Who Speaks Louder? Optimizing the Extraction of Information from Gravitational Waves for future Cosmological Analyses

Gravitational waves have emerged as a robust cosmological probe in an era where precision measurements increasingly challenge the completeness of the standard ΛCDM model. Future detectors upgrades will lead to a significant growth in the number of detections, with catalogs expected to reach orders of 10^5 events. These advances will enable tighter constraints on both cosmological parameters (e.g., the Hubble constant) and on the population properties of compact binaries. While such large data sets promise unprecedented statistical power, they also raise practical challenges: analyzing entire catalogs may become computationally prohibitive, and the information content of individual events may not be uniformly distributed. This thesis investigates how event selection strategies can optimize the extraction of information from large GW catalogs. Using simulated dark siren datasets derived from the MICE galaxy simulation with spectroscopic redshift uncertainties, cosmological and astrophysical parameters are inferred within a hierarchical Bayesian framework implemented in the CHIMERA pipeline. The analysis focuses on the impact of progressively stricter SNR thresholds, examining how constraints on cosmology and binary population parameters evolve as the event sample is reduced. Correlation studies and principal component analysis are employed to identify the dominant factors governing the informativeness of GW events. The analysis highlights how catalog subsampling can balance accuracy and computational demands, outlining strategies that will be increasingly important in the era of third–generation detectors.