Construction of adaptive biomass networks using the Physarum polycephalum strategy
Construction of adaptive biomass networks using the Physarum polycephalum strategy
Description of the
Physarum polycephalum is a slime mould, i.e. a larger, unicellular, amoeboid organism. The active search for food is positively chemotactic. Once a food source has been identified, veins contract rhythmically and then relax again in order to transport the available cytoplasm. If one of the two phases lasts longer, the organism moves in the corresponding direction.
Tero et al. analysed slime moulds and derived a mathematical model for the adaptive development of networks. This model was applied, for example, to the current railway network in the Tokyo region. Initially, 36 food sources were selected to represent (to scale) the geographical position of cities in the Tokyo area. The starting point for Physarum polycephalum was Tokyo. After initially growing over a large area, direct connections were formed (after 8 hours) between the various food sources. After about a day, the fungus had reproduced the actual rail network almost congruently.
The mathematical model presented by Tero et al. is to be transferred to the application scenario of optimising biomass networks as part of the final thesis. This requires, for example, an extension of the method to include different feed types (supply and demand), which must, however, still be transported via/through the same network.
In addition to designing and realising a suitable simulation environment and possibly a GUI, the focus is on creating a data format for specifying (at least) the slime mould's food sources and the simulation parameters. An evaluation of the simulation results can be limited to an implementation analogous to Tero et al.
As the work outlined is characterised by a high practical component, experience in the adaptation and implementation of algorithms is absolutely essential. At a later stage, the realised software is to be linked to existing systems, so that realisation of this work in Java would be an advantage. If the candidate prefers another programming language, he/she must also implement appropriate interfaces that enable methods/functions to be called from Java.
The exact design of the thesis must be agreed with the supervisor.
Atsushi Tero, Seiji Takagi, Tetsu Saigusa, Kentaro Ito, Dan P. Bebber, Mark D. Fricker, Kenji Yumiki, Ryo Kobayashi, Toshiyuki Nakagaki:
Rules for Biologically Inspired Adaptive Network Design
Science, 327(5964):439-442, 2010