You are on page 1of 54

Applications of Evolutionary

Computation 21st International


Conference EvoApplications 2018
Parma Italy April 4 6 2018 Proceedings
1st Edition Kevin Sim
Visit to download the full and correct content document:
https://textbookfull.com/product/applications-of-evolutionary-computation-21st-internat
ional-conference-evoapplications-2018-parma-italy-april-4-6-2018-proceedings-1st-ed
ition-kevin-sim/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Artificial Life and Evolutionary Computation 13th


Italian Workshop WIVACE 2018 Parma Italy September 10
12 2018 Revised Selected Papers Stefano Cagnoni

https://textbookfull.com/product/artificial-life-and-
evolutionary-computation-13th-italian-workshop-wivace-2018-parma-
italy-september-10-12-2018-revised-selected-papers-stefano-
cagnoni/

Applications of Evolutionary Computation 19th European


Conference EvoApplications 2016 Porto Portugal March 30
April 1 2016 Proceedings Part I 1st Edition Giovanni
Squillero
https://textbookfull.com/product/applications-of-evolutionary-
computation-19th-european-conference-evoapplications-2016-porto-
portugal-march-30-april-1-2016-proceedings-part-i-1st-edition-
giovanni-squillero/

Big Data Cloud and Applications Third International


Conference BDCA 2018 Kenitra Morocco April 4 5 2018
Revised Selected Papers Youness Tabii

https://textbookfull.com/product/big-data-cloud-and-applications-
third-international-conference-bdca-2018-kenitra-morocco-
april-4-5-2018-revised-selected-papers-youness-tabii/

Reversible Computation 10th International Conference RC


2018 Leicester UK September 12 14 2018 Proceedings
Jarkko Kari

https://textbookfull.com/product/reversible-computation-10th-
international-conference-rc-2018-leicester-uk-
september-12-14-2018-proceedings-jarkko-kari/
Dependable Software Engineering Theories Tools and
Applications 4th International Symposium SETTA 2018
Beijing China September 4 6 2018 Proceedings Xinyu Feng

https://textbookfull.com/product/dependable-software-engineering-
theories-tools-and-applications-4th-international-symposium-
setta-2018-beijing-china-september-4-6-2018-proceedings-xinyu-
feng/

Swarm Intelligence 11th International Conference ANTS


2018 Rome Italy October 29 31 2018 Proceedings Marco
Dorigo

https://textbookfull.com/product/swarm-intelligence-11th-
international-conference-ants-2018-rome-italy-
october-29-31-2018-proceedings-marco-dorigo/

Product Lifecycle Management to Support Industry 4 0


15th IFIP WG 5 1 International Conference PLM 2018
Turin Italy July 2 4 2018 Proceedings Paolo Chiabert

https://textbookfull.com/product/product-lifecycle-management-to-
support-industry-4-0-15th-ifip-wg-5-1-international-conference-
plm-2018-turin-italy-july-2-4-2018-proceedings-paolo-chiabert/

Discovery Science 21st International Conference DS 2018


Limassol Cyprus October 29 31 2018 Proceedings Larisa
Soldatova

https://textbookfull.com/product/discovery-science-21st-
international-conference-ds-2018-limassol-cyprus-
october-29-31-2018-proceedings-larisa-soldatova/

Database and Expert Systems Applications 29th


International Conference DEXA 2018 Regensburg Germany
September 3 6 2018 Proceedings Part II Sven Hartmann

https://textbookfull.com/product/database-and-expert-systems-
applications-29th-international-conference-dexa-2018-regensburg-
germany-september-3-6-2018-proceedings-part-ii-sven-hartmann/
Kevin Sim
Paul Kaufmann et al. (Eds.)
LNCS 10784

Applications of
Evolutionary Computation
21st International Conference, EvoApplications 2018
Parma, Italy, April 4–6, 2018
Proceedings

123
Lecture Notes in Computer Science 10784
Commenced Publication in 1973
Founding and Former Series Editors:
Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen

Editorial Board
David Hutchison
Lancaster University, Lancaster, UK
Takeo Kanade
Carnegie Mellon University, Pittsburgh, PA, USA
Josef Kittler
University of Surrey, Guildford, UK
Jon M. Kleinberg
Cornell University, Ithaca, NY, USA
Friedemann Mattern
ETH Zurich, Zurich, Switzerland
John C. Mitchell
Stanford University, Stanford, CA, USA
Moni Naor
Weizmann Institute of Science, Rehovot, Israel
C. Pandu Rangan
Indian Institute of Technology Madras, Chennai, India
Bernhard Steffen
TU Dortmund University, Dortmund, Germany
Demetri Terzopoulos
University of California, Los Angeles, CA, USA
Doug Tygar
University of California, Berkeley, CA, USA
Gerhard Weikum
Max Planck Institute for Informatics, Saarbrücken, Germany
More information about this series at http://www.springer.com/series/7407
Kevin Sim Paul Kaufmann et al. (Eds.)

Applications of
Evolutionary Computation
21st International Conference, EvoApplications 2018
Parma, Italy, April 4–6, 2018
Proceedings

123
Editors

see next page

ISSN 0302-9743 ISSN 1611-3349 (electronic)


Lecture Notes in Computer Science
ISBN 978-3-319-77537-1 ISBN 978-3-319-77538-8 (eBook)
https://doi.org/10.1007/978-3-319-77538-8

Library of Congress Control Number: 2018935897

LNCS Sublibrary: SL1 – Theoretical Computer Science and General Issues

© Springer International Publishing AG, part of Springer Nature 2018


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.

Printed on acid-free paper

This Springer imprint is published by the registered company Springer International Publishing AG
part of Springer Nature
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Volume Editors

Kevin Sim Francisco Fernández de Vega


Edinburgh Napier University University of Extremadura
UK Spain
k.sim@napier.ac.uk fcofdez@unex.es
Paul Kaufmann Kyrre Glette
Mainz University University of Oslo
Germany Norway
paul.kaufmann@gmail.com kyrrehg@ifi.uio.no
Gerd Ascheid J. Ignacio Hidalgo
RWTH Aachen University Universidad Complutense de Madrid
Germany Spain
gerd.ascheid@ice.rwth-aachen.de hidalgo@ucm.es
Jaume Bacardit Julien Hubert
Newcastle University Vrije Universiteit Amsterdam
UK Netherlands
jaume.bacardit@newcastle.ac.uk jghubert@gmail.com
Stefano Cagnoni Giovanni Iacca
University of Parma RWTH Aachen University
Italy Germany
cagnoni@ce.unipr.it giovanni.iacca@gmail.com
Carlos Cotta Oliver Kramer
Universidad de Málaga University of Oldenburg
Spain Germany
ccottap@lcc.uma.es oliver.kramer@uni-oldenburg.de
Fabio D’Andreagiovanni Michalis Mavrovouniotis
CNRS, UTC - Sorbonne University Nottingham Trent University
France UK
d.andreagiovanni@hds.utc.fr michalis.mavrovouniotis@ntu.ac.uk
Federico Divina Antonio M. Mora García
Universidad Pablo de Olavide Sevilla Universidad Internacional de La Rioja
Spain Spain
fdivina@upo.es antoniomiguel.mora@unir.net
Anna I. Esparcia-Alcázar Trung Thanh Nguyen
Universitat Politècnica de València Liverpool John Moores University
Spain UK
esparcia@upv.es T.T.Nguyen@ljmu.ac.uk
VI Volume Editors

Robert Schaefer Neil Urquhart


AGH University of Science Edinburgh Napier University
and Technology UK
Poland n.urquhart@napier.ac.uk
schaefer@agh.edu.pl
Mengjie Zhang
Sara Silva Victoria University of Wellington
Faculdade de Ciências, Universidade New Zealand
de Lisboa mengjie.zhang@ecs.vuw.ac.nz
Portugal
sara@fc.ul.pt
Alberto Tonda
INRA
France
alberto.tonda@grignon.inra.fr
Preface

This volume contains the proceedings of EvoApplications 2018, the International


Conference on the Applications of Evolutionary Computation. The event was held in
Parma, Italy, during April 4–6.
EvoAPPS, as it is familiarly called, is part of Evo*, the leading event on bio-inspired
computation in Europe. EvoAPPS aimed to show the applications of the research,
ranging from proofs of concept to industrial case studies. At the same time, under the
Evo* umbrella, EuroGP focused on the technique of genetic programming, EvoCOP
targeted evolutionary computation in combinatorial optimization, and EvoMUSART
was dedicated to evolved and bio-inspired music, sound, art and design. The pro-
ceedings for all of these co-located events are available in the LNCS series.
This edition combines research from 14 different domains: business analytics and
finance (EvoBAFIN track); computational biology (EvoBIO track); communication
networks and other parallel and distributed systems (EvoCOMNET track); complex
systems (EvoCOMPLEX track); energy-related optimization (EvoENERGY track);
games and multi-agent systems (EvoGAMES track); image analysis, signal processing,
and pattern recognition (EvoIASP track); real-world industrial and commercial envi-
ronments (EvoINDUSTRY track); knowledge incorporation in evolutionary compu-
tation (EvoKNOW track); continuous parameter optimization (EvoNUM track);
parallel architectures and distributed infrastructures (EvoPAR track); evolutionary
robotics (EvoROBOT track); nature-inspired algorithms in software engineering and
testing (EvoSET track); and stochastic and dynamic environments (EvoSTOC track).
This year we received 84 high-quality submissions, most of them well suited to fit in
more than one track. We selected 36 papers for full oral presentation, while a further 23
works were presented as posters. All contributions, regardless of the presentation
format, appear as full papers in this volume (LNCS 10784).
Many people contributed to this edition: We express our gratitude to the authors for
submitting their works, and to the members of the Program Committees for devoting
such a big effort to review papers pressed by our tight schedule.
The papers were submitted, reviewed, and selected through the MyReview
conference management system, and we are grateful to Marc Schoenauer (Inria,
Saclay-Île-de-France, France) for providing, hosting, and managing the platform.
We would also like to thank the local organizing team led by Stefano Cagnoni and
Monica Mordonini from the University of Parma, Italy, for providing such an enticing
venue and arranging an array of additional activities for delegates.
We would like to acknowledge Pablo García Sánchez (University of Cádiz, Spain)
for his continued support in maintaining the Evo* website and handling publicity.
We credit the invited keynote speakers, Una-May O’Reilly (MIT Computer Science
and Artificial Intelligence Laboratory, USA) and Penousal Machado (Computational
Design and Visualization Lab at the University of Coimbra, Portugal), for their
fascinating and inspiring presentations.
VIII Preface

We are grateful to the support provided by SPECIES, the Society for the Promotion
of Evolutionary Computation in Europe and Its Surroundings, and its individual
members (Marc Schoenauer, President; Anna I. Esparcia-Alcázar, Secretary and
Vice-President; Wolfgang Banzhaf, Treasurer) for the coordination and financial
administration.
And last but not least, we express our continued appreciation to Jennifer Willies for
her ongoing support and expertise as one of the founding organizers of Evo* and also
to Anna I Esparcia-Alcázar, from Universitat Politècnica de València, Spain, whose
considerable efforts in managing and coordinating Evo* helped to build our unique,
vibrant, and friendly atmosphere.

