The New Milky Way

Principal investigator: Sofia Feltzing, Lund University
Co-investigator: Paul Barklem, Andreas Korn and Nikolai Piskunov, Astronomy and Space Physics
Project title: The New Milky Way
Grant amount: SEK 34 000 000 during six years

In the last twenty years our understanding of the Galaxy has increased immensely. This important progress in our knowledge has impact not only for our understanding of the local universe but in our studies of the universe as a whole. The European Gaia-satellite mission, due for launch in 2013, will provide us with a truly enormous catalogue of positions and velocities of stars in our Galaxy. This will initiate a paradigm shift in both our understanding of the Milky Way as a galaxy as well as how we do Galactic research. Astronomers at Lund University have been instrumental in realising the Gaia-satellite mission and Sweden has invested heavily in the pre-launch preparations. In order to fully exploit the fantastic opportunity offered by Gaia, complementary ground-based spectroscopic observations are necessary. Our collaboration of astronomers at Lund and Uppsala Universities are in a unique position to take a leading role in this ground-based follow-up because of the combination of in-house knowledge of Gaia, our world-leading expertise in analysis of stellar spectra, and established position in Galactic astronomy. In this application we are asking for funding to exploit the Gaia catalogue and to develop revolutionary new analysis tools for large-scale, ground-based spectroscopic surveys. We also request funds for hardware for (part of) a European multi-object spectrograph.

Galaxy formation and evolution is a key topic in contemporary astrophysics. Understanding the formation and evolution of galaxies is foremost about understanding how the baryons distribute themselves within our universe, where the large scale structures are dominated by dark matter. As baryons largely reside in stellar disks in galaxies, understanding these disks is the key to solving the question of galaxy formation across the ages of the universe. Over recent years, significant progress has been made by studying galaxy evolution across space and time. This has been done through major extra-galactic observational efforts (e.g., Kauffmann et al. 2003, Tremonti et al. 2004) as well as through the development of highly successful numerical simulations of large pieces of the universe (e.g., the Millennium simulation, Springel et al. 2005). These efforts have shown the universe to be a highly dynamical place where individual galaxies undergo many accretion and merger events. In the late 1980’s and early 1990’s, the Milky Way was seen as a rather static galaxy, built from a few large and simple building blocks – the disk, the bulge, and the halo. Recently, this picture has undergone dramatic changes.

Although the change started earlier, the shift from a static to a dynamical Milky Way took on a new dimension when in 1994 a small group of researchers trained their multi-object spectrograph on star fields in the direction of the Galactic bulge. The bulge was thought to be a simple spherical concentration of stars. To their surprise they found, in addition to the expected normal distribution of stellar velocities, a component with a completely different velocity. These stars were found to be brightest in a small galaxy in the direction of the stellar constellation Sagittarius (Ibata et al. 1994). This galaxy is being ripped apart by the Milky Way. Its debris has now been traced across
the whole sky (Belokurov et al. 2007) and is direct evidence that late accretion happens in regular galaxies, like the Milky Way.

With the advent of the Hipparcos satellite and its catalogue of stellar positions, parallaxes, and proper motions for 118,218 stars (Perryman, Lindegren et al. 1997), the next steps in fully appreciating the dynamical nature of the Milky Way were taken. It became feasible to select stars based on their motion within the Milky Way for spectroscopic studies of elemental abundances. Interestingly, these studies of Hipparcos stars have revealed that disk stars that are kinematically hot (have large non-circular velocity components) have different elemental abundances compared to stars on orbits similar to that of the sun. This indicates different histories for these two stellar populations (Bensby et al. 2003, 2004; see also Bovy et al. 2012 for a different view based on a different type of data). Apart from accretion events in the halo and complex disk structures, we have also found that the central part of the Milky Way is not a spherical bulge. Indeed, it hosts a bar and complex stellar populations, perhaps even including an unexpectedly young stellar population (Bensby et al. 2013).

This, however, barely scratches the surface of the immensely rich stellar system that makes up our Milky Way galaxy. For example, the studies that find differences between the stellar disks are based on small samples of a few hundred stars in the solar neighbourhood (within 100 pc) and we have essentially no detailed elemental abundances at all for stars between us and the Galactic bulge, i.e. a full 8 000 pc (or 24 000 light years) stretches out in a virtual knowledge desert where we have little information about the properties of the stellar disk(s). As the understanding of formation and evolution of galaxies in a cosmological context requires understanding of the stellar disks, this is a severe handicap to any further progress. With the project applied for here, we aim to overcome this limitation.

