An Interpretable Machine Learning Framework for Modeling High-Resolution Spectroscopic Data

Abstract

Comparison of echelle spectra to synthetic models has become a computational statistics challenge, with over ten thousand individual spectral lines affecting a typical cool star echelle spectrum. Telluric artifacts, imperfect line lists, inexact continuum placement, and inflexible models frustrate the scientific promise of these information-rich datasets. Here we debut an interpretable machine-learning framework "blasé" that addresses these and other challenges. The semi-empirical approach can be viewed as "transfer learning" -- first pre-training models on noise-free precomputed synthetic spectral models, then learning the corrections to line depths and widths from whole-spectrum fitting to an observed spectrum. The auto-differentiable model employs back-propagation, the fundamental algorithm empowering modern Deep Learning and Neural Networks. Here, however, the 40,000+ parameters symbolize physically interpretable line profile properties such as amplitude, width, location, and shape, plus radial velocity and rotational broadening. This hybrid data-/model- driven framework allows joint modeling of stellar and telluric lines simultaneously, a potentially transformative step forwards for mitigating the deleterious telluric contamination in the near-infrared. The blasé approach acts as both a deconvolution tool and semi-empirical model. The general purpose scaffolding may be extensible to many scientific applications, including precision radial velocities, Doppler imaging, chemical abundances, and remote sensing. Its sparse-matrix architecture and GPU-acceleration make blasé fast. The open-source PyTorch-based code includes tutorials, Application Programming Interface (API) documentation, and more. We show how the tool fits into the existing Python spectroscopy ecosystem, demonstrate a range of astrophysical applications, and discuss limitations and future extensions.

Accepted for publication in ApJ; Open source code available at https://github.com/gully/blase

References in corpus (21)

• Array Programming with NumPy
• The Radial Velocity Experiment (RAVE): first data release
• Molecfit: A general tool for telluric absorption correction. I. Method and application to ESO instruments
• Precision Stellar Characterization of FGKM Stars using an Empirical Spectral Library
• exoplanet: Gradient-based probabilistic inference for exoplanet data & other astronomical time series
• Uniform Atmospheric Retrieval Analysis of Ultracool Dwarfs I: Characterizing Benchmarks, Gl570D and HD3651B
• Correcting for Telluric Absorption: Methods, Case Studies, and Release of the TelFit Code
• Placing the spotted T Tauri star LkCa 4 on an HR diagram
• A search for TiO in the optical high-resolution transmission spectrum of HD 209458b: Hindrance due to inaccuracies in the line database
• Detection of sodium in the atmosphere of WASP-69b
• No Sun-like dynamo on the active star $\zeta$ Andromedae from starspot asymmetry
• The EXPRES Stellar Signals Project II. State of the Field in Disentangling Photospheric Velocities
• Measuring precise radial velocities on individual spectral lines. III. Dependence of stellar activity signal on line formation temperature
• TESS Observations of the Luhman 16AB Brown Dwarf System: Rotational Periods, Lightcurve Evolution, and Zonal Circulation
• The CARMENES search for exoplanets around M dwarfs -- A deep learning approach to determine fundamental parameters of target stars
• A 1.46-2.48 $\mu$m Spectroscopic Atlas of a T6 Dwarf (1060 K) Atmosphere with IGRINS: First Detections of H${}_2$S and H${}_2$, and Verification of H${}_2$O, CH${}_4$, and NH${}_3$ Line Lists
• The SDSS/APOGEE Catalog of HgMn Stars
• Phase Retrieval and Design with Automatic Differentiation
• The He I $\lambda$10830 $\mathring{A}$ line as a probe of winds and accretion in young stars in Lupus and Upper Scorpius
• Kernel Phase and Coronagraphy with Automatic Differentiation
• The Sonora Brown Dwarf Atmosphere and Evolution Models I. Model Description and Application to Cloudless Atmospheres in Rainout Chemical Equilibrium

Cited by in corpus (4)

• DSPS: Differentiable Stellar Population Synthesis
• Differentiable Modeling of Planet and Substellar Atmosphere: High-Resolution Emission, Transmission, and Reflection Spectroscopy with ExoJAX2
• Interpreting anomaly detection of SDSS spectra
• Panchromatic View of the Frigid Jovian Exoplanet COCONUTS-2 b