Researchers from LMU, the ORIGINS Excellence Cluster, the Max Planck Institute for Extraterrestrial Physics (MPE), and the ORIGINS Data Science Lab (ODSL) have made a crucial breakthrough within the evaluation of exoplanet atmospheres. Using physics-informed neural networks (PINNs), they’ve managed to model the complex light scattering within the atmospheres of exoplanets with greater precision than has previously been possible. This method opens up recent opportunities for the evaluation of exoplanet atmospheres, especially with regard to the influence of clouds, and will significantly improve our understanding of those distant worlds.
When distant exoplanets pass in front of their star, they block a small portion of the starlight, while a fair smaller portion penetrates the planetary atmosphere. This interaction results in variations in the sunshine spectrum, which mirror the properties of the atmosphere resembling chemical composition, temperature, and cloud cover. To find a way to investigate these measured spectra, nevertheless, scientists require models which might be able to calculating hundreds of thousands of synthetic spectra in a short while. Only by subsequently comparing the calculated spectra with the measured ones will we obtain information in regards to the atmospheric composition of the observed exoplanets. And what’s more, the highly detailed recent observations coming from the James Webb Space Telescope (JWST) necessitate equally detailed and sophisticated atmospheric models.
Rapid solving of complex equations due to AI
A key aspect of exoplanet research is the sunshine scattering within the atmosphere, particularly the scattering off clouds. Previous models were unable to satisfactorily capture this scattering, which led to inaccuracies within the spectral evaluation. Physics-informed neural networks offer a decisive advantage here, as they’re able to efficiently solving complex equations. Within the just-published study, the researchers trained two such networks. The primary model, which was developed without taking light scattering into consideration, demonstrated impressive accuracy with relative errors of mostly under one percent. Meanwhile, the second model incorporated approximations of so-called Rayleigh scattering — the identical effect that makes the sky seem blue on Earth. Although these approximations require further improvement, the neural network was in a position to solve the complex equation, which represents a crucial advance.
Interdisciplinary collaboration
These recent findings were possible due to a singular interdisciplinary collaboration between physicists from LMU Munich, the ORIGINS Excellence Cluster, the Max Planck Institute for Extraterrestrial Physics (MPE) and the ORIGINS Data Science Lab (ODSL), which is specialized in the event of latest AI-based methods in physics. “This synergy not only advances exoplanet research, but additionally opens up recent horizons for the event of AI-based methods in physics,” explains lead creator of the study David Dahlbüdding from LMU. “We would like to further expand our interdisciplinary collaboration in the long run to simulate the scattering of sunshine off clouds with greater precision and thus make full use of the potential of neural networks.”