February 2018 Kevin Sim Giovanni Iacca


Paul Kaufmann Michael Kampouridis
Gerd Ascheid Oliver Kramer
Jaume Bacardit Michalis Mavrovouniotis
Stefano Cagnoni Antonio M. Mora García
Carlos Cotta Trung Thanh Nguyen
Fabio D’Andreagiovanni Fernando Otero
Federico Divina Robert Schaefer
Anna I. Esparcia-Alcázar Sara Silva
Francisco Fernández de Vega Alberto Tonda
Kyrre Glette Neil Urquhart
Julien Hubert Mengjie Zhang
J. Ignacio Hidalgo
Organization

EvoApplications Coordinator
Kevin Sim Edinburgh Napier University, UK

EvoApplications Publication Chair


Paul Kaufmann Mainz University, Germany

Local Chairs
Stefano Cagnoni University of Parma, Italy
Monica Mordonini University of Parma, Italy

Publicity Chair
Pablo García Sánchez University of Cádiz, Spain

EvoBAFIN Chairs
Michael Kampouridis University of Kent, UK
Fernando Otero University of Kent, UK

EvoBIO Chairs
Federico Divina Universidad Pablo de Olavide, Seville, Spain
Jaume Bacardit Newcastle University, UK

EvoCOMNET Chairs
Fabio D’Andreagiovanni CNRS, UTC - Sorbonne University, France
Giovanni Iacca RWTH Aachen University, Germany

EvoCOMPLEX Chairs
Carlos Cotta Universidad de Málaga, Spain
Robert Schaefer AGH University of Science and Technology, Poland

EvoENERGY Chairs
Paul Kaufmann Mainz University, Germany
Oliver Kramer University of Oldenburg, Germany
X Organization

EvoGAMES Chairs
Antonio M. Mora García Universidad Internacional de La Rioja, Spain
Alberto Tonda INRA, France

EvoIASP Chairs
Stefano Cagnoni University of Parma, Italy
Mengjie Zhang Victoria University of Wellington, New Zealand

EvoINDUSTRY Chairs
Kevin Sim Edinburgh Napier University, UK
Neil Urquhart Edinburgh Napier University, UK

EvoKNOW Chairs
Giovanni Iacca RWTH Aachen University, Germany
Gerd Ascheid RWTH Aachen University, Germany

EvoNUM Chair
Anna I. Esparcia-Alcázar Universitat Politècnica de València, Spain

EvoPAR Chairs
Francisco Fernández University of Extremadura, Spain
de Vega
J. Ignacio Hidalgo Universidad Complutense de Madrid, Spain

EvoROBOT Chairs
Kyrre Glette University of Oslo, Norway
Julien Hubert Vrije Universiteit Amsterdam, The Netherlands

EvoSET Chairs
Anna I. Esparcia-Alcázar Universitat Politècnica de València, Spain
Sara Silva Faculdade de Ciências, Universidade de Lisboa,
Portugal
Organization XI

EvoSTOC Chairs
Michalis Mavrovouniotis Nottingham Trent University, UK
Trung Thanh Nguyen Liverpool John Moores University, UK

Program Committees
Bahriye Basturk Akay Erciyes University, Turkey [EvoINDUSTRY]
Jhon Amaya UNET, Venezuela [EvoCOMPLEX]
Jacopo Aleotti University of Parma, Italy [EvoIASP]
Michele Amoretti University of Parma, Italy [EvoIASP]
Anca Andreica Universitatea Babeş-Bolyai, Romania
[EvoCOMPLEX]
Jarosław Arabas Warsaw University of Technology, Poland
[EvoKNOW]
Antonio Fernández Ares Universidad de Granada, Spain [EvoGAMES]
Ignacio Arnaldo PatternEx, USA [EvoPAR]
María Arsuaga-Ríos CERN [EvoINDUSTRY]
Jason Atkin University of Nottingham, UK [EvoINDUSTRY]
Joshua Auerbach Champlain College, USA [EvoROBOT]
Lucia Ballerini University of Edinburgh, UK [EvoIASP]
Tiago Baptista Universidade de Coimbra, Portugal [EvoCOMPLEX]
Thomas Bauschert Technical University Chemnitz, Germany
[EvoCOMNET]
Vitoantonio Bevilacqua Politecnico di Bari, Italy [EvoIASP]
Hans-Georg Beyer Vorarlberg University of Applied Sciences, Austria
[EvoNUM]
Leonardo Bocchi University of Florence, Italy [EvoIASP]
János Botzheim Tokyo Metropolitan University, Japan [EvoKNOW]
Juergen Branke University of Warwick, UK [EvoSTOC]
Nicolas Bredeche Institut des Systèmes Intelligents et de Robotique,
France [EvoROBOT]
Jörg Bremer University of Oldenburg, Germany [EvoENERGY]
Cédric Buche ENIB, France [EvoGAMES]
Doina Bucur University of Twente, The Netherlands [EvoCOMNET,
EvoKNOW]
Aleksander Byrski AGH University of Science and Technology, Poland
[EvoCOMPLEX]
Raúl Lara Cabrera Universidad Autónoma de Madrid, Spain
[EvoGAMES]
David Camacho Universidad Autónoma de Madrid, Spain
[EvoGAMES]
Fabio Caraffini De Montfort University, UK [EvoKNOW]
Hui Cheng Liverpool John Moores University, UK [EvoSTOC]
Luca Chiaraviglio University of Rome Tor Vergata, Italy [EvoCOMNET]
Francisco Chicano Universidad de Málaga, Spain [EvoSET]
XII Organization

Anders Christensen University Institute of Lisbon, ISCTE-IUL, Portugal


[EvoROBOT]
Antonio Della Cioppa University of Salerno, Italy [EvoIASP]
Myra Cohen University of Nebraska, USA [EvoSET]
José Manuel Colmenar Universidad Rey Juan Carlos, Spain [EvoPAR]
Stefano Coniglio University of Southampton, UK [EvoCOMNET]
Ernesto Costa University of Coimbra, Portugal [EvoSTOC]
Sam Cramer University of Kent, UK [EvoBAFIN]
Antonio Córdoba Universidad de Sevilla, Spain [EvoCOMPLEX]
Anthony Clark Michigan State University, USA [EvoROBOT]
Christian Darabos University of Pennsylvania, USA [EvoBIO]
Stephane Doncieux Institut des Systèmes Intelligents et de Robotique,
France [EvoROBOT]
Bernabé Dorronsoro Universidad de Cádiz, Spain [EvoCOMPLEX]
Jitesh Dundas Indian Institute of Technology, India [EvoBIO]
Marc Ebner Ernst Moritz Arndt University, Greifswald, Germany
[EvoIASP]
Aniko Ekart Aston University, UK [EvoINDUSTRY]
Kai Olav Ellefsen University of Oslo, Norway [EvoROBOT]
Andries P. Engelbrecht University of Pretoria, South Africa [EvoSTOC]
Şima Etaner-Uyar Istanbul Technical University, Turkey [EvoNUM]
Ivanoe De Falco ICAR - CNR, Italy [EvoIASP]
Thomas Farrenkopf Technische Hochschule Mittelhessen, Germany
[EvoINDUSTRY]
Carlos Fernandes University of Lisbon, Portugal [EvoCOMPLEX]
Gianluigui Folino ICAR-CNR, Italy [EvoPAR]
Francesco Fontanella University of Cassino, Italy [EvoIASP]
Gordon Fraser University of Sheffield, UK [EvoSET]
Alex Freitas University of Kent, UK [EvoBIO]
Mario Garza Liverpool John Moores University, UK [EvoSTOC]
Gregory Gay University of South Carolina, USA [EvoSET]
Mario Giacobini Universitá di Torino, Italy [EvoBIO]
Raffaele Giancarlo Università degli Studi di Palermo, Italy [EvoBIO]
Rosalba Giugno University of Verona, Italy [EvoBIO]
Michael Guckert Technische Hochschule Mittelhessen, Germany
[EvoINDUSTRY]
Evert Haasdijk VU University Amsterdam, The Netherlands
[EvoROBOT]
Ahmed Hallawa RWTH Aachen University, Germany [EvoKNOW]
Heiko Hamann University of Lübeck, Germany [EvoROBOT]
Jin-Kao Hao University of Angers, France [EvoBIO]
Emma Hart Edinburgh Napier University, UK [EvoINDUSTRY]
Jacqueline Heinerman VU University Amsterdam, The Netherlands
[EvoROBOT]
Daniel Hernández Instituto Tecnológico Nacional, Mexico [EvoPAR]
Organization XIII

Malcom Heywood Dalhousie University, Canada [EvoBAFIN]


Ronald Hochreiter WU Vienna University of Economics and Business,
Austria [EvoBAFIN]
Rolf Hoffmann Technical University Darmstadt, Germany
[EvoCOMNET]
Ting Hu Memorial University, Canada [EvoBIO]
Joost Huizinga University of Wyoming, USA [EvoROBOT]
Andreas Kassler Karlstad University, Sweden [EvoCOMNET]
Shayan Kavakeb AECOM, UK [EvoSTOC]
Graham Kendall University of Nottingham, UK [EvoINDUSTRY]
Mario Koeppen Kyushu Institute of Technology, Japan [EvoIASP]
Wacław Kuś Silesian University of Technology, Poland
[EvoCOMPLEX]
Fergal Lane University of Limerick, Ireland [EvoKNOW]
William B. Langdon University College London, UK [EvoNUM, EvoPAR]
Juan Luis Jiménez Laredo Université du Havre, France [EvoCOMPLEX,
EvoPAR]
Antonio Fernández Leiva Universidad de Málaga, Spain [EvoGAMES]
Charly Lersteau Liverpool John Moores University, UK [EvoSTOC]
Changhe Li China University of Geosciences, China [EvoSTOC]
Antonios Liapis University of Malta, Malta [EvoGAMES]
Federico Liberatore Universidad Carlos III, Spain [EvoGAMES]
Piotr Lipinski University of Wroclaw, Poland [EvoBAFIN]
Francisco Luna Universidad de Málaga, Spain [EvoPAR]
Evelyne Lutton INRA, France [EvoIASP]
Chenjie Ma Fraunhofer Institute for Wind Energy and Energy
System Technology, Germany [EvoENERGY]
Tobias Mahlmann Lund University, Sweden [EvoGAMES]
Carlo Mannino Sintef, Norway [EvoCOMNET]
Elena Marchiori Radboud Universiteit van Nijmegen, The Netherlands
[EvoBIO]
Ingo Mauser Karlsruhe Institute of Technology, Germany
[EvoENERGY]
Michalis Mavrovouniotis Nottingham Trent University, UK [EvoSTOC]
Michael Mayo University of Waikato, New Zealand [EvoBAFIN]
Vinícius Veloso de Melo UNIFESP-SJC, Brazil [EvoKNOW]
Tim Menzies University of Nebraska, USA [EvoSET]
Juan Julián Merelo Universidad de Granada, Spain [EvoGAMES,
EvoCOMPLEX, EvoNUM]
Martin Middendorf University of Leipzig, Germany [EvoENERGY]
Wiem Mkaouer University of Michigan, USA [EvoSET]
Maizura Mokhtar Heriot-Watt University, UK [EvoENERGY]
Jean-Marc Montanier Softbank Robotics Europe, France [EvoROBOT]
Roberto Montemanni IDSIA, Switzerland [EvoCOMNET]
Jared Moore Grand Valley State University, USA [EvoROBOT]
Vincent Moulton University of East Anglia, UK [EvoBIO]
XIV Organization

