Peter Willendrup

And you were right this updates the situation to: * kdtree does not function on GPU - less of a problem 🟡 * regular provides the same output on CPU...

One of the typical things that _could_ give this type of difference is if some variable initially in the `COMPONENT` scope (i.e. the Table object) gets modified during `TRACE` -...

Actual a VERY obvious candidate is the "R_SWAP"... macro. I am afraid every neutron will either need its own copy of the interpolator - or a function with one or...

I think my choice will in fact be to effectively disable "kdtree" on GPU. Exit with a warning if anything other than "default" / "regular" is chosen. The other stuff...

Heh, interestingly the size of flipfield.dat is not what it claims to be... ``` wc -l flipfield.dat 151 flipfield.dat (mcstas-3.x-dev/miniconda3) data $ head -2 flipfield.dat # 300 -0.1 -0.1 -0.5...

Ha! The remaining issue with the flipfield.dat file is apparently one of "irregular" array dimensions! Apparently there is an underlying assumption not only that the bins along each of the...

@sq-meng thanks for the ping-pong! :) I believe we have arrived at a good solution, so marking "ready for release". (This is my hack to help when I write a...

@farhi this stuff is likely also important in several places in McXtrace, do try it out when you feel like it. In essence: * @sq-meng provided bug fixes that could...

https://github.com/McStasMcXtrace/McCode/blob/main/mcstas-comps/examples/Tests_polarization/Test_Pol_Tabled/ now has a couple of remeshing-tools and the input files alongside the instrument.

Ah, of course! I have simply overlooked the non-consistent axes bins... (.... But of course resampling to a regular 6x6x6 worked :-) ) I will remodel the part now insisting...