Improvements of the laser envelope extension

[MaxThevenet]

This PR proposes the following improvements to the EXT_LaserEnvelope extension:

• isLaserEnvelope is a required attribute, rather than optional.
• The standard supports both z and t representations for the longitudinal axis. Attributes isSpatial contains this info.
• The standard supports both electric field (V/m) and vector potential a0. Attributes isNormalized contains this info.

While the first point is minor and the second point has been discussed in the context of LASY, the last point is relatively new. Supporting the normalized vector potential (or at least the SI vector potential) can be important as this is what appears in the ponderomotive force, and what is implemented in envelope solvers. As the conversion from field to vector potential can be non-trivial in a simulation code (it required the knowledge of the local frequency, which comes from a derivative of the phase, and can be cumbersome to implement in a robust way), I believe it is useful for the standard to accept both. An alternative approach would be to add, as an attribute, another array containing the local frequency, so the reconstruction in simulation codes is made easier.

[ax3l]

This sounds great. My only thought is, do we want to have a non-binary option instead of isNormalized? I think that normalizing by:

• peak V/m
• peak a0
• total energy
• V/m @fwhm
• a0 @ FWHM
• ...

are other options often used in experimental measurements. We could stay extensible if we only standardize a few strings for now: normalization=

• "peak_a0"
• "peak_Vm"
• "energy_J"
• "fwhm_a0"
• "fwhm_Vm"
• ...

[RemiLehe]

Hm, I think that the name isNormalized might be confusing: it maybe suggests that this just changes the field by a normalizing constant, whereas that is not the case.
The fields that are stored with isNormalized = false and isNormalized = true are not simply proportional to each other - since in one case the field is the electric field and in the other case it is the vector potential.

Maybe we could have something more explicit like, envelopeField = 'electric_field' or envelopeField = 'normalized_vector_potential'. (Btw, having this single metadata could replace both isEnvelope and isNormalized.)

[MaxThevenet]

@ax3l Thanks, I think that, as Remi pointed out, the name isNormalized was confusing: vector potential and electric field are two different things, as seen on the figure below (from a realistic FBPIC simulation, these are temporal profiles, sorry there's no axes). I think we can start with just the 2 options, and maybe add some others later, but maybe it's good to be quite restrictive here, to make the standard lightweight?

@RemiLehe Thanks, this is better indeed. So to check if an openPMD file complies with the EXT_LaserEnvelope, one would check if envelopeField is present, and which of the 2 possible values it has? I'll make the change, and we can have another round before merging.

[MaxThevenet]

I made the changes :)

[ax3l]

This is great, thank you! :)