Two new Monte Carlo methods for point flux estimation Original Research Article

Considering Monte Carlo methods for point flux estimation, we share the opinion of Lux and Koblinger that the problem is not yet solved satisfactorily [7]. The intention of this article is to proceed in this area. We present two new methods, both based on non-analog sampling schemes in which the point of interest is explicitly considered. As a result, a simulated particle preferably moves to the point, and we consider this as an important feature of a useful method. Both methods generally yield partial scores for which an order-dependent upper bound can be given for physically relevant cases. Furthermore, one of our methods (detector emphasis technique) yields for a special case bounded partial scores. Together with a technical condition this is sufficient for all moments to exist, and we therefore think that this case is treated satisfactorily by this method. We finally investigate a simple model, comparing our new methods with established ones. Both methods clearly yield higher performances and more reliable error estimations, but further technical means (weight windows) are necessary to compensate for the drawbacks of non-analog sampling.