Clustering-based halo mass assignment for high-redshift galaxies: method, validation, and application to JWST

Abstract

We present a clustering-based method for inferring halo masses of high-redshift galaxies, validated using the COLIBRE simulations combined with the HOMA empirical model, which assumes star formation to be proportional to halo accretion rate and is calibrated to JWST observations. Our approach matches the two-point correlation function of galaxies in stellar mass bins to reference halo clustering, establishing the stellar-to-halo mass relation over an optimal radial range 0.5 < r_p/cMpc < 1.0. Validation against true halo masses shows minimal bias, with scatter below 0.3 dex for volumes down to (50 cMpc)^3 and redshifts to z=12, even for photometric data. Cross-validation using the native COLIBRE population shows that the method is robust to the different galaxy-halo prescriptions. Survey volume dominates the error budget: field-to-field variations in the clustering amplitude vary by factors of 3 (2) for photometric (spectroscopic) samples, translating to 0.3-0.5 dex uncertainty in halo masses. Splitting the sample by properties such as SFR, colour, and age mitigates assembly bias, offering a key advantage over abundance matching. Application to JWST (JADES) samples at z=6 and 10 yields halo masses of log M_h/M_sun = 10.52_{-0.21}^{+0.12} and 9.91_{-0.34}^{+0.19} for M_UV < -17 galaxies, with linear biases of b_h = 4.2_{-0.41}^{+0.26} and 7.7_{-1.16}^{+0.77}, respectively. Current data cannot distinguish between star formation models. Our Roman Deep Tier forecasts indicate that at z=10, the inferred halo masses for M_UV < -21 galaxies differ by 0.5 dex between bursty and non-bursty models, but the expected number of pairs limits the constraining power. Our framework provides robust, empirically-calibrated halo masses essential for interpreting JWST observations and constraining galaxy formation during reionization.