CLUSTERING, CODING, SWITCHING, HIERARCHICAL ORDERING, AND CONTROL IN A NETWORK OF CHAOTIC ELEMENTS

A Network of chaotic elements is investigated with the use of globally coupled maps. A simple coding of many attractors with clustering is shown. Through the coding, the attractors are organized so that their change exhibits bifurcation-like phenomena. A precision-dependent tree is constructed which leads to the similarity of our attractor with those of spin-glasses. Hierarchical dynamics is constructed on the tree, which leads to the dynamical change of trees and the temporal change of effective degrees of freedom. By a simple input on a site, we can switch among attractors and tune the strength of chaos. A threshold on a cluster size is found, beyond which a peculiar "posi-nega" switch occurs. Possible application to biological information processing is discussed with the emphasis on the fuzzy switch (chaotic search) and hierarchical code (categorization). © 1990.