Monte Carlo has been used for radiation and particle transport since the first gener- ation of digital computers. The Geant family is the mainstay of High Energy physics experiments over the last 30+ years; Geant4 is the backbone of the simulation of the LHC experiments at CERN, used to discover the Higgs Boson in 2012. It has been incorporated into innovative tools for medical research in imaging and radiotherapy, in the simulation of the effects of radiation on spacecraft and many other application areas.
We give an overview of some of the underpinnings of Geant4, with emphasis on its use of pseudorandom number generators, including the recent advances including the inclusion of the MIXMAX generator, and the modeling of geometry in a common geometry engine, VecGeom.
We also discuss the future of detector simulation, including the advanced require- ments for pseudo-random number generations the GeantV project, which seeks to use CPU’s caches and vector instructions (or accelerators including GPU and Xeon Phis) to obtain a large increase in computing performance. GeantV requires vec- torised implementation and improvements in geometry, physics processes and pseu- dorandom number generation. We discuss the additional requirements for PRNGs coming from the need to obtain the same simulated tracks even when the order in which particles are processed can change in a multi-threaded, fine-grain parallel pro- gram. The needs for very large periods, splitting and excellent statistical properties, which are driving forces for the continued development of MIXMAX are detailed.