Characterizing large-scale structure tracers with machine learning

• Ano de defesa: 2025
• Autor(a) principal: Rodrigues, Natália Villa Nova
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: aprendizado de máquina; conexão halo-galáxia; estrutura em grande escala; halo-galaxy connection; large-scale structure; levantamentos fotométricos; machine learning; photometric surveys
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/43/43134/tde-11092025-192533/

In the era of high-precision cosmology, a major challenge is to efficiently and accurately model large-scale structure in the nonlinear regime, and to extract the maximum information from the vast datasets provided by modern surveys. In this context, machine learning methods have gained attention for their flexibility in handling large and complex datasets. In this work, we address the modeling of large-scale structure in three main directions. First, on the observational side, we develop methods to classify quasars, galaxies, and stars in the miniJPAS narrow-band photometric survey. Second, we investigate the halo-galaxy connection, a key element in linking cosmological and galaxy formation models with observables. Third, we study how to model the distribution of dark matter halos based on information from the underlying dark matter field. Across different contexts, we explore how machine learning can contribute to the characterization of halos, galaxies, and quasars, which act as tracers of the underlying matter distribution. Together, these efforts contribute to the broader goal of modeling the large-scale structure of the Universe and addressing fundamental questions in cosmology.