Disentangling dark energy and cosmic tests of gravity from weak lensing systematics

TL;DR

This paper analyzes how astrophysical and measurement systematics affect constraints on dark energy and gravity modifications from upcoming large-scale structure surveys, emphasizing the importance of modeling and cross-correlations.

Contribution

It demonstrates the impact of systematics on cosmological constraints and offers a fitting function for the matter power spectrum in modified gravity models.

Findings

Systematic uncertainties can reduce constraints by a factor of four.

Cross-correlations help recover constraining power.

Constraints are maintained if fewer than 36 parameters describe systematics.

Abstract

We consider the impact of key astrophysical and measurement systematics on constraints on dark energy and modifications to gravity on cosmic scales. We focus on upcoming photometric "Stage III" and "Stage IV" large scale structure surveys such as DES, SuMIRe, Euclid, LSST and WFIRST. We illustrate the different redshift dependencies of gravity modifications compared to intrinsic alignments, the main astrophysical systematic. The way in which systematic uncertainties, such as galaxy bias and intrinsic alignments, are modelled can change dark energy equation of state and modified gravity figures of merit by a factor of four. The inclusion of cross-correlations of cosmic shear and galaxy position measurements helps reduce the loss of constraining power from the lensing shear surveys. When forecasts for Planck CMB and Stage IV surveys are combined, constraints on the dark energy equation of state and modified gravity model are recovered, relative to those from shear data with no systematic uncertainties, if fewer than 36 free parameters in total are used to describe the galaxy bias and intrinsic alignment models as a function of scale and redshift. To facilitate future investigations, we also provide a fitting function for the matter power spectrum arising from the phenomenological modified gravity model we consider.