Astronomical spectra are some of the most information-dense types of data in our field, providing a wealth of diagnostics about physical processes for a variety of sources, including stars, nebulae, planets and galaxies. Fitting spectral data to extract that information, however, requires sophisticated synthetic models and a careful statistical treatment of both models and data. The Starfish framework, originally developed by Czekala et al. (2015) is an extensible likelihood framework for hierarchical Bayesian spectroscopic inference based on stellar synthetic spectra. We present here a new version of this tool, rewritten from the ground up as open-source python framework in an effort to create a more accessible piece of software. The new framework is faster, dynamic, and applies software engineering best practices with proper documentation, unit testing, and continuous integration. We demonstrate the software by fitting the near-IR spectra of sample stars obtained with the SpeX spectrograph at the NASA/IRTF facility.
|Published - 1 Jan 2020