The onion universe: all sky lightcone simulations in spherical shells

Galaxy surveys provide a large-scale view of the universe that typically has a limited line of sight or redshift resolution. The lack of radial accuracy in these surveys can be modelled by picturing the universe as a set of concentric radial shells of finite width around the observer, i.e. an onion-like structure. We present a new N-body simulation with 2048³ particles developed at the Marenostrum supercomputer with the GADGET-2 code. Using the lightcone output we build a set of angular maps that mimic this onion-like representation of the universe. The onion maps are a highly compressed version of the raw data (i.e. a factor of >1000 smaller size for arcminute resolution maps) and they provide a new and powerful tool to exploit large-scale structure observations. We introduce two basic applications of these maps that are especially useful for constraining dark energy properties: the baryon acoustic oscillations (BAOs) in the galaxy power spectrum and all-sky maps of the weak lensing distortion. In particular, from the matter density maps, we determine the smallest scale where linear theory and the Gaussianity of the error analysis applies. Using the weak lensing maps, we measure the convergence power spectra and compare it to halo fit predictions. We also discuss mass resolution effects and error determinations. As a further application, we compute the variance and higher order moments of the maps. We show that sampling variance on scales of few degrees is quite large, resulting in a significant (25 per cent at 10 arcmin scales) bias in the variance. We caution that current lensing surveys such as the COSMOS HST should take into account this bias and extra sampling error in their clustering analyses and inferred cosmological parameter constraints. Finally, we test the importance of projection effects in the weak lensing mass reconstruction. On the mean, the mass calibration works well but it exhibits a large non-Gaussian scatter what could induce a large bias in the recovered mass function.