| Abstract: In my thesis, we investigated the application of machine learning techniques to the analysis of stellar photometric and spectroscopic data, with particular emphasis on RR Lyrae stars and stellar parameter estimation. We developed artificial neural network (ANN) based interpolators to generate theoretical light curves of RRab stars in the V and I bands from a precomputed grid of pulsation models. The trained ANN interpolators were used to predict light curves of RRab stars in the Magellanic Clouds, and a good agreement was found between the predicted and observed light curves. This approach provided a fast and efficient method for constructing smooth model grids over a wide range of physical parameters.
We also carried out an extensive photometric study of RR Lyrae stars in the globular cluster M3 using 3140 optical CCD images spanning 35 years from multiple astronomical data archives. Periods of 238 RR Lyrae stars, including 178 RRab, 49 RRc, and 11 RRd variables, were rederived using multiband periodogram analysis. We derived the distance to M3 using Period Wesenheit relations calibrated with theoretically predicted relations from the literature. The physical parameters of 79 non-Blazhko RRab stars were determined through ANN-based comparisons between observed and theoretical light curves.
In addition, in my thesis, we applied ANN techniques to spectroscopic data to derive atmospheric parameters of stars in the globular cluster NGC 6397. The ANN was trained to interpolate within the parameter space of the Göttingen Spectral Library and applied to MUSE spectra obtained at the Very Large Telescope using the ULySS analysis framework. The derived atmospheric parameters were in excellent agreement with previously published results. |