Jean-Baptiste Mouret Inria Larsen Team, France [EvoROBOT]


Nysret Musliu Vienna University of Technology, Austria
[EvoINDUSTRY]
Enrico Natalizio UTC - Sorbonne University, France [EvoCOMNET]
Boris Naujoks TH Cologne University of Applied Sciences, Germany
[EvoNUM]
Antonio Nebro Universidad de Málaga, Spain [EvoCOMPLEX]
Ferrante Neri De Montfort University, UK [EvoNUM, EvoIASP,
EvoKNOW, EvoSTOC]
Geoff Nitschke University of Cape Town, South Africa [EvoROBOT]
Rafael Nogueras Universidad de Málaga, Spain [EvoCOMPLEX]
Stefano Nolfi Institute of Cognitive Sciences and Technologies, Italy
[EvoROBOT]
Gustavo Olague CICESE, México [EvoPAR]
Carlotta Orsenigo Politecnico di Milano, Italy [EvoBIO]
Ender Ozcan University of Nottingham, UK [EvoINDUSTRY]
Ben Paechter Edinburgh Napier University, UK [EvoINDUSTRY]
Peter Palensky Technical University of Delft, The Netherlands
[EvoENERGY]
Antonio González Pardo Universidad Autónoma de Madrid, Spain
[EvoGAMES]
Anna Paszyńska Jagiellonian University, Poland [EvoCOMPLEX]
Riccardo Pecori eCampus University, Italy [EvoCOMNET]
David Pelta University of Granada, Spain [EvoSTOC]
Raffaele Perego ISTI-CNR, Italy [EvoPAR]
Sanja Petrovic University of Nottingham, UK [EvoINDUSTRY]
Nelishia Pillay University of KwaZulu-Natal, South Africa
[EvoINDUSTRY]
Clara Pizzuti ICAR-CNR, Italy [EvoBIO]
Mihai Polceanu ENIB, France [EvoGAMES]
Riccardo Poli University of Essex, UK [EvoIASP]
Arkadiusz Poteralski Silesian University of Technology, Poland
[EvoCOMPLEX]
Simon Powers Edinburgh Napier University, UK [EvoINDUSTRY]
Petr Pošík Czech Technical University in Prague, Czech Republic
[EvoNUM]
Mike Preuss University of Münster, Germany [EvoGAMES,
EvoNUM]
Mauricio Resende Amazon, USA [EvoCOMNET]
Jose Carlos Ribeiro Politechnique Institute of Leiria, Portugal [EvoPAR]
Hendrik Richter Leipzig University of Applied Sciences, Germany
[EvoSTOC]
Florentino Fernández Universidad de Vigo, Spain [EvoBIO]
Riverola
Simona Rombo Università degli Studi di Palermo, Italy [EvoBIO]
Organization XV

Claudio Rossi Universidad Politecnica de Madrid, Spain


[EvoROBOT]
Guenter Rudolph University of Dortmund, Germany [EvoNUM]
Mohammed Salem University of Mascara, Algeria [EvoGAMES]
Pablo Mesejo Santiago University of Granada, Spain [EvoIASP]
Pablo García Sánchez Universidad de Cádiz, Spain [EvoGAMES]
Sanem Sariel Istanbul Technical University, Turkey
[EvoINDUSTRY]
Thomas Schmickl University of Graz, Austria [EvoROBOT]
Sevil Sen Hacettepe University, Turkey [EvoCOMNET]
Chien-Chung Shen University of Delaware, USA [EvoCOMNET]
Anabela Simões Institute Polytechnic of Coimbra, Portugal [EvoSTOC]
Moshe Sipper Ben-Gurion University, Israel [EvoGAMES]
Stephen Smith University of York, UK [EvoIASP]
Maciej Smołka AGH University of Science and Technology, Poland
[EvoCOMPLEX]
Andy Song RMIT, Australia [EvoIASP]
Andreas Steyven Edinburgh Napier University, UK [General]
Giovanni Squillero Politecnico di Torino, Italy [EvoGAMES, EvoIASP]
Jose Santos Reyes Universidad de A Coruña, Spain [EvoBIO]
Ke Tang University of Science and Technology of China, China
[EvoNUM]
Andrea Tettamanzi University of Nice Sophia Antipolis/I3S, France
[EvoBAFIN]
Renato Tinós Universidade de São Paulo, Brazil [EvoSTOC]
Julian Togelius New York University, USA [EvoGAMES]
Krzysztof Trojanowski Cardinal Stefan Wyszyński University in Warsaw,
Poland [EvoSTOC]
Wojciech Turek AGH University of Science and Technology, Poland
[EvoCOMPLEX]
Ryan Urbanowicz University of Pennsylvania, USA [EvoBIO]
Andrea Valsecchi University of Granada, Spain [EvoIASP]
Leonardo Vanneschi Universidade Nova de Lisboa, Portugal [EvoIASP]
Sebastien Varrete Université du Luxemburg, Luxemburg [EvoPAR]
Nadarajen Veerapen University of Stirling, UK [EvoINDUSTRY]
Francisco Goméz Vela Pablo de Olavide University, Spain [EvoBIO]
José Manuel Velasco Universidad Complutense de Madrid, Spain [EvoPAR]
Marco Villani University of Modena and Reggio Emilia, Italy
[EvoCOMNET]
Rafael Villanueva Universitat Politecnica de Valencia, Spain [EvoPAR]
Tanja Vos Open University, The Netherlands [EvoSET]
Jaroslaw Was AGH University of Science and Technology, Poland
[EvoCOMNET]
Simon Wells Edinburgh Napier University, UK [EvoINDUSTRY]
David White University College London, UK [EvoSET]
XVI Organization

Bing Xue University of Wellington, New Zeland [EvoBIO,


EvoIASP]
Anil Yaman Technical University of Eindhoven, The Netherlands
[EvoKNOW]
Shengxiang Yang De Monfort University, UK [EvoINDUSTRY,
EvoSTOC]
Georgios Yannakakis University of Malta, Malta [EvoGAMES]
Danial Yazdani Liverpool John Moores University, UK [EvoSTOC]
Aleš Zamuda University of Maribor, Slovenia [EvoKNOW]
Nur Zincir-Heywood Dalhousie University, Canada [EvoCOMNET]
Contents

EvoBAFIN

Multi-objective Cooperative Coevolutionary Algorithm with Dynamic


Species-Size Strategy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Karoon Suksonghong and Kittipong Boonlong

EvoBIO

Task Classification Using Topological Graph Features for Functional


M/EEG Brain Connectomics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Javier Del Ser, Eneko Osaba, and Miren Nekane Bilbao

Feature Selection for Detecting Gene-Gene Interactions in Genome-Wide


Association Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Faramarz Dorani and Ting Hu

Fitness Functions Evaluation for Segmentation of Lymphoma Histological


Images Using Genetic Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Thaína A. A. Tosta, Paulo Rogério de Faria, Leandro Alves Neves,
and Marcelo Zanchetta do Nascimento

Mutual Information Iterated Local Search: A Wrapper-Filter Hybrid


for Feature Selection in Brain Computer Interfaces . . . . . . . . . . . . . . . . . . . 63
Jason Adair, Alexander E. I. Brownlee, and Gabriela Ochoa

Automatic Segmentation of Neurons in 3D Samples of Human


Brain Cortex. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
G. Mazzamuto, I. Costantini, M. Neri, M. Roffilli, L. Silvestri,
and F. S. Pavone

Analysis of Relevance and Redundance on Topoisomerase 2b (TOP2B)


Binding Sites: A Feature Selection Approach . . . . . . . . . . . . . . . . . . . . . . . 86
Pedro Manuel Martínez García, Miguel García Torres,
Federico Divina, Francisco Antonio Gómez Vela,
and Felipe Cortés-Ledesma

EvoCOMNET

Multimodal Transportation Network Design Using Physarum


Polycephalum-Inspired Multi-agent Computation Methods . . . . . . . . . . . . . . 105
Rishi Vanukuru and Nagendra R. Velaga
XVIII Contents

Improving Multi-objective Evolutionary Influence Maximization


in Social Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Doina Bucur, Giovanni Iacca, Andrea Marcelli,
Giovanni Squillero, and Alberto Tonda

Social Relevance Index for Studying Communities in a Facebook


Group of Patients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Laura Sani, Gianfranco Lombardo, Riccardo Pecori,
Paolo Fornacciari, Monica Mordonini, and Stefano Cagnoni

A Fast Metaheuristic for the Design of DVB-T2 Networks. . . . . . . . . . . . . . 141


Fabio D’Andreagiovanni and Antonella Nardin

EvoCOMPLEX

A Genetic Algorithm for Community Detection in Attributed Graphs . . . . . . 159


Clara Pizzuti and Annalisa Socievole

Maximizing the Effect of Local Disturbance in the Dynamics


of Opinion Formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
Long Him Cheung, Ka Wai Cheung, and Kwok Yip Szeto

Accelerating the Computation of Solutions in Resource Allocation


Problems Using an Evolutionary Approach and Multiagent
Reinforcement Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Ana L. C. Bazzan

EvoENERGY

Achieving Optimized Decisions on Battery Operating Strategies


in Smart Buildings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
Jan Müller, Mischa Ahrens, Ingo Mauser, and Hartmut Schmeck

Phase-Space Sampling of Energy Ensembles with CMA-ES . . . . . . . . . . . . . 222


Jörg Bremer and Sebastian Lehnhoff

Many-Objective Optimization of Mission and Hybrid Electric Power


System of an Unmanned Aircraft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
Teresa Donateo, Claudia Lucia De Pascalis, and Antonio Ficarella

Evolving Controllers for Electric Vehicle Charging . . . . . . . . . . . . . . . . . . . 247


Martin Pilát

Network Coordinated Evolution: Modeling and Control of Distributed


Systems Through On-line Genetic PID-Control Optimization Search . . . . . . . 256
Holm Smidt, Matsu Thornton, and Reza Ghorbani
Contents XIX

EvoGAMES

Piecemeal Evolution of a First Person Shooter Level . . . . . . . . . . . . . . . . . . 275


Antonios Liapis

Online-Trained Fitness Approximators for Real-World Game Balancing. . . . . 292


Mihail Morosan and Riccardo Poli

Recomposing the Pokémon Color Palette . . . . . . . . . . . . . . . . . . . . . . . . . . 308


Antonios Liapis

Mapping Chess Aesthetics onto Procedurally Generated


Chess-Like Games . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
Jakub Kowalski, Antonios Liapis, and Łukasz Żarczyński

