On cosmological bias due to the magnification of shear and position samples in modern weak lensing analyses

TL;DR

This paper demonstrates that magnification effects in galaxy surveys significantly bias cosmological parameter estimates in weak lensing analyses, emphasizing the need to model magnification accurately.

Contribution

It emulates magnification effects on key observables using simulations and assesses their impact on cosmological inference in Stage-IV-like surveys.

Findings

Magnification causes >1σ biases in key cosmological parameters.

Significant biases are found in cosmic shear and 3x2pt analyses due to magnification.

Magnification affects the redshift distribution mean, which could help identify contamination.

Abstract

The magnification of galaxies in modern galaxy surveys induces additional correlations in the cosmic shear, galaxy-galaxy lensing and clustering observables used in modern lensing "3x2pt" analyses, due to sample selection. In this paper, we emulate the magnification contribution to all three observables utilising the SLICS simulations suite, and test the sensitivity of the cosmological model, galaxy bias and redshift distribution calibration to un-modelled magnification in a Stage-IV-like survey using Monte-Carlo sampling. We find that magnification cannot be ignored in any single or combined observable, with magnification inducing $>1\sigma$ biases in the $w_0-\sigma_8$ plane, including for cosmic shear and 3x2pt analyses. Significant cosmological biases exist in the 3x2pt and cosmic shear from magnification of the shear sample alone. We show that magnification induces significant biases in the mean of the redshift distribution where a position sample is analysed, which may potentially be used to identify contamination by magnification.