Author: Edelen, A.L.
Paper Title Page
Machine Learning Demonstrations on Accelerators  
  • A.L. Edelen
    SLAC, Menlo Park, California, USA
  Machine learning has been used in various ways to improve acclerator operation including the development of surrogate models to improve real-time modeling, advanced optimization of accelerator operating configurations such as quadrupole or undulator strengths, development of virtual diagnostics to ’measure’ accelerator and beam parameters, and prognostics to improve operating time.  
slides icon Slides THXBA1 [31.075 MB]  
Export • reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)  
FRXBA4 Maximizing 2-D Beam Brightness Using the Round to Flat Beam Transformation in the Ultralow Charge Regime 986
SUPLM02   use link to see paper's listing under its alternate paper code  
  • F.W. Cropp V, P.E. Denham, J. Giner Navarro, E.T. Liu, P. Musumeci
    UCLA, Los Angeles, USA
  • N. Burger, L. Phillips
    PBPL, Los Angeles, USA
  • A.L. Edelen, C. Emma
    SLAC, Menlo Park, California, USA
  Funding: This work is supported by the United States National Science Foundation award PHY-1549132 (the Center for Bright Beams)
We seek to maximize the 2-D beam brightness in an RF photoinjector operating in an ultralow charge (<1 pC) regime by implementing the FBT. Particle tracking simulations suggest that in one dimension, normalized projected emittances smaller than 5 nm can be obtained at the UCLA Pegasus facility with up to 100 fC beam charge. A tunable magnetic field is put on the cathode. Three skew quadrupoles are used to block-diagonalize the beam matrix and recover the vastly different eigenemittances as the projected emittances. Emittance measurement routines, including grid-based, pepperpot-based and quad scan routines, have been developed for on-line calculation of the 4-D beam matrix and its eigenemittances. Preliminary measurements are in agreement with simulations and indicate emittance ratios larger than 10 depending on the laser spot size on the cathode. Fine tuning the quadrupole gradients for the FBT has a significant effect on the 2-D beam brightness. We have made concrete steps toward computer minimization and machine learning optimization of the quadrupole gradients in order to remove the canonical angular momentum from the beam and achieve the target normalized projected emittances.
slides icon Slides FRXBA4 [3.059 MB]  
DOI • reference for this paper ※  
About • paper received ※ 28 August 2019       paper accepted ※ 05 December 2019       issue date ※ 08 October 2019  
Export • reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)