dark and cold for the poor little girl?"
"Oh, I don't know that," said Solomon. "I don't suppose folks feels cold when they are
dead; anyhow, we must cover her up warm."

But poor Dot's heart was very full; and, sitting on the grass beside the little girl's grave,
she began to cry and sob as if her heart would break.

"Don't cry, Dot," said the old man; "maybe the little girl knows nothing about it—maybe
she's asleep like."

But Dot's tears only flowed the faster. For she felt sure if the little girl were asleep, and
knew nothing about it, as old Solomon said, she would be waking up some day, and
then how dreadful it would be for her.

"Come, Dot," said Solomon at last, "I must fill it up."

Then Dot jumped up hastily. "Please, Mr. Solemn, wait one minute," she cried, as she
disappeared amongst the bushes.

"Whatever is she up to now?" said the old grave-digger.

She soon came back with her pinafore full of daisies. She had been gathering them all
the morning, and had hid them in a shady place under the trees. Then, with a little sob,
she threw them into the deep grave, and watched them fall on the little coffin. After this
she watched Solomon finish his work, and did not go home till the little girl's grave was
made, as old Solomon said, "all right and comfortable."

CHAPTER IV

LILIAN AND HER WORDS

DOT took a very great interest in "her little girl's grave," as she called it. She was up
early the next morning; and as soon as her mother had washed her, and given her her
breakfast, she ran to the quiet corner in the cemetery to look at the new-made grave. It
looked very bare, Dot thought, and she ran away to gather a number of daisies to
spread upon the top of it. She covered it as well as she could with them, and she patted
the sides of the grave with her little hands, to make it more smooth and tidy. Dot
wondered if the little girl knew what she was doing, and if it made her any happier to
know there were daisies above her.

She thought she would ask Solomon; so when she had finished she went in search of
him. He was not far away, and she begged him to come and look at what she had done
to her little girl's grave. He took hold of Dot's hand, and she led him to the place.

"See, Mr. Solemn," she said, "haven't I made my little girl pretty?"

"Aye," he answered; "you have found a many daisies, Dot."

"But, Mr. Solemn," asked Dot anxiously, "do you think she knows?"

"Why, Dot, I don't know—maybe she does," he said, for he did not like to disappoint
her.

"Mr. Solemn, shall I put you some daisies at the top of your grave?" said Dot, as they
walked away.

Solomon made no answer. Dot had reminded him so often of his own grave, that he had
sometimes begun to think about it, and to wonder how long it would be before it would
have to be made. He had a vague idea that when he was buried, he would not come to
an end.

He had heard of heaven and of hell; and though he had never thought much about
either of them, he had a kind of feeling that some day he must go to one or other. Hell,
he had heard, was for bad people, and heaven for good ones; and though Solomon tried
to persuade himself that he belonged to the latter class, he could not quite come to that
opinion. There was something in his heart which told him all was not right with him, and
made the subject an unpleasant one. He wished Dot would let it drop, and not talk to
him any more about it; and then he went into a reverie about Dot, and Dot's daisies,
and all her pretty ways.

It was the afternoon of the same day, and Dot was sitting beside her little girl's grave,
trying to make the daisies look more pretty by putting some leaves among them, when
she heard footsteps crossing the broad gravel path. She jumped up, and peeped behind
the trees to see who was coming. It was the lady and gentleman whom she had seen at
the funeral, and they were coming to look at their little grave. Dot felt very shy, but she
could not run away without meeting them, so she hid behind a hawthorn bush at the
other side.

The little girl's papa and mamma came close to the grave, and Dot was so near that, as
they knelt down beside it, she could hear a great deal of what they were saying. The
lady was crying very much, and for some time she did not speak. But the gentleman
said—

"I wonder who has put those flowers here, my dear; how very pretty they are!"

"Yes," said the lady, through her tears; "and the grave was full of them yesterday."

"How pleased our little girl would have been!" said he. "She was so fond of daisies! Who
can have done it?"

