Attractor dominance patterns in sparsely connected Boolean nets (Q2640962)

The author considers the generalized cellular automata which are lattices of N sites affected by themselves and two other, not necessarily different, sites. The state of each site at time \(t+1\) is given by a binary function T on three arguments. For a fixed lattice size N and rule T, a lattice connection table is chosen at random. Then a state is chosen equiprobably and its limit cycle under the produced cellular automaton is found. Finally another state at a distance \(\Delta\) from the cycle is chosen at random and the cellular automaton is restarted at the displaced state. The displaced state is counted as dominated by the cycle, if its trajectory reencounters the original cycle. The procedure is repeated more than 1000 times and the percentage \(D_{\Delta}\) of the displaced states at distance \(\Delta\) that are dominated by the cycle is computed. \(D_{\Delta}\) as the function of \(\Delta\) is called the dominance function. The author classifies various dominance functions into three main classes: coherent and bipolar, which seem to be most frequent, and also diffuse class. The three trivial rules T have representatives in each of the classes. Increasing lattice size appears to move most, but, importantly, not all dominance patterns toward the extremes shown by the three trivial rules.