next up previous index
Next: P4.3 Intensity Scaling and Up: Session P4. Data Analysis Previous: P4.1 Adaptive, Lossless Compression   Author Index


P4.2 An application of the coordinate descent algorithm to the determination of fine structure in the mass-luminosity relation and to the explanation for the nature of ``overmassive'' stars

Oleg Malkov, Dana Kovaleva, Anatoly Piskunov (Institute of Astronomy, Moscow)

The main objectives of the project are to ``improve'' observed data on masses and/or luminosities of binary components (within the inherent uncertainty limits imposed by observation/calibration errors), and to improve the precision of the mass-luminosity relation for moderate mass stars. This can be done by examining stellar models and taking into account the evolutionary status of the components, effects of chemical composition, and the constraints arising from the common origin of the components (equal age and chemical composition).

The task has been solved by using observational data from the Catalogue of Astrophysical Parameters of Binary Stars and current theoretical stellar models for different Pop I chemical compositions. We have designed a technique for simultaneous minimization of the discrepancies with observed masses/radii (within the observational errors), while assuring equal ages and chemical compositions of binary system components. Coordinate descent and other algorithms were used for optimization of observed and theoretical parameters.

As another application, the technique was used for decomposition of some low-mass binary components suspected of being unresolved. To explain the shift of low-mass binary components from their ``normal'' positions in the mass-luminosity plane, we calculate the most probable mass and luminosity values of hidden components. Some observational parameters are also estimated for the components.


next up previous index
Next: P4.3 Intensity Scaling and Up: Session P4. Data Analysis Previous: P4.1 Adaptive, Lossless Compression   Author Index
adass@ncsa.uiuc.edu