hide
Free keywords:
-
Abstract:
The Cannon is a flexible, data-driven spectral-modeling and parameter- inference framework, demonstrated on high-resolution, near-infrared SDSS-IV APOGEE spectra of giant stars to estimate stellar labels (Teff , logg, [Fe/H], and detailed abundances) to precisions higher than the APOGEE model-grid pipelines. The lack of reliable atmospheric models in the near infrared for temperatures cooler than ~3550 K motivates the extension of this approach to M-dwarf stars, but with calibrated training labels from optical surveys that overlap with APOGEE. Training two models (one for temperature/metallicity and one for spectral type) spanning temperatures 2860 < Teff < 4130 K, metallicities -0.5 < [Fe/H] < 0.5 dex, and classifications M0 to M9, we demonstrate predictive accuracies (in cross-validation) of 77 K, 0.09 dex and 0.9 subtypes, respectively. We apply our models to 10,311 sources selected from the overlap of APOGEE and Gaia DR2, and compare our Cannon-derived Teff against color-temperature relations, finding better agreement than pipeline measurements, and compare [Fe/H] against isochrone models in Gaia color-magnitude space. Finally we use empirically-calibrated relations to estimate radii from K-band absolute magnitudes and Cannon temperatures.