This paper studies the use of coevolution to search for a cellular automaton (CA) solving the well-known density classification task. The Coevolutionary Genetic Algorithm (CGA) coevolves two non-interbreeding populations which interact as predator and prey. The main purpose of this paper is to illustrate some of the intricacies involved in the use of coevolution to solve a given task. Concepts from standard GA theory can be used to understand these problems. A simple modification is proposed to significantly increase the performance. 1 INTRODUCTION In nature, predator-prey interactions provide a driving force towards complexity. This because there is a strong evolutionary pressure for prey to defend themselves better (e.g. by running quicker, growing bigger shields, better camouflage ...) in response to which future generations of predators develop better attacking strategies (e.g. stronger claws, better eye-sight ...). Such arms races are characterized by an inverse fitness interactio...

CA rules can be classied automatically for a spectrum of ordered, complex and chaotic dynamics, by a measure of the variance of input-entropy over time. Rules that support interacting gliders and related complex dynamics can be identied, giving an unlimited source for further study. The distribution of rule classes in rule-space can be shown. A byproduct of the method allows the automatic \ltering" of CA space-time patterns to show up gliders and related emergent congurations more clearly. The classication seems to correspond to our subjective judgment of space-time dynamics. There are also approximate correlations with global measures on convergence in attractor basins, characterized by the distribution of in-degree sizes in their branching structure, and to the rule parameter, Z. Based on computer experiments using the software Discrete Dynamics Lab (DDLab)[22], this paper explains the methods and presents results for 1d CA. 1 Introduction Cellular automata (CA) are a much stud...

We define several qualitative classes of cellular automata (CA) behavior, based on various statistical measures, and describe how the space of all cellular automata is organized. As a cellular automaton...

A major impediment of cellular automata (CA) stems from the difficulty of utilizing their complex behavior to perform useful computations. Recent studies by [ Packard, 1988, Mitchell et al., 1994b ] have shown that CAs can be evolved to perform a computational task. In this paper non-uniform CAs are studied, where each cell may contain a different rule, in contrast to the original, uniform model. We describe experiments in which non-uniform CAs are evolved to perform the computational task using a local, co-evolutionary algorithm. For radius r = 3 we attain peak performance values of 0:92 comparable to those obtained for uniform CAs (0:93 \Gamma 0:95). This is notable considering the huge search spaces involved, much larger than the uniform case. Smaller radius CAs (previously unstudied in this context) attain performance values of 0:93 \Gamma 0:94. For r = 1 this is considerably higher than the maximal possible uniform CA performance of 0:83, suggesting that nonuniformity reduces con...

We describe a framework for exploring the evolution of adaptive behaviors in response to different physical environment structures. We focus here on the evolving behavior-generating mechanisms of individual creatures, and briefly mention some approaches to characterizing different environments in which various behaviors may prove adaptive. The environments are described initially as simple two-dimensional grids containing food arranged in some layout. The creatures in these worlds can have evolved sensors, internal states, and actions and action-triggering conditions. By allowing all three of these components to evolve, rather than prespecifying any of them, we can explore a wide range of behavior types, including “blind ” and memoryless behaviors. Our system is simple and well-defined enough to allow complete specification of the range of possible actiontypes (including moving, eating, and reproducing) and their effects on the energy levels of the creature and the environment (the bioenergetics of the world). Useful and meaningful ways of characterizing the structures of environments in which different behaviors will emerge remain to be developed. 1

We investigate the dynamics of competitive co-evolution in the framework of two miniature mobile robots, a predator with a vision system and a faster prey with proximity sensors. Both types of robots are controlled by evolutionary neural networks. A variety of efficient chase-escape behaviors emerge in few generations. These results are analyzed in terms of variable fitness landscapes and selection criteria. A new vision of artificial evolution as generation and maintainance of adaptivity is suggested and contrasted with the theory and practice of mainstream evolutionary computation. In a second stage, different types of ontogenetic changes applied to the robot controllers are compared and the results are analyzed in the context of competitive co-evolution. It is shown that predators benefit from forms of directional changes whereas prey attempt to exploit unpredictable behaviors. These results and their effect on coevolutionary dynamics are then considered in relation to open-ended evolution in unpredictably changing environments.

We use a very simple description of human driving behavior to simulate traffic. The regime of maximum vehicle flow in a closed system shows nearcritical behavior, and as a result a sharp decrease of the predictability of travel time. Since Advanced Traffic Management Systems (ATMSs) tend to drive larger parts of the transportation system towards this regime of maximum flow, we argue that in consequence the traffic system as a whole will be driven closer to criticality, thus making predictions much harder, A simulation of a simplified transportation network supports our argument.

. In this paper, we review the literature on statistical long-range correlation in DNA sequences. We examine the current evidence for these correlations, and conclude that a mixture of many length scales #including some relatively long ones# in DNA sequences is responsible for the observed 1=f -like spectral component. We note the complexity of the correlation structure in DNA sequences. The observed complexity often makes it hard, or impossible, to decompose the sequence into a few statistically stationary regions. We suggest that, based on the complexityof DNA sequences, a fruitful approach to understand long-range correlation is to model duplication, and other rearrangement processes, in DNA sequences. One model, called #expansion-modi#cation system", contains only point duplication and point mutation. Though simplistic, this model is able to generate sequences with 1=f spectra. We emphasize the importance of DNA duplication in its contribution to the observed long-rang...

Introduction Every linear cellular automaton ae can be associated with a regular language L(ae) of finite words: L(ae) is the collection of all finite subwords of configurations that arise after one application of the global map of the cellular automaton. Discussions of the language theoretic aspects of linear cellular automata and sofic systems, in particular with respect to their relation to the topology of the space of configurations, can be found in [8], [10] and [7]. In this paper, we will study two measures of complexity associated with L(ae) that are based on minimal finite state machines of a certain type. The first is simply the size of the minimal automaton for L(ae), or, equivalently, the number of left quotients of this language. For the second measure, one can exploit the fact that the languages L(ae) are no

The field of artificial life is enriching both the content and method of philosophy. One example of the impact of artificial life on the content of philosophy is the light it sheds on the perennial philosophical question of the nature of emergent pheonomena in general. Another second example is the way it highlights and promises to explain the suppleness of mental processes. Artificial life's computational thought experiments also provide philosophy with a methodological innovation. The limitations of the central arguments in Stephen Jay Gould's Wonderful Life and Daniel Dennett's Darwin's Dangerous Idea illustrate the value of this new method.