A neural network trained on 46 million stellar models dates the Pleiades at 113 million years

By simulating the pulsations of rotating δ Scuti stars, two astronomers have determined the age of the cluster from five stars observed by TESS, in line with conventional methods.

Stars of type δ Scuti vibrate in modes whose frequencies depend on their mass, composition, rotation and age. Exploiting these frequencies one by one to date young stars requires the computation of countless models, which makes complete statistical exploration very expensive. Anuj Gautam and Simon J. Murphy propose a neural emulator trained on about 46 million rotationally aware pulsation patterns: from four parameters, it predicts classical observables and mode frequencies, including the doubling caused by rotation.

An age consistent with lithium

Coupled with Bayesian sampling, the tool was applied to five Scuti δ of the Pleiades cluster observed by the TESS satellite. The individual ages, between 107 and 117 million years ago, agree; combined, they give 113.3 ± 4.5 million years, in agreement with dating by the lithium depletion limit (112 ± 5 million years). The authors also measured the sensitivity of the method to physical assumptions, with median shifts of 2 to 5 million years, and see it as a complementary method for dating young stars.