An implementation of radiative transfer in the cosmological simulation code GADGET

TL;DR

This paper introduces a new, efficient, and photon-conserving radiative transfer implementation within the GADGET cosmological simulation code, enabling on-the-fly modeling of reionisation alongside galaxy formation.

Contribution

It develops a novel anisotropic diffusion operator and an implicit solver for radiative transfer in SPH, integrated into GADGET for dynamic cosmological simulations.

Findings

Accurate tests with static and cosmological density fields

Good agreement with analytical solutions and other codes

Efficient on-the-fly integration during simulations

Abstract

We present a novel numerical implementation of radiative transfer in the cosmological smoothed particle hydrodynamics (SPH) simulation code {\small GADGET}. It is based on a fast, robust and photon-conserving integration scheme where the radiation transport problem is approximated in terms of moments of the transfer equation and by using a variable Eddington tensor as a closure relation, following the `OTVET'-suggestion of Gnedin & Abel. We derive a suitable anisotropic diffusion operator for use in the SPH discretization of the local photon transport, and we combine this with an implicit solver that guarantees robustness and photon conservation. This entails a matrix inversion problem of a huge, sparsely populated matrix that is distributed in memory in our parallel code. We solve this task iteratively with a conjugate gradient scheme. Finally, to model photon sink processes we consider ionisation and recombination processes of hydrogen, which is represented with a chemical network that is evolved with an implicit time integration scheme. We present several tests of our implementation, including single and multiple sources in static uniform density fields with and without temperature evolution, shadowing by a dense clump, and multiple sources in a static cosmological density field. All tests agree quite well with analytical computations or with predictions from other radiative transfer codes, except for shadowing. However, unlike most other radiative transfer codes presently in use for studying reionisation, our new method can be used on-the-fly during dynamical cosmological simulation, allowing simultaneous treatments of galaxy formation and the reionisation process of the Universe.