eStarSpec: using eScience to map the Milky Way

A key goal of contemporary astrophysics is to discover the history of our Milky Way galaxy: when and where did its stars form, and how did they come to be where they are today? To do that the Gaia satellite, funded by ESA, is measuring the distances to around a billion stars. To complete the data from Gaia we also need the stars’ radial velocities – their speeds measured along the line of sight – and their chemical compositions. Both can be derived from absorption lines that are visible when the light from a star is split up into its different wavelengths: the star’s spectrum.

Aims
The aim of this project is to produce a pipeline to analyse the spectra of stars that will be observed by 4MOST. 4MOST is an instrument being built for the VISTA telescope at the European Southern Observatory (ESO), and will observe spectra of around 10 000 stars in the Milky Way each night. This is a much greater data rate than any previous spectrograph, and so new tools need to be developed to analyse the spectra. Current approaches, even for large surveys, involve large amounts of hands-on work, and are not well suited for use in high-performance computer systems. We aim to produce a new pipeline, based on a development of current approaches, that is faster, reliable, and capable of autonomous operation in supercomputer centres.

Methods
The core of a project is a collaboration between groups at Lund and Uppsala universities. At Lund University we have experts on the use of high-performance computing in astrophysics. The galactic project scientist for 4MOST, professor Sofia Feltzing, is based at Lund. The group at Uppsala University are expert on detailed stellar spectroscopy, and the developers of one of the leading spectral synthesis codes, Spectroscopy Made Easy (SME). Working together, and starting with SME, we will develop a code suitable to analyse large numbers of spectra in an automated fashion in a computing centre. One idea that we intend to investigate early on is to use a data-driven technique. This has the potential to allow us to obtain the surface temperature and gravitational field strength of the star much more quickly than is possible with traditional approaches.

Research group

PI:Ross Church
Dept of Astronomy & Theoretical Physics, Lund University
Paul Barklem
Dept. Physics & Astronomy, Uppsala University
Melvyn B. Davies, Sofia Feltzing, Anders Johansen, Andreas Korn, Nikolai Piskunov
Dept. Astronomy & Theoretical Physics, Lund University
Andreas Korn, Nikolai Piskunov
Dept. Physics & Astronomy, Uppsala University

Links and references