J/AJ/159/182 APOGEE Net, YSOs parameters through deep learning (Olney+, 2020)
APOGEE Net: improving the derived spectral parameters for young stars through
deep learning.
Olney R., Kounkel M., Schillinger C., Scoggins M.T., Yin Y., Howard E.,
Covey K.R., Hutchinson B., Stassun K.G.
<Astron. J., 159, 182 (2020)>
=2020AJ....159..182O 2020AJ....159..182O
ADC_Keywords: YSOs; Stars, dwarfs; Spectra, infrared; Abundances, [Fe/H];
Effective temperatures
Keywords: Astroinformatics ; Computational methods ; Young stellar objects ;
Low mass stars ; Stellar classification
Abstract:
Machine learning allows for efficient extraction of physical
properties from stellar spectra that have been obtained by large
surveys. The viability of machine-learning approaches has been
demonstrated for spectra covering a variety of wavelengths and
spectral resolutions, but most often for main-sequence (MS) or evolved
stars, where reliable synthetic spectra provide labels and data for
training. Spectral models of young stellar objects (YSOs) and low-mass
MS stars are less well-matched to their empirical counterparts,
however, posing barriers to previous approaches to classify spectra of
such stars. In this work, we generate labels for YSOs and low-mass MS
stars through their photometry. We then use these labels to train a
deep convolutional neural network to predict logg, Teff, and Fe/H for
stars with Apache Point Observatory Galactic Evolution Experiment
(APOGEE) spectra in the DR14 data set. This "APOGEE Net" has produced
reliable predictions of logg for YSOs, with uncertainties of within
0.1dex and a good agreement with the structure indicated by pre-MS
evolutionary tracks, and it correlates well with independently derived
stellar radii. These values will be useful for studying pre-MS stellar
populations to accurately diagnose membership and ages.
Description:
Apache Point Observatory Galactic Evolution Experiment (APOGEE) is a
high-resolution (R∼22500) near-infrared (1.51-1.7µm spectrograph
mounted on the Sloan Foundation 2.5m telescope. We restrict the
current analysis only to the data Realease 14.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
table4.dat 159 235999 Results for combined evaluation
--------------------------------------------------------------------------------
See also:
III/279 : RAVE 5th data release (Kunder+, 2017)
I/345 : Gaia DR2 (Gaia Collaboration, 2018)
J/AJ/130/1145 : YSO near-infrared properties (Doppmann+, 2005)
J/ApJ/794/125 : IN-SYNC. I. APOGEE stellar parameters (Cottaar+, 2014)
J/ApJ/807/27 : IN-SYNC. III. RVs of IC348 stars (Cottaar+,2015)
J/ApJ/799/136 : IN-SYNC. II. Candidate young stars in NGC1333 (Foster+,2015)
J/ApJ/804/64 : Empirical and model parameters of 183 M dwarfs (Mann+, 2015)
J/ApJ/808/16 : The Cannon, a new approach to determine abund. (Ness+, 2015)
J/AJ/152/141 : Solar neighborhood. XXXVII. RVs M dwarfs (Benedict+, 2016)
J/ApJ/822/81 : PTF stellar rot. periods for Pleiades members (Covey+, 2016)
J/ApJ/818/59 : IN-SYNC. IV. YSOs in Orion A (Da Rio+, 2016)
J/AJ/151/144 : ASPCAP weights for 15 APOGEE chemi. elements (Garcia+, 2016)
J/A+A/589/A80 : APOGEE strings (Hacar+, 2016)
J/AJ/153/136 : Planets and their host stars with Gaia plxs (Stassun+, 2017)
J/ApJ/851/26 : Teff, metallicity & Ti abundance of M dwarfs (Veyette+,2017)
J/MNRAS/478/4513 : GALAH Survey DR2 (Buder+, 2018)
J/A+A/610/A77 : Orion Integral Filament ALMA+IRAM30m N2H+(1-0) (Hacar+,2018)
J/AJ/156/18 : APOGEE DR14,Binary comp. of evolved stars (Price-Whelan+, 2018)
J/AJ/156/84 : APOGEE-2 survey of Orion Complex. II. (Kounkel+, 2018)
J/ApJ/869/72 : IN-SYNC. VIII. YSOs NGC 1333, IC 348 & Orion A (Yao+, 2018)
J/AJ/157/196 : Close companions around young stars (Kounkel+, 2019)
Byte-by-byte Description of file: table4.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 18 A18 --- APOGEE APOGEE Identifier (2MHHMMSSss+DDMMSSs)
20- 32 F13.9 deg RAdeg Right Ascension in decimal degrees (J2000)
34- 46 F13.9 deg DEdeg [-32.7/87.6] Declination in decimal degrees
(J2000)
48- 54 F7.4 [cm/s2] loggL [-0.003/11.7]? Input pre-porcessing label for
log surface gravity
56- 63 F8.5 [cm/s2] logg [-0.6/5.8] Predicted log surface gravity
65- 71 F7.5 [cm/s2] e_logg [0.0001/1.6] Uncertainty in logg
73- 82 F10.4 K TeffL [2523/12478]? Input pre-processing label for
effective temperature
84- 92 F9.4 K Teff [2566/6700] Predicted effective temperature
94-102 F9.4 K e_Teff [0.11/1618] Uncertainty in Teff
104-111 F8.5 [Sun] [Fe/H]L [-2.6/0.68]? Input pre-processing label for
metallicity
113-120 F8.5 [Sun] [Fe/H] [-1.79/0.8] Predicted metallicity
122-128 F7.5 [Sun] e_[Fe/H] [0.0001/0.61] Uncertainty in [Fe/H]
130-147 F18.13 --- SNR [8.77e-11/4138] Signal-to-Noise
149-153 A5 --- MLset Data set description (1)
155-159 A5 --- Type Data type (2)
--------------------------------------------------------------------------------
Note (1): In order to predict Teff, logg, and Fe/H simultaneously, it
was necessary to normalize these target values; normalizing the
targets put the losses (and gradients during training) onto a
comparable scale. To normalize, we calculated the mean (µ) and
standard deviation (σ) of each target variable using the
training set and then standardized all prediction targets across all
sources and all sets; train (176417 occurrences), development (22785
occurrences), and test (23066 occurrences). See section 3.1 of the
article for mor details.
Note (2): In training APOGEE Net, we adopt labels for Teff, logg, and
Fe/H from the Payne's outputs for 215342 stars with Teff>4000K or
logg<3.5. As the Payne does not yet produce reliable labels for stars
with lower Teff and higher logg than these limits, in the next
section, we use empirically calibrated photometric relationships to
generate new labels for 4480 low-mass MS stars and 2446 YSOs. More
details about Payne labels in section 2.2 of the article.
--------------------------------------------------------------------------------
History:
From electronic version of the journal
(End) Prepared by [AAS], Coralie Fix [CDS], 16-Sep-2020