Automated classification and stellar parameterization

Abstract

Different approaches for automated spectral classification are critically reviewed. We describe in detail ANN based methods which are very efficient in quick handling of the large volumes of data generated by different surveys. We summarize the application of ANN in various surveys covering UV, visual and IR spectral regions and the accuracies obtained. We also present the preliminary results obtained with medium resolution spectra (R ˜ 1000) for a modest sample of stars using the 2.3 m Vainu Bappu Telescope at Kavalur observatory, India. Our sample contains uniform distribution of stars in temperature range 4500 to 8000 K, log g range of 1.5 to 5.0 and [Fe/H] range of 0 to -3. We have explored the application of artificial neural network for parameterization of these stars. We have used a set of stars with well determined atmospheric parameters for training the networks for temperature, gravity and metallicity estimations. We could get an accuracy of 200 K in temperature, 0.4 in log g and 0.3 dex in [Fe/H] in our preliminary efforts.