TREVR: A NEW APPROACH TO RADIATIVE TRANSFER IN ASTROPHYSICS SIMULATIONS

Abstract

In this thesis we present TREVR (Tree-based Reverse Ray Tracing), a general algo- rithm for computing the radiation field, including absorption, in astrophysical sim- ulations. TREVR is designed to handle large numbers of sources and absorbers; it is based on a tree data structure and is thus suited to codes that use trees for their gravity or hydrodynamics solvers (e.g. Adaptive Mesh Refinement). It achieves com- putational speed while maintaining a specified accuracy via controlled lowering of resolution of both sources and rays from each source. TREVR computes the radiation field in O(N log(N)) time without absorption and O (Nlog(N)log(N)) time with absorption. These claims are substantiated by mathematically predicting and testing the algorithm’s general scaling. The scalings arise from merging sources of radiation according to an opening angle criterion and walking the tree structure to trace a ray to a depth that gives the chosen accuracy for absorption. The absorption-depth refinement criterion is unique to TREVR and is presented here for the first time. We provide a suite of tests demonstrating the algorithm’s ability to accurately compute fluxes, ionization fronts and shadows. Two novel test cases are presented here for the first time as part of this suite.