Optimal Importance Sampling Densities for Estimating Several Expectations at Once

Simulation studies often estimate several quantities from a single sample, and the best importance sampling density depends on how the errors on these quantities are combined. We characterise the optimal auxiliary density for a general error criterion. We show that it is the nominal density reweighted by a weighted length of the vector of estimated functions, where the weights are set by the error criterion, or by the curvature of the loss when the criterion comes from a loss on the errors. We also provide a lower bound on the best achievable error. The optimal density is explicit for smooth losses, including the usual losses on covariance matrices. We also derive allocation laws for the cells of a histogram and characterise the minimax density for the probabilities of several events. When there is no closed form, we show how to compute the optimal density numerically.

Authors

Publication Details

Journal
arXiv (Cornell University)
Published
2026-11-16
DOI
https://doi.org/10.48550/arxiv.2611.20831
Primary Topic
Markov Chains and Monte Carlo Methods
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Optimal Importance Sampling Densities for Estimating Several Expectations at Once

Jérôme Morio, Jason Beh
arXiv (Cornell University)
Markov Chains and Monte Carlo Methods
preprint

Optimal Importance Sampling Densities for Estimating Several Expectations at Once

Jérôme Morio, Jason Beh
preprint en

Abstract

Simulation studies often estimate several quantities from a single sample, and the best importance sampling density depends on how the errors on these quantities are combined. We characterise the optimal auxiliary density for a general error criterion. We show that it is the nominal density reweighted by a weighted length of the vector of estimated functions, where the weights are set by the error criterion, or by the curvature of the loss when the criterion comes from a loss on the errors. We also provide a lower bound on the best achievable error. The optimal density is explicit for smooth losses, including the usual losses on covariance matrices. We also derive allocation laws for the cells of a histogram and characterise the minimax density for the probabilities of several events. When there is no closed form, we show how to compute the optimal density numerically.

arXiv (Cornell University)
Markov Chains and Monte Carlo Methods
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon