Orbit response matrix for BEND.FSE

Moderators: michael_borland, soliday

Post Reply
petrenko
Posts: 43
Joined: 09 Jun 2008, 02:53
Contact:

Orbit response matrix for BEND.FSE

Post by petrenko » 13 Apr 2009, 09:40

Dear Michael,

I think there is a bug in the correction_matrix_output command then BEND.FSE is used as steering element. Closed_orbit command gives a different result.

Please take a look at this example script.

I would like to use BEND.FSE in a response matrix fitting. Is it supposed to be used that way? We have a horizontal correction as additional coil in 45-degree dipoles.

Thanks,
Alexey.
Attachments
hrm.png
hrm.png (10.4 KiB) Viewed 5656 times
PAR.tar.gz
(23.24 KiB) Downloaded 693 times

michael_borland
Posts: 1927
Joined: 19 May 2008, 09:33
Location: Argonne National Laboratory
Contact:

Re: Orbit response matrix for BEND.FSE

Post by michael_borland » 13 Apr 2009, 11:54

Alexey,

This is an interesting result. I think the problem is that for orbit response matrix computation, elegant assumes that the kick delivered by the element occurs at the end of the element. This isn't exactly right, but is usually "good enough."

I'll have to think about how to solve this. There isn't a quick fix or workaround that occurs to me.

The only thing I can suggest for now is to compute the response matrix outside of elegant based on difference orbits.

By the way, this is what is done for the APS model, but for other reasons. It was found that the response matrix from elegant can differ from the difference orbit from an experiment because in the latter case we have significant orbit variation in sextupoles.

--Michael

petrenko
Posts: 43
Joined: 09 Jun 2008, 02:53
Contact:

Re: Orbit response matrix for BEND.FSE

Post by petrenko » 14 Apr 2009, 01:46

I guess the effect of corrector length should be important if there is some significant betatron phase advance along that steering element.

I've found a simple workaround: the effect of FSE in a bend can be replaced with two HKICKs at the beginning and at the end of the magnet. For a coasting beam (fixed_length=0) the resulting closed orbit will differ from the exact solution only inside the bend -- that will actually produce some small path length error and for a bunched beam in a small ring can be taken into account as well.

Here is the expression for the required HKICK values:
HKICK1 = HKICK2 = -(FSE/(R*sqrt(K)))*( 1-cos(L*sqrt(K)) )/sin(L*sqrt(K)),
where K = K1 + 1/R^2, R -- bending radius, L-- bend length, K1 -- quadrupole component.
Attachments
APS.tar.gz
1 KICK vs 2 KICK approximation for BEND.FSE in the APS
(568.61 KiB) Downloaded 615 times
Screenshot-MPL Outboard Driver-150.png
1 KICK vs 2 KICK approximation for BEND.FSE in the APS

Post Reply