Vkick acts in the horizontal plane

[Laurie Nevay]

Vkick and hkick both kick in the horizontal plane. However, in debug mode, the global field is specified correctly in each case (along z in hkick and along x in vkick). It would seem this is ignored by the stepper. Perhaps to do with the stepper using an internal parameter rather than the field value in combination with the rotated placement.

In any case, this is why LHC tracking is currently wrong as IP1 (launch point) uses a vertical bump that is correct not long into the lattice.

[Laurie Nevay]

fix issue #101 - hkick now bends in vertical - was too complicated to fix in v0.9-rc. Ultimately should edit BDSDipoleStepper which only works in the horizonal locally just now.

→ <<cset e2af20831ffa>>

[Laurie Nevay]

BDSDipoleStepper ignores the field value supplied and uses it's own internal parameter for the field strength. This doesn't really present any advantage as vector cross products are still used (using a unit y vector) and then appropriately normalised. The equations used fix its usage to a specific dimension. H and V kickers are unique in that they define a uniform dipole field that's not aligned to local unit y - aligned to local unit x in the case of the vkicker.

Recently, I changed BDSKicker to simply create a different vector B Field rather than rotate the beam pipe volume, which result in rotated aperture (wrong), or an inconsistent aperture parameter meaning (not preferred). As a consequence of this, the field definition was correct in global coordinates, but BDSDipoleStepper ignored this, resulting in always bending in the horizontal plane.

BDSDipoleStepper should be edited with caution as it has large implications. Adapting it for any vector field is non-trivial and therefore the simpler fix for this bug is to use another Geant4 stepper for H and V Kickers that will use the correctly specified global field. For this, I've used G4HelixExplicitEuler, which is recommended for constant and uniform fields.