The OpenMOC method of characteristics neutral particle transport code

• An open source method of characteristics neutron transport code has been developed.
• OpenMOC shows nearly perfect scaling on CPUs and 30× speedup on GPUs.
• Nonlinear acceleration techniques demonstrate a 40× reduction in source iterations.
• OpenMOC uses modern software design principles within a C++ and Python framework.
• Validation with respect to the C5G7 and LRA benchmarks is presented.

The method of characteristics (MOC) is a numerical integration technique for partial differential equations, and has seen widespread use for reactor physics lattice calculations. The exponential growth in computing power has finally brought the possibility for high-fidelity full core MOC calculations within reach. The OpenMOC code is being developed at the Massachusetts Institute of Technology to investigate algorithmic acceleration techniques and parallel algorithms for MOC. OpenMOC is a free, open source code written using modern software languages such as C/C++ and CUDA with an emphasis on extensible design principles for code developers and an easy to use Python interface for code users. The present work describes the OpenMOC code and illustrates its ability to model large problems accurately and efficiently.