The BINGO project: Cosmological constraints from simulated HI intensity maps

• Ano de defesa: 2021
• Autor(a) principal: Motta, Pablo Cesar Benevides de Carvalho Rossas
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: 21cm; amostragem aninhada; angular power spectra; BINGO; Cosmologia; cosmological constraints; Cosmology; espectro de potência angular; estimação de parâmetros; HI.; Monte Carlo; nested sampling; parameter estimation; vínculos cosmologicos
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/43/43134/tde-02092021-133803/

One of the main cosmological challenges today is to explain dark energy. A recent project which aims to understand dark energy properties is BINGO (Baryon Acoustic Oscillations from Integrated Neutral Gas Observation). BINGO is an IM instrument designed to measure BAO in the radio band, in the frequency range of 980MHz - 1260MHz ( 0.13 < z < 0.45 ), through the measurement of the 21cm line of emission. The optical configuration consists of two static dishes of ~40m of diameter and 28 feed horns disposed in a configuration called Double-Rectangular arrangement. In this work, we use the nested sampling Monte Carlo method to study cosmological constraints in BINGO. Since BINGO has not released any real data yet, we use some recently published HIR4 mocks of the HI Intensity Mapping signal. The technique requires the use of objects such as the mixing matrix, which mainly contains information on the BINGO observation area, and the covariance matrix, that relates to the statistical uncertainties. We provide a large discussion on these topics, as well as on the angular power spectrum formalism. The covariance matrix is computed through FLASK lognormal simulations, and we compare it to a theoretical model. We do not take into account any noise or foreground removal. We placed constraints separately on three redshift bins (from a total of 30) over the BINGO range, and we combined them through a Planck likelihood. We have successfully recovered the fiducial parameters, so our results seem to validate the HIR4 mocks. The next step will be to extend this analysis to the 30 BINGO bins all together. This task involves a high computational cost, so a large computer cluster is required.