Published July 2025 | Version Published
Journal Article

Pointing Accuracy Improvements for the South Pole Telescope with Machine Learning

  • 1. ROR icon University of Chicago
  • 2. ROR icon Fermilab
  • 3. ROR icon University of Melbourne
  • 4. ROR icon Institut d'Astrophysique de Paris
  • 5. ROR icon Sorbonne University
  • 6. ROR icon Argonne National Laboratory
  • 7. ROR icon Stanford University
  • 8. ROR icon SLAC National Accelerator Laboratory
  • 9. ROR icon University of California, Berkeley
  • 10. ROR icon CEA Saclay
  • 11. ROR icon University of Illinois Urbana-Champaign
  • 12. ROR icon High Energy Accelerator Research Organization
  • 13. ROR icon McGill University
  • 14. ROR icon Canadian Institute for Advanced Research
  • 15. ROR icon Princeton University
  • 16. ROR icon University of Colorado Boulder
  • 17. ROR icon University of California, Los Angeles
  • 18. ROR icon Michigan State University
  • 19. ROR icon University of California, Davis
  • 20. ROR icon Northwestern University
  • 21. ROR icon Korea Advanced Institute of Science and Technology
  • 22. ROR icon Case Western Reserve University
  • 23. ROR icon University of Arizona
  • 24. ROR icon National Center for Supercomputing Applications
  • 25. ROR icon University of Toronto
  • 26. ROR icon California Institute of Technology
  • 27. ROR icon Harvard-Smithsonian Center for Astrophysics

Abstract

In this paper, we present improvements to the pointing accuracy of the South Pole Telescope (SPT) using machine learning. The ability of the SPT to point accurately at the sky is limited by its structural imperfections, which are impacted by the extreme weather at the South Pole. Pointing accuracy is particularly important during SPT participation in observing campaigns with the Event Horizon Telescope (EHT), which requires stricter accuracy than typical observations with the SPT. We compile a training dataset of historical observations of astronomical sources made with the SPT-3G and EHT receivers on the SPT. We train two XGBoost models to learn a mapping from current weather conditions to two telescope drive control arguments — one which corrects for errors in azimuth and the other for errors in elevation. Our trained models achieve root mean squared errors on withheld test data of 2.′′14 in cross-elevation and 3.′′57 in elevation, well below our goal of 5′′ along each axis. We deploy our models on the telescope control system and perform further in situ test observations during the EHT observing campaign in April 2024. Our models result in significantly improved pointing accuracy: for sources within the range of input variables where the models are best trained, average combined pointing error improved 33%, from 15.′′9 to 10.′′6. These improvements, while significant, fall shy of our ultimate goal, but they serve as a proof of concept for the development of future models. Planned upgrades to the EHT receiver on the SPT will necessitate even stricter pointing accuracy which will be achievable with our methods.

Copyright and License

© 2025 World Scientific Publishing Co Pte Ltd.

Acknowledgement

The South Pole Telescope program is supported by the National Science Foundation (NSF) through Awards OPP-1852617 and OPP-2332483. Partial support is also provided by the Kavli Institute of Cosmological Physics at the University of Chicago. Argonne National Laboratory's work was supported by the U.S. Department of Energy, Office of High Energy Physics, under Contract DE-AC02-06CH11357. The University of California, Davis group acknowledges support from Michael and Ester Vaida. Work at Fermi National Accelerator Laboratory, a DOE-OS, HEP User Facility managed by the Fermi Research Alliance, LLC, was supported under Contract No. DE-AC02-07CH11359. The Melbourne authors acknowledge support from the Australian Research Council's Discovery Project scheme (No. DP210102386). The Paris group has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program (Grant Agreement No. 101001897), and funding from the Centre National d'Etudes Spatiales. The SLAC group is supported in part by the Department of Energy at SLAC National Accelerator Laboratory, under Contract DE-AC02-76SF00515. PMC, TMC, AEL, and DPM acknowledge support from NSF Award AST-2034306. Preprint of an paper submitted for consideration in the Journal of Astronomical Instrumentation 2024.

Additional details

Related works

Is new version of
Discussion Paper: arXiv:2412.15167 (arXiv)

Funding

National Science Foundation
OPP-1852617
National Science Foundation
OPP-2332483
United States Department of Energy
DE-AC02-06CH11357
United States Department of Energy
DE-AC02-07CH11359
Australian Research Council
DP210102386
European Union
101001897
Centre National d'Études Spatiales
United States Department of Energy
DE-AC02-76SF00515
National Science Foundation
AST-2034306

Dates

Accepted
2025-04-23
Available
2025-07-02
Published

Caltech Custom Metadata

Caltech groups
Division of Physics, Mathematics and Astronomy (PMA)
Publication Status
Published