The Gaia satellite and ground-based follow-up

The European Gaia satellite (first proposed by Lindegren & Perryman, 1996) is the most important, currently-funded, project in Galactic astronomy. Gaia will provide highly accurate parallaxes and proper motions for a billion objects across the entire sky. Gaia will give some astrophysical information and radial velocities for the brightest 15% of the billion objects. Elemental abundances will, however, only be available for the brightest 0.3% of the billion stars. To make a major breakthrough in understanding the evolution of the Milky Way and its components, detailed chemical information as well as radial velocities for as many stars as possible are essential. This can only be achieved through complementary, ground-based observations using new multi-object spectrographs, where the spectra of up to a thousand stars over several square degrees of the sky can be taken in parallel. No such facilities exist today. By combining positional information from Gaia combined with low-resolution spectra we obtain the full three dimensional velocities. High-resolution spectra provide the stellar parameters (e.g., surface temperature) as well as all the crucial elemental abundances from which the star formation and assembly history of the Milky Way can be re-constructed (see below).

Chemical clocks

The history of a galaxy is imprinted in its stars: not only in their positions and velocities, but in their chemical compositions. This chemical imprint is more robust and samples significantly earlier times than stellar motions. The elemental abundances in atmospheres of long-lived stars remain largely unperturbed over time and act as time-capsules showing the mixture of elements present in the gas from which the stars formed. Different elements are released to the interstellar medium by stars of different masses and therefore on different timescales. Thus, the abundance ratios measured in stars provide cosmic clocks enabling the reconstruction of the history of star formation and gas accretion for the Milky Way. The classical clock is oxygen abundance relative to iron. Oxygen is produced in massive stars and expelled through core-collapse supernovae into the interstellar medium on short timescales. Iron is mainly produced in supernovae type Ia, which expel material on much longer timescales. To obtain the relevant elemental abundances we require a high-resolution multi-object spectrograph that can observe statistically interesting samples of stars with relatively short observing times.

This proposal

A grant from the Knut and Alice Wallenberg foundation for this research will enable a significant and unique contribution from Swedish researchers to a deeper understanding of the Milky Way, our home in the Universe, and to our understanding of galaxy formation in general.

Research environment and implementation

Project implementation

Sofia Feltzing will devote 50% of her time to The New Milky Way project, assuming the overall management and leading projects. Her contribution is fully financed by Lund University, as is that of Co-PI Lennart Lindegren. The Co-PIs in Uppsala will commit 3.2 FTE (full-time equivalent) to the project, of which half is financed by Uppsala University. The final Co-PI, Thomas Bensby, is a Swedish Research Council-funded researcher 2011-2014 with no guaranteed funding beyond 2014. We apply for 80% of his salary 2015-2018. The Co-PIs are joined by the expert team members. They will contribute 7.5 FTE, of which Lund and Uppsala Universities pay 2.6 FTE and 1 FTE is provided through a Marie Curie grant (details in budget text). In order to fully implement the project, however, additional manpower is required. We are not able to carry out these ambitious projects with the staff listed; their expertise is key to the success of the project, but additional resources are necessary. We estimate that three two year postdocs and one four year PhD student are needed as additional man-power.

Three Multi-Object Spectrographs (MOS)

As a response to ESO’s call in 2010, two proposals for the next-generation of multi-object spectrographs emerged: 4MOST (de Jong et al. 2012) and MOONS (Cirasuolo et al. 2012). A third proposal for a Northern spectrograph, WEAVE, was put forward by British and Dutch astronomers. In May 2013, ESO will select one of 4MOST and MOONS for implementation. Funding decisions for WEAVE will be taken at a similar time. As Feltzing, Korn, and Ryde are members of all three science teams, our involvement in a multi-object spectrograph to provide ground-based follow-up for Gaia is secured. We stress that the The New Milky Way project is independent of which spectrograph is implemented. We also stress that the selection of instrument will happen spring 2013 and hence will be a known factor to the KAW foundation at the time of funding decisions.

