Using autoencoders and deep transfer learning to determine the stellar parameters of 286 CARMENES M dwarfs

Abstract

Deep learning (DL) techniques are a promising approach among the set of methods used in the ever-challenging determination of stellar parameters in M dwarfs. In this context, transfer learning could play an important role in mitigating uncertainties in the results due to the synthetic gap (i.e. difference in feature distributions between observed and synthetic data). We propose a feature-based deep transfer learning (DTL) approach based on autoencoders to determine stellar parameters from high-resolution spectra. Using this methodology, we provide new estimations for the effective temperature, surface gravity, metallicity, and projected rotational velocity for 286 M dwarfs observed by the CARMENES survey. Using autoencoder architectures, we projected synthetic PHOENIX-ACES spectra and observed CARMENES spectra onto a new feature space of lower dimensionality in which the differences between the two domains are reduced. We used this low-dimensional new feature space as input for a convolutional neural network to obtain the stellar parameter determinations. We performed an extensive analysis of our estimated stellar parameters, ranging from 3050 to 4300 K, 4.7 to 5.1 dex, and -0.53 to 0.25 dex for Teff, logg, and [Fe/H], respectively. Our results are broadly consistent with those of recent studies using CARMENES data, with a systematic deviation in our Teff scale towards hotter values for estimations above 3750 K. Furthermore, our methodology mitigates the deviations in metallicity found in previous DL techniques due to the synthetic gap. We consolidated a DTL-based methodology to determine stellar parameters in M dwarfs from synthetic spectra, with no need for high-quality measurements involved in the knowledge transfer. These results suggest the great potential of DTL to mitigate the differences in feature distributions between the observations and the PHOENIX-ACES spectra.

Accepted in A&A

References in corpus (82)

