Bounded next-event estimation and Monte Carlo forward-adjoint correlated coupling

摘要

Bounded next-event estimation coupling of forward and adjoint Monte Carlo histories is possible by biasing the track-length selection of either the forward or adjoint calculation such that the 1/r(2) term is eliminated from the estimator. This method is analytic, involving only elementary functions and minimal computer resources. A first-adjoint-collision response distribution from the detector can be created from the general forward-adjoint coupling procedure for use with the standard forward next-event point estimator with no 1/r(2) term. This estimation is applicable for both a point detector or a point sampled from a finite detector volume. The truncated, first-adjoint-collision version of the general coupling method requires no actual adjoint calculation, but adjoint scattering probabilities must be made available in the forward estimation procedure. Various aspects of the estimator are investigated, and some simple calculational comparisons with standard methods are presented.