Downing, Keith L. “Modular designer chemistries for artificial life.” In Proceedings of the 3rd Annual Conference on Genetic and Evolutionary Computation , pp. 845-852. 2001.
The vast majority of artificial life (alife) systems lack underlying chemical models that exhibit both a) realistic relationships between biomass production/consumption and energy, and b) strong restrictions on the legal combinations of atomic units. In many cases, biomass is simply accumulated resource, occasionally paid for by an energy tax. This suffices for many purposes, but when the alife system serves as a testbed for microbiological investigations, more realism is desirable. However, real biochemistry is so complex that one naturally turns to the alife discipline of abstract/artificial chemistries for rich, yet manageable chemical backdrops; but with little success.
This paper introduces MD-CHEM, a simple algorithm for generating random chemistries that a) meet user specifications, b) capture important biomass-energy relations, and c) easily plug into alife simulators. Runs of these chemistries in our METAMIC simulator illustrate how a) organisms evolve to exploit the energetic potential of the chemistry and environment, and b) chemistries restrict population density and diversity.