Review McStasToX

[tkittel]

Review of McStasToX: https://github.com/mccode-dev/McStasToX

This is an early version of the package, still without documentation, testing and CI. The main purpose of this review is to ensure the package has a healthy structure so that it can be:

• Easily expanded to more functionality (writing nexus files, exporting to other formats, read more geometries)
• Easily maintained

It is foreseen that the code surrounding the package for documentation, CI and so forth will follow the scipp template which can be applied retroactively. This is partly to ensure easy collaboration with the scipp team.

The module is currently two files, so it is natural to split the review in two as well:

• NeXus reading in ReadNeXus
• Interface and transformation in LoadFile

The demo notebooks show examples of how to use the code.

[damianmr76]

Found some strange results in the demo provided in McStasToX. While the original simulation with 10^6 works well and gives expected results. If done with 10^8 trajectories, inconsistent results start to appear. The figure where the diffraction patterns comparison between a single detector bank and all the data doesn't give the correct outcome. If done with 2*10^8 trajectories, then an index error appears.

[willend]

@damianmr76 please include some files and screenshots to allow reproducing your findings. :-)

[damianmr76]

McStasToX.zip
Here are the notebooks I have used. The only change is the number of the initial trajectories in the simulations

[willend]

And version of McStas + McStasScript are "latest conda", platform is VISA?

[damianmr76]

McStas 3.5.24 with latest conda in Ubuntu 22.04 (from Linux subsystem in Windows 10). It's also reproducible in VISA.

[damianmr76]

Couldn't reproduce it in Windows as I haven't been able to install McStas in Windows through conda