Skip to main navigation Skip to search Skip to main content

An importance sampling method based on martingale with applications to rare event probability

  • Yue Qiu*
  • , Hong Zhou
  • , Yueqin Wu
  • *Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

It usually takes long time to simulate rare event using traditional Monte Carlo method, while importance sampling techniques can effectively reduce the simulation time and improve simulation efficiency. A new implementation for importance sampling method to estimate rare event probability in simulation models is proposed. The optimal importance sampling distributions was obtained by making use of the martingale constructed by likelihood ratio. The computation results were compared with the importance sampling based on cross-entropy, the importance sampling based on minimizing variance and crude Monte Carlo method. Numerical experiments had been conducted and the results indicate that the method can effectively estimate the rare event probabilities.

Original languageEnglish
Title of host publicationProceedings of the 7th World Congress on Intelligent Control and Automation, WCICA'08
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages4041-4045
Number of pages5
ISBN (Print)9781424421145
DOIs
StatePublished - 2008
Event7th World Congress on Intelligent Control and Automation, WCICA'08 - Chongqing, China
Duration: 25 Jun 200827 Jun 2008

Publication series

NameProceedings of the World Congress on Intelligent Control and Automation (WCICA)

Conference

Conference7th World Congress on Intelligent Control and Automation, WCICA'08
Country/TerritoryChina
CityChongqing
Period25/06/0827/06/08

Keywords

  • Importance sampling
  • Likelihood ratio
  • Martingale
  • Rare event

Fingerprint

Dive into the research topics of 'An importance sampling method based on martingale with applications to rare event probability'. Together they form a unique fingerprint.

Cite this