Unbiased estimators of coincidence and correlation in non-analogous Monte Carlo particle transport

The conventional non-analogous Monte Carlo methods are optimized to preserve the mean value of the distributions. Therefore, they are not suited to non-Boltzmann problems such as the estimation of coincidences or correlations. This paper presents a general method called history splitting for the non-analogous estimation of such quantities. The basic principle of the method is that a non-analogous particle history can be interpreted as a collection of analogous histories with different weights according to the probability of their realization. Calculations with a simple Monte Carlo program for a pulse-height-type estimator prove that the method is feasible and provides unbiased estimation. Different variance reduction techniques have been tried with the method and Russian roulette turned out to be ineffective in high multiplicity systems. An alternative history control method is applied instead. Simulation results of an auto-correlation (Rossi-α) measurement show that even the reconstruction of the higher moments is possible with the history splitting method, which makes the simulation of neutron noise measurements feasible.