FEL2019 - List of Keywords (target)

Paper	Title	Other Keywords	Page
WEB03	Application of Machine Learning to Beam Diagnostics	optics, network, diagnostics, controls	311
	E. Fol, R. Tomás CERN, Meyrin, Switzerland J.M. Coello de Portugal PSI, Villigen PSI, Switzerland G. Franchetti GSI, Darmstadt, Germany
	Machine Learning (ML) techniques are widely used in science and industry to discover relevant information and make predictions from data. The application ranges from face recognition to High Energy Physics experiments. Recently, the application of ML has grown also in accelerator physics and in particular in the domain of diagnostics and control. The target is to provide an overview of ML techniques and to indicate beam diagnostics tasks where ML based solutions can be efficiently applied to complement or potentially surpass existing methods. Besides, a short summary of recent works will be given demonstrating the great interest for use of ML concepts in beam diagnostics and latest results of incorporating these concepts into accelerator problems, with the focus on beam optics related application.
	Slides WEB03 [5.721 MB]
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-FEL2019-WEB03
About •	paper received ※ 20 August 2019 paper accepted ※ 27 August 2019 issue date ※ 05 November 2019
Export •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

THP013	User Operation of Sub-Picosecond THz Coherent Transition Radiation Parasitic to a VUV FEL	electron, linac, radiation, FEL	621
	S. Di Mitri, N. Adhlakha, E. Allaria, L. Badano, G. De Ninno, P. Di Pietro, G. Gaio, L. Giannessi, G. Penco, A. Perucchi, P. Rebernik Ribič, E. Roussel, S. Spampinati, C. Spezzani, M. Trovò, M. Veronese Elettra-Sincrotrone Trieste S.C.p.A., Basovizza, Italy G. De Ninno University of Nova Gorica, Nova Gorica, Slovenia L. Giannessi ENEA C.R. Frascati, Frascati (Roma), Italy S. Lupi Coherentia, Naples, Italy S. Lupi Sapienza University of Rome, Roma, Italy F. Piccirilli IOM-CNR, Trieste, Italy E. Roussel PhLAM/CERLA, Villeneuve d’Ascq, France E. Roussel PhLAM/CERCLA, Villeneuve d’Ascq Cedex, France
	Coherent transition radiation is enhanced in intensity and extended in frequency spectral range by the electron beam manipulation in the beam dump beam line of the FERMI FEL, by exploiting the interplay of coherent synchrotron radiation instability and electron beam optics [1]. Experimental observations at the TeraFERMI beamline [2] confirm intensity peaks at around 1 THz and extending up to 8.5 THz, for up to 80 µJ pulse energy integrated over the full bandwidth. By virtue of its implementation in an FEL beam dump line, this work might stimulate the development of user-oriented multi-THz beamlines parasitic and self-synchronized to VUV and X-ray FELs. [1] S. Di Mitri et al., Scientific Reports, 8, 11661 (2018). [2] A. Perucchi et al., Synch. Rad. News 4, 30 (2017).
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-FEL2019-THP013
About •	paper received ※ 29 July 2019 paper accepted ※ 27 August 2019 issue date ※ 05 November 2019
Export •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

THP061	Bayesian Optimisation for Fast and Safe Parameter Tuning of SwissFEL	FEL, undulator, feedback, controls	707
	J. Kirschner, A. Krause, M. Mutný, M. Nonnenmacher ETH, Zurich, Switzerland A. Adelmann, N. Hiller, R. Ischebeck PSI, Villigen PSI, Switzerland
	Parameter tuning is a notoriously time-consuming task in accelerator facilities. As tool for global optimization with noisy evaluations, Bayesian optimization was recently shown to outperform alternative methods. By learning a model of the underlying function using all available data, the next evaluation can be chosen carefully to find the optimum with as few steps as possible and without violating any safety constraints. However, the per-step computation time increases significantly with the number of parameters and the generality of the approach can lead to slow convergence on functions that are easier to optimize. To overcome these limitations, we divide the global problem into sequential subproblems that can be solved efficiently using safe Bayesian optimization. This allows us to trade off local and global convergence and to adapt to additional structure in the objective function. Further, we provide slice-plots of the function as user feedback during the optimization. We showcase how we use our algorithm to tune up the FEL output of SwissFEL with up to 40 parameters simultaneously, and reach convergence within reasonable tuning times in the order of 30 minutes (< 2000 steps).
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-FEL2019-THP061
About •	paper received ※ 13 August 2019 paper accepted ※ 27 August 2019 issue date ※ 05 November 2019
Export •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

