Branchless Collisions for Reducing Spatial Correlations in Continuous Energy Monte Carlo Power Iteration

Abstract

Branchless collisions are customarily used in time-dependent Monte Carlo simulations and have been recently introduced in Monte Carlo power iteration. Previous works have shown that branchless collisions are very efficient in quenching spatial correlations in a multi-group framework, and a preliminary work was previously conducted in order to extend the use of branchless collisions to continuous energy. In that work, two variants of branchless collisions were introduced: branchless on the material, and branchless on the isotope. In this work, we analyse in depth the effect of using branchless collisions in realistic systems on spatial correlations, and we show that performing branchless collisions on the material consistently results in a smaller effect of spatial correlations than when using a standard branching collision algorithm.