Variance-Reduction Methods for Monte Carlo Simulation of Radiation Transport
MetadataShow full item record
Monte Carlo simulationStatistical uncertaintiesVariance-reduction methodsSplitting and Russian rouletteAnt colony algorithmsInteraction forcingDelta scattering
García-Pareja S, Lallena AM and Salvat F (2021) Variance-Reduction Methods for Monte Carlo Simulation of Radiation Transport. Front. Phys. 9:718873. [doi: 10.3389/fphy.2021.718873]
SponsorshipFinancial support from the Spanish Ministerio de Ciencia, Innovación y Universidades/Agencia Estatal de Investigación/ European Regional Development Fund (ERDF) of European Union (projects nos. RTI2018-098117-B-C22 and PID2019- 104888GB-I00) and the Junta de Andalucía (projects nos. FQM387 and P18-RT-3237) is gratefully acknowledged.
After a brief description of the essentials of Monte Carlo simulation methods and the definition of simulation efficiency, the rationale for variance-reduction techniques is presented. Popular variance-reduction techniques applicable to Monte Carlo simulations of radiation transport are described and motivated. The focus is on those techniques that can be used with any transport code, irrespective of the strategies used to track charged particles; they operate by manipulating either the number and weights of the transported particles or the mean free paths of the various interaction mechanisms. The considered techniques are 1) splitting and Russian roulette, with the ant colony method as builder of importance maps, 2) exponential transform and interaction-forcing biasing, 3) Woodcock or delta-scattering method, 4) interaction forcing, and 5) proper use of symmetries and combinations of different techniques. Illustrative results from analog simulations (without recourse to variance-reduction) and from variance-reduced simulations of various transport problems are presented.