2021-10-28, 12:15–12:30, Grand Ballroom
The analysis of absorption features along the line of sight to distant sources is an invaluable tool for observational cosmology, giving a direct insight into the physical and chemical state of the inter- and circumgalactic medium, which can be used to constrain both the cosmic evolution (abundance of primordial elements, reionization, etc.) and fundamental physics (dark matter, variability of constants, general relativity, etc.). Such endeavour entails two basic requirements: the accessibility of bright QSOs as background beacons, to be observed with current and future earth-based facilities (e.g. the ESO VLT and ELT), and the availability of software resources to properly analyze the data, extracting the relevant information in a reliable and reproducible way. In this talk, we will present the latest results we obtained in both direction within the QUBRICS project (QUasars as BRIght beacons for Cosmology in the Southern hemisphere). We will describe how machine learning techniques (canonical component analysis, probabilistic random forest) were applied to detect hundreds of previously unknown QSOs in the southern hemisphere, and how state-of-the art software like QSFit and Astrocook was integrated in the analysis of the targets, opening up new possibilities for the next era of intergalactic medium observations.
Solutions for workflow management and reproducibility, Understanding and improving machine learning, Building accessible and friendly user interfaces