Evolving a TORCS Modular Fuzzy Driver Using Genetic Algorithms . . . . . . 342


Mohammed Salem, Antonio Miguel Mora, Juan Julian Merelo,
and Pablo García-Sánchez

Self-adaptive MCTS for General Video Game Playing. . . . . . . . . . . . . . . . . 358


Chiara F. Sironi, Jialin Liu, Diego Perez-Liebana, Raluca D. Gaina,
Ivan Bravi, Simon M. Lucas, and Mark H. M. Winands

Deceptive Games . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 376


Damien Anderson, Matthew Stephenson, Julian Togelius,
Christoph Salge, John Levine, and Jochen Renz

EvoIASP

Evolution of Convolutional Highway Networks . . . . . . . . . . . . . . . . . . . . . 395


Oliver Kramer

Adapting Bagging and Boosting to Learning Classifier Systems . . . . . . . . . . 405


Yi Liu, Will N. Browne, and Bing Xue

An Automatic Feature Extraction Approach to Image Classification


Using Genetic Programming. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
Ying Bi, Bing Xue, and Mengjie Zhang

Improving Evolutionary Algorithm Performance for Feature Selection


in High-Dimensional Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 439
N. Cilia, C. De Stefano, F. Fontanella, and A. Scotto di Freca

CGP4Matlab - A Cartesian Genetic Programming MATLAB Toolbox


for Audio and Image Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455
Rolando Miragaia, Gustavo Reis, Francisco Fernandéz,
Tiago Inácio, and Carlos Grilo
XX Contents

Can the Relevance Index be Used to Evolve Relevant Feature Sets? . . . . . . . 472
Laura Sani, Riccardo Pecori, Emilio Vicari, Michele Amoretti,
Monica Mordonini, and Stefano Cagnoni

Towards Evolutionary Super-Resolution. . . . . . . . . . . . . . . . . . . . . . . . . . . 480


Michal Kawulok, Pawel Benecki, Daniel Kostrzewa,
and Lukasz Skonieczny

Evolvable Deep Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497


Jakub Nalepa, Grzegorz Mrukwa, and Michal Kawulok

Estimation of the 3D Pose of an Object Using Correlation Filters


and CMA-ES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 506
Juan Carlos Dibene, Kenia Picos, Victor H. Díaz-Ramírez,
and Leonardo Trujillo

EvoINDUSTRY

Evaluating the Performance of an Evolutionary


Tool for Exploring Solution Fronts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523
Neil B. Urquhart

A Classifier to Identify Soft Skills in a Researcher Textual Description . . . . . 538


Antonia Azzini, Andrea Galimberti, Stefania Marrara, and Eva Ratti

Toward the Online Visualisation of Algorithm Performance


for Parameter Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 547
David J. Walker and Matthew J. Craven

Integrating Evolution Strategies into Genetic Algorithms with Fuzzy


Inference Evaluation to Solve a Steelmaking and Continuous Casting
Scheduling Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 561
Eduardo Salazar

Automatic Generation of Constructive Heuristics for Multiple Types


of Combinatorial Optimisation Problems with Grammatical Evolution
and Geometric Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 578
Christopher Stone, Emma Hart, and Ben Paechter

EvoKNOW

Rotation Invariance and Rotated Problems: An Experimental Study


on Differential Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597
Fabio Caraffini and Ferrante Neri
Contents XXI

EvoNUM

Multi-strategy Differential Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 617


Anil Yaman, Giovanni Iacca, Matt Coler, George Fletcher,
and Mykola Pechenizkiy

A Generic Framework for Incorporating Constraint Handling Techniques


into Multi-Objective Evolutionary Algorithms. . . . . . . . . . . . . . . . . . . . . . . 634
Hiroaki Fukumoto and Akira Oyama

EvoPAR

A CPU-GPU Parallel Ant Colony Optimization Solver for the Vehicle


Routing Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 653
Antón Rey, Manuel Prieto, J. I. Gómez, Christian Tenllado,
and J. Ignacio Hidalgo

EvoROBOT

Evolving Artificial Neural Networks for Multi-objective Tasks . . . . . . . . . . . 671


Steven Künzel and Silja Meyer-Nieberg

Revolve: A Versatile Simulator for Online Robot Evolution . . . . . . . . . . . . . 687


Elte Hupkes, Milan Jelisavcic, and A. E. Eiben

Search Space Analysis of Evolvable Robot Morphologies . . . . . . . . . . . . . . 703


Karine Miras, Evert Haasdijk, Kyrre Glette, and A. E. Eiben

Combining MAP-Elites and Incremental Evolution to Generate Gaits


for a Mammalian Quadruped Robot. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 719
Jørgen Nordmoen, Kai Olav Ellefsen, and Kyrre Glette

Evolving a Repertoire of Controllers for a Multi-function Swarm . . . . . . . . . 734


Sondre A. Engebråten, Jonas Moen, Oleg Yakimenko,
and Kyrre Glette

HyperNTM: Evolving Scalable Neural Turing Machines


Through HyperNEAT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 750
Jakob Merrild, Mikkel Angaju Rasmussen, and Sebastian Risi

EvoSET

Investigating the Evolvability of Web Page Load Time . . . . . . . . . . . . . . . . 769


Brendan Cody-Kenny, Umberto Manganiello, John Farrelly,
Adrian Ronayne, Eoghan Considine, Thomas McGuire,
and Michael O’Neill
XXII Contents

Late Acceptance Hill Climbing for Constrained Covering Arrays . . . . . . . . . 778


Mosab Bazargani, John H. Drake, and Edmund K. Burke

Search-Based Temporal Testing in an Embedded Multicore Platform. . . . . . . 794


Komsan Srivisut, John A. Clark, and Richard F. Paige

EvoSTOC

Robust Evolutionary Optimization Based on Coevolution. . . . . . . . . . . . . . . 813


Steffen Limmer and Tobias Rodemann

On the Use of Repair Methods in Differential Evolution for Dynamic


Constrained Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 832
Maria-Yaneli Ameca-Alducin, Maryam Hasani-Shoreh,
and Frank Neumann

Prediction with Recurrent Neural Networks in Evolutionary


Dynamic Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 848
Almuth Meier and Oliver Kramer

A Multi-objective Time-Linkage Approach for Dynamic Optimization


Problems with Previous-Solution Displacement Restriction . . . . . . . . . . . . . . 864
Danial Yazdani, Trung Thanh Nguyen, Juergen Branke,
and Jin Wang

A Type Detection Based Dynamic Multi-objective


Evolutionary Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 879
Shaaban Sahmoud and Haluk Rahmi Topcuoglu

General

CardNutri: A Software of Weekly Menus Nutritional Elaboration


for Scholar Feeding Applying Evolutionary Computation . . . . . . . . . . . . . . . 897
Rafaela P. C. Moreira, Elizabeth F. Wanner, Flávio V. C. Martins,
and João F. M. Sarubbi

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 915


EvoBAFIN
Multi-objective Cooperative Coevolutionary
Algorithm with Dynamic Species-Size Strategy

Karoon Suksonghong1(&) and Kittipong Boonlong2


1
Faculty of Management and Tourism, Burapha University, Saen Suk, Thailand
karoon@buu.ac.th
2
Faculty of Engineering, Burapha University, Saen Suk, Thailand
kittipong@buu.ac.th

Abstract. Although numbers of heuristic algorithms are successfully devel-


oped for solving portfolio optimization problems, this is not for all cases of the
large-scale ones. A large-scale portfolio optimization involves dealing with the
large search space and dense variance-covariance matrix associated with the
problem. This paper proposed a new multi-objective algorithm for solving a
large-scale optimization problem based upon the notion of cooperative coevo-
lutionary algorithms (CCA). The new problem decomposition scheme was
designed by allowing the species-size to be dynamically adjusted as the runs
progress. This scheme enhances capability of traditional CCA in dealing with
non-separable optimization problem. The collaborator selection method was
modified to allow the proposed CCA to perform in a multi-objective (MO) op-
timization framework. Additionally, the proposed algorithm, named as
“DMOCCA”, was implemented for solving large-scale portfolio optimization
problem with cardinality constraint using the real-world data set having scale up
to 2196 dimensions. Moreover, its performances were benchmarked with those
of the SPEA-II and MOPSO.

Keywords: Cardinality  Cooperative coevolutionary


Dynamic problem decomposition  Large-scale optimization
Portfolio selection

1 Introduction

Intuitively, one plausible way to handle a large domain problem is to employ a


divide-and-conquer strategy. The cooperative coevolutionary algorithm (CCA) pro-
posed by Potter and De Jong [1] offers a convincing framework for decomposing and
solving a large-scale problem at the same time. CCA is able to focus the search on
relevant areas while simultaneously making adaptive changes between interacting and
evolving activities. Basically, CCA involves the use of multiple species to represent a
complete solution. By decomposing a full solution vector into multiple species, the
search space in which algorithm has to explore is significantly reduced. For instance, in
the case that a solution is encoded into a binary chromosome of length 100, the number
of possible solutions is 1.27  1030 = 2100. On the other hand, by dividing a full
solution into 20 species and each species is represented by a 5-bit binary string, the

© Springer International Publishing AG, part of Springer Nature 2018


K. Sim and P. Kaufmann (Eds.): EvoApplications 2018, LNCS 10784, pp. 3–17, 2018.
https://doi.org/10.1007/978-3-319-77538-8_1
4 K. Suksonghong and K. Boonlong

number of possible solutions is reduced to only 20  25 = 640 which allows an


algorithm to discover the optimal solutions faster and more efficient.
Potter and De Jong [1] applied this framework for solving the optimization prob-
lems having decision variables up to 30. The experimented results revealed that CCA
outperformed the conventional GA on the separable problems; however, it performed
worse on the non-separable problems. Subsequently, Liu et al. [2] developed the fast
evolutionary programming (FEP) and integrated it with CCA. The resulting algorithm
was successfully implemented for solving the optimization functions having decision
variables up to 1,000. They reported that, however, the proposed algorithm was trapped
in the local optima when dealing with non-separable functions. It is clear that the
interdependency among variables raises difficulties in CCA execution.
It is argued that effectiveness of CCA depends heavily on problem decomposition
method [3, 4]. In the past decade, several works proposed new problem decomposition
strategies to address interdependency issues. For example, Van den Bergh and
Engelbrecht [3], who integrated CCA with particle swarm optimization (PSO) algo-
rithms, decomposed n decision variables into K species. Each species contains s indi-
viduals, i.e. n = K  s. Therefore, each species represents s variables rather than one
variable as the case of the original CCA. However, the members in each species were
not altered as iterations progress. Shi et al. [4] proposed the cooperative coevolutionary
differential evolution (CCDE) with splitting-in-half decomposition strategy. In their
work, the problem was divided into two subcomponents and evolution of these sub-
components was performed separately. Similarly, the members of these two subcom-
ponents were not changed throughout the algorithm runs. The common underlying idea
of these two strategies is to increase the number of variables in each species, i.e.
species-size, by expecting that the interacting variables will be placed within the same
species. However, these two static decomposition strategies require a priori knowledge
about the structure of the problem at hand and perform poorly when the problem size is
very large [5, 6].
Instead of using static decomposition strategy, the species-size and the species
member that can be dynamically adjusted during run time using rules pre-specified by
algorithm settings could be useful for alleviating problems mentioned earlier. Yang et al.
[5] proposed a new dynamic decomposition strategy, the so-called “random grouping”,
based on probability assignment and integrated this idea with CCDE for solving
non-separable problem with 1,000 variables. This decomposition method allows group
size, i.e. species-size, to be varied throughout the course of optimization. They
demonstrated that the more often the grouping size is adjusted, the higher the probability
the interacting variables will be placed within the same group In the time since this
remarkable decomposition strategy was published, the superiorities of dynamic
decomposition CCA for solving non-separable functions were widely reported [7–9].
In this paper, we developed multi-objective CCA (MOCCA) for solving a
large-scale optimization problem. To enhance the capability of the proposed MOCCA,
a new problem decomposition scheme based upon dynamic species-size strategy and
the dynamic process were developed. Concurrently, collaborator selection scheme was
designed in order to allow CCA to perform in multi-objective framework. Then, the
proposed MOCCA, denoted by “DMOCCA”, was tested for solving the large-scale
cardinality constraint mean-variance portfolio optimization problem (CCMVPOP) with
Multi-objective Cooperative Coevolutionary Algorithm 5

problem size up to 2196 variables. Lastly, its performance was benchmarked with that
of the SPEA-II [10] and MOPSO [11].
This paper is organized as follows. The detailed developments of the MOCCA with
dynamic species-size strategy are described in Sect. 2. Section 3 discusses and for-
mulates the portfolio optimization problem. The experimented results are reported in
Sect. 4 and conclusion is given in Sect. 5.

