Determination of stellar fundamental parameters using Artificial Intelligence techniques.

1
Determination of stellar fundamental parameters
using Artificial Intelligence techniques. Luis
M. Sarro. Dpto. Inteligencia Artificial, UNED /
Spanish VO, Madrid, Spain Enrique Solano.
INTA-LAEFF / Spanish VO, P.O. Box 50727, 28080
Madrid, Spain
Understanding the physical processes of stars
requires the determination of a number of
fundamental parameter. The advances in
astronomical instrumentation as well as in
computational capabilities have provided the
means to make, for the first time, accurate and
precise comparisons between extensive libraries
of synthetic spectra and large observational
databases covering wide wavelength ranges at high
resolutions. In this framework the classical
methodology, where a high degree of human
intervention is present, has proven to be quite
inappropriate and automatic methods constitute
the only approach to guarantee repeatability and
high speed in the determination of the stellar
fundamental parameters. In this poster we
describe the techniques used for the
determination of the effective temperature,
surface gravity, metallicity and projected
rotational velocity of the star with planet
HD142. A comparison with the results obtained
by different authors using detailed analysis
demonstrates the potential capabilities of our
methodology.
Methodology and Results
Bayesian ensemble of Artificial Neural Networks


Determination of stellar physical parameters in
the framework of the Virtual Observatory
Determinations from detailed analysis
A critical issue to ensure efficiency in the
determination of the stellar parameters from
spectroscopy is the accessibility of the
theoretical models. In the context of the
Virtual Observatory, the Spanish VO is playing a
leading role in the definition (in collaboration
with ESA-VO) and implementation of VO-compliant
access protocols for theoretical datasets, in
particular for stellar model atmospheres. We have
already implemented a VO access protocol for the
ATLAS9 Kurucz model atmospheres and we are in
contact with a number of developers and
publishers of libraries of synthetic stellar
spectra for their adaptation to the Virtual
Observatory standards and requirements. In a
medium term, our system will be implemented as a
web service allowing the user to choose in a
straightforward way the theoretical models and/or
libraries with which to compare the observations.

• REFERENCES
• Hobson, Bridle Lahav, 2002, MNRAS, 335,
  377-388
• Hornik, Stinchcombe H. White, 1989, Neural
  Networks, 2, 359--366

Comparison between the observed spectrum and
the best model.