Inferring host-galaxy properties of LIGO-Virgo-KAGRA's black holes

Observations of gravitational waves from binary black hole (BBH) mergers have measured the redshift evolution of the BBH merger rate. The number density of galaxies in the Universe evolves differently with redshift based on their physical properties, such as their stellar masses and star formation rates. In this work we show that the measured population-level redshift distribution of BBHs sheds light on the properties of their probable host-galaxies. We first assume that the hosts of BBHs can be described by a mixture model of galaxies weighted by stellar mass or star formation rate, and find that we can place upper limits on the fraction of mergers coming from a stellar mass weighted sample of galaxies. We then constrain parameters of a physically motivated power-law delay-time distribution using GWTC-3 data, and self-consistently track galaxies in the \textsc{UniverseMachine} simulations with this delay-time model to infer the probable host-galaxies of BBHs over a range of redshifts. We find that the inferred host-galaxy distribution at redshift $z=0.21$ has a median star formation rate $\sim 0.9\,M_\odot\mathrm{yr}^{-1}$ and a median stellar mass of $\sim 1.9 \times 10^{10}\,M_\odot$. We also provide distributions for the mean stellar age, halo mass, halo radius, peculiar velocity, and large scale bias associated with the host-galaxies, as well as their absolute magnitudes in the B- and ${ \rm K_s}$-bands. Our results can be used to design optimal electromagnetic follow-up strategies for BBHs, and also to aid the measurement of cosmological parameters using the statistical dark siren method.