Tesis profesional presentada por Pablo Huerta Ocaña [pablo.huertaoa@udlap.mx]

Miembro del Programa de Honores. Licenciatura en Actuaría. Departamento de Actuaría, Física y Matemáticas. Escuela de Ciencias, Universidad de las Américas Puebla.

Jurado Calificador

Director: Dra. Milagros Zeballos Rebaza
Presidente: Dr. Miguel Ánguel Reyes Cortés
Secretario: Dr. Rubén Blancas Rivera

Cholula, Puebla, México a 5 de diciembre de 2023.

Resumen

This research focuses on the phenomenon of cosmological redshift, a crucial aspect in contemporary cosmology and our understanding of the universe. The study uses photometric redshift determination techniques based on artificial intelligence, utilizing the ugriz photometric bands extracted from the Sloan Digital Sky Survey (SDSS) and the infrared bands from Wide-field Infrared Survey Explorer (WISE) as input. The primary objective is to demonstrate the effectiveness of neural networks in solving regression problems, as compared to empirical techniques. The work introduces ANNz2, a code developed by Sadeh et al. (2016), and compares it with a neural network implemented in Keras, as well as the empirical method for determining redshifts from the SDSS. The results exhibit superior performance of neural networks in the range of 0:0 < z < 0:8 when compared to the empirical method of the SDSS. Additionally, it is identified that the ugriz+WISE bands are not sufficient for predicting redshifts in the interval of 0:8 < z < 1:5, although they still outperform other empirical techniques.

Keywords: Galaxies: distances and redshifts - Catalogs: SDSS - Large-scale structure of Universe - Methods: data analysis - Methods: numerical

Table of content

Portada

Índices

Capítulo 1. Introduction

Capítulo 2. Theoretical Framework

  • 2.1 Cosmological redshift
  • 2.2 Galaxies
  • 2.3 Photometry
  • 2.4 Artificial neural networks

Capítulo 3. Metodology

  • 3.1 ANNz2
  • 3.2 SDSS method
  • 3.3 Keras Neural network
  • 3.4 Metrics

Capítulo 4. Input data

  • 4.1 Sloan Digital Sky Survey data
  • 4.2 Wide-field Infrared Survey Explorer data

Capítulo 5. Results

  • 5.1 SDSS experiments
  • 5.2 WISE experiments

Capítulo 6. Conclusions

Referencias

Huerta Ocaña, P. 2023. Redshift estimation via artificial intelligence methods. Tesis Licenciatura. Actuaría. Departamento de Actuaría, Física y Matemáticas, Escuela de Ciencias, Universidad de las Américas Puebla. Diciembre. Derechos Reservados © 2023.