A Look At Multilevel Splitting

This paper gives a non-technical overview of our work on the efficiency of a simple multilevel splitting technique for estimating rare event probabilities. For a particular class of stochastic models, this method is asymptotically optimal provided the splitting factor is properly chosen. However, for more general classes of processes, the method fails to be asymptotically optimal unless the splitting is done in a manner consistent with the dominant way in which the rare event happens (the large deviations behavior). In the absence of such consistency, the method also exhibits a kind of bias in which the estimate is too small with high probability. We briefly contrast our approach with one used in nuclear physics.

By: Paul Glasserman (Columbia Univ.), Philip Heidelberger, Perwez Shahabuddin (Columbia Univ.) and Tim Zajic (Columbia Univ.)

Published in: RC20692 in 1997


This Research Report is available. This report has been submitted for publication outside of IBM and will probably be copyrighted if accepted for publication. It has been issued as a Research Report for early dissemination of its contents. In view of the transfer of copyright to the outside publisher, its distribution outside of IBM prior to publication should be limited to peer communications and specific requests. After outside publication, requests should be filled only by reprints or legally obtained copies of the article (e.g., payment of royalties). I have read and understand this notice and am a member of the scientific community outside or inside of IBM seeking a single copy only.


Questions about this service can be mailed to reports@us.ibm.com .