Vie artificielle Où la biologie recontre l'informatique

Citation

Adamatzky, A. (2003), "Vie artificielle Où la biologie recontre l'informatique", Kybernetes, Vol. 32 No. 7/8. https://doi.org/10.1108/k.2003.06732gae.001

Vie artificielle Où la biologie recontre l'informatique

Vie artificielle Où la biologie recontre l'informatique

Illustré ave JavaJean-Philippe Rennard (http://www.rennard.org)Vuibert (http://www.vuibert.com)Paris2002408 pp.ISBN 2-7117-8694-3Review DOI 10.1108/03684920310483261

Every generation of scientists is eager to gain immortality. The easiest way to do it is to find a novel field of science. If you lack creativity to construct the new field – just rename the old one. What was called cybernetics in 1950s became bionics and bio-cybernetics in 1970s and then was subsequently renamed to artificial life in 1980s. Due to stronger flux of scientists between disciplines and the emerging trend towards academic Renaissance, this newborn nomen was lucky enough to accommodate computer science, biology, physics, chemistry and engineering. Rapidly artificial life is becoming a theory of everything, and therefore, newcomers feel themselves completely disoriented in this soup of weird ideas, concepts, algorithms and implementations. This is why a good introductory book would be highly appreciated.

Jean-Philippe Rennard undertook a painful task of systematizing and popularising basic concepts of the artificial life. His efforts resulted in the very accessible and "politically correct" overview of the field. Right from the title page – Vie artificielle – and then all over the text, Dr Rennard keeps us reminding that "artificial life" is first of all "life" and only then "artificial".

The text is subdivided into seven chapters – that logically introduce fundamental concepts of artificial life – and two appendices – that provide biological foundations of evolution and basics of computer graphics. Essential concepts of artificial life are unfolded in the first part (chapters 1-4), from defining "artificial" to emergence to self-replication and recursions. Second part (chapters 5-7) deals with bio-inspired computations and natural distributed computing.

First chapter tries to answer the question "What is artificial life?" It starts with a short and pleasurable excursion in the history of artificial life, from Golem and Frankenstein to mechanical duck. Then, the entelechy is derived from thermodynamics and Maturana-Verala's autopoietic systems, self- maintaining and self-reproducing autonomous processes. The notion of life is considered in the context of universality, thus universal Turing machine is discussed in detail. To the end of the chapter, reader finds himself reading about Java classes and methods, which will be used in simulation examples all over the book.

Second chapter is about emergence and its role in the framework of complexity and non-linearity. Cellular automata are motherhood of emergence. This is why rest of the chapter deals with these one- and two-dimensional lattices of locally connected finite-state machines. After briefly mentioning Ulam and von Neumann, author introduces Conway's game of life. This species of cellular automata exhibits remarkably rich spatio-temporal dynamics and yet bounded growth of its configurations, and thus instantiates a non-trivial (i.e. emergence-based) mathematical model of artificial life. A reader will not only enjoy walking in the zoo of famous patterns – gliders and glider guns, oscillators and breathers, but also will be given a chance to implement his own cellular-automaton models based on Java classes described by the end of the chapter.

Universality and self-replication are two essential attributes of biological life. They are looked at through a prism of cellular-automaton lattices in third chapter of the book. A computational universality, i.e. an ability to implement any logical function, is introduced in the chapter using game of life cellular automata. Author shows how to represent truth-values in states of mobile compact patters, or gliders, travelling on the lattice. Then basic logical gates are technically implemented via collisions of glider streams, and cellular-automaton model of universal Turing machine is exemplified. Wolfram and Heudine phenomenological classifications are introduced to demonstrate emergence of non-trivial (now coined as "computation at the edge of chaos") properties. Rest of the chapter, lavishly illustrated, is devoted to Neumann's self-replicating automaton, its version simplified by Codd and Langton's loops. Typically, for this book we can take a delight in studying Java implementation of Morita-Imai's cellular-automaton self- replicating system.

Life is recursive is the thesis of fourth chapter. The chapter introduces recursive algorithms and mathematics of self-similarity, which then exploited constructing biomorphs and L-systems. Biomorphs give us rather aesthetic insight of "would-life" life forms invented by Dawkins, while Lindenmayer's systems offer a formal-language-based approach to imitation of growth and form. Customary, we can also get hands on L-systems Java applet, analysed in full details.

Fifth chapter shows how to do optimisation using genetics and evolution. The chapter starts with detailed informal introduction to genetic algorithms followed by excursion in the field of evolutionary programming and nature- inspired parallel optimisation and genetic programming. A brief discussion of advances in evoware, including evolutionary electronics, programmable field array, nicely indicates promising ways of evolving in silica. A detailed study of Java classes gives us a chance to implement our own computer experiments in natural optimisation.

Most up-to-date concepts of bio-inspired distributed intelligence – from social insects to robotic collectives – are condensed in sixth chapter. This is the chapter about swarms, natural and artificial. The chapter starts (a concept of emergence in collectives is introduced) and finishes (a Java applet is vivisected) with Reynold's boids. Author shows how self-organization and stigmergy (interaction via environment) can be applied to design sensible algorithms for optimisation on graphs. This is exemplified by the approximation of shortest path in a colony of ants. Animats and Brainteberg vehicles serve as a prelude to overview collective and evolutionary robotics. Author, quite reasonably selected two demonstrations of collective robots: pioneer Beckers-Holland- Deneubourg experiments in collectives sorting in robot swarms and Kube- Zhang findings in cooperative activities in robotic collectives. All usual suspects are presented in the sections in evolutionary robotics – Sims's virtual creatures, Pollack's evolved robots, and Yim modular self-configurable robots.

The last, seventh, chapter is about the origination of virtual life. Initially, readers are warmed up by Core Wars, where tiny assembler programs fight for computer memory resources. Then, an exciting concept of programmable matter is introduced. Unfortunately, we could not find any references to Toffoli-Margolus ideas of programmable matter; however, we have enjoyed a brief introduction to Rasmussen's self-assembling automata and Dittrich's artificial chemistry.

There are two appendices. The first one introduces the basics of biological evolution, from Lamarck and Darwin to Kimura and Kauffman, and informal excurse into elementary genetics and cell biology. Second appendix teaches how to write Java applets for visualisation.

Another advantages of the book include quite comprehensive glossary and exhaustive list of references. Extensive bibliography will give a quick source of references to professionals.

The book is self-consistent, accessible to amateurs and comprehensive in subject representation. Those thinking to enter this exciting field of cross- disciplinary research will appreciate the text. The book is a lovely piece of work – fun to browse through and pleasure to read. It is a must, if you need an informal introduction to artificial life, digital biology and bio-inspired computation.

Frenchmen have everything that others lack – healthy food, comfortable climate and wonderful wines – now they have got Jean-Philippe Rennard's book on artificial life. Let us hope that the book will be promptly translated into English so that the rest of the world will enjoy it.

Andrew AdamatzkyComputing, Engineering and Mathematical Sciences, University of the West of EnglandE-mail: Andrew.Adamatzky@uwe.ac.uk