On improved estimation for importance sampling

David Firth
2011 Brazilian Journal of Probability and Statistics  
The standard estimator used in conjunction with importance sampling in Monte Carlo integration is unbiased but inefficient. An alternative estimator is discussed, based on the idea of a difference estimator, which is asymptotically optimal. The improved estimator uses the importance weight as a control variate, as previously studied by Hesterberg (Ph.D. Dissertation, Stanford University (1988); Technometrics 37 (1995) 185-194; Statistics and Computing 6 (1996) 147-157); it is routinely
more » ... and can deliver substantial additional variance reduction. Finite-sample performance is illustrated in a sequential testing example. Connections are made with methods from the survey-sampling literature.
doi:10.1214/11-bjps155 fatcat:qidihl7xxfcxhapkoe4kxfvmkq