| The accelerating expansion of the universe is one of the main topics of modern cosmology. It may stem from a new component, so-called dark energy, which would make up 70% of the energy content of the universe. To study its nature through its equation of state, one can measure a standard ruler given by baryonic acoustic oscillations (BAO) at various redshifts or for different slices of the universe. The BAO signal was confirmed in 2012 by the BOSS project (SDSS-III) and then by the eBOSS project (SDSS-IV), both with galaxies and Ly-alpha absorbers revealed in quasar spectra. During my thesis, I was involved in the preparation of the next generation of BAO experiments, the new Dark Energy Spectroscopic Instrument (DESI) program. This project will perform a 3D survey of several tens of millions of galaxies and quasars with the 4-meter Mayall telescope located at Kitt Peak in Arizona. To this end, the Mayall Telescope has been upgraded with a reconfigurable focal plane made up of 5000 robotically controlled fiber tips that collect light from the observed objects. The light is transmitted down to 10 spectrograph modules which are able to simultaneously measure the spectra of the 5000 objects. Despite these unique capabilities, it is first necessary to select the targets to be pointed using data coming from extensive imaging surveys of the sky probed in several photometric bands. Among these targets, quasars are very bright and distant. They are ideal candidates for mapping the structures of the universe at high redshift. In this talk, I will introduce the science conducted by DESI using BAO measurement. I will then describe the general architecture of the instrument and the activities I participated in for the construction of the spectrograph and the alignment of the CCD sensors. I will also discuss the algorithms that I have developed for the selection of quasar candidates based on their photometry properties by making use of machine learning tools. |