Galactic Archaeology with Convolutional Neural-Networks: Realising the potential of Gaia and 4MOST

• Applicant: Dr. Guillaume Guiglion, Ph.D.
• Subject Area: Astrophysics and Astronomy
• Term: since 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 511098263

Unraveling the chemo-dynamical processes involved in the formation of the Milky Way (MW) and its surrounding environment is currently one of the biggest challenges in modern astronomy. Such processes are fundamental to understanding the evolution of not only our Galaxy, but also its satellite galaxies such as the Large and small Magellanic clouds (LMC/SMC). Detailed observations in these nearby environments can then provide critical insights into the general processes of galaxy evolution in the larger context of the Λ Cold Dark Matter model. Since the photospheric abundances of FGK stars are stable over long periods of cosmic time, they can be used as a fossil record to look back time and study the environment in which the star was born. Thus, the relative chemical abundances of stars allow us to meticulously disentangle stellar populations and to put robust constraints on the nucleosynthetic origins of various elements in the Universe. This is why it is so important to obtain precise abundance measurements for these stars. The community has put strong efforts into building spectroscopic surveys for Galactic Archaeology investigations, such as RAVE, and APOGEE. The next generation of spectroscopic surveys, such as 4MOST, and Gaia-RVS will gather several tens of millions of spectra, opening new challenges for precisely and accurately measuring abundances. With such large volumes of data it is critical to employ efficient analysis techniques, and Machine-Learning (ML) methods have been shown in recent years to have incredible potential for processing and and analyzing such data sets. With this in mind, I plan to develop a highly sophisticated, cutting-edge ML based analysis pipeline, implementing a Convolutional Neural-Network (CNN) capable of precisely and rapidly determining abundances and atmospheric parameters using the data products of the Gaia-RVS space mission and the 4MOST consortium survey. In addition, ages, masses and kinematics will also be computed. With the powerful synergy of chemistry, kinematics and ages, the stage will be set to address highly pertinent and impactful scientific questions: I will put chemo-dynamical constraints on the MW halo/disc interface, and I will investigate the chemical-evolution of lithium and neutron-capture elements in the MW and the LMC/SMC using 4MOST abundances.

• DFG Programme: Research Grants