\(\partial\textrm{SGP4}\) Documentation#
dsgp4 is a differentiable SGP4 program written leveraging the PyTorch machine learning framework: this enables features like automatic differentiation and batch propagation (across different TLEs) that were not previously available in the original implementation. Furthermore, it also offers a hybrid propagation scheme called ML-dSGP4 where dSGP4 and ML models can be combined to enhance SGP4 accuracy when higher-precision simulated (e.g. from a numerical integrator) or observed (e.g. from ephemerides) data is available.
For more details on the model and results, check out our publication: Acciarini, Giacomo, Atılım Güneş Baydin, and Dario Izzo. “Closing the Gap Between SGP4 and High-Precision Propagation via Differentiable Programming” (2024) Vol. 226(1), pages: 694-701
The authors are Giacomo Acciarini, Atılım Güneş Baydin, Dario Izzo. The main developer is Giacomo Acciarini (giacomo.acciarini@gmail.com).
Getting Started