Goal of my PhD

PhD project

Machine Learning from Complex Disk Models

Planets form in so-called protoplanetary disks. These regions are investigated to improve our understanding of planet formation and analyse the material that will make up planets. The analysis is done by the comparison of models to observations. However, here lie many challenges, especially regarding the computational speed of complex disk models.

During my PhD, we tackled this problem in different ways, including Machine Learning and the development of new disk models. We created Neural networks that can predict the spectral energy distributions (SEDs) of protoplanetary disks within milliseconds. This allowed us to perform a full Bayesian analysis for 30 well-known protoplanetary disks to determine their physical disk properties, including uncertainties and degeneracies. Additionally, we created a disk model that can describe the molecular line emission as well as the dust continuum as seen by the James Webb Space Telescope. We benchmarked this model and apply it to a JWST spectrum to extract the conditions under which molecules emit in protoplanetary disks.