search

annayqho / TheCannon

a data-driven method for determining stellar parameters and abundances from stellar spectra
MIT License
39 stars 16 forks source link

Purge training set of objects w/ sketchy APOGEE labels #51

Closed annayqho closed 9 years ago

annayqho commented 9 years ago

From the Holtzman et al. paper: -- Warning on Teff for giants outside the range of those used to derive the calibration (3500 < Teff < 6000) -- COLORTE_BAD means that the deviation between the ASPCAP temperature and the col-temp relation from another paper is > 1000K -- STAR_BAD: "stars with this bit should not be used for parameter/abundance analysis" -- In the allStar file, "TEFF, LOGG, PARAM_M_H, PARAM_ALPHA_M" are the recommended values for these quantities -- STAR_BAD bit: "stars with unreliable parameters." STAR_WARN uses less stringent criteria. -- "ASPCAP results at cooler temperatures Teff < 4000 K may be more uncertain" -- Before using an alpha-enhancement value, confirm that it's not near a grid edge, which would be indicated by the ELEMENTFLAG bitmask in which GRIDEDGE_BAD or GRIDEDGE_WARN is set -- Below Teff < 3800, CALRANGE_WARN is set in the ELEMFLAG bitmask because the calibration is uncertain -- In STARFLAG, the persistence bit indicates uncertain abundances -- Abundance measurements are less certain at cool (Teff<4000) warm stars and logg < 0 -- ALPHAFE_WARN, M_H_WARN, METALS_WARN

davidwhogg commented 9 years ago

These comments / cuts should be put into a function that applies them. The function should be commented so it is fully human-readable, and the comment header should cite out to the Holtzman paper.