J/AJ/169/270 Learning-based classification of LAMOST CP1 & CP2 stars (Lu+, 2025)
Deep Learning-based Identification of CP1 and CP2 Stars from LAMOST DR11.
Lu Z., Kong X., Zhang Y., Yi Z., Liu M.
<Astron. J., 169, 270 (2025)>
=2025AJ....169..270L 2025AJ....169..270L
ADC_Keywords: Stars, peculiar; Stars, Am; Stars, Ap; MK spectral classification;
Models; Spectra, optical; Spectra, infrared
Keywords: Chemically peculiar stars ; Am stars ; Ap stars
Abstract:
Chemically peculiar (CP) stars, particularly CP1 (metallic-line) and
CP2 (magnetic Ap/Bp) stars, are crucial for studying microscopic
processes in stellar atmospheres, such as atomic diffusion and
magnetic interactions. In this study, we present ODCNNnet, a deep
learning-based classification model designed to identify CP1 and CP2
stars from low-resolution spectra. The model achieves an accuracy of
99.81% and a recall of 99.43% for CP1 stars, and an accuracy of 98.82%
with a recall of 99.41% for CP2 stars. Applying ODCNNnet to B- to
early F-type stars in LAMOST DR11, we identified 33,049 CP1 stars and
5901 CP2 stars, including 5287 newly identified CP1 stars and 1153
newly identified CP2 stars, validated through cross matching with the
MKCLASS tool. We further analyzed the stellar parameters of CP stars.
The effective temperature (Teff) of CP1 stars ranges from 6000K to
9000K, peaking at 7500K, while CP2 stars span 6000K to 13,000K, with
peaks near 8900K and 11,900K. The surface gravity (logg) of CP1 stars
is concentrated between 3.4 and 4.6dex, peaking at 3.9dex, while CP2
stars range from 3.5 to 4.6dex. The [Fe/H] of CP1 stars is generally
between -1.0 and 0.8dex, peaking around -0.1dex, whereas CP2 stars
exhibit a broader distribution from -1.5 dex to 1.0dex, with peaks
around -0.25dex and 0.75dex. This study provides a valuable data set
for investigating the physical properties of CP stars and their
underlying formation mechanisms.
Description:
The LAMOST, also known as the Guo Shoujing Telescope, is a large-scale
reflecting Schmidt telescope located at Xinglong Observatory, operated
by the National Astronomical Observatories of China. With an effective
aperture ranging from 3.6 to 4.9m and a wide field of view of 5°,
LAMOST is capable of simultaneously acquiring spectra from up to 4000
celestial objects through its focal plane, which houses 4000 optical
fibers. As of the 11th data release (DR11, V/162), LAMOST has obtained
11,944,094 low-resolution spectra, covering a wavelength range from
3700Å to 9000Å, from ultraviolet to near-infrared, with a
spectral resolution of approximately 1800 at 5500Å.
This study aims to develop a three-class classifier to effectively
distinguish CP1, CP2, and non-CP1/CP2 stars using low-resolution
spectra from LAMOST DR11 (V/162) and DR12. To construct and evaluate
the classifier, we curated three distinct data sets: CP1 stars, CP2
stasr, and non-CP1/CP2 stars.
The CP1 star data set was compiled from two catalogs: one from
Shang+2022 (J/ApJS/259/63) containing 17,986 CP1 stars, and the other
from Tian+2023 (J/ApJS/266/14) with 21,600 CP1 stars. To improve the
accuracy of the model, we selected a subset of 5299 common samples
obtained by cross referencing these two catalogs. This intersection
was used as the CP1 training set, ensuring that the data set was
reliable and representative. For the CP2 star data set, we sourced
data from two key catalogs: one from Shang+2022, which lists 2708 CP2
stars, and the other from Shi+2023 (J/ApJ/943/147), which includes
2700 CP2 stars. To ensure adequate data for training, we removed
duplicate entries, performed data cleaning, and then combined the two
catalogs to generate a final data set of 3799 unique CP2 stars.
The non-CP1/CP2 star data set was curated from B-type to early F-type
stars in the LAMOST DR9 low-resolution spectra. CP1 and CP2 stars were
excluded, and duplicate entries were addressed by retaining only the
spectra with the highest signal-to-noise ratio (S/N) in the g band.
From this refined data set, we randomly selected 9000 non-CP1/CP2 star
spectra as the training set for non-CP1/CP2 stars.
