Getting the record from the disc: A study of the Milky Way disc with WEAVE and Gaia – MWDisc

The project consists of three complementary components: the construction of homogenized catalogs, the determination of stellar ages, and the chemo-dynamical modeling of the Galactic disc. Following the instrumental delays of WEAVE, the methods were deployed and validated on existing public surveys, ensuring publishable intermediate results and optimal preparation for future WEAVE data exploitation. This evolution has strengthened both the methodological scope and the transferability of the developed tools.

The first component involves the homogenization of spectroscopic catalogs from different surveys. Supervised machine learning methods were developed to place atmospheric parameters and chemical abundances on a single reference scale, relying on cross-survey overlapping stars and reference stars. These techniques were successfully tested on APOGEE and GALAH, integrating physical constraints to ensure robustness and interpretability. This approach will be directly transferable to WEAVE data as soon as it becomes available.

The second component focuses on the determination of stellar ages. The project developed Bayesian isochrone-fitting methods combining spectroscopic parameters, multi-band photometry, and Gaia parallaxes, while explicitly accounting for extinction laws. New age–metallicity and age–[alpha/Fe] relations were obtained from APOGEE data, allowing for the study of the atypical population characterized by young ages and high alpha-abundance. The methods were designed to be modular so that they remain applicable to various datasets.

The third component concerns dynamical modeling. Perturbative models in angle-action coordinates were developed to describe the disc's response to non-axisymmetric perturbations (bar, spiral arms), including the treatment of resonances and time-varying amplitudes. Numerical codes for perturbed distribution functions and self-gravitating secular diffusion (Fokker-Planck type) were applied to mono-age and mono-abundance populations to explore the observable signatures of radial migration and variations in the bar's rotation speed.

The overall work provides a coherent framework linking reference catalogs, dating methods, dynamical modeling, and global Galactic interpretation, with the ultimate goal of producing robust constraints on the evolution of the Milky Way disk and sustainable tools for the community.

Despite the delays with WEAVE, the project has produced results across all its components, ensuring immediate impact and preparing for the exploitation of future data.

Regarding homogenisation methods, we developed and validated cross-catalogue homogenisation techniques based on machine learning, placing stellar parameters and chemical abundances on a common scale across reference surveys. Successfully tested on APOGEE and GALAH, these methods demonstrated their ability to integrate physical constraints- such as abundance correlations from common nucleosynthetic channels- and will be central to WEAVE analyses. Two papers have been published (Thomas+ 2023, Turchi+ 2025), with another currently in preparation (Gran+ in prep.).

In terms of the chemical evolution of the disc, new inference pipelines combining spectroscopy, multi-band photometry, and Gaia parallaxes were developed, accounting for various extinction laws and photometric calibrations. This work enabled the analysis of seemingly young alpha-rich stars, showing that they are predominantly "blue straggler" type objects -in this case, thick disc stars that acquired mass via transfer from a companion star (Cerqui+ 2023). Crucially, these efforts produced new age–metallicity and age–[alpha/Fe] relations at different Galactic radii (Cerqui+ 2025), demonstrating how the chemical evolution of the disc depends on Galactic radius and how radial migration induced by the bar can be used to estimate its age (Haywood+ 2024). Finally, it was shown that stars with higher metallicity than the interstellar medium (migrated stars) are systematically older than lower-metallicity stars formed locally (Kordopatis+ 2025).

Regarding the internal dynamics of the disc, new perturbative models in angle-action coordinates were developed, incorporating the bar and spiral arms, variable amplitudes, and resonance treatment (Li+ 2024, Yuan+ 2024). The PERDIGAL code was produced and validated (al Kazwini+ 2022). Test-particle simulations showed that a bar with a decelerating rotation speed can reproduce the main phase-spiral structures observed by Gaia. In parallel, the proof of concept for a time-dependent self-gravitating linear response was published, paving the way for secular diffusion models of the disc.

Other results and publications include papers on stellar dating and the identification of robust spectral diagnostics according to stellar type (Kordopatis+ 2023a,b), the characterisation of new globular clusters (Gran+ 2024), the study of the metal-poor disc (Fernandez-Alvar+ 2024, Gonzalez Rivera+ 2024), and models of perturbed dynamical response (Khalil+ 2025). Two further papers (Gran+ 2026a,b) have been submitted analysing accreted halo stars using P.I. data from the VLT.

The project establishes methodological and numerical foundations that are directly exploitable by upcoming large-scale spectroscopic surveys. The tools developed for cross-catalogue homogenisation, stellar age determination, and dynamical modelling constitute a ready-to-use framework for the joint analysis of Gaia and WEAVE, followed by other surveys such as 4MOST.

In the short term, the progressive availability of WEAVE data will allow for the direct and in-depth application of these methods, providing a significant shift in statistical scale. The combination of homogenised chemical abundances, robust ages, and dynamical observables will pave the way for quantitative constraints on the formation history of the disc, the efficiency of radial migration, the effects of extragalactic stellar accretion, and the temporal properties of the Galactic bar and spiral arms.

Developments in angle-action modelling and self-gravitating perturbative response can be extended to multi-population and multi-temporal perturbation models. This will enable a direct physical interpretation of the fine substructures expected to be revealed by the fourth Gaia catalogue (scheduled for December 2026).

Finally, the methods for catalogue combination are transferable to other astrophysical contexts and data science fields dealing with heterogeneous and biased measurements. They promote the production of high-value-added catalogues for the wider scientific community.

Published papers:
Al Kazwini et al., A&A 658, A50 (2022)
Cerqui et al., A&A 676, A108 (2023)
Kordopatis et al., A&A 669, A104 (2023)
Kordopatis et al. A&A 674, A104 (2023)
Li et al. in print to MNRAS, doi:10.1093/mnras/stad2199
Gran et al. accepted to A&A

In a hierarchically-formed Universe, the Milky Way is a test-bed to study in details the mechanisms that shape galaxies. The synergy between the Gaia space satellite and the ground-based spectroscopic survey WEAVE gives access, for the first time, to more than thirty tracers of the past of our Galaxy for a million stars of the extended Solar neighbourhood, and to a dozen of tracers for another two million stars outside of it. Our project concerns the study of the Galactic disc, a structure that encodes both internal (e.g. stellar radial migration) and external (e.g. accretion events) mechanisms that come into play in the chemo-dynamical evolution of our Galaxy.

We have built a versatile team with nodes in Nice, Paris and Strasbourg, including experts in Galaxy evolution, simulations and modelling. The team members are heavily involved in both WEAVE and Gaia in order to extract the maximum of information available in those combined catalogues. Over the course of the four-year MWDisc ANR project, and alongside to the accumulation of the WEAVE data (starting in Q1 2021), we aim to produce added-value catalogues for the WEAVE stellar targets (containing homogeneous stellar chemical abundances, ages, orbits and extinctions) and models associated to the diffusion of mono-age populations by time-varying perturbations and superposition of perturbations (associated to the spiral arms and the Galactic bar). These models and catalogues will allow us, in turn, to evaluate the star formation history in various regions of the disc, put constraints on the merger tree of the Milky Way (including the analysis of existing simulations of ours), to link the geometrical properties of the thin and thick disc with their chemical counterparts and finally to characterize the efficiency of radial migration throughout the disc.