A study of photometric redshifts and intrinsic alignments with PAUS and Euclid

Abstract

The field of cosmology is undergoing an era of unprecedented new observational data, with galaxy surveys collecting information from billions of galaxies. This will enable the scientific community to perform precise measurements of the large-scale structure of the universe, enhancing our understanding about the cosmological model that best describes it. This thesis focuses on photometric redshift (photo-z) estimation and the study of intrinsic alignments (IA) with the Physics of the Accelerating Universe Survey (PAUS) and Euclid. First, we employ PAUS to estimate the distances to galaxies, via photo-z, for ∼1.8 million objects in an area of ∼50 deg2 and down to iAB < 23. We employ a template-fitting code to estimate photo-z and implement a novel zero-point calibration, which does not require spectroscopic redshifts. We define an innovative weighting scheme, where the photo-z estimates from PAUS and broad-band (BB) data are combined. This allows us to achieve photo-z accuracies one order of magnitude better than BB estimates for bright objects, while reducing the number of outliers and the bias. We analyse the photo-z accuracy as a function of the galaxy colour, finding similar performances for red and blue objects. Additionally, we provide calibrated probability density distributions of the photo-z for each object. Next, using the previously computed photo-z, we measure the galaxy clustering (GC) and the IA of galaxies in the PAUS fields, for 400000 objects in the redshift range 0.1 < z < 1.0 . We analyse the dependence on colour, redshift and luminosity and constrain the GC and IA parameters for each scenario. The results show significant detection of IA for red galaxies, with an increase in the signal at high redshifts and luminosities. In contrast, blue galaxies show null or weak alignments. In general, our results are consistent with previous literature, extending the IA amplitude-luminosity relation towards fainter objects. Finally, we implement the Tidal Alignment and Tidal Torquing (TATT) IA model in Euclid’s likelihood pipeline. Then, we perform forecasts on how IA affects cosmological parameter inference, demonstrating that the mismodelling of the IA effect might lead to biases in the cosmological parameters. Therefore, we highlight the need of adding flexible IA models for accurate cosmological inference in upcoming galaxy surveys.

Date of Award	15 Oct 2024
Original language	English
Awarding Institution	Autonomous University of Barcelona
Supervisor	Enrique Gaztanaga (Supervisor) & Martin Crocce (Supervisor)