From the Precision Era towards the Accuracy Era of Cosmology with DESI

[eng] Despite the successes of the cosmological ΛCDM model and having entered the "Precision Era of Cosmology" there are still open questions. The principal model ingredients, ΛCDM, contribute to ~95% of the total energy density of the universe, but their underlying nature is still completely unknown. This lack of understanding is the main science driver behind many experimental and observational missions as well as theoretical efforts within the field of fundamental physics. Furthermore, different cosmological observations favor different parameter values, where the most famous discrepancy is the up to (depending on the considered dataset) ~5σ "tension" between model-dependent early-time and direct late-time measurements of the Hubble constant H0. The Dark Energy Spectroscopic Instrument (DESI) survey is one of these campaigns. As the name indicates, it was launched to unravel the mystery of dark energy by measuring millions of distant galaxy and quasar spectra to create the largest, three-dimensional map of the large scale structure of the universe ever obtained. From that map, the DESI collaboration aims to extract both the expansion history and the growth rate of structures history throughout cosmic time. The expansion history is obtained via the so-called standard ruler technique: distances (in function of redshift) are measured in units of a characteristic scale, the standard ruler, which is an imprint of the gravity-pressure waves in early universe leading to the so-called baryon acoustic oscillations (BAO). The growth rate of structures is traced by the measurement of the anisotropy of galaxy clustering along and across the line-of-sight, which is induced by the peculiar velocities of galaxies impacting the redshift measurements from their spectra. As a consequence, distances inferred from these redshifts are distorted, hence this effect is called redshift-space distortions (RSD). Both the BAO and RSD observables deliver a pristine probe of the late-time dynamics of the universe. In the first part of this thesis we present a method to blind the galaxy catalogs to mimic different BAO and RSD signals. Upcoming DESI data will benefit from blinding in order to remove the impact of confirmation bias on cosmological results. We explore two blinding shifts at the catalog level, perturbing individual galaxy positions within the galaxy clustering catalog along the line of sight. The first one is a purely geometrical shift based on a different expansion law. In the second one redshifts are shifted depending on the galaxy density field mimicking RSD with a modified growth rate. We test both blinding shifts by performing BAO and full shape RSD analyses on original and blinded galaxy mocks. In the second part, we elevate the established way how BAO and RSD analyses are performed towards including another observable, the shape of the clustering signal as function of galaxy separations. While the BAO and RSD incorporate the horizontal and vertical information respectively in the clustering signal, the shape captures the "diagonal" information. We find that this technique called ShapeFit is sufficient to obtain cosmological constraints as tight as direct model fits to galaxy two-point statistics while preserving the advantages of model-independence of the standard BAO and RSD analyses. Both parts of this thesis stress the importance of model-agnosticism in the context of large surveys and cosmological tensions. They play a crucial role for the DESI survey cosmological analysis providing a road to transition from the "Precision Era" to the "Accuracy Era" of cosmology.[spa] El trabajo de esta tesis, basada en proyectos para el Instrumento Espectroscópico de Energía Oscura (DESI), consiste de dos partes. En la primera parte de esta tesis presentamos un método de cegado en los catálogos de galaxias que enmascara las señales reales de “oscilaciones acústicas de bariones” (BAO) y “distorsiones del espacio de corrimiento al rojo” (RSD). Los futuros datos de DESI se beneficiarán de este cegado para eliminar el impacto del sesgo de confirmación en los resultados cosmológicos. Exploramos dos tipos de desplazamientos para este cegado a nivel de catálogo, perturbando las posiciones individuales de las galaxias a lo largo de la línea de visión. El primero es un desplazamiento puramente geométrico basado en una ley de expansión del universo diferente. En el segundo, los corrimientos al rojo se desplazan en función del campo de densidad de las galaxias imitando una señal de RSD con una tasa de crecimiento de estructura modificada. Demostramos que ambos desplazamientos distorsionan las señales de forma coherente al realizar análisis de BAO y RSD en los catálogos originales y ciegos. En la segunda parte, mejoramos el estado del arte en que se realizan los análisis BAO y RSD incluyendo un nuevo observable, la “forma” de la señal de agrupamiento en función de las separaciones de las galaxias. Mientras que las señales de BAO y el RSD incorporan la información horizontal y vertical respectivamente en la señal de agrupamiento, la forma capta la información "diagonal". Encontramos que esta técnica llamada ShapeFit es suficiente para obtener mediciones cosmológicas tan precisas como los ajustes directos del modelo a las estadísticas de dos puntos de las galaxias, preservando al mismo tiempo las ventajas de la independencia del modelo de los análisis estándar BAO y RSD. Ambas partes de esta tesis subrayan la importancia del agnosticismo hacia el modelo en el contexto de los grandes sondeos y las tensiones cosmológicas. Estas, despeñan un papel crucial para el análisis cosmológico del sondeo DESI, y proporcionan un camino para la transición entre la "Era de la Precisión" a la "Era de la Exactitud" de la cosmología