many-turn tracking with time dependent modulations

[martin]

Dear all,

I intend to do many-turn ring-tracking with time dependent modulation of some elements. I run into this bug with numerical errors in the longitudinal position. The bug does not occur if I set change_t=1 in one of the cavities (RFCA). However I then have no time depenance and cannot use commands like modulate_elements.

Is there a way to cirmumvent this issue?

Or is it maybe possible to modulate elements in a turn-dependent way (such as in ramp_elements) but with arbitrary rpn expressions and/or input files?

Thanks in advance!

best,
martin

[petrenko]

Dear Martin,

I've recently made tracking simulations of bunch rotation (RF-gymnastics) in the SPS using Elegant.
The attached example might be helpful to you since I'm using &modulate_elements with external file here.

Best regards,
Alexey

----------------------------------
SPS (LHC injector) has two RF systems: the main one with 200 MHz cavities and another one with 800 MHz cavities. This tracking simulates RF voltage change in both cavities according to the program given in the RF_program.sdds file.

[michael_borland]

Martin,

It would help if you could post your input files, or send them to me via email.

--Michael

[martin]

Thanks for your replies. The problem I have occurs in an electron storage ring with synchroton radiation. The components are basically:

a ring with 1st order matrix elements,
a cavity (RFCA with change_t=0),
and sychroton radiation: sr: sreffects, damping=1, qexcitation=1, losses=1

The result is that for a single particle, the logitudinal ellipse (delta vs s) is beeing deformed significantly after 200k-1M truns, depending on the bunch length and other values. I also see that effect in an increase of sigma s (Ss in watch mode=parameter), it groths indefinitely for high turn numbers (>200k).

Both effects do not occur if i set change_t=1 in the cavity: Ss stays constant and the ellipse (delta vs s) also keeps it shape indefinitely, as it should be in an equilibrium of damping and excitation.

I will post an example file later.

Best,
Martin

[martin]

I attached the input files (it is a modification of the PAR/bunchLengthening example). And a plot of sigma_s of the tracking once with change_t set in the cavity and once without. It seems the black curve is the wrong one, is it not?

The thing is, I'd like to look at the bunch length while applying time/turn dependend modulations.

[michael_borland]

Martin,

The Ss parameter is often misleading as it doesn't mean what people think it means. It isn't equal to the rms bunch length, but rather the rms spread in effective distance traveled at the instantaneous velocity.
http://www.aps.anl.gov/Accelerator_Syst ... ode13.html

You should use the St parameter, which is the spread in arrival time at the observation point. In that case, I think you'll find the behavior is reasonable for both change_t=0 and 1.

Probably it would be best for everyone's sanity if I just removed Ss or simply redefine it as St*<beta>*c.

--Michael

[martin]

Dear Michael,

yes you are absolutly right! I misinterpreted the meaning of s. St is the quantity I wanted and it gives comparable results for both change_t=0 and 1. Thank you for your help!

Martin