In this thesis, we investigate methods by which discrete event network diffusion simulators may execute without the restriction of lockstep or near lockstep synchronicity. We develop a discrete event simulator that allows free clock drift between threads, develop a differential equations model to approximate communication cost of such a simulator, and propose an algorithm by which we leverage information gathered in the natural course of simulation to redistribute agents to parallel threads such that the burden of communication is lowered during future replicates.
Dissertation
On the parallelization of network diffusion models
University of Iowa
Doctor of Philosophy (PhD), University of Iowa
Summer 2017
DOI: 10.17077/etd.4411rgfu
Abstract
Details
- Title: Subtitle
- On the parallelization of network diffusion models
- Creators
- Patrick Rhomberg - University of Iowa
- Contributors
- Alberto Maria Segre (Advisor)Philip Polgreen (Committee Member)Bruce Ayati (Committee Member)Colleen Mitchell (Committee Member)Sriram Pemmaraju (Committee Member)
- Resource Type
- Dissertation
- Degree Awarded
- Doctor of Philosophy (PhD), University of Iowa
- Degree in
- Applied Mathematical and Computational Sciences
- Date degree season
- Summer 2017
- DOI
- 10.17077/etd.4411rgfu
- Publisher
- University of Iowa
- Number of pages
- xv, 151 pages
- Copyright
- Copyright © 2017 Patrick Rhomberg
- Language
- English
- Description illustrations
- color illustrations
- Description bibliographic
- Includes bibliographical references (pages 144-147).
- Public Abstract (ETD)
This study investigates ways to reduce the computational cost of epidemic-like simulations. We model one of the most expensive portions of large simulations – communication between computers – and explore ways to distribute workload across computers so that this communication cost is reduced.
- Academic Unit
- Interdisciplinary Graduate Program in Applied Mathematical & Computational Sciences
- Record Identifier
- 9983776729502771
Metrics
320 File views/ downloads
150 Record Views