Statistical tools in cosmology: model selection and covariance matrix comparison

• Ano de defesa: 2021
• Autor(a) principal: Ferreira, Tassia Andrade
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Espírito Santo; BR; Doutorado em Astrofísica, Cosmologia e Gravitação; Centro de Ciências Exatas; UFES; Programa de Pós-Graduação em Astrofísica, Cosmologia e Gravitação
• Palavras-chave em Português: cosmologia; subject.br-rjbn; Astronomia
• Link de acesso: http://repositorio.ufes.br/handle/10/14957

Albeit ΛCDM’s fame as the concordance model, there are many interesting myster ies worth exploring, such as the nature of dark energy. Here, we test the viability of several classes of scenarios of the dark sector with linear and non-linear inter acting terms. To do so, we use a Bayesian model selection with data from type Ia supernovae, cosmic chronometers, cosmic microwave background and two sets of baryon acoustic oscillations measurements: 2-dimensional angular measurements (BAO2), and 3-dimensional angle-averaged measurements (BAO3). On the other hand, we consider covariance matrices, which are important tools for parameter estimation. We explore ways of compressing them by analysing their eigenvalues and signal-to-noise ratio, by employing a tomographic compression and, lastly, with the Massively Optimized Parameter Estimation and Data compression (MOPED). We find that MOPED is a powerful tool in the comparison of covariance matrices and, towards that end, we build a python code that uses a fast Monte Carlo simulation to obtain comprehensible values for differences between two covariance matrices. This method thus eliminates the need for a full cosmological analysis as we relate its output to the corresponding parameter constraints.