• SciPy 1.0--Fundamental Algorithms for Scientific Computing in Python
• Array Programming with NumPy
• Gaia Data Release 2. Summary of the contents and survey properties
• The Seventh Data Release of the Sloan Digital Sky Survey
• PARSEC: stellar tracks and isochrones with the PAdova and TRieste Stellar Evolution Code
• Gaia Data Release 3: Summary of the content and survey properties
• A grid of MARCS model atmospheres for late-type stars I. Methods and general properties
• Intrinsic Colors, Temperatures, and Bolometric Corrections of Pre-Main Sequence Stars
• A new extensive library of PHOENIX stellar atmospheres and synthetic spectra
• New evolutionary models for pre-main sequence and main sequence low-mass stars down to the hydrogen-burning limit
• X-shooter, the new wide band intermediate resolution spectrograph at the ESO Very Large Telescope
• How to Constrain Your M Dwarf: measuring effective temperature, bolometric luminosity, mass, and radius
• Improving PARSEC models for very low mass stars
• The HARPS search for southern extra-solar planets XXXI. The M-dwarf sample
• PARSEC evolutionary tracks of massive stars up to $350M_{\odot }$ at metallicities 0.0001$\le Z\le$0.04
• alpha-, r-, and s-process element trends in the Galactic thin and thick disks
• Stellar Diameters and Temperatures II. Main Sequence K & M Stars
• ESPRESSO@VLT -- On-sky performance and first results
• BANYAN. XI. The BANYAN $\Sigma$ multivariate Bayesian algorithm to identify members of young associations within 150 pc
• Determining stellar atmospheric parameters and chemical abundances of FGK stars with iSpec
• The SEGUE Stellar Parameter Pipeline. I. Description and Initial Validation Tests
• New PARSEC evolutionary tracks of massive stars at low metallicity: testing canonical stellar evolution in nearby star forming dwarf galaxies
• Spectrum radial velocity analyser (SERVAL). High-precision radial velocities and two alternative spectral indicators
• The Cannon: A data-driven approach to stellar label determination
• Metallicity and Temperature Indicators in M dwarf K band Spectra: Testing New & Updated Calibrations With Observations of 133 Solar Neighborhood M dwarfs
• The SEGUE Stellar Parameter Pipeline. II. Validation with Galactic Globular and Open Clusters
• Trumpeting M Dwarfs with CONCH-SHELL: a Catalog of Nearby Cool Host-Stars for Habitable ExopLanets and Life
• The CARMENES search for exoplanets around M dwarfs: High-resolution optical and near-infrared spectroscopy of 324 survey stars
• The SEGUE Stellar Parameter Pipeline. III. Comparison with High-Resolution Spectroscopy of SDSS/SEGUE Field Stars
• Gaia Data Release 3: Analysis of RVS spectra using the General Stellar Parametriser from spectroscopy
• Spectro-Thermometry of M dwarfs and their candidate planets: too hot, too cool, or just right?
• The CARMENES search for exoplanets around M dwarfs: Different roads to radii and masses of the target stars
• M dwarfs: effective temperatures, radii and metallicities
• Prospecting in late-type dwarfs: a calibration of infrared and visible spectroscopic metallicities of late-K and M dwarfs spanning 1.5 dex
• CARMENES input catalogue of M dwarfs. I. Low-resolution spectroscopy with CAFOS
• Metal-Rich M-dwarf Planet Hosts: Metallicities with K-Band Spectra
• CARMENES input catalogue of M dwarfs V. Luminosities, colours, and spectral energy distributions
• The SEGUE Stellar Parameter Pipeline. IV. Validation with an Extended Sample of Galactic Globular and Open Clusters
• Stellar Diameters and Temperatures V. Eleven Newly Characterized Exoplanet Host Stars
• Flat-relative optimal extraction. A quick and efficient algorithm for stabilised spectrographs
• Galactic spiral structure revealed by Gaia EDR3
• The 10 parsec sample in the Gaia era
• CARMENES input catalogue of M dwarfs: III. Rotation and activity from high-resolution spectroscopic observations
• An Application of Deep Neural Networks in the Analysis of Stellar Spectra
• M Dwarf Metallicities and Giant Planet Occurrence: Ironing Out Uncertainties and Systematics
• Deriving the stellar labels of LAMOST spectra with Stellar LAbel Machine (SLAM)
• The CARMENES search for exoplanets around M dwarfs. Activity indicators at visible and near-infrared wavelengths
• The CARMENES search for exoplanets around M dwarfs. Stellar atmospheric parameters of target stars with SteParSyn
• The CARMENES search for exoplanets around M dwarfs - Photospheric parameters of target stars from high-resolution spectroscopy
• Metallicity of M dwarfs IV. A high-precision [Fe/H] and Teff technique from high-resolution optical spectra for M dwarfs
• The Nearby, Young, Argus Association: Membership, Age, and Dusty Debris Disks
• The CARMENES search for exoplanets around M dwarfs -- Photospheric parameters of target stars from high-resolution spectroscopy. II. Simultaneous multiwavelength range modeling of activity insensitive lines
• An empirical calibration to estimate cool dwarf fundamental parameters from H-band spectra
• Prospecting in Ultracool Dwarfs: Measuring the Metallicities of Mid- and Late-M Dwarfs
• Stellar parameters of early M dwarfs from ratios of spectral features at optical wavelengths
• Application of Convolutional Neural Networks for Stellar Spectral Classification
• Cool Subdwarf Investigations (CSI) I: New Thoughts for the Spectral Types of K and M Subdwarfs
• Calibrating the metallicity of M dwarfs in wide physical binaries with F-, G-, and K- primaries - I: High-resolution spectroscopy with HERMES: stellar parameters, abundances, and kinematics
• Temperatures and Metallicities of M dwarfs in the APOGEE Survey
• Accurate Gravities of F, G, and K stars from High Resolution Spectra Without External Constraints
• Beyond the Hubble Sequence -- Exploring Galaxy Morphology with Unsupervised Machine Learning
• A discontinuity in the $T_{\mathrm{eff}}$-radius relation of M-dwarfs
• SteParSyn: A Bayesian code to infer stellar atmospheric parameters using spectral synthesis
• Stellar Characterization of M-dwarfs from the APOGEE Survey: A Calibrator Sample for the M-dwarf Metallicities
• ODUSSEAS: A machine learning tool to derive effective temperature and metallicity for M dwarf stars
• A physically motivated and empirically calibrated method to measure effective temperature, metallicity, and Ti abundance of M dwarfs
• Deep Learning improves identification of Radio Frequency Interference
• Stellar parametrization of LAMOST M dwarf stars
• The CARMENES search for exoplanets around M dwarfs -- A deep learning approach to determine fundamental parameters of target stars
• Determination of spectroscopic parameters for 313 M dwarf stars from their APOGEE Data Release 16 H-band spectra
• An autoencoder of stellar spectra and its application in automatically estimating atmospheric parameters
• Unsupervised feature-learning for galaxy SEDs with denoising autoencoders
• Chemical Properties of the Local Galactic Disk and Halo. I. Fundamental Properties of 1,544 Nearby, High Proper-Motion M dwarfs and subdwarfs
• Lick Spectral Indices for Super Metal-rich Stars
• Parameterizing Stellar Spectra Using Deep Neural Networks
• Understanding Physical Properties of Young M-dwarfs: NIR spectroscopic studies
• VEXAS: VISTA EXtension to Auxiliary Surveys -- Data Release 2: Machine-learning based classification of sources in the Southern Hemisphere
• The CARMENES search for exoplanets around M dwarfs -- A deep transfer learning method to determine Teff and [M/H] of target stars
• Representation learning for automated spectroscopic redshift estimation
• Characterization of Low Mass K2 Planet Hosts Using Near-Infrared Spectroscopy
• Bayesian decomposition of the Galactic multi-frequency sky using probabilistic autoencoders
• Calibrating the metallicity of M dwarfs in wide physical binaries with F-, G-, and K-primaries -- II: Carbon, oxygen, and odd-Z iron-peak abundances of the primary stars

Cited by in corpus (6)

• Euclid Quick Data Release (Q1) -- Spectroscopic search, classification and analysis of ultracool dwarfs in the Deep Fields
• The stellar activity-rotation-age relationship under the lens of asteroseismology
• Performance of the Stellar Abundances and atmospheric Parameters Pipeline adapted for M dwarfs I. Atmospheric parameters from the spectroscopic module
• Refined M-type Star Catalog from LAMOST DR10: Measurements of Radial Velocities, $T_{\text{eff}}$, log $g$, [M/H] and [$\alpha$/M]
• Towards model-free stellar chemical abundances. Potential applications in the search for chemically peculiar stars in large spectroscopic surveys
• The CARMENES search for exoplanets around M dwarfs. A homogeneous catalogue of projected rotational velocities accounting for limb-darkening