Little Dot heard all this from her hiding-place, and she felt very pleased that she had
made her little girl's grave so pretty.

The lady cried a great deal as she sat by the grave; but just before they left, Dot heard
the gentleman say—

"Don't cry, dearest; remember what our little Lilian said the night before she died."

"Yes," said the lady, "I will not forget."

And she dried her eyes, and Dot thought she tried to smile as she looked up at the blue
sky. Then she took a bunch of white violets which she had brought with her, and put
them in the middle of the grave, but she did not move any of Dot's daisies, at which she
looked very lovingly and tenderly.

As soon as they were gone, Dot came out from behind the hawthorn bush. She went up
to her little girl's grave, and kneeling on the grass beside it she smelt the white violets
and stroked them with her tiny hand. They made it look so much nicer, she thought; but
she felt very glad that the lady had liked her daisies. She would gather some fresh ones
to-morrow.

Dot walked home very slowly. She had so much to think over. She knew her little girl's
name now, and that she was fond of daisies. She would not forget that. Dot felt very
sorry for the poor lady; she wished she could tell her so. And then she began to wonder
what it was that her little girl had said the night before she died. It must be something
nice, Dot thought, to make the lady wipe her eyes and try to smile. Perhaps the little
girl had said she did not mind being put into the dark hole. Dot thought it could hardly
be that, for she felt sure she would mind it very much indeed. Dot was sure she would
be very frightened if she had to die, and old Solomon had to dig a grave for her. No, it
could not be that which Lilian had said. Perhaps Solomon was right, and the little girl
was asleep. If so, Dot hoped it would be a long, long time before she woke up again.

Solomon had left his work, or Dot would have told him about what she had seen. But it
was tea-time now, and she must go home. Her mother was standing at the door looking
out for her, and she called to the child to be quick and come in to tea.

Dot found her father at home, and they began their meal. But little Dot was so quiet,
and sat so still, that her father asked her what was the matter. Then she thought she
would ask him what she wanted to know, for he was very kind to her, and generally
tried to answer her questions.

So Dot told him about her little girl's grave, and what the lady and gentleman had
talked about, and she asked what he thought the little girl had said, which had made
her mother stop crying.

But Dot's father could not tell her. And when Dot said she was sure she would not like to
be put in a hole like that, her father only laughed, and told her not to trouble her little
head about it: she was too young to think of such things.

"But my little girl was only just about as big as me," said Dot, "'cause Mr. Solemn told
me so."

That was an argument which her father could not answer, so he told Dot to be quick
over her supper, and get to bed. And when she was asleep, he said to his wife that he
did not think the cemetery was a good place for his little girl to play in—it made her
gloomy. But Dot's mother said it was better than the street, and Dot was too light-
hearted to be dull long.

And whilst they were talking little Dot was dreaming of Lilian, and of what she had said
the night before she died.
 CHAPTER V

DOT'S BUSY THOUGHTS

A DAY or two after, as Dot was putting fresh daisies on the little grave, she felt a hand
on her shoulder, and looking up she saw her little girl's mamma. She had come up very
quietly, and Dot was so intent on what she was doing that she had not heard her. It was
too late to run away; but the lady's face was so kind and loving that the child could not
be afraid. She took hold of Dot's little hand, and sat down beside her, and then she said
very gently—

"Is this the little girl who gathered the daisies?"

"Yes," said Dot shyly, "it was me."

The lady seemed very pleased, and she asked Dot what her name was, and where she
lived. Then she said—

"Dot, what was it made you bring these pretty flowers here?"

"Please," said the child, "it was 'cause Mr. Solemn said she was ever such a little girl—
maybe about as big as me."

"Who is Mr. Solemn?" asked the lady.

 "IS THIS THE LITTLE GIRL WHO GATHERED THE DAISIES?"

"It's an old man—him as digs the graves; he made my little girl's grave," said Dot,
under her breath, "and he filled it up and all."

The tears came into the lady's eyes, and she stooped down and kissed the child.

