Exploring the cosmological large scale structure with deep learning

• Grant number: 24/18415-7
• Support Opportunities: Scholarships in Brazil - Program to Stimulate Scientific Vocations
• Start date: January 06, 2025
• End date: February 24, 2025
• Field of knowledge: Physical Sciences and Mathematics - Astronomy - Extragalactic Astrophysics
• Principal Investigator: Clecio Roque de Bom
• Grantee: Diogo Pereira de Lima Carvalho
• Host Institution: Centro Brasileiro de Pesquisas Físicas (CBPF). Ministério da Ciência, Tecnologia e Inovação (Brasil)

Abstract

Physical cosmology is entering a crucial era in which the significant advances of the last two decades will be rigorously tested through a new generation of cosmological surveys that may confirm the existing paradigm or reveal the need for new physics. Determining the redshifts of galaxies and quasars is essential for large-scale studies of the universe, but the necessary spectroscopy is expensive and unfeasible for all observable galaxies.Machine learning (ML) has shown a promising endeavour in astronomy. ML techniques have demonstrated efficiency in recovering galaxy properties from photometric data, allowing large-scale studies with greater coverage and depth than spectroscopic surveys. Furthermore, the application of ML significantly accelerates the processing and analysis of large datasets, optimizing resources and expanding our capacity to explore the universe on a large scale, contributing to advances in understanding the formation and evolution of cosmic structures. (AU)