Bálint Aradi

@xavbdlt Thanks for making all those comparisons. I think, if we want to move forward, we would need a more systematic approach (as also suggested by Jörg Hutter on the...

Very good idea! Maybe, we should couple it with the possibility of providing simplified input for the Monkhorst-Pack scheme: ``` KPointsAndWeights = MonkhorstPack { GridDensity [lenghtUnit] = d1 d2 d3...

@yurivict I've never thought about that aspect. A simple timestamp in the filename would be enough, wouldn't it? (e.g. `slako-packed-20220101.tar.xz`)

Thanks, I've started a similar calculation and will try to reproduce the behavior.

Just a quick feedback: I've got a similar crash on our development server using the conda openmpi library, so at least the problem is reproducible. The error message, however, comes...

@chem-william Could you as a quick check test, whether you get different behavior if you use an old-style driver? If I use `Driver = ConjugateGradient { MovedAtoms = 17:-192}` I...

Just a quick feedback: I still have no idea, where the error comes from. On the other hand, I managed to run your system with an intel/mkl/impi binary and it...

Unfortunately, this is not limited to the MPI/Compiler combinations available on Conda. I can reproduce it on dpdevel also with gcc 11.2 & openmpi 4.1. Also there, it predictably stops...

No, in theory, CMake is supposed to find it automatically, when a dependency on OpenMP is detected. (Our targets should depend on OpenMP.) Two things: * Could you try to...

Actually, you should always adjust/extend `CMAKE_PREFIX_PATH` according to the desired libraries, as otherwise CMake often fails to find them, or even worse, finds an other version of it...