Statistics in large galaxy redshift surveys

Abstract

This thesis focuses on modeling and measuring pairwise statistics in large galaxy redshift surveys. The first part focuses on two point correlation function measurements relevant to the Euclid and DESI BGS surveys. Two point measurements in these surveys will have small statistical errors, so understanding and correcting for systematic bias is particularly important. We use point processes to build catalogues with analytically known two point, and for the first time, 3-point correlation functions for use in validating the Euclid clustering pipeline.
We build and summarise a two point correlation function code, , and show it successfully recovers the two point correlation function of a DESI BGS mock catalogue. The second part of this thesis focuses on work related to the PAU Survey (PAUS), a unique narrow band wide field imaging survey.
We present a mock catalogue for PAUS based on a physical model of galaxy formation implemented in an N-body simulation, and use it to quantify the competitiveness of the narrow band imaging for measuring novel spectral features and galaxy clustering. The mock catalogue agrees well with observed number counts and redshift distributions. We show that galaxy clustering is recovered within statistical errors on two-halo scales but care must be taken on one halo scales as sample mixing can bias the result.
We present a new method of detecting galaxy groups, Markov clustering (MCL), that detects groups using pairwise connections. We explain that the widely used friends-of-friends (FOF) algorithm is a subset of MCL. We show that in real space MCL produces a group catalogue with higher purity and completeness, and a more accurate cumulative multiplicity function, than the comparable FOF catalogue. MCL allows for probabilistic connections between galaxies, so is a promising approach for catalogues with mixed redshift precision such as PAUS, or future surveys such as 4MOST-WAVES.