Trajectory correction issue

[Sjostrom]

Hello Michael,

I've been running trajectory correction simulations with Elegant v24.1Beta1 on Ubuntu 11.10 (I compiled it from the available source this summer). The trajectory correction algorithm converges nicely after applying the error model and I get realistic kick strengths and trajectory deviations. This behaviour was reproduced by another colleague using Linux but installing from the available binaries instead. However, sending the same input files to several colleagues running Elegant v24.0.1 on Windows (installed via the available binaries) the simulation diverges and results in an extremely long bunch length. At this point the simulation fails due to the bunch being longer than the defined linac wake fields.

As the only change made between the versions was altering all '/' to '\' for the file name references in the max4.ele file, I am somewhat confused at the result: I didn't see anything regarding trajectory correction listed under known bugs. Did something change in Elegant's trajectory correction routines between the two versions?

Regards,
/Magnus Sjöström


PS
Note that the file daresburyastrain.sdds, which contains the beam input data, is missing in the archive. It is about 4 MB, which is the reason I could not add it. I can mail it if needed.

[michael_borland]

Magnus,

I ran your files with elegant 24.1, 24.0.1 and 23.1.1 on linux, with no differences. It is possible there's an unidentified bug that affects the windows version. I'll have a colleague try your inputs with version 24.1 on Windows, to see if the problem persists in the latest version.

--Michael

[michael_borland]

Magnus,

We've confirmed the problem with the latest version on elegant on Windows and are looking for the cause.

Meanwhile, you might try explicitly setting the number of singular values allowed in the correction and/or reducing the number of correctors (set STEERING=0 on some correctors until there are more BPMs than correctors).

--Michael

[Sjostrom]

Hi Michael,

thank you for the tips as well as the rapid response. I'll forward the former to my windows-using colleagues; hopefully it will allow them to run the simulations for the time being.

/Magnus