CROCODILE (CROss-COrrelation retrievals of Directly-Imaged self-Luminous Exoplanets) runs atmospheric retrievals of directly observed gas giant exoplanets by adopting adequate likelihood functions. The code makes use of petitRADTRANS (ascl:2207.014) and PyMultiNest (ascl:1606.005) and provides a statistical framework to interpret the photometry, low-resolution spectroscopy, and medium (and higher) resolution cross-correlation spectroscopy.

ml4ptp produces physically consistent pressure-temperature (PT) profiles that do not require explicit assumptions about the functional form of the PT profiles. Atmospheric retrievals (AR) of exoplanets typically rely on a combination of a Bayesian inference technique and a forward simulator to estimate atmospheric properties from an observed spectrum. A key component in simulating spectra is the PT profile, which describes the thermal structure of the atmosphere. AR pipelines commonly use ad hoc fitting functions here that limit the retrieved PT profiles to simple approximations, but still use a relatively large number of parameters. ml4ptp uses fewer parameters than other methods while achieving better fit quality and reducing computational cost.

fm4ar (flow matching for atmospheric retrievals) infers atmospheric properties of exoplanets from observed spectra. It uses flow matching posterior estimation (FMPE) for its machine learning (ML) approach to atmospheric retrieval; this approach provides many of the advantages of neural posterior estimation (NPE) while also providing greater architectural flexibility and scalability. The package uses importance sampling (IS) to verify and correct ML results, and to compute an estimate of the Bayesian evidence. fm4ar's ML models are conditioned on the assumed noise level of a spectrum (i.e., error bars), thus making them adaptable to different noise models.