17–22 May 2026
C.I.D
Europe/Zurich timezone

Machine-Learning–Assisted Bayesian Uncertainty Quantification for Accelerator Digital Twin Modeling and Control

MOP6310
18 May 2026, 16:00
2h
C.I.D

C.I.D

Deauville, France
Poster Presentation MC6.D13: Instrumentation: Artificial Intelligence Poster session

Speaker

Georg Hoffstaetter (Cornell University (CLASSE))

Description

Digital twins of particle accelerators are used to plan and control operations and to design data collection campaigns. Accurate modeling typically requires knowledge of quantities that are hard to measure directly, e.g., magnet alignments, power supply transfer functions, magnet nonlinearities, and stray fields. In this work we introduce parameters to describe these effects and use Bayesian methods to probabilistically estimate their values and uncertainties by calibrating digital twins to beam measurements performed at the AGS and its Booster at Brookhaven National Laboratory. The inference is computationally accelerated using a machine learning emulator of the physical accelerator digital twin trained to a perturbed-parameter ensemble of Bmad simulations. The result is a joint posterior distribution over the parameters constrained by the data and taking into account beam monitor errors. Incorporating estimates of these parameters into the digital twin is shown to result in a significant improvement in the quality of the model and provides error bars on the model parameters and predictions.

Funding Agency

Work supported by Brookhaven Science Associates, LLC under Contract No. DE-SC0012704 with the U.S. Department of Energy and by DOE-NP No. DE SC-0024287.

In which format do you inted to submit your paper? LaTeX

Author

Weijian Lin (Brookhaven National Laboratory)

Co-authors

Christopher Kelly (Brookhaven National Laboratory) Eiad Hamwi (Cornell University (CLASSE)) Georg Hoffstaetter (Cornell University (CLASSE)) Kevin Brown (Brookhaven National Laboratory) Nathan Urban (Brookhaven National Laboratory)

Presentation materials

There are no materials yet.