Reimplementation of ParallelCyclotronTracker

We should re-implement the tracker such that it is independent of the integrator in order to avoid duplicated code. Adding new integrators is then also simplified.

created branch 124-reimplementation-of-parallelcyclotrontracker

I introduced a new subdirectory Steppers that contains all integrators, like RK-4, LF-2 etc. The MTS part has still to be updated. But here is already a comparison between the previous and new implementation using the RingCyclotron regression test.

Attention: This test uses just RK-4 ... i.e. I have to test also all other steppers.

I finished my implementation for the moment. The ParallelCyclotronTracker is still a hack but I tried to remove duplicated code. I ran subsequent tests to verify my changes:

• RestartTest-2

• RingCyclotron

• RingCyclotronMTS

• MultiBunch.tar

• MultiBunchMTS.tar

• RingCyclotronTunes.tar

• RingCyclotronSingleParticle

I noticed that the bunch position is wrong for a LF-2 simulation (see plot below). This is now fixed in the new version. In case of a multi-bunch simulation using MTS it's still wrong. Furthermore, the multi-bunch simulation using MTS of the master branch crashes.

I have no explanation for the discrepancies between the old and new LF-2 interface (see plots below). All plots are in Results-ParallelCyclotronTracker.tar.

@adelmann If you agree, I merge with the master. However, I do not close this issue since there's more work to be done.

@frey_m yes go on and merge

mentioned in issue #110 (closed)

mentioned in issue regression-tests#57 (closed)

mentioned in issue #227

mentioned in issue #277 (closed)

added OPAL-cycl label

mentioned in merge request !129 (merged)

@frey_m can we close?

Since it is part of the list #227, we can close. It can be reopened if necessary.

closed