4MOST is a massively multi-plexed spectrograph for ESO’s 4-meter VISTA telescope on Paranal, Chile. The baseline design is a Field-of-View of 3 square-arcminutes (goal: 5). It will have one low-resolution spectrograph with 1 500 fibres (goal: 3 000) and a high-resolution component with 150 fibres (goal: 300). The Galactic project will survey 10 000 square degrees (goal: 20 000) over 5 years. In total 6 – 20×10⁶ stars will be observed. The project is led by R. de Jong, AIP, Potsdam.

WEAVE is the Northern hemisphere counterpart to 4MOST; to be placed on the 4-meter WHT telescope on La Palma, Spain. Many technological solutions, survey implementation, and data analysis will be transferred from 4MOST. The project is led by G. Dalton, RAL, Oxford.

MOONS is a NIR spectrograph that can look into dust-obscured regions of the Galaxy. It will be placed on ESO’s 8-meter VLT on Paranal, Chile. The large aperture allows deeper observations at the cost of covering a smaller portion of the sky. The Field-of-View is 0.14 square-arcminutes. It will have a low-resolution spectrograph with 500 fibres and a high-resolution component with 500 fibres. The Galactic project will survey 1 000 square degrees over 5 years. In total 2×10⁶ stars will be observed. The project is led by M. Cirasuolo, ATC, Edinburgh.

Detailed project plan

The New Milky Way project concerns the formation and evolution of the Milky Way and the primary tools needed to read the history of the Galaxy. The necessary data are obtained through the following steps:

1. Positions and proper motions of stars: Gaia-satellite.
2. Radial velocities for a statistically significant fraction of the 85% of the one billion stars observed by Gaia with no radial velocities from the mission itself. For this a low-resolution multi-object spectrograph is required.
3. Elemental abundances for a statistically significant fraction of the stars brighter than V=16. Although the low-resolution spectrograph allows crude abundance determinations (“metallicities”), a high-resolution multi-object spectrograph is required for precise elemental abundances for statistically interesting samples.
4. In addition, dedicated analysis software must be developed. This also includes further research into the understanding of the stellar atmosphere as an astrophysical phenomenon.

The New Milky Way project focuses on items 3 and 4 in this list.

A) Unravelling the formation of the Milky Way – Advanced stellar abundance analysis techniques for large surveys

For a full understanding of the chemical and dynamical evolutionary processes that have shaped the Milky Way, we require samples of stars with precise chemical abundances and kinematic information. These samples must be very large in order to distinguish the four known Galactic components (bulge/bar, halo, thick and thin disc) and sufficiently sample these components both throughout the Galaxy in space and throughout their complete chemical history; i.e. sample locations along and perpendicular to the Galactic plane, as well as cover the whole metallicity range. The expected differences in abundances which need to be resolved to distinguish these components are often below 0.2 dex, and it is thus crucial to obtain abundances with a precision better than 0.1 dex (Lindegren & Feltzing 2013). Such information will permit us to disentangle the formation paths of each of the Galactic components and to understand their interrelationship. New structures are likely to be discovered, especially if even higher precision can be attained.

Gaia will provide the precise kinematic information needed, and so the possibility to accurately distinguish populations and associate stars to common formation sites rests with the ability to obtain precise elemental abundances for large samples of stars from the analysis of high-resolution spectra. The Lund-Uppsala collaboration, together with international collaborators, is particularly well placed to tackle the challenge of developing the next generation of abundance analysis tools. Our longstanding track record of leading the development in modelling stellar atmospheres and associated analysis tools (e.g., Gustafsson 1974, Edvardsson et al. 1993, Asplund et al. 1997, Gustafsson et al. 2008; Valenti & Piskunov 2012); many ground-breaking studies in abundance studies of the Milky Way (e.g., Edvardsson et al. 1993, Bensby et al. 2003, 2004, Barklem et al. 2005, Korn et al. 2006, Luck & Heiter 2006); theoretical and laboratory work on relevant atomic data and their application (e.g., Barklem et al. 2010, Hartman et al. 2010, Lind et al. 2010) in combination with our deep involvement in the abundance analysis of the first major high-resolution spectroscopic survey, the Gaia-ESO Survey, mean that we are in a very strong position ready to develop the relevant tools for detailed abundance analysis for large spectroscopic surveys.

The New Milky Way will cement the world-leading positions of our separate groups, as well as position Swedish Milky Way researchers as a strong, coherent research unit, which is seen as a natural partner in pan-European and international endeavours such as 4MOST and spectroscopic instruments for the European extremely large telescope (E-ELT). The New Milky Way will also guarantee that Sweden gains maximal scientific returns from its investments in the Gaia satellite.

