Variance Reduction in the Presence of Initial-Condition Bias

Abstract

We consider the application of antithetic variates and control variates variance reduction techniques (VRTs) in the context of steady-state simulation experiments when initial-condition bias is present. We show that, by appropriately modifying the experiment design, incorporating a VRT can improve both point and interval-estimator performance. Guidelines for modifying the experiment design are given.

Handled by the Department of Simulation.

Notes

Handled by the Department of Simulation.