Abstract From available data, we show strong positive spatial correlations in the qubits of a D-Wave 2000Q quantum annealing chip that are connected to qubits outside their own unit cell. Then, by simulating the dynamics of three different spin networks and two different initial conditions, we then show that correlation between nodes is affected by a number of factors. The different connectivity of qubits within the network means that information transfer is not straightforward even when all the qubit-qubit couplings have equal weighting. Connected nodes behave even more dissimilarly when the couplings’ strength is scaled according to the physical length of the connections (here to simulate dipole-dipole interactions). This highlights the importance of understanding the architectural features and potentially unprogrammed interactions/connections that can divert the performance of a quantum system away from the idealised model of identical qubits and couplings across the chip.

Abstract Reservoir Computing is a computing model ideal for performing computation on varied physical substrates. However, these physical reservoirs can be difficult to scale up. We propose joining various reservoirs together as an approach to solving this problem, simulating physical reservoirs with Echo State Networks (ESNs). We investigate various methods of combining ESNs to form larger reservoirs, including a method that we dub Restricted ESNs . We provide a notation for describing Restricted ESNs, and use it to benchmark a standard ESN against restricted ones. We investigate two methods to keep the weight matrix density consistent when comparing a Restricted ESN to a standard one, which we call overall consistency and patch consistency . We benchmark restricted ESNs on NARMA10 and the sunspot prediction benchmark, and find that restricted ESNs perform similarly to standard ones. We present some application scenarios in which restricted ESNs may offer advantages over standard ESNs. We then test restricted ESNs on a version of the multi-timescale Multiple Superimposed Sines tasks, in order to establish a baseline performance that can be improved upon in further work. We conclude that we can scale up reservoir performance by linking small homogeneous subreservoirs together without significant loss in performance over a single large reservoir, justifying future work on using heterogeneous subreservoirs for greater flexibility.

The field called Artificial Life (ALife) coalesced following a workshop organized by Chris Langton in September 1987 (Langton, 1988a). That meeting drew together work that had been largely carried out from the 1950s through to the 1980s. A few years later, Langton became the founding editor of this journal, Artificial Life, which started its life with Volume 1, Issue 1_2 in the (northern) winter of 1993/1994.1 This current issue therefore begins the 30th volume and 30th year of Artificial Life. We think this is a milestone worth celebrating!In the proceedings of that first workshop, Langton famously defined ALife as the study of “life as it could be,” of “possible life,” in contrast to biology’s study of “life as we know it to be” (on Earth). His stated aim was to derive “a truly general theoretical biology capable of making universal statements about life wherever it may be found and whatever it may be made of ” (Langton, 1988b, p. xvi).Central to ALife is a debate that has continued sporadically since the field’s formal inception. This explores what should be properly counted as living phenomena and what should not. As a field, ALife has sustained a diverse range of opinions on this topic and has simultaneously investigated even contradictory perspectives. As the field gained attention in the mid-1990s, this was partially responsible for criticism that ALife had a lack of focus, a looseness with metaphor and the use of terminology, and a lack of grounding in fundamental biology (Horgan, 1995; Smoliar, 1995). Such risks were explicitly preempted by Nils Aall Barricelli (1962) as he pioneered digital evolution in the 1950s and 1960s. Similar criticisms, however, have also been leveled at biology for failing to agree on the necessary and sufficient characteristics, hence a definition, of life itself. This is something considered even by Aristotle; countless books, including those by Schrödinger (1944), Monod (1971), Rosen (1991), and Morange (2008), have been written on the subject in the 2,300 years since that time. Hence it is hardly surprising that ALife, a relative newcomer to the debate, has struggled with the same issue, especially because it proposes to broaden biology’s definition, whatever that may be.Possibly as a result of ALife’s diversity, the field resembles more a tangled bramble of subdisciplines than some neat “Tree of Life” (Figure 1). Most of its current topics build directly on work published in the earliest workshop proceedings (Langton, 1988a; Langton et al., 1992) and the first double issue of this journal. Indeed, we recommend that newcomers to the discipline, and even more established members, (re-)read those early works. They reveal that the relationships between the topics were as tangled then as they are now. Arguably, this is because tight classification seems less interesting to the community as a whole than exploring links between fields; the nodes of this complex network are less important than its edges.We venture to claim that ALife is an outlet for the parts of ourselves that are “scientific misfits,” the parts that do not fit naturally into a single academic discipline. This does not undermine the field’s validity, the quality of its practitioners, or the rigor with which ALife research can be conducted; rather, it highlights a strength, namely, that we cherish the opportunity to be creative, to think outside the silos that today’s science can inflict regarding “value”: departmental foci and territorial stakes, government funding priorities, industry needs, financial reward, initiation of hot topics and trends, promotion paths, employment opportunities, and publication citation metrics. ALife is arguably not a straightforward way to meet such goals.We have found anecdotally during decades of ALife conferences, meetings, and email exchanges that a diversity of views and explorations is a key feature that maintains individuals’ interest in the ALife research community. As a community, we prize novelty, innovation, and open-endedness in approaches to our work as much as we prize them in the outcomes of our research. We have also been historically accepting of a wide range of opinions, have been willing to openly debate them, and have welcomed explorations made by hobbyists or professional academics in their spare time beyond that devoted to “serious” research in traditional and better-supported fields. As a result, ALife remains fresh and fun at 30 plus years, and attendance at an ALife conference can be as exciting as entering a new field, even for those of us who have been involved a long time.But, if the field is so diverse, does it have a common goal? Yes, and this does not seem to have changed much in the last 30 years, as the word cloud from 1993–1994 (Figure 1) illustrates. We would add, though, that it is not simply to study life as it could be using a diversity of approaches; it is also to design it, build it, play with it, and enjoy it from a variety of perspectives. The playful aspect has been criticized repeatedly, especially but not exclusively in the contexts of synthetic biology and cloning, where the manipulated media match those of natural biology. One criticism leveled at practitioners in this subdomain is a perceived need to “play God” (Douglas et al., 2013), usually a male one (throughout Helmreich, 1998), with Frankensteinian connotations of upsetting the natural order (Kember, 2003, p. 56). There may be some elements of this in ALife. But our (one male and one female) editors’ sense of the dominant engagement between world builder and worlds here is one of curiosity and wonder, rather than a drive for control or mastery. As pointed out in some accounts of the God-likeness of ALife world builders (Kember, 2003, p. 97), a lack of control is in fact frequently encountered; it may even be desirable (Brooks & Flynn, 1989; Kelly, 1994). Hence the sense of exploration and playfulness not only endures; arguably, it is fundamental.Playfulness is something we frequently see in submissions to the journal. In fact, we wish that some projects and submissions were less “tinkering” and better structured! We need our journal’s submissions to follow at least the main requirements of rigorous science, including well-stated hypotheses; clear context setting; reproducability; careful analysis; and detailed exploration of outcomes, benefits, and their implications. Yet even in the more formal submissions, playfulness and enjoyment need not be diminished; they underpin thoughtful and well-contextualized research in a collective and individual effort to (a) understand the aspects of biological, ecological, and societal systems that facilitate or generate interactions or behaviors typically understood to be characteristics of living systems and (b) replicate the interactions and behaviors of life in media other than those from which natural organisms are constructed.Ultimately, then, our journal’s scope remains more or less constant after 30 years. Studies of life’s behaviors and interactions realized through technology are always welcome. The media might change. There may remain long-standing disagreement in the community whether something is “really alive” (Pattee, 1988). There is sometimes a tendency even to ignore or set aside this question and delve recklessly into biological metaphors; we have been warned repeatedly. But all of this has been shown over the years to matter far less to our community than whether our novel study systems behave in ways we deem to be relevant and engaging.With the design of engaging systems a high priority, it is little wonder that ALife continues to be enthusiastically embraced and broad: compare Figure 2, a title-term plot (1993–2023), and Figure 1, a keyword cloud (1993–1994). Its topics have been surveyed before (Aguilar et al., 2014), even since the early days (Farmer & Belin, 1992), but here we attempt to manage ALife’s bramble of subdisciplines with a simple taxonomy and 30 years’ hindsight. We then discuss the articles published in this issue of the journal, before concluding with a discussion of some important ethical and practical concerns for future ALife technologies.Today’s interdisciplinary study of ALife covers a variety of aspects.2 Not all researchers in these areas would consider themselves to be studying ALife per se; nevertheless, here we describe a simple taxonomy of research topics in this context. As noted, these topics do not form a neat hierarchical tree but rather form a tangled meshwork of interconnected and interrelated aspects that encompass ALife activities. Our classification is not exhaustive. A complete and detailed scope would be short-lived, because the field shifts with changes in technology, understanding, and culture; its boundaries are ill defined, and as subareas mature, they may calve off, form new (sub-)disciplines; and acquire new names.This topic considers individual artificial entities that represent or behave like artificial organisms. It can be roughly broken down into three areas: wet, robotic, and virtual ALife (Bedau, 2003; Farmer & Belin, 1992).Wet ALife covers artificial entities made from “wet” chemicals and their reactions. Examples include chemical systems like motile droplets (Čejková et al., 2017); minimal protocells constructed either bottom-up, from biochemical components, or top-down, by stripping out nonvital components from natural cells (Rasmussen et al., 2009); and natural organisms modified by synthetic biology techniques, such as gene engineering (Hanczyc, 2020).Robotic ALife covers artificial entities engineered from “dry” components. The robot bodies may be “hard,” mostly rigid, stiff mechanical components with relatively few degrees of freedom, or “soft,” mostly pliable, elastic components with very many degrees of freedom (Lee et al., 2017; Yasa et al., 2023). The “brains” of these systems are often some form of neural controller implemented in a standard computer. The area includes evolutionary robotics, especially when allowing body and brain to coevolve (Doncieux et al., 2015; Nolfi & Floreano, 2000); swarm robotics, a population of homogeneous or heterogeneous robots cooperating to achieve a task, from nano-scale to macro-scale systems (Schranz et al., 2020); and cognitive robotics, focusing on the intelligence, learning, and adaptation of robots (Cangelosi & Asada, 2022).Virtual ALife covers artificial entities that exist in a virtual environment as software running on a computer or over a network of computers. The research goal is for these entities to actually live, not to be mere simulations of life (Pattee, 1988). What this means, and whether it is possible, has been a subject of some philosophical debate, especially in the early days of ALife. Life is a dynamic process; “life is a verb, not a noun” (Gilman, 1904, chap. X). Does that process need to be physically realized in carbon-based chemistry, or can it be realized in different material substrates undergoing different physical processes—or can it even be virtual? Does the matter matter? The computational viewpoint tends to say that it does not (Langton, 1986; Ray, 1992); the biological viewpoint tends to disagree (Maturana & Varela, 1980; Morange, 2008; Rosen, 1991). Pattee (1988) explored criteria for realizations of virtual life.Some of the issues around virtual ALife are sidestepped by simulated ALife, which makes no claims about the simulation being alive, and by embodied ALife, which focuses on embodiment and situatedness as necessary components of a living entity (Clark, 1997); a form of embodiment may be possible even in virtual systems (Stepney, 2007). Consideration of embodied ALife includes aspects of growth, morphogenesis, and development (Doursat et al., 2012) and more generally of self-construction, self-assembly, and autopoiesis (McMullin, 2004; Varela et al., 1974).In addition to these of the material or of the and computational aspects of the and are also of the whole living the in a through its and in to and population behaviors and and on to intelligence, all have an computational aspect (Stepney, and all of the subareas can be for in evolutionary swarm et al., & might be at the of and swarm ALife, the individuals’ bodies and swarm collective can result from swarm et al., the tangled of the that is for all parts of our not only for artificial organisms do and in complete They exist as and in of including some and in This of ALife of artificial its with artificial organisms of organisms undergoing with and in the context of use simple of these to for of their use can often have and surprising et al., natural evolution is much more in of population of from gene to gene the of development and interactions between and defined more natural of evolution are the of ALife evolutionary as of in the including the articles by et et et and also such as collective in or and & of biological like these can drive in swarm simulations can these behaviors can from only in of their other and an environment of entities and all of which of different ways of making a that can themselves and & 2008; et al., This of different to and through and & and & and et al., as the can facilitate and interactions et al., and all which to of and & and are also fundamental to biology. ALife study systems can from natural is in their to researchers to far from naturally et al., of the this can us to what are necessary for life and what are and of the single of natural life that we have to ALife can not only the of it can design and play very different from those that have or could on to and life’s ALife main to do and virtual worlds a to the of complex evolutionary and techniques, including complex systems and can general of and to us understand them, by natural life has is its ” p. the diversity of natural organisms and their ways of making a The study of this the of evolution et al., et al., or open-endedness in general et al., open-endedness started of is the question of the of life et al., it possible for systems to to being what do from to life living entities do of do in evolution & which the of new of and and is a to the of that can Does this the and evolution of and evolution & may be considered in the context of and as as physical of for and & of et al., and also into this of ALife main in this area is the use of a variety of systems to build simulations of include 1986; et al., & artificial & and like 1992), & and et al., A of to and to its or in such simulations and in physical systems et al., & is a to the field from diverse have and and behaviors using many media and by et al., like and remain and are not universal or constant We therefore to our from different or the same or as we the of an an engineered has a and including in and 2003; & our attention as they the attention of and who them at the in whatever and they and whatever the for the Hence some have of ALife that the of the field by and studying work that has on those we its have ALife work since the field’s into and a variety of (Aguilar et al., 2003, & set the context for today’s and us of the of the to which we may be to ALife researchers what might be the of these What is (Bedau, 2007). life is physical in is no or “life living systems to be different from One goal of ALife is to life in a way that is not on of natural life but that would to all the possible of life and to necessary and sufficient for it & discussion topics ALife, all on in other of this relationships between and whether embodiment is or for the of the of and the of in boundaries and therefore living to of such as and the grounding and of what it for an to sense or its the between and whether biological evolution is in fact as as what and to and the between novelty, and ALife is a for and ethical issues of and living physical and virtual organisms are also (Douglas et al., & consider these as as the and societal of especially if they directly with and The Artificial Life conference includes a on and that we We submissions to the on ALife, and the have not been as often as we would traditional academic articles and software are a common way to opinions and debate issues to ALife, this field in has a body of into the relationships between and technology through ALife may for on sometimes considered as of the of the field, but they also on more work from the in and computer and from ALife’s 2015; 2004; et al., In ALife’s techniques, such as evolutionary virtual and have found many in computer and media the focusing explicitly on ALife’s ALife were for and before their with ALife was for and are of our field that were first as computer of by to biological in the of computer ALife research often and and for new ALife and use virtual may be to ALife, but their software might on or even on computer As with science, are and for complex are in relevant ALife research have been to the general and the of ALife especially through and has its and synthetic biology natural is a but are also many of evolutionary that for In our simulation can many complex phenomena and us societal and They can also be to and that the of living organisms. There are many for ALife systems to with biological, ecological, or societal This is but has risks and careful of and are but beyond our field are relatively few and far between to a current the of computer and include work as the ALife conference for But sometimes when the into other such as synthetic or evolutionary and we may of ALife is into other that to our work even as a community, we do not always the this issue, we of articles to the and of ALife and have also other submissions that fit our We articles relevant to the the et al., in their and The of a on the need for in and biological and its in synthetic biology. This the need for in computational and with biological and in and a a can use to build the need for This the use of simulation to the of complex behaviors from simple in for Artificial for for and us on a and artificial intelligence, the approaches and the more and neural network relevant to an Artificial Life ethical issues are but not the and in Artificial of a at an of a gene network and the complex This a that can to biological and to complex that could be into an artificial to of the on to the of and robots by and This material from the and early to the and long of the et al., in of of Artificial Life in and about or by ALife. is a topic at the ALife conferences, but less so in the journal, and so we its this issue with The by science Artificial The for a from 30 years The by in and from an ALife can we that our research makes to and to our over the 30 This has not been a stated of researchers in ALife over the 30 years. do we the world do we systems that directly with biological, ecological, and systems to their researchers in ALife would do to the issue of failing to with technology, innovation, and change. As are like other either a or a a not Our a many benefits, but we also many We have long that the is to of We have that are more the of for and We have since the that our use of chemical the that chemical and and that and are We have technology has changes in the through the in all of these and in been and at the complex new have or where often not and have to with change. In some has been perceived to be for have and in the regarding In other may to be rather than and for the use of modified fresh issue to artificial life is what to do about the of artificial intelligence, like and like systems are to be the work of and they undermine the of media that aim to something they are for and and in their on and they Life could follow the of artificial We could we have a physical capable of of goal and and only then to its and its in our this is to The of swarm robotics, one interest of ALife to is an of a research we is and should have been in general to be and some are for this to than following the ALife should do something that us as a of ALife for societal and as are a in the Indeed, of ALife systems to with natural and to better manage and them is a natural aspect of our outcomes, is We should also be and for the and of artificial living systems to or research and It is better to be because ALife’s technology has more than have at least to no is But in this is not a simple goal to We and between different and and their needs, and we have to and and we have to the of those of and other This is It is the and of we as researchers to with complex and systems are to if we were to our field to We are to do are we This is a future the ALife community should before the 30 years at years, we would be to our

