Reversible Boolean networks. II: Phase transitions, oscillations, and local structures (Q5946575)

The paper studies extensions of so called Kauffman nets, a model for dynamical systems that has been used in social science, physics, economics, etc. A Kauffman net consists of a vector of \(N\) Boolean variables that evolves over time. A Boolean variable in this paper is assumed to take on values \(-1,1\) (hence also the name spin). The value of the \(i\)th variable at time \(t\) is denoted by \(\sigma_i^t\). The state of the net at time \(t\) is \(\Sigma^t=(\sigma_1^t,\dots,\sigma_n^t)\). Now, \(\sigma_i^{t+1}\) is given by a \(K\)-ary Boolean function \(F_i\) where the inputs are chosen among the \(n\) variables \(\sigma_1^t,\dots,\sigma_n^t\); this is denoted by the authors by \(\sigma_i^{t+1}=F_i(\Sigma^t)\), slightly misusing common mathematical notation: formally, \(F_i\) here is an \(N\)-ary Boolean function that has only \(K\) non-fictive inputs. The functions \(F_i\) are fixed, and a starting vector \((\sigma_1^0,\dots,\sigma_n^0)\) is picked. The present paper now studies a time-reversible extension of Kauffman nets, where for the computation of the state at time \(t+1\) both the state at time \(t\) and time \(t-1\) are needed; more precisely: \(\sigma_i^{t+1}=F_i(\Sigma^t)\sigma_i^{t-1}\). This extension of Kauffman nets was introduced in a precursor of this paper by the same authors [Reversible Boolean networks. I: Distribution of cycle lengths, Physica D 149, 11-29 (2001; Zbl 1004.94041)]. Issues studied in this paper are distribution of cycle length and behavior of the Hamming distance in the time-reversible model. A crucial parameter in the results obtained is the number \(K\) above. As an example, numerical results show that an enormous change of the behaviour of the Hamming distance around \(K\approx 1.62\) can be observed.