Predicting the 21-cm field with a hybrid effective field theory approach

A detection of the 21-cm signal can provide a unique window of opportunity for uncovering complex astrophysical phenomena at the epoch of reionization and placing constraints on cosmology at high redshifts, which are usually elusive to large-scale structure surveys. In this work, we provide a theoretical model based on a quadratic bias expansion capable of recovering the 21-cm power spectrum with high accuracy sufficient for upcoming ground-based radio interferometer experiments. In particular, we develop a hybrid effective field theory model in redshift space that leverages the accuracy of $N$-body simulations with the predictive power of analytical bias expansion models, and test it against the Thesan suite of radiative transfer hydrodynamical simulations. We make predictions of the 21-cm brightness temperature field at several distinct redshifts, ranging between $z=6.5$ and 11, thus probing a large fraction of the reionization history of the Universe (${x}_{\mathrm{HI}}=0.3\ensuremath{\sim}0.9$), and compare our model to the ``true'' 21-cm brightness in terms of the correlation coefficient, power spectrum and modeling error. In practice, we use the neutral hydrogen field as a proxy for the 21-cm emission field, which in the $z\ensuremath{\lesssim}10$ redshift range and for standard reionization is an excellent approximation. We find percent-level agreement at large and intermediate scales, $k\ensuremath{\lesssim}0.5h/\mathrm{Mpc}$, and favorable behavior down to small scales, $k\ensuremath{\sim}1h/\mathrm{Mpc}$, outperforming pure perturbation-theory-based models. To put our findings into context, we show that even in the absence of any foreground contamination the thermal noise of a futuristic HERA-like experiment is comparable with the theoretical uncertainty in our model in the allowed ``wedge'' of observations, providing further evidence in support of using hybrid-effective-field-theory-based models to approximate a range of cosmological observables.