We start this issue with two research articles.The first, by Bullinaria, "Simulating the Effect of Environmental Change on Evolving Populations," uses simulations to study how populations can adapt to rapid environmental changes.It demonstrates the effect of model parameters on system robustness and the consequent care needed when modeling environmental change.The second article, by Collard, "Processionary Caterpillars at the Edge of Complexity," explores a model of the collective behavior of pine tree processionary caterpillars.The article begins by modeling the caterpillars' actual behaviors, then is extended to explore some more complex collective behaviors.Following these contributions, we include a series of articles forming a special issue of extended papers from the Artificial Life conference held online and in Trento, Italy, in 2022 (Holler et al., 2022).Thanks to Silvia Holler, Barbora Hudcová, Richard Löffler,

We start this issue with two research articles. The first is by Rainwater, on “Self-Organization and Phase Transitions in Driven Cellular Automata.” In this article, the author explores the altered dynamics of Conway’s Game of Life cellular automaton that occur when cells are forced to remain in the “on” state, which would not normally remain on if they were simply following the standard rules for state transitions. The second article is by Milano and Nolfi, on “Interaction Rules Supporting Effective Flocking Behavior.” They investigate using more complex interaction rules for the individual agents and the effect of field of view on flocking and aggregation behaviors. Somewhat counterintuitively, models with restricted sensory input from a restricted field of view are shown to aggregate more effectively than models with access to more sensory information.Following these, we include a series of articles forming a special issue on open-ended evolution (OEE), which presents work that either follows on from the fourth Workshop on OEE, held at the Artificial Life conference in Prague and online in 2021, or has been developed since then to build on the two previous special issues on OEE (Artificial Life, 25[1–2]). Alastair Channon, Mark Bedau, Norman Packard, and Tim Taylor serve here as guest editors of the special issue.

