The discovery of planets similar to ours is at the forefront of concerns in planetary science today. For almost 30 years now, since the first exoplanet was discovered (Mayor & Queloz, 1995), more than 5000 new exoplanets have been detected using various observation techniques. Despite the advancements in technology in this field, detecting Earth-sized planets at temperate distances from their stars remains a significant challenge. With upcoming missions such as LIFE, whose objective is to discover such planets, it is crucial to prepare for these findings. Understanding the types of systems in which these planets form and being able to predict which systems have the highest probability of hosting an Earth-like planet will ultimately save observation time and ensure successful detections.
Davoult et al. (in prep) have investigated the properties of systems harboring Earth-like planets. They utilized populations of synthetic systems calculated using the Bern model (a planetary system formation model developed in Bern over the past decade by Y. Alibert, W. Benz, C. Mordasini, etc.) to study the correlations between the presence of an Earth-like planet and the properties of visible planets within a system. This study has produced a list of properties that seem to be linked to the presence of an Earth-like planet. The objective of the proposed project is to build upon the results of Davoult et al. and incorporate them into a Machine Learning model (Random Forest) to gain a clear understanding of systems that host an Earth-like planet versus those that do not. The project will also involve investigating different parameters and understanding the correlations found by connecting them to the theoretical physics behind planet formation and the Bern model.
The student will be co-supervised by Jeanne Davoult (Ph.D. candidate) and Yann Alibert (Professor). The internship will take place at the University of Bern, where the student will have an assigned workspace. The supervision of the project is flexible and can be discussed with the student. Proficiency in Python is necessary, but no experience in machine learning is expected for the successful completion of the project.
Contact : jeanne.davoult@unibe.ch; yann.alibert@unibe.ch