Attributes PC, ENERGY and GAMMA of BEAM command not described

There are several possible definitions for PC

1. $PC = mc^2 \frac{\sqrt{\sum_i p_{x,i}^2 + p_{y,i}^2 + p_{z,i}^2}}{N}$
2. $PC = mc^2 \frac{\sum_i p_{z,i}}{N}$

Currently the second definition is implemented. This has the advantage that mean longitudinal momentum is well defined and thereby the reference particle has the same velocity as the centroid. The disadvantage is that the kinematic energy of the bunch is to high. The definition for PC should be chosen and the two attributes should be described in the manual for 1.6 and 1.9.

changed the description

assigned to @snuverink_j

@snuverink_j thanks for having a look on that, I am a bit overloaded at the moment

The (documented) source code says Particle momentum in GeV/c.

Then this sets the reference (Partdata) as follows (energy scale = 1):

double pc = Attributes::getReal(itsAttr[PC]) * energy_scale;
reference.setP(pc);

This does in PartData:

double e = sqrt(p * p + mass * mass);
beta = p / e;
gamma = e / mass;

So everywhere in Beam / PartData it is assumed that PC is the full particle momentum, not the longitudinal one. From this it seems to me that the Distribution class should use the first definition.

On the other hand of course we can change it, or introduce a new attribute if this is convenient.

Note that the Beam attributes ENERGY and GAMMA also assume the full energy / gamma in the same way as PC, and not just the longitudinal part. This is in contrast to what the Distribution class does with it and seems counter-intuitive.

changed title from Attributes PC and ENERGY of BEAM command not described to Attributes PC, ENERGY and GAMMA of BEAM command not described

I have implemented a correction but not pushed yet. Have to think how to include OFFSETPX, OFFSETPY and OFFSETPZ. Do they increase/decrease the energy? If not, does OFFSETPZ make sense? Or should we have OFFSETEKIN or similar? How does a user specify a bunch that has the nominal energy but a momentum that isn't tangential to the reference trajectory?

mentioned in issue #184 (closed)

I have documented the undocumented attributes in the Wiki manual: https://gitlab.psi.ch/OPAL/src/wikis/Manual/beam-command.

mentioned in commit b0eb1f8e

I've added the attribute OFFSETP to add a momentum shift relative to the nominal momentum. OFFSETPZ is imho redundant since the longintudinal component of the momentum is fixed with PC, OFFSETPX and OFFSETPY.

I think we need to make sure this is 100% documented. What about compatibility to other codes: astra, elegant etc ? Would it maybe wise to discuss this on the mailing list ?

My 5 cents: the distribution should not change the total energy of the beam. Because of symmetry in generating the distribution, as far as I remember, this should not happen.

I totally agree, the number of attributes in the distribution command are way too many.

TBC

@adelmann What did you mean with "Because of symmetry in generating the distribution, as far as I remember, this should not happen."?

When we create the pz distribution, the we create a symmetric distribution, around the mean.

This is true but if rms_px > 0 or rms_py > 0 then mean_pz != mean_p. So far we computed mean_pz from Ekin which in most cases yields wrong results.

added Remarks and removed ~78 labels

How to progress on this? This is mostly relevant for low energy.

In Elegant it seems one can only define P0 : Central momentum of the bunch. Not sure about the actual implementation. Elegant also has an ENERGY element, which can define central energy and P0.

I've implemented and pushed (sometime end of November) a correction that ensures that the mean momentum (not mean Pz) is equal to P0. For me this is fixed and the issue can be closed.

ok thanks, for me it is also concluded: the chosen solution is consistent and documented.

@adelmann : shall we close? Or would you like to keep this open and discuss it further at some point?

closed