In closing Artificial Life’s 30th anniversary volume, we would like to take the opportunity to look ahead. Some readers may feel that the field’s emergent wandering is desirable. It has been interesting (and fun!) most of the time, and valuable some of the time. Who knows what may come of our field in the future? Maybe yet more novelty will emerge. If we wander like a flock of boids looking to our nearest neighbors, addressing the interests of connected practitioners, without oversight or planning, where are we all likely to end up? We may travel toward somewhere even more interesting. Or, like a glider on Conway’s grid, we may find ourselves reinvented, in very much the same state as we were 30 years ago, but somewhere new on a grid that has changed deeply around us. Given what we Artificial Life researchers know about successful search of immense spaces, neither pure exploitation nor pure exploration yields the best outcome. If we do want to find ourselves located somewhere important, we need to balance exploratory wandering with agreement on the functions we will use to assess the value of new places we discover: a quality–diversity approach.Most of us are guided by our intuition, by discussions with colleagues, by papers we read and seminars we attend. So we are not conducting a random walk or an aimless drift. But meta-level guidance of our field can still, potentially at least, aid us in shifting the field into a region where we want it to go. Some should wander off to explore in their own directions, ignoring trends and guides. That is an aspect of our field that many of us treasure. But if everybody does this all the time, we will remain a small, eccentric niche in a vast research web. The odds of our field having a centralized and substantial impact in this way seem low at best and negligible on average. If we adopt this approach for the next 30 years, we may continue to spin off valuable research while our core retains its diversity and its arguably “alternative” mind-set, exploring the unknown unknowns. This is potentially valuable. But it is not the only thing we should do.Meta-level guidance can be provided by workshop themes, calls for articles on special issues, textbooks, PhD supervisors advising cohorts of students, competitions that motivate researchers to tackle specific problems, and position papers like this editorial. So here goes.The Artificial Life journal should publish more translation research. We should publish work that applies the principles, understanding, and techniques of our discipline to tackle real, urgent, complex problems that impact human society and Earth’s other physical and ecological systems. Our field is full of diverse, well-read, well-educated, relatively privileged individuals. Many of us work at institutions that receive government funding derived from citizens’ taxes. We have a moral obligation to do something valuable. This is essential, especially now, to maintain fair and equitable processes of governance, access to health care and education, resource distribution, and environmental protection and sustainability under an unstable global sociopolitical situation and an increasingly unstable climate.For Artificial Life to publish such important work as this, we first need to receive it. We know the work is out there. Sometimes it is published in the discipline where the translational impact is most keenly felt, perhaps in robotics, ecology, social science, or epidemiology. Much, much more can be done. We would welcome articles on translational topics in Artificial Life where the links between real-world impacts and our field’s fundamentals can be explored. So, we encourage you to think about how you, your collaborators, and your students can tackle societal and environmental issues for everyone’s benefit without lining the pockets of multinational tech firms. So many deep problems need to be solved immediately. How can we justify turning our backs? We would love to receive, review, and publish your most impactful research in Artificial Life. Send it along!Of course, our views on where Artificial Life ought to go are just two of many. In this issue, Beer offers his personal reflections on Artificial Life’s past, and an opinion on the field’s potential directions, in “(A)Life as It Could Be.” Another of our long-term researchers, Harvey, holds views partly in opposition to those of Beer. He has expressed these in a “Comment on Randall D. Beer’s ‘A(Life) as It Could Be.’ ” We, like these authors, hope their contrasting texts stir up some debate.One of the strengths of the discipline is the wide range of knowledge and techniques that can be brought to bear on problems: ALife provides inspiration and tools both for rich construction media and for a wide range of design approaches. Several articles in this issue demonstrate that breadth of approach.Rusin and Medvet provide insight into one of the major themes of ALife, embodied intelligence as demonstrated in robots, in their article “How Perception, Actuation, and Communication Impact the Emergence of Collective Intelligence in Simulated Modular Robots.” They take an evolutionary approach to design, using modularity to restrict the design and optimization search spaces. Practical robots need to be simple to build but also highly functional; the authors show how these competing objectives may be achieved.Dubey et al. are also interested in optimal design of engineered structures, here using a case study of bridge trusses. In “Evolving Novel Gene Regulatory Networks for Structural Engineering Designs,” they apply an evo-devo approach as an improved way to automate the search for better structural engineering designs. They exploit many of the techniques of ALife—evo-devo, GRNs, neural networks, and genetic programming—to develop an approach applicable to both 2-D and 3-D structures.Continuing with the theme of embodied agent control, Langer and Ay present “Outsourcing Control Requires Control Complexity.” Here the focus is on one of the promised features of embodiment: the capability to outsource some complexity, through interactions between agent body and environment. The authors demonstrate that effective outsourcing still requires enough complexity in the controller to exploit the environmental capabilities.In his article “Emergence and Criticality in Spatiotemporal Synchronization: The Complementarity Model,” Scirè investigates the core ALife concepts of emergence and self-organization through the lens of core ALife tools, criticality and dissipative dynamical systems, here with N coupled 2-D oscillators. The spatiotemporal dynamics of the studied system exhibit a number of emergent properties, including a variety of power law–style avalanches. The author notes how these fundamental properties may have application in some origin of life theories.An important feature of translating from abstract models to the real world is to remember the noise and limited precision of the latter. Thresholding and fuzzy limits are a way to model such constraints. Lawry, in his article “Heterogeneous Thresholds, Social Ranking, and the Emergence of Vague Categories,” applies these ideas to social decision-making scenarios via learning fuzzy categories.In a submission close to Artificial Life’s central idea to shed light on biology via modeling, Bull provides “On Recombination.” His letter offers a potential explanation of the relationship between the evolution of sex, recombination via meiosis, and the fitness benefits for individuals gained due to the resultant smoothing of the fitness landscape that it generates.ALife researchers have long sought to understand, model, and potentially realize cognition by creating embodied agents. This interest predates the field’s formal inception, and Braitenberg’s (1984) Vehicles provided something of a cybernetics-originated prototype in this area, published well before the first conference in Artificial Life (Langton, 1989) and prior to the journal’s first issue, yet still influential today. In “New Directions (and Insights) in Braitenberg Vehicles and Cognitive Science,” Alicea reviews Hotton and Yoshimi’s recent book, The Open Dynamics of Braitenberg Vehicles, which takes that early work in a new direction.Following in this established line of inquiry, Adami, in “How Brains Perceive the World,” focuses, for the fourth installment of his five-essay series, on building a sense of timing and an attention mechanism in artificial brains.We hope that you enjoy this issue, and the next 30 volumes, of Artificial Life.

We argue that attempting to quantify open-endedness misses the point: The nature of open-endedness is such that an open-ended system will eventually move outside its current model of behavior, and hence outside any measure based on that model. This presents a challenge for analyzing Artificial Life systems, leading us to conclude that the focus should be on understanding the mechanisms underlying open-endedness, not simply on attempting to quantify it. To demonstrate this, we apply several measures to eight long experimental runs of the spatial version of the Stringmol automata chemistry. These experiments were originally designed to examine the hypothesis that spatial structure provides a defense against parasites. The runs successfully show this defense, but also show a range of innovative, and possibly open-ended, behaviors involved in countering a parasitic arms race. Commencing with system-generic measures, we develop and use a variety of measures dedicated to analyzing some of these innovations. We argue that a process of analysis, starting with system-generic measures but going on to system-specific measures, will be needed wherever the phenomenon of open-endedness is involved.

Abstract This tutorial covers physical reservoir computing from a computer science perspective. It first defines what it means for a physical system to compute, rather than merely evolve under the laws of physics. It describes the underlying computational model, the Echo State Network (ESN), and also some variants designed to make physical implementation easier. It explains why the ESN model is particularly suitable for direct physical implementation. It then discusses the issues around choosing a suitable material substrate, and interfacing the inputs and outputs. It describes how to characterise a physical reservoir in terms of benchmark tasks, and task-independent measures. It covers optimising configuration parameters, exploring the space of potential configurations, and simulating the physical reservoir. It ends with a look at the future of physical reservoir computing as devices get more powerful, and are integrated into larger systems.

The Reservoir Computing (RC) paradigm is a supervised machine learning approach that makes use of the inherent processing capacity of dynamical systems. Using the system's transient response to an external input, delayed chaotic systems offer rich dynamics for information processing, and have therefore been recognised as ideal systems for reservoir computing. A distinctive feature of delay-line reservoirs is their single-input/single-output structure, which makes them efficient for physical implementation. However, this also presents a significant limitation to multi-input tasks, as the sequence of information in the time-multiplexed input stream is not obvious. Here, we propose enhancing the input masking process used in delay-feedback RCs to mix multiple inputs in the time domain. We investigate two approaches: ‘interleaved’ and ‘sequential’, of injecting multi-input signals into a delay-line reservoir without modifying its topology. Further, we propose a novel task for RC, which inherently requires multiple inputs, to evaluate our approach: the control of a forced Van der Pol oscillator system. We use the trained reservoir as a controller to regulate the nonlinear dynamics of the Van der Pol system by constraining its trajectory to a circle. We find that, with careful choice of model parameters and offset masking scheme, the ‘sequential’ method outperforms the ‘interleaved’ method on this task.

