On variance reducing multipliers for Monte Carlo integration (Q1395204)

The authors consider the evaluation of an absolutely convergent integral \(I=\int_Gf(x)p(x) dx\), where \(G\) is an \(n\)-dimensional domain and \(p(x)\) is a given probability density. The crude Monte Carlo method for evaluating \(I\) is the estimator \(\Theta_N=\frac 1N \sum_{i=1}^Nf(\xi_i)\), where \(\xi_1,\ldots,\xi_N\) are independent realizations of \(\xi\). Three well known variance reduction techniques are considered and their estimators compared with \({\Theta}_N\).