Machine learning based phase space tomography using kicked beam turn-by-turn centroid data in a storage ring

Aug 6, 2026

Upload by Anda

Kicked transverse beam measurements in a storage ring enable phase space reconstruction when the centroid time series reflects both lattice focusing and the Fourier transform of the projected 1D beam density along the kick angle. Under nonlinear focusing, a 2D computational tomography approach is presented using only turn-by-turn beam position monitor data, avoiding rotating beam angles through repeated destructive measurements. The work demonstrates multiple tomography strategies, including machine learning methods, compares their performance, and evaluates reliability. A potential extension to 4D computational tomography is discussed.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages17

Reading Time43 min

AccessFree to download

Introduction

Phase-space importance in accelerator operation

Core idea of kicked-beam tomography via centroid Fourier relation

Limitations of conventional reconstruction methods

Model discrepancies and reliability challenges

Lawrence Berkeley National Laboratory LBL Publications Title Machine learning based phase space tomography using kicked beam turn-by-turn centroid data in a storage ring Permalink [https://escholarship.org/uc/item/8269q2s1](https://escholarship.org/uc/item/8269q2s1) Journal Physical Review Accelerators and Beams, 26(10) ISSN 1098-4402 Authors Hwang, Kilean Mitchell, Chad Ryne, Robert Publication Date 2023-10-02 DOI 10.1103/physrevaccelbeams.26.104601 Copyright Information This work is made available under the terms of a Creative Commons Attribution License, available at [https://creativecommons.org/licenses/by/4.0/](https://creativecommons.org/licenses/by/4.0/) Peer reviewed [eScholarship.org](eScholarship.org) Powered by the California Digital Library University of California Machine learning based phase space tomography using kicked beam turn-by-turn centroid data in a storage ring Kilean Hwang* Facility for Rare Isotope Beams, Michigan State University, East Lansing, Michigan, USA Chad Mitchell and Robert Ryne Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, California 94720, USA (Received 22 May 2023; accepted 2 October 2023; published 23 October 2023) When a charged-particle bunch in a storage ring is kicked to a large transverse offset, the time series describing the dynamics ofthe bunch centroid is determined both by the lattice focusing and by the Fourier transform of the 1D density profile of the bunch projected along the angle of the kick. In the presence of nonlinear focusing, we show that this fact can be exploited to enable 2D phase space reconstruction of the bunch (computational tomography) based only on turn-by-turn beam position monitor data. We demonstrate various tomography methods based on this principle, including machine learning methods, and discuss their advantages and disadvantages, and measure of reliability. We also mention a possible extension to 4D phase space computational tomography. DOI: 10.1103/PhysRevAccelBeams.26.104601 I. INTRODUCTION Information regarding the particle beam phase-space density can be essential for understanding and solving various beam dynamics issues that an accelerator physicist may encounter during accelerator operation. In addition, detailed knowledge of the beam phase-space density is essential to the success of advanced phase space manipulation techniques in FEL light sources [1], to characterizing beam halo and predicting beam loss in high-intensity proton linacs and rings [2,3], and to implementing effective diagnostics in storage rings with complex nonlinear dynamics [4] . The cornerstone of our tomography method lies in the theoretical finding that is described in this paper and a previous work [5] . It shows that, in the limit of substantial kick strength, the temporal evolution of the beam centroid can be directly related to the Fourier transform of the 1D beam profile along the angle of the applied kick. Here, a kicked beam refers to single or multiple bunches of particles that have been given a one-time transverse momentum kick by an external force. Consequently, employing a multishot measurement involving multiple kicks at various angles within the phase-space domain *[hwang@frib.msu.edu](hwang@frib.msu.edu) Published by the American Physical Society under the terms of the Creative Commons Attribution 4.0 International license. Further distribution of this work must maintain attribution to the author(s) and the published article’s title, journal citation, and DOI. yields two-dimensional (2D) phase-space density distribution of the beam. However, the theoretical prediction is limited to large kicks that can result in the loss of the beam hitting the beam pipe. In cases where the kicks are sufficiently small to avoid inducing beam loss, the connection between beam profile and the temporal evolution of the beam centroid becomes too complex for theoretical analysis. In such a

FAQs

How does the kicked-beam method relate centroid motion to the beam density profile?

With sufficiently strong kicks, the centroid time evolution directly corresponds to the Fourier transform of the 1D beam profile projected along the kick angle. This provides the basis for reconstruction from centroid measurements.

Why is machine learning useful when kicks are small?

For smaller kicks that avoid beam loss, the relationship between beam profile and centroid time evolution becomes too complex for straightforward theoretical treatment. Machine learning can model this complex mapping from turn-by-turn data.

What data source does the proposed 2D tomography rely on?

The reconstruction is performed using only turn-by-turn beam position monitor data, rather than camera images or wire/laser scanner profiles at multiple rotated angles.

What practical issues affect conventional phase-space tomography accuracy?

Conventional techniques depend on physics models that predict phase-space rotation; nonlinear optics and discrepancies between model and real machine reduce precision. Additional model errors accumulated during beam transport and distant measurement locations further degrade reliability.

Aug 6, 2026

Upload by Anda