Abstract
Using synthetic Lyman-$\alpha$ (Ly$\alpha$) forest data from the Dark Energy Spectroscopic Instrument (DESI) survey, we present a study of the impact of errors in quasar redshift estimation on Lyman-$\alpha$ correlation functions. Quasar redshift estimates suffer from large uncertainties of several hundred $\mathrm{km,s^{-1}}$ due to the broadness of emission lines and intrinsic velocity offsets among different lines. We inject Gaussian random redshift errors into mock quasar catalogues and measure both the Ly$\alpha$ auto-correlation and the Ly$\alpha$–quasar cross-correlation functions. We find that redshift errors lead to a smearing of the baryon acoustic oscillation (BAO) feature in the radial direction, while shifts in the peak position remain negligible. However, we observe a significant unphysical correlation at small transverse separations, whose amplitude increases with the size of the redshift errors. We interpret this contamination as arising from the broadening of emission lines in the estimated mean quasar continuum induced by redshift errors, combined with the unrealistically strong small-scale clustering of quasars in the simulations.