Applying Geant4's Importance Biasing to improve the efficiency of SuperCDMS background simulations

Abstract: Experiments searching for extremely rare events surround their sensitive detectors with several layers of different shielding materials to protect them from external radiation and to achieve their low-background requirements to be able to observe a potential signal. Standard Monte Carlo simulations that propagate particles through the thick shielding, usually do not penetrate the shield in sufficient numbers to properly model the external background, which is crucial for understanding the experiment's background composition. Geant4 is a widely used toolkit to simulate the passage of particles through matter and it offers various biasing techniques, among them being importance biasing, which has been intensively explored for application in background simulations for the SuperCDMS experiment. In this article, the basic working principle of importance biasing is explained. Furthermore, we provide guidance for developers for their own implementation of a biasing scheme. A new track property, the "biasing index", is introduced to allow different track topologies to be distinguished. Validation studies and optimal parameters for biasing gammas and neutrons are presented and caveats are discussed. In this work, simulations run with importance biasing achieved an efficiency boost of about $\mathcal{O}(10^4)$ for gammas and up to 500 for neutrons. By applying these techniques, we show that energy distributions simulated with and without importance biasing are consistent with each other within statistical uncertainty at a fraction of the consumed computing time.