The low redshift Lyman-$\alpha$ Forest as a constraint for models of AGN feedback


Submitted on 20 April 2022


We study the low redshift Lyman-$\alpha$ Forest in the Illustris and IllustrisTNG (TNG) cosmological simulations to demonstrate their utility in constraining aspects of sub-grid models of feedback from active galactic nuclei (AGN). The two simulations share an identical Ultraviolet Background prescription and similar cosmological parameters, but TNG features an entirely reworked AGN feedback model. Therefore a comparison of these simulations is useful to assess the effects of an altered AGN sub-grid model on the low redshift Lyman-$\alpha$ Forest. We find significant differences in the IGM temperature-density relation between the two simulations due to changes in the gas heating rate due to AGN. We investigate Lyman-$\alpha$ Forest observables such as the column density distribution function, flux PDF, and Doppler width ($b$-parameter) distribution. Due to the AGN radio mode model, the original Illustris simulations have a factor of 2-3 fewer absorbers than TNG at column densities $N_{\rm HI}< 10^{15.5}$ cm$^{-2}$. We show that TNG is in much better agreement with the observed $z=0.1$ flux power spectrum than Illustris. The differences in the amplitude and shape of the flux PDF and power spectrum between Illustris and TNG cannot be attributed to simple changes in the photoheating rate. We also compare the simulated Forest statistics to UV data from the Cosmic Origins Spectrograph (COS) and find that neither simulation can reproduce the slope of the absorber distribution. Both Illustris and TNG also produce significantly smaller $b$-parameter distributions than observed in the COS data, possibly due to unresolved or missing sources of turbulence.


Comment: Submitted to ApJL, comments welcome

Subjects: Astrophysics - Astrophysics of Galaxies; Astrophysics - Cosmology and Nongalactic Astrophysics