Consistency of cosmic shear analyses in harmonic and real space

Abstract

Recent cosmic shear studies have reported discrepancies of up to 1 sigma on the parameter S-8 = sigma(8) root Omega(m)/0.3 between the analysis of shear power spectra and two-point correlation functions, derived from the same shear catalogues. It is not a priori clear whether the measured discrepancies are consistent with statistical fluctuations. In this paper, we investigate this issue in the context of the forthcoming analyses from the third year data of the Dark Energy Survey (DES Y3). We analyse DES Y3 mock catalogues from Gaussian simulations with a fast and accurate importance sampling pipeline. We show that the methodology for determining matching scale cuts in harmonic and real space is the key factor that contributes to the scatter between constraints derived from the two statistics. We compare the published scales cuts of the KiDS, Subaru-HSC, and DES surveys, and find that the correlation coefficients of posterior means range from over 80 per cent for our proposed cuts, down to 10 per cent for cuts used in the literature. We then study the interaction between scale cuts and systematic uncertainties arising from multiple sources: non-linear power spectrum, baryonic feedback, intrinsic alignments, uncertainties in the point spread function, and redshift distributions. We find that, given DES Y3 characteristics and proposed cuts, these uncertainties affect the two statistics similarly; the differential biases are below a third of the statistical uncertainty, with the largest biases arising from intrinsic alignment and baryonic feedback. While this work is aimed at DES Y3, the tools developed can be applied to Stage-IV surveys where statistical errors will be much smaller.