Paper

Title

Other Keywords

Page

WEB03

Application of Machine Learning to Beam Diagnostics

optics, network, diagnostics, controls

311

E. Fol, R. Tomás
CERN, Meyrin, Switzerland
J.M. Coello de Portugal
PSI, Villigen PSI, Switzerland
G. Franchetti
GSI, Darmstadt, Germany

Machine Learning (ML) techniques are widely used in science and industry to discover relevant information and make predictions from data. The application ranges from face recognition to High Energy Physics experiments. Recently, the application of ML has grown also in accelerator physics and in particular in the domain of diagnostics and control. The target is to provide an overview of ML techniques and to indicate beam diagnostics tasks where ML based solutions can be efficiently applied to complement or potentially surpass existing methods. Besides, a short summary of recent works will be given demonstrating the great interest for use of ML concepts in beam diagnostics and latest results of incorporating these concepts into accelerator problems, with the focus on beam optics related application.

Slides WEB03 [5.721 MB]

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-FEL2019-WEB03

About •

paper received ※ 20 August 2019 paper accepted ※ 27 August 2019 issue date ※ 05 November 2019

Export •

reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

THP013

User Operation of Sub-Picosecond THz Coherent Transition Radiation Parasitic to a VUV FEL

electron, linac, radiation, FEL

621

S. Di Mitri, N. Adhlakha, E. Allaria, L. Badano, G. De Ninno, P. Di Pietro, G. Gaio, L. Giannessi, G. Penco, A. Perucchi, P. Rebernik Ribič, E. Roussel, S. Spampinati, C. Spezzani, M. Trovò, M. Veronese
Elettra-Sincrotrone Trieste S.C.p.A., Basovizza, Italy
G. De Ninno
University of Nova Gorica, Nova Gorica, Slovenia
L. Giannessi
ENEA C.R. Frascati, Frascati (Roma), Italy
S. Lupi
Coherentia, Naples, Italy
S. Lupi
Sapienza University of Rome, Roma, Italy
F. Piccirilli
IOM-CNR, Trieste, Italy
E. Roussel
PhLAM/CERLA, Villeneuve d’Ascq, France
E. Roussel
PhLAM/CERCLA, Villeneuve d’Ascq Cedex, France

Coherent transition radiation is enhanced in intensity and extended in frequency spectral range by the electron beam manipulation in the beam dump beam line of the FERMI FEL, by exploiting the interplay of coherent synchrotron radiation instability and electron beam optics [1]. Experimental observations at the TeraFERMI beamline [2] confirm intensity peaks at around 1 THz and extending up to 8.5 THz, for up to 80 µJ pulse energy integrated over the full bandwidth. By virtue of its implementation in an FEL beam dump line, this work might stimulate the development of user-oriented multi-THz beamlines parasitic and self-synchronized to VUV and X-ray FELs.
[1] S. Di Mitri et al., Scientific Reports, 8, 11661 (2018).
[2] A. Perucchi et al., Synch. Rad. News 4, 30 (2017).

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-FEL2019-THP013

About •

paper received ※ 29 July 2019 paper accepted ※ 27 August 2019 issue date ※ 05 November 2019

Export •

reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

THP061

Bayesian Optimisation for Fast and Safe Parameter Tuning of SwissFEL

FEL, undulator, feedback, controls

707

J. Kirschner, A. Krause, M. Mutný, M. Nonnenmacher
ETH, Zurich, Switzerland
A. Adelmann, N. Hiller, R. Ischebeck
PSI, Villigen PSI, Switzerland

Parameter tuning is a notoriously time-consuming task in accelerator facilities. As tool for global optimization with noisy evaluations, Bayesian optimization was recently shown to outperform alternative methods. By learning a model of the underlying function using all available data, the next evaluation can be chosen carefully to find the optimum with as few steps as possible and without violating any safety constraints. However, the per-step computation time increases significantly with the number of parameters and the generality of the approach can lead to slow convergence on functions that are easier to optimize. To overcome these limitations, we divide the global problem into sequential subproblems that can be solved efficiently using safe Bayesian optimization. This allows us to trade off local and global convergence and to adapt to additional structure in the objective function. Further, we provide slice-plots of the function as user feedback during the optimization. We showcase how we use our algorithm to tune up the FEL output of SwissFEL with up to 40 parameters simultaneously, and reach convergence within reasonable tuning times in the order of 30 minutes (< 2000 steps).

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-FEL2019-THP061

About •

paper received ※ 13 August 2019 paper accepted ※ 27 August 2019 issue date ※ 05 November 2019

Export •

reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)