Producing simulated catalogues for next generation galaxy surveys

Abstract

Current and future galaxy surveys will be able to map the large-scale structure of the Universe with unprecedented detail and measure cosmological parameters with exquisite precision. In order to develop the science cases and the analysis pipelines, it is necessary an accurate modelling of the non-linear gravitational evolution. This thesis presents a methodology for producing accurate mock catalogues, much faster than conventional methods (2-3 orders of magnitude) and with a realistic observational geometry. First, we present the optimization of a quasi N-body method in the compromise between accuracy and computational cost. We studied how variations in the code parameter space have and impact on the accuracy of observables such as the halo abundance and distribution and matter clustering. We propose optimal parameter configurations for achieving high accuracy as compared to exact N-body simulations and we explore different calibration techniques to match even better the latter. The next step is mimicking the geometry of real astrophysical observations, in which distant objects are seen in the past light cone. We introduce ICE-COLA, a simulation code developed for this thesis that implements the production of light cone catalogues on-the-fly. The user can generate three different kinds of data types. The first contains all the information while the others store high-level data catalogues ready to use to model galaxy surveys. This enables large compression factors of ∼ 2 orders of magnitude of the volume of data to be stored. In particular, the code can generate halo catalogues in the light cone and two-dimensional projected matter density maps in spherical concentric shells around the observer. We produce large light cone simulations with the new method developed and we show the validation of the catalogues. In particular, we model for the first time weak gravitational lensing with an approximate method and we show that we can resolve most of the scales probed by current lensing experiments. Finally we extend the results to halo mock catalogues with weak lensing quantities, which represents the first success in the ability of modelling galaxy clustering and weak lensing observables consistently in a fast simulation.

Date of Award	20 Jul 2016
Original language	English
Supervisor	Héctor Martín Crocce (Director), Eduard Masso Soler (Tutor) & Pablo Fosalba Vela (Director)