USING NORMALITY TESTS TO ASSESS THE VALIDITY OF MONTE CARLO CONFIDENCE INTERVALS

摘要

One means of indicating the uncertainty on Monte Carlo results is through the use of confidence intervals (CIs), the accurate determination of which requires a knowledge of the distribution of the random variates. The statistical procedures used by the MC21 Monte Carlo code is based on the assumption that the random variates-i.e., the tally values for each batch-are normally distributed. Tests have been implemented in the code to assess this assumption, and to warn the user if it is likely incorrect for a particular result.