Dottorato - PhD thesis
Area Tematica: RSN 1: Galassie e Cosmologia - RSN 5: Tecnologie avanzate e strumentazione-
Referente: Giorgio Calderone (
Titolo: QUBRICS: bright, high-redshift quasars for Cosmology
Decorrenza: 03.11.2025
The QUBRICS project applies state-of-the-art machine learning (ML) techniques to the discovery and classification of luminous,high-redshift quasars, using extensive photometric databases. As astronomical surveys such as Euclid and the Rubin Observatory introduce a new era of high-volume data acquisition, the need for advanced computational methodologies to process and analyze these vast datasets becomes increasingly critical. This PhD fellowship provides an opportunity to contribute to the technological evolution of ML-driven astrophysics, developing and implementing novel data-mining algorithms while utilizing high-performance computing resources to optimize quasar identification pipelines. Computational and ML Methodologies The selected PhD candidate will acquire expertise in cutting-edge data science applications within astrophysics, including: - Database Management: Maintenance and continuous optimization of large-scale photometric and spectroscopic datasets, ensuring seamless integration with expanding astronomical surveys. - Quasar Candidate Selection: Implementation of probabilistic classification models, including Probabilistic Random Forest, eXtreme Gradient Boosting, and other deep learning frameworks to enhance the accuracy of quasar detection. - Photometric Redshift Estimation: Exploration and deployment of Gaussian Processes, StratLearn, and other Bayesian approaches for refining redshift prediction models. - Target Prioritization Strategies: Development of multi-objective optimization frameworks that balance dataset completeness with the minimization of unnecessary spectroscopic follow-ups. - Proposal Writing for Spectroscopic Follow-Up: Formulating observing strategies that maximize scientific output while optimizing telescope time allocation. - Observational Execution: Coordination and participation in spectroscopic observations in both visitor and delegated visitor modes at leading observatories. Scientific Applications of ML-Driven Quasar Discovery By refining and implementing these technological advancements, the research will contribute to key astrophysical investigations, including: - Big Bang Nucleosynthesis: Assessing primordial element abundances through high-redshift quasar spectra. - Variation of Fundamental Constants: Investigating potential time variations in physical constants via absorption features in quasar spectra. - Sandage Test for Cosmic Redshift Drift: Applying precision measurements of distant quasar evolution to detect the universal expansion rate. - Galaxy Formation and Evolution: Using Damped Lyman-alpha systems to infer early-stage galactic formation mechanisms. - Origins and Growth of Supermassive Black Holes: Exploring black hole accretion processes and the role of quasars in shaping the cosmic UV background. - Broad Line Region (BLR) Dynamics: Studying ionization, kinematics, and structural properties of the BLR in quasars, providing insights into accretion physics. Supervisors: - Giorgio Calderone (P.S. - INAF) - S.Cristiani (INAF) - M.Girardi (UniTs)
