Bounding the variance in Monte Carlo experiments (Q1197912)

This paper describes a method or obtaining a worse-case bound on the variance in a Monte Carlo experiment by using available lower and upper bounds on outcomes within disjoint strata in the corresponding sample space. This bound enables the potential user ofor obtaining a worse-case bound on the variance in a Monte Carlo experiment by using available lower and upper bounds on outcomes within disjoint strata in the corresponding sample space. This bound enables the potential user of the Monte Carlo method to determine an upper bound on how long to run the experiment to ensure the desired accuracy. Moreover, a naive upper bound which is derivable without resorting to mathematical programming is presented. An example representing a stochastic low network with 10 nodes and 25 arcs is given that illustrates the considerably tighter upper bound that the proposal oow network with 10 nodes and 25 arcs is given that illustrates the considerably tighter upper bound that the proposal of the paper offers when compared to the naive upper bound.