When globular clusters and dwarf galaxies venture too close to the Milky Way, tidal forces tear them apart, creating spectacular stellar streams—cosmic rivers of stars that stretch across hundreds of thousands of light-years. These streams are among our most sensitive probes of dark matter, revealing invisible substructures through their gravitational influence on stellar trajectories.
StreamSculptor brings cutting-edge computational physics to stellar stream modeling, combining both perturbation theory and direct N-body simulation in a unified JAX framework. The toolkit offers GPU acceleration for computationally intensive simulations, automatic differentiation for parameter optimization and uncertainty quantification, and support for custom time-dependent galactic potentials. Researchers can model auxiliary physics along stellar trajectories and seamlessly switch between perturbative approximations for speed and full simulations for precision.
Built for the era of Gaia and upcoming surveys like LSST, StreamSculptor enables astronomers to fit complex models to observed stellar streams, constraining both the Milky Way’s gravitational potential and the properties of dark matter subhalos. The differentiable architecture makes it particularly powerful for Bayesian inference and machine learning applications, opening new avenues for discovering the invisible scaffolding of our galaxy.
⭐ Stars: 11
💻 Language: Python
🔗 Repository: jnibauer/streamsculptor