Release v0.7

[jank324]

Checklist for v0.7 release features

• Vectorisation (Vectorised simulations #116)
• Add space charge (Add space-charge to Cheetah #142)
• Add nice interface for space charge (delayed to later release) (Nice interface for space charge #160)
• Add custom transfer map element (Batched execution #100)
• Fix issue with default length (Vectorised simulations #116, [Vectorised branch] Tracking crashes with single Marker, or only elements without lengths between non skippable elemtents #143, length cannot be a NN parameter #164)
• Add uniformly distributed beam initialisation (Add particle generation for 3D uniform ellispoid distribution #146)
• Fix NumPy 2 compatibility (NumPy v2.0.0 compatibility #177)

Let's discuss what features we want to have ready before releasing v0.7.

[jank324]

One thought, now that we are a couple more people contributing to Cheetah, we should use this release to establish common ground on certain foundational design choices, e.g. the coordinate system. So if there is something like that you thing needs to be discussed, please bring it up here.

[cr-xu]

From the discussion before, the current plan is to

• Fix remaining issues, review the code and merge Add space-charge to Cheetah #142
• Adapt the changes and make sure SC implementation is still correct in Change Cheetah particle coordinate to Canonical variables #163, then merge it into master
• Implement Nice interface for space charge #160

[jank324]

I have some proposals on revising the plan for v0.7.0, considering that the version of Cheetah on master is by now very different from the most recent stable release you would get from PyPI ( v0.6.3 without vectorisation and such).

• Add space-charge to Cheetah #142 is merged.
• Change Cheetah particle coordinate to Canonical variables #163 should definitely be merged before, being a bit of a breaking change and almost done.
• I would suggest to defer Nice interface for space charge #160 to v0.7.1
• I think Enable automatic broadcasting #208 I think is an important item we are working on right now, so it would be nice to have this in v0.7.0. But, considering that it will probably take some time, it might also make sense to make that part of v0.7.1.
• add BmadQuadrupole element #153 is almost ready, but it really doesn't matter if this appears in v0.7.0 or v0.7.1.
• I'm not sure where we stand with length cannot be a NN parameter #164.
• Fix twiss plot #213 is a bug fix for the vectorisation, so it should be merged before v0.7.0 ... but it's ready, so that should be easy.
• MAD-X integration #197 and Add a new combined Corrector element #207 can become part of a later release.

So I guess we could release v0.7.0 as soon as #213 is merged and #163 is finalised and merged. The only open question is if we would want to wait for #208.

[cr-xu]

I believe #164 is already fixed with #209.

[jank324]

#227 will have to be fixed before release as well.

[jank324]

#241 also needs fixing before this release!

[jank324]

#246 also goes on the list.

[Hespe]

#250 should probably also be addressed before the release.

[jank324]

List of things that need doing before we can finish this release:

• Enable automatic broadcasting #208
• Fix twiss plot #213
• Splitting elements is incompatible with vectorization #227
• Vectorised aperture tracking is broken #241
• Broadcasting of rectangular bending magnets fails #246
• Lattice pruning broken by vectorization #250
• Update documentation for v0.7 #256
• Update LatticeJSON version tag #257
• Broadcasting error in corrector elements #264
• Vectorized tracking for TransverseDeflectingCavity #270
• Error in Elegant import #274
• plot_reference_particle_traces crashes for non-leaf tensors #285
• Plotting routines fail for segments that contain tensors with require_grad=True #287
• Tracking fails with TransverseDeflectingCavity if embedded in a segment #290
• Add note to changelog that versions are now permanently archived on Z… #295
• nn.Parameter provided at initialization does not appear in Element.parameters #302

[cr-xu]

I really think #208 should be merged asap as it is the most breaking change, otherwise a lot of the PRs still need to adapt to it afterwards.

[roussel-ryan]

Need to address the following before 0.7.0, sorry I don't know how to do the pretty formatting

• Distribution plotting #292
• ParticleBeam.__getitem__ method #291

[jank324]

I would also like to sort the following out before release:

• Update acknowledgements #267