Problem with symmetries - "You gained too much charge"
If you run these input files the calculation stops in the first iteration with the error "You gained too much charge".
Input and output file:
Here I paste the MT charges for spin 1:
l-like charge atom s p d f total --> 1 0.05899 2.88805 0.03415 0.12132 3.10281 --> 2 1.13345 2.28644 2.26961 0.02407 5.71706 --> 3 0.79805 1.41614 24.94528 0.01330 27.17614 --> 4 0.79607 1.21032 24.94336 0.01352 26.96665 --> 5 0.79332 1.42920 5.06667 0.01308 7.30565 --> 6 0.79590 1.21131 5.06068 0.01337 7.08462
The Selenium atoms of groups 3 and 4 pick up way too much charge in the d orbitals. As the d orbitals of Se also have an LO I also looked at the eigenvalues to see whether the additional charge comes from the LOs. But it seems that there are the correct number of d states in the eigenvalue spectrum at the LO energies.
Edit: I noticed that the SOC is on in these input files. However, this does not change the behaviour (without SOC these inputs still iterate)
This is BUG because:
Symmetries should not change the results this drastically
Ideas for fixes
If you try to reproduce the problem with the MaX4 version the code gives the error
**juDFT-Error* Error message:nonsymmorphic symmetries not yet implemented for films! Error occurred in subroutine:r_inpXML Error from PE:0/32
So it maybe to do with these kinds of symmetries