Lecture Notes in Computer Science 3469 Commenced Publication in 1973 Founding and Former Series Editors: Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen Editorial Board David Hutchison Lancaster University, 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, Switzerland John C. Mitchell Stanford University, CA, USA Moni Naor Weizmann Institute of Science, Rehovot, Israel Oscar Nierstrasz University of Bern, Switzerland C. Pandu Rangan Indian Institute of Technology, Madras, India Bernhard Steffen University of Dortmund, Germany Madhu Sudan Massachusetts Institute of Technology, MA, USA Demetri Terzopoulos New York University, NY, USA Doug Tygar University of California, Berkeley, CA, USA Moshe Y. Vardi Rice University, Houston, TX, USA Gerhard Weikum Max-Planck Institute of Computer Science, Saarbruecken, Germany
Alden H. Wright Michael D. Vose Kenneth A. De Jong Lothar M. Schmitt (Eds.) Foundations of Genetic Algorithms 8th International Workshop, FOGA 2005 Aizu-Wakamatsu City, Japan, January 5-9, 2005 Revised Selected Papers 13
Volume Editors Alden H. Wright University of Montana Computer Science Department, Missoula, MT 59812, USA E-mail: alden.wright@umontana.edu Michael D. Vose University of Tennessee Computer Science Department, 203 Claxton Complex, 1122 Volunteer Blvd. Knoxville, TN 37996-3450, USA E-mail: vose@cs.utk.edu Kenneth A. De Jong George Mason University Computer Science Department, 4400 University Drive, Fairfax, VA 22030, USA E-mail: kdejong@cs.gmu.edu Lothar M. Schmitt University of Aizu Aizu-Wakamatsu, Fukushima 965-8580, Japan E-mail: LMSchmitt@yahoo.com Library of Congress Control Number: 2005928446 CR Subject Classification (1998): F.1-2, I.2, I.2.6, I.2.8, D.2.2 ISSN 0302-9743 ISBN-10 3-540-27237-2 Springer Berlin Heidelberg New York ISBN-13 978-3-540-27237-3 Springer Berlin Heidelberg New York This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation, broadcasting, reproduction on microfilms or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer. Violations are liable to prosecution under the German Copyright Law. Springer is a part of Springer Science+Business Media springeronline.com Springer-Verlag Berlin Heidelberg 2005 Printed in Germany Typesetting: Camera-ready by author, data conversion by Olgun Computergrafik Printed on acid-free paper SPIN: 11513575 06/3142 5 4 3 2 1 0
Preface The 8th Workshop on the Foundations of Genetic Algorithms, FOGA-8, was held at the University of Aizu in Aizu-Wakamatsu City, Japan, January 5 9, 2005. This series of workshops was initiated in 1990 to encourage further research on the theoretical aspects of genetic algorithms, and the workshops have been held biennially ever since. The papers presented at these workshops are revised, edited and published as volumes during the year following each workshop. This series of (now eight) volumes provides an outstanding source of reference for the theoretical work in this field. At the same time this series of volumes provides a clear picture of how the theoretical research has grown and matured along with the field to encompass many evolutionary computation paradigms including evolution strategies (ES), evolutionary programming (EP), and genetic programming (GP), as well as the continuing growth in interactions with other fields such as mathematics, physics, and biology. A tradition of these workshops is organize them in such a way as to encourage lots of interaction and discussion by restricting the number of papers presented and the number of attendees, and by holding the workshop in a relaxed and informal setting. This year s workshop was no exception. Thirty-two researchers met for 3 days to present and discuss 16 papers. The local organizer was Lothar Schmitt who, together with help and support from his university, provided the workshop facilities. After the workshop was over, the authors were given the opportunity to revise their papers based on the feedback they received from the other participants. It is these revised papers that are included in this volume and follow the order in which they were presented at the workshop. In addition to these 16 papers, there were 2 invited talks: an opening presentation by Alden Wright and a closing presentation by Kenneth De Jong. These slides-only presentations are not included in this volume, but can be obtained from the authors upon request. A brief summary of these presentations is provided here. Alden Wright opened the workshop with a presentation titled Can Evolutionary Computation Theory Have Significance Outside of EC? and subtitled Can EC Theory Help Us To Understand Evolution?. The field of artificial life has been successful in reaching a wide audience with claims that artificial life experiments can give insight into natural evolution. Wright asked if EC theory can do the same? He proposed that EC theory might be relevant to some challenges in evolutionary research 1. These included: Analysis of the evolution of rates of mutation and recombination. Do optimal rates evolve? Analysis of the evolution of the information content of genomes. 1 Some of these challenges came from the website: http://evonet.sdsc.edu/evoscisociety/chall and oppors in e res.htm
