Page 1 of 2

CLEAN in Pelegant

Posted: 20 Apr 2011, 12:48
by ding
Hello,

I noticed that if I set CLEAN such as CLEAN, DELTALIMIT =1.5, it won't work with Pelegant, though works well in Elegant. However, put a larger number like CLEAN, DELTALIMIT=1.8 it works. I am confused with it.

thanks,
Yuantao

Re: CLEAN in Pelegant

Posted: 21 Apr 2011, 07:58
by michael_borland
Yuantao,

Can you post some files that show the problem so I can test it more easily?

Thanks--Michael

Re: CLEAN in Pelegant

Posted: 21 Apr 2011, 17:32
by ding
Michael,

while I set up a very simple example, it works for both elegant and Pelegant. I need some time to look into it first before asking help again.

thanks,
Yuantao

Re: CLEAN in Pelegant

Posted: 27 May 2011, 13:52
by ding
Hi Michael,

I uploaded an example, where RCOL is used in the middle of the chicane. When run Pelegant, i can have output at the location after the RCOL, and also at the end of the chicane. But actually the Charge parameter was not correct in the watchout files (still 20pC as before collimation). But after that, it won't continue to run into the following linacs. This is similar with CLEAN element I met. When there is a large amount of particle loss, it seems Pelegant has problems.

For the same setup runing with Elegant, everything works correctly.

thanks,
Yuantao

Re: CLEAN in Pelegant

Posted: 31 May 2011, 08:34
by ywang25
Yuntao,

Can you upload the wake files:
ZWAKEFILE="/afs/slac.stanford.edu/u/rl/ding/elegant/wakefiles/Sz_p05um_10mm.sdds", &
TRWAKEFILE="/afs/slac.stanford.edu/u/rl/ding/elegant/wakefiles/Sx_p05um_10mm.sdds", &
ZWAKEFILE="/afs/slac.stanford.edu/u/rl/ding/elegant/wakefiles/Sz_p05um_10mm.sdds", &
TRWAKEFILE="/afs/slac.stanford.edu/u/rl/ding/elegant/wakefiles/Sx_p05um_10mm.sdds", &

Thanks,

Yusong

Re: CLEAN in Pelegant

Posted: 31 May 2011, 08:53
by ywang25
Yuntao,

I realized you were using the AFS file system. Please be aware the AFS is not listed as a supported file system for parallel I/O, which is required for the latest version of Pelegant.

You can see previous discussions at:
viewtopic.php?f=18&t=161&p=583&hilit=afs#p583

Yusong

Re: CLEAN in Pelegant

Posted: 31 May 2011, 11:37
by ding
Hi Yusong,

You may try with this uploaded lte file, where I removed the wakes. If I also remove the LSC, then it works. It looks to me Wakes or LSC will cause trouble, and they are both related with Charge.

I realized I can not work Pelegant on AFS for writing purpose. So I work with NFS, and only these wake files are saved in AFS space for read only. Without the CLEAN of RCOL element it works fine in my NFS.
thanks,
Yuantao

Re: CLEAN in Pelegant

Posted: 01 Jun 2011, 13:15
by ywang25
Yuantao,

You can use the attached file to replace the file under the elegant directory for the version 24.0 to fix the charge issue. The problem is caused by the delay of updating the total particle number across all the processors for minimizing communication overhead. The incorrect charge happens only in the intermediate output files and it would not affect the correctness of the final output. Unless you use the intermediate output file as the input of the next step, it should not cause a problem. Let me know if you find other issues.

Yusong

Re: CLEAN in Pelegant

Posted: 01 Jun 2011, 13:37
by ding
Thank you Yusong. Do I need to re-compile the Pelegant after replacing the file?
Yuantao

Re: CLEAN in Pelegant

Posted: 01 Jun 2011, 13:43
by ywang25
You do need recompile the code. The fix I just provided should only affect the watch point output file. I will provide a better fix soon.

Yusong