Files in this item



application/pdf2002_melby.pdf (2Mb)Restricted to U of Illinois


Title:Adaptation to the Edge of Chaos and Critical Scaling in Self-adjusting Dynamical Systems
Author(s):Melby, Paul Christian
Doctoral Committee Chair(s):Hubler, Alfred W.
Department / Program:Physics
Subject(s):dynamical systems
self-adjusting systems
edge of chaos
Abstract:We present a mechanism for adaptation in dynamical systems. Systems which have this mechanism are called called self-adjusting systems. The control parameters in a self-adjusting system are slowly varying, rather than constant. The dynamics of the control parameters are governed by a lowpass filtered feedback from the dynamical variables. We apply this model to several systems, numerically, analytically, and experimentally, and examine the behavior of the control parameters. We observe a high probability of finding the parameter at the boundary between periodicity and chaos. We therefore find that self-adjusting systems adapt to the edge of chaos. In addition, we find that noise in the system drives the parameter away from the edge of chaos on very long timescales so that chaos is suppressed in the system. We show that, with the presence of noise, the parameter can re-enter the chaotic regime. This is called a chaotic outbreak in the system and we find that the distribution of outbreaks is a power-law with the duration of the outbreak. We then study the robustness of adaptation to the edge of chaos by examining the effect of a control force being applied to the parameter. We find the behavior to be very robust, except for very large control forces. Finally, we look at systems of coupled maps and show that adaptation to the edge of chaos occurs in systems of higher dimensions, as well.
Issue Date:2002
Genre:Dissertation / Thesis
Other Identifier(s):4591801
Rights Information:©2002 Melby
Date Available in IDEALS:2012-06-05

This item appears in the following Collection(s)

Item Statistics

  • Total Downloads: 2
  • Downloads this Month: 0
  • Downloads Today: 0