基于梯度提升决策树算法测定LAMOST红巨星分支恒星的化学丰度

打开文本图片集

中图分类号：P145.9 文献标志码：A 文章编号：1673-5072（2025）03-0311-07

Abstract：Taking high-precision asteroseismological Red Giant Branch （RGB） stars from APOKASC-2 with high-resolution spectroscopic parameters as a training sample，We developed a method to estimate the chemical abundances of LAMOST DR5 Red Giant Branch stars precisely.Based on the Gradient Boosting Decision Tree algorithm（GBDT）and the basic atmospheric parameters （ Teff∖logg∖[Fe/H] ），four chemical abundances，[C/Fe]、[Mg/Fe]、[Ca/Fe]and [Ni/Fe]，were predicted in this work.The current results are shown on the test set with median errors between O.O2 and 0.04dex . The comparison with other works indicates that the method used in this work is feasible，providing a stage validation for the future promotion of this method in determining the chemical abundance of Red Giant Branch stars.

Keywords：Milky Way disk;Red Giant Branch stars;chemical abundance;machine learning;star catalog

星系的形成与演化是当今天体物理学的重大科学问题，银河系是在多星族六维相空间下研究星系物理的重要参考与基石[1-5]。（剩余9451字）