On the parallelization of network diffusion models

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 ap...

Full description

Bibliographic Details
Main Author: Rhomberg, Patrick
Other Authors: Segre, Alberto Maria
Format: Others
Language:English
Published: University of Iowa 2017
Subjects:
Online Access:https://ir.uiowa.edu/etd/5831
https://ir.uiowa.edu/cgi/viewcontent.cgi?article=7309&context=etd
id ndltd-uiowa.edu-oai-ir.uiowa.edu-etd-7309
record_format oai_dc
spelling ndltd-uiowa.edu-oai-ir.uiowa.edu-etd-73092019-10-13T04:35:00Z On the parallelization of network diffusion models Rhomberg, Patrick 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. 2017-08-01T07:00:00Z dissertation application/pdf https://ir.uiowa.edu/etd/5831 https://ir.uiowa.edu/cgi/viewcontent.cgi?article=7309&context=etd Copyright © 2017 Patrick Rhomberg Theses and Dissertations eng University of IowaSegre, Alberto Maria Dynamics Simulation Modeling Workload Balancing Applied Mathematics
collection NDLTD
language English
format Others
sources NDLTD
topic Dynamics
Simulation Modeling
Workload Balancing
Applied Mathematics
spellingShingle Dynamics
Simulation Modeling
Workload Balancing
Applied Mathematics
Rhomberg, Patrick
On the parallelization of network diffusion models
description 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.
author2 Segre, Alberto Maria
author_facet Segre, Alberto Maria
Rhomberg, Patrick
author Rhomberg, Patrick
author_sort Rhomberg, Patrick
title On the parallelization of network diffusion models
title_short On the parallelization of network diffusion models
title_full On the parallelization of network diffusion models
title_fullStr On the parallelization of network diffusion models
title_full_unstemmed On the parallelization of network diffusion models
title_sort on the parallelization of network diffusion models
publisher University of Iowa
publishDate 2017
url https://ir.uiowa.edu/etd/5831
https://ir.uiowa.edu/cgi/viewcontent.cgi?article=7309&context=etd
work_keys_str_mv AT rhombergpatrick ontheparallelizationofnetworkdiffusionmodels
_version_ 1719264527511453696