The objectives were addressed through three different approaches.
By taking spectra of stars we could measure their velocities and the amount of different elements each star contains using the characteristic features produced when each element absorbs starlight. We took spectra of stars located in the Milky Way's bar using the European VLT telescope in Chile. When combined with measurements from the Gaia satellite, we could see how each star was moving in three dimensions and, by comparing to models, we could see which of these stars were part of the Milky Way's bar. Interestingly, we found that the stars in the bar contained large quantities of iron. Because iron-rich stars formed more recently than their iron-poor counterparts, this is telling us how the bar formed and evolved. It could be that these younger stars in the bar formed in the bar, or it could be that the Milky Way's bar is growing, and capturing young stars as it does. Both scenarios are possible, and we hope that better data over the coming few years from a new generation of large surveys will tell us which narrative is correct.
To understand how the dark matter is distributed we used a rare and special type of star called RR Lyrae. The Gaia satellite has measured proper motions (meaning the motion across the sky) for more than a billion stars across the Galaxy. However, unfortunately for most of these stars we do not have the distances we need to turn these proper motions into real velocities. RR Lyrae are a type of pulsating star where we can estimate accurate distances. We therefore used a sample of almost 16,000 RR Lyrae measured by Gaia to measure the velocities of the stars which make up the Milky Way's stellar halo. We expect our galaxy to be evolving slowly, on a timescale of billions of years, and so we used these velocities to tell us about the forces holding our Galaxy together. For example if the velocities of the stars are high, then the gravitational forces must be strong, otherwise the Milky Way would not be held together and would fly apart in a relatively short timescale. We were therefore able to measure the gravitational forces across the Milky Way using this sample, and we could even do so in two-dimensions giving us a direction to forces at each point in space (see attached plot). The forces we found are too strong to be produced by the stars and matter that we can see, with the remainder produced by unseen dark matter. Interestingly, because the forces point nearly towards the galactic center, this means that the dark matter that surrounds us must be mostly spherical in shape.
The project also developed tools that will allow others to build on this work. A web-based tool that lets anyone make a mock survey of the Milky Way, taking fake measurements of a model Milky Way, will allow others both to design fruitful new surveys, as well as help to understand already taken data. Furthermore, a software package was developed that uses machine learning techniques that are normally used to train neural networks to categorize images to instead make models of the Milky Way. The model of the Milky Way is analogous to the neural network - it is trained until it can correctly predict the data that we take on the Milky Way. This package has been made open source to allow others to use and improve on its methods.
