Abstract:

We introduce multi-level Artificial Chemistries as a way of tackling difficult problems in the evolution of complexity. We present two algorithms for moving between levels of abstraction in a multi-level Artificial Chemistry. (1) Moving upwards from a low-level description to a high-level description involves making approximations. We discuss these, and provide an algorithm to perform the approximations. (2) Moving downwards is more problematic. We discuss the issues involved in moving down, including conservation of mass. We present an algorithm to generate constraints that any lowlevel implementation of the system must satisfy. These constraints can be used to: obtain information about the system; automatically generate a low-level implementation of the system; guide a search for suitable low-level implementations of the system.