MaStar Caveats

• This is not a caveat: please note the wavelengths for the spectra are given in vacuum.
• The MaStar spectra released in DR15/DR16 are NOT corrected for foreground dust extinction.
• All spectra have been corrected to the restframe according to the reported heliocentric velocity (HELIOV), regardless of whether the HELIOV derived is robust or not. To find out which HELIOV is considered unreliable, one needs to look in the GOODVISITS or ALLVISITS extensions of the mastarall file or in the mastar-goodspec file (or mastar-goodvisits table on SkyServer). Those entries that have a nonzero V_ERRCODE column, or equivalently have bit 6 (BADHELIORV) set in the MJDQUAL flag (see the bitmask table for MASTAR_QUAL here) are considered to have unreliable velocity measurements. There are also a number of entries whose V_ERRCODE are zero but have an velocity uncertainty (VERR) of 999.0 km/s. These are objects for which we have only one good exposure and cannot make an error estimate from multiple exposures. These velocities could also be problematic. A few cases have HELIOV being exactly zero for which no shift to restframe is done.
• The spectral resolution varies with wavelength. The resolution array is given along with each visit spectrum. We provide a description of how to deal with the changing resolution in Yan et al. (2019).
• The spectra do NOT have the same spectral resolution array --- they differ from spectrum to spectrum. The multiple visit spectra for the same star also can differ in their resolution arrays.
• A small fraction of the good quality visit spectra still contain emission lines. These could either come from HII regions around the star or from stellar flares. These can be identified as having bit 8 (EMLINE) set in the MJDQUAL flag (see the bitmask table for MASTAR_QUAL here).
• A small fraction of the good quality visit spectra may be affected by scattered light. They can be identified as those having bit 2 (HIGHSCAT) set in the MJDQUAL flag (see the bitmask table for MASTAR_QUAL here). This issue does not always affect the data significantly. Thus, we have ignored this bit when selecting the good quality visit spectra. If one notices strange features in the spectra, it would be worth checking if this bit is set.

We have discovered that the Pan-STARRS1 (PS1) photometry we provided for some stars are unreliable. They scatter away from the stellar locus on a color-color diagram. There are multiple reasons for this. For some objects, it is because we took the mean magnitude among multiple epochs of PS1 photometry. If one epoch is problematic, it could affect the mean value significantly. Switching to median would make a much more robust estimate. For some other objects, it is because they are variable stars and even the median may not necessarily give a consistent color. In addition, we were using a slightly outdated calibration. We have now switched to the latest calibration, switched to using median magnitudes. In this file, we list the corrected version of the PS1 magnitudes for all good stars with photometry coming from PS1. We also provide the error estimates for the median magnitudes which can be used to reject objects with unreliable photometry. One could apply a threshold of 0.01 or 0.03 mag on these magnitude errors to select a clean and robust sample, depending on how conservative one chooses to be. The photometry we provide is slightly different from those available in PS1 Data Release 1 due to a slight difference in calibration.

The following non-stellar objects are targeted by mistake. Some are still in the final summary files. Please exclude them in your science analysis.

• MaNGA ID 3-148801340 is a galaxy.
• MaNGA ID 4-11104 is a galaxy. It is already excluded from the GOODSTARS and GOODVISITS extensions of the mastarall file and the mastar-goodspec file.
• MaNGA ID 4-15433 is a galaxy. It is already excluded from the GOODSTARS and GOODVISITS extensions of the mastarall file and the mastar-goodspec file.

MaNGAID provides a convenient identifier to identify unique MaStar targets, and to identify all repeated observations associated with each unique star. Usually, each unique MaNGAID represents a unique star, except for those in the following lists.

• 27-1856 is the same star as 4-19555
• 4-17885 is the same star as 4-840
• 4-823 is the same star as 3-134064467
• 4-21098 is the same star as 4-20929
• 4-20872 is the same star as 4-20905
• 4-10681 is the same star as 3-12454116
• 4-12664 is the same star as 3-95914294