How should inconsitencies between laser centroid, focus and propagtaion directio be handled?

[PrometheusPi]

The laser definition requires the attribututes (all 3D vectors):

• propagation_direction
• focal_position
• centroid_position

To be consistent (the laser should propagate from centroid_position to focal_position or should have propagated from focal_position to centroid_position. Thus, the difference between focal_position and centroid_position should be parallel (or anti-parallel) to the propagation_direction.
According to https://picmi.readthedocs.io/en/latest/standard/laser_profiles.html#picmistandard.PICMI_GaussianLaser this does not seem to be a requirement.
Is there an undocumented consistency check, or do I misunderstand something here?

If not, how should the case of (focal_position - centroid_position) not being parallel to propagation_direction be handled.

cc'ing @steindev @BrianMarre @chillenzer

[chillenzer]

From my perspective (coming less from the physics), geometrically the propagation_direction simply seems to be redundant information that should not be exposed in the interface but derived in the backend -- precisely as you wrote -- as the vector pointing from centroid to focus.

[PrometheusPi]

In principle, it could also point from focus position to centroid position if you want to initialize an (already) defocusing laser. Thus, the redundancy of the propagation directions boils down to one bit (bool) of information and not a unit vector in 3D space.

[chillenzer]

Okay, I see. Yes, something like a bool defocussing or better an enum propagation = {focussing, defocussing} sounds reaonable for that.

Another issue is with lasers (like plane waves) that don't actually have a focus. One could still use that information to define the propagation direction but it would, of course, be quiet arbitrary.