Parabola focus shows strange effects

[em-archer]

I am having an issue with using lasy and axiprop (@hightower8083) to calculate known focal profiles for short focal lengths (but still well above the paraxial limit). In the figure attached, I show the result for three different focal length lenses: 400mm, 200mm and 100mm. The pulse shows weird effects on the side of the focus to the lens. These get worse if the resolution of the grid is decreased, but increasing the resolution of the grid doesn't make these effects disappear. I haven't really found any advice or guidelines about what grid specifications are required for axiprop, so I tried to use grids that make sense i.e. resolving the spatial profile without taking hours to run.

Bottom left of the figure shows a 'close to paraxial limit' calculation case where the calculation completely fails.

Bottom right plots what the focus looks like based on a correct calculation from the analytic theory and an alternative diffraction integral. As you can see, these effects are not present.

I am not sure whether these issues are due to the definition of the lens in lasy or due to the grids / method used in axiprop.

[MaxThevenet]

Note: this PR in Axiprop may help: hightower8083/axiprop#33

[hightower8083]

hi @em-archer

thank you reporting this issue! I'm not sure what is your case, but I assume that you use a PropagatorResampling to bring the field from the focusing optics to the far field using some custom new grid (r_axis_new argument in axiprop)? do you have a script (lasy or axiprop) that reproduces this problem?

I think, I have seen before a similar problem for the cases of short f0 and at that time i didn't find a good solution with the PropagatorResampling. This was solved by using the PropagatorResamplingFresnel which implements the diffraction integral, and intrinsically can handle such cases without unnecessarily over-resolving the nearfield. This propagator works well, but is a bit less straightforward to use with the arbitrary input-output grids, and afaik there are some plans to bring it to lasy but its not yet done.

I can help you to setup the case of interest if you provide the input.

[hightower8083]

UPD (@em-archer ):

I've checked a case similar to yours and can reproduce the issue with PropagatorResampling

which get fixed when using the PropagatorResamplingFresnel

Here is an axiprop script that shows the problem.

Note, that you can always brute-force the propagator using larger Nr, but ultimately it will bring you to the limit of Nr ~ Rmax/dr with dr being your final resolution (e.g. dr=Rmax_new/Nr_new~w0/8), and these are very large numbers.

I don't know what are the plans on integrating PropagatorResamplingFresnel (and its equivalent PropagatorFFT2Fresnel) in LASY, but I guess that will take some time. Maybe we can discuss this next meeting @MaxThevenet ?

[em-archer]

Hi @hightower8083,

Thanks so much for your reply and for clarifying that PropagatorResampling is causing the issue. In response to your original message, that is exactly what I was doing. I use the defaults in LASY's laser.propagate(), passing a second grid recalculated for the focal region (see original screenshot for details on the input / output grids), which causes PropagatorResampling to be called. Because I am calling this via another package written on top of LASY, I'd have had to write a MWE from scratch to circumvent this; so it's great that you can replicate what I see without this. Thanks for saving me some time and for providing the direct comparison of PropagatorResampling and PropagatorResamplingFresnel.

I am now using one of my own diffraction codes to propagate the fields outside LASY, so there is no rush on my side. Happy to help where I can with bringing these propagators into LASY. I am not sure whether we can flag this as a known bug in the interim?