The star formation history and the stellar initial mass function of the Milky Way disc. The population synthesis Besançon Galaxy Model in the Gaia era

Abstract

AIMS:

We develop a new theoretical framework to generate Besançon Galaxy Model Fast Approximate Simulations (BGM FASt) to address fundamental questions of the Galactic structure and evolution performing multi-parameter inference. The flexibility of BGM FASt allows the inference of fundamental parameters related to the stellar initial mass function (IMF), the star formation history (SFH), the density distribution, the kinematics and the chemo-dynamics, among others. BGM FASt allows the study of different Milky Way (MW) components. In this thesis we are focused in a first application of our strategy to simultaneously infer the IMF and the SFH of the MW disc.

METHOD:

BGM FASt is based on a reweighing scheme, that uses a specific pre-sampled simulation. We use BGM FASt together with an approximate Bayesian computation algorithm to obtain the posterior probability distribution function of the inferred parameters, by automatically comparing synthetic versus observed data. Our full strategy is codified to run on Apache Spark and Hadoop, suited to deal with large surveys. BGM FASt is implemented in the big data infrastructure known as Gaia Data Analytics Framework (GDAF) at the University of Barcelona.

To evaluate the performance of BGM FASt we execute a set of validation tests comparing density, colour, mass and age distributions of BGM FASt versus BGM standard simulations. We present two scientific cases that compare synthetic versus Tycho-2 colour-magnitude diagrams. We obtain for the first time using BGM an IMF and SFH of the thin disc by exploring a 6-Dimensional parameter space.

We use Gaia data-release 2 magnitudes, colours, and parallaxes for stars with G<12 to explore a parameter space with 15 dimensions. This includes simultaneously the IMF and, for the first time, a non-parametric SFH for the Galactic disc.

RESULTS:

The set of tests applied show a very good agreement between equivalent simulations performed with BGM FASt and standard BGM. It has resulted to be 10000 times faster. We demonstrate it is a very valuable tool to perform multi-parameter inference using large catalogues. The two scientific demonstration cases of our strategy applied to Tyhco-2 data gives us, for the first time using BGM, a full 6D posterior probability distribution function of the parameters involved in the IMF and the SFH of the thin disc component.

Using Gaia DR2 we find an imprint of a star formation burst 2-3 Gyr ago in the Galactic thin disc domain. Our results show a decreasing trend followed by a Star Formation Rate (SFR) enhancement starting at about 5 Gyr ago and continuing until about 1 Gyr ago. This enhancement is detected with high statistical significance by discarding the null hypothesis of an exponential SFH. The timescale and the amount of stellar mass generated during this SFR enhancement event lead us to hypothesise that its origin, currently under investigation, is not intrinsic to the disc. When we adopt a non-parametric SFH the resulting IMF for the thin disc has an alpha3 of approximately 2 for masses M larger than 1.53 Msun and alpha2 approximately 1.3 for the mass range between 0.5 and 1.53Msun.

CONCLUSIONS:

BGM FASt has allowed us to increase our knowledge about the IMF and the SFH of the MW disc. Our results have shown that the evolution of the SFR with time is much more complex than a simple mathematical exponential decreasing shape. We have seen how the imposition of a mathematical shape for the SFH has a clear impact into the derivation of the IMF at high masses. The good performance of our whole strategy opens very promising perspectives, among them, the possibility to study whether the IMF variates with the time or not.