MC/DC on GitHub Read the Docs JOSS Paper

Monte Carlo / Dynamic Code (MC/DC) is the primary deliverable for our center. It includes novel transport methods focused for transient neutron transport as well as an acceleration and abstraction software engineering scheme to make MC/DC a platform for rapid methods development in the felid. While MC/DC is where we deploy our novel algorithms and rapidly develop numerical methods for transient transport, it is also intended to live beyond the length of the center. It includes a testing library (unit, regression, verification, performance), continuous integration (via GitHub actions and chron jobs), issue tracking, a documentation site, and published PR and contribution processes.

CEMeNT research implemented in MC/DC

• Fully transient transport
• iQMC hybrid Monte Carlo iterative methods
• Embedded UQ
• Python acceleration and abstraction techniques for CPUs and GPUs (Nvidia and AMD)
• Asynchronous GPU scheduling
• Automatic transient weight window production
• Hash based random number generation for complete reproducibility
• Hybrid Woodcock-delta tracking methods
• Compressed sensing
• Multi-level Monte Carlo

Features

• Multigroup physics
  • Capture
  • Isotropic scattering
  • Fission (prompt and delayed)
• Continuous energy physics
  • NJOY generated point-wise data,
    1. Room temperature
    2. Assumed linear interpolation
  • Capture (MT=102-117)
  • Fission (prompt and delayed)
  • Scattering (non-capture & non-fission)
    1. Isotropic elastic scattering in COM
    2. Free gas, constant XS model for thermal scattering
  • Supports almost all nuclides
• Geometry
  • Surface-tracking
  • Woodcock-delta tracking
  • Quadric CSG surface
  • Multi-level lattice
  • Fully transient geometrical elements
  • Time dependent release of delayed neutron production drift
• Simulation modes
  • Fixed-source (time-dependent)
  • k-Eigenvalue
  • Iterative quasi-Monte Carlo
• Running modes: Python, Numba (CPU and GPU)
• Parallel support
  • MPI
  • GPU SIMT on Nvidia and AMD GPUs/APUs (via Harmonize)
  • Domain decomposition
  • Automatic node balancing
  • Reproducibility on all targets (via hash-based RNG seeding)