Presentation #102.04 in the session Poster Session.
Bayesian Atmospheric Radiative Transfer (BART, Harrington et al. 2022, PSJ; Cubillos et al. 2022, PSJ; Blecic et al. 2022, PSJ) is an open-source, reproducible-research code for atmospheric composition and structure retrieval. Its modular design and high degree of user control allow it to be configured to mimic the physics hardwired into many other codes. Users can configure BART to reproduce published work and then explore the effect of different choices for line lists, included species, thermal-profile models, Bayesian samplers, and cloud parameterizations, among others. By implementing the approaches of differing authors, BART has resolved controversies, such as the interpretation of WASP-12 b eclipse data (Himes et al. 2022, PSJ). BART comes with its own verification test package, BARTTest, which can be applied to any radiative-transfer or retrieval code. In developing BART, we learned valuable lessons in statistics and spectral modeling, such as how many Bayesian steps to run to achieve a given posterior accuracy. BART’s radiative-transfer code, Transit, has trained a machine-learning surrogate model that accelerates retrieval by a factor of ~100 with full Bayesian accuracy (Himes et al. 2022, PSJ). Challener et al. (2022, PSJ) analyzed WASP-34 b eclipses with BART. This research was supported by the NASA Fellowship Activity Grant 80NSSC20K0682, NASA Exoplanets Research Program grant NNX17AB62G, NASA Planetary Atmospheres grant NNX12AI69G, and NASA Astrophysics Data Analysis Program grant NNX13AF38G, held by J.H. J.B. held NASA Earth and Space Science Fellowship NNX12AL83H. I.D.-D. and J.B. held NASA Exoplanets Research Program grant NNX17AC03G. P.E.C. was supported by the Fulbright Program for Foreign Students. P.M.R. acknowledges support from CONICYT project Basal AFB-170002.