A PINN approach learns galactic gravitational potentials from acceleration data, achieving sub-percent errors on simulations while outperforming analytic models and retaining interpretability via structured priors.
Constraining the Milky Way Halo Potential with the GD-1 stellar stream
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
We use ESA/Gaia astrometry together with SEGUE and LAMOST measurements of the GD-1 stellar stream to explore the improvement on the Galactic gravitational potential that these new data provide. Assuming a realistic universal model for the dark matter halo together with reasonable models of the baryonic components, we find that the orbital solutions for GD-1 require the circular velocity at the Solar radius to be $V_{\rm circ}(R_\odot) =244 \pm 4\,{\rm km\, s^{-1}}$, and also that the density flattening of the dark halo is $q_{\rho}=0.82^{+0.25}_{-0.13}$. The corresponding Galactic mass within 20kpc was estimated to be $M_{\rm MW}(<20kpc)=2.5 \pm 0.2 \times 10^{11}\, M_\odot$. Moreover, Gaia's excellent proper motions also allowed us to constrain the velocity dispersion of the GD-1 stream in the direction tangential to the line of sight to be $<2.30\,{\rm km\, s^{-1}}$ (95 % confidence limit), confirming the extremely cold dynamical nature of this system.
fields
astro-ph.GA 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Reconstructing Galactic Gravitational Potentials from Stellar Kinematics with Physics-Informed Neural Networks
A PINN approach learns galactic gravitational potentials from acceleration data, achieving sub-percent errors on simulations while outperforming analytic models and retaining interpretability via structured priors.