Constructing Cellular Automaton Models from Observation Data

We propose a construction method of cellular automaton models based on observation data of phenomena. Because variables and states of cellular automata should be discrete, we discretize the observation data. Under the given number of neighbors and possible states of a site, we estimate the rules of cellular automata by the statistic analysis. In this paper, we apply this method to the discretized diffusion equation and Burgers equation. Both deterministic and stochastic models, which have three neighbors and 2-8 possible states, are obtained. Results of the models are compared with that of the original equations.