Energy as a driver of diversity in open-ended evolution Tim Hoverd and Susan Stepney YCCSA, University of York, YO10 5GE, UK tim.hoverd@cs.york.ac.uk Abstract We investigate the consequences of introducing an energy model into open ended evolutionary simulations. We pro- pose a metamodel for simulations that incorporate an energy model and apply that model by extending Turk’s Sticky Feet model. We show that introducing an energy model produces simulations with measurably increased diversity of the simu- lated population. Introduction We are interested in open ended evolution and in particu- lar evolution within systems that are open to a simulated energy flux, open to changes in the simulated environment, and open to the representation of evolutionary mechanisms. In this paper we focus on energy flux, which allows us to represent many aspects of real world systems, such as the availability of food supplies, and different means of making a living within an environment, be they predatory or sessile. In order to investigate these issues we have chosen to extend Turk’s Sticky Feet [10] model. This gives a sim- ple mechanism for implementing mobility and experiment- ing with open-ended evolution. A Sticky Feet simulation is a collection of simulated creatures moving in a 2D do- main. Each such creature is a graph of springs connecting together feet. Motion is achieved as a consequence of sim- ple harmonic oscillation of the springs, which pushes the feet around within the simulation space. The coefficient of friction experienced by the feet is modulated–at times slippy, at times sticky–which results overall in motion through the space. Each creature has a heart and a mouth, each of which is a distinguished type of foot. The heart represents the crea- ture’s ‘essence’. The mouth–when it happens upon another creatures’s heart–allows the former creature to eat the latter, removing it from the simulation. The likelihood of a crea- ture happening upon another is facilitated to some extent by the springs being equipped with sensors, which may mod- ulate the oscillation of the spring when in the presence of another creature’s heart. This allows a creature to turn to- wards another, with the chance that it might then be able to consume the target. When a creature is consumed the eater produces a single offspring, which may be a mutation of the parent. Mutations that include additional feet, springs and sensors allow the creatures to evolve in a manner that even- tually produces offspring that are better adapted to hunting for and eating other creatures. A Sticky Feet world is one in which creatures evolve to improve their performance at consuming other creatures, and therefore being able to pass on their genome. As such, it provides some aspect of a model of open ended evolution. We use this term here in the sense of an evolutionary sys- tem where components continue to evolve new forms con- tinuously, rather than halting when some ‘optimal’ or stable position is reached [9]. Sticky Feet [10] works in this manner, as there is no over- all fitness function and all creature behaviour is expressed in a single large environment rather than relying on artifi- cial two-creature tournaments. As such it is representative of many aspects of real-world evolution. There is, though, no mechanism for sticky feet creatures to pass on their genomes other than by consuming other creatures. That is, the simulation is closed to the develop- ment of non-predatory behaviour. This is useful from the point of view of maintaining a constant sized simulation, but is not representative of real world evolution where popula- tion sizes can change dramatically. Natural evolution–that which operates in the world around us–is different in essence from the sticky feet model in that success does not entirely derive from hunting and reproduc- tion. Creatures in natural environments must be able to ex- tract some sort of living from that environment, supported either by consuming other creatures, or by turning some flux in the world, for example sunlight or the chemical nutrients consumed by extremophiles, into food. This argument is essentially that famously made by Malthus in 1798 [7], which led Darwin towards the prin- ciple of natural selection [2]. Although Malthus discussed the availability of food we generalise this to the availability of energy. This is a limited resource although the environ- ment is continually bathed in an energy flux. This flux may