Ray-tracing 3D dust radiative transfer with DART-Ray: code upgrade and public release

TL;DR

The paper introduces an extensively upgraded version of the DART-Ray 3D dust radiative transfer code, featuring optimizations, parallelization, dust self-heating, and new output capabilities, making it more efficient and versatile for astrophysical modeling.

Contribution

The paper presents a major upgrade to the DART-Ray code, including optimizations, parallelization, dust self-heating, and new output formats, with public release and adaptable input tools.

Findings

Optimized ray-angular density and source function for efficiency.

Implemented hybrid MPI+OpenMP parallelization schemes.

Validated with benchmark models including dust self-heating effects.

Abstract

We present an extensively updated version of the purely ray-tracing 3D dust radiation transfer code DART-Ray. The new version includes five major upgrades : 1) a series of optimizations for the ray-angular density and the scattered radiation source function; 2) the implementation of several data and task parallelizations using hybrid MPI+OpenMP schemes; 3) the inclusion of dust self-heating; 4) the ability to produce surface brightness maps for observers within the models in HEALPix format; 5) the possibility to set the expected numerical accuracy already at the start of the calculation. We tested the updated code with benchmark models where the dust self-heating is not negligible. Furthermore, we performed a study of the extent of the source influence volumes, using galaxy models, which are critical in determining the efficiency of the DART-Ray algorithm. The new code is publicly available, documented for both users and developers, and accompanied by several programmes to create input grids for different model geometries and to import the results of N-body and SPH simulations. These programmes can be easily adapted to different input geometries, and for different dust models or stellar emission libraries.