Simulation performance versus stochasticity in large-scale cellular automata models

Due to a growing interest in chemical and biological phenomena, simulation of reaction-diffusion processes on micro-level becomes urgently wanted. Asynchronous cellular automata are promising mathematical models to be used as a base for creating computer simulation systems, which gives reason for the investigation of their capability. In particular, since the micro-level simulation requires a very large size of cellular automata, the performance of simulation is important, especially, because parallel implementation is inevitable. In this connection, the dependence of simulation performance on the CA stochasticity (the degree of randomness) is studied in this paper. Much attention is being given to the contradiction between the stochasticity and the parallelization efficiency. It is shown how the proper choice of the CA stochasticity may help to achieve the acceptable performance. The results of a 3D process of diffusion limited aggregation are presented to assert the dependencies obtained.