propagating "multi_electron_waves" ?

[marcocamma]

Dear developers,

I have used "multi_electron_stack" method of the undulator class to get Electric fields for different electrons in the electron bunch. Watching how the different electrons build up the total intensity is actually quite instructive. I have also used the functions provided by the coherent module to measure the coherent lengths and so on.

Now I would like to do the same "measurements" after propagating through some optics (slits, mirrors, etc.). Is there a way I can do so with XRT ? I have tried to play a bit but failed likely because I have yet understood how XRT does wave propagation.

Any help to get me started would be appreciated

[kklmn]

Technically, it is possible to propagate multi_electron_stack fields. We would need to loop over all electrons and for each electron create a Beam object with the given field distribution from a slice of the stack and give it starting points and directional cosines for each pixel of the stacked image. It would take a handful of lines. Then we'd propagate that Beam in a usual way. But I don't see a reason, there are better mechanisms.

Firstly, I don't like a gridded version of field sampling because the positional periodicity may introduce an unwanted grating effect. Secondly, with a grid, you'd need to check the convergence with respect to the grid steps in the two directions, and this is a problem in itself. I view the Monte Carlo sampling as a big advantage compared to the gridded sampling of SRW. In the MC sampling, you only care about the total number of samples, which you can assess by examining a dark filed area somewhere on the periphery.

Please look at the examples of 14_SoftiMAX for wave propagation. For any plot, you can specify fluxKind='EsPCA' (this one is for the s polarization) that will create a stack array yourPlot.field3D of the shape (repeats, xaxis.bins, yaxis.bins), and this is what you want, I think. Please confirm.

Good luck with it!

[marcocamma]

Thanks a lot for your suggestions and suggestions. I understand your concerns but I will give it a shot anyway.

I guess that what you are suggesting is to use multi_electron_stack to get the fields. for each "electron" then one would:

• copy the Es to the beam.Es
• fix the beam.(a,b,c) for all pixels to the electron "pointing"
• propagate the beam as shown in the examples

The only doubt I have is how to get the electron "pointing". I don't think that multi_electron_stack can return it. Is there a smart way or should I create my own "multi_electron_stack" to return this information ?

I will try in the next days when I will find some time and I will report back. thanks

[kklmn]

The x,y,z will be the global coordinates of the stack pixels. The a,b,c are much less important. They can only slightly affect the calculated Kirchhof integral. I studied this a long time ago and concluded that in most cases the directions can be taken as paraxial: a=c=0 and b=1. In the propagation sequence, we assign the directions as being originated from the center of the previous optical element, in your case -- the source, independently of the actual trajectory of a given electron. I've never seen an example where this rule may break. You can also experiment with the a,b,c values by giving the origin a large transverse kick, much bigger than the electron beam size. I expect that you will see no effect of it.

[marcocamma]

Hi again, I finally found some time to get started. I run in some issues though at the preliminary step of running normal "Monte Carlo" propagation.

This script (https://gist.github.com/marcocamma/76053e9c516cef1b9a7c03b76add92b7) calculates the diffraction from a slit at different distances.

The problem is that I get data only every second row and column (see attached screenshot below). The script is usually run with --nrays 1e5 or --nrays 1e6. But the problem appears for all nrays.

It can be seen in:

• the plots
• the data saved by the plot instances
• the "total4D" array I save by hand.

Here the job output in case it can help job_output.txt, Do you have any idea what can be the reason causing it ? thanks a lot as usual, marco

[kklmn]

Hi! Make sure rw.diffract() is really calculated for the screen pixels. In other words, when you do screen.prepare_wave(), check that x and z coordinates are those expected.

[marcocamma]

I tried to verify what you have asked (see code below) and found that Es and x attributes of the screens look fine. No 'zero' values are present. The x and y values are what I would have expected. I used the very same machine I used when having problems. It seems that the problem comes from the plot instance. I will keep trying to understand, any help is appreciated, marco

import simple_diffract
bl = simple_diffract.build_beamline()
d = simple_diffract.run_process(bl)
b=d["screen_at_100m"].Es
x=d["screen_at_100m"].x

[kklmn]

I don't understand your answer.

Lines 132-133:

        xm, zm, _ = prepare_mesh(bl,screen)
        wave = screen.prepare_wave(bl.slit1, xm, zm)

make shure xm and zm are really the screen pixels coordinates you want.

[marcocamma]

Problem solved. There was an inconsistency between the screen and the plot instances (same number of bins but different ranges). Sorry for having bothered you.