2 Multi-objective CCA with Dynamic Problem


Decomposition

Although the prominent ability of CCA to divide the problem into smaller search space
and to increase the influences of the evolutionary operators on these smaller subcom-
ponents is widely recognized, CCA with static decomposition strategy is inapplicable
for solving a non-separable problem. With the presence of interrelation between vari-
ables, the use of dynamic decomposition strategy that allows CCA to adjust species-size
and species member during the algorithm runs could be the efficient way to handle this
type of problem [8]. Li and Yao [9] demonstrated that by changing species-size fre-
quently (the so-called “grouping size” in their work), the probability that two interacting
variables will be placed into the same subcomponent significantly increases.
In the literature, two dynamic problem decomposition schemes based upon
dynamic species-size strategy were proposed. Firstly, in Yang et al. [7], a set of
potential species-size is subjectively identified. Then, at the beginning of generation,
species-size is randomly selected from this potential set based upon probabilities. The
higher probability will be rewarded to a species-size having better past performance.
Thus, species-size that works well is more likely to be used in the next generation.
Secondly, Li and Yao [9] adjusted species-size by comparing the fitness values before
and after each generation run. If the fitness value is greater than the previous run, the
next generation run will be executed using the same species-size. In contrast, if fitness
value turns worse, species-size will be changed by randomly choosing a new
specie-size from pre-determined set of species-size. The common idea of these two
schemes is to allow species-size that performs well to be used for the next generation
run. The criterion for changing species-size is straightforward since the term “better” or
“worse” can be objectively compared by scalar value. In the other words, the existing
dynamic problem decomposition strategies were designed for solving SOOP.

2.1 Dynamic Species-Size


Our proposed dynamic species-size method allows CCA to handle both SOOP and
MOOP. According to our method, a set of dynamic species-size (SS) is firstly identi-
fied. This set is generated by a geometric series which the first term is 1 and common
ratio is 2, i.e. {1, 2, 4, 8, 16,…}. Thus, a set of number of species (NS) is the result of
division where numerator is number of variable (N) and denominator is species-size,
i.e. NS ¼ N=SS. The algorithm starts by randomly selecting a species-size from SS.
Then variables assignment is conducted in chronological sequence from the first spe-
cies to the last species. This implies that variables will be selected and allocated to the
6 K. Suksonghong and K. Boonlong

first species until it is fully filled. Then, allocation process moves on to the second
species and so forth. It should be noted that number of variables assigned to the last
species may not equal to that of other previous species if fraction from the division of
N=SS exists. The managing rule for this case is that if the remainder is equal or greater
than half of species-size; all residual variables will be allocated to the next species,
otherwise they will be combined with the previous species.
For example, a set of number of species (NS) for a problem having 100 variables is
defined as NS = 100  {1, 2, 4, 8, 16, 32, 64} = {100, 50, 25, 13, 6, 3, 1}. Considering
the case that the chosen SS is 8, thus NS is 12.5. By allocating 8 variables to each of the
first 12 species, the remaining variables are 4 which equal to half of species-size.
Therefore, algorithm assigns them to the 13th species. In contrast, if the selected SS is 16,
the computed NS is 6.25. By appointing 16 variables to each of the first 6 species, the
remainder is 4 which is smaller than half of species-size. Therefore, algorithm combines
them with the 6th species. The aim of the proposed procedure described above is to
provide the diversity of species-size ranging from largest to smallest size. Although a set
of SS is still needed to be supplied, it is identified based on simple rule and does not
require sophisticated setting from user. It can be thought of that CCA that adopts the
largest species-size from SS is analogous to the traditional multi-objective optimization
algorithms whose solutions are represented by only one species containing all decision
variables. Whereas, the classical CCA adopts the smallest species-size, i.e. SS = 1.

2.2 Dynamic Process


The proposed DMOCCA begins by randomly selecting a species-size from SS. The
selection without replacement is employed in this scheme to ensure that all
species-sizes are chosen for execution. For dynamic mechanism, the selected
species-size will be used in the next iterations if solutions obtained from the current
iteration are better than those of the previous iteration. Otherwise, species-size will be
adjusted. The comparison of solutions between iterations for MOOP is different from
that for SOOP since there is no the best solution to a MOOP. Instead, a set of
non-dominated solutions is obtained in each algorithm run. In this paper, we employed
“convergence detection” as a method for comparing the solution sets of current and
previous iteration and as a trigger criterion for adjusting of species-size. The conver-
gence detection process can be described as follows.
Given Ai and Ai–1 are non-dominated solution sets of the current iteration and
previous iteration, respectively. The condition for solutions convergence is proposed as:

CðAi ; Ai1 Þ  C ðAi1 ; Ai Þ ð1Þ

where CðAi ; Ai1 Þ is coverage ratio of solution set Ai over solution set Ai1 while
C ðAi1 ; Ai Þ is the reverse value of C ðAi ; Ai1 Þ. The solution coverage, C, which is
used for comparing any two sets of solutions, is mathematically expressed as:

jfai1 2 Ai1 ; 9ai 2 Ai : ai  ai1 gj


C ðAi ; Ai1 Þ ¼ ð2Þ
jAi1 j
Multi-objective Cooperative Coevolutionary Algorithm 7

where ai  ai1 means that solution ai covers, i.e. dominates solution ai1 and
C ðAi ; Ai1 Þ 2 ½0; 1. In the case that C ðAi ; Ai1 Þ ¼ 1, all solutions in set Ai1 are
dominated by solutions in set Ai . In contrast, C ðAi ; Ai1 Þ ¼ 0 represents the situation
that none of solutions in set Ai1 are covered by those of set Ai . From this sense,
C ðAi ; Ai1 Þ [ C ðAi1 ; Ai Þ implies that solutions set of the current iteration is better
than that of the previous iteration. Whereas, CðAi ; Ai1 Þ  C ðAi1 ; Ai Þ indicates
solutions convergence condition that activates algorithm to change species-size.

2.3 Collaborator Selection Method


According to CCA framework, the fitness of individuals of the considered species can
be evaluated after combining them with individuals of the other species, the so-called
“collaborator”. In practice, it is widely recognized that collaborator selection method is
crucial for CCA performance. As illustrated in Fig. 1(a), the classical CCA which is
usually applied for SOOP selects collaborators from the current best individual.
Therefore, all individuals of the evaluated species are cooperated with the similar set of
collaborators. In this paper, the new collaborator selection scheme based upon elitism
strategy is presented and its procedure is demonstrated in Fig. 1(b). For example,
considering a problem having four variables and species-size equals one. At the end of
iteration, algorithm obtains a set of non-dominated solutions and keeps them in archive.
In the case that species 1 containing variable x1 is evaluated, algorithm chooses col-
laborators from the current archive and combines them with the evaluated species with
random ordering. We found that this proposed method enhances search ability since it
increases probability of exploring to new searching area that eventually helps avoid
trapping within the local optima.

x1 x2 x3 x4 Current best solution x1 x2 x3 x4 Current archive


x1,1 x2,1 x3,1 x4,1 x1,b x2,b x3,b x4,b x1,1 x2,1 x3,1 x4,1 x1,A1 x2,A1 x3,A1 x4,A1
x1,2 x2,2 x3,2 x4,2 x1,2 x2,2 x3,2 x4,2 x1,A2 x2,A2 x3,A2 x4,A2



x1,P x2,P x3,P x4,P Collaborators x1,P x2,P x3,P x4,P x1,AP x2,AP x3,AP x4,AP
Evaluated Evaluated

f (x1, x2, x3, x4) f (x1, x2, x3, x4)


Random
f (x1,1, x2,A2 order
f (x1,1, x2,b, x3,b, x4,b) 2,A2 , x3,A2
3,A2 , x4,A2
4,A2 )

f (x1,2, x2,b, x3,b, x4,b) f (x1,2, x2,A5


2,A5 , x3,A5
3,A5 , x4,A5
4,A5 )


f (x1,P, x2,b, x3,b, x4,b) f (x1,P, x2,A


2,A3
3
, x3,A
3,A3
3
, x4,A
4,A3
3
)

(a) Collaborator selection method for (b) New collaborator selection scheme for
single-objective optimization problem multi-objective optimization problem

Fig. 1. Collaborator selection schemes


Another random document with
no related content on Scribd:
sinua silmiin, tietämättä etten tapaisi sinua seuraavana päivänä —
en enää koskaan!»

— »Niin, mutta lapseni», sanoi hän, »et saa alati miettiä tuota yhtä
ja samaa.»

— Sitten olin tekevinäni hänelle jonkun pienen palveluksen ja hän


kiitti minua siitä. Hän oli aina niin kiitollinen jokaisesta
vähäpätöisimmästäkin asiasta. Se vaivasi minua, sillä minä tein aina
liian vähän hänen hyväksensä. Enkä ollut koskaan saanut oikein
sanotuksi mitä tunsin häntä kohtaan. Mutta viime yönä voin
vihdoinkin tehdä sen: »Äiti, et saa koskaan kiittää minua, minulle
tuottaa tuskaa, kun teet sen. Se on kuin taakka povellani. Sillä
minähän olen sinulle ikuisessa kiitollisuudenvelassa — eikä
päinvastoin! Sinä tiedät etten voinut koskaan tehdä mitään
hyväksesi, enkä edes kiittänyt siitä mitä sinä teit minulle!»… Olin niin
iloinen kun vihdoinkin sain sanoa sen! — — —

*****

— Eikö teistä käy ikäväksi istua tätä kaikkea kuuntelemassa?

