Realistic systematic biases induced by residual intrinsic alignments in cosmic shear surveys

Abstract

We study the parameter estimation bias induced by intrinsic alignments on a Euclid-like weak lensing survey. For the intrinsic alignment signal we assume a composite alignment model for elliptical and spiral galaxies using tidal shearing and tidal torquing as the alignment generating mechanism, respectively. The parameter estimation bias is carried out analytically with a Gaussian bias model and through running Monte-Carlo-Markov-chains on synthetic data including the alignment signal with a likelihood only including the cosmic shear signal. In particular, we study the impact of $II$ and $GI$ alignment terms individually as well as the more realistic situation where both types of alignment are present, and investigate the scaling of the estimation biases with varying strength of the alignment signal. Our results show that intrinsic alignments can cause substantial biases in cosmological parameters even if the coupling of galaxies to the ambient large is small. Especially $GI$-contributions strongly bias key cosmological parameters such as the dark energy equation of state. We also correct the analytic expression for the Gaussian bias model and find that the biases induced by intrinsic alignments are not accurately recovered by the simple analytic model.