A new scale in the bias expansion

The fact that the spatial nonlocality of galaxy formation is controlled by some short length scale like the Lagrangian radius is the cornerstone of the bias expansion for large-scale-structure tracers. However, the first sources of ionizing radiation between $z\approx 15$ and $z\approx 6$ are expected to have significant effects on the formation of galaxies we observe at lower redshift, at least on low-mass galaxies. These radiative-transfer effects introduce a new scale in the clustering of galaxies, i.e. the finite distance which ionizing radiation travels until it reaches a given galaxy. This mean free path can be very large, of order $100\,h^{-1}\,{\rm Mpc}$. Consequently, higher-derivative terms in the bias expansion could turn out to be non-negligible even on these scales: treating them perturbatively would lead to a massive loss in predictivity and, for example, could spoil the determination of the BAO feature or constraints on the neutrino mass. Here, we investigate under what assumptions an explicit non-perturbative model of radiative-transfer effects can maintain the robustness of large-scale galaxy clustering as a cosmological probe.