Overview

BeAR is an open-source computer program that can perform retrieval calculations for observational data of planetary atmospheres as well as brown dwarfs. It is the successor of the previous retrieval code HELIOS-r2 (Kitzmann et al. 2020) that has been successfully applied for the characterisation of data from ground-based and space telescopes for a wide range of objects.

BeAR is currently able to perform retrievals for emission, transmission, and secondary-eclipse spectra. It can use low to mid-resolution spectral data, as well as photometric data points. It has support for constant temperature and abundance profiles of atmospheric constituents but can also retrieve vertically-varying profiles if the observational data is good enough.

While the core of BeAR is written in C++ and CUDA, it also comes with a Python interface, pyBeAR, that allows the forward model and retrieval calculations to be performed in Python. The interface allows BeAR to be coupled to parameter space explorers or machine-learning algorithms written in Python. The stand-alone C++ version is currently only able to use the MultiNest sampler for the retrieval calculations, while the Python interface also supports the use of other sampling algorithms, such as Dynesty or MCMC.