SIMBAD references

Context. Characterizing the interiors of gaseous giant exoplanets is currently one of the main objectives in exoplanetary sciences. In particular, the planetary heavy-element mass provides a critical constraint on planet formation from exoplanetary systems. However, gas giant exoplanets show large diversities in thermal states and their interior properties vary across a wide magnitude range. Forward modeling of their interiors exhibits a larger degeneracy with respect to rocky exoplanets.
Aims. We applied machine learning techniques based on mixture density networks (MDNs) to investigate the interiors of gaseous giant exoplanets. We aim to provide a well-trained MDN for quick and efficient predictions.
Methods. Based on our current knowledge of gas giants in the Solar System, we discussed an effect of model uncertainties on planetary interiors and presented a data set for gas giants with masses between 0.1 and 10 Jupiter masses using two-layer interior models. Then, MDNs were constructed to train the generated data set and their performance was evaluated in order to achieve a well-trained one.
Results. The MDN using planetary mass and radius as inputs exhibits the well-known degeneracy of interior models. The surface temperature of a planet bears constraints on the thermal state of planetary interiors, and adding it as additional input considerably breaks the degeneracy of possible interior structures. The MDN with inputs of mass, radius, and surface temperature is found to show excellent performance in predicting the interior properties of gaseous giant exoplanets, although these interior properties span over a very wide range. We also applied the well-trained MDN to four gas giants in the Solar System and beyond. The MDN predictions are in good agreement with the interior model solutions within the observational and systematic uncertainties.
Conclusions. We offer a convenient and powerful tool available online providing knowledge of the interiors of gaseous giant exoplanets in addition to rocky exoplanets, which could be helpful for our understanding of planet formation in diverse protoplanetary environments.