Previous abstract Next abstract
We explore the generic behavior and the possibility of structural phase transitions in spatially-extended automata network models of various star formation scenarios. Included are the expansion of dense shells driven by supernovae or stellar winds, heating and cooling of the gas, and a star formation threshold depending on the gas density and temperature. The model architecture is a hybrid of cellular automata and neural networks in which the qualitative features of hydrodynamic simulations (e.g. shell expansion and collisions) are abstracted into the network rules and connections. Particular attention is given to the transient behavior of the models. Many complex spatially extended systems exhibit very long transients, in which the collective spatio-temporal behavior is highly irregular and intermittent, before settling into any equilibrium state. The duration of such transients is a function of the system size. Our preliminary results suggest that the transient time for these model galaxies is in fact quite long when the network is near a critical point. This effect could have important consequences in interpreting the evolution of high redshift galaxies, as well as galaxies at the current epoch.