|Time:||16:00 23 Feb 2005|
|Participants:||T. Asaka, H. Braun, H. Burkhardt, R. Corsini, P. Eliasson, J. Resta Lopez, D. Schulte, F. Tecker, F. Zimmermann|
Daniel started by mentioning that he has been talking to Catherina about Elegant. She had told him that it may not be that easy to install. She will come to CERN to help and give some information about the code.
Hans reported from the wiggler workshop that there are many codes that can be used for simulations of wigglers. It would also be possible to implement wigglers in MAD. It is not straight-forward though since the wiggler cannot be described as a sum of multipoles. Instead it could be done by generating a fieldmap describing the wiggler. Maxim could take care of this, but in that case he would like it to be a part of his thesis and would have to finish before the end of May.
Takao and Javier had come to the conclusion that in SAD the SR-effects are calculated both in quads and in bending magnets while in MAD only the SR in the bendings is simulated.
Helmut proposed that Javier should talk to John Jowett to find out if there are any options for calculating the SR in the quads with MAD8.
Takao and Javier also showed plots of the particle distribution at the IP after tracking through the BDS taking SR into account (see link below). The plots obtained with SAD differed quite much from the ones obtained with MAD, especially in the vertical plane. Frank T mentioned that he knows about a problem with the initial energy in the particular version of MAD8 that Javier is using. Frank Z said it might also be a problem with the position of the waist, but according to Javier the α- and β-values at the IP agree between the two codes. Daniel proposed that phase space plots should be made to sort out the problems. Frank Z also proposed that the simulations should be done without energyspread to see if there is a chromaticity problem.
SAD particle distribution, x
SAD particle distribution, y
MAD particle distribution, x
MAD particle distribution, y
Peder showed a plot of emittance versus gain at the end of the main linac (see link below). This showed that for 40 dependant feedbacks the optimal gain is 0.02. The optimal gain agreed with what Daniel obtained some years ago when he simulated independant feedbacks. At that time Daniel assumed a bpm resolution of 100nm. The dependant feedbacks were less sensitive to noise and a bpm resolution of 400nm gave comparable emittances to the ones in the old simulations.
Emittance vs gainDaniel pointed out that a model of the emittance growth, where the different contributions are separated, should be created:
Δ = Δεuncorrτ + Δεcorrt + Δεfeed(σres, g).
Possibly the luminosity could be modeled in a similar fashion.
Daniel then explained that he will have to make a new lattice design since the girder length will be 2.06m instead of 2.23m. He also said that Igor has created a new model of the PETS that Daniel has included in Placet. When he ran simulations with this new model the beam was killed though.
Helmut has been trying to speed up his old SR routines even further and is also for the moment working on adding these routines to the GEANT4 code, since the table method used now is not good enough. There were some different opinions about whether this is really the case. Some people thought that the table method is good enough. Daniel proposed that a case-study should be done to see which is the best solution.
Finally Daniel mentioned that it would be intesting to find some good routines for random number generation with the possibility to save and recall seeds.
Maintained by: Peder Eliasson
Last updated: 14 Mar 2005