Application of the stochastic oscillator to assess source convergence in Monte Carlo criticality calculations

A novel method for assessing source convergence in Monte Carlo criticality calculations is presented here. The method is based on the stochastic oscillator, an indicator that is commonly used in the technical analysis of financial markets, and entails performing a posterior diagnostic test on the Shannon entropy of the source distribution. The stochastic oscillator takes advantage of the fact that when a scalar time series is increasing (or decreasing) monotonically, its value will be higher (or lower) than the previous values. Extensive testing on the OECD/NEA source convergence benchmark suite shows that the stochastic oscillator diagnostic performs very well. The relative merits of this method compared to previous approaches are discussed. (authors)