Spectral classification of stars based on LAMOST spectra

TL;DR

This paper presents a spectral classification method for stars using LAMOST data, combining line indices and SVM algorithms, achieving up to 90% accuracy for certain star types and highlighting the advantages of line index-based classification.

Contribution

The study introduces a spectral classification approach that maps spectral line features to MK classes and compares SVM-based classification with line index methods.

Findings

Achieves up to 90% classification completeness for A and G stars.

SVM classification accuracy drops to about 50% for OB and K stars.

Line indices provide a natural, continuous stellar classification.

Abstract

In this work, we select the high signal-to-noise ratio spectra of stars from the LAMOST data andmap theirMK classes to the spectral features. The equivalentwidths of the prominent spectral lines, playing the similar role as the multi-color photometry, form a clean stellar locus well ordered by MK classes. The advantage of the stellar locus in line indices is that it gives a natural and continuous classification of stars consistent with either the broadly used MK classes or the stellar astrophysical parameters. We also employ a SVM-based classification algorithm to assignMK classes to the LAMOST stellar spectra. We find that the completenesses of the classification are up to 90% for A and G type stars, while it is down to about 50% for OB and K type stars. About 40% of the OB and K type stars are mis-classified as A and G type stars, respectively. This is likely owe to the difference of the spectral features between the late B type and early A type stars or between the late G and early K type stars are very weak. The relative poor performance of the automatic MK classification with SVM suggests that the directly use of the line indices to classify stars is likely a more preferable choice.