The realization of competitive abundance analysis tools for large spectroscopic surveys requires advances on two fronts:
1) Understanding of the stellar atmosphere as an astrophysical phenomenon Advances in the understanding and modelling of stellar atmospheres and their spectra are required to achieve the desired precision in stellar abundances across such large samples. Systematic errors presently exceed 0.1 dex (see, e.g. Asplund 2005). These errors are due to a) uncertainties in atomic and molecular data used in the models, and b) simplifying assumptions of classical models (e.g., local thermodynamic equilibrium (LTE), 1D, and plane-parallel geometry). The systematic errors often vary significantly with stellar parameters, especially metallicity, and thus cannot be appreciably mitigated by using differential analysis techniques in large-scale projects such as the groundbased Gaia follow-up. Our collaboration has significant, often world-leading, expertise regarding atomic data and processes relevant to stellar spectra, non-LTE modelling of stellar spectra, and 3D-modelling of stellar atmospheres and their spectra. To improve the modelling to the required level for all elements of interest across a wide stellar parameter space is indeed extremely challenging, labour intensive, and requires significant infrastructure, especially computing. For example, building an accurate non-LTE model of spectral line formation for a single element currently requires months of full time work to calculate and/or assemble relevant atomic data, implement and test the model in radiative transfer codes, and finally calculate model spectra for a wide range of stellar parameters. However, the timescales for such work are decreasing. With sufficient resources it is reasonable to believe that methodical efforts on advanced modelling will result in abundances with systematic errors 0.05 dex for astrophysically important elements within the coming years.
2) Dedicated analysis software for large surveys Development of the methodology for robust, automated analysis of large samples (millions) of stars is needed. Stellar spectra are complex and difficult to interpret even for experienced spectroscopists, especially in the case of spectra with moderate signal-to-noise ratios (as will be the case for ground-based Gaia follow-up and is usually the case for ground-breaking projects with state-of-the-art telescopes pushing the limits to fainter and fainter stars). The development of theory, algorithms, and codes for automated methods for analysis by spectrum synthesis is in its infancy, but our groups have been at the forefront of this development. In particular, we have developed the Spectroscopy Made Easy (SME) program (Valenti & Piskunov 1996 and 2012) and its applications: e.g., in Barklem et al. (2005) and the Gaia-ESO Survey where we already are extending SME beyond classical models, e.g., including NLTE effects. The Gaia-ESO Survey, with its 10⁵ stellar spectra, provides us with an excellent opportunity to test the methodologies needed to analyse millions of stellar spectra. This hands-on experience combined with our excellence in the understanding and analysis of stellar spectra puts us at a particular advantage to make a ground-breaking contribution to future large-scale surveys focusing on Gaia follow-up.

Team and implementation

Barklem and Feltzing will assume overall leadership. Crucial contributions to aspects of the project will come from Korn, Lind, Bensby, Bergemann, Edvardsson, Eriksson, Hartman, Piskunov, and Ryde. This project involves significant international collaboration, especially with Asplund’s groups at MPA (Bergemann) and ANU. Implementation, testing, and application of models and analysis codes will be done both at Lund and Uppsala. The project will be supported by two postdocs as well as the expert researchers. Lund University provides access to the necessary multi-core computing facilities through Lunarc.

