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

A Hybrid Physics and Data-Driven Modeling Framework for Accelerators with Automatic Differentiation

THV5303
21 May 2026, 16:00
2h
C.I.D

C.I.D

Deauville, France
Invited poster MC5.D11: Code Developments and Simulation Techniques Poster session

Speaker

Yue Hao (Facility for Rare Isotope Beams)

Description

In modern accelerator modeling, many lattice components can be accurately described using established first-principles physics. However, certain intricate effects, such as complex boundary conditions, collective interactions, and self-fields, remain difficult to model reliably and efficiently from theory alone. At the same time, high-resolution beam position and profile measurements provide rich information about these poorly understood dynamics. In this paper, we present a hybrid modeling framework with built-in automatic differentiation, designed to seamlessly integrate physics-based lattice models with data-driven representations of complex effects. This approach improves predictive accuracy, enables gradient-based optimization, and offers a practical path toward more faithful digital twins of accelerator systems.

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

Author

Yue Hao (Facility for Rare Isotope Beams)

Co-authors

Christian Ratcliff (Facility for Rare Isotope Beams) Helena Alamprese (Facility for Rare Isotope Beams) Ji Qiang (Lawrence Berkeley National Laboratory) Jinyu Wan (Facility for Rare Isotope Beams) William Fung (Facility for Rare Isotope Beams)

Presentation materials

There are no materials yet.