Automated Parallel Calculation of Collaborative Statistical Models
Large scale statistical data analysis in particle physics
The project leader’s team developed a fast Monte Carlo simulation program for the interaction of low-energetic charged particles with matter. There is great interest for this code, from industry and science. The code, however, is developed and optimized for the team’s own special purpose, which is the simulation of electron images of lithographically defined lines in resist and the shot noise limited resolution of the lithography process.
The code reliably predicts secondary electron yields, as compared to experimental values. To enable third parties to use the code the team proposes to transfer it to an open source platform, in use by a large community. The proposed platform is Geant4, a well-known and worldwide collaboration of scientists including CERN, Geneva, that study the passage of high-energy particles through matter. Geant4 so far did not contain a low-energy module for charged particle scattering. The challenge of the project is to implement the team’s home-built geometry definition into Geant4, the fast tracking of electrons, and improve the physical modeling. Efficient computing will be needed and, possibly, a hybrid shared-memory/ distributed-memory type of parallelization and/or GPU-based vector processing.