Steve Russel: Current Status of the AMAS code OPAL-T and its Application to the 250 MeV PSI-FEL Injector Room: WBGB 019 Selected questions and discussion: BP: What happened to these Los Alamos projects you mentioned? SRu: They were not funded. Instead of APT, a nuclear reactor was chosen, in the case of AWT it was dropped altogether. TS: How come Rob Ryne could simulate a billion particles ten years, while we are struggling with 100 million today? SRu: Well, it probably was not quite a billion back then(?) Followup: Rob Ryne started his foray into massive particle simulations with the development of the code IMPACT about 10 years ago in the mid 1990s. In the late 1990s this code was rewritten to take advantage of new parallel computing techniques. He did not immediately achieve 1 billion particles. In fact, it was not until more recently that this was achieved (e.g. J. Qiang, I. Pogorelov and R. Ryne, Proceedings of the 2007 Particle Accelerator Conference, p. 3522 (2007)).

Review on the 1st Full Start-to-End Simulations for the PSI-XFEL project with CTF3 RF gun Part-I : Yujong Kim - Design Concepts of Linac and its Performance Room: WBGB 019 Selected questions and discussion: VS: How well do we have to control the laser profile? YK: This was discussed with CH, we can reach a rise and fall time of 0.5 ps, in my simulation I have assumed 0.7 ps. MD: A very hard edge does not really help for low emittance, 0.7 ps is certainly enough! JYR: Despite the wakefields you still operate LINAC2 on-crest? YK: Yes. MP: What have you budgeted for the total length? YK: After BC2 diagnostics, I allow for 60 m for DIAG3, dog leg and switch yard. VS (on p. 17): Why do we have such a nice Gauss profile? YK: We have the X-band linearizer (FLASH does not)! AO: How does your number of S-band tubes compare to that used by Rene Bakker? YK: I need 90 tubes for 6 GeV, don't know the number in Rene's design. Clearly when running off-crest you need more tubes than on-crest. JYR (on p. 59): Be careful about the Japanese 40 MV/m S-band results, there is no information on the breakdown rate! SR: You are using a lot of off-crest acceleration to get rid of the chirp. Have you considered using X-band instead? YK: The RF jitter tolerance would be too severe for X-band acceleration! We might consider an X-band structure without RF to add short-range wakefield. MD: At the very end of the linac, you are far less sensitive to jitter. VS: The undulator has 5 mm aperture, we will need precise control of the beam anyway! JYR: But if you run off-crest, you are more sensitive to jitter... JYR: What is the amplitude stability? YK: We need 0.01 percent stability in RF amplitude (phase 0.01 degree) before BC2, after that it will be less critical. SRe (on p. 49): With 40 degrees off-crest acceleration you are wasting a lot of energy! YK: It depends on the chirp, i.e. FEL performance, decision must come from the user side! AA: What is your error bar on the 0.33 emittance? YK: From the convergence with particle number I guess it is around 0.4 (50k give 0.3416, 200k give 0.3454) AA: What grid are you using? YK: In ASTRA 15 radial and 20 longitudinal, for CSR studies 150 time slices in the bunch compressor. FLP: So you're assuming 0.2 initial emittance for the CERN gun, but what if it is 0.3? Will it be a problem for the project? YK: All slice parameters go up, so we will end up at 0.4 rather than 0.3. The facility would become longer correspondingly

Part-II : Sven Reiche - Design Concepts of Undulators and FEL Performance Room: WBGB 019 Selected questions and discussion: FLP: Why not go to the Si edge (14 nm)? SRe: Then you have a tunability problem, cannot cover all the way up from 1.8 nm to 14 nm! BP: The seeded undulator line is saturated? SRe: Yes, it saturates at 20 m. But we may need an additional undulator module before. YK: We can turn off X-band for short pulses. SRe: This works for VUV, I doubt it will work for X-ray because you run into overcompression. FLP: Do we need 50% overhead in the undulator? What about the unusued undulator section? SRe: This is a typical overhead for FELs. The unused undulator sections are just openened. You would open the first ones, not the last ones. But at the beginning we will certainly need the full 60 m! LR: Have you benchmarked your undulator simulation against data from FLASH? SRe: In principle it is benchmared, but keep in mind that the limitation is mainly in the quality of the input parameters, not the model. Therefore you can explain almost any result with the simulation! MP: If we reach the baseline 0.2 mm mrad, how could we use the 50% overhead? Tapering? SRe: The unused undulator modules can be used for taper to enhance the emitted power but is only suitable for linear polarized radiation because the last three undulator modules, which are of APPLE type, are not effective anymore to define the polarization. YK: According to your simulation 1.5 kA gives better performance than 2.7 kA, since nonlinear chirp and energy spread are smaller for 1.5 kA. SR: With 2.7 kA you have more margin for error, but overall worse performance. VS: We need a quantitative table for tolerance margins! SR quickly lists a few numbers... There's a pay-off between diagnostics and beam stability. The requirements for energy and current jitter are rather relaxed. The strongest constraint comes from transverse jitter (alignment error). AA: How does your simulation depend on uncorrelated energy spread? SR: Right now I am using 500 keV intrinsic energy spread. But there is a much bigger effect from the betatron amplitude (acts like an energy spread, which completely masks the intrinsic energy spread) FLP: The wakefields you have shown at the beginning seem pretty high? SRe: They are not high at all, they are a factor 10 larger at LCLS! The tolerance means how much energy spread etc. I can allow. Typically of order 50 keV/m. The tolerance is within the bandwidth of FEL. A slight positive taper is benefitial to the FEL performance because it compensates the gradual energy loss of the electron beam during the FEL amplification.