In the revised Table 2, we present the updated results: 31,350 CP1
stars and 5744 CP2 stars, including 6421 newly identified CP1 stars
and 951 newly identified CP2 stars.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
table2.dat 147 37094 The Catalogs of CP1 and CP2 Stars in the LAMOST
DR11 (v1.0) and DR12 (v0) (revised from Erratum)
--------------------------------------------------------------------------------
See also:
I/355 : Gaia DR3 Part 1. Main source (Gaia Collaboration, 2022)
III/162A : General Catalogue of Ap and Am stars (Renson+ 1991)
V/162 : LAMOST DR11 catalogs (Luo+, 2026)
J/A+A/499/567 : BV differential photometry of HR 7224 (Krticka+, 2009)
J/MNRAS/449/1401 : Am stars candidates from LAMOST DR1 (Hou+, 2015)
J/AJ/151/13 : LAMOST-Kepler MKCLASS spectral classification (Gray+, 2016)
J/MNRAS/480/2953 : Catalogue of CP stars (HgMn, ApBp, AmFm) (Ghazaryan+, 2018)
J/ApJS/242/13 : Am stars from LAMOST DR5 (Qin+, 2019)
J/A+A/640/A40 : 1002 mCP stars from LAMOST DR4 (Hummerich+, 2020)
J/ApJS/259/63 : Chemically peculiar stars CP1 & CP2 from LAMOST (Shang+, 2022)
J/ApJ/943/147 : Ap stars from LAMOST DR9 sp. and Gaia DR3 (Shi+, 2023)
J/ApJS/266/14 : Am stars selected in LAMOST DR8-DR10 (Tian+, 2023)
J/ApJS/272/43 : Am and Ap star candidates from LAMOST DR10 (Yang+, 2024)
http://zenodo.org/records/15590979 : Data from this study (Lu+, 2025)
Byte-by-byte Description of file:table2.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 10 I10 --- ObsID [405074/1247205195] Unique Spectra ID from
LAMOST
12- 24 F13.9 deg RAdeg Right ascension (J2000)
26- 37 F12.9 deg DEdeg Declination (J2000)
39- 40 A2 --- SpTypeL LAMOST derived spectral type (1)
42- 49 F8.2 K Teff [4786/13307] LAMOST effective
temperature
51- 57 F7.3 K e_Teff [3.6/777]? Teff uncertainty
59- 63 F5.3 [cm/s2] logg [2.1/4.9] LAMOST log surface gravity
65- 69 F5.3 [cm/s2] e_logg [4e-3/1.11]? logg uncertainty
71- 76 F6.3 [-] [Fe/H] [-2.5/0.995] LAMOST metallicity
78- 82 F5.3 [-] e_[Fe/H] [2e-3/0.662]? [Fe/H] Uncertainty
84- 107 A24 --- SpTypeM MKCLASS code derived spectral type
109- 113 A5 --- qflag Quality flag for SpT_mkclass if any
(from "poor" to "vgood") (2)
115- 117 A3 --- subClass Subclass of CP type stars (3)
119 I1 --- fNew [0/1] "0"=Known stars; "1"=New stars (4)
121- 139 I19 --- GaiaDR3 ? Gaia DR3 (I/355) source identifier as
obtained from LAMOST
141- 147 A7 --- Version LAMOST data release version, DR11 v1
(V/162) or DR12 v0 (5)
--------------------------------------------------------------------------------
Note (1): Spectral types occurrences as follows:
A* = 22599 occurrences
F* = 14492 occurrences
G* = 3 occurrences
Note (2): Quality flags occurrences as follows:
fair = 159 occurrences
good = 6042 occurrences
poor = 44 occurrences
vgood = 6581 occurrences
N/A = 24268 occurrences
Note (3): Subclass of chemically peculiar stars as follows:
CP1 = Metallic-line (31350 occurrences)
CP2 = Ap/Bp stars (5744 occurrences)
Note (4): Occurrences as follows:
0 = Known star (29722 occurrences)
1 = New star (7372 occurrences)
Note (5): LAMOST data release versions occurrences as follows:
DR11 = 33694 occurrences
DR12 = 3400 occurrences
--------------------------------------------------------------------------------
History:
From electronic version of the journal
References:
Lu et al. Original paper 2025AJ....169..270L 2025AJ....169..270L
Lu et al. Erratum 2025AJ....170...62L 2025AJ....170...62L
(End) Prepared by [AAS], Robin Leichtnam [CDS] 11-Feb-2026