Extragalactic background light studies with gamma rays via Markov chain Monte Carlo methods

• Ano de defesa: 2023
• Autor(a) principal: Xavier, Matheus Genaro Dantas
• 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: análise Bayesiana; Bayesian analysis; estimação de sinal; Extragalactic Background Light; gamma rays; Hamiltonian Monte Carlo; Luz Extragaláctica de Fundo; Monte Carlo Hamiltoniano; raios gama; signal estimation
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/43/43134/tde-10102023-101716/

The Extragalactic Background Light (EBL) is the second most intense background radiation field in the universe, being the product of the integrated stellar emission and light reprocessed by dust throughout the history of structure formation and cosmic evolution. The precise spectral shape of the EBL is not completely known, as direct measurements are difficult to make due to dominant foregrounds. However, it is possible to probe the EBL indirectly using gamma rays, since, during their propagation over cosmological distances, very high energy (VHE) photons can interact with the EBL producing electron-positron pairs. This dissertation explores the use of Markov Chain Monte Carlo methods to obtain simultaneous constraints on the EBL and intrinsic spectral parameters of gamma-ray sources. Thus, the fundamental goal is to reconstruct the posterior probability density of parameters characterising the EBL and the intrinsic flux emission of active galactic nuclei (AGNs), in a Bayesian approach. The first part of the work is mainly concerned with validating the methodology with a sample of synthetic BL Lacs observed with the instrument configuration of the future Cherenkov Telescope Array (CTA). In this controlled scenario, we investigate the impacts on EBL constraints by progressively including more spectra in the likelihood function. We identify a consistent improvement in uncertainties by combining different sources in the analysis, while also being capable of recovering the spectral indices of all intrinsic spectra. We further explore the impacts of increasing the observation time of the sources and possible systematic effects associated to the choice of EBL modelling. In the second part, we analyse a sample of 65 real spectra from 36 AGN observed by various Imaging Atmospheric Cherenkov Telescopes, obtaining constraints on the EBL and intrinsic parameters that are consistent to other results found in the literature. We identify the Markarian 501 flare data as essential for constraining the far-Infrared part of the EBL, while the combination of all other sources provided robust constraints on the mid-Infrared. Such analysis was possible through the use of the Hamiltonian Monte Carlo method, which is very efficient in a parameter space with a high number of dimensions. Two other extensions are also explored. With the synthetic sample, we discuss the possibility of constraining the Hubble constant while probing the EBL. We also present a Bayesian method for signal estimation in On/Off measurements and perform a preliminary analysis based on H.E.S.S. data. Such method allows improved signal estimation without performing selection cuts on data, which could be useful for improving VHE measurements and detection of faint sources.