Search of closed orbit in a ring with BGGEXP elements

[Nagaoka]

Hello,

I am investigating the closed orbit distortion possibly created by dipoles which are all represented by BGGEXP elements obtained from numerical field data fitted by the GGE method available in elegant.

I have used the following command block for the closed orbit search:

&closed_orbit
output="%s.clo",
output_monitors_only = 0;
closed_orbit_accuracy = 1e-9,
closed_orbit_iterations = 500,
iteration_fraction = 0.1
fixed_length = 1
tracking_turns = 1
immediate = 1
&end

However, I get the following error message in running it;

error: closed orbit did not converge to better than 4.494233e+307 after 500 iterations (requirement is 1.000000e-07)
Trying secondary, tracking-based method for orbit determination (1 turn).
error: closed orbit did not converge to better than 4.494233e+307 after 500 iterations (requirement is 1.000000e-07)
statistics: ET: 02:03:40.663 CP: 154.40 BIO:0 DIO:0 PF:0 MEM:201735

Despite this message, the result that I see in "%s.clo" appears all normal with the found trajectory well fulfilling the periodicity, as well as showing good convergence in the closed orbit search (please see the attachment).

I would therefore like to know whether I can trust the result obtained and whether you see why I get this error message.

Thanking you in advance for your kind help,
R. Nagaoka

[michael_borland]

Ryutaro,

I would be suspicious of this result. First, the convergence warning indicates that the normal algorithm had a problem. Second, you have tracking_turns=1, which asks elegant to use a fall-back multi-turn tracking method; however, you only asked it to use 1 turn, which isn't reliable.

It sometimes helps to play with the iteration parameters for &closed_orbit, e.g., try

Code: Select all

iteration_fraction = 0.5
closed_orbit_iterations = 500

I would suspect that your tunes are close to an integer, which can happen when assembling a lattice from GGE elements, since they never match the usual hard-edge elements precisely. To avoid this, we typically tune the STRENGTH parameter on the GGE elements to match the linear matrices of the hard-edge elements, one by one.

--Michael

[Nagaoka]

Dear Michael,

I thank you very much for your helpful feedback despite your supposedly very high workload during this period!
I shall carefully follow your suggestions and see what I shall get.

Thanking you again and my apologies for my late feedback to your message,
Ryutaro