23–28 Aug 2026
America/Los_Angeles timezone

Machine Learning Optimization of E-Beam Transport in the ORGAD Accelerator

MOP34
24 Aug 2026, 16:00
2h
Poster Presentation Session 3: FEL theory and Machine Learning Monday Poster Session

Speaker

Amir Weinberg (Tel Aviv University)

Description

The 6MeV ORGAD Accelerator at Ariel University is driving a THz Superradiant FEL. The radiation is emitted superradiantly (in proportion to the number of electrons squared) in a rectangular waveguide-undulator section. The condition for the spontaneous emission to emit superradiantly is that the duration of the electron bunch σ_t is much shorter than the optical period (2π/ω) of the radiation as the beam propagates along the undulator. This requires proper choice of the acceleration phase and energy chirp in the hybrid RF gun. In addition, the parameters of the coils and quads along the transport linemust be adjusted to keep the beam cross-section dimensions small enough to enter the waveguide. Both goals are susceptible to space-charge limitations which tend to expand the beam in both transverse and longitudinal dimensions. This requires optimization of many transport parameters that are mutually dependent. In this paper we suggest an optimization method, using machine learning Bayesian optimization of the entire beamline with the goal of attaining minimal bunch duration in the undulator under limitation of good transport along the beamline. Full 3D general particle tracer (GPT) simulations, including space-charge effect, were applied using realistic field-maps which were measured in the lab. Thirty simulation iterations sufficed to arrive to optimal beam transport design.

Funding Agency

ISF -Israel Science Foundation

I have read and accept the Privacy Policy Statement Yes

Author

Amir Weinberg (Tel Aviv University)

Co-authors

Ariel Nause (Ariel University) Avraham Gover (Tel Aviv University) Mr Eyal Farchi (Ariel University) Leon Feigin (Ariel University)

Presentation materials

There are no materials yet.