Adrian Thompson
Centre for Computational Neuroscience and Robotics,
University of Sussex, Brighton BN1 9QH, UK.
adrianth@sussex.ac.uk
Recent experiments in evolutionary electronics have shown how artificial evolution can craft extremely efficient electronic circuits by manipulating a real physical silicon medium. Each individual circuit is physically instantiated in a reconfigurable chip (FPGA) for its fitness evaluation, so evolution can exploit all of the natural physical properties exhibited by the electronic medium, resulting in circuits well tailored to it. This can only be done properly by rigorously rejecting conventional design methods. Artificial evolution is then faced with a similar problem to that encountered in nature: how to construct a system from processes which all vary with temperature, such that the system can perform adequately over a wide range of temperatures? It is beneficial to do this in a more natural way than simply forbidding all analogue continuous-time dynamics, as conventional digital design does. Engineering proposals are formulated by analysing the correspondences between nature and evolutionary electronics -- some of these are promising and surprising. There are wider implications for ALife, in that thermal considerations cannot be as easily ignored as `implementation details' as might have been thought.