Kpca-Ent:A New Method for Automatic Measurement of Stellar Atmospheric Physical Parameters

Abstract

The sixth-period spectral observation task of the Large Sky Area Multi-Object Fibre Spectroscopic Telescope (LMAOST) had been completed, and the sixth dataset had been released in public named LAMOST-DR6, and then massive amounts of data are in urgent need of automated methods to improve the processing efficiency. In this paper, an automatic method named Kpca-Ent is proposed to measure the stellar atmospheric physical parameters—effective temperature Teff, surface gravity log g and metallicity [Fe/H] based on the spectral information. This method consists of three steps mainly. Firstly, we do the preprocessing for each spectrum, which includes denoising and normalization. And secondly, the kernel principal component analysis algorithm is used to extract the features which are highly relevant with the stellar atmospheric physical parameters. Finally, we apply the elastic-net regression model which has been trained to measure the three stellar parameters, here the extracted features from last step are as the inputs of this model. Furthermore, in order to test the effect of this method, we select 30,000 stellar spectra from LAMOST-DR6 randomly to do a series of experiments. And our method is capable of delivering stellar parameters with a precision of 66.99 K for Teff, 0.1193 dex for log g, 0.0773 dex for [Fe/H]. The results show that this method can measure the stellar parameters based on the stellar spectra accurately and effectively.