arxiv.org/abs/2409.04557 A Deep Learning Framework for TTV Characterization: A Case Study of the Kepler-88 System
Researchers built a Transformer-based deep learning model that infers the mass and orbit of an unseen outer planet from transit timing and duration variations. Tested on Kepler-88 with a million simulations, it recovered semi-major axis and eccentricity to within about 10 percent and planet mass to roughly 10-30 percent.
— via arXiv — Astro-ph, Chen Chen, Jiapeng Gao, Gongjie Li, Lingkai Kong, Molei Tao
