Relying on the complementarity of Gaia proper motions with radial velocities of the RAVE survey, we attempt to constrain the kinematics of the Milky Way disc. Based on the population synthesis model, we simulate the observations, applying the detailed selection functions of the observations. The dynamics is described using a global gravitational potential computed from the mass distribution of the population model, approximated by a Stäckel potential (Bienaymé et al. 2015). We explore a set of free parameters (solar motion, age - velocity dispersion of the disc as a function of age, the velocity gradients, vertex deviation) using a Markov Chain Monte Carlo method. We show that the fitted model reproduces very well the radial velocity and proper motion distributions, allowing to constrain the thin and thick disc secular evolution with time.

The body of photometric and astrometric data on stars in the Galaxy has been growing very fast in recent years (Hipparcos/Tycho, OGLE-3, 2-Mass, DENIS, UCAC2, SDSS, RAVE, Pan Starrs, Hermes, . . .) and in two years ESA will launch the Gaia satellite, which will measure astrometric data of unprecedented precision for a billion stars. On account of our position within the Galaxy and the complex observational biases that are built into most catalogues, dynamical models of the Galaxy are a prerequisite full exploitation of these catalogues. On account of the enormous detail in which we can observe the Galaxy, models of great sophistication are required. Moreover, in addition to models we require algorithms for observing them with the same errors and biases as occur in real observational programs, and statistical algorithms for determining the extent to which a model is compatible with a given body of data.

JD5 reviewed the status of our knowledge of the Galaxy, the different ways in which we could model the Galaxy, and what will be required to extract our science goals from the data that will be on hand when the Gaia Catalogue becomes available.

I summarize the main characteristics of cosmological galaxy formation models and discuss some of the main results obtained in the last years.

From NLTE computations of the magnesium abundance in a sample of extremely metal-poor giants we derive [Mg/Fe]=+0.7, leading to [Al/Mg]=−0.80 and [Na/Mg]=−0.85 in the early Galaxy. The ratio [O/Mg] should be near to the solar value. Measurements of nitrogen abundances derived from the analysis of the NH band in eight more stars confirm the large scatter of the ratios [N/Fe] and [N/O] in the early Galaxy.