On asymptotically efficient simulation of large deviation probabilities

A.B. Dieker, M.R.H. Mandjes

Research output: Contribution to journalArticleAcademicpeer-review

22 Citations (Scopus)

Abstract

Let $\{\nu_{\varepsilon}, \varepsilon >0\}$ be a family of probabilities for which the decay is governed by a large deviation principle, and consider the simulation of $\nu_{\varepsilon_0}(A)$ for some fixed measurable set $A$ and some $\varepsilon_0>0.$ We investigate the circumstances under which an exponentially twisted importance sampling distribution yields an asymptotically efficient estimator. Varadhan's lemma yields necessary and sufficient conditions, and these are shown to improve on certain conditions of Sadowsky. This is illustrated by an example to which Sadowsky's conditions do not apply, yet for which an efficient twist exists.
Original languageUndefined
Article number10.1239/aap/1118858638
Pages (from-to)539-552
Number of pages14
JournalAdvances in applied probability
Volume37
Issue number2
DOIs
Publication statusPublished - 2005

Keywords

  • EWI-17609
  • Asymptotic efficiency
  • IR-70163
  • Large deviation
  • METIS-224185
  • Importance sampling

Cite this