ICE-COLA: fast simulations for weak lensing observables

Abstract

Approximate methods to full N-body simulations provide a fast and accurate solution to the development of mock catalogues for the modeling of galaxy clustering observables. In this paper we extend ICE-COLA (Izard et al. 2016), based on an optimized implementation of the approximate COLA method, to produce weak lensing maps and halo catalogues in the light cone using an integrated and self consistent approach. We show that despite the approximate dynamics, the catalogues thus produced enable an accurate modeling of weak lensing observables one decade beyond the characteristic scale where the growth becomes non-linear. In particular, we compare ICE-COLA to the MICE-GC N-body simulation for some fiducial cases representative of upcoming surveys and find that, for sources at redshift $z=1$, their convergence power spectra agree to within one percent up to high multipoles (i.e., of order $1000$). The corresponding shear two point functions, $\xi_{+}$ and $\xi_{-}$, yield similar accuracy down to $2$ and $20$ arcmin respectively, while tangential shear around a $z=0.5$ lens sample is accurate down to $4$ arcmin. We show that such accuracy is stable against an increased angular resolution of the weak lensing maps. Hence, this opens the possibility of using approximate methods for the joint modeling of galaxy clustering and weak lensing observables and their covariance in ongoing and future galaxy surveys.