Thora ei vastannut, katsoi vain häneen.

Kun Samuel Stern ei enää jatkanut, kysyi Thora: — Eikö ollut vielä
muutakin?

— Sitten oli kuin kävisi kaikki jälleen pimeäksi ja raskaaksi. Minä


kyselin äitiäni kaikkialla. Joku sanoi minulle että hän oli poissa.
Minusta se kuului naurettavalta. Vihdoin hän tuli. »Rakas äiti, kuinka
en ole nähnyt sinua niin pitkään aikaan? Oli hyvä että tulit, sillä
minun täytyy kertoa sinulle tuhansista seikoista! Mutta miksi tulet niin
hiljaa? Kuinka on laitasi? Nukutko nyt paremmin? Muistatko että
sanoit viimeisenä yönä: 'On niin vaikeata, — en saa hetkeksikään
unta, en päivällä enkä yöllä!' Onko sinun nyt parempi olla? Onko
parempi, kun pitkä, hiljainen yö on tullut? Onko totta tuo, että
kuolleitten uni on sikeä, matala heidän multapieluksensa!?» — — —

*****

— »Äiti, olen niin iloinen, kun saan kuulla äänesi on niin tyhjää,
kun et enää ole täällä — on niin tyhjää, kun en tiedä palaatko sinä
pian jälleen!» Sitten oli kuin olisin käynyt hänen haudallansa, ja hän
saattoi minua kotiin. »Älä sure», sanoi hän. »Etkö tunne, että minä
olen luonasi?» — — —

— Silloin vavahteli sydämeni. »Et saa enää koskaan mennä


luotani!»

— Kun tulimme vanhan kotimme portille, pelästyin minä, sillä näin


että hänellä oli siivet. »Tulethan sinäkin sisään!» sanoin hänelle.
Silloin hän hymyili hiljaa. »Rakas lapseni, etkö tiedä että minä olen
taistellut taisteluni loppuun?» Ja hän kääntyi minun puoleeni ja
toivotti minulle siunaten rauhaa — — —

— Kurotin käteni hänen jälkeensä — mutta silloin heräsin. — — —

*****

— Nyt kun hän on poissa, olen ruvennut uskomaan että täytyy


vielä olla jotakin haudan tuolla puolen — — —

— Olisin niin kernaasti tahtonut pitää hänet luonani, että olisin


voinut pitää hänestä parempaa huolta hänen vanhoilla päivillänsä
kuin ajattelemattomuudessani olin tähän saakka pitänyt. Hän oli
luvannut tulla luokseni tänä kesänä — — —

— Kuinka hän ajatteli minua kuolemaansa saakka! Hän tahtoi


säästää minua. Ei kukaan saanut sanoa, kuinka hänen laitansa oli.
Äidin rakkaus on ainoa, jota ei mikään — ei edes oma huonoutemme
— voi meiltä riistää! — —

— No niin, — ja nytkö sitten on lähdettävä kotiin pelaamaan


shakkia ladyn kanssa?

— Eipä suinkaan! Olisi parempi että menisitte kotiin lepäämään.


Olette väsyksissänne!

— Äiti sanoi aina: »Ei saa väsyä!» Hän ei väsynyt koskaan. Sillä
hän vei meidät harhaan, meidät ajattelemattomat ja heikot olennot —
siinä määrin, että me luulimme hänen olevan jotain hyvää, joka ei
koskaan loppuisi. Niin, hän petti minut siinä määrin, että minä
sydämeni sanomattomassa yksinkertaisuudessa luulin, että hän oli
kuolematon… Kuuluu uskomattomalta, mutta kun muut vanhat
ihmiset kuolivat, niin minusta se oli paikallaan — mutta äitiin ei
kuoleman ollut kajoominen, se ei saanut lähestyä häntä.— — —

*****

— Muistatteko vielä — ei, ettehän te nähnyt häntä koskaan!… Hän


kuunteli mielellään sotilassoittoa ulkona vapaan, sinisen taivaan alla.
Se näytti päästävän sidottuja voimia hänessä valloilleen, hänen
silmiinsä tuli loistetta. Hänen mielessänsä asui kaiho ylös
avaruuteen, mutta hän oli kahlehdittu. Niin laski hän siipensä
kokoon. Hänellä oli paljon voimaa ja tuollaista hiljaista,
järkkymätöntä rohkeutta, jota ei mikään murra. Hän taipui
kuormiensa alle ja kantoi ne. Hän oli voittamaton — kaikki hän kantoi
ja vei voittoon. Siksi olikin Napoleon hänen sankarinsa. — — —

*****

— Kun tulin suuremmaksi ja sain hieman ymmärrystä, aloin tuntea


nöyrää kunnioitusta katsoessani hänen kasvoihinsa. Kaikki mitä
tiesin kärsimyksistä, jokapäiväisestä surusta ja huolesta, kävi kuin
mitättömäksi noiden rohkeiden, tyynten silmien katseesta, noiden
siunattujen sinisilmien katseesta, jotka tiesivät niin paljon. —

— Hänen kaunis, soinnukas äänensä merkitsi niin paljon meille


lapsille. Se toi mieleemme sellaisen turvallisuuden tunteen. Ja kun
hän lauloi tai luki meille, oli se meille juhlahetki.

— Hänen äänensä ei käynyt koskaan vanhaksi kuten muiden


ihmisten. Se säilytti aina lempeän, kirkkaan sointinsa. — — —

— On väärin, etten voi suoda hänelle lepoa. Kuolemahan on


kuninkaallinen palkka. Se on elämän kruunu. Pitäisihän minun suoda
se hänelle.

Thora Thammers katsoi häntä liikutettuna rukoilevin katsein. —


Niin, niin kyllä! kuiskasi hän.

— Minun on niin vaikea olla! Eräs seikka, joka minua vaivaa


pahemmin päivällä kuin yöllä, on se, etten ollut hänen luonansa
hänen kuollessansa — etten ollut pyyhkimässä tuskanhikeä hänen
otsastansa ja nostanut häntä pystyyn vuoteessa ja hoidellut häntä.
Hän arveli itse että olisi parempi minulle etten näkisi hänen kärsivän.
Mutta nyt saan kokea seuraukset. Näen kaikki alati edessäni; kärsin
hänen tuskansa — — — Minä olen aina ollut noin huono ja heikko.
Minua on aina vaivannut jonkinmoinen liiallinen tunteellisuus…
Sentähden minä puren hampaani yhteen ja lyön kaikki leikiksi. Mutta
en nyt enää — nyt kärsin kärsittäväni!

Hänen kasvoillensa laskeutui tuskan ilme. Hän nousi ja telki


muutaman askeleen, mutta palasi jälleen ja kävi istumaan.

— On vielä toinenkin asia, joka on alkanut minua vaivata — — —

Thora Thammers nousi ja laski kätensä hänen käsivarrellensa.

— Nyt menemme kotiin! Onhan äitinne sanonut, että te ette saa


ajatella tuota kaikkea! Te tiedätte varsin hyvin, ettei hän ikinä olisi
tahtonut että te kiusaisitte itseänne noilla ajatuksilla… Mutta luulen
että on parempi, että olette niistä puhunut. Teidän on varmaankin nyt
helpompi vapautua niistä.

Samuel Stern katsahti ylös, hämmästyneenä hänen varmasta


äänensävystänsä.

Hän nousi, ja he alkoivat kulkea tietä alas.

— Minä olen raukka — enkös olekin? Mutta kyllä minun tulee


helpompi olla. Te tiedätte että tulee helpompi, kun ensi viikolla
ryhdyn kirjeitä vastaamaan ja kun sitten tulen työhön konttoriin — —

He näkivät mäen alla lady Hammondin, tulossa heitä vastaan.


Samuel
Sternin kasvot kävivät yrmeiksi.

— Näettekö, kuinka hän on iloinen ja tyytyväinen! Hän ihmettelee


kuollaksensa sitä etteivät kaikki ihmiset ole tyytyväisiä oloonsa. Yksi
paha on tosin olemassa, sen hän myöntää, — se että väliin voi
pitkästyä aikaansa.

— Entä lordi! Oletteko kuullut, että hän aikoo kirjoittaa kirjan


Norjasta?

— Aikooko hän kirjoittaa? No, kenties se käy paremmin; kaikki,


mitä hän puhuu, on kyllä takaperoista!

Nyt tuli lady Hammond heitä vastaan, viittoillen, loistaen innosta.


Hän oli odottanut Samuel Sterniä shakkilaudan ääressä… Nyt
tuumiskeli hän, kuinka kostaisi hänelle!

20.

Metsolaan oli tullut vielä muutamia vieraita. Nyt alkoi Marit


Hennerudin kolmas tämänkesäinen toimikausi.

Tulokkaat olivat metsästäjiä ja kalastajia. Ja alhaalta kylästä oli


tullut muutamia Maritin ystävättäriä, jotka nyt olivat saaneet hiukan
lomaa.

Niiden joukossa oli pari nuorta hauskaa talonpoikaisemäntää,


joiden kanssa Thora seurusteli mielellänsä. Heidän luonteva,
vaatimaton hienotunteisuutensa, henkevyytensä ja rehellisyytensä
vaikutti virkistävästi ja vahvistavasti kuin vuori-ilma.

Thora Thammers tunsi itsensä voimakkaammaksi kuin moneen


vuoteen. Oli kuin voisi hän vasta nyt levätä — ja tehdä työtä. Rouva
Harder oli kerran sanonut, että sen, joka tahtoo saada toimeen
jotakin, on alettava omasta itsestänsä. Siihen tapaan oli Thora nyt
alkanut melkein tietämättänsä. Hänestä tuntui ettei hän oikeastaan
itse tehnyt mitään — mutta hänen sisimmissään oli työ käynnissä…
hänen katkeruutensa suli pois ja katosi olemattomiin, hänen
näköpiirinsä laajeni.

Hänen täytyi melkein nauraa. Vasta nyt hän näet eräänä päivänä
huomasi, ettei Sjurilla ollut juuri mitään vaatteita.

Nyt ei hänellä enää ollut aikaa pitkiin kävelyretkiin. Hän alkoi kutoa
sukkia ja neuloa paitoja. Hän huomasi että sekin oli hyödyksi hänelle
itsellensä.

Päivät menivät menojansa. Samuel Sterniä ei hän nähnyt kuin


harvoin. Hänelle oli lähetetty paljon konttorityötä. Sen lisäksi teki hän
metsästysretkiä. Voi kulua päivämääriä, ilman että Thora näki häntä.

Thora ei enää hakenut hänen seuraansa muulloin kuin


huomatessaan ettei hänen pitäisi olla yksin. Sen havaitsi hän hänen
kasvoistansa.

Eräänä päivänä, kun Thora istui ulkona töineen, tuli Samuel Stern
hänen luoksensa.

— Missä te nykyisin oleksitte päivät päästänsä? Näen teitä vain


ani harvoin!… Te olette langennut kaikenmoisiin hyveisiin!

Thora katsahti häneen hymyillen.

