Nuclear Science and Engineering / Volume 72 / Number 1 / October 1979 / Pages 108-110
Technical Note / dx.doi.org/10.13182/NSE79-A19313
Articles are hosted by Taylor and Francis Online.
The problem of analyzing the variance in a Monte Carlo calculation is addressed. The “batch” analysis method is considered, where one uses the average of contributions of k particles as an independent random variable, versus the “one-particle” analysis method, where k = 1. It is shown through general statistical considerations that the “one-particle” method yields more reliable estimates of the variance.