English

Reversible boolean networks II: Phase transition, oscillation, and local structures

Disordered Systems and Neural Networks 2016-08-31 v2 Statistical Mechanics Dynamical Systems Adaptation and Self-Organizing Systems Cellular Automata and Lattice Gases Computational Physics

Abstract

We continue our consideration of a class of models describing the reversible dynamics of NN Boolean variables, each with KK inputs. We investigate in detail the behavior of the Hamming distance as well as of the distribution of orbit lengths as NN and KK are varied. We present numerical evidence for a phase transition in the behavior of the Hamming distance at a critical value Kc1.65K_c\approx 1.65 and also an analytic theory that yields the exact bounds on 1.5Kc2.1.5 \le K_c \le 2. We also discuss the large oscillations that we observe in the Hamming distance for K<KcK<K_c as a function of time as well as in the distribution of cycle lengths as a function of cycle length for moderate KK both greater than and less than KcK_c. We propose that local structures, or subsets of spins whose dynamics are not fully coupled to the other spins in the system, play a crucial role in generating these oscillations. The simplest of these structures are linear chains, called linkages, and rings, called circuits. We discuss the properties of the linkages in some detail, and sketch the properties of circuits. We argue that the observed oscillation phenomena can be largely understood in terms of these local structures.

Keywords

Cite

@article{arxiv.cond-mat/0009019,
  title  = {Reversible boolean networks II: Phase transition, oscillation, and local structures},
  author = {S. N. Coppersmith and Leo P. Kadanoff and Zhitong Zhang},
  journal= {arXiv preprint arXiv:cond-mat/0009019},
  year   = {2016}
}

Comments

31 pages, 15 figures, 2 tables