VI Preface Analysis of genic selection and of conflict within genomes (e.g., segregation distortion, evolution of gene expression, etc.). How does evolution maintain the great complexity of organisms while also allowing rapid evolution in some areas? How is it possible that phenotypic variations do not destroy brittle interactions between the subsystems of an organism while still allowing for the variablity that allows for evolutionary innovations? Wright suggested that investigation of the genotype-phenotype map might give insight into the last two challenges. Kenneth De Jong closed the workshop with a presentation titled Unifying EC Theory. In this presentation he argued that developing a more unified framework for EC theory was important for further progress in the field. This was based on the observation that historically the field has evolved around a number of EA demes (GAs, ESs, etc.), resulting in deme-specific terminology and theory. We now have deme-independent EC toolkits that provide creative mixing and matching of EA components, but we have no theory to guide EA design at this level. De Jong outlined a strategy for developing such a theory. He suggested that we need to clearly distinguish between EAsasdynamicalsystemsandEAsas problem solvers. Adopting a dynamical systems view allows us to answer questions about trajectories, fixed points, etc., and makes contact with a large body of existing theoretical work in evolutionary biology, evolutionary game theory, and dynamical systems theory. Adopting a problem-solving view allows us to answer questions about the effectiveness of EAs for optimization, search, machine learning, etc., and makes contact with a large body of existing theoretical work from computer science, operations research, and artificial intelligence. De Jong argued that in both of these areas it is important to find a middle ground between theories that are too abstract to be helpful and too specific to be applicable to new situations. He gave several examples of how that might be done using a top-down strategy. He concluded by noting that several of the papers presented were nice examples of this middle theoretical ground, and expressed the hope that he would see more of them at the next FOGA. In between those 2 presentations 16 papers were presented on a wide range of theoretical evolutionary computation topics. We hope that you find them as interesting and provocative as we did. We fully expect that these papers will serve as a catalyst for further progress to be reported at the next FOGA workshop in 2007. March 2005 Alden Wright Michael Vose Kenneth De Jong Lothar Schmitt
Organization Program Co-chairs Alden Wright (University of Montana, Missoula, USA) Michael Vose (University of Tennessee, USA) Kenneth De Jong (George Mason University, USA) Lothar Schmitt (University of Aizu, Japan) Reviewers Lee Altenberg Anne Auger Thomas Baeck Wolfgang Banzhaf Hans-Georg Beyer Jürgen Branke Anthony Bucci Forbes Burkowski Kenneth De Jong Rolf Drechsler Stefan Droste Anton Eremeev Larry Eshelman James Foster Marcus Gallagher Oliver Giel Jens Gottlieb Walter Gutjahr Jonathan Hallam Nikolaus Hansen William Hart Jun He Robert Heckendorn Jeff Horn Christian Igel Yaochu Jin Jens Jägersküpper Thomas Jansen Tim Kovacs William Langdon Anthony Liekens Sean Luke Evelyne Lutton Keith Matthias Nicholas McPhee Peter Merz Martin Middendorf Heinz Mühlenbein Bart Naudts Silja Meyer-Nieberg Una-May O Reilly Riccardo Poli Adam Prugel-Bennett Colin Reeves Franz Rothlauf Jonathan Rowe Günter Rudolph Dave Schaffer Lothar Schmitt Marc Schoenauer Hans-Paul Schwefel Jonathan Shapiro Jim Smith William Spears Peter Stadler Chris Stephens Matthew Streeter Dirk Thierens Marc Toussaint Ingo Wegener Karsten Weicker Nicole Weicker Darrell Whitley Paul Wiegand Carsten Witt Alden Wright Annie Wu Xin Yao Sponsoring Institutions International Society for Genetic and Evolutionary Computation (ISGEC) University of Aizu, Japan
VIII Organization University of Aizu Institutional Support President T. Ikegami, Vice-President S. Tsunoyama Professor R.H. Fujii, Professor L. Pichl K. Ishikawa, Y. Nagashima, M. Nanaumi K. Doi, S. Ito, M. Oouchi, K. Takeyasu FOGA 2005 Poster FOGA 2005 Participants
Table of Contents Genetic Algorithms for the Variable Ordering Problem of Binary Decision Diagrams... 1 Wolfgang Lenders and Christel Baier Gray, Binary and Real Valued Encodings: Quad Search and Locality Proofs... 21 Darrell Whitley and Jonathan E. Rowe A Comparison of Simulated Annealing with a Simple Evolutionary Algorithm... 37 Thomas Jansen NP-Completeness of Deciding Binary Genetic Encodability... 58 Andreas Blass and Boris Mitavskiy Compact Genetic Codes as a Search Strategy of Evolutionary Processes... 75 Marc Toussaint Asymptotic Convergence of Some Metaheuristics Used for Multiobjective Optimization... 95 Mario Villalobos-Arias, Carlos A. Coello Coello, and Onésimo Hernández-Lerma Running Time Analysis of a Multiobjective Evolutionary Algorithm on Simple and Hard Problems...112 Rajeev Kumar and Nilanjan Banerjee Tournament Selection, Iterated Coupon-Collection Problem, and Backward-Chaining Evolutionary Algorithms...132 Riccardo Poli A Schema-Based Version of Geiringer s Theorem for Nonlinear Genetic Programming with Homologous Crossover...156 Boris Mitavskiy and Jonathan E. Rowe Coarse Graining Selection and Mutation...176 Jonathan E. Rowe, Michael D. Vose, and Alden H. Wright Perturbation Theory and the Renormalization Group in Genetic Dynamics...192 Christopher R. Stephens, Adolfo Zamora, and Alden H. Wright Optimal Weighted Recombination...215 Dirk V. Arnold
X Table of Contents On the Prediction of the Solution Quality in Noisy Optimization...238 Hans-Georg Beyer and Silja Meyer-Nieberg Rigorous Runtime Analysis of the (1+1) ES: 1/5-Rule and Ellipsoidal Fitness Landscapes...260 Jens Jägersküpper Population Sizing of Dependency Detection by Fitness Difference Classification...282 Miwako Tsuji, Masaharu Munetomo, and Kiyoshi Akama The Deceptive Degree of the Objective Function...300 Yun-qiang Li Author Index...315