B) Exploiting the first Gaia data to start unravelling the Milky Way

The Gaia mission ends 2018 with the final catalogue arriving 2020. However, already 28 months after launch the first parallaxes, transverse motions, radial velocities, and photometry will be released. The precision of parallaxes and transverse motions will be a factor 2-3 less good than the final data, but still a huge improvement as it provides distances to a billion new objects. For 10⁵ of these stars the Gaia-ESO Survey (Gilmore et al. 2012) will by then provide ground-based radial velocities enabling full 3D kinematics to be derived. The Gaia-ESO Survey is currently observing using the high-resolution GIRAFFE and UVES spectrographs on ESO’s VLT in Chile. With UVES thousands of stars within 1 kpc of the sun are observed, while the GIRAFFE observations cover 10⁵ turn-off stars in the stellar disk(s) and halo at larger distances and turn-off and giant stars in 100 open clusters. We lead one of the abundance analysis groups in the survey. The Gaia-ESO Survey will do an important differential study of the stellar properties in the disk. Due to the short wavelength coverage of GIRAFFE the surface gravity is poorly constrained by the spectral analysis. Adding the first Gaia parallaxes to these data will completely change the situation. With precise astrometric distances, reliable surface gravities will be derived and the abundance analysis greatly improved, increasing the precision. The distances will provide age determinations for all turn-off stars – such a large sample of turn-off stars with known properties has never been studied before. These data are a rich ground for new discoveries and will pave the way to develop new interpretational methodologies for large-scale studies (e.g., Bovy et al. 2012, Liu & van de Ven 2012). This exciting data-set will provide the basis for many studies. We are in particular planning work along two lines of investigation:
1) Stellar populations in the disk The presence or lack of radial gradients in elemental abundances and ages in the stellar disks provide fundamental constraints on models of galaxy formation and evolution (e.g., Spitoni & Matteucci 2012). The re-analysed data from the Gaia-ESO Survey provides the first comprehensive data-set which does include both ages as well as detailed elemental abundances for dwarf stars from which a detailed study of radial variations can be conducted. In addition, the improved distances and space motions of open clusters will allow a first mapping of their elemental abundances in combination with their motion in the disk. An in-depth comparison of cluster and surrounding field stars will for the first time put the open clusters properly into context and answer the question how well the open clusters are able to trace the underlying stellar population in the disk, e.g., with respect to radial abundance gradients.
2) Improving the astrometric solution Astrometry normally estimates five of the six astrometric parameters accurately. By including independent radial-velocity data from another instrument the sixth component can be determined and may improve the astrometric solution due to parameter correlations. Developing such combined solutions for Gaia is of great interest.

Team and implementation

The team will be led by Lindegren and Bensby. Hobbs will lead the work on the astrometric solution. With our in-house knowledge on the quality of the Gaia data (Lindegren, Hobbs, Korn) and our good understanding of the spectroscopic data from the Gaia-ESO Survey (Korn, Bensby, Lind, Heiter, Feltzing, Ryde) we are ideally positioned to take advantage of the first Gaia data. One post-doc and one PhD student will be assigned to these projects.

C) Participation in a consortium for the realisation of a multi-object spectrograph with a high-resolution component

The baseline design for the next-generation multi-object spectrographs for Gaia follow-up is a low-resolution spectrograph that provides radial velocities for tens of millions of stars matching the quality of the transverse motions measured by Gaia. This will enable a kinematic decomposition of the Milky Way in a way never done before. Furthermore, a high-resolution component is required to obtain detailed elemental abundances for a subset of stars for the full chemodynamical view of Milky-Way evolution. Abundance ratios provide the necessary time variable. We have repeatedly shown that elemental abundances with sufficient precision are key observables to dissect the solar neighbourhood into the different stellar populations (Edvardsson et al. 1993; Bensby et al. 2003, 2004; Lindegren & Feltzing 2013).

Members of our team have a strong track-record in development of data reduction techniques that are of central importance to the next generation of massively multi-plexed spectrographs, e.g., optimal extraction with removal of fiber crosstalk (Piskunov & Valenti, 2002). In addition, the sheer amount of spectra from such spectrographs will require automatic quality control, database-style archiving of intermediate and final products with cross-reference keys between calibrations and science data. In this respect it is not dissimilar to Gaia itself and hence the deep involvement by members of our team in the data-processing of the Gaia data will provide an invaluable reference for the development of the data-reduction tools to be developed for the new spectrograph.

The final spectroscopic survey will be realised after the end of this project. The aim is to have the instrument on the telescope in 2017/18 and the Galactic surveys will run over 5 years. As members of the instrument consortium, we are in an excellent position to do ground-breaking research on the assembly history of the Milky Way (notably the formation of the thick disk), on extremely metal-poor stars, which will be identified and studied in quite some number with this survey, and on the chemo-dynamical nature of the Galactic bulge. Although the data will eventually be public, being inside the consortium will have the great advantage of first-hand access and understanding the instrument-specific uncertainties and how to mitigate them in the analysis. Precise Gaia data will be available by the start of the survey, providing an unprecedented basis for selection of targets for statistical studies of specific volume elements of the Milky Way.
Cost: It is estimated that a high-resolution spectrograph in addition to the baseline low-resolution spectrograph will incur a minimum cost of 1.5 M€, likely not larger than 3 M€, depending on the exact construction (see also budget text). A Swedish contribution of 15 MSEK will help to realize the all important high-resolution mode for one new multi-object spectrograph.

