Is there a small-world in the small world of ants?
UMR 5263 CLLE (ERSS)
Last modified: December 19, 2007
How to measure the complexity of artificial simulation or ecosystem? How to know whether a simulation is more complex than another one? How to use simulation to understand self-organization?
This article presents a new approach to measure the complexity of artificial life simulation. To achieve that, one identifies in the simulation each parameter that seems relevant to measure simulation complexity and self-organization. During the simulation, the evolution of these parameters is used to build a graph. We analyze the evolution of the latter using four properties defining a small-world. These properties give the information about the rank of organization of the simulation.
In the results, we give many examples where it is possible to apply this method and we explain a successful application on an ant simulation. We compare different ant simulations with different parameters and we compare their evolutions and the different complexity thanks to properties of small-worlds. Producing small-worlds with all their properties is a difficult problem for generally the data come from the real life. This study also emphasizes that those ant simulations are an alternative method to produce randomly generated small-worlds.
These artificial small-worlds are interesting and could be compare to real small-worlds and help us to understand the complexity of the self-organization. In perspective, it could be interesting to see if we can establish some relations between the information theory and the properties of the small-worlds for a similar simulation.