A new tool for modeling cosmic ray emissions
The CROMA tool uses GPU acceleration to model how cosmic rays behave in space.
Researchers created CROMA, a new solver for modeling non-thermal emission. It uses GPU acceleration to process cosmological magnetohydrodynamical (MHD) simulations.
The code tracks cosmic ray proton and electron spectra along Lagrangian tracer trajectories. It includes processes like re-acceleration and secondary-particle production.
The tool solves time-dependent Fokker-Planck (FP) equations. These equations model the spectra of cosmic rays including cooling and hadronic interactions.
The solver uses the Chang-Cooper finite-difference scheme. This reduces the FP equation for each tracer into an independent tridiagonal system.
The system uses a CUDA implementation of the parallel cyclic reduction (PCR) algorithm on NVIDIA GPUs. This increased throughput by a factor of 10 on the Leonardo Booster system.
This fast solver helps with high-resolution simulations. It can be used to compare results with next-generation radio observations and multi-messenger applications.
Key facts
- CROMA increases per-node throughput by a factor of 10 on the Leonardo Booster system.
- The tool models cosmic ray proton and electron spectra using Lagrangian tracer trajectories.
- The solver uses a CUDA implementation of the parallel cyclic reduction (PCR) algorithm.