Updates to HDF5 metadata

[jsoishi]

• Fixes issue HDF5 dims labels are incorrect for vector quantities #190
• addes scale names to scale objects as well as tasks
• addes coordinate scales coords for tensor component dimensions populated from coordinate name strings from the CoordinateSystem object

[kburns]

Looks good. Only issue I can think of is that this could be misleading if we save vectors or tensors in coeff space, since there the coefficients may correspond to spin/regularity components, not the coordinate components like in grid space.

[jsoishi]

That's a good point. We could do something like s_1/2, s_-1/2, r_0, r_1, etc in that case, basis dependent?

[kburns]

Yeah or maybe just {-1, 0, +1} or even {-, 0, +}, and you have to know its spin or regularity based on the context? Any thoughts @lecoanet?

[lecoanet]

Our ordering is {-, +, 0}. Using {-, +, 0} for both spin and regularity bases would be fine. But {s-, s+, s0} and {r-, r+, r0} might be a bit more descriptive. Don't think we ever have spin 1/2.

[jsoishi]

Too much QM teaching...

…

On Tue, Apr 26, 2022, 23:26 lecoanet ***@***.***> wrote: Our ordering is {-, +, 0}. Using {-, +, 0} for both spin and regularity bases would be fine. But {s-, s+, s0} and {r-, r+, r0} might be a bit more descriptive. Don't think we ever have spin 1/2. — Reply to this email directly, view it on GitHub <#192 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA7SHYNBTV3XPBPQXNH52RDVHCXXVANCNFSM5UMDX5TQ> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[kburns]

@jsoishi want to push this to a separate branch on the main repo?

[kburns]

Marking as a draft since we're still WIP.