Nuclear Science and Engineering / Volume 187 / Number 2 / August 2017 / Pages 142-153
Technical Paper / dx.doi.org/10.1080/00295639.2017.1312944
This work shows that summary statistics for fixed-source Monte Carlo problems sampled by batch are subject to more variability than are statistics sampled by individual histories. This is most pronounced when the number of batches used is small. Using individual histories, or a larger number of smaller batches, allows for better precision in estimating the variability of the underlying quantity being calculated when the histories are uncorrelated.