RayUI RayX footprint differences

[julixH]

The output was generated with rayUI and rayX using the following .rml files:

• ellipsoid.rml
• mirror.rml
• grating.rml

Additional alignment errors were manually added (not included in the .rml files, since rayX does not recognize these):

"Ellipsoid": { 
    "translationYerror": 20
},
"Plane Mirror": { 
    "rotationXerror": 500,
    "rotationYerror": 500
}

Issue:

• For ellipsoid.rml and grating.rml, the output differs significantly between rayUI and rayX.
• For mirror.rml, the outputs are approximately similar, but only when reflectivityType = 0.
• Differences are present even without alignment errors, although they are less significant in that case.

This shows that rayx is not fully supporting the METRIX beamline yet.

Expected Behavior:

Consistent simulation results between rayUI and rayX, especially for identical input configurations.

rayXrayUI.zip

[Atraxus]

Edited for easier reading

[Atraxus]

[Atraxus]

@julixH, <param id="gratingEfficiency" enabled="T">0.5</param> set this to 1. This might fix the problem with the grating.

Also, can you please check the axis labels again? There seems to be a mismatch between the 2d histograms and the projected ones.

[treegardel]

Regarding the misalignment of the ellipsoid:
In RAY-UI there is a very special parameter for the element: “Coordinate System” under “Alignment Error”, which can be set either to “Ellipsoid” or to “Mirror (Standard)”. To be able to make a comparison with RAYX, you must set it to “Mirror (Standard)”. I only added the other option, “Ellipsoid”, to ensure backwards compatibility.

[julixH]

I replaced these values in the rml

Ellipsoid:
<param id="misalignmentCoordinateSystem" comment="Mirror (standard)" enabled="T">1</param>

Grating:
<param id="gratingEfficiency" enabled="T">1</param>

and for all elements:
<param id="reflectivityType" comment="Derived by Material" enabled="T">0</param>

Now I get good results for Slit, Cylinder and Mirror.

I set all translation/rotation erros = 0 and only use the parameters in the rml. I still get slightly different results with Ellipsoid and Grating but in both cases the difference very much depends on the run:

1. Ellipsoid
   

2. Grating:
   

[julixH]

I put all rml and output files (generated not using any translation/rotation errors) in a github project:

Here are all rml files:
beamlines

Here is the output:
output

The results seem similar, but as I said above, for grating and ellipsoid, there are sometimes small differences.

[julixH]

To analyze if each element of the METRIX_U41_G1_H1_318eV_PS_MLearn_v114.rml beamline behaves the same for rayUI and rayX, I copied the parameters of the elements in the metrix beamline into the beamlines with only one element (all parameters except the world position/orientation).

Here are all new rml files:
beamlines_metrix

Here is the output:
output_metrix

Here are the parameter differences from the beamlines above to the new beamlines:
parameter differences

I did not use any rotation/translation errors.
For almost all elements, the results are different... o.o

[treegardel]

How are the hist projections are calculated from the footprints in the comparison of the Ellipsoid and Grating? (#368 (comment)) In the case of "hist projection x" (which should called hist projection y, or?), the scale of the x-axes does not fit to the footprint. Each bin has a certain x-value and a certain width, are these definitions for both cases the same? Especially in the footprint of the ellipsoid, I cannot see the relatively large offset in the y-direction that can be seen in the projection.

[julixH]

Ah you are right, I used the wrong axis labels in the hist-y plot. I fixed that axis labels and updated the output.

[julixH]

For the metrix beamline elements, I guess the world coordinates and x/y/z direction to not match the other parameters. I updated the value in distancePreceding to worldPosition.z. But which parameters are responsible for the x/y/z direction?

[julixH]

I changed <param id="misalignmentCoordinateSystem" comment="Mirror (standard)" enabled="T">1</param> in the complete metrix beamline and now the results look the same for all sampled configurations specified in the paper Offset Finding of Beamline Parameters.
So I think this issue is not relevant anymore :)

[treegardel]

In the case of the grating there is still a difference in the maximal hight in the hist projection y. Looking at the raw footprint data it seems that in the case of RAYX there are few rays which are further away from zero (max y = -0.175) than in the case of RAY-UI (max y = -0.13). In case your bin size depends on this extrema and you choose in both cases the same number of bins, this could explain the height difference. As in the case of RAY-UI the bin width would be smaller.

[treegardel]

Last open question in this issue: is RAYX throwing away rays (using Monte Carlo) for tracing the reflection or is RAYX multiplying the reflectivity to the E-Field?