AI and the World's Hard Problems · Astronomy & space · arXiv · 2026-10-23
Tighter Dark Matter Constraints from the Projected Mass Method: A Neural Network Enhanced Method for Galaxy Groups and Clusters
The classical Projected Mass Estimator tends to overestimate galaxy-group masses when only a few satellite galaxies are available. The authors trained residual neural networks on simulated groups from the IllustrisTNG cosmological simulation to correct it, with separate networks for each satellite sample size. In tests on simulated halos, the overestimate fell from about 1.3–1.5 times to 1.02 times the true mass, and the Milky Way's total mass was estimated at about 1.14 × 10^12 solar masses, with a lower range of 0.8–0.95 × 10^12 from the brightest satellites.