Improved predictive sampling using quasi-Monte Carlo with application to neutron-cross-section evaluation

In a recent T-16 study, directly measured neutron cross sections for 239Pu with about 1.5% standard error have been combined with a highly accurate criticality measurement (0.2%) to improve our knowledge of the cross sections. I demonstrate the usefulness of generating 'quasi-random' point sets using Centroidal Voronoi Tessellation for the purpose of sampling the posterior distribution to predict uncertainties in subsequent criticality simulations.