The Maunder Model and Catalog: Stellar Rotation, Bimodal Activity, and Magnetic Braking in Kepler Main-Sequence Stars

概要

We present The Maunder, a machine learning pipeline and resulting catalog of rotation periods for 148,746 main-sequence stars in the Kepler field. To overcome single-catalog systematics and the simulation-to-reality gap, our architecture employs a hybrid training objective: a joint-embedding self-supervised loss applied to all light curves, combined with a supervised loss trained strictly on cross-catalog consensus labels. By processing multi-scale time- and frequency-domain inputs over rolling…