Gaia ICAP-SA-VSWG workshop
Friday October 8 2004
Paris Observatory campus (Salle de l'Atelier, entrance at 77 Avenue Denfert Rochereau),
France. Meeting beginning 9h30
Provisional Programme
The meeting will have the form of a round table of Working Group objectives. The subjects will be briefly introduced, then the actions
will be planned.
Variable Star Working Group topics
- Variable star characteristics ( Juan Fabregat on Be stars )
- Compilation of surveys (Jan Cuypers)
- Estimation of the number of variable stars observed (Giuseppe Bono)
- Study of other future surveys
- Study of classification (Vasily Belokurov) see
Vasily's webpage
- Tools available
- New studies of sampling law
- Statistical tests
- Procedure for programme verification/validation
|
New subjects
- Introduction of variability in Becancon Galactic model (fraction of variable stars)
- Comparison of "time scales"/period search methods
- Visualisation of data
- Follow up observations from the ground (Michel Boer)
|
ICAP Discussion topics, draft, CBJ 30.08.04
- 1. What kind of top-level classification hierarchy should
we use for classification?
- hierarchical vs. parallel (see ICAP-CBJ-007)
- 2. How to overcome the the problem of systematic errors in AP
determination, in particular for the 'weak' APs, those which only
weakly effect the photometric data (e.g. logg, [Fe/H])
- 3. Extension of parametrization methods to determining alpha element
abundances
- from photometry and/or spectroscopy?
- which synthetic data can we use?
- 4. Determination of interstellar extinction
- star-by-star or using a 3D Galactic extinction model?
- a single extinction law (value of R_v) or distribution?
- 5. How to estimate uncertainties in APs derived from data
- statistical estimates based on a validation set?
- bootstrapping (ICAP-PW-004)?
- parametrizing posterior probability distribution?
- 6. Machine learning methods for multidimensional classification and
parametrization
- minimum distance / nearest neighbour methods
- performance of various method (see ICAP-AB-004)
- how to address the problem of degeneracy between APs
(see ICAP-CBJ-007)
- 7. How do we calibrate the physical parametrization models? Do we rely
entirely on synthetic spectra, or should these be adjusted based on
real spectra? If the latter, how should this be done? (see
ICAP-CBJ-002, ICAP-AB-002 and ICAP-CS-001)
- 8. We have focused on supervised parametrization schemes, i.e.
determining physical parameters for stars directly from the
data. These necessarily depend on specific stellar models.
Should we also consider:
- unsupervised schemes in order to find natural groupings
in the data
- empirical classification schemes, i.e. not directly going
to physical parameters, but rather to MK system-like
empirical parameters
- 9. Should we look also at morphological classification of image data,
e.g. to identify extended objects (mostly galaxies)? This may be
useful for onboard detection.
- 10. It is generally accepted that parallaxes should be included where
available in the AP determination. How should these be introduced
into the parametrization models? Should they rely on empirical or
theorerical mass-luminosity-radius relations (or a mix)?
- 11. How should we combine photometric and spectroscopic data
in the classifications?
- keep them entirely separate, i.e. two independent determinations?
- combine them on an equal footing, i.e. just expand the size of
the input data space?
- 12. How do we best detect astrometrically-unresolved binary stars from
their composite photometry? (We will have large numbers of such
objects, so detecting them is important for mass function studies
etc.) See ICAP-PW-003 and ICAP-PW-002.
- 13. How should we deal with peculiar stars?
- identify them as "non-normal" (outliers) and ignore them?
- attempt to classify them? If so, what
classification/parametrization scheme?
- 14. Should we incorporate additional non-Gaia data into the
classification for individual objects, such as 2MASS?
|
Last update (Laurent Eyer): October 12, 2004