Computational Science Technical Note CSTN-114

CSTN Home

Optimal Data Structures for Spatially Localised Agent-Based Automata and Hybrid Systems

C. J. Scogings and K. A. Hawick

Archived August 2010, Revised March 2012

Abstract

Agent-based systems and cellular automata are two closely related model formulations that are heavily used in studying complex systems. They are both formulated as microscopically simple rule-based models that are applied to individual cells or agents in a collection, where the spatially localised neighbourhood of other cells or agents is used as input to update each one. We have experimented with a range of models including classic cellular automata, through more sophisticated multi-state automata, flocking models, and stochastic-agent models and animat agent-based predator-prey models. We discuss algorithmic commonalities and code implementation patterns that have emerged as common properties of these models and describe how we have experimented with optimal data structures to support spatially localised models of this class. We show how concurrency and model correctness issues are affected by different data structures in addition to their effect on model update computational efficiency.

Keywords: agent-based models; cellular automata; hybrid model; spatial locality; rule-set; efficient update; neighbourhood.

Full Document Text: PDF version.

Citation Information: BiBTeX database for CSTN Notes.

BiBTeX reference:

@INPROCEEDINGS{CSTN-114,
  author = {C. J. Scogings and K. A. Hawick},
  title = {Optimal Data Structures for Spatially Localised Agent-Based Automata
	and Hybrid Systems},
  booktitle = {Proc. Int. Conf. on Artificial Intelligence and Soft Computing},
  year = {2012},
  pages = {221-227},
  address = {Napoli, Italy},
  month = {25-27 June},
  publisher = {IASTED},
  institution = {Computer Science, Massey University},
  timestamp = {2010.09.09}
}


[ CSTN Index | CSTN BiBTeX ]