State-dependent Importance Sampling and Large Deviations (with P. Glynn and J. C. Liu)

Summary:

This paper revisits the ideas behind the fact that large deviations techniques suggest many changes-of-measure that are applicable in importance sampling. Some of them are not optimal (despite being useful in the proof of large deviations results).

Bibtex:

@INPROCEEDINGS {BlanGlyLiu06,

AUTHOR={ J. Blanchet and P. Glynn and J. C. Liu},

title = {State-dependent Importance Sampling and large Deviations},

booktitle = {Proceedings of the 1st international conference on Performance evaluation methodolgies and tools},

series = {valuetools '06},

year = {2006},

isbn = {1-59593-504-5},

location = {Pisa, Italy},

articleno = {20},

url = {http://doi.acm.org/10.1145/1190095.1190120},

doi = {http://doi.acm.org/10.1145/1190095.1190120},

acmid = {1190120},

publisher = {ACM},

address = {New York, NY, USA},

keywords = {Importance sampling, Large Deviations, Random Walks, Rare-event Simulation},

}