The ongoing mission Gaia and its planned sequel Gaia-NIR provide accurate astrometric information for over a billion Galactic and extragalactic stars.
We know now the spatial position of the source and two components of the velocity vector. However, spectroscopic measurements of the third component, radial velocity (RV), are missing for a large majority of targets since they are too faint for Gaia's RVS spectrograph and the ground-based surveys can observe only a limited number of stars.
Using AI to infer stars’ RV
This study proposes to infer a star's RV using artificial intelligence (AI) techniques (a) from astrometry and (b) from astrometry and complementary properties of the source.
We will test if absolute magnitude, colour, flux variability, and likely membership in individual Galactic components can improve the RV inference accuracy.
Approach (a) has been recently attempted but it has never been validated with Gaia DR3 RVs and/or by ground-based measurements.
Approach (b) adds knowledge of photometry and physical classification flags from Gaia DR3 to astrometry as an input for the AI approach.
The proposal's goal is twofold: (1) to test the published astrometric RV inferences with RVs recently measured by Gaia DR3 and ground-based spectroscopic surveys, (2) to explore possible improvements to the RV inference scheme by addition of priors from photometry, Gaia classification flags, and spatial distribution of Galactic components.
Partners
- Faculty of Mathematics and Physics, University of Ljubljana
- European Space Agency (ESA)