Accessible generative artificial intelligence (AI) tools like large-language models (LLMs) (e.g., ChatGPT,1 Minerva2) are raising a flurry of questions about the potential and implications of generative algorithms and the ethical use of AI-generated text in a variety of contexts, including open science (Bugbee & Ramachandran, 2023), student assessment (Heidt, 2023), and medicine (Harrer, 2023). Similarly, among the graphic and visual arts communities, the use of generative image synthesis algorithms (e.g., DALL-E,3 Midjourney,4 Stable Diffusion5) that take text prompts as input and produce works in the style of a particular human artist, or no artist who ever lived, are causing consternation and posing challenging questions (Murphy, 2022; Plunkett, 2022). The use of generative AI to create deep fakes has also been in the spotlight (Ruiter, 2021), as has its role in answering scientific research questions directly (Castelvecchi, 2023).To our minds, the questions these technologies are raising do not seem to be of a fundamentally different character to questions asked about AI for many years. They largely concern (a) what is possible, (b) what is right, and (c) the implications of the technology’s use. For instance,Can AI generate documentary “evidence” that is indistinguishable from reality? Can AI generate artifacts that are competitive with (or superior to) those made by a human?How is the concept of “truth” confused or undermined by the output of these technologies? Is it ethical to load examples of a human’s art to generate a model for replicating their style?Who owns the intellectual property rights of AI-generated artifacts? Who is accountable if an AI-generated artifact causes harm? Will I lose my job to an AI; does AI make creative writers or artists redundant?The sense of urgency and public interest in these questions has increased as generative models become practical and ubiquitous tools with levels of skill or expertise that appear to be of above average human competency.Of general relevance to academic publishers, including Artificial Life, is that LLMs can compose text structured in the form of research paper sections or even whole articles. And these texts are often (currently) hard to distinguish from human-written text, even by software trained to detect text generated by language models (LMs). When questioned (on February 20, 2023) “What should a journal editor do about LMs?” You.com’s chatbot suggested,A journal editor should ensure that all submitted manuscripts adhere to the journal’s style guide, and that the language used is appropriate for the journal’s audience. They should also ensure that any language models (LMs) used to assist with the writing process are properly cited and that any generated text is labeled as such. Additionally, the journal editor should ensure that the LMs are used in accordance with the journal’s policies and ethical considerations.That’s a generic response that begins tangentially to the questioner’s intended focus. As any human reader can immediately see, the chatbot is not aware of the social context of the question, nor has it had the opportunity to gauge the questioner’s level of expertise. But the text is “good enough”—it does address relevant issues, and it’s an answer that probably took you a few seconds to read, interpret, and critique. It may take you more time to read and analyze the text and this paragraph than it took for us and the chatbot to generate it. Was this therefore a waste of your time, or ours? Is the chatbot wasting your time? Are we? Some journals and publishers have drafted formal policies that require the use of LLMs for writing submissions to be explicitly acknowledged (e.g., at Springer-Nature; “Tools Such as ChatGPT,” 2023). In practice, their use (or misuse) may be very difficult to detect.Artificial Life, and its publisher MIT Press generally, is also adopting the policy that any use of generative AI, for any part of a submitted work, including but not limited to text, images, sound, data, mathematics, logic, reasoning, programming code, or algorithms, must be prominently, explicitly, and unambiguously labeled and its source formally cited (e.g., via a name, manufacturer, URL, version number, or access date).Journals and publishers have also moved to prevent LLMs from being listed as authors on articles. For instance, Springer-Nature’s policy was online earlier this year, and, although it has now seemingly been removed from its original location, variants of it have been incorporated into the authorship policies of some journals:Large Language Models (LLMs), such as ChatGPT, do not currently satisfy our authorship criteria. Notably an attribution of authorship carries with it accountability for the work, which cannot be effectively applied to LLMs. Use of an LLM should be properly documented in the Methods section (and if a Methods section is not available, in a suitable alternative part) of the manuscript. (Nature, 2023)In a policy that remains online at time of writing, the journal Science states,Text generated from AI, machine learning, or similar algorithmic tools cannot be used in papers published in Science journals, nor can the accompanying figures, images, or graphics be the products of such tools, without explicit permission from the editors. In addition, an AI program cannot be an author of a Science journal paper. A violation of this policy constitutes scientific misconduct. (Science, 2023)Artificial Life and MIT Press are taking an approach in alignment with those of the editorial boards (and publishing house legal teams) of such journals: that authorship is associated with responsibility and accountability for an article. However, for Artificial Life, the issue doesn’t stop there.The implications of generative AI are relevant to a broad spectrum of society. But an interest in generative computational processes is arguably at the center of Artificial Life research. How might LLMs be specifically relevant to Artificial Life, as opposed to the subdiscipline (yes, that’s ironic) of AI? Here are a few ideas.The production of novelty can be explored through the use of an LM that continually takes as its input text composed by humans, other LMs, and its own output. Such a system is relevant to our field’s interests in feedback loops, open-endedness, and the emergence of complexity. Is this system engaged in language acquisition through “social” interactions?LMs might be used to explore questions related to the emergence of meaning in language. Can meaning be generated by an LM, or is it specific to living things? Can LMs evolve to be better interpreters and writers? How does the text LMs generate change the way humans produce and use language?If any work responding to these questions was presented in the form of a formal research experiment documented in an article, then this would naturally fall within the scope of human-authored research. However, an LLM-generated poem, song, or essay can be of value to researchers in Artificial Life exploring these topics (even if it isn’t very good; Cave, 2023). The coauthorship of such a work by an LM and a human as a way of communicating ideas about Artificial Life would be interesting to consider. In this case, the text contributed by the LM would need to be quoted as an “example” within the text of a submission made by a human author who determined that it was worthy of submission. Even though the text itself is a direct, self-referential, and, we would hope, revealing exploration of an LM system’s quirks, capabilities, or limitations, the determination of its relevance must, for the time being at least, be made by and attributed to a human.There is a precedent for such work, much of which has been explored under the banners of cybernetic, generative, and Artificial Life art (e.g., see many historical examples in Benthall, 1972; Ohlenschläger, 2012; Reichardt, 1968; Whitelaw, 2004). In such contexts, the art is published. Associated commentary and/or explanations may come later, and these might not be authored by the same system or person who made the original work.An interview, debate, discussion, or duet between a human and an AI, or between several computer programs, can also challenge our ideas about living systems and their exchange of information, the use of language, or the production of improvised movement and sound. The inclusion of extracts from discussions with computer chatbots dates back at least to the advent of ELIZA in the 1960s: “Men are all alike—IN WHAT WAY—They’re always bugging us about something or other” (Weizenbaum, 1966, p. 36). Likewise collaborative improvisations performed by robots, algorithms, and humans have an established place in music (Bown, 2011; Eldridge, 2005). As far as we know, the journal hasn’t published such works previously. But we could.For something along these lines to be published today, as with any contribution, it would of course need to provide novel perspective or insight. However, the main point here is that in these scenarios, even though we might intuitively feel that the generative AI system warrants the status of contributor at the level of coauthor, we have to insist on a human author of the submission. They would have ultimate responsibility for the work produced by the generative algorithm so that, for instance, if the LLM’s poetry influenced thought in a positive and productive way, or if it incited violence, we would have somebody accountable to thank or blame. If the article’s publication required the payment of an open access fee, we would also have somebody from whom to extract the payment!We haven’t yet received any submissions made by generative AI (as far as we know). But this issue contains novel work by human authors. In fact, we have recently published a spate of varied special issues reporting on the research presented at human gatherings, some in person, some online. These have covered a wide range of exciting activity in the Artificial Life community: Issue 28:2 has extended versions of selected papers from the 2019 Artificial Life conference; 28:3 is a collection of articles on embodied intelligence; 28:4 is the Artificial Life 2021 conference special issue; 29:1 explores agent-based models of human behavior. We extend our thanks to all the guest editors for their hard work in handling the selection and review of articles for their issues. New ideas for special issues in the subdomains of Artificial Life are always welcome. If you have an idea, please contact us.After that run of special issues, we welcome you back to a general issue of contributed research articles that cover a wide range of Artificial Life topics—including distributed control, emergence, dynamical systems, self-organization, game theory, artificial chemistry, and biocomputing—addressed through theory, models, simulations, and physical experiments.We start with a letter from Bull and Liu, on “A Generalised Dropout Mechanism for Distributed Systems.” They use a modified NK model to sharpen the criteria for determining when local control is more beneficial than global control. Next, we have an article from Gershenson, on “Emergence in Artificial Life.” He uses the difference in information present at different levels of a system as the basis for a new definition of emergence, one of the fundamental components of ALife.The article from Howison et al., “On the Stability and Behavioral Diversity of Single and Collective Bernoulli Balls,” describes a platform for investigating how dynamical systems may be used as the basis for designing a variety of agent behaviors. This platform, both in simulation and as a physical system, comprises a collection of “Bernoulli balls” in an airflow, interacting with each other and with the flow. The aim is to develop a dynamical system with a diverse set of possible behaviors.Ichinose et al. present “How Lévy Flights Triggered by the Presence of Defectors Affect Evolution of Cooperation in Spatial Games.” Lévy flights model a kind of random motion with both small and big displacements. Here Lévy flights are combined with game theory concepts in an agent-based model. The authors investigate how the presence of defectors changes the optimal behaviors.Next, Scott and Pitt investigate “Interdependent Self-Organizing Mechanisms for Cooperative Survival.” Complex survival games, where cooperation is needed to survive intermittent catastrophes, need complex strategies. Here the authors look at social self-organization, which, as any complex domain, has aspects that can make the situation better in some cases and, in other cases, worse. They conclude that such systems need to be able to reflect on their own operation through some kind of self-model.Sienkiewicz and Jędruch tell us about “DigiHive: Artificial Chemistry Environment for Modeling of Self-Organization Phenomena.” This two-dimensional continuous space simulation environment supports experiments with the goal of facilitating open-ended simulations. It steers more toward natural physical and biological systems (e.g., it includes energy conservation), rather than toward the more abstract operation of some other artificial chemistries. The authors describe the rationale and operation of the system and use it to investigate aspects of self-organization and self-replication in cellular-like systems.Finally, Svahn and Prokopenko examine “An Ansatz for Computational Undecidability in RNA Automata.” An ansatz is an “educated guess” about the form of the solution to a problem that can be used to provide a stepping-stone to finding the solution. Here the approach uses the known computational power of a set of automaton models as the form of solution and shows how RNA behaviors map to these models to demonstrate the computational power of this biological form of computing.

We start this issue with a research article by Keith L. Downing, “The Evolution of Conformity, Malleability, and Influence in Simulated Online Agents.” The author uses an agent-based model to investigate the potential of positive feedback in personalized recommender systems for driving a population to polarization and conformity.Following this, we include a series of articles forming a special issue from the LIFELIKE Computing Systems Workshop, held at the Artificial Life conference in Prague/Online, 2021. The workshop was co-organized by Peter Lewis, Anthony Stein, Jean Botev, and Sven Tomforde. Lewis serves here as guest editor of the special issue.We close this brief editorial to issue 29:4 by noting that our next issue will be the first of volume 30. We will be celebrating 30 years of Artificial Life by hosting several special anniversary articles giving personal perspectives on the history, development, and future of our subject. We are looking forward to publishing an amazing set of pieces.

Abstract Deep learning methods are recognised as state-of-the-art for many applications of machine learning. Recently, deep learning methods have emerged as a solution to the task of automatic music generation (AMG) using symbolic tokens in a target style, but their superiority over non-deep learning methods has not been demonstrated. Here, we conduct a listening study to comparatively evaluate several music generation systems along six musical dimensions: stylistic success, aesthetic pleasure, repetition or self-reference, melody, harmony, and rhythm. A range of models, both deep learning algorithms and other methods, are used to generate 30-s excerpts in the style of Classical string quartets and classical piano improvisations. Fifty participants with relatively high musical knowledge rate unlabelled samples of computer-generated and human-composed excerpts for the six musical dimensions. We use non-parametric Bayesian hypothesis testing to interpret the results, allowing the possibility of finding meaningfulnon-differences between systems’ performance. We find that the strongest deep learning method, a reimplemented version of Music Transformer, has equivalent performance to a non-deep learning method, MAIA Markov, demonstrating that to date, deep learning does not outperform other methods for AMG. We also find there still remains a significant gap between any algorithmic method and human-composed excerpts.

