Jeff JONES
Abstract. The single celled organism Physarum polycephalum has been shown to be efficient in the construction and minimisation of dynamical nutrient transport networks resembling proximity graphs. Previous research has investigated the evolution of networks from both fully connected networks, and from inoculation at a single node. We present a model multi-agent population which collectively approximates the network behaviours of Physarum. We demonstrate spontaneous transport network formation and evolution and show that the collective population also exhibits second order quasi-physical emergent properties. These properties allow the collective population to be considered as a virtual computing material – a synthetic plasmodium. This ‘substance’ is used as an unconventional method to approximate spatially represented geometry problems. We demonstrate three different methods for the construction, evolution and minimisation of Physarum-like transport networks which approximate Steiner trees, relative neighbourhood graphs, convex hulls and concave hulls. The results span the Toussaint hierarchy of proximity graphs and suggest that the foraging and minimising behaviours of Physarum reflect interplay between maximising area covered and minimising distance for transport. We hope that the properties and behaviours of the synthetic virtual plasmodium may be useful in future physical instances of unconventional computing devices, and may also provide some clues to the generation of emergent computation behaviours by Physarum.