Hybrid Radiation Hydrodynamics scheme with gravity tree-based adaptive optimization algorithm
Cheryl S. C. Lau, Maya A. Petkova, Ian A. Bonnell
arXiv:2404.17084v1 Announce Type: new
Abstract: Modelling the interaction between ionizing photons emitted from massive stars and their environment is essential to further our understanding of galactic ecosystems. We present a hybrid Radiation-Hydrodynamics (RHD) scheme that couples an SPH code to a grid-based Monte Carlo Radiative Transfer code. The coupling is achieved by using the particle positions as generating sites for a Voronoi grid, and applying a precise mapping of particle-interpolated densities onto the grid cells that ensures mass conservation. The mapping, however, can be computationally infeasible for large numbers of particles. We introduce our tree-based algorithm for optimizing coupled RHD codes. Astrophysical SPH codes typically utilize tree-building procedures to sort particles into hierarchical groups (referred to as nodes) for evaluating self-gravity. Our algorithm adaptively walks the gravity tree and transforms the extracted nodes into pseudo-SPH particles, which we use for the grid construction and mapping. This method allows for the temporary reduction of fluid resolution in regions that are less affected by the radiation. A neighbour-finding scheme is implemented to aid our smoothing length solver for nodes. We show that the use of pseudo-particles produces equally accurate results that agree with benchmarks, and achieves a speed-up that scales with the reduction in the final number of particle-cell pairs being mapped.arXiv:2404.17084v1 Announce Type: new
Abstract: Modelling the interaction between ionizing photons emitted from massive stars and their environment is essential to further our understanding of galactic ecosystems. We present a hybrid Radiation-Hydrodynamics (RHD) scheme that couples an SPH code to a grid-based Monte Carlo Radiative Transfer code. The coupling is achieved by using the particle positions as generating sites for a Voronoi grid, and applying a precise mapping of particle-interpolated densities onto the grid cells that ensures mass conservation. The mapping, however, can be computationally infeasible for large numbers of particles. We introduce our tree-based algorithm for optimizing coupled RHD codes. Astrophysical SPH codes typically utilize tree-building procedures to sort particles into hierarchical groups (referred to as nodes) for evaluating self-gravity. Our algorithm adaptively walks the gravity tree and transforms the extracted nodes into pseudo-SPH particles, which we use for the grid construction and mapping. This method allows for the temporary reduction of fluid resolution in regions that are less affected by the radiation. A neighbour-finding scheme is implemented to aid our smoothing length solver for nodes. We show that the use of pseudo-particles produces equally accurate results that agree with benchmarks, and achieves a speed-up that scales with the reduction in the final number of particle-cell pairs being mapped.