In this issue we are pleased to share with you a diverse set of reading materials. Sadly, we mark with an obituary the passing of Julian Miller, a researcher whose loss has been keenly felt within the community of Artificial Life researchers. He shall be sorely missed.On a much brighter note, the second installment of Chris Adami’s column exploring how artificial evolution might facilitate the design of General Intelligence is to be found within the pages of this issue. Adami explains how the indirect encoding of artificial brains to facilitate neuro-evolution might be managed. He discusses approaches to choosing an appropriate neuron, how to connect neurons to create a functioning network, how to train the network, and how the different options scale up to high levels of complexity. Drawing such connections between the techniques of Artificial Life and the concerns of Artificial Intelligence is key (we feel) to enhancing the recognition that embodiment, developmental processes, and evolutionary processes all have a role to play in the emergence of natural intelligence – to overlook this whilst striving for artificial general intelligence is likely problematic.Simon Penny, an artist long engaged in Artificial Life art and robotics, provides for us a critical review of a new book by Sofian Audry, Art in the Age of Machine Learning (MIT Press 2021). The title might seem to be slightly out of line with Artificial Life’s main focus, perhaps even more suited to an AI readership, but, as Penny points out, this isn’t necessarily the case. In fact, by presenting both the practical artistic-technological concerns of the day, and the philosophical issues these raise with respect to agency, creativity and art-making by machines, Audry is in fact delving into areas that should concern us as researchers of Artificial Life.A topic infrequently explored within the pages of this journal is the impact that Artificial Life has on human relationships. In Uncanny Beauty: Aesthetics of Companionship, Love, and Sex Robots, Paolo Euron enters this space by examining “physical beauty according to the artistic, cultural, and philosophical traditions”, of sexbots. Since Euron focuses on the visual appearance of these humanoid robots, with this article we have adopted a new approach for the Artificial Life journal to widen the perspective. The text is therefore supported by commentaries the editors have sought from alternative points of view. Thomas Arnold provides comment on Euron’s work from the perspective of Human-Robot Interaction by assessing the ethics of sex robots and how concepts of human trust, dignity, and autonomy potentially influence our interactions with such machines. Maria O’Sullivan examines how human interactions with sexbots relate to gender power relations and our expectations and human norms of intimacy and vulnerability. She also considers the very real dangers now widely associated with the commodification of beauty and the potential for moral harm that may result from an increase in the ubiquity or use of sexbots. We hope that you find the article and commentaries thought provoking.The issue also includes five other intriguing research articles.In Computation by Convective Logic Gates and Thermal Communication, Bartlett et al. demonstrate a simulation of an embodied computational system of Boolean NOR gates that can be realised in a convective fluid. This demonstrates that computation can be achieved in relatively simple physical contexts.The article Morphological Development at the Evolutionary Timescale: Robotic Developmental Evolution, by Benureau and Tani, flips the usual biological timescales (slow evolution, faster developmental processes) to one where morphological development is slower than evolution, and applies this in the context of robot design. Early in the evolutionary process, robots are ‘babies’, and only later on in evolution have they developed into ‘adults’. The authors investigate this novel approach to evolve diverse gaits.In Effect of Environmental Change Distribution on Artificial Life Simulations, Bullinaria explores the effect of environmental change (modelled as coloured noise) on evolutionary outcomes. He carefully dissects two factors: the average size, and the distribution of changes, to show that results are not as sensitive to noise colour as previously thought.Monte Carlo Physarum Machine: Characteristics of Pattern Formation in Continuous Stochastic Transport Networks, authored by Elek et al., takes inspiration from slime mould growth to develop an algorithm for reconstructing networks from sparse data. The authors demonstrate their algorithm by applying it to a case study rarely seen in the Artificial Life arena: cosmological data sets of the large-scale distribution of gas and dark matter in the universe.In From Dynamics to Novelty: An Agent-Based Model of the Economic System, Recio et al. explicitly include time evolution in a new model of a complex self-organising economy. They realise their model in an agent-based simulation of an economy that can generate new technologies and products, and investigate the dynamics of the resulting system.We hope you enjoy the issue!

Julian Francis Miller1It is with great sadness that we report the death of our colleague and friend, Julian Miller.Julian’s work is well known throughout the Artificial Life community: His Cartesian genetic programming (CGP) and in materio computing are foundational concepts. He also made contributions in morphological computing and neurocomputing, all based on his fascination with evolution as a means of attacking and solving problems. Like many in the ALife community, he had an interdisciplinary career, commencing with a first degree in Physics and a PhD in Mathematics, followed by research in Natural Computing and material computing at the universities of Napier, Birmingham, and York in the UK.Julian invented CGP (Miller, 1999), a way of encoding graph programs (functional nodes connected by edges) in a string of integers, allowing the string to be evolved in the standard way, with the graph (located on a Cartesian grid, hence its name) produced as the result of a genotype to phenotype mapping. From this simple beginning, Julian and his students continued to develop the approach, and other researchers joined in. Ten years later, the field had grown significantly, with many researchers both using CGP in their own work and extending the original concept. Indeed, the field had grown enough that Julian could edit an entire book on the topic (Miller, 2011). Ten years later still, the field shows no signs of abating, and Julian wrote a 40-page review for Genetic Programming and Evolvable Machines on CGP’s status, its many variants, and its future prospects (Miller, 2020).Julian was also a pioneer in the field of in materio computing (Miller & Downing, 2002), which exploits the physical properties of unconventional materials, such as liquid crystals (Harding & Miller, 2004) and carbon nanotubes (Miller et al., 2014), to perform computation intrinsically, in what he dubbed a “Field Programmable Matter Array.” His original work used evolutionary algorithms directly to configure the materials. Later, he also used Reservoir Computing as a more abstract model for getting these materials to compute (Dale et al., 2017). This is another field with explosive growth, so much so that some authors have even published on the name of the domain itself (Ricciardi & Milano, 2022). Julian was there from the start, contributing his insights and ideas throughout.CGP and in materio computing may be what Julian is best known for, but these contributions were embedded in a deeper research program of understanding development as a fundamental component of evolving embodied computation. From growing a self-repairing “French-flag” organism (Miller, 2004), to assembling complex structures through Artificial Chemistries (Faulconbridge et al., 2011), to the idea of the “software garden” (Miller, 2018), Julian felt that both growth and evolution are essential concepts in complex systems.His Festschrift Inspired by Nature (Stepney & Adamatzky, 2018) was a (slightly late) 60th birthday present from his many academic colleagues. It includes chapters contributed by a wide range of authors who have built on and been inspired by his many research interests. The text covers evolution and hardware, CGP applications, chemistry, and development. Julian retired in 2016, but he did not stop his research. He used the freedom from the quotidian constraints of an academic job to pursue a new interest. He was bringing together his discoveries in evolution, development, networks, and computation to develop a new neural model to evolve programs that build, or grow, neural networks. His most recent publication on that topic has only just appeared (Miller, 2022).Julian’s retirement also allowed him to spend time with his recently acquired beloved new family. His wife Gabi remembers him thus: “He was a loving and generous-hearted husband, a wise step-father to my three grown adults and much loved Grandpa to our four grand-children. Jules will be sadly missed, but also lovingly remembered by all whose life he touched.” Many further tributes to Julian from his colleagues can be found in the latest SIGEVO newsletter (Ochoa, 2022).

November 02 2021 Editorial: News from the New Co-Editors in Chief Alan Dorin, Alan Dorin Monash University, Computational and Collective Intelligence Group, Department of Data Science and AI, Faculty of Information Technology. [email protected] Search for other works by this author on: This Site Google Scholar Susan Stepney Susan Stepney University of York, Department of Computer Science. [email protected] Search for other works by this author on: This Site Google Scholar Author and Article Information Alan Dorin Monash University, Computational and Collective Intelligence Group, Department of Data Science and AI, Faculty of Information Technology. [email protected] Susan Stepney University of York, Department of Computer Science. [email protected] Online Issn: 1530-9185 Print Issn: 1064-5462 © 2021 Massachusetts Institute of Technology2021Massachusetts Institute of Technology Artificial Life (2021) 27 (2): 73–74. https://doi.org/10.1162/artl_e_00350 Cite Icon Cite Permissions Share Icon Share Twitter LinkedIn Views Icon Views Article contents Figures & tables Video Audio Supplementary Data Search Site Citation Alan Dorin, Susan Stepney; Editorial: News from the New Co-Editors in Chief. Artif Life 2021; 27 (2): 73–74. doi: https://doi.org/10.1162/artl_e_00350 Download citation file: Ris (Zotero) Reference Manager EasyBib Bookends Mendeley Papers EndNote RefWorks BibTex toolbar search Search nav search search input Search input auto suggest search filter All ContentAll JournalsArtificial Life Search Advanced Search It has been almost 20 years, and 20 volumes, since Mark Bedau took on the role of Editor in Chief of Artificial Life. After diligently maintaining the journal day-to-day and steering it strategically over the long term, Mark has decided to retire from this role. He was instrumental in cementing the Artificial Life journal as a dependable and valuable publication venue for research in our field. But also, during his lengthy term, Mark was willing to trust would-be guest editors with the independence to manage their own special issues. He remained consistently keen to innovate and to explore new directions for the journal that capitalised on our field’s inherent multidisciplinarity, curiosity, and diversity. Mark also led ISAL (International Society for Artificial Life) during the years 2001–2015. In the mid-2000s, he was also Chief Operating Officer of ProtoLife SRL... © 2021 Massachusetts Institute of Technology2021Massachusetts Institute of Technology You do not currently have access to this content.

Abstract We introduce the concept of Semantic Neutral Drift (SND) for genetic programming (GP), where we exploit equivalence laws to design semantics preserving mutations guaranteed to preserve individuals’ fitness scores. A number of digital circuit benchmark problems have been implemented with rule-based graph programs and empirically evaluated, demonstrating quantitative improvements in evolutionary performance. Analysis reveals that the benefits of the designed SND reside in more complex processes than simple growth of individuals, and that there are circumstances where it is beneficial to choose otherwise detrimental parameters for a GP system if that facilitates the inclusion of SND.

