Distributed Individual-Based Simulation

  • Authors:
  • Jiming Liu;Michael B. Dillencourt;Lubomir F. Bic;Daniel Gillen;Arthur D. Lander

  • Affiliations:
  • University of California, Irvine 92697;University of California, Irvine 92697;University of California, Irvine 92697;University of California, Irvine 92697;University of California, Irvine 92697

  • Venue:
  • Euro-Par '09 Proceedings of the 15th International Euro-Par Conference on Parallel Processing
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Individual-based simulations are an important class of applications where a complex system is modeled as a collection of autonomous entities, each having its own identify and behavior in the underlying simulated space. The main drawback of such simulations is that they are extremely compute-intensive. We consider the class of individual-based simulations where the simulated entities interact with one another indirectly through the underlying simulated space, significant performance improvement is attainable through parallelism on a network of machines. We present a data distribution and an approach to reduce the communication overhead, which leads to significant performance improvements while preserving the accuracy of the simulation.