General conditions for bounded relative error in simulations of highly reliable Markovian systems
Keyword(s):
We establish a necessary condition for any importance sampling scheme to give bounded relative error when estimating a performance measure of a highly reliable Markovian system. Also, a class of importance sampling methods is defined for which we prove a necessary and sufficient condition for bounded relative error for the performance measure estimator. This class of probability measures includes all of the currently existing failure biasing methods in the literature. Similar conditions for derivative estimators are established.
1996 ◽
Vol 28
(03)
◽
pp. 687-727
◽
2012 ◽
Vol 22
(06)
◽
pp. 1250052
◽
2019 ◽
Vol 33
(10)
◽
pp. 1950086
1988 ◽
Vol 11
(1)
◽
pp. 27-36
2009 ◽
Vol 86
(1)
◽
pp. 45-60
◽
2016 ◽
Vol 62
(3)
◽
pp. 267-272
2007 ◽
Vol 38
(3)
◽
pp. 267-276
◽
1990 ◽
Vol 33
(4)
◽
pp. 482-488
◽