issue with reference_correction and epitch/eyaw misalignments

[GPenn]

Hi,

I'm seeing a large discrepancy between tracking with and without REFERENCE_CORRECTION in CSBEND elements, when there is also an epitch or eyaw misalignment. The case with REFERENCE_CORRECTION=0 seems like a closer fit to AT calculations, and the final offset is much smaller which seems more reasonable. For a typical bend magnet with epitch=1e-5, there is about a 1 micron difference in final position of the particle (starting on axis). This is with elegant 2021.2.0.

A minimal set of example files is attached. The same size discrepancy appears if I set eyaw instead of epitch, and it also doesn't matter if I use malign_method = 1 or 2. If there is just an etilt, then the difference in tracking results is much smaller.

Thanks,
Gregg

[michael_borland]

Gregg,

Thanks for reporting the bug. The attached file should fix it.

--Michael

[GPenn]

Thank you, that did the trick!

-Gregg