— Ja te puolestanne laiskuuteen! Olette nyt istuskellut liian kauan


kuistillanne aivan joutilaana… Katsokaas, tässä on muuan kirja,
jonka olin aikonut tuoda teille. Se teidän pitää lukea. Alkakaa jo nyt
heti!
Samuel Stern katsoi häneen ihmeissään, mutta otti kirjan ja läksi
pois.

Pian ei hän enää hämmästynyt mistään. Asiat alkoivat olla sillä


kannalla, että hän aina teki kuten Thora käski. Vähitellen,
kummankaan huomaamatta, oli Thora saanut häneen nähden
vaikutusvallan, jommoista hänellä ei ollut koskaan ennen ollut.

Thorasta oli, kuin jäisi Omar Pashasta aina vähemmän jäljelle,


kuin tulisi hänen tilalleen se Samuel Stern, jonka hän kerran oli
tuntenut.

Heidän oli tullut tavaksi mennä joka päivä yhdessä kävelemään,


jos Samuel Stern oli kotosalla. Väliin oli myöskin lady Hammond tai
joku toinen mukana.

Ensi kerralla tuntui Thorasta tosin hiukan oudolta että joku muu oli
mukana. Mutta pian hän tottui siihen. Heistä molemmista alkoi se jo
tuntua aivan luonnolliselta.

Mutta puhuivathan he sentään tuttavallisemmin keskenään


ollessaan kahden.

Thora kertoeli lapsistansa ja Dyvekestä. Ja kun Dyvekeltä tuli


kirjeitä, luki hän ne aina Samuel Sternille. Se virkisti häntä. Tuo
raikas, pulppuava nuoruudenilo vaikutti heihin kuin leivon liverrys.

Ja kun he näkivät postin saapuvan mäen juurelle, oli Samuel Stern


yhtä innokas kuin Thora saamaan nuo kirjeet.

Hän oli myöskin alkanut puhua Thoran kanssa kauppa-asioistaan


heidän ollessaan kahden — ja sitten tuli aina lisäksi jotain hänen
äidistänsä, ikäänkuin hänen mainitsemisensa olisi tuottanut hänelle
tyydytystä. Ja joskaan hän ei häntä maininnut, huomasi Thora hänen
katseestansa ja äänensävystänsä, milloin äiti oli hänen
ajatuksissansa. Oli kuin hän aina olisi ollut heidän luonansa — kun
he olivat kahden.

Thora Thammers iloitsi siitä. Hän voi kumminkin tuottaa hänelle


lohdutusta. Hänestä tuntui kuin ansaitsemattomalta onnelta, että
Samuel Stern tahtoi puhua hänen kanssaan äidistänsä.

Mutta naisen tarkalla havaintokyvyllä hän huomasi että kaikki


heidän entisyyttään koskeva oli kuin kulunut pois hänen
sydämestänsä.

Mikä tavaton, armelias kyky unhottaa!

Hän iloitsi siitäkin, sillä se oli oikeudenmukaista. Hänestä oli, kuin


varttuisi hänkin tuon tietoisuuden nojassa.

Hän varoisi kyllä herättämästä enää mitään entistä henkiin.

Nyt oli kai Samuel Stern täydelleen antanut hänelle anteeksi. Hän
turvautui lohdullisena tuohon ajatukseen. Lady Hammondkin muiden
muassa oli huomannut että lordi Stern, kuten hän häntä nimitti, oli
käynyt suopeammaksi muille eikä ollut enää niin ärtyisä kuin ennen.

Eräänä päivänä saattoi hän lady Hammondin ihastuksiin,


lupaamalla lähteä mukaan kalastusretkelle. Lady Hammond arveli
että tuosta hauskasta muutoksesta sai kiittää shakkipeliä, sillä hän
oli jo aikoja huomannut sen vaikuttavan jalostavasti. Mutta omaa
lordiansa ei hänen ollut koskaan onnistunut saada sitä koettamaan.

Tuo kalastusretki kävi pitemmäksi kuin oli ajateltu — ja vielä


romantillisemmaksi kuin Ldy Hammond oli kuvitellut.
He olivat eksyneet metsässä ja heidän oli täytynyt viettää kaksi
yötä tunturipaimenten risumajasissa.

Lady Hammond oli haltioissaan. Lordi merkitsi muistiinpanoja


kirjaansa varten — Samuel Stern väsyi koko seikkailuun ja halusi
kotiin.

Sinä iltana, jolloin he vihdoinkin saapuivat Metsolaan, märkinä,


saaliilla kuormitettuina, olivat he olleet poissa melkein viikkokauden.

Lordi Stern oli huonolla tuulella, väitti lady Hammond.

Mutta sitä ei Marit Hennerud voinut huomata. Hän piti hänestä niin
hellää huolta, kuin olisi ollut hänen äitinsä, ja juhlallisemmin ei liesi
ollut loimunnut vielä koskaan.

Vihdoin tuli rouva Thammerskin seurusteluhuoneeseen ja lausui


heidät tervetulleiksi. Hän oli kuullut että he olivat palanneet, mutta
hänen oli ensin pitänyt lopettaa jokin työ.

Syötiin oivallista kalaa, lepäiltiin ja lämmiteltiin sekä pidettiin


hauskaa, sillä lady Hammond kertoeli ihmeellisellä norjalais-
englantilaisella sekamelskallaan kaikista retken seikkailuista.

Samuel Stern istui hiljaa tuijottaen valkeaan pää kätten varassa.

Niin kiintynyt ei lady Hammond ollut aineeseensa, ettei olisi


samalla tarkannut kuinka miellyttävästi tuo kaunismuotoinen pää
kiharatukkineen kuvastui valkeata vastaan.

Hänen oma lordinsa oli aivan kaljupäinen.


Samuel Stern loi tuon tuostakin ihmettelevän katseen Thora
Thammersiin. Hänestä tuntui että hän oli muuttunut varsin
huomattavasti. Tai kenties hän vasta nyt, oltuansa poissa, huomasi
tuon muutoksen — — —

Vihdoin sanoi hän siitä jotain Thoralle.

Tämä kääntyi hänen puoleensa.

— Joku on sanonut: »Sorrow has its reward. It never leaves us,


where it found us.»… [Kärsimyksellä on palkkansa mukanaan. Se ei
jätä meitä koskaan sille asteelle, missä kohtasi meidät.] Toivon niin
olevan. Olihan muutos kylläkin tarpeen — — —

Sitten hän lisäsi hiljaa: Luulen että opin hiukan joka päivä —
teidän äidistänne!

Hän kääntyi pois — hänen silmänsä kimmelsivät kuin kyynelissä.

*****

Päivät tulivat ja menivät hiljalleen — loistavan kirkkaina ja tyyninä.

Kullanhohtoisina ympäröivät lehtipuut pihaa. Saatiin nauttia


tuollaisesta jälkikesästä, jommoisen syyskuu väliin lahjoittaa
lopettajaisikseen.

Tänä vuonna sitä kesti harvinaisen kauan.

Lady Hammond ei tahtonut lähteä Metsolasta, kun siellä vielä oli


niin kaunista, ja hän oli saanut Marit Hennerudin lupaamaan että hän
viipyisi siellä vielä jonkun aikaa. Hän oli jo vuokrannut itselleen
yhden Metsolan huviloista seuraavaksi kesäksi, ja hän koetti
houkutella rouva Thammersia lupaamaan että hänkin tulisi sinne
silloin.

Mutta rouva Thammers oli pudistanut päätänsä, hymyillen


taipumatonna.

Oikeastaan sai Thora aina hetkisen kuluttua kyllänsä lady


Hammondin seurasta. Siksi käytti hän aina ensi tilaisuutta
päästäkseen pakoon, — hänestä oli melkein väärin istua sisällä
juttelemassa tuollaisen ihanan sään vallitessa kuin nyt viime päivinä.

Hän meni nummelle sille paikalle missä Samuel Stern oli istunut
äitinsä kanssa.

Hän meni sinne nykyisin usein — sillä näköala sieltä oli


ihmeellinen nyt kuulakassa syysvalaistuksessa — ja myöskin siksi
että tuo paikka muistoineen liittyi hänen hiljaiseen sielunelämäänsä.

Kuinka aurinko paistoikaan korkealla, kirkkaalla taivaalla! Oli kuin


koko nummi olisi ollut tulessa, kuin olisi kaikkialla sinkoillut säkeniä
ilmaan.

Thora kävi istumaan kivellensä, keskelle tulvivaa päivänpaistetta.


Ja auringonsäteet tulivat tuttavallisina ja lempeinä tuomaan säästöön
jäänyttä kesän suloutta.

Thora nautti suurten, tyynten tunturien katselemisesta. Hänestä ne


soivat hänelle sisäistä voimaa ja hivenen omaa ikuista rauhaansa.

Rauhan saavuttaminen oli nyt oleva hänen pyrkimyksenänsä. Aina


enemmän rauhaa! [Geerge Sand.] Se oli nyt oleva hänenkin
mielilauseensa.
Puut näyttivät kulkevan häntä kohden noilta pitkiltä rinteiltä,
rauhallisina iloiten päivänpaisteesta… Ne olivat hänen ystäviänsä, —
hänen tuli kiittää niitä, sillä nekin olivat auttaneet häntä.

Punertavana seisoi pieni vaivaiskoivu hänen rinnallansa ja


vieressä muutamia vaalean kullanhohteen värittämää haapoja.
Vieno tuulenhenkäys pudotteli niiden lehtiä. Yksitellen ne liitelivät
maahan hänen ympärillään, synnyttäen hiljaista sipinää.

Äkkiä tuli tuulahdus, tuoden mukanaan kosken kohinaa, niin että


kuulosti kuin olisi se ollut aivan lähellä.

Tuo ei ollut hänelle mieleen. Tuo ääni ei soveltunut tänne, missä


tunsi olevansa niin valmis elämään nähden. Hänkin oli nyt alkanut
tuntea olevansa valmis — — —

Mieluummin hän kuunteli joen ääntä! Sillä oli luja tahtonsa, mutta
se ei ollut enää niin kiihkeä. Mitä se keväällä oli sanonut
hillittömyydessään, sen oli se jo unohtanut — — —

— Mitä sanot, joki, ohirientäessäsi? Onko kaikki lopuksi kääntyvä


hyväksi?

Ja joki suhisi hitaasti ja tyynesti: — Ei hätää! Päämääräänsä


saapuu kukin aikanansa!

— Niin, niin kyllä! vastasi Thora omiin ajatuksiinsa.

— Senhän tiedämme niin hyvin!

Mutta hän istui kumminkin paikallaan sitä kuunnellen. Hänestä se


lausui myönnyttelynsä niin varmasti ja vakuuttavasti, koko tuon
tyynen, tuntureita kohden kohoavan nummen kuultavaksi. — Niin,
niin kyllä! toisteli se vitkalleen ja varmasti.

21.

Thora ei tiennyt kuinka pitkälle iltapäivä oli kulunut. Hän otaksui


että oli myöhäistä ja aikoi juuri lähteä nummelta, kun näki Samuel
Sternin saapuvan sinne ja suuntaavan matkansa suoraan häntä
kohden.

Oli niin harvinaista, että he nykyisin kohtasivat toisensa nummella.


