Effect of baryonic feedback on two- and three-point shear statistics: prospects for detection and improved modelling

Accurate knowledge of the effect of feedback from galaxy formation on the matter distribution is a key requirement for future weak lensing experiments. Recent studies using hydrodynamic simulations have shown that different baryonic feedback scenarios lead to significantly different two-point shear statistics. In this paper we extend earlier work to three-point shear statistics. We show that, relative to the predictions of dark matter only models, the amplitude of the signal can be reduced by as much as 30-40% on scales of a few arcminutes. We find that baryonic feedback may affect two- and three-point shear statistics differently and demonstrate that this can be used to assess the fidelity of various feedback models. In particular, upcoming surveys such as Euclid might be able to discriminate between different feedback models by measuring both second- and third-order statistics. Because it will likely remain impossible to predict baryonic feedback with high accuracy from first principles, we argue in favour of phenomenological models that can capture the relevant effects of baryonic feedback processes in addition to changes in cosmology. We construct such a model by modifying the dark matter-only halo model to characterise the generic effects of energetic feedback using a small number of parameters. We use this model to perform a likelihood analysis in a simplified case in which two- and three-point shear statistics are measured between 0.5 and 20 arcmin and in which the amplitude of fluctuations, sigma8, the matter density parameter, Om, and the dark energy parameter, w0, are the only unknown free parameters. We demonstrate that for weak lensing surveys such as Euclid, marginalising over the feedbac parameters describing the effects of baryonic processes, such as outflows driven by feedback from star formation and AGN, may be able to mitigate the bias affecting Om, sigma8 and w0.