Sensitivity Analysis in Monte Carlo Simulation of Stochastic Activity Networks

Stochastic activity networks (SANs) such as those arising in Project Evaluation Review Technique (PERT) and Critical Path Method (CPM) are an important classical set of models in operations research. We focus on sensitivity analysis for stochastic activity networks when Monte Carlo simulation is employed. After a brief aside reminiscing on Saul’s influence on the author’s career and on the simulation community, we review previous research for sensitivity analysis of simulated SANs, give a brief summary overview of the main approaches in stochastic gradient estimation, derive estimators using techniques not previously applied to this setting, and address some new problems not previously considered. We conclude with some thoughts on future research directions.