Marvin Friede

Have you checked all paths as well as the output of `shell pkg-config --libs dftd4`? It seems like DFT-D4 is still not found.

Additional info (stack trace from latest build, 3.7.0): ``` forrtl: severe (174): SIGSEGV, segmentation fault occurred Image PC Routine Line Source dftd4 00000000021CCB5A Unknown Unknown Unknown dftd4 00000000023E6240 Unknown Unknown...

Interestingly, the CN calculation in D3 also uses the reductions ([here](https://github.com/dftd3/simple-dftd3/blob/50245008d0bf596e6da98a3f7bdeec4e6bc284fa/src/dftd3/ncoord.f90#L151)).

> Publishing scientific software in a dedicated journal is definitely beneficial for the community. It certainly provides a clear, citable reference instead of relying on a GitHub link or snapshot,...

The build fails for the second Python version. The workflow tries to compile for Python 3.8 - 3.12. I am not sure if this is the correct way or if...

@xhochy I couldn't figure out a fix. That's why I restricted the recipe to only build for Python 3.11. I plan on updating the feedstock (with changes outside the recipe...

Note that the bond energy and torsion energy change. Also note that the total energy for 6.5.0 is positive. This might be caused by a bug in the EEQ electrostatic...

We should also consider pinning tblite to a release version and not to HEAD (both meson and cmake).

Thanks for the comprehensive report. I can confirm your findings. The error starts appearing with 3.0.0. Maybe related to #197. This issue, however, only pointed out differences due to a...

Please adapt the minimum CMake version in the [README](https://github.com/dftd4/dftd4/tree/main?tab=readme-ov-file#building-from-source) and the [docs](https://github.com/dftd4/dftd4/blob/main/doc/recipe/installation.rst).