Forecasting a Cosmological 21 cm Stacking Detection with Early HIRAX Data

Abstract

Post-reionization 21 cm intensity mapping provides a powerful technique for efficiently mapping the Universe on large scales. The Hydrogen Intensity and Real-time Analysis eXperiment (HIRAX) is a compact and highly redundant radio interferometer designed to exploit this technique. We present a forward modelling pipeline for forecasting a statistical detection of the cosmological 21 cm signal with HIRAX through cross-correlation with an external source catalogue. We simulate the cosmological signal using the halo model and propagate it, together with Galactic and extragalactic foreground emission and instrumental noise, through a model of an early-stage 36-dish HIRAX array configuration. We generate simulated visibilities, apply delay-domain foreground filtering, and construct cross-correlation-ready ringmaps for an observing time of 120 days with a single zenith pointing over the frequency range from 700 to 800 MHz, covering approximately 3547 deg$^2$ of the sky. We investigate the prospects for an early statistical signal detection by stacking these ringmaps at the known three-dimensional positions of sources from an overlapping external survey. Specifically, we quantify how delay-domain foreground filtering, the number of stacking sources, catalogue completeness, and source redshift uncertainties affect the detection significance. We find that redshift precision is the most critical requirement: while spectroscopic redshifts with $σ_z / (1+z) \sim 0.001$ preserve a significant stacking signal even under realistic foreground filtering and substantial catalogue incompleteness, photometric redshift uncertainties of $σ_z / (1+z) \sim 0.01$ strongly suppress the signal for realistic delay cutoffs. Applying the framework to a Euclid-like spectroscopic galaxy sample, we recover a minimum signal-to-noise ratio of 11 for our most conservative delay cutoff of 100 ns.