About

The intergalactic medium (IGM) - observed primarily in absorption in the spectra of luminous quasars - represents the largest repository of baryons in the Universe. It is an excellent probe of the thermal and ionisation state of intergalactic gas over a vast swathe of cosmic time, stretching all the way from the local Universe at z~0 to the epoch of reionisation at redshifts z>5. It is, furthermore, an unrivalled tracer of the structure of the cosmic web on scales of ~10-100 comoving kpc that is probed observationally by the Lyman-alpha forest. Over the last few years, our team of researchers have used the Sherwood simulation suite to address a wide range of open questions pertaining to reionisation, the IGM and the "small-scale crisis" of cold dark matter. However, a limitation of Sherwood was that it assumed a pre-computed, spatially homogeneous ultraviolet background for heating and ionising the baryons in the IGM. This approach neglects the large-scale fluctuations in the ionisation state of intergalactic gas expected approaching the reionisation epoch. This limits the utility of our simulations when interpreting the Lyman-alpha forest at z>5 in the context of reionisation models. It has also been argued that neglecting ionisation and thermal fluctuations in IGM simulations could weaken existing Lyman-alpha forest constraints on the free streaming length of dark matter, leaving the door open for alternatives to cold dark matter.

The Sherwood-Relics simulations redress this situation by including a self-consistent model for cosmological radiative transfer. This enables us to properly include the effect of the inhomogeneous nature of reionisation on the thermal and ionisation history of the IGM, to explore a wider range of reionisation histories, and to push our accurate modelling of Lyman-alpha forest data toward the reionisation era at z >5. This is crucial for further improving measurements of the free streaming of dark matter, as well as constraining the thermal state of the IGM. We achieve this through a combination of P-Gadget-3 cosmological hydrodynamical simulations and radiative transfer simulations performed with the GPU accelerated code ATON. A major advantage of our novel "hybrid radiation-hydrodynamics" approach is that we use well-calibrated models of patchy reionisation obtained with ATON to construct hydrodynamical simulations that are fully coupled to the thermal pressure associated with inhomogeneous IGM heating. This approach provides consistent IGM temperatures and pressure smoothing for these patchy reionisation models. With Sherwood-Relics, we are therefore able to accurately model the IGM both during and after reionisation at the mass resolution required for correctly capturing the properties of the Lyman-alpha forest.

The image above shows the neutral hydrogen density (top panels) and gas temperature (bottom panels) in one of the Sherwood-Relics simulations assuming either a homogeneous UV background (left, following the approach used in our earlier Sherwood project), a UV background obtained with cosmological radiative transfer applied in post-processing using ATON (middle), and the new hybrid radiation-hydrodynamics approach we have developed for Sherwood-Relics (right). The hybrid model has a similar structure of ionised and neutral regions compared to the post-processed model, but crucially it also accounts for the shock-heating of gas in high density regions, self-consistently captures the smoothing of the gas distribution by thermal pressure, and follows the advection of thermal energy in the gas flows.

The Sherwood-Relics simulations were performed with a modified version of the cosmological smoothed particle hydrodynamics code P-Gadget-3, developed by Volker Springel. P-Gadget-3 is an updated and extended version of the publicly available Gadget-2 code. The cosmological radiative transfer in Sherwood-Relics was followed the moment-based, M1-closure code ATON, developed by Dominique Aubert. The 50 million hours of supercomputer time used to run the Sherwood-Relics simulations was awarded through the PRACE (Partnership for Advanced Computing in Europe) 16th regular access call, and the 12th call of the UK's DiRAC (Distributed Research using Advanced Computing) facility. Funding was also provided by The Royal Society, the Science and Technology Facilities Council (STFC) and the European Research Council (ERC).