Thomas Kittelmann

Just my 2 cents concerning the MPI issue: I would think it is best is if each process simply uses the mcpl skip ahead functionality to pick out a subset...

I think the cell volume and masses are both OK (after all, 4 aluminium atoms should have a mass of about 4*27=108g/mole), but you are leaving out a factor of...

It indeed seems related to the kind of per-cell versus per-atom issues we noticed today...

I can add to what @marquezj wrote, that since last week NCrystal is also available on conda-forge (since @paulromano suggested this during our meeting).

@paulromano I added comments concerning the two non-obvious fixmes in the PR. Let us know if you have any specific wishes concerning this (or anything else), and we will try...

So @paulromano, unless I overlooked something I believe @marquezj took care of all the changes requested so far (including applying clang-format).

FYI we haven't forgotten this, but some other work came in the way. We hope to spend some time on this on Friday.

I just pushed a few updates, mainly to use a new py API method I have just added in NCrystal 3.4.1, which helps getting the basic material composition in Python...

Outstanding issues are tests, documentation and CI as requested. We will have a look at those tomorrow.

@paulromano I added a `tools/ci/gha-install-ncrystal.sh` script now (can most likely be improved a bit in the future). Should we also update the `gha-install.sh` and `gha-install.py` files? Also, will it be...