MaStar Target Selection

Target selection for MaStar is a multi-tiered process that uses spectroscopically and photometrically determined parameters from previous catalogs to ensure that MaStar provides comprehensive coverage in stellar parameter space (T_eff, log(g), [Fe/H], [α/Fe], etc.). Broadly speaking, the aim of the target selection process is to avoid undersampling rare stars and avoid oversampling common stars. The selection also evolved during the six years of the survey as more information became available, such as the distance estimates based on Gaia data and the extinction estimates from the 3D dust map. As a result, the target selection is neither volume-limited nor magnitude-limited. We discourage the use of the library for statistical analysis. It is only meant to serve as a template library for stellar spectra.

In order to achieve this goal, we make use of a statistical weighting scheme that favors targets based on the number of stars with similar parameters available in MaStar’s footprint (i.e., “neighbors” in parameter space). Target candidates with fewer neighbors are favored over those with many neighbors. This inverse-density based weighting scheme is continuously adjusted over the course of the survey to give higher priority to under-dense regions of the parameter space.

Additionally, candidate targets are prioritized based on the reliability of their source parameters. In general, spectroscopically determined parameters (selected from APOGEE, SEGUE, and LAMOST) are much more accurate, and are favored over those that only have photometrically-determined parameters (selected from APASS and PanSTARRS1). Selection of targets with spectroscopically determined parameters is further optimized by prioritizing based on additional details of their source catalogs, such as spectral resolution, flux calibration, etc. In general, this means that APOGEE is favored over SEGUE, and SEGUE is favored over LAMOST.

In fields where stars with spectroscopically determined parameters are not available, stars with photometrically determined parameters are used to patch the hot and cold ends of the parameter space. This is necessary due to certain range restrictions of the spectroscopic catalogs.

Once Gaia data and Gaia-based distance estimates become available, we take advantage of the more accurate color and luminosity information to patch the underpopulated regions of Hertzsprung-Russell diagram. These include the very hot end of the main sequence, blue supergiants, red supergiants, carbon stars, cool subdwarfs, very cool dwarfs, white dwarfs, and extreme horizontal branch stars. We also identify and target likely members of open clusters and globular clusters based on position and proper motion measurements from Gaia.

The magnitude range of our targets also evolved during the survey. During the first 4 years, in order to ensure sufficient signal-to-noise (S/N) and avoid saturation, we only accept targets ranging between 12.5 and 17.5 in either g-band or i-band. This results in poor sampling of very luminous stars as those stars would have to be far out of the Milky Way or behind significant amount of dust to be sufficiently faint to be targeted. Thus, in the last 2 years, in addition to our standard exposure time (900 seconds), we adopted shorter exposure times, ranging from 28 seconds to 250 seconds, to improve the targeting of luminous stars.

We also rule out stars with nearby neighbors on sky to eliminate contamination from brighter sources. These limits are imposed by using photometry and astrometry from APASS and PanSTARRS1 initially, and from Gaia later, for every target (regardless of the source parameters used for the selection process discussed previously).