No abstract

Some statistical features of steady states of a Kauffman-like model for prebiotic evolution are reported from computational studies. We postulate that the interesting "lifelike" states will be characterized by a nonequilibrium distribution of species and a time variable species self-correlation function. Selecting only such states from the population of final states produced by the model yields the probability of the appearance of such states as a function of a parameter p of the model. p is defined as the probability that a possible reaction in the the artificial chemistry actually appears in the network of chemical reactions. Small p corresponds to sparse networks utilizing a small fraction of the available reactions. We find that the probability of the appearance of such lifelike states exhibits a maximum as a function of p: at large p, most final states are in chemical equilibrium and hence are excluded by our criterion. At very small p, the sparseness of the network makes the probability of formation of any nontrivial dynamic final state low, yielding a low probability of production of lifelike states in this limit as well. We also report results on the diversity of the lifelike states (as defined here) that are produced. Repeated starts of the model evolution with different random number seeds in a given reaction network lead to final lifelike states which have a greater than random likelihood of resembling one another. Thus a form of "convergence" is observed. On the other hand, in different reaction networks with the same p, lifelike final states are statistically uncorrelated. In summary, the main results are (1) there is an optimal p or "sparseness" for production of lifelike states in our model-neither very dense nor very sparse networks are optimal--and (2) for a given p or sparseness, the resulting lifelike states can be extremely different. We discuss some possible implications for studies of the origin of life.

No abstract

No abstract

No abstract