Impact of baryonic feedback on HSC Y1 weak lensing non-Gaussian statistics

Abstract: Baryonic feedback is a major systematic in weak lensing cosmology. Its most studied effect is the suppression of the lensing power spectrum, a second-order statistic, on small scales. Motivated by the growing interest in statistics beyond the second order, we investigate the effect of baryons on lensing non-Gaussian statistics and the resulting biases in the matter clustering amplitude $S_8 = \sigma_8\sqrt{\Omega_m/0.3}$. We focus on the Subaru Hyper Suprime-Cam Year 1 (HSC-Y1) data which, with its high source number density, closely resembles those expected from the upcoming Euclid and Rubin LSST. We study four non-Gaussian statistics -- peak counts, minimum counts, the probability distribution function, and the scattering transform -- in addition to the usual power spectrum. We first estimate the biases in $S_8$ using mock observations built from the IllustrisTNG and BAHAMAS hydrodynamical simulations and theoretical models built from dark matter-only simulations. We find up to $1\sigma$ bias in $S_8$ when the smallest scales (2 arcmin) and the highest feedback level are considered. We then analyze the HSC-Y1 data and compare the $S_8$ obtained for each statistic with different smoothing scales or scale cuts. As we expect that baryons mostly affect the small scales, comparing the results obtained from including and excluding small scales can indicate the level of impact from baryons. With HSC data, we find only minor ($\leq0.5\sigma$) differences in $S_8$ for all statistics, even when considering very small scales (2 arcmin). Our results suggest that the effect of baryons is insignificant at the level of HSC-Y1 down to 2~arcmin for all statistics examined here, or it is canceled by other scale-dependent systematics.