Détection et caractérisation d'exoplanètes à l'aide d'algorithmes keplerian // detection and characterization of exoplanets using keplerian-stacker algorithms

Topic description

La majorité des 7 exoplanètes découvertes l'ont été par des méthodes indirectes (vitesse radiale, photométrie, astrométrie), limitant l'accès à leurs propriétés physiques et atmosphérique. L'imagerie directe, reste contrainte par le contraste des instruments actuels (SPHERE, GPI : ~10⁶ à mas), ne détectant que des géantes gazeuses jeunes et massives, et échouant pour les planètes proches de la ligne de glace ( 10 h), durant lesquels le mouvement orbital dégrade fortement le rapport signal/bruit.
Le projet Keplerian-Stacker propose une approche nouvelle en transformant le mouvement orbital en atout. Plutôt que de s'appuyer sur des détections mono-époque, cette méthode exploite la dynamique képlérienne des exoplanètes (Le Coroller et al.,, ; OCTOFITTER : Thompson et al. ) pour combiner des observations multi-époques et multi-techniques (ex: imagerie directe, vitesses radiales). En ajustant simultanément le flux et les paramètres orbitaux sur l'ensemble des données, ces algorithmes détectent des planètes avec un S/N 10 AU, and failing for planets close to the ice line ( 10 hours), during which orbital motion severely degrades the signal-to-noise ratio.
The Keplerian-Stacker algorithms offers a novel approach by turning orbital motion into an advantage. Instead of relying on single-epoch detections, this method leverages the Keplerian dynamics of exoplanets (Le Coroller et al.,, ; OCTOFITTER: Thompson et al. ) to combine multi-epoch and multi-technique observations (e.g., direct imaging, radial velocities). By simultaneously fitting the flux and orbital parameters across all data, these algorithms detect planets with S/N < 2, otherwise lost in the noise. This approach reduces the observation time required for large surveys and maximizes scientific return by enabling robust characterization (atmosphere, orbit) of fainter targets previously inaccessible.
Main objectives of the thesis:
1.We will jointly infer planetary flux and orbital parameters using a unified MCMC framework, explicitly modeling reflected light variations as a function of phase, a groundbreaking approach for characterizing atmospheres and surfaces. The student will develop new likelihood functions to combine high-contrast imaging, astrometry (ex: GAIA), and radial velocities, enabling a coherent analysis of all data. Additionally, K-Stacker will be adapted for the first time to cross-correlation techniques in high-dispersion spectroscopy (Snellen et al. ) and will integrate an N-body orbital fitting module (Beust et al., ), pushing the detection limits of current multi-epoch algorithms.
2.Design and train machine learning models to identify orbital signatures, complementing K-Stacker's classical algorithms to improve detection sensitivity, especially for low-mass planets.
3.Develop a modular software infrastructure: Create a complete reduction pipeline (RAW→ASDI→KS) to validate the algorithms (1. and 2.), assess optimal multi-epoch observation strategies, and process real observational data. The student will implement this infrastructure on the LAM computing cluster to automate the processing of large datasets and generate detection statistics (ROC curves).
This project will directly support the goals of the Roman Space Telescope by enabling the detection of Jupiter-like planets around nearby stars by the end of the thesis (e.g., observing eps Eri b with RST). It will also lay the groundwork for HWO observations, with the ambition to characterize dozens of exoplanets near the ice line, many of which would remain undetectable without these algorithms. By combining indirect and direct methods, this work will contribute to the search for biosignatures, a central objective of HWO, and advance our understanding of planetary system architectures. The K-Stacker algorithm will also be applicable to the characterization of rocky planets using ground-based observations (ELT-PCS).
------------------------------------------------------------------------------------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Début de la thèse : 01/09/
WEB :

Funding category

Funding further details

Programmes de l'Union Européenne de financement de la recherche (ERC, ERASMUS)