Team and implementation

For our team, the actual work in relation to the spectrograph will not start until 2016 when work on dedicated data-reduction software will be ramping up. This is one of our areas of unique expertise (Piskunov). This timescale fits naturally with the ones for on-going instrument-related projects (notably Gaia and CRIRES) thus freeing university-funded staff to put more time into this project. The PI will lead the project for the first years to be replaced by Piskunov and Korn. They will be joined by expert team members as required.

Return on investment

In return for the investment of funds and human resources, we will be able to lead one of the science cases carried out with the high-resolution spectrograph. In addition, we will be well placed to make significant contributions to the other science cases, both with respect to detailed elemental-abundance studies as well as kinematic studies coupled to Gaia. In summary, we are convinced that the return on investment of this project is high: for European leadership in optical astronomy and the Lund-Uppsala team of astrophysicists alike.

References

Asplund, M., 2005, ARA&A 43, 481
Asplund, M., Gustafsson, B., Kiselman, D., Eriksson, K., 1997, A&A, 318, 521
Barklem, P.S., Christlieb, N., Beers, T.C., Hill, V, et al., 2005, A&A, 439, 129
Barklem, P.S., Belyaev, A.K., Dickinson, A.S., Gadéa, F.X., 2010 A&A 519, A20
Belokurov, V., Zucker, D.B., Evans, N.W., Gilmore, G., et al., 2006, ApJ, 642, L137
Bensby, T., Feltzing, S., Lundström, I., 2003, A&A, 410, 527
Bensby, T., Feltzing, S., Lundström, I, 2004, A&A, 415, 155
Bensby, T., Yee, J.C., Feltzing, S., Johnson, J.A., et al., 2013, A&A, 549, A147
Bovy, J., Rix, H.-W., Liu, C., Hogg, D.W., et al., 2012, ApJ, 753, 148
Cirasuolo et al. (incl. Feltzing, Ryde), 2012, SPIE, 8446, also arXiv:1208.5780
de Jong, R., et al. (incl. Feltzing, Korn), 2012, SPIE, 8446, also arXiv:1206.6885
Edvardsson, B., Andersen, J., Gustafsson, B., Lambert, D.L., et al., 1993, A&A, 275, 101
Gilmore, G. et al. (incl. Feltzing, Korn), 2012, Msgr, 147, 25
Gustafsson, B., Bell, R.A., Eriksson, K., Nordlund, A., 1974, A&A, 42, 407
Gustafsson, B., Edvardsson, B., Eriksson, K., Jørgensen, U.G., et al., 2008, A&A, 486, 951
Hartman, H., Nilsson, H., Engström, L., Lundberg, H., et al., 2010, PhRvA, 82, 2512
Ibata, R.A., Gilmore, G., Irwin, M.J., 1995, Nature, 370, 194
Kauffmann, G. Heckman, T.M., White, S.D.M., Charlot, S., et al., 2003, MNRAS, 341, 33
Korn, A.J., Grundahl, F., Richard, O., Barklem, P.S, et al., 2006, Nature, 442, 657
Lind, K., Asplund, M., Barklem, P.S., Belyaev, A.K., 2011, A&A 528 A103
Lindegren, L & Feltzing, S., 2013, submitted to A&A
Lindegren, L. & Perryman, M.A.C., 1996 A&AS, 116, 579
Liu, C. & van de Ven, G., 2012, MNRAS, 425, 2144
Luck, R.E. & Heiter, U., 2006, AJ, 131, 3069
Perryman, M.A.C., Lindegren, L., Kovalesvsky, J., Høg, E., et al., 1997, A&A, 323, L49
Piskunov, N. Valenti, J.A., 2002, A&A, 385, 1095
Spitoni, E. & Matteucci, F., 2012, A&A, 531, A72
Springel, V., White, S.D.M., Jenkins, A., Frenk, C.S., et al. 2005, Nature, 435, 629
Tremonti, C.A., Heckamn, T.M., Kauffmann, G., Brinchmann, J., et al., ApJ, 613, 898
Valenti, J.A. & Piskunov, N., 2012, Astrophysics Source Code Library, record ascl:1202.013
Valenti, J.A. & Piskunov, N., 1996, A&AS, 118, 595 10