Efficient Simulation of Rare Events in one-dimensional systems using a parallelised cloning algorithm

  • Tobias Brewer

Student thesis: Doctoral ThesisPhD

Abstract

We consider the population dynamics that are implemented by the cloning algorithmfor analysis of large deviations of time-averaged quantities. We consider exclusionprocesses acting on particles on one-dimensional lattices such as the simple symmetricexclusion process and the Fredkin Process. We use large deviation theoryto quantify the probabilities of rare events. To achieve this we adapt a numericalalgorithm which employs a combination of biased cloning and simulation of modi-ed dynamics. We establish its accuracy within particular regimes, determine whichcongurations are likely to produce rare events and quantify the convergence of thealgorithm with respect to algorithmic parameters. We investigate the eciency andspeed-up obtained when using dierent parallelisation techniques to implement thealgorithm which involves complex communication patterns between systems.
Date of Award1 May 2019
LanguageEnglish
Awarding Institution
  • University of Bath
SupervisorStephen Clark (Supervisor), Robert Jack (Supervisor) & Russell Bradford (Supervisor)

Cite this

Efficient Simulation of Rare Events in one-dimensional systems using a parallelised cloning algorithm
Brewer, T. (Author). 1 May 2019

Student thesis: Doctoral ThesisPhD