SUBSTELLAR Pizza Seminar
"Bayesian methods for UCD detection and characterization. An application to Euclid data"
By Dr Javier Olivares (UNED), IAC, 28/11/2023
The first SUBSTELLAR Pizza Seminar was held today at the IAC by Dr Javier Olivares (UNED). His presentation was focused on Bayesian methods for UCD detection and characterization. An application to Euclid data.
In this presentation, there were overviewed three Bayesian methods for the characterization and detection of UCDs. Kalkayotl and Sakam are twocodes that work with astrometry and photometry, respectively. These two methods can be used to infer the properties of the stellar system to which a UCD belongs. Once the properties of the group have been established, these can be used to infer with better precision the age, mass, position and velocity of the UCD. In the context of Euclid, where only infrared photometry and sky positions will be available, the Bayesian methods can be used to transfer knowledge from the bright and well-characterized stellar members to UCDs with large uncertainties and possibly missing values. The third method, Miec, is a Bayesian hierarchical model created to identify members of stellar clusters with differential extinction. Possible modifications to this code are suggested to fit the Euclid data requirements. Finally, it was also discussed some AI methods that could be of use for UCD detection in the Euclid context.
Dr Olivares is an astrophysicist whose research focuses on the analysis of stellar systems in the solar vicinity (star-forming regions, stellar associations, and open clusters) particularly on the characterisation of the brown dwarf and planetary-mass domain of the initial mass function. He did a joint PhD in astrophysics and artificial intelligence (AI) and has spent the last six years as a postdoc researcher. First, contracted by the ERC-founded COSMIC-DANCe project he developed tools that mine catalogues of millions of sources and identify members of stellar systems down to the planetary mass regime. Then, as a postdoc at the IAC (Instituto de Astrofísica de Canarias) his task was to mine the billion-source Gaia catalogue to identify and characterize members in nearby stellar systems and fine-tune the search to ejected low-mass stars and brown dwarfs located in the tidal tails of these systems.
Currently, Dr Olivares is a postdoc at the Artificial Intelligence Department of the National Distance Education University of Spain (UNED), where he won a grant to exploit current AI developments and apply them to the Astrophysics and Earth science domains. During his first year as a postdoc, he and his students joined two AI paradigms (artificial neural networks and Bayesian hierarchical models) in an age inference method that successfully combines isochrones and Lithium depletion dating techniques.
See the presentation here Bayesian methods for UCD detection and characterization. An application to Euclid Data