A new method for the inversion of atmospheric parameters of A/Am stars

Abstract

We present an automated procedure that derives simultaneously the effective temperature $T_{eff}$, the surface gravity logg, the metallicity [Fe/H], and the equatorial projected rotational velocity vsini for "normal" A and Am stars. The procedure is based on the principal component analysis inversion method of Paletou et al. (2015a). A sample of 322 high resolution spectra of F0-B9 stars, retrieved from the Polarbase, SOPHIE, and ELODIE databases, were used to test this technique with real data. We have selected the spectral region from 4400-5000\AA\ as it contains many metallic lines and the Balmer H$\beta$ line. Using 3 datasets at resolving powers of R=42000, 65000 and 76000, about 6.6x$10^6$ synthetic spectra were calculated to build a large learning database. The Online Power Iteration algorithm was applied to these learning datasets to estimate the principal components (PC). The projection of spectra onto the few PCs offered an efficient comparison metric in a low dimensional space. The spectra of the well known A0- and A1-type stars, Vega and Sirius A, were used as control spectra in the three databases. Spectra of other well known A-type stars were also employed in order to characterize the accuracy of the inversion technique. All observational spectra were inverted and atmospheric parameters derived. After removal of a few outliers, the PCA-inversion method appears to be very efficient in determining $T_{eff}$, [Fe/H], and vsini for A/Am stars. The derived parameters agree very well with previous determinations. Using a statistical approach, deviations of around 150 K, 0.35 dex, 0.15 dex, and 2 km/s were found for $T_{eff}$, logg, [Fe/H], and vsini with respect to literature values for A-type stars. The PCA-inversion proves to be a very fast, practical, and reliable tool for estimating stellar parameters of FGK and A stars, and deriving effective temperatures of M stars.