DRIFT vs CSRDRIFT vs EDRIFT vs LSCDRIFT

Moderators: cyao, michael_borland

Post Reply
frank_stulle
Posts: 13
Joined: 04 Jun 2009, 04:58

DRIFT vs CSRDRIFT vs EDRIFT vs LSCDRIFT

Post by frank_stulle » 19 Jan 2010, 04:10

Hello,
besides LSC in the LSCDRIFT and CSR in the CSRDRIFT, what are the differences between these drifts?

Best regards
Frank

michael_borland
Posts: 1933
Joined: 19 May 2008, 09:33
Location: Argonne National Laboratory
Contact:

Re: DRIFT vs CSRDRIFT vs EDRIFT vs LSCDRIFT

Post by michael_borland » 19 Jan 2010, 08:05

Frank,

The DRIFT element type is a matrix implementation, up to second order. (The second order terms are the path-length dependence on slope.)

EDRIFT is an Exact DRIFT. Instead of using a matrix it just computes the exact path length effect. On modern hardware, square roots are very fast so I'd guess there is little performance reduction from this. CSRDRIFT and LSCDRIFT, in addition to including CSR and LSC, also perform exact drifting.

--Michael

frank_stulle
Posts: 13
Joined: 04 Jun 2009, 04:58

Re: DRIFT vs CSRDRIFT vs EDRIFT vs LSCDRIFT

Post by frank_stulle » 19 Jan 2010, 08:19

Thanks, that's already good to know. But now I am in trouble, because you say EDRIFT, CSRDRIFT and LSCDRIFT all use exact drifting. I have a lattice with long drifts and arcs. The difference between using DRIFT and CSRDRIFT is negligible. But when I use EDRIFT I get some strange emittance growth in the following arcs. I will prepare some test case which shows this behavior.

Frank

frank_stulle
Posts: 13
Joined: 04 Jun 2009, 04:58

Re: DRIFT vs CSRDRIFT vs EDRIFT vs LSCDRIFT

Post by frank_stulle » 21 Jan 2010, 06:28

Well... I just made a stupid mistake in my input files and then made a wrong interpretation of the cause of the strange output I got. So, forget what I wrote, everything is fine.

Frank

michael_borland
Posts: 1933
Joined: 19 May 2008, 09:33
Location: Argonne National Laboratory
Contact:

Re: DRIFT vs CSRDRIFT vs EDRIFT vs LSCDRIFT

Post by michael_borland » 21 Jan 2010, 07:58

Frank,

No problem. It is always good to test any program and I'm happy to hear it was alright in the end.

--Michael

Post Reply