Tammo van der Heide

@aradi @bhourahine No irregularities on buildbot. The minimum deviation between all eigenvalues actually is 9.163e-8 a.u.

> Due to the incaccuracy of stored Hessian, the eigenmodes, which I am pretty sure should be degenerated, split off slightly. So, the current treatment of the gauge should be...

@aradi I have substituted the large `C24O6H8_eigvec` test case with `C6H6_eigvec`.

@aradi Does this work for you?

@aradi Should address your comments now, but the hard coded OS dependent relative CWD path in `getParamSearchPaths()` is not nice.

@aradi Rebased on main.

@bhourahine Switched to Python `>=3.10`.

> LGTM, can you rebase on main (any reason not to also squash the 2 commits together?) @bhourahine Rebased on main. Strictly spoken the second commit is an unrelated bug...

Hi @andrewshyichuk, while `slateratom` and `sktwocnt` are interfaced with `libXC`, we only support a limited number of xc-functionals. An extension of this list in principle is straight-forward and requires changes...

@aradi Is this information reliable? As it contradicts my (guessed) specification of >=3.8. https://github.com/dftbplus/dftbplus/blob/a2eca8caab77096e4150fc25b20b84c56aeda568/tools/dptools/README.rst?plain=1#L63