Orbit Correction Accuracy

[fliller]

Mike,

I have a ring where I'm attempting to look at the tune shift vs. the accuracy of the orbit correction. I've attempted to vary the "correction_accuracy" parameter in the &correct command. I vary the accuracy from 1 to 1e-6 and I see no difference in orbit output. Is there anything eles I need to change? I've attached the files I'm using.

Ray

[michael_borland]

Ray,

I think the only problem is that your corrector and monitor arrangement is not ideal. If I remove the line that restricts the number of singular values, the condition number in the horizontal plane is very large. This indicates that the response of one or more correctors is very similar. You should try moving the correctors around.

--Michael

[fliller]

Michael,

I agree. We know that the corrector arraingement has a degeneracy in it. We've found that this helps keep the orbit deviations to a minimum. We can move one of the correctors to the opposite side of a quad to break the degeneracy but the orbit deviations are somewhat larger.

I'm confused though why this results in the "correction accuracy" parameter not working properly. Would I need to further restrict the number of simgular values??

Ray

[michael_borland]

Ray,

Sorry, I didn't understand your initial question. Let me try again.

The orbit correction loop in elegant is as follows:
while (n_xy_passes not exhausted) {
iterate x orbit n_iterations times
iterate y orbit n_iterations times
if desired accuracy achieved, then break out of loop
}

In your run, n_iterations=100, so elegant converges very well in the first pass through the loop. The accuracy parameter is always met.

--Michael

[fliller]

Michael,

That's just the bit that I needed! Seems I can get it to work O.K. now. Thanks.

Ray