Beam lost in a single turn. Why does it happens? Is it just a numerical error?

[Gyeongsu]

Many studies shows that the dynamic aperture with RF reduces the dynamic aperture area compared to the dynamic aperture without RF. I tried to find out why does it happens. I tracked the particles with elegant. First I put the watch element at some place and plot the x or y along the # of turn. I realized that the particle suddenly lost in a single turn. For example, the particle ID 1319 has normal amplitude at 141th turn and it is lost at 142th turn. Therefore, I wanted to know what happens to the lost particle in a single turn.

1. I want to know how to easily record the trajectory of the lost particle in a single turn. I don't know the easy way. So I just put many watch elements in the lattice and combined them to get the information of the tracking particles along s.

2. As far as my simulation, the amplitude of the lost particle grows really fast. Is this a numerical error? or the real physics? If it really happens why?

3. Is it possible to get a tune from a single turn trajectory? If there is many watch elements in a lattice, I can get many points of (x,s) in a single turn. With this data, running sddsfft or sddsnaff seems not appropriate, because the amplitude of the x is proportional to the root of the betax, not uniform. Is there any other way to get a turn by turn tune information?

I attached my simulation files. Your answer or any advice will be really helpful. Thanks.

[michael_borland]

Using WATCH elements is a very resource-intensive way to do this. A simpler way is to track one particle at a time and record the centroid data. Using the wrap_around=0 setting requests that the centroid information vs s gets "unwrapped" so you can see all the turns. You can then plot (Cx, Cxp) or (s, Cx) to see what's happening in more detail. What I see is that the particle coordinates don't go instantaneously to large values, but build up to it. It does happen pretty quickly though. I think we need to recall that when amplitudes start to get large, the kicks from sextupoles (which go like x^2-y^2 in the horizontal) grow quickly.

The attached script shows how to do what I described, including getting the tunes for each turn.

--Michael

[Gyeongsu]

Thank you for your reply.

I think the centroid motion with the lost particle's initial condition can't ensure the same tracking results.

I revised some of your script and tried to check the single particle motion in all position s. In that case, the recorded data of the watch elements and centroid motion have totally different results. As far as I know, the ".lost" file records the data(s, Pass, particleID, x, xp, y, yp, etc.) of the lost moment. I attatched the script that compares the results of the tracking data. One is watch element data and the other is centroid data. In the ".lost" file, particleID 516 are lost at Pass 33. The data of the watch elements show the same results. However, the centroid motion by tracking the single particle with the same initial condition gives the different results. In this case, the particleID 516 is lost at Pass 27 and the data(s, x, xp, y, yp, etc.) are also different with the ".lost" file.
All output file of the watch elements are heavy. Therefore I didn't attached this files.

Can I believe the results of centroid motion tracking results?
If I can't, the only way to check a single turn trajectory of the lost particle is using lots of watch elements?