Probing cosmology with an eye on Rubin: from Strong Lensing to the Large Scale Structure of the Universe
Name: RENAN ALVES DE OLIVEIRA
Publication date: 11/04/2024
Examining board:
Name |
Role |
|---|---|
| CLÉCIO DE BOM | Examinador Externo |
| HERMANO ENDLICH SCHNEIDER VELTEN | Examinador Interno |
| MARTIN MAKLER | Presidente |
| RAUL ABRAMO | Examinador Externo |
| VALERIO MARRA | Examinador Interno |
Summary: In 2024, the Vera C. Rubin Observatory will begin observing the Universe for the next ten years. Two key cosmological observables that Rubin will probe are gravitational lensing and the large-scale structure of the Universe. In this thesis, we derive analytical solutions for strongly lensed images that can be useful for generating fast simulations and as a starting point in parameter searches for lens inversion. Then, we obtain an expression in closed form for the magnification cross-section, which can be used to predict the abundance of highly magnified sources. Next, we focus on real data and assemble an extensive compilation of Strong Lensing candidate systems from the literature containing over 30,000 unique objects. We cross-match this sample with the current major photometric and spectroscopic catalogs. As preparation for Rubin, we generate image cutouts for these systems in most current wide-field surveys with subarcsecond seeing, namely DES, HSC, KiDS, CFHTLens, RCSLens, and CS82. This sample dubbed the “Last Stand Before Rubin” (LaStBeRu), has a myriad of applications, from using archival data to selections for follow-up projects and training of machine learning algorithms. As an application, we have performed a test of General Relativity (GR) with these data, combining information from strong lensing and velocity dispersions, which allow one to set constraints on the Post-Newtonian parameter PPN. From the LaStBeRu database, we were able to provide the first independent test of PPN from previous results and for the first time only for systems identifiable in ground-based images. We can obtain the most stringent constraint on PPN by combining these data with the current samples. Moreover, we have obtained new spectroscopic data for systems selected from LaStBeRu, which were used to obtain the first end-to-end determination of PPN. It is also the first determination derived purely from ground-based data and the first to use self-consistent priors. Our results are consistent with GR at the 1- level and with the previous results from the literature. Finally, in the context of the large structure, we present two neural emulators capable of making fast predictions for the density, displacement, and velocity fields of dark matter particles without necessarily having to run expensive N-body simulations. We compared these emulators with another fast method for the same task, showing that neural emulators provide the best results.
