Organised by:
in collaboration with:



social media:


Accesso Utenti

Contact image
CUCEA - Universidad de Guadalajara, Zapopan
Miscellaneous Information:

Abstract Reference: 30248
Identifier: P1.9
Presentation: Poster presentation
Key Theme: 1 Reduction and Analysis Algorithms for Large Databases and Vice-versa

Computational Intelligence for Stellar Magnetic Fields Parameter Determination

Córdova Barbosa Juan P., Navarro Jiménez Silvana G., Ramírez Vélez Julio C.

Nowadays there are plenty of astronomical databases available, containing enormous quantities of raw data. Hence, analysis and automatic extraction of relevant information from these has become a crucially important task. In this work, we present the first results of applying an algorithm which enables automatic determination of a couple of parameters from polarized stellar spectra: effective temperature (Teff) and mean longitudinal magnetic field (Heff). Our method is based on supervised learning for artificial neural networks. For this purpose, for each stellar atmospheric model we first generated a synthetic database of polarized stellar spectra using the code COSSAM. The database consists of 200 different magnetic models each one corresponding to a different combination of the model parameters (7 free parameters). Then, we characterize the performance of the algorithm for the inference of the parameters of interest, Heff and Teff, at different levels of signal-to-noise ratio. Considering 10 different atmospheric models, a total of 2000 individual spectra were synthesized with the code COSSAM which were used to conduct the proper training of our neural network. In this work we will present the first results of the network performance under a supervised regime. Our final goal is to achieve a good efficiency in the code to retrieve the Heff and Teff parameters, to subsequently apply it to a big database of real objects (