Abstract We explore the effect of connectivity and topology on the dynamical behaviour of Reservoir Computers. At present, considerable effort is taken to design and hand-craft physical reservoir computers. Both structure and physical complexity are often pivotal to task performance, however, assessing their overall importance is challenging. Using a recently developed framework, we evaluate and compare the dynamical freedom (referring to quality) of neural network structures, as an analogy for physical systems. The results quantify how structure affects the behavioural range of networks. It demonstrates how high quality reached by more complex structures is often also achievable in simpler structures with greater network size. Alternatively, quality is often improved in smaller networks by adding greater connection complexity. This work demonstrates the benefits of using dynamical behaviour to assess the quality of computing substrates, rather than evaluation through benchmark tasks that often provide a narrow and biased insight into the computing quality of physical systems.

Abstract We have developed abstraction/representation (AR) theory to answer the question “When does a physical system compute?” AR theory requires the existence of a representational entity (RE), but the vanilla theory does not explicitly include the RE in its definition of physical computing. Here we extend the theory by showing how the RE forms a linked complementary model to the physical computing model. We show that the RE does not need to be a human brain, by demonstrating its use in the case of intrinsic computing in a non-human RE: a bacterium.

We introduce MetaChem, a language for representing and implementing artificial chemistries. We motivate the need for modularization and standardization in representation of artificial chemistries. We describe a mathematical formalism for Static Graph MetaChem, a static-graph-based system. MetaChem supports different levels of description, and has a formal description; we illustrate these using StringCatChem, a toy artificial chemistry. We describe two existing artificial chemistries-Jordan Algebra AChem and Swarm Chemistry-in MetaChem, and demonstrate how they can be combined in several different configurations by using a MetaChem environmental link. MetaChem provides a route to standardization, reuse, and composition of artificial chemistries and their tools.

We introduce a form of neutral Horizontal Gene Transfer (HGT) to Evolving Graphs by Graph Programming (EGGP). We introduce the µ × λ evolutionary algorithm, where µ parents each produce λ children who compete with only their parents. HGT events then copy the entire active component of one surviving parent into the inactive component of another parent, exchanging genetic information without reproduction. Experimental results from 14 symbolic regression benchmark problems show that the introduction of the µ × λ EA and HGT events improve the performance of EGGP. Comparisons with Genetic Programming and Cartesian Genetic Programming strongly favour our proposed approach.

The EvoEvo project was a 2013–2017 FP7 European project aiming at developing new evolutionary approaches in information science and producing novel algorithms based on the current understanding of molecular and evolutionary biology, with the ultimate goals of addressing open-ended problems in which the specifications are either unknown or too complicated to express, and of producing software able to operate even in unpredictable, varying conditions. Here we present the main rationals of the EvoEvo project and propose a set of design rules to evolve adaptive software systems. These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

The Reservoir Computing (RC) framework is said to have the potential to transfer onto any input-driven dynamical system, provided two properties are present: (i) a fading memory, and (ii) input separability. A typical reservoir consists of a fixed network of recurrently connected processing units; however recent hardware implementations have shown reservoirs are not ultimately bound by this architecture. Previously, we have demonstrated how the RC framework can be applied to randomly-formed carbon nanotube composites to solve computational tasks. Here, we apply the RC framework to an evolvable substrate and compare performance to an already established in materia training technique, referred to as evolution in materia. The results show that by adding the programmable reservoir layer, reservoir computing in materia can significantly outperform the original evolution in materia implementation. This suggests the RC framework offers improved performance, even across non-temporal tasks, when combined with the evolution in materia technique.

An enormous quantity may be termed “astronomical,” referencing the huge span of time since the Big Bang (∼1017 s), the huge size of the universe (∼1027 m), or the huge amount of material in the observable universe (∼1080 atoms). Yet these quantities pale into insignificance compared to those generated by combinatorics, where numbers are combined using multiplication and exponentiation, leading to an explosion in their size. The number of possible proteins of the typical length of eukaryotic proteins is 20400 ∼ 10520 (although not all of these would have a sensible shape or function); the number of possible memory configurations of a mere 1 kB of RAM is 28×210∼ 10103; the number of books in Borges' Library of Babel is more than 10106 (yet hardly any are interesting books), they can be shelved in ∼10106 to the power 10106 ways, and even the library's catalogue is huge; and so on.Daniel Dennett, in his book Darwin's Dangerous Idea, uses a clever trick to remind us of the sheer scales involved. He builds up an intuition, or possibly more of a feeling, of such sizes, then dubs these “Vast,” with a capital V. Ever after, the term Vast evokes that sheer scale.Within the Vastness of all possibilities, only a subset is somehow interesting: Most is mere noise. This subset may be Vast in its own right, yet vanishingly small relative to the Vastness of all possibilities. How to find such vanishingly small needles in the Vastness of a combinatoric haystack?One technique might be dubbed “search and construct.” Search for a useful set of atoms, primitives, components, that form the basis of the Vast combinatorial space. Then use rules and processes to define or generate only those constructs with interesting structure and behavior within that space. For the Library of Babel, the atoms are characters, and the Vastness is all possible books of these characters. But what rules delimit the subspace of interesting books, books that are grammatical, readable, and worthwhile? There is chunking to form higher-level components: words. There are syntactic restrictions on the form of sentences, and further semantic restrictions to be meaningful. But to go further, to construct the subset that is literature, say, requires as yet uncodified human creativity. For computer programming, constructing a member of the interesting subset is a slightly easier task. The primitives are the relevant high-level language constructs and their syntactic constraints, the rules include well-formedness constraints and patterns, yet there is still much creativity needed to construct useful programs.Many researchers turn to the natural world for inspiration. Evolution is one process that explores these interesting possibilities. It can be considered part of a process that searches for genomes, then constructs phenotypes. Interestingness here is viability. A range of artificial evolutionary algorithms take inspiration from these natural processes. In nature, the starting point for evolution is already something quite complex: An organism, even a single-celled organism, is nontrivial, not a random collection of molecules. Can we find a mechanism for generating this initial complexity?Underlying life is chemistry. Chemistry is combinatorics par excellence. From a small set of atoms, chemical bonding laws produce a Vast set of molecules with structure and behavior. It has chunking: Atoms can form small molecular building blocks, such as DNA bases and amino acids, that are themselves the components in higher-level constructions. Good blocks can be searched and selected for by evolution. As we have seen from the protein example above, the larger molecules produced are still a vanishingly small subset of the potential Vastness. Not all combinations of atoms can form stable molecules, and not all molecules that can form have a function or structure that can contribute to further construction.Artificial chemistry (AChem) takes such ideas from natural chemistry, in order to generate and explore a variety of forms of combinatoric Vastness in silico. If we think of AChems as a generic form of “search and construct” processes, and as rule-based novelty generators, we can see that they can be applied not simply to chemical problems, but to a whole range of domains where such processes are needed and used, including computing, dynamical systems, language and music, and modeling in silico and in vitro complex systems.An AChem provides three components for virtual world explorations. First, there is the material, the virtual atoms and molecules, that provides the Vast combinatorial space of potential structures. Then there are the reaction rules, the analogs of the laws of nature in our virtual world, which define how the material combines and dissociates, and possibly even how the space it occupies is restructured (as with P-systems). These rules implicitly define a subspace of possible structures in that Vastness. Finally, there is the algorithm, which lays out our explicit experimental setup to explore that implicit subspace, anywhere from exhaustive search to pouring some virtual stuff in a virtual bucket and watching what happens.Nature provides just the one particular kind of material (real-world atoms and molecules) and one set of rules (chemical bonding and reactions) that say which molecules are possible, and which are not. The only freedom the scientist has is in the algorithm: the experimental setup that controls which molecules encounter which others, under what environmental conditions. Despite its real-world constraints, chemistry provides all the richness and complexity sufficient for life itself.The playpen of AChem is even richer, since we also have the freedom to choose different basic material, and different rules. Yet it has the corresponding downside in that we now have to implement the rules of our virtual world.This new book forms a comprehensive introduction to many different facets of the discipline of AChem. The plurality in its title, Artificial Chemistries, indicates the diversity of approaches covered. It covers the why, how, and what of the choices of material, rule, and algorithm, and their consequences. For the beginning student, it provides a wide-ranging review of the subject, and its 1000-item bibliography is a marvelous resource in its own right, providing entry into the relevant scientific literature. For the practicing AChemist, it provides an invaluable reference material on all topics in the discipline.Despite its comprehensive nature, this book is no mere annotated bibliography: Its structure provides a narrative unity for the discipline. Part I comprises four foundational chapters, laying out the philosophy and scope of the subject, illustrated with some simple example AChems. It includes a primer on basic concepts from chemistry, such as chemical reactions, the law of mass action, equilibrium, chemical bonds, and catalysis. It also covers differential equation modeling and computational techniques.Part II comprises four chapters covering the natural-world inspiration. It starts with the chemistry of life: that of biochemistry and large organic molecules, including proteins, RNA, and DNA. The level of detail is useful for showing the underlying complexity and richness of the chemical processes that are frequently abstracted as mere string concatenation. It would probably do students good to review this material again once they have designed their initial AChem, to help them appreciate the simplifications they have made. The next chapter discusses simple cells, including their structure with lipid walls, and their dynamics in terms of metabolism. It includes discussion of autopoeisis, Robert Rosen's ideas on organization in living systems, origin-of-life theories, and more. All this is necessarily brief, as each topic has deservedly book-length treatment elsewhere, and so things can get quite dense in places: The Rosen section in particular will probably be incomprehensible to anyone who has not already encountered the material. But the bibliography will guide the curious reader to further explanations. Next come chapters on evolution and open-ended systems. Open-endedness is the holy grail of AChems: Not only can they explore a Vast configuration space, they may be able to grow this very space by opening up new possibilities and dimensions through their own contingent development. These chapters contain a mix of fairly standard material given added value by being filtered through an AChem perspective—for example, evolutionary dynamics is discussed in terms of chemical reactions—and some quite deep and provocative concepts.Part III comprises three chapters of massive literature review, documenting AChems and categorizing them into rewriting systems, automata, and bio-inspired. In rewriting systems the reaction rules state how a particular string or other representation is systematically changed into a new form; these include lambda calculi, P-systems, L-systems, and the like. Automata AChems comprise molecules whose atoms are assembly-language-level computational instructions: Molecular behavior is given by the execution of these fragments. These include specific systems such as Tierra and Avida, as well as more generic systems such as cellular automata, von Neumann constructors, and all the way up to Turing machines. The bio-inspired AChems hold more closely to biological mechanisms, such as enzyme reactions, RNA binding, shape-based lock-and-key binding, genetic networks, and swarms. These chapters demonstrate a strength and weakness of AChems: the ability to build yet another arbitrary complex system. Some of these AChems have been examined in detail over a long period of time by research groups; others exist in only an article or two resulting from a single doctoral student project. These chapters can be used as a reference to find specific AChems, or as source material for developing new AChems, hopefully as a synthesis and unification of existing ones. Their comprehensive nature can be a problem on occasion: A whole algorithm may be covered in a single spare sentence. Yet the vast bibliography leads on to more detail.Part IV comprises four chapters focusing on the global dynamics of general AChems. While parts II and III will be best for students, this part will be of most value to more experienced researchers. First is a chapter on organization theory, written with Pietro Speroni di Fenizio. This looks at conditions for and properties of closed sets of molecules: sets where each molecule is produced by members of the set, and so the reaction network is closed. The following chapter discusses the dynamics of such organizations: effects of reaction rates and probabilities on their construction and maintenance. Next comes a chapter dealing with what for me is the raison d'être of AChems: emergence. It provides a discussion of relevant topics: self-organization, nonequilibrium thermodynamics, chaos, downward causation, all as they are relevant to AChems. Several deep and important concepts are each outlined in half a page, and the chapter covers a stunning range of topics. The final chapter in this part continues the theme of emergence by discussing constructive dynamical systems: how AChems can produce novelty.Part V comprises five chapters on applications of AChems to a wide range of domains. Here we get discussion of everything from robotics to unconventional computation, from nuclear physics to economics, from modeling biological systems to synthetic biology.The book also includes an appendix giving details of the PyCell AChem package, which provides an immediate entry to computational AChems.This book is really three or more significant books rolled into one, as needed to cover the breadth of the subject. There are interdisciplinary issues here: A practitioner needs to know a lot about a wide range of subjects. As such, it is a remarkable work of scholarship, bringing together a whole host of diverse information, and synthesizing it into a coherent and valuable account of the discipline of artificial chemistry. I learned a lot from reading it—not just the material that was new to me, but also new ways of looking at known material, and the valuable syntheses of a wide range of concepts. The authors should be commended for their impressive contribution to the field. Any AChemist, ALifer, or, more generally, nature-inspired computer scientist or engineer will find Artificial Chemistries a valuable addition to their research bookshelf.