Dot was beginning to feel quite at home with the little girl's mamma, and she stroked
the lady's soft glove with her tiny hand.

They sat quite still for some time. Dot never moved, and the lady had almost forgotten
her—she was thinking of her own little girl. The tears began to run down her cheeks,
though she tried to keep them back, and some of them fell upon Dot as she sat at her
feet.

"I was thinking of my little girl," said the lady, as Dot looked sorrowfully up to her face.

"Please," said Dot, "I wonder what your little girl said to you the night before she died?"
She thought perhaps it might comfort the lady to think of it, as it had done so the other
day.

The lady looked very surprised when Dot said this, as she had had no idea that the little
girl was near when she was talking to her husband.

"How did you know, Dot?" she asked.

"Please, I couldn't help it," said little Dot; "I was putting the daisies."

"Yes?" said the lady, and she waited for the child to go on.

"And I ran in there," said Dot, nodding at the hawthorn bush. "I heard you—and,
please, don't be angry."

"I am not angry," said the lady.

Dot looked in her face, and saw she was gazing at her with a very sweet smile.

"Then, please," said little Dot, "I would like very much to know what the little girl said."

"I will tell you, Dot," said the lady. "Come and sit on my knee."

There was a flat tombstone close by, on which they sat whilst the girl's mamma talked
to Dot. She found it very hard to speak about her child, it was so short a time since she
had died. But she tried her very best, for the sake of the little girl who had covered the
grave with daisies.

"Lilian was only ill a very short time," said the lady; "a week before she died she was
running about and playing—just as you have been doing to-day, Dot. But she took a bad
cold, and soon the doctor told me my little girl must die."

"Oh," said Dot, with a little sob, "I am so sorry for the poor little girl!"

"Lilian wasn't afraid to die, Dot," said the lady.

"Wasn't she?" said Dot. "I should be frightened ever so much—but maybe she'd never
seen Mr. Solemn bury anybody; maybe she didn't know she had to go into that dark
hole."

"Listen, Dot," said the lady, "and I will tell you what my little girl said the night before
she died."

"'Mamma,' she said, 'don't let Violet and Ethel think that I'm down deep in the
cemetery; but take them out, and show them the blue sky and all the white clouds, and
tell them, Little sister Lilian's up there with Jesus.' Violet and Ethel are my other little
girls, Dot."

"Yes," said Dot, in a whisper; "I saw them at the funeral."

"That is what my little girl said, which made me stop crying the other day."

Dot looked very puzzled. There was a great deal that she wanted to think over and to
ask Solomon about.

The lady was obliged to go home, for it was getting late. She kissed the child before she
went, and said she hoped Dot would see her little girl one day, above the blue sky.

Dot could not make out what the lady meant, nor what her little girl had meant the
night before she died. She wanted very much to hear more about her, and she hoped
the lady would soon come again.

"Mr. Solemn," said Dot the next day, as she was in her usual place on the top of one of
Solomon's graves, "didn't you say that my little girl was in that long box?"

"Yes," said Solomon—"yes, Dot, I said so, I believe."

"But my little girl's mamma says she isn't in there, Mr. Solemn, and my little girl said so
the night before she died."

"Where is she, then?" said Solomon.

"She's somewhere up there," said Dot, pointing with her finger to the blue sky.

"Oh, in heaven," said Solomon. "Yes, Dot, I suppose she is in heaven."

"How did she get there?" said Dot. "I want to know all about it, Mr. Solemn."

"Oh, I don't know," said the old man. "Good folks always go to heaven."

"Shall you go to heaven, Mr. Solemn, when you die?"

"I hope I shall, Dot, I'm sure," said the old man. "But there, run away a little; I want to
tidy round a bit."

Now, Solomon had very often "tidied round," as he called it, without sending little Dot
away; but he did not want her to ask him any more questions, and he hoped she would
forget it before she came back.

But Dot had not forgotten. She had not even been playing; she had been sitting on an
old tombstone, thinking about what Solomon had said. And as soon as he had finished