EDP Sciences
Free access
Volume 436, Number 1, June II 2005
Page(s) 379 - 390
Section Instruments, observational techniques, and data processing
DOI http://dx.doi.org/10.1051/0004-6361:20041974

A&A 436, 379-390 (2005)
DOI: 10.1051/0004-6361:20041974

Analysis of medium resolution spectra by automated methods - Application to M 55 and $\omega$ Centauri

P. G. Willemsen1, M. Hilker1, A. Kayser1, 2 and C. A. L. Bailer-Jones3

1  Sternwarte der Universität Bonn, Auf dem Hügel 71, 53121 Bonn, Germany
    e-mail: willemse@astro.uni-bonn.de
2  Astronomisches Institut der Universität Basel, Venusstrasse 7, 4102 Binningen, Switzerland
3  Max-Planck-Institut für Astronomie, Königstuhl 17, 69117 Heidelberg, Germany

(Received 8 September 2004 / Accepted 8 February 2005 )

We have employed feedforward neural networks trained on synthetic spectra in the range 3800 to 5600 Å with resolutions of ~2-3 Å to determine metallicities from spectra of about 1000 main-sequence turn-off, subgiant and red giant stars in the globular clusters M 55 and $\omega$ Cen. The overall metallicity accuracies are of the order of 0.15 to 0.2 dex. In addition, we tested how well the stellar parameters $\log\,g$ and  $T_{\rm eff}$ can be retrieved from such data without additional colour or photometric information. We find overall uncertainties of 0.3 to 0.4 dex for $\log\,g$ and 140 to 190 K for  $T_{\rm eff}$. In order to obtain some measure of uncertainty for the determined values of [Fe/H], $\log\,g$ and  $T_{\rm eff}$, we applied the bootstrap method for the first time to neural networks for this kind of parametrization problem. The distribution of metallicities for stars in $\omega$ Cen clearly shows a large spread in agreement with the well known multiple stellar populations in this cluster.

Key words: globular clusters: individual: $\omega$ Centauri -- globular clusters: individual: M 55 -- stars: fundamental parameters -- methods: data analysis -- techniques: spectroscopic -- stars: abundances

SIMBAD Objects

© ESO 2005