Deployment and validation of predictive 6-dimensional beam diagnostics through generative reconstruction with standard accelerator elements

TL;DR

The paper addresses the challenge of obtaining high-dimensional beam phase-space information by introducing a GPSR framework that uses only standard accelerator components to reconstruct the full 6D beam phase space $$(x, x', y, y', z, \delta)$$. It demonstrates, through simulations and a PAL-XFEL experiment, that quadrupole scans, dispersive measurements, and RF phase scans suffice to recover the complete phase-space distribution, including nonlinear features, with predictions of independent downstream measurements closely matching ground-truth observables. The approach avoids specialized diagnostics like transverse deflecting cavities, enabling broader applicability and paving the way for predictive diagnostics along beamlines and in digital twin frameworks. The results show robust reconstruction and predictive capability across configurations, highlighting the method's potential to improve accelerator performance with reduced hardware requirements.

Abstract

Understanding the 6-dimensional phase space distribution of particle beams is essential for optimizing accelerator performance. Conventional diagnostics such as use of transverse deflecting cavities offer detailed characterization but require dedicated hardware and space. Generative phase space reconstruction (GPSR) methods have shown promise in beam diagnostics, yet prior implementations still rely on such components. Here we present the first experimental implementation and validation of the GPSR methodology, realized by the use of standard accelerator elements including accelerating cavities and dipole magnets, to achieve complete 6-dimensional phase space reconstruction. Through simulations and experiments at the Pohang Accelerator Laboratory X-ray Free Electron Laser facility, we successfully reconstruct complex, nonlinear beam structures. Furthermore, we validate the methodology by predicting independent downstream measurements excluded from training, revealing near-unique reconstruction closely resembling ground truth. This advancement establishes a pathway for predictive diagnostics across beamline segments while reducing hardware requirements and expanding applicability to various accelerator facilities.

Figures (4)

• Figure 1: a. Schematics of the GPSR method PhysRevLett.130.145001. b: Cartoon showing an evolution of beam energy chirp by the accelerating field. c. Schematic view of the first bunch compressor diMitri2018beam section at the PAL-XFEL. Training and test datasets are measured at the screen 1 walasek2011scintillatingKim:IPAC2016-WEYB01. The lattice used for the reconstruction training process is in the dashed orange box. d. Validation section. The longitudinal phase space is measured at screen 2 using S-band TCAV and spectrometer diMitri2015layout [e. measured longitudinal phase space for BC and XLIN OFF]. Measured beam charge and reference energy for the reconstruction process are 250 pC and 260.5 MeV, respectively.
• Figure 2: Comparison between measured beam distributions and predictions at the screen 1. Training dataset is highlighted with orange box. $k$ indicates the quadrupole magnetic field strength in T/m. Contour lines indicate 90 and 50th percentiles, where solid(dashed) lines represent the measurement(prediction). Background image shows the measurement for a, while it illustrates the prediction for b.
• Figure 3: Comparison of LPSs and $(y,\delta)$ correlations at the screen 2. In case of the LPS, the beam head is placed at positive $z$ while the tail is at negative $z$. Background image indicates the measured distribution. The average kinetic energy is indicated in each sub-figure, as the $\delta$ is distributed with respect to zero. White dashed(yellow solid), projected histogram indicates predicted(measured) density distribution. White contour lines indicate 90 and 50th percentiles. Top row a-e: LPS where BC ON. Middle row f-j: LPS where BC OFF. Bottom row k-o: $(y,\delta)$ correlations with BC OFF.
• Figure 4: a. Reconstructed 6-dimensional phase space. Contour lines indicate 90 and 50th percentile, while different line styles represent three cases in the error study. "Reference" indicates the nominal case. Note that only 90th percentile is shown for $(z-\delta)$ distribution. Blue histogram indicates the 1-dimensional histogram in each coordinate. b. 3-dimensional scatter/density plot of the reconstructed beam phase space.