3x2pt in the context of large-scale structure analysis with stage-IV surveys

Abstract

The field of cosmology is currently at the cusp of a new era of unprecedented precision thanks to the latest generation of cosmological surveys. Stage-IV surveys are expected to provide sub-percent constraints on the cosmological parameters of Lambda CDM, the so-called standard model of cosmology. Not only that, the wealth of data that will be generated by these surveys will allow for the exploration of alternative cosmological models beyond Lambda CDM, with early results from the DESI survey already showing promising hints at new physics regarding the nature of dark energy. Stage-IV surveys will provide imaging, photometric and spectroscopic measurements of billions of galaxies, allowing for the characterisation of the dynamics of structure formation and evolution in the late Universe. The development of an adequate methodology of analysis for these upcoming surveys is crucial in order to fully extract the cosmological information present in the data without incurring in biases in the inferred parameters of the model. _x000D_ _x000D_ The works presented in this thesis have been carried out in the context of the stage-IV surveys. Three works explore the combined use of galaxy clustering, weak lensing and galaxy-galaxy lensing, the so-called 3x2pt analysis, in the methodology of analysis of large-scale structure with stage-IV-like simulated data. _x000D_ _x000D_ The first work is devoted to the optimisation of the selection of tomographic redshift bins with the specific objective of improving the constraints on the parameters of a dynamical dark energy model. We have developed a method to robustly and efficiently optimise the tomographic configuration for 3x2pt analysis through an iterative sampling of the source and lens galaxy samples. The application of this optimisation method results in improvements in the constraints of the dynamical dark energy parameters w0 and wa of 25% compared to the widely used case of redshift bins of equal width. _x000D_ _x000D_ The second work investigates the cosmological bias that may arise when using 3x2pt analysis while assuming an incorrect model of gravity in the modelling. In particular, we measure the bias in the inferred cosmology from assuming General Relativity in the modelling of a modified gravity simulated galaxy mock. We used a set of twin galaxy mocks generated with two different models of gravity: General Relativity and the Hu-Sawicki formulation of f(R) gravity. The Hu-Sawicki model is a particular form of f(R) gravity proposed to explain the accelerated expansion of the Universe. The analysis of the modified gravity mock while assuming General Relativity in the modelling resulted in very significant biases in the inferred cosmology. The bias in the marginalised posteriors of {Omega_m, sigma_8} is of 12 sigma and so is the bias for the S_8 parameter. Our work shows that assuming General Relativity in the modelling of the 3x2pt data vectors when analysing a modified gravity Universe can result in very significant biases. _x000D_ _x000D_ The objective of the third work, still ongoing, is to assess the feasibility of modelling the 3x2pt data vectors as measured in the Flagship 2 simulation up to non-linear scales using the SHAMe emulator. The SHAMe model is an extension of the Subhalo Abundance Matching method created to populate dark matter simulations with galaxies. It provides a sophisticated modelling of the galaxy-halo connection which results in accurate clustering and lensing observables even up to non-linear scales. We have tested the use of the SHAMe emulator for the projected correlation function and the galaxy-galaxy lensing function, achieving accordance with the corresponding measurements from the Flagship 2 simulation up to very small scales of 1 arcmin2. If successful, our project could pave the way for the use of the SHAMe emulator in the analysis of Euclid data.

Date of Award	13 Feb 2026
Original language	English
Awarding Institution	Universitat Autònoma de Barcelona (UAB)
Supervisor	Isaac Tutusaus Lleixa (Director) & Pablo Fosalba Vela (Director)