Title: Atomic Hydrogen Shows its True Colours: Correlations between HI and Galaxy Colour in Simulations

Abstract: Intensity mapping experiments are beginning to measure the spatial distribution of neutral atomic hydrogen (HI) to constrain cosmological parameters and the large-scale distribution of matter. However, models of the behaviour of HI as a tracer of matter are complicated by galaxy evolution. In this work, we examine the clustering of HI in relation to galaxy colour, stellar mass, and HI mass in IllustrisTNG at $z$ = 0, 0.5, and 1. We compare the HI-red and HI-blue galaxy cross-power spectra, finding that HI-red has an amplitude 1.5 times higher than HI-blue at large scales. The cross-power spectra intersect at $\approx 3$ Mpc in real space and $\approx 10$ Mpc in redshift space, consistent with $z \approx 0$ observations. We show that HI clustering increases with galaxy HI mass and depends weakly on detection limits in the range $M_{\mathrm{HI}} \leq 10^8 M_\odot$. In terms of $M_\star$, we find blue galaxies in the greatest stellar mass bin cluster more than blue galaxies in other stellar mass bins. Red galaxies in the greatest stellar mass bin, however, cluster the weakest amongst red galaxies. These trends arise due to central-satellite compositions. Centrals correlate less with HI for increasing stellar mass, whereas satellites correlate more, irrespective of colour. Despite the clustering relationships with stellar mass, we find that the cross-power spectra are largely insensitive to detection limits in HI and galaxy surveys. Counter-intuitively, all auto and cross-power spectra for red and blue galaxies and HI decrease with time at all scales in IllustrisTNG. We demonstrate that processes associated with quenching contribute to this trend. The complex interplay between HI and galaxies underscores the importance of understanding baryonic effects when interpreting the large-scale clustering of HI, blue, and red galaxies at $z \leq 1$.