Siksi jäi Thora paikoilleen.

Samuel Stern kävi istumaan tavalliselle paikallensa.

— No, milläs mielellä te nykyisin olette? Olen huomannut teidän


käyttävän shakkipeliä ja minua tekosyinä, päästäksenne pakoon.

Hymyillen myönsi Thora hänen olevan oikeassa.

— Kuinka kauan te vielä viivytte Metsolassa?

— Ladymme on houkutellut minut lupaamaan että olen täällä vielä


jonkun aikaa, niin kauan kuin parantola on auki…

— Se on oikein, — te näytätte paljoa terveemmältä! — — — Ja


sittenhän teillä on täällä niin paljon tehtävää.

— Tarkoitatte että luen Sjurin kanssa… Niin, onhan siinäkin syytä


viipyä täällä, luulen että hänellä on siitä hyötyä.
— Ja sittenhän hän käy hienona kuin sulhaspuvussa joka päivä…
Meidän on kiittäminen teitä paljosta, sekä Sjurin että minun… Mutta
minä olen mustasukkainen Sjurille… Paidat ja sukat saa hän pitää
kernaasti minun puolestani, mutta miksi te ette ole lukenut minunkin
kanssani?

— Te olette tänään hyvällä tuulella.

— Oikeastaan on asianlaita aivan päinvastoin! Minun täytyy lähteä


pois tänä iltana — jo ennen illallista. Olen jo lausunut kiitokset ja
hyvästit tuolla alhaalla — ja jättänyt tervehdyksen teille!… Mutta
olipa onni, että tapasin teidät täällä!

Thora hymyili hiukan väkinäisesti. — Sehän oli merkityksetön


sattuma, kun te kerran olitte jättänyt tervehdyksen!

— Kuten tahdotte! Mutta minulle se oli onnellinen sattuma… Saan


kiittää teitä kaikesta, mitä te olette ollut Sjurille ja minulle!

— Minusta te olette niin leikkisällä päällä, sanoi Thora lyhyesti.

— Niin, tavallisesti on ollut toisin… Mutta vakavasti puhuen, oli


hauskaa että vielä tapasimme toisemme — eikö niin?

Thora ei vastannut mitään. Häntä loukkasi että hyvästijättö oli ollut


vain sattuman varassa. Sitä ei hän katsonut ansainneensa.

Mutta hän hillitsi mielenliikutuksensa. Kääntyessään Samuel


Sternin puoleen oli hän jälleen aivan tyyni.

— Kyllä, olen iloinen että sain tavata teitä ennen lähtöänne…


kiitollinen siitä että saan sanoa teille hyvästit!
— Mutta sallikaa minun nyt vielä sanoa muutama sana! Haluaisin
kernaasti tietää erään seikan… Meidän välillemme ei saa jäädä
mitään katkeruutta!

Hän oli käynyt hiukan kalpeammaksi kuin tavallisesti. Mutta hän


istui rauhallisena, kädet ristissä polven ympärillä. Hän vältti Samuel
Sternin katsetta.

Tämä oli käännähtänyt häntä kohden, ja hänen silmistänsä tuikahti


nopea katse Thoraan. Hän teki liikkeen kuin lähteäksensä — mutta
jäi sitten paikoilleen katsoen vaieten Thoraan.

Thora taisteli hetken, tyyntyäksensä täysin.

Sitten hän jatkoi: — Halusin sanoa teille jotain. Nyt luulen voivani
tehdä sen. Mehän olemme ystävät — eikö niin?

Samuel Stern nyökkäsi, ja Thora jatkoi hiukan epäröiden:

— Eräänä päivänä, kun puhuitte äidistänne ja mikä onni on


omistaa jonkun ihmisen rakkaus… sanoitte te, että on aina oikeutettu
saamaan tietää sen, jos on voittanut jonkun rakkauden… että
tietoisuus siitä voisi olla hyödyksi ja tukena elämässä — ja että
kenelläkään ei ollut oikeutta salata sellaista tunnetta. Teistä tuntui
raskaalta, ettette ollut koskaan saanut sanoa sitä äidillenne! — — —

— Silloin ajattelin: »Tuo on aivan totta!» Sentähden tahdon nyt


sanoa sanottavani: Aina, sekä silloin kun luulin rakastavani Arvidia
että myöhemmin… aina pidin ainoastaan teistä… teihin oli sydämeni
kiintynyt, te olitte haaveitteni esineenä! En sano tätä puolustuksena
itselleni… ja voihan tämä myöskin kuulua ihmeelliseltä. Mutta ehkä
tämä saattaa teidät käsittämään kaiken sen, mikä olennossani on
ollut katkeraa! —

— Nythän voin sanoa tämän kaiken, nyt, kun kaikki on


unohdettua! Mutta nyt teidänkin puolestanne täytyy sanoa, että olette
täysin antanut minulle anteeksi!

Samuel Stern teki torjuvan kädenliikkeen.

Thora jatkoi: — Ja näettehän te, etten minä ole enää onneton.


Olen oppinut paljon… edistynyt entisestäni… Olen tyytyväinen olooni
— kun te vain sanotte, että olette antanut minulle anteeksi!

Samuel Stern pudisti päätänsä.

— Teidänhän on annettava minulle anteeksi! Te tahdotte olla


ystäväni… Mutta se ei riitä minulle… Katsokaa, nyt olen jälleen
alkanut pitää teistä, mutta taaskin toisella tavalla kuin ennen…
Rauhoittukaa, ei teidän tarvitse pelästyä — en kiusaa teitä enää
koskaan! —

— Mutta tämä sinun täytyy saada tietää: noina pitkinä vuosina,


jolloin käyskentelin yksin, loin usein katseeni ovelle, kohotin päätäni
ja odotin etkö tulisi — käännähdin huoneessani katsomaan, etkö
hiipisi luokseni minua yllättämään — — —

— Pitkän taipaleen olen saanut kulkea elämäni tietä, ehtiäkseni


niin pitkälle, että nyt tyydyn siihen, että tiedän sinun olevan
olemassa… on ihanaa ajatella sitä — niin ei ainakaan ole yksin!

— Oli aika, jolloin olisin tahtonut vihata sinua. Nyt olen ehtinyt
edemmäksi — olen onnellinen siitä, että rakkauteni on niin suuri. En
sano: »Jää luokseni, sillä päivä ehtii ehtooseen, jää luokseni,
valokseni, kun suuri pimeys tulee.» Ei, minä sanon: »Mene! Mene
luotani! Tahdon rukoilla että jotain hyvää tulee osaksesi!»

— Tuo, että on nähnyt sydämensä tunteen kasvavan suureksi ja


ihanaksi… tietoisuus, että on ehtinyt siihen, ettei enää toivo mitään
itselleen, vaan ainoastaan toiselle, se saattaa aavistamaan sielujen
salaista yhteenkuuluvaisuutta, — saattaa uskomaan, että meissä on
jotain kunnioitusta ansaitsevaa, jotain siitä Jumalan kirkkaudesta,
joka on kuluttava tuli — — —

— Te tiedätte, että kansan piti pestä vaatteensa, ennenkuin Herra


tuli alas Sinain vuorelle. Ja asetettiin raja, ettei kukaan kävisi liian
lähelle pyhää — ja hukkuisi. Ja kansa vapisi ja seisoi loitompana —
——

— Monet meistä ihmisistä ovat hukkuneet — siksi että ovat


saastuttaen koskeneet pyhään! — — —

Hänen äänensä sointi oli käynyt lempeäksi. Thorasta oli, kuin olisi
vuolas virta hellyyttä tulvehtinut häntä vastaan.

— Sano ainoastaan, rakkaani, minne menet, niin että tiedän missä


elät ja hengität! Et aavistakaan mitä minulle merkitsee tieto, että sinä
olet olemassa!… Miksi kalpenet? Onhan kaikki, mitä välillämme on
ollut, vaihtunut rauhaan! — — Löytäköön jalkasi tien sinne, missä
sinun on hyvä olla!

Thora oli istunut pää alas painuneena. Nyt loi hän katseensa ylös,
ja se kuvasti tuskaa, hänen puhuessaan värähtelevin äänin:

— Suuntasin askeleeni tänne, siksi että arvelin että jos voisin


tehdä jotain sinun hyväksesi, niin olisi se parasta minulle! — — —
— Nyt en minä lähde sinun luotasi, vaan sinä minun luotani! On
sittenkin niin, että aika kaikki parantaa! Ja kun kammiosi ei enää ole
tyhjä, vaan siinä asuu kaksi onnellista, nuorta ihmistä — silloin on
tapahtunut se, mikä minulle on parasta. Silloin tiedän, että sinä olet
antanut minulle anteeksi!

He istuivat molemmat hiljaa.

Sitten nousi Samuel Stern ja ojensi molemmat kätensä Thoralle.

— Sano, että nyt ainakin uskot minua! Etkö usko minua


vieläkään?
Tahdon, että teet sen!

Hän kävi kiivaaksi.

Thora sulki silmänsä ja nojautui tukea hakien tuohon suureen


kiveen.

— Miksi en uskoisi sinua! Sinä sanoit täällä kerran, että olet


unohtanut kaikki tyyni, ja minä uskon sen! Olenhan ehtinyt kesän
kuluessa tulla siitä vakuutetuksi, olen nähnyt ja oivaltanut sen
sydämelläni ja sielullani ja kaikilla aistimillani — niin että sen kyllä
uskon täydelleen! — — —

Hiljaa lisäsi hän:

— Sinun täytyy antaa anteeksi kaikki mitä minä silloin sanoin!…


Nyt en enää ajattele sillä tavoin. Nyt ajattelen ainoastaan, kuinka
onnellinen olisin, jos sinun olisi hyvä olla!

Mielenliikutus valtasi hänet, ja hän koetti turhaan sitä vastustaa.


Samuel Stern istui hänen viereensä.

— Sano nyt minulle kautta tosimman totuuden, etkö uskonut


minua silloinkin, kun sanoit ettet voisi tehdä sitä koskaan… ja oliko
totta että olit unohtanut minut?

Thora kävi kuolonkalpeaksi.

— Se ei ollut totta! Mutta minä pelkäsin omaa itseäni. En luottanut


itseeni. Siksi täytyi minun pelastautua pakoon.

Samuel Stern nousi ja kulki muutaman askeleen. Mieli järkytettynä


katsoi Thora hänen jälkeensä.

Hän palasi, kumartui Thoraa kohden ja katsoi hänen


hermostuneesti vavahteleviin kasvoihinsa.

— Enkö minä voisi päästä Suden paikalle? Sehän ei ole täällä!

Ja hän anasti Suden paikan, laskien päänsä Thoran polvelle.

Saiko hän tehdä niin? Hän kysyi vielä kerran nöyrästi ja


kaihoisasti.

Thora ei voinut vastata. Hänen mieltänsä vavahdutti hiljainen


riemuntunne, joka purkaantui kyyneliin.

Samuel Stern ei puhunut enää mitään.

Silloin kumartui Thora häntä kohden. Hän halusi nähdä hänen


kasvonsa.

Samuel Stern itki.

You might also like