15–18 Dec 2026
Duke University
America/New_York timezone
Artificial Intelligence for the Electron Ion Collider (2026)

Machine learning based momentum scale determination and correction at sPHENIX

Not scheduled
20m
LSRC B101 Love Auditorium (Duke University)

LSRC B101 Love Auditorium

Duke University

Durham, NC 27710
Calibration, Monitoring, and Experimental Control in Streaming Environments Calibration, Monitoring, and Experimental Control in Streaming Environments

Speaker

Hao-Ren Jheng (Massachusetts Institute of Technology)

Description

A neural-network-based method is developed in this work to derive residual per-track momentum-scale corrections for the sPHENIX charged-particle reconstruction. The method addresses both charge-even and charge-odd momentum biases using kinematic constraints from reconstructed $K_{s}^{0}$ and $\Lambda^{0}$ decays. The correction is derived from reconstructed data and applied after track reconstruction, after which the standard secondary-vertex reconstruction and selection chain is rerun on the corrected tracks using the sPHENIX KFParticle package. The procedure is designed to reduce residual momentum-scale biases that remain after detector and tracking calibration. As the underlying calibration evolves, the neural-network-based correction can be retrained for subsequent data productions, with progressively smaller residual corrections expected as the calibration improves. After correction, the reconstructed $K_{s}^{0}$ and $\Lambda^{0}$ mass peaks move closer to their reference values, their mass resolutions improve, and the momentum dependence of the reconstructed masses is reduced. Independent validation with a resonance not used to derive the correction shows consistent improvements in the reconstructed mass scale and resolution. Closure and distortion-injection studies further show that the method produces negligible corrections for well-calibrated samples and recovers imposed momentum-scale distortions. The method is intended for integration into the sPHENIX tracking reconstruction framework, providing a data-driven procedure for deriving and updating residual single-track momentum-scale corrections as detector calibration and data production evolve.

Author

Hao-Ren Jheng (Massachusetts Institute of Technology)

Co-authors

Mr Alexander Patton (Massachusetts Institute of Technology) Prof. Cameron Dean (Iowa State University) Prof. Gunther Roland (Massachusetts Institute of Technology)

Presentation materials

There are no materials yet.