PM-colonies

Abstract

A colony is meant to be a symbol manipulating system consisting of as simple as possible components which behave in a cooperative way such that the collective competence is strictly larger (even significantly larger) than the component competences. We introduce here colonies whose agents can only perform point mutation (hence the abbreviation PM) transformations of the common string (which represents the environment of the colony), in a vicinity of the agent. In this way, important notions of this area, such as  localization of agents, parallelism, lack of internal representation, agent interaction, come into stage in a very natural way. In contrast with the simplicity of the involved agents, the behavior of PM-colonies is quire intricate: many problems concerning the "life" of a colony are not algorithmically solvable, the number of agents in the colony or simultaneously present in the environment defines infinite hierarchies of languages, etc. Such results show that the behaviour of PM-colonies is not predictable and that their behaviour is significantly synergetic.