The capacity of pluripotent embryonic stem cells to differentiate into any cell type in the body makes them invaluable in the field of regenerative medicine. However, because of the complexity of both the core pluripotency network and the process of cell fate computation it is not yet possible to control the fate of stem cells. We present a theoretical model of stem cell fate computation that is based on Halley and Winkler's Branching Process Theory (BPT) and on Greaves et al.'s agent-based computer simulation derived from that theoretical model. BPT abstracts the complex production and action of a Transcription Factor (TF) into a single critical branching process that may dissipate, maintain, or become supercritical. Here we take the single TF model and extend it to multiple interacting TFs, and build an agent-based simulation of multiple TFs to investigate the dynamics of such coupled systems. We have developed the simulation and the theoretical model together, in an iterative manner, with the aim of obtaining a deeper understanding of stem cell fate computation, in order to influence experimental efforts, which may in turn influence the outcome of cellular differentiation. The model used is an example of self-organization and could be more widely applicable to the modelling of other complex systems. The simulation based on this model, though currently limited in scope in terms of the biology it represents, supports the utility of the Halley and Winkler branching process model in describing the behaviour of stem cell gene regulatory networks. Our simulation demonstrates three key features: (i) the existence of a critical value of the branching process parameter, dependent on the details of the cistrome in question; (ii) the ability of an active cistrome to "ignite" an otherwise fully dissipated cistrome, and drive it to criticality; (iii) how coupling cistromes together can reduce their critical branching parameter values needed to drive them to criticality.

This paper describes the artificial epigenetic network, a recurrent connectionist architecture that is able to dynamically modify its topology in order to automatically decompose and solve dynamical problems. The approach is motivated by the behavior of gene regulatory networks, particularly the epigenetic process of chromatin remodeling that leads to topological change and which underlies the differentiation of cells within complex biological organisms. We expected this approach to be useful in situations where there is a need to switch between different dynamical behaviors, and do so in a sensitive and robust manner in the absence of a priori information about problem structure. This hypothesis was tested using a series of dynamical control tasks, each requiring solutions that could express different dynamical behaviors at different stages within the task. In each case, the addition of topological self-modification was shown to improve the performance and robustness of controllers. We believe this is due to the ability of topological changes to stabilize attractors, promoting stability within a dynamical regime while allowing rapid switching between different regimes. Post hoc analysis of the controllers also demonstrated how the partitioning of the networks could provide new insights into problem structure.

Computational reflection uses software architectures that are capable of self- modification at runtime. These systems have implementations between two extremes: procedural reflection, in which unlimited self-modification is available at the expense of infinite recursion; and declarative reflection, which uses pre-defined metrics to drive the self-modification and is hence limited in scope. Biological processes also exploit the concept of reflection, where natural selection drives the process of modification. The concept of a program in computing has an analogy with an individual member of a species. The process of life is discretised into a series of autonomous systems, each of which creates modified versions of itself as offspring. This paper unifies the concept of computational reflection with biological systems via a new analysis of von Neumanns Universal Constructor. The result is a bio-reflective architecture that is capable of unconstrained self-modification without the problems of infinite recursion that exist in the computational counterparts. The new architecture is a blueprint for applications in Artificial Life studies, Evolutionary Algorithms, and Artificial Intelligence.

We present a subsymbolic Artificial Chemistry (ssAChem) in which all properties relevant to bonding are emergent from the underlying dynamical system (an RBN). We explore this ssAChem by evolving a seed set of atomic particles and showing the type of composite particles the system can produce.

Automata chemistries are good vehicles for experimentation in open-ended evolution, but they are by necessity complex systems whose low-level properties require careful design. To aid the process of designing automata chemistries, we develop an abstract model that classifies the features of a chemistry from a physical (bottom up) perspective and from a biological (top down) perspective. There are two levels: things that can evolve, and things that cannot. We equate the evolving level with biology and the non-evolving level with physics. We design our initial organisms in the biology, so they can evolve. We design the physics to facilitate evolvable biologies. This architecture leads to a set of design principles that should be observed when creating an instantiation of the architecture. These principles are Everything Evolves, Everything's Soft, and Everything Dies. To evaluate these ideas, we present experiments in the recently developed Stringmol automata chemistry. We examine the properties of Stringmol with respect to the principles, and so demonstrate the usefulness of the principles in designing automata chemistries.

Our long term goal is to develop an open-ended reflective software architecture to support open-ended evolution. Here we describe a preliminary experiment using reflection to make simple programs evolved via Grammatical Evolution robust to mutations that result in coding errors. We use reflection in the domain of grammatical evolution (GE) to achieve a novel means of robustness by autonomously repairing damaged programs, improving continuity in the search and allowing programs to be evolved effectively using soft grammars. In most implementations of GE, individuals whose programs encounter errors are assigned the worst possible fitness; using the techniques described here, these individuals may be allowed to continue evolving. We describe two different approaches to achieving robustness through reflection, and evaluate their effectiveness through a series of experiments carried out on benchmark regression problems. Results demonstrate a statistically significant improvement on the fitness of the best individual found during evolution

Biological organisms exist within environments in which complex nonlinear dynamics are ubiquitous. They are coupled to these environments via their own complex dynamical networks of enzyme-mediated reactions, known as biochemical networks. These networks, in turn, control the growth and behavior of an organism within its environment. In this paper, we consider computational models whose structure and function are motivated by the organization of biochemical networks. We refer to these as artificial biochemical networks and show how they can evolve to control trajectories within three behaviorally diverse complex dynamical systems: 1) the Lorenz system; 2) Chirikov's standard map; and 3) legged robot locomotion. More generally, we consider the notion of evolving dynamical systems to control dynamical systems, and discuss the advantages and disadvantages of using higher order coupling and configurable dynamical modules (in the form of discrete maps) within artificial biochemical networks (ABNs). We find both approaches to be advantageous in certain situations, though we note that the relative tradeoffs between different models of ABN strongly depend on the type of dynamical systems being controlled.

A novel bio-inspired architecture comprising three layers is introduced for a six-legged robot in order to generate adaptive rhythmic locomotion patterns using environmental information. Taking inspiration from the intracellular signalling processes that decode environmental information, and considering the emergent behaviours that arise from the interaction of multiple signalling pathways, we develop a decentralised robot controller composed of a collection of artificial signalling networks. Crosstalk, a biological signalling mechanism, is used to couple such networks favouring their interaction. We also apply nonlinear oscillators to model gait generators, which induce symmetric and rhythmical locomotion movements. The trajectories are modulated by a coupled artificial signalling network, which yields adaptive and stable robotic locomotive patterns. Gait trajectories are converted into joint angles by means of inverse kinematics. The architecture is implemented in a simulated version of the real robot T-Hex. Our results demonstrate the ability of the architecture to generate adaptive and periodic gaits.

Abstract The issues surrounding the question of atomicity, both in the past and nowadays, are briefly reviewed, and a picture of an ACID (atomic, consistent, isolated, durable) transaction as a refinement problem is presented. An example of a simple air traffic control system is introduced, and the discrepancies that can arise when read-only operations examine the state at atomic and finegrained levels are handled by retrenchment. Non-ACID timing aspects of the ATC example are also handled by retrenchment, and the treatment is generalised to yield the Retrenchment Atomicity Pattern . The utility of the pattern is confirmed against a number of different case studies. One is the Mondex Electronic Purse, its protocol treated as a conventional atomic transaction. Another is the recovery protocol of Mondex, viewed as a compensated transaction (leading to the view that compensated transactions in general fit the pattern). A final one comprises various unruly phenomena occurring in the implementations of software transactional memory systems, which can frequently display non-ACID behaviour. In all cases the Atomicity Pattern is seen to perform well.

In studying complex systems, agent-based simulations offer the possibility of directly modelling components in an environment. However, the scientific value of agent-based simulations has been limited by inadequate scientific rigour. The paper focuses on agent-based simulations that are used in biological and bio-medical research. Starting from a review of best practice in simulation engineering, the paper identifies some of the key activities in developing complex systems simulations that support scientific research, and how these contribute to the essential development of mutual trust among developers and scientists. Examples from the authors' own experience illustrate how a range of studies have manifested these key activities, and identifies some successes and problems encountered.

L-systems are parallel generative grammars that can model branching structures. Taking a graphical object and attempting to derive an L-system describing it is a hard problem. Grammatical Evolution (GE) is an evolutionary technique aimed at creating grammars describing the legal structures an object can take. We use GE to evolve L-systems, and investigate the effect of elitism, and the form of the underlying grammar.

The bacterial genome is well understood by biologists. Although its efficiency and adaptability should make it a good model for evolutionary algorithms, the bacterial genome is tightly coupled with the components of the bacterial metabolism, referred to here as the metabolome. This paper explores an approach to modelling an artificial bacterial metabolome in an efficient and modular manner, so that analogues of bacterial genome organisation and gene regulation can be implemented in evolutionary algorithms. We propose a particulate model of bacterial metabolic pathways in which the constituents drift in a fixed, limited space and obey a limited set of biologically plausible reaction rules. The potential of this model is demonstrated by creating a network that is capable of appropriate behavioural switching that can be observed in bacteria.

An evolutionary algorithm automatically discovers suitable solutions to a problem, which may lie anywhere in a large search space of candidate solutions. In the case of genetic programming, this means performing an efficient search of all possible computer programs represented as trees. Exploration of the search space appears to be constrained by structural mechanisms that exist in genetic programming as a consequence of using trees to represent solutions. As a result, programs with certain structures are more likely to be evolved, and others extremely unlikely. We investigate whether the graph representation used in Cartesian genetic programming causes an analogous biasing effect, imposing natural limitations on the class of solution structures that are likely to be evolved. Representation bias and structural bias are identified: the rarer ldquoregularrdquo structures appear to be easier to evolve than more common ldquoirregularrdquo ones.

We propose a simple approach to visualising the time behaviour of Random Boolean Networks (RBNs), and demonstrate the approach by examining the effect of canalising functions for K > 2 networks.

We discuss applications of an underlying anonymous infrastructure to enforce fair behaviour on participants in a distributed resource-sharing system. This approach aims to prevent users from forming self-rewarding cliques in order to gain unfair advantages in the use of shared resources.

Abstract. Ten years ago the Mondex electronic purse was certified to ITSEC Level E6, the highest level of assurance for secure systems. This involved building formal models in the Z notation, linking them with refinement, and proving that they correctly implement the required security properties. The work has been revived recently as a pilot project for the international Grand Challenge in Verified Software. This paper records the history of the original project and gives an overview of the formal models and proofs used.

The new breed of reconfigurable integrated circuits (ICs) offer switched-capacitor based analogue circuits whose functionality can be altered during run-time. Rapidly changing the functionality of an analogue circuit provides interesting opportunities for control systems. It also opens a large design space in which decisions have to be made regarding the frequency and form of reconfiguration. We present a bio-inspired architecture to facilitate the automated search for circuits on these platforms, based on well-established evolutionary algorithms. Unlike previous attempts at evolving single solutions, our genomes contain multiple solutions and use feedback provided by the circuit output to trigger reconfiguration of the IC.

We present what we believe is the first attempt to physically reconstruct the exploratory mechanism of genetic regulatory networks. Feedback plays a crucial role during developmental processes and its mechanisms have recently become much clearer due to evidence from evolutionary developmental biology. We believe that without similar mechanisms of interaction and feedback, digital genomes cannot guide themselves across functional search spaces in a way that fully exploits a domain's resources, particularly in the complex search domains of real-world physics. Our architecture is designed to let evolution utilise feedback as part of its mechanism of exploration.

The issues surrounding the question of atomicity, both in the past and nowadays, are briefly reviewed, and a picture of an ACID (atomic, consistent, isolated, durable) transaction as a refinement problem is presented. An example of a simple air traffic control system is introduced, and the discrepancies that can arise when read-only operations examine the state at atomic and finegrained levels are handled by retrenchment. Non-ACID timing aspects of the ATC example are also handled by retrenchment, and the treatment is generalised as the retrenchment Atomicity Pattern. The utility of the pattern is confirmed against a different case study, the Mondex Electronic Purse.

Recent years have seen the application of evolutionary and other nature-inspired search approaches to achieve human-competitive results in cryptography and cryptanalysis. We have also seen the emergence of quantum computation as a tremendously exciting computational paradigm with significant potential applications in these areas. To date there seems to have been no synergistic application of these techniques in these fields. All applications are geared to the effective exploitation of one computational paradigm or another. Nature-inspired search and quantum computing can, however, be combined to achieve results neither is capable of individually. All that is needed is that classical search get 'close enough' for quantum search to take over and solve the residual problem. This observation has significant implications for the security of crypto-systems and our understanding of the power and usefulness of nature-inspired and quantum search.

The Mondex Electronic Purse is an outstanding example of industrial scale formal refinement, and was the first verification to achieve ITSEC level E6 certification. A formal abstract model and a formal concrete model were developed, and a formal refinement was hand-proved between them. Nevertheless, certain requirements issues were set beyond the scope of the formal development, or handled in an unnatural manner. The retrenchment tower pattern is used to address one such issue in detail: the use of a hash function rather than a total injective function when clearing the highly constrained purse logs. A retrenchment is constructed from the lowest level model to a model using a hash, and is then lifted to create two refinement developments, working at different levels of detail, and connected via retrenchments. The tower development is appropriately validated, vindicating the design used.

The Mondex electronic purse is an outstanding example of industrial scale formal refinement, and was the first verification to achieve ITSEC level E6 certification. A formal abstract model and a formal concrete model were developed, and a formal refinement was hand-proved between them. Nevertheless, certain requirements issues were set beyond the scope of the formal development, or handled in an unnatural manner. The retrenchment tower pattern is used to address one such issue in detail: the finiteness of the purse log (which records unsuccessful transactions). A retrenchment is constructed from the lowest level model of the purse system to a model in which logs are finite, and is then lifted to create two refinement developments of the purse, working at different levels of detail, and connected via retrenchments, forming the tower. The tower development is appropriately validated, vindicating the design used

We show how Genetic Programming (GP) can be used to evolve useful quantum computing artefacts of increasing sophistication and usefulness: firstly specific quantum circuits, then quantum programs, and finally system-independent quantum algorithms. We conclude the paper by presenting a human-competitive Quantum Fourier Transform (QFT) algorithm evolved by GP.

In this paper, we show how genetic programming (GP) can be used to evolve system-size-independent quantum algorithms, and present a human-competitive Quantum Fourier Transform (QFT) algorithm evolved by GP.

Boolean function design is at the heart of cryptography, and is the subject of a great deal of theoretical research. We have use a simulated annealing approach to find functions with particular desirable cryptographic properties; for functions of a small number of variables, results with properties as good as (and sometimes better than) the best so far have been achieved. The success of this approach is very sensitive to the cost function chosen; here we investigate this property, and describe a meta-search approach to finding the most effective cost function for this class of problems.

Substitution boxes are important components in many modern day block and stream ciphers. Their study has attracted a great deal of attention over many years. The development of a variety of cryptosystem attacks has lead to the development of criteria for resilience to such attacks. Some general criteria such as high nonlinearity and low autocorrelation have been proposed (providing some protection against attacks such as linear cryptanalysis and differential cryptanalysis). There has been little application of evolutionary search to the development of s-boxes. In This work we show how a cost function that has found excellent single-output Boolean functions can be generalised to provide improved results for small s-boxes.

When formal modelling is done we must validate both the model and the assumptions. Formal techniques tend to concentrate on the former. We examine how fault injection (specification mutation) and model checking can help address the latter, in particular, the effects of failure. We find that, in contrast with software testing, where they are a problem, "equivalent mutants" are valuable for specification validation.

Multiple anagramming is a general method for the cryptanalysis of transposition ciphers, and has a graph theoretic representation. Inspired by a partially mechanised approach used in World War II, we consider the possibility of a fully automated attack. Two heuristics based on measures of natural language are used - one to recognise plaintext, and another to guide construction of the secret key. This is shown to be unworkable for cryptograms of a certain difficulty due to random variation in the constructive heuristic. A solver based on an ant colony optimisation (AGO) algorithm is then introduced, increasing the range of cryptograms that can be treated; the pheromone feedback provides a mechanism for the recognition heuristic to correct the noisy constructive heuristic.

Abstract As critical applications grow in size and complexity, high level languages, rather than better-trusted assembly languages, will be used in their development. This adds potential for extra errors to creep in, especially in the now necessary compiler. To avoid these new errors, it is necessary to have a formal specification of the high level language, and a formal development of its compiler. We outline what we believe is a practical route for achieving a demonstrably correct compiler, and describe a prototype compiler we have built by this route for a small, but non-trivial, language.

Abstract Computing facilities networked together but controlled by different administrations pose a problem of access control. Who decides who can use what? We specify a formal model for an access control system which allows users and services from different administrations to communicate with each other, while still allowing the administrators to retain control of their own parts of the network. The model, written in the Z specification language, has been developed as the access control system for ADMIRAL, though it is not specific to ADMIRAL. It provides a framework for administrators to build access control systems to meet their differing requirements. A system based on the model would allow users to log in to a distributed computing system and to make requests for services in any part of the system, without having to provide any more information about themselves. After this initial log in all subsequent access control decisions are handled automatically, and remain invisible to the user unless access is refused.ö We also discuss the experience we have had animating this model in Prolog.