doi:10.3808/jei.200300008
Copyright © 2022 ISEIS. All rights reserved

Emergent Computation and Modelling: Complex organization and Bifurcation within Environmental Bounds (COBWEB)

N. Suh1,B. Bass2*,E. Chan1 and N. Toller1

  1. Division of the Environment, University of Toronto, 33 Willcocks Street, Toronto, Ontario M5S 3E8, Canada
  2. Adaptation & Impacts Research Group, Environment Canada at the University of Toronto, Institute for Environmental Studies, 33 Willcocks Street, Toronto, Ontario M5S 3E8, Canada

*Corresponding author. Email: brad.bass@ec.gc.ca

Abstract


Currently, emergent computation (EC) is a relatively new approach for understanding ecosystem dynamics. Central to this approach is the idea that high-level ecosystem dynamics emerging from low-level interactions of individual agents. In an effort to further understand this dynamic new field, this paper will examine the conceptual background which researchers use to EC models and review the main classes of EC models: cellular automata, genetic algorithms, classifier systems, and neural networks. Finally, the paper will introduce and discuss the results of trials investigating the boundaries of Complex Organization and Bifurcation within Environmental Bounds (COBWEB) model. COBWEB obeys emergent computation characteristics and is a general simulation platform developed to support a large number of independent agents, each encoded by a GA, in a 2-D environment with resources encoded by a cellular automaton. COBWEB displays non-linear behaviour and is characterized by five main attractors: predator-prey cycling between resources and a dominant agent, agent elimination, unlimited growth of agents, predator-prey cycling with no dominant agent and unpredictability.

Keywords: Agent-based modeling, anticipatory agents, emergent computation, genetic algorithms


Full Text:

PDF

Supplementary Files:

Refbacks

  • There are currently no refbacks.