Interior Structure Modelling of Low-Mass Exoplanets with Constraints

Kurzfassung

Understanding the interior structure of low-mass exoplanets is a fundamental challenge in planetary science, particularly given the inherent degeneracies in the modeling of their internal compositions. This thesis addresses this challenge by focusing on the water and gas layers, as these volatiles significantly influence planetary radii by altering density and pressure gradients, introducing substantial uncertainty in existing models. The primary objective is to reduce this degeneracy and provide clearer insights into the compositional and structural characteristics of these planets, which are crucial for interpreting observational data and assessing habitability. To tackle this issue, I employ a machine-learning model based on Mixture Density Networks (MDNs) called ExoMDN to study the interior structures of low-mass exoplanets. Trained on a database of 5.6 million precomputed synthetic interior structures, ExoMDN enables rapid probabilistic inference of planetary interiors in under a second. Unlike traditional methods, which require extensive modeling for each planet and can take several days to compute, Exo-MDN’s efficiency makes it ideal for large-scale studies. This advantage allows me to comprehensively explore planetary interiors that would otherwise be computationally expensive with conventional approaches. This thesis contributes to the field by providing a deeper understanding of the relationship between planetary composition, volatile content, and observable properties such as mass and radius. The results have significant implications for future observational missions like PLATO, which aim to characterize exoplanets with high precision and explore their potential habitability. Furthermore, this work underscores the importance of integrating constraints from planet formation and atmospheric